42 Commits
Author SHA1 Message Date
saidsurucuandClaude Opus 4.8 a062237474 docs: add local uv copy-paste install for Antigravity; bump to 0.2.1
README'ye Antigravity için lokal uvx kurulumunu otomatik yapan
kopyala-yapıştır komutu eklendi (~/.gemini/config/mcp_config.json).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-18 18:51:19 +03:00
Said Sürücü b69eda77af Merge pull request #28 from hburaktasyurek/tool-limit-alignment
Tool Routing: align descriptions with runtime behavior and reduce Bedesten request overhead
2026-06-16 11:53:09 +03:00
Hasan Burak Taşyürek 3927dcee8f feat(deep-research): reduce Bedesten request overhead
Use search-result metadata for Deep Research previews instead of fetching every candidate document.

This keeps the compatibility tools within upstream Bedesten rate limits and updates the README to match the current active tool set.
2026-06-13 00:07:03 +03:00
Said Sürücü 3768104679 Merge pull request #27 from Baijack-star/docs-remote-mcp-troubleshooting
Document remote MCP troubleshooting
2026-06-08 18:53:18 +03:00
saidsurucuandClaude Opus 4.7 aa580ffafc docs: update Pro version URL to yargi.betaspacestudio.com
Replace https://yargi-mcp-pro-production.up.railway.app with
https://yargi.betaspacestudio.com in:
- README.md Pro announcement at the top
- mcp_server_main.py rate-limit messages in search_bedesten_unified
  and get_bedesten_document_markdown (4 occurrences)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-06-02 18:02:50 +03:00
saidsurucuandClaude Opus 4.7 931eb3ca8f docs(readme): announce Yargı MCP Pro at the top
Add a top-of-README callout pointing to the professional version that
combines mevzuat and içtihat in a single MCP server:
https://yargi-mcp-pro-production.up.railway.app

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-31 16:53:39 +03:00
saidsurucuandClaude Opus 4.7 d258ad2375 Merge fix/anayasa-document-url-host: force correct host for AYM document URLs
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-27 14:06:24 +03:00
saidsurucuandClaude Opus 4.7 061887f870 fix(anayasa): force correct host for AYM document URLs by path
get_anayasa_document_unified 404'd when a /ND/ (Norm Denetimi) path was
supplied on the bireysel host (kararlarbilgibankasi) instead of the norm
host. Detection was netloc-first and passed the wrong-domain URL through
unchanged.

Add normalize_anayasa_document_url(): classify by path (/ND/ vs /BB/) and
re-key the host to the canonical domain, preserving query/fragment. Route
get_document_unified() off the normalized result. Verified end-to-end: the
previously-404 URL now returns the decision markdown.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-27 14:06:09 +03:00
saidsurucu ac611f840c Merge fix/kik-v2-dynamic-security-headers: KİK v2 dynamic request signing 2026-05-26 12:19:52 +03:00
saidsurucuandClaude Opus 4.7 c938f10ba2 fix(kik): generate v2 request-signing headers per-request
The KİK v2 API (ekapv2.kik.gov.tr) validates a timestamp embedded in the
X-Custom-Request-Ts header and rejects stale values with HTTP 401
"İstek zaman aşımına uğradı." The client previously sent hardcoded, captured
header values, so once that timestamp aged out every search 401'd across all
three decision types (uyusmazlik/duzenleyici/mahkeme).

Replicate the Angular HTTP interceptor: AES-192-CBC/PKCS7 encrypt a fresh uuid4
GUID and the current epoch-millis timestamp with the environment.r8fact key and
a random IV, regenerated on every request.

Verified live: all three decision types return results with hataKodu "0".

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-26 12:19:48 +03:00
Baijack-star 1356c4d020 Document remote MCP troubleshooting 2026-05-22 07:39:44 +08:00
saidsurucuandClaude Opus 4.7 5392435c7a docs(bedesten): point rate-limit message to Yargı MCP Pro beta
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-18 17:44:22 +03:00
saidsurucuandClaude Opus 4.7 96a5a538b2 perf(server): unblock event loop on rate-limit waits and markitdown
Two complementary changes to mitigate intermittent TLS handshake
timeouts and "notifications/cancelled: Bad Request" seen against the
single-worker uvicorn deployment.

1. bedesten rate-limiter back-pressure
   - Add optional ``max_wait`` to ``_TokenBucket.acquire``: if the next
     wait would exceed it, raise ``BedestenRateLimited`` immediately
     instead of sleeping. After a server-side 429 the bucket pauses for
     up to 30s; previously a queued request sat in ``asyncio.sleep``
     for that whole window, holding the worker slot and pushing the
     MCP client past its cancellation timeout.
   - ``search_bedesten_unified`` / ``get_bedesten_document_markdown``
     catch ``BedestenRateLimited`` and reuse the existing structured
     429-style response, so callers get a fast, clean retry signal.
   - Tunable via ``BEDESTEN_RATE_MAX_WAIT_S`` (default 8.0s).

2. Offload sync markitdown conversions to a thread
   - Every ``markitdown.convert*`` call site is now wrapped in
     ``asyncio.to_thread(...)`` across 14 modules (bedesten, yargitay,
     danistay, anayasa norm + bireysel, uyusmazlik, emsal, rekabet,
     gib, kvkk, sayistay, bddk, sigorta_tahkim, kik_v2). PDF / large
     HTML parsing was stalling the event loop for seconds, which on a
     single-worker deployment delayed every other in-flight request
     and queued new TLS handshakes until they timed out.

Verified locally:
- ``ast.parse`` + ``importlib.import_module`` on all 15 modified files
- ``mcp_server_main.create_app()`` constructs successfully
- New ``_TokenBucket.acquire(max_wait=...)`` smoke-tested across 6
  paths: capacity-available, no-arg backward compat, max_wait raise,
  max_wait wait+succeed, ``penalize_until`` + max_wait fast-raise.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-11 14:31:23 +03:00
saidsurucu 26aa3dacc6 fix(bedesten): include itemTypeList in fetch metadata search
The fetch tool's metadata lookup constructed BedestenSearchData without
the required itemTypeList field, causing a Pydantic validation error and
losing the chance to enrich the response with a proper title.
2026-05-08 21:58:30 +03:00
saidsurucuandClaude Opus 4.7 58457b076f feat(bedesten): client-side rate limiter with 429 back-pressure
Probed the live API (2026-05-08): the per-IP limit is 10 requests in a
rolling 30s window, with HTTP 429 + Retry-After: 30 on the 11th call.

Add a token bucket inside BedestenApiClient (default capacity=1, refill
1 token / 3.5s — strict serialization, no burst) so we stay below the
threshold by default. When the server still returns 429 (e.g. the egress
IP is shared with other clients), pause the whole bucket for the
Retry-After window so queued in-flight requests wait gracefully instead
of hammering. Tunable via BEDESTEN_RATE_CAPACITY / BEDESTEN_RATE_REFILL_S.

Verified: 14 concurrent requests after a clean cooldown -> 13 OK,
1 stray 429 (bucket auto-paused 22.5s, then drained cleanly).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-08 21:10:37 +03:00
saidsurucuandClaude Opus 4.7 4521e1de85 fix(bedesten): suggest yargi-cli as fallback in 429 message
When the Bedesten API rate-limits us, point the model at the local
yargi-cli tool (https://github.com/saidsurucu/yargi-cli) so the user
has a working alternative while waiting out the limit.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-08 20:30:49 +03:00
saidsurucuandClaude Opus 4.7 8f04010c57 fix(bedesten): return structured 429 response instead of raising
Bedesten API can intermittently return HTTP 429 Too Many Requests.
Previously the tool raised, leaving the LLM with an unhandled error.
Now search_bedesten_unified returns a dict with error="rate_limit_exceeded"
and get_bedesten_document_markdown returns a BedestenDocumentMarkdown
whose markdown_content describes the rate limit, so the model can
inform the user and retry. Non-429 errors still propagate.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-08 20:18:12 +03:00
saidsurucuandClaude Opus 4.7 7ed9c25687 docs(readme): announce migration to yargimcp.surucu.dev
Add a prominent banner at the top of the README and inline notices
near the connection instructions stating the server has moved to
https://yargimcp.surucu.dev/mcp. The old https://yargimcp.fastmcp.app/mcp
endpoint is now a migration stub that returns only a notice tool.
Update Claude Desktop and Google Antigravity config URLs to the new host.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-08 00:43:32 +03:00
saidsurucuandClaude Opus 4.7 4fdc7a3689 fix(deploy): make migration_app entrypoint a FastMCP instance
The Dokploy FastMCP build pipeline runs `fastmcp inspect <module>:app`
and expects `app` to be a FastMCP instance, not a Starlette ASGI app.
Drop the `mcp.http_app()` wrapper and bind the FastMCP instance to
`app` directly so `fastmcp inspect` and `fastmcp run --transport http`
both work. Verified locally with `fastmcp inspect` and end-to-end MCP
initialize over `fastmcp run`.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-08 00:39:02 +03:00
saidsurucuandClaude Opus 4.7 1538a4c145 feat(deploy): add migration stub MCP app pointing to new URL
migration_app.py is a minimal FastMCP server with a single
migration_notice tool. Intended for the deprecated endpoint so
existing MCP clients learn the server has moved to
https://yargimcp.surucu.dev/mcp and instruct the user to update
their client configuration.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-08 00:29:32 +03:00
saidsurucuandClaude Opus 4.7 a24def2e66 feat(deploy): add minimal ASGI app and Dockerfile for simple deploys
Mirrors the mevzuat-mcp pattern: a thin app.py exposing
mcp.http_app() with a /health route, no FastAPI/CORS/OAuth wrapper.
Dockerfile builds on python:3.12-slim and runs uvicorn directly.
Removes Dockerfile/fly.toml entries from .dockerignore so the new
Dockerfile is included in the build context.

Existing asgi_app.py (api.yargimcp.com production) is untouched.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-07 13:13:22 +03:00
saidsurucuandClaude Opus 4.7 6b781b61d2 docs(semantic_search): recommend multilingual-e5-large for Turkish (#22)
Different embedding model families need different prompt prefixes —
Gemini wants "task: ... | query: ..." and "title: ... | text: ...",
e5 wants "query: ..." / "passage: ...", and using the wrong one
silently degrades retrieval quality. Add EMBEDDING_PROMPT_STYLE
(gemini/e5/raw) so the prefix matches the chosen model.

Defaults: gemini for OpenRouter (matches the existing default
google/gemini-embedding-001), e5 for the local provider (matches
the recommended multilingual-e5-large setup). Both override via
env var or constructor.

Update README and .env.example to recommend intfloat/multilingual-
e5-large served by HuggingFace Text Embeddings Inference (one
docker run) as the Turkish-optimized local setup, with a clear env
var reference table. Ollama and OpenRouter remain documented as
alternatives.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 01:55:42 +03:00
saidsurucuandClaude Opus 4.7 fb29146755 feat(semantic_search): support local OpenAI-compatible embedding servers (#22)
Adds a LocalEmbedder that targets any OpenAI-compatible embedding
endpoint (Ollama, llama.cpp, vLLM, LM Studio, ...). Zero new
Python dependencies — reuses the existing openai SDK with a
custom base_url. Defaults to Ollama at http://localhost:11434/v1
with nomic-embed-text @ 768 dims; override via env vars for other
servers/models (e.g. bge-m3 @ 1024 dims for better Turkish).

Refactors the shared encode/similarity logic into a private base
class so OpenRouterEmbedder and LocalEmbedder don't duplicate ~50
lines. OpenRouter keeps its ranking headers; local sends none.

Adds get_embedder() factory selecting the provider based on
EMBEDDING_PROVIDER (local) or OPENROUTER_API_KEY presence, and
is_semantic_search_available() that returns True for either path.
mcp_server_main now uses these so the semantic_search tool is
exposed when only a local server is configured.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 01:50:47 +03:00
saidsurucuandClaude Opus 4.7 42731a2c03 feat(semantic_search): make embedding model configurable (#22)
google/gemini-embedding-001 became paid on OpenRouter, leaving
users without credit unable to run the semantic_search tool. The
old code hardcoded the model and 3072 dimensions in three places.

Make OpenRouterEmbedder accept model/dimension via constructor
args or OPENROUTER_EMBEDDING_MODEL / OPENROUTER_EMBEDDING_DIMENSION
env vars, with the previous values as backward-compatible defaults.
Switch the VectorStore and the response payload in mcp_server_main
to read embedder.dimension instead of the hardcoded 3072 so a
configured non-Gemini model does not produce shape mismatches.

Bad dimension input (non-int or non-positive) now raises a clear
ValueError instead of a downstream shape error.

Documented the new env vars in .env.example.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 01:46:33 +03:00
saidsurucuandClaude Opus 4.7 ae5d590cca fix(sayistay): surface clear error when upstream WAF returns 418 (#23)
Verified 2026-05-03 against a real Chrome browser: POSTs to
/KararlarGenelKurul/DataTablesList consistently return HTTP 418
with the WAF block page "Bilgi Güvenliği Politikaları Gereği
Kısıtlanmıştır", regardless of headers, cookies, CSRF token, or
form payload. The block is server-side at sayistay.gov.tr and
cannot be worked around client-side. The Temyiz Kurulu and Daire
endpoints are unaffected (29k/22k records still return normally).

Detect the 418 + WAF marker in all three search methods and raise
a clear RuntimeError explaining it is an upstream restriction,
instead of the cryptic "Client error '418 I'm a teapot'" that
hides the real situation from MCP clients.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 01:41:35 +03:00
saidsurucuandClaude Opus 4.7 ee544dc603 fix(rekabet): return parsed decisions instead of empty array (#24)
Token-optimization commit e34d81b tightened RekabetDecisionSummary
fields from Optional[str]/Optional[HttpUrl] to plain str with ""
defaults, but client.py kept passing None for unparsed cells and
HttpUrl(...) for URLs. Pydantic v2 rejected both, the broad
except Exception swallowed every row, and decisions came back []
while total_records_found stayed populated.

Default unparsed string fields to "" and pass URL strings directly
to the model. Verified against the live API for empty args,
PdfText filter, and KararTuru filter.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 01:33:52 +03:00
saidsurucuandClaude Opus 4.7 355f505da9 docs: Update README for GİB özelge module
- Intro paragraph: add GİB Özelgeleri to institution list
- Feature bullets: add GİB Özelgeleri entry with supported filters
- Tool list: new "GİB Özelge Araçları" subsection covering
  search_gib_ozelge and get_gib_ozelge_document_markdown
- Counts: tool total 22 → 24, institution total 14 → 15

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-18 18:01:57 +03:00
saidsurucuandClaude Opus 4.7 4c06a5926b Add GİB özelge (tax rulings) MCP module
Introduces two tools backed by the gib.gov.tr public JSON API
(reverse-engineered from the Next.js SPA chunks):

- search_gib_ozelge: keyword, ozelgeNo, kanunNo, date-range, paging
  over 18k+ Revenue Administration tax rulings. Simple YYYY-MM-DD
  dates are auto-expanded to ISO 8601 to satisfy the backend.
- get_gib_ozelge_document_markdown: fetch a single ruling by numeric
  id and return 5000-char paginated Markdown with a metadata header
  block (title, ozelgeNo, tarih, kanun, kaynak).

Also prunes stale auth/Fly.io-era entries from uv.lock.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-18 17:55:14 +03:00
saidsurucu 2cec4dccd6 Fix user_id not defined error in bedesten search 2026-04-03 15:28:48 +03:00
saidsurucu 5c2e9cc92b Add .serena to gitignore and remove from repo 2026-04-03 02:10:18 +03:00
saidsurucu a66a3f2053 Remove auth and Fly.io deployment 2026-04-03 02:08:56 +03:00
saidsurucu a4d9e2e53d Remove mcp_auth_http_simple.py 2026-04-03 02:02:44 +03:00
saidsurucu 036e49a928 docs: Update README with Sigorta Tahkim Komisyonu (14 institutions, 22 tools) 2026-03-09 22:40:10 +03:00
saidsurucu 7f78f87508 feat: Add Sigorta Tahkim Komisyonu MCP module (3 tools)
Add Insurance Arbitration Commission integration with Tavily search
and direct PDF download for 64 quarterly journal issues (2010-2025).

Tools:
- search_sigorta_tahkim_decisions: Search via Tavily API
- get_sigorta_tahkim_document_markdown: PDF download + paginated markdown
- search_within_sigorta_tahkim_issue: Keyword search within individual
  decisions of a journal issue, with Turkish İ/I case folding support

Total tools: 25 (was 22)
2026-03-09 22:17:27 +03:00
saidsurucu d8805cb93b fix: Reject null JSON-RPC IDs per MCP spec 2025-11-25
Monkey-patch JSONRPCNotification to use extra="forbid" so that
requests with "id": null are no longer misclassified as notifications
(202 Accepted). They now correctly fail validation and return a
-32600 Invalid Request error.
2026-02-16 01:29:17 +03:00
saidsurucuandClaude Opus 4.5 28ff2e39a5 fix: Update Bedesten document source_url to mevzuat.adalet.gov.tr format
Changed source_url from API endpoint (bedesten.adalet.gov.tr/document/{id})
to user-facing URL (mevzuat.adalet.gov.tr/ictihat/{id}) for direct browser access.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-27 20:02:41 +03:00
Said Sürücü fd08637ca2 Update README.md 2026-01-15 11:00:21 +03:00
saidsurucu a5e6baeec8 fix: FastMCP 2.12+ auth import compatibility 2026-01-15 00:59:54 +03:00
saidsurucu 8818a7809a docs: Add Google Antigravity setup instructions 2025-12-27 11:05:52 +03:00
saidsurucu 12d51e3735 fix: Emsal API null safety 2025-12-26 21:07:42 +03:00
Said Sürücü efe962abf1 Update README with new application udfcevir.com
Added a new application for professional conversion from Word to UDF.
2025-12-26 14:29:07 +03:00
saidsurucu 1d73265f10 feat: ChatGPT App compliance updates 2025-12-25 17:42:22 +03:00
54 changed files with 2300 additions and 4275 deletions
-3
View File
@@ -181,6 +181,3 @@ site
# Production logs
**/logs/*.log.*
**/Dockerfile
**/Dockerfile
fly.toml
+37 -2
View File
@@ -74,11 +74,46 @@ JWT_SECRET_KEY=your_jwt_secret_key_here
# SEMANTIC SEARCH SETTINGS (Optional)
# =============================================================================
# OpenRouter API Key for semantic search functionality
# Embedding provider for the semantic_search tool.
# Pick exactly one of: OpenRouter (hosted) or Local (your own server).
# --- Option A: OpenRouter (hosted, default) -----------------------------------
# Get your API key from: https://openrouter.ai/keys
# If not set, semantic search tool will be disabled
# If neither this nor EMBEDDING_PROVIDER=local is set, semantic search is off.
OPENROUTER_API_KEY=sk-or-v1-your_openrouter_api_key_here
# Optional: override the OpenRouter embedding model and dimension.
# Defaults: google/gemini-embedding-001 at 3072 dims (paid on OpenRouter).
# Pick any model from https://openrouter.ai/models?modality=embedding
# and set the dimension to that model's output size — they must match.
# OPENROUTER_EMBEDDING_MODEL=google/gemini-embedding-001
# OPENROUTER_EMBEDDING_DIMENSION=3072
# --- Option B: Local OpenAI-compatible server (no API key required) ----------
# Recommended for Turkish: intfloat/multilingual-e5-large served by HuggingFace
# Text Embeddings Inference (TEI). One-line setup:
#
# docker run -p 8080:80 ghcr.io/huggingface/text-embeddings-inference:latest \
# --model-id intfloat/multilingual-e5-large
#
# Then uncomment the block below. Other model families work too — set
# EMBEDDING_PROMPT_STYLE to match: e5 / gemini / raw.
#
# EMBEDDING_PROVIDER=local
# LOCAL_EMBEDDING_BASE_URL=http://localhost:8080/v1
# LOCAL_EMBEDDING_MODEL=intfloat/multilingual-e5-large
# LOCAL_EMBEDDING_DIMENSION=1024
# EMBEDDING_PROMPT_STYLE=e5
# LOCAL_EMBEDDING_API_KEY= # most local servers ignore this
#
# Ollama fallback (if you prefer Ollama and don't need top Turkish quality):
# ollama serve && ollama pull nomic-embed-text
# EMBEDDING_PROVIDER=local
# LOCAL_EMBEDDING_BASE_URL=http://localhost:11434/v1
# LOCAL_EMBEDDING_MODEL=nomic-embed-text
# LOCAL_EMBEDDING_DIMENSION=768
# EMBEDDING_PROMPT_STYLE=raw # nomic uses its own search_query/search_document
# =============================================================================
# USAGE INSTRUCTIONS
# =============================================================================
+3
View File
@@ -1,3 +1,6 @@
# Serena
.serena/
# Byte-compiled / optimized / DLL files
__pycache__/
*.py[cod]
-1
View File
@@ -1 +0,0 @@
/cache
-84
View File
@@ -1,84 +0,0 @@
# list of languages for which language servers are started; choose from:
# al bash clojure cpp csharp csharp_omnisharp
# dart elixir elm erlang fortran go
# haskell java julia kotlin lua markdown
# nix perl php python python_jedi r
# rego ruby ruby_solargraph rust scala swift
# terraform typescript typescript_vts yaml zig
# Note:
# - For C, use cpp
# - For JavaScript, use typescript
# Special requirements:
# - csharp: Requires the presence of a .sln file in the project folder.
# When using multiple languages, the first language server that supports a given file will be used for that file.
# The first language is the default language and the respective language server will be used as a fallback.
# Note that when using the JetBrains backend, language servers are not used and this list is correspondingly ignored.
languages:
- python
# the encoding used by text files in the project
# For a list of possible encodings, see https://docs.python.org/3.11/library/codecs.html#standard-encodings
encoding: "utf-8"
# whether to use the project's gitignore file to ignore files
# Added on 2025-04-07
ignore_all_files_in_gitignore: true
# list of additional paths to ignore
# same syntax as gitignore, so you can use * and **
# Was previously called `ignored_dirs`, please update your config if you are using that.
# Added (renamed) on 2025-04-07
ignored_paths: []
# whether the project is in read-only mode
# If set to true, all editing tools will be disabled and attempts to use them will result in an error
# Added on 2025-04-18
read_only: false
# list of tool names to exclude. We recommend not excluding any tools, see the readme for more details.
# Below is the complete list of tools for convenience.
# To make sure you have the latest list of tools, and to view their descriptions,
# execute `uv run scripts/print_tool_overview.py`.
#
# * `activate_project`: Activates a project by name.
# * `check_onboarding_performed`: Checks whether project onboarding was already performed.
# * `create_text_file`: Creates/overwrites a file in the project directory.
# * `delete_lines`: Deletes a range of lines within a file.
# * `delete_memory`: Deletes a memory from Serena's project-specific memory store.
# * `execute_shell_command`: Executes a shell command.
# * `find_referencing_code_snippets`: Finds code snippets in which the symbol at the given location is referenced.
# * `find_referencing_symbols`: Finds symbols that reference the symbol at the given location (optionally filtered by type).
# * `find_symbol`: Performs a global (or local) search for symbols with/containing a given name/substring (optionally filtered by type).
# * `get_current_config`: Prints the current configuration of the agent, including the active and available projects, tools, contexts, and modes.
# * `get_symbols_overview`: Gets an overview of the top-level symbols defined in a given file.
# * `initial_instructions`: Gets the initial instructions for the current project.
# Should only be used in settings where the system prompt cannot be set,
# e.g. in clients you have no control over, like Claude Desktop.
# * `insert_after_symbol`: Inserts content after the end of the definition of a given symbol.
# * `insert_at_line`: Inserts content at a given line in a file.
# * `insert_before_symbol`: Inserts content before the beginning of the definition of a given symbol.
# * `list_dir`: Lists files and directories in the given directory (optionally with recursion).
# * `list_memories`: Lists memories in Serena's project-specific memory store.
# * `onboarding`: Performs onboarding (identifying the project structure and essential tasks, e.g. for testing or building).
# * `prepare_for_new_conversation`: Provides instructions for preparing for a new conversation (in order to continue with the necessary context).
# * `read_file`: Reads a file within the project directory.
# * `read_memory`: Reads the memory with the given name from Serena's project-specific memory store.
# * `remove_project`: Removes a project from the Serena configuration.
# * `replace_lines`: Replaces a range of lines within a file with new content.
# * `replace_symbol_body`: Replaces the full definition of a symbol.
# * `restart_language_server`: Restarts the language server, may be necessary when edits not through Serena happen.
# * `search_for_pattern`: Performs a search for a pattern in the project.
# * `summarize_changes`: Provides instructions for summarizing the changes made to the codebase.
# * `switch_modes`: Activates modes by providing a list of their names
# * `think_about_collected_information`: Thinking tool for pondering the completeness of collected information.
# * `think_about_task_adherence`: Thinking tool for determining whether the agent is still on track with the current task.
# * `think_about_whether_you_are_done`: Thinking tool for determining whether the task is truly completed.
# * `write_memory`: Writes a named memory (for future reference) to Serena's project-specific memory store.
excluded_tools: []
# initial prompt for the project. It will always be given to the LLM upon activating the project
# (contrary to the memories, which are loaded on demand).
initial_prompt: ""
project_name: "yargi-mcp"
included_optional_tools: []
+22
View File
@@ -469,6 +469,28 @@ doc8 = await get_kvkk_document_markdown(decision_url="https://www.kvkk.gov.tr/Ic
- **Fallback Token**: If not set, uses a limited free token automatically
- KVKK search tools will work without configuration (with rate limits)
### Rate Limits
| API | Rate Limit | Notes |
|-----|------------|-------|
| Bedesten Unified | ~10 req / 30s window per source IP (measured 2026-05-08); 11th req → HTTP 429 with `Retry-After: 30`. Client uses an internal token bucket (default 1 token, refill 1/3.5s) plus 429 back-pressure (whole bucket pauses for the Retry-After window). Override via `BEDESTEN_RATE_CAPACITY` / `BEDESTEN_RATE_REFILL_S`. |
| Yargıtay Primary | Unknown | Official government API |
| Danıştay | Unknown | Official government API |
| Anayasa Mahkemesi | Unknown | Constitutional Court API |
| KİK v2 | Unknown | Public Procurement Authority API |
| Rekabet Kurumu | Unknown | Competition Authority API |
| Sayıştay | Unknown | Court of Accounts API |
| Uyuşmazlık | Unknown | Jurisdictional Disputes Court API |
| Emsal | Unknown | UYAP Precedent Database API |
| KVKK (Brave) | 1,000/month | Brave Search API free tier limit |
| BDDK | Unknown | Banking Regulation API |
**Recommendations:**
- Implement client-side caching for repeated queries
- Use pagination parameters to limit result sizes
- Space out requests during bulk operations
- Consider implementing retry logic with exponential backoff
### OAuth Authentication Configuration
The server uses **Clerk JWT tokens** for all authentication. **Cross-origin authentication** is implemented using Bearer JWT tokens as per Clerk's best practices.
+42 -23
View File
@@ -1,34 +1,53 @@
# -------- BASE IMAGE ---------------------------------------------------------
# Use Python 3.12 slim image
FROM python:3.12-slim
# -------- Runtime setup ----------------------------------------------------
# Set working directory
WORKDIR /app
# Copy dependency manifests first for layer-cache
COPY pyproject.toml poetry.lock* requirements*.txt* ./
# Install system dependencies (gcc/g++ kept in case any wheel falls back to source build)
RUN apt-get update && apt-get install -y --no-install-recommends \
gcc \
g++ \
&& rm -rf /var/lib/apt/lists/*
# Fast, deterministic install with `uv`
RUN pip install --no-cache-dir uv && \
uv pip install --system --no-cache-dir . && \
uv pip install --system --no-cache-dir .[asgi,saas]
# Copy project metadata first for better Docker layer caching
COPY pyproject.toml ./
COPY README.md ./
# Cache buster - force rebuild
ARG CACHE_BUST=202510061202
RUN echo "Cache bust: $CACHE_BUST"
# Copy entry points
COPY app.py ./
COPY asgi_app.py ./
COPY mcp_server_main.py ./
# Copy application source
COPY . .
# Copy MCP modules and shared packages
COPY anayasa_mcp_module ./anayasa_mcp_module
COPY bddk_mcp_module ./bddk_mcp_module
COPY bedesten_mcp_module ./bedesten_mcp_module
COPY danistay_mcp_module ./danistay_mcp_module
COPY emsal_mcp_module ./emsal_mcp_module
COPY gib_mcp_module ./gib_mcp_module
COPY kik_mcp_module ./kik_mcp_module
COPY kvkk_mcp_module ./kvkk_mcp_module
COPY rekabet_mcp_module ./rekabet_mcp_module
COPY sayistay_mcp_module ./sayistay_mcp_module
COPY sigorta_tahkim_mcp_module ./sigorta_tahkim_mcp_module
COPY uyusmazlik_mcp_module ./uyusmazlik_mcp_module
COPY yargitay_mcp_module ./yargitay_mcp_module
COPY semantic_search ./semantic_search
# -------- Environment ------------------------------------------------------
ENV PYTHONUNBUFFERED=1
ENV ENABLE_AUTH=true
ENV PORT=8000
# -------- Health check -----------------------------------------------------
HEALTHCHECK --interval=30s --timeout=10s --start-period=10s --retries=3 \
CMD python -c "import httpx, os, sys; r=httpx.get(f'http://localhost:{os.getenv(\"PORT\",\"8000\")}/health'); sys.exit(0 if r.status_code==200 else 1)"
# Install the package with ASGI extras (uvicorn + starlette)
RUN pip install --no-cache-dir -e ".[asgi]"
# Expose port
EXPOSE 8000
# -------- Entrypoint -------------------------------------------------------
CMD ["uvicorn", "asgi_app:app", "--host", "0.0.0.0", "--port", "8000", "--proxy-headers"]
# Set environment variables
ENV PORT=8000
ENV PYTHONUNBUFFERED=1
# Health check
HEALTHCHECK --interval=30s --timeout=10s --start-period=40s --retries=3 \
CMD python -c "import httpx; httpx.get('http://localhost:8000/health', timeout=5)" || exit 1
# Run the ASGI application
CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "8000"]
-1
View File
@@ -1 +0,0 @@
web: uvicorn asgi_app:app --host 0.0.0.0 --port $PORT
+172 -42
View File
@@ -1,8 +1,24 @@
# Yargı MCP: Türk Hukuk Kaynakları için MCP Sunucusu
> ## ✨ Profesyonel Sürüm Hazır: Yargı MCP Pro
>
> **Mevzuat ve içtihatı tek bir MCP sunucusunda birleştiren** profesyonel sürüm yayında:
>
> 👉 **https://yargi.betaspacestudio.com**
> ## 🚨 SUNUCU YENİ ADRESE TAŞINDI
>
> **Yeni Remote MCP adresi:** `https://yargimcp.surucu.dev/mcp`
>
> **Eski adres** (`https://yargimcp.fastmcp.app/mcp`) **artık kullanım dışıdır** — yalnızca taşındığını bildiren bir uyarı tool'u döner.
>
> **Yapmanız gereken:** MCP istemcinizdeki (Claude Desktop, 5ire, Google Antigravity, ChatGPT vb.) sunucu URL'sini yukarıdaki yeni adresle güncelleyin.
## Word'den UDF'ye profesyonel dönüşüm için yeni uygulamam [udfcevir.com](https://udfcevir.com) adresinde!
[![Star History Chart](https://api.star-history.com/svg?repos=saidsurucu/yargi-mcp&type=Date)](https://www.star-history.com/#saidsurucu/yargi-mcp&Date)
Bu proje, çeşitli Türk hukuk kaynaklarına (Yargıtay, Danıştay, Emsal Kararlar, Uyuşmazlık Mahkemesi, Anayasa Mahkemesi - Norm Denetimi ile Bireysel Başvuru Kararları, Kamu İhale Kurulu Kararları, Rekabet Kurumu Kararları, Sayıştay Kararları, KVKK Kararları ve BDDK Kararları) erişimi kolaylaştıran bir [FastMCP](https://gofastmcp.com/) sunucusu oluşturur. Bu sayede, bu kaynaklardan veri arama ve belge getirme işlemleri, Model Context Protocol (MCP) destekleyen LLM (Büyük Dil Modeli) uygulamaları (örneğin Claude Desktop veya [5ire](https://5ire.app)) ve diğer istemciler tarafından araç (tool) olarak kullanılabilir hale gelir.
Bu proje, çeşitli Türk hukuk kaynaklarına (Yargıtay, Danıştay, Emsal Kararlar, Uyuşmazlık Mahkemesi, Anayasa Mahkemesi - Norm Denetimi ile Bireysel Başvuru Kararları, Kamu İhale Kurulu Kararları, Rekabet Kurumu Kararları, Sayıştay Kararları, KVKK Kararları, BDDK Kararları, GİB Özelgeleri ve Sigorta Tahkim Komisyonu Kararları) erişimi kolaylaştıran bir [FastMCP](https://gofastmcp.com/) sunucusu oluşturur. Bu sayede, bu kaynaklardan veri arama ve belge getirme işlemleri, Model Context Protocol (MCP) destekleyen LLM (Büyük Dil Modeli) uygulamaları (örneğin Claude Desktop veya [5ire](https://5ire.app)) ve diğer istemciler tarafından araç (tool) olarak kullanılabilir hale gelir.
---
@@ -10,19 +26,76 @@ Bu proje, çeşitli Türk hukuk kaynaklarına (Yargıtay, Danıştay, Emsal Kara
### ✅ Kurulum Gerektirmez! Hemen Kullan!
🔗 **Remote MCP Adresi:** `https://yargimcp.fastmcp.app/mcp`
🔗 **Remote MCP Adresi:** `https://yargimcp.surucu.dev/mcp`
### Claude Desktop ile Kullanım
> ⚠️ **Eski adres** `https://yargimcp.fastmcp.app/mcp` **artık kullanım dışıdır** — yalnızca taşındığını bildiren bir uyarı tool'u döner. Lütfen yukarıdaki yeni adresi kullanın.
### Claude Desktop ile Kullanım (Ücretli abonelik gerekir)
1. **Claude Desktop'ı açın**
2. **Settings → Connectors → Add Custom Connector**
3. **Bilgileri girin:**
- **Name:** `Yargı MCP`
- **URL:** `https://yargimcp.fastmcp.app/mcp`
- **URL:** `https://yargimcp.surucu.dev/mcp`
4. **Add** butonuna tıklayın
5. **Hemen kullanmaya başlayın!** 🎉
> 💡 **İpucu:** Remote MCP sayesinde Python, uv veya herhangi bir kurulum yapmadan doğrudan Claude Desktop üzerinden Türk hukuk kaynaklarına erişebilirsiniz!
### Google Antigravity ile Kullanım (Lokal `uv` Kurulumu — Kopyala-Yapıştır)
> **Ön Gereksinimler:** Bilgisayarınızda **Python**, **`uv`** ([kurulum](https://docs.astral.sh/uv/getting-started/installation/)) ve **Node.js** ([indir](https://nodejs.org/en/download)) kurulu olmalı. (Node.js yalnızca aşağıdaki kurulum komutunu çalıştırmak için gerekir; MCP'yi `uvx` çalıştırır.)
Aşağıdaki **bloğun tamamını** terminale yapıştırın. Komut, Antigravity'nin okuduğu `~/.gemini/config/mcp_config.json` dosyasını sizin yerinize oluşturur/günceller (varsa diğer sunucularınız korunur):
**macOS / Linux** (Terminal):
```bash
node - <<'YARGI'
const fs=require("fs"),os=require("os"),path=require("path");
const dir=path.join(os.homedir(),".gemini","config"),file=path.join(dir,"mcp_config.json");
fs.mkdirSync(dir,{recursive:true});
let cfg={};try{cfg=JSON.parse(fs.readFileSync(file,"utf8"))}catch{}
if(typeof cfg!=="object"||cfg===null||Array.isArray(cfg))cfg={};
if(typeof cfg.mcpServers!=="object"||cfg.mcpServers===null)cfg.mcpServers={};
cfg.mcpServers["yargi-mcp"]={command:"uvx",args:["yargi-mcp"]};
fs.writeFileSync(file,JSON.stringify(cfg,null,2)+"\n");
console.log("yargi-mcp eklendi -> "+file);
YARGI
```
**Windows** (PowerShell):
```powershell
@'
const fs=require("fs"),os=require("os"),path=require("path");
const dir=path.join(os.homedir(),".gemini","config"),file=path.join(dir,"mcp_config.json");
fs.mkdirSync(dir,{recursive:true});
let cfg={};try{cfg=JSON.parse(fs.readFileSync(file,"utf8"))}catch{}
if(typeof cfg!=="object"||cfg===null||Array.isArray(cfg))cfg={};
if(typeof cfg.mcpServers!=="object"||cfg.mcpServers===null)cfg.mcpServers={};
cfg.mcpServers["yargi-mcp"]={command:"uvx",args:["yargi-mcp"]};
fs.writeFileSync(file,JSON.stringify(cfg,null,2)+"\n");
console.log("yargi-mcp eklendi -> "+file);
'@ | node -
```
Komut `yargi-mcp eklendi -> ...` çıktısını verdiğinde kurulum tamamlanmıştır. Antigravity'yi (açıksa kapatıp) yeniden başlatın; `yargi-mcp` araçları otomatik yüklenir.
> 💡 **İpucu:** Lokal kurulumda hukuk kaynaklarına erişim doğrudan bilgisayarınızda `uvx yargi-mcp` ile çalışır; uzaktan sunucuya ihtiyaç duymaz.
### Remote MCP Sorun Giderme
`https://yargimcp.surucu.dev/mcp` bir web sayfası değil, Streamable HTTP MCP uç noktasıdır. Tarayıcıda açınca veya düz `curl` ile GET isteği atınca `406 Not Acceptable` ve `Client must accept text/event-stream` benzeri bir yanıt görmek normaldir; bu, sunucunun kapalı olduğu anlamına gelmez. MCP istemcisi `Accept: application/json, text/event-stream` başlığıyla JSON-RPC isteği göndermelidir.
Hızlı sağlık kontrolü için tarayıcıda şu adresleri açabilirsiniz:
- `https://yargimcp.surucu.dev/health` — servis sağlık durumu
Claude.ai veya başka bir istemci "araç yok" gibi davranırsa:
1. Connector'ı kaldırıp yeniden ekleyin.
2. URL olarak önce `https://yargimcp.surucu.dev/mcp` deneyin; istemciniz yönlendirmeleri takip etmiyorsa `https://yargimcp.surucu.dev/mcp/` deneyin.
3. Eski `https://yargimcp.fastmcp.app/mcp` adresinin istemci ayarlarında veya önbellekte kalmadığından emin olun.
4. İstemcinin remote/Streamable HTTP MCP desteklediğini ve `text/event-stream` kabul ettiğini kontrol edin.
---
@@ -52,6 +125,8 @@ Bu proje, çeşitli Türk hukuk kaynaklarına (Yargıtay, Danıştay, Emsal Kara
* **Sayıştay:** 3 karar türü ile kapsamlı denetim kararlarına erişim + **8 Daire Filtreleme** + **Tarih Aralığı & İçerik Arama** (Genel Kurul yorumlayıcı kararları, Temyiz Kurulu itiraz kararları, Daire ilk derece denetim kararları)
* **KVKK (Kişisel Verilerin Korunması Kurulu):** Brave Search API ile veri koruma kararlarını arama; uzun karar metinlerini (5.000 karakterlik) sayfalanmış Markdown formatında getirme + **Türkçe Arama** + **Site Hedeflemeli Arama** (kvkk.gov.tr kararları)
* **BDDK (Bankacılık Düzenleme ve Denetleme Kurumu):** Bankacılık düzenleme kararlarını arama; karar metinlerini Markdown formatında getirme + **Optimized Search** + **"Karar Sayısı" Targeting** + **Spesifik URL Filtreleme** (bddk.org.tr/Mevzuat/DokumanGetir)
* **GİB (Gelir İdaresi Başkanlığı) Özelgeleri:** Resmi vergi özelgelerini arama (18.000+ özelge: KDV, Kurumlar, Gelir, ÖTV, Damga vb.); tam metni sayfalanmış Markdown formatında getirme + **Keyword + Özelge No + Kanun No + Tarih Aralığı** + **Otomatik ISO 8601 Dönüşümü** + **Metadata Başlık Bloğu**
* **Sigorta Tahkim Komisyonu:** Hakem Karar Dergisi (64 sayı, 2010-2025) içindeki sigorta tahkim kararlarını arama; dergi PDF'lerini Markdown formatında getirme + **Sayı İçi Karar Arama** + **Türkçe Büyük/Küçük Harf Desteği** + **Relevance Scoring**
* Karar metinlerinin daha kolay işlenebilmesi için Markdown formatına çevrilmesi.
* Claude Desktop uygulaması ile `fastmcp install` komutu kullanılarak kolay entegrasyon.
@@ -156,20 +231,40 @@ Yargı MCP'yi Gemini CLI ile kullanmak için:
---
<details>
<summary>🧠 <strong>Semantik Arama (Opsiyonel - OpenRouter API)</strong></summary>
<summary>🧠 <strong>Semantik Arama (Opsiyonel)</strong></summary>
Yargı MCP, **semantik arama** özelliği ile kararları anlamsal olarak sıralayabilir. Bu özellik opsiyoneldir ve `OPENROUTER_API_KEY` ayarlandığında otomatik olarak etkinleşir.
Yargı MCP, **semantik arama** özelliği ile kararları anlamsal olarak sıralayabilir. Opsiyoneldir; iki yoldan biri yapılandırıldığında otomatik etkinleşir:
- **Yerel** (önerilen, ücretsiz): kendi makinenizdeki OpenAI-uyumlu embedding sunucusu (HuggingFace TEI, llama.cpp, Ollama, vLLM, LM Studio…)
- **Hosted**: OpenRouter API anahtarı
### Semantik Arama Nasıl Çalışır?
1. `initial_keyword` ile Bedesten API'den 100 karar çekilir
2. `query` ile bu kararlar embedding modeli kullanılarak anlamsal olarak sıralanır
3. En alakalı kararlar döndürülür
### OpenRouter API Anahtarı Alma
1. [OpenRouter](https://openrouter.ai/) sitesine gidin
2. Hesap oluşturun ve API anahtarı alın (ücretsiz kredi ile başlayabilirsiniz)
### Önerilen Türkçe Kurulumu (Yerel — `multilingual-e5-large`)
### Claude Desktop için Yapılandırma
`intfloat/multilingual-e5-large` Türkçe için kıyas ettiğimiz açık kaynak modeller arasında en iyilerinden. HuggingFace'in **Text Embeddings Inference (TEI)** sunucusuyla tek komutta ayağa kalkar ve OpenAI-uyumlu API sunar:
```bash
docker run -p 8080:80 ghcr.io/huggingface/text-embeddings-inference:latest \
--model-id intfloat/multilingual-e5-large
```
Sonra Yargı MCP'ye şu env vars'ları geçirin:
```bash
EMBEDDING_PROVIDER=local
LOCAL_EMBEDDING_BASE_URL=http://localhost:8080/v1
LOCAL_EMBEDDING_MODEL=intfloat/multilingual-e5-large
LOCAL_EMBEDDING_DIMENSION=1024
EMBEDDING_PROMPT_STYLE=e5
```
> ⚠️ **Önemli:** `EMBEDDING_PROMPT_STYLE=e5` şart — e5 modelleri `query:` / `passage:` öneki bekleyecek şekilde eğitilmiştir; yanlış önek sessizce kaliteyi düşürür.
#### Claude Desktop örneği (yerel TEI)
```json
{
"mcpServers": {
@@ -177,42 +272,67 @@ Yargı MCP, **semantik arama** özelliği ile kararları anlamsal olarak sırala
"command": "uvx",
"args": ["yargi-mcp"],
"env": {
"OPENROUTER_API_KEY": "sk-or-v1-xxx..."
"EMBEDDING_PROVIDER": "local",
"LOCAL_EMBEDDING_BASE_URL": "http://localhost:8080/v1",
"LOCAL_EMBEDDING_MODEL": "intfloat/multilingual-e5-large",
"LOCAL_EMBEDDING_DIMENSION": "1024",
"EMBEDDING_PROMPT_STYLE": "e5"
}
}
}
}
```
### 5ire için Yapılandırma
Tool ayarlarında **Environment Variables** alanına ekleyin:
### Alternatif 1: Ollama (yerel, daha hafif kurulum)
```bash
ollama serve
ollama pull nomic-embed-text # 768 dim, İngilizce ağırlıklı
```
```bash
EMBEDDING_PROVIDER=local
LOCAL_EMBEDDING_BASE_URL=http://localhost:11434/v1
LOCAL_EMBEDDING_MODEL=nomic-embed-text
LOCAL_EMBEDDING_DIMENSION=768
EMBEDDING_PROMPT_STYLE=raw
```
> Ollama kütüphanesinde `multilingual-e5-large` doğrudan yok; Türkçe için TEI yolu daha doğru sonuç verir.
### Alternatif 2: OpenRouter (hosted)
```bash
OPENROUTER_API_KEY=sk-or-v1-xxx...
# İsteğe bağlı — varsayılan google/gemini-embedding-001 (3072 dim, ÜCRETLİ)
# OPENROUTER_EMBEDDING_MODEL=...
# OPENROUTER_EMBEDDING_DIMENSION=...
# EMBEDDING_PROMPT_STYLE=gemini # varsayılan
```
### Gemini CLI için Yapılandırma
```json
{
"mcpServers": {
"yargi_mcp": {
"command": "uvx",
"args": ["yargi-mcp"],
"env": {
"OPENROUTER_API_KEY": "sk-or-v1-xxx..."
}
}
}
}
```
API anahtarınızı [openrouter.ai/keys](https://openrouter.ai/keys) adresinden alın. Varsayılan model `google/gemini-embedding-001` artık ücretli — ücretsiz bir model seçerseniz `OPENROUTER_EMBEDDING_MODEL`, `OPENROUTER_EMBEDDING_DIMENSION` ve uygun `EMBEDDING_PROMPT_STYLE` değerlerini birlikte ayarlayın.
> 💡 **Not:** `OPENROUTER_API_KEY` ayarlanmazsa semantik arama aracı görünmez, diğer 19 araç normal şekilde çalışmaya devam eder.
### Yapılandırma Referansı
| Env Var | Açıklama | Örnek |
|---|---|---|
| `EMBEDDING_PROVIDER` | `local` ise yerel sunucu, boş ise OpenRouter | `local` |
| `EMBEDDING_PROMPT_STYLE` | `gemini` / `e5` / `raw` — modelin beklediği önek | `e5` |
| `LOCAL_EMBEDDING_BASE_URL` | Yerel sunucunun OpenAI-uyumlu URL'i | `http://localhost:8080/v1` |
| `LOCAL_EMBEDDING_MODEL` | Model adı | `intfloat/multilingual-e5-large` |
| `LOCAL_EMBEDDING_DIMENSION` | Modelin çıktı boyutu (mutlaka eşleşmeli) | `1024` |
| `OPENROUTER_API_KEY` | OpenRouter anahtarı (sadece hosted için) | `sk-or-v1-…` |
| `OPENROUTER_EMBEDDING_MODEL` | OpenRouter model id'si | `google/gemini-embedding-001` |
| `OPENROUTER_EMBEDDING_DIMENSION` | OpenRouter modelinin çıktı boyutu | `3072` |
> 💡 **Not:** Hiçbir embedding sağlayıcı yapılandırılmazsa semantik arama aracı görünmez, diğer 24 araç normal şekilde çalışır.
</details>
<details>
<summary>🛠️ <strong>Kullanılabilir Araçlar (MCP Tools)</strong></summary>
Bu FastMCP sunucusu **19 temel MCP aracı** + **1 opsiyonel semantik arama aracı** sunar (token verimliliği için optimize edilmiş):
Bu FastMCP sunucusu **26 aktif MCP aracı** + **1 opsiyonel semantik arama aracı** sunar (token verimliliği için optimize edilmiş):
### **Yargıtay Araçları (Birleşik Bedesten API - Token Optimized)**
*Not: Yargıtay araçları token verimliliği için birleşik Bedesten API'ye entegre edilmiştir*
@@ -237,8 +357,8 @@ Bu FastMCP sunucusu **19 temel MCP aracı** + **1 opsiyonel semantik arama arac
8. `get_anayasa_document_unified(document_url, page_number)`: AYM kararlarını birleşik belge getirme - **sayfalanmış Markdown** içeriği
### **KİK (Kamu İhale Kurulu) Araçları**
9. `search_kik_decisions(karar_tipi, ...)`: KİK (Kamu İhale Kurulu) kararlarını arar.
10. `get_kik_document_markdown(karar_id, page_number)`: Belirli bir KİK kararını, Base64 ile encode edilmiş `karar_id`'sini kullanarak alır ve **sayfalanmış Markdown** içeriğini getirir.
9. `search_kik_v2_decisions(decision_type, karar_metni, karar_no, basvuran, idare_adi, baslangic_tarihi, bitis_tarihi)`: KİK v2 API ile uyuşmazlık, düzenleyici ve mahkeme kararlarını arar.
10. `get_kik_v2_document_markdown(gundemMaddesiId)`: Arama sonucundaki `gundemMaddesiId` ile KİK karar metnini Markdown formatında getirir.
### **Rekabet Kurumu Araçları**
    * `search_rekabet_kurumu_decisions(KararTuru: Literal[...], ...) -> RekabetSearchResult`: Rekabet Kurumu kararlarını arar. `KararTuru` için kullanıcı dostu isimler kullanılır (örn: "Birleşme ve Devralma").
    * `get_rekabet_kurumu_document(karar_id: str, page_number: Optional[int] = 1) -> RekabetDocument`: Belirli bir Rekabet Kurumu kararını `karar_id` ile alır. Kararın PDF formatındaki orijinalinden istenen sayfayı ayıklar ve Markdown formatında döndürür.
@@ -246,22 +366,32 @@ Bu FastMCP sunucusu **19 temel MCP aracı** + **1 opsiyonel semantik arama arac
---
* **Sayıştay Araçları (3 Karar Türü + 8 Daire Filtreleme):**
* `search_sayistay_genel_kurul(karar_no, karar_tarih_baslangic, karar_tamami, ...)`: Sayıştay Genel Kurul (yorumlayıcı) kararlarını arar. **Tarih aralığı** (2006-2024) + **İçerik arama** (400 karakter)
* `search_sayistay_temyiz_kurulu(ilam_dairesi, kamu_idaresi_turu, temyiz_karar, ...)`: Temyiz Kurulu (itiraz) kararlarını arar. **8 Daire filtreleme** + **Kurum türü** + **Konu sınıflandırması**
* `search_sayistay_daire(yargilama_dairesi, web_karar_metni, hesap_yili, ...)`: Daire (ilk derece denetim) kararlarını arar. **8 Daire filtreleme** + **Hesap yılı** + **İçerik arama**
* `get_sayistay_genel_kurul_document_markdown(decision_id: str)`: Genel Kurul kararının tam metnini Markdown formatında getirir
* `get_sayistay_temyiz_kurulu_document_markdown(decision_id: str)`: Temyiz Kurulu kararının tam metnini Markdown formatında getirir
* `get_sayistay_daire_document_markdown(decision_id: str)`: Daire kararının tam metnini Markdown formatında getirir
* **Sayıştay Araçları (Birleşik API, 3 Karar Türü + 8 Daire Filtreleme):**
* `search_sayistay_unified(decision_type, start, length, ...)`: `genel_kurul`, `temyiz_kurulu` veya `daire` kararlarını tek araçla arar. `length` 1-100 aralığındadır.
* `get_sayistay_document_unified(decision_id, decision_type)`: Birleşik arama sonucundaki karar ID'si ve karar türüyle tam metni Markdown formatında getirir.
* **KVKK Araçları (Brave Search API + Türkçe Arama):**
* `search_kvkk_decisions(keywords, page, pageSize, ...)`: KVKK (Kişisel Verilerin Korunması Kurulu) kararlarını Brave Search API ile arar. **Türkçe arama** + **Site hedeflemeli** (`site:kvkk.gov.tr "karar özeti"`) + **Sayfalama desteği**
* `search_kvkk_decisions(keywords, page)`: KVKK (Kişisel Verilerin Korunması Kurulu) kararlarını Brave Search API ile arar. **Türkçe arama** + **Site hedeflemeli** (`site:kvkk.gov.tr "karar özeti"`) + **Sayfalama desteği**. Sonuç sayısı sunucuda 10 olarak sabitlenmiştir.
* `get_kvkk_document_markdown(decision_url: str, page_number: Optional[int] = 1)`: KVKK kararının tam metnini **sayfalanmış Markdown** formatında getirir (5.000 karakterlik sayfa)
### BDDK Araçları
* `search_bddk_decisions(keywords, page)`: BDDK (Bankacılık Düzenleme ve Denetleme Kurumu) kararlarını arar. **"Karar Sayısı" targeting** + **Spesifik URL filtreleme** (`bddk.org.tr/Mevzuat/DokumanGetir`) + **Optimized search**
* `get_bddk_document_markdown(document_id: str, page_number: Optional[int] = 1)`: BDDK kararının tam metnini **sayfalanmış Markdown** formatında getirir (5.000 karakterlik sayfa)
### GİB (Gelir İdaresi Başkanlığı) Özelge Araçları (Resmi GİB JSON API)
* `search_gib_ozelge(keywords, ozelgeNo, kanunNo, ozelgeStartDate, ozelgeEndDate, page, pageSize)`: GİB özelgelerini (Türk Gelir İdaresi Başkanlığı vergi özelgeleri) arar — **18.000+ özelge** (KDV, Kurumlar, Gelir, ÖTV, Damga, VUK vb.). **Keyword + Özelge No + Kanun No + Tarih Aralığı** + **Otomatik ISO 8601 Dönüşümü** (`YYYY-MM-DD` girdileri otomatik olarak full ISO 8601'e çevrilir)
* `get_gib_ozelge_document_markdown(ozelge_id: int, page_number: int = 1)`: Belirli bir özelgenin tam metnini **sayfalanmış Markdown** formatında getirir (5.000 karakterlik sayfa) + **Metadata başlık bloğu** (Başlık, Sayı, Tarih, Kanun, Kaynak URL)
### Sigorta Tahkim Komisyonu Araçları (Tavily Search API + PDF)
* `search_sigorta_tahkim_decisions(keywords, page)`: Sigorta Tahkim Komisyonu kararlarını Tavily Search API ile arar. **Site hedeflemeli** (`sigortatahkim.org`) + **Sayfalama desteği**. Sonuç sayısı sunucuda 10 olarak sabitlenmiştir.
* `get_sigorta_tahkim_document_markdown(issue_number: str, page_number: int)`: Hakem Karar Dergisi sayısının PDF'ini indirip **sayfalanmış Markdown** formatında getirir (5.000 karakterlik sayfa). 64 sayı (2010-2025)
* `search_within_sigorta_tahkim_issue(issue_number: str, keyword: str, max_results: int)`: Belirli bir dergi sayısı içindeki kararları anahtar kelime ile arar. **Türkçe İ/I desteği** + **Relevance scoring** + **Excerpt** ile sonuç
### Yardımcı ve Uyumluluk Araçları
* `check_government_servers_health()`: Yargı kaynaklarının erişilebilirliğini kontrol eder.
* `search(query)`: ChatGPT Deep Research uyumluluğu için Bedesten destekli kaynaklarda arama yapar.
* `fetch(id)`: ChatGPT Deep Research uyumluluğu için tek bir Bedesten belge ID'sinin tam metnini getirir.
</details>
---
@@ -276,8 +406,8 @@ Bu FastMCP sunucusu **19 temel MCP aracı** + **1 opsiyonel semantik arama arac
- **Korunan İşlevsellik:** %100 özellik desteği devam ediyor
**GENEL İSTATİSTİKLER:**
- **Toplam Mahkeme/Kurum:** 13 farklı hukuki kurum (KVKK dahil)
- **Toplam MCP Tool:** 19 temel araç + 1 opsiyonel semantik arama aracı
- **Toplam Mahkeme/Kurum:** 15 farklı hukuki kurum (GİB Özelgeleri ve Sigorta Tahkim Komisyonu dahil)
- **Toplam MCP Tool:** 26 aktif araç + 1 opsiyonel semantik arama aracı
- **Daire/Kurul Filtreleme:** 87 farklı seçenek (52 Yargıtay + 27 Danıştay + 8 Sayıştay)
- **Tarih Filtreleme:** Birleşik Bedesten API aracında ISO 8601 formatında tam tarih aralığı desteği
- **Kesin Cümle Arama:** Birleşik Bedesten API aracında çift tırnak ile tam cümle arama (`"\"mülkiyet kararı\""` formatı)
+2 -1
View File
@@ -1,6 +1,7 @@
# anayasa_mcp_module/bireysel_client.py
# This client is for Bireysel Başvuru: https://kararlarbilgibankasi.anayasa.gov.tr
import asyncio
import httpx
from bs4 import BeautifulSoup, Tag
from typing import Dict, Any, List, Optional, Tuple
@@ -302,7 +303,7 @@ class AnayasaBireyselBasvuruApiClient:
elif "Karar Tarihi" in key and not karar_tarihi_from_page: karar_tarihi_from_page = value
elif "Resmi Gazete Tarih / Sayı" in key: resmi_gazete_info_from_page = value
full_markdown_content = self._convert_html_to_markdown_bireysel(html_content_from_api)
full_markdown_content = await asyncio.to_thread(self._convert_html_to_markdown_bireysel, html_content_from_api)
if not full_markdown_content:
return AnayasaBireyselBasvuruDocumentMarkdown(
+2 -1
View File
@@ -1,6 +1,7 @@
# anayasa_mcp_module/client.py
# This client is for Norm Denetimi: https://normkararlarbilgibankasi.anayasa.gov.tr
import asyncio
import httpx
from bs4 import BeautifulSoup
from typing import Dict, Any, List, Optional, Tuple
@@ -309,7 +310,7 @@ class AnayasaMahkemesiApiClient:
official_gazette_from_page = rg_text_content.replace("Resmî Gazete tarih ve sayısı:", "").replace("Resmi Gazete tarih/sayı:", "").strip()
full_markdown_content = self._convert_html_to_markdown_norm_denetimi(html_content_from_api)
full_markdown_content = await asyncio.to_thread(self._convert_html_to_markdown_norm_denetimi, html_content_from_api)
if not full_markdown_content:
return AnayasaDocumentMarkdown(
+65 -15
View File
@@ -2,8 +2,8 @@
# Unified client for both Norm Denetimi and Bireysel Başvuru
import logging
from typing import Optional
from urllib.parse import urlparse
from typing import Optional, Tuple
from urllib.parse import urlparse, urlunparse
from .models import (
AnayasaUnifiedSearchRequest,
@@ -18,6 +18,51 @@ from .bireysel_client import AnayasaBireyselBasvuruApiClient
logger = logging.getLogger(__name__)
# Canonical hosts per decision type. Norm Denetimi (/ND/) documents live on the
# "norm" subdomain; Bireysel Başvuru (/BB/) documents on the plain subdomain.
# Callers (or upstream search links) sometimes supply the wrong host for a given
# path, which makes the AYM server return 404. We re-key the host off the path.
_NORM_HOST = "normkararlarbilgibankasi.anayasa.gov.tr"
_BIREYSEL_HOST = "kararlarbilgibankasi.anayasa.gov.tr"
def normalize_anayasa_document_url(document_url: str) -> Tuple[Optional[str], str]:
"""Detect the AYM decision type from the URL path and force the correct host.
Detection is path-based (``/ND/`` vs ``/BB/``) because the path is
unambiguous, whereas the supplied host may be wrong. Query params and
fragment are preserved (they are harmless for document fetches).
Returns ``(decision_type, normalized_url)`` where ``decision_type`` is
``"norm_denetimi"``, ``"bireysel_basvuru"``, or ``None`` if it cannot be
determined (URL returned unchanged in that case).
"""
parsed = urlparse(document_url)
path = parsed.path or ""
if "/ND/" in path:
decision_type, host = "norm_denetimi", _NORM_HOST
elif "/BB/" in path:
decision_type, host = "bireysel_basvuru", _BIREYSEL_HOST
else:
# Fall back to host-based detection when the path is uninformative.
if "normkararlarbilgibankasi" in parsed.netloc:
return "norm_denetimi", document_url
if "kararlarbilgibankasi" in parsed.netloc:
return "bireysel_basvuru", document_url
return None, document_url
normalized = urlunparse((
parsed.scheme or "https",
host,
parsed.path,
parsed.params,
parsed.query,
parsed.fragment,
))
return decision_type, normalized
class AnayasaUnifiedClient:
"""Unified client that handles both Norm Denetimi and Bireysel Başvuru searches."""
@@ -80,13 +125,19 @@ class AnayasaUnifiedClient:
async def get_document_unified(self, document_url: str, page_number: int = 1) -> AnayasaUnifiedDocumentMarkdown:
"""Unified document retrieval that auto-detects the appropriate client."""
# Auto-detect decision type based on URL
parsed_url = urlparse(document_url)
if "normkararlarbilgibankasi" in parsed_url.netloc or "/ND/" in document_url:
# Norm Denetimi document
result = await self.norm_client.get_decision_document_as_markdown(document_url, page_number)
# Auto-detect decision type from the path and force the correct host.
# This repairs malformed URLs (e.g. a /ND/ path on the bireysel host),
# which otherwise 404 against the AYM server.
decision_type, normalized_url = normalize_anayasa_document_url(document_url)
if normalized_url != document_url:
logger.info(
f"AnayasaUnifiedClient: Normalized document URL "
f"'{document_url}' -> '{normalized_url}'"
)
if decision_type == "norm_denetimi":
result = await self.norm_client.get_decision_document_as_markdown(normalized_url, page_number)
return AnayasaUnifiedDocumentMarkdown(
decision_type="norm_denetimi",
source_url=result.source_url,
@@ -96,11 +147,10 @@ class AnayasaUnifiedClient:
total_pages=result.total_pages,
is_paginated=result.is_paginated
)
elif "kararlarbilgibankasi" in parsed_url.netloc or "/BB/" in document_url:
# Bireysel Başvuru document
result = await self.bireysel_client.get_decision_document_as_markdown(document_url, page_number)
elif decision_type == "bireysel_basvuru":
result = await self.bireysel_client.get_decision_document_as_markdown(normalized_url, page_number)
return AnayasaUnifiedDocumentMarkdown(
decision_type="bireysel_basvuru",
source_url=result.source_url,
@@ -110,7 +160,7 @@ class AnayasaUnifiedClient:
total_pages=result.total_pages,
is_paginated=result.is_paginated
)
else:
raise ValueError(f"Cannot determine document type from URL: {document_url}")
+35
View File
@@ -0,0 +1,35 @@
"""
ASGI application for Yargı MCP Server (simple deployment variant).
This is a minimal ASGI application that can be run with:
uvicorn app:app --host 0.0.0.0 --port 8000
The MCP server will be available at:
http://localhost:8000/mcp/
For the FastAPI-wrapped variant with CORS and extra metadata routes,
see asgi_app.py instead.
"""
from starlette.responses import JSONResponse
from mcp_server_main import create_app
mcp = create_app()
@mcp.custom_route("/health", methods=["GET"])
async def health_check(request):
"""Health check endpoint for monitoring services (Fly.io, Render, etc.)."""
return JSONResponse({
"status": "healthy",
"service": "Yargı MCP Server",
"version": "0.2.1",
})
# Create ASGI app directly from FastMCP server
app = mcp.http_app()
# Endpoints:
# - /mcp/ - MCP server (Streamable HTTP transport, default FastMCP path)
# - /health - Health check for monitoring
+22 -465
View File
@@ -2,98 +2,32 @@
ASGI application for Yargı MCP Server
This module provides ASGI/HTTP access to the Yargı MCP server,
allowing it to be deployed as a web service with FastAPI wrapper
for OAuth integration and proper middleware support.
allowing it to be deployed as a web service with FastAPI wrapper.
Usage:
uvicorn asgi_app:app --host 0.0.0.0 --port 8000
"""
import os
import time
import logging
import json
from datetime import datetime, timedelta
from fastapi import FastAPI, Request, HTTPException, Query
from fastapi.responses import JSONResponse, HTMLResponse, Response
from fastapi.exception_handlers import http_exception_handler
import logging
from fastapi import FastAPI, Request
from fastapi.responses import JSONResponse
from starlette.middleware import Middleware
from starlette.middleware.cors import CORSMiddleware
from starlette.middleware.base import BaseHTTPMiddleware
# Import the proper create_app function that includes all middleware
from mcp_server_main import create_app
# Conditional auth-related imports (only if auth enabled)
_auth_check = os.getenv("ENABLE_AUTH", "false").lower() == "true"
if _auth_check:
# Import MCP Auth HTTP adapter (OAuth endpoints)
try:
from mcp_auth_http_simple import router as mcp_auth_router
except ImportError:
mcp_auth_router = None
# Import Stripe webhook router
try:
from stripe_webhook import router as stripe_router
except ImportError:
stripe_router = None
else:
mcp_auth_router = None
stripe_router = None
# OAuth configuration from environment variables
CLERK_ISSUER = os.getenv("CLERK_ISSUER", "https://clerk.yargimcp.com")
BASE_URL = os.getenv("BASE_URL", "https://api.yargimcp.com")
CLERK_SECRET_KEY = os.getenv("CLERK_SECRET_KEY")
CLERK_PUBLISHABLE_KEY = os.getenv("CLERK_PUBLISHABLE_KEY")
# Setup logging
logger = logging.getLogger(__name__)
# Configure CORS and Auth middleware
# Configure CORS
cors_origins = os.getenv("ALLOWED_ORIGINS", "*").split(",")
# Import FastMCP Bearer Auth Provider
from fastmcp.server.auth import BearerAuthProvider
from fastmcp.server.auth.providers.bearer import RSAKeyPair
# Create MCP app
mcp_server = create_app()
# Import Clerk SDK at module level for performance
try:
from clerk_backend_api import Clerk
CLERK_SDK_AVAILABLE = True
except ImportError:
CLERK_SDK_AVAILABLE = False
logger.warning("Clerk SDK not available - falling back to development mode")
# Configure Bearer token authentication based on ENABLE_AUTH
auth_enabled = os.getenv("ENABLE_AUTH", "false").lower() == "true"
bearer_auth = None
if CLERK_SECRET_KEY and CLERK_ISSUER:
# Production: Use Clerk JWKS endpoint for token validation
bearer_auth = BearerAuthProvider(
jwks_uri=f"{CLERK_ISSUER}/.well-known/jwks.json",
issuer=None,
algorithm="RS256",
audience=None,
required_scopes=[]
)
else:
# Development: Generate RSA key pair for testing
dev_key_pair = RSAKeyPair.generate()
bearer_auth = BearerAuthProvider(
public_key=dev_key_pair.public_key,
issuer="https://dev.yargimcp.com",
audience="dev-mcp-server",
required_scopes=["yargi.read"]
)
# Create MCP app with Bearer authentication
mcp_server = create_app(auth=bearer_auth if auth_enabled else None)
# Create MCP Starlette sub-application with root path - mount will add /mcp prefix
# Create MCP Starlette sub-application
mcp_app = mcp_server.http_app(path="/")
@@ -104,7 +38,7 @@ class UTF8JSONResponse(JSONResponse):
headers = {}
headers["Content-Type"] = "application/json; charset=utf-8"
super().__init__(content, status_code, headers, **kwargs)
def render(self, content) -> bytes:
return json.dumps(
content,
@@ -119,46 +53,22 @@ custom_middleware = [
CORSMiddleware,
allow_origins=cors_origins,
allow_credentials=True,
allow_methods=["GET", "POST", "OPTIONS", "DELETE"],
allow_headers=["Content-Type", "Authorization", "X-Request-ID", "X-Session-ID"],
allow_methods=["GET", "POST", "OPTIONS"],
allow_headers=["Content-Type", "X-Request-ID", "X-Session-ID"],
),
]
# Create FastAPI wrapper application
app = FastAPI(
title="Yargı MCP Server",
description="MCP server for Turkish legal databases with OAuth authentication",
description="MCP server for Turkish legal databases",
version="0.1.0",
middleware=custom_middleware,
default_response_class=UTF8JSONResponse, # Use UTF-8 JSON encoder
redirect_slashes=False # Disable to prevent 307 redirects on /mcp endpoint
default_response_class=UTF8JSONResponse,
redirect_slashes=False,
)
# Add auth-related routers to FastAPI (only if available)
if stripe_router:
app.include_router(stripe_router, prefix="/api/stripe")
if mcp_auth_router:
app.include_router(mcp_auth_router)
# Custom 401 exception handler for MCP spec compliance
@app.exception_handler(401)
async def custom_401_handler(request: Request, exc: HTTPException):
"""Custom 401 handler that adds WWW-Authenticate header as required by MCP spec"""
response = await http_exception_handler(request, exc)
# Add WWW-Authenticate header pointing to protected resource metadata
# as required by RFC 9728 Section 5.1 and MCP Authorization spec
response.headers["WWW-Authenticate"] = (
'Bearer '
'error="invalid_token", '
'error_description="The access token is missing or invalid", '
f'resource="{BASE_URL}/.well-known/oauth-protected-resource"'
)
return response
# FastAPI health check endpoint - BEFORE mounting MCP app
@app.get("/health")
async def health_check():
"""Health check endpoint for monitoring"""
@@ -167,119 +77,33 @@ async def health_check():
"service": "Yargı MCP Server",
"version": "0.1.0",
"tools_count": len(mcp_server._tool_manager._tools),
"auth_enabled": os.getenv("ENABLE_AUTH", "false").lower() == "true"
}
# Add explicit redirect for /mcp to /mcp/ with method preservation
@app.api_route("/mcp", methods=["GET", "POST", "HEAD", "OPTIONS"])
async def redirect_to_slash(request: Request):
"""Redirect /mcp to /mcp/ preserving HTTP method with 308"""
from fastapi.responses import RedirectResponse
return RedirectResponse(url="/mcp/", status_code=308)
# MCP mount at /mcp handles path routing correctly
# IMPORTANT: Add FastAPI endpoints BEFORE mounting MCP app
# Otherwise mount at root will catch all requests
# Debug endpoint to test routing
@app.get("/debug/test")
async def debug_test():
"""Debug endpoint to test if FastAPI routes work"""
return {"message": "FastAPI routes working", "debug": True}
# Clerk CORS proxy endpoints
@app.api_route("/clerk-proxy/{path:path}", methods=["GET", "POST", "PUT", "DELETE", "OPTIONS"])
async def clerk_cors_proxy(request: Request, path: str):
"""
Proxy requests to Clerk to bypass CORS restrictions.
Forwards requests from Claude AI to clerk.yargimcp.com with proper CORS headers.
"""
import httpx
# Build target URL
clerk_url = f"https://clerk.yargimcp.com/{path}"
# Forward query parameters
if request.url.query:
clerk_url += f"?{request.url.query}"
# Copy headers (exclude host/origin)
headers = dict(request.headers)
headers.pop('host', None)
headers.pop('origin', None)
headers['origin'] = 'https://yargimcp.com' # Use our frontend domain
try:
async with httpx.AsyncClient() as client:
# Forward the request to Clerk
if request.method == "OPTIONS":
# Handle preflight
response = await client.request(
method=request.method,
url=clerk_url,
headers=headers
)
else:
# Forward body for POST/PUT requests
body = None
if request.method in ["POST", "PUT", "PATCH"]:
body = await request.body()
response = await client.request(
method=request.method,
url=clerk_url,
headers=headers,
content=body
)
# Create response with CORS headers
response_headers = dict(response.headers)
response_headers.update({
"Access-Control-Allow-Origin": "*",
"Access-Control-Allow-Methods": "GET, POST, PUT, DELETE, OPTIONS",
"Access-Control-Allow-Headers": "Content-Type, Authorization, Accept, Origin, X-Requested-With",
"Access-Control-Allow-Credentials": "true",
"Access-Control-Max-Age": "86400"
})
return Response(
content=response.content,
status_code=response.status_code,
headers=response_headers,
media_type=response.headers.get("content-type")
)
except Exception as e:
return JSONResponse(
{"error": "proxy_error", "message": str(e)},
status_code=500,
headers={"Access-Control-Allow-Origin": "*"}
)
# FastAPI root endpoint
@app.get("/")
async def root():
"""Root endpoint with service information"""
return {
"service": "Yargı MCP Server",
"description": "MCP server for Turkish legal databases with OAuth authentication",
"description": "MCP server for Turkish legal databases",
"endpoints": {
"mcp": "/mcp",
"health": "/health",
"status": "/status",
"stripe_webhook": "/api/stripe/webhook",
"oauth_login": "/auth/login",
"oauth_callback": "/auth/callback",
"oauth_google": "/auth/google/login",
"user_info": "/auth/user"
},
"transports": {
"http": "/mcp"
},
"supported_databases": [
"Yargıtay (Court of Cassation)",
"Danıştay (Council of State)",
"Danıştay (Council of State)",
"Emsal (Precedent)",
"Uyuşmazlık Mahkemesi (Court of Jurisdictional Disputes)",
"Anayasa Mahkemesi (Constitutional Court)",
@@ -288,140 +112,12 @@ async def root():
"Sayıştay (Court of Accounts)",
"KVKK (Personal Data Protection Authority)",
"BDDK (Banking Regulation and Supervision Agency)",
"Bedesten API (Multiple courts)"
"Bedesten API (Multiple courts)",
"Sigorta Tahkim Komisyonu (Insurance Arbitration Commission)",
],
"authentication": {
"enabled": os.getenv("ENABLE_AUTH", "false").lower() == "true",
"type": "OAuth 2.0 via Clerk",
"issuer": CLERK_ISSUER,
"providers": ["google"],
"flow": "authorization_code"
}
}
# OAuth 2.0 Authorization Server Metadata - MCP standard location
@app.get("/.well-known/oauth-authorization-server")
async def oauth_authorization_server_root():
"""OAuth 2.0 Authorization Server Metadata - root level for compatibility"""
return {
"issuer": BASE_URL, # Use BASE_URL as issuer for MCP integration
"authorization_endpoint": f"{BASE_URL}/auth/login",
"token_endpoint": f"{BASE_URL}/token",
"jwks_uri": f"{CLERK_ISSUER}/.well-known/jwks.json",
"response_types_supported": ["code"],
"grant_types_supported": ["authorization_code", "refresh_token"],
"token_endpoint_auth_methods_supported": ["client_secret_basic", "none"],
"scopes_supported": ["read", "search", "openid", "profile", "email"],
"subject_types_supported": ["public"],
"id_token_signing_alg_values_supported": ["RS256"],
"claims_supported": ["sub", "iss", "aud", "exp", "iat", "email", "name"],
"code_challenge_methods_supported": ["S256"],
"service_documentation": f"{BASE_URL}/mcp",
"registration_endpoint": f"{BASE_URL}/register",
"resource_documentation": f"{BASE_URL}/mcp"
}
# Claude AI MCP specific endpoint format - suffix versions
@app.get("/.well-known/oauth-authorization-server/mcp")
async def oauth_authorization_server_mcp_suffix():
"""OAuth 2.0 Authorization Server Metadata - Claude AI MCP specific format"""
return {
"issuer": BASE_URL, # Use BASE_URL as issuer for MCP integration
"authorization_endpoint": f"{BASE_URL}/auth/login",
"token_endpoint": f"{BASE_URL}/token",
"jwks_uri": f"{CLERK_ISSUER}/.well-known/jwks.json",
"response_types_supported": ["code"],
"grant_types_supported": ["authorization_code", "refresh_token"],
"token_endpoint_auth_methods_supported": ["client_secret_basic", "none"],
"scopes_supported": ["read", "search", "openid", "profile", "email"],
"subject_types_supported": ["public"],
"id_token_signing_alg_values_supported": ["RS256"],
"claims_supported": ["sub", "iss", "aud", "exp", "iat", "email", "name"],
"code_challenge_methods_supported": ["S256"],
"service_documentation": f"{BASE_URL}/mcp",
"registration_endpoint": f"{BASE_URL}/register",
"resource_documentation": f"{BASE_URL}/mcp"
}
@app.get("/.well-known/oauth-protected-resource/mcp")
async def oauth_protected_resource_mcp_suffix():
"""OAuth 2.0 Protected Resource Metadata - Claude AI MCP specific format"""
return {
"resource": BASE_URL,
"authorization_servers": [
BASE_URL
],
"scopes_supported": ["read", "search"],
"bearer_methods_supported": ["header"],
"resource_documentation": f"{BASE_URL}/mcp",
"resource_policy_uri": f"{BASE_URL}/privacy"
}
# OAuth 2.0 Protected Resource Metadata (RFC 9728) - MCP Spec Required
@app.get("/.well-known/oauth-protected-resource")
async def oauth_protected_resource():
"""OAuth 2.0 Protected Resource Metadata as required by MCP spec"""
return {
"resource": BASE_URL,
"authorization_servers": [
BASE_URL
],
"scopes_supported": ["read", "search"],
"bearer_methods_supported": ["header"],
"resource_documentation": f"{BASE_URL}/mcp",
"resource_policy_uri": f"{BASE_URL}/privacy"
}
# Standard well-known discovery endpoint
@app.get("/.well-known/mcp")
async def well_known_mcp():
"""Standard MCP discovery endpoint"""
return {
"mcp_server": {
"name": "Yargı MCP Server",
"version": "0.1.0",
"endpoint": f"{BASE_URL}/mcp",
"authentication": {
"type": "oauth2",
"authorization_url": f"{BASE_URL}/auth/login",
"scopes": ["read", "search"]
},
"capabilities": ["tools", "resources"],
"tools_count": len(mcp_server._tool_manager._tools)
}
}
# MCP Discovery endpoint for ChatGPT integration
@app.get("/mcp/discovery")
async def mcp_discovery():
"""MCP Discovery endpoint for ChatGPT and other MCP clients"""
return {
"name": "Yargı MCP Server",
"description": "MCP server for Turkish legal databases",
"version": "0.1.0",
"protocol": "mcp",
"transport": "http",
"endpoint": "/mcp",
"authentication": {
"type": "oauth2",
"authorization_url": "/auth/login",
"token_url": "/token",
"scopes": ["read", "search"],
"provider": "clerk"
},
"capabilities": {
"tools": True,
"resources": True,
"prompts": False
},
"tools_count": len(mcp_server._tool_manager._tools),
"contact": {
"url": BASE_URL,
"email": "support@yargi-mcp.dev"
}
}
# FastAPI status endpoint
@app.get("/status")
async def status():
"""Status endpoint with detailed information"""
@@ -431,159 +127,20 @@ async def status():
"name": tool.name,
"description": tool.description[:100] + "..." if len(tool.description) > 100 else tool.description
})
return {
"status": "operational",
"tools": tools,
"total_tools": len(tools),
"transport": "streamable_http",
"architecture": "FastAPI wrapper + MCP Starlette sub-app",
"auth_status": "enabled" if os.getenv("ENABLE_AUTH", "false").lower() == "true" else "disabled"
}
# Simplified OAuth session validation for callback endpoints only
async def validate_clerk_session_for_oauth(request: Request, clerk_token: str = None) -> str:
"""Validate Clerk session for OAuth callback endpoints only (not for MCP endpoints)"""
try:
# Use Clerk SDK if available
if not CLERK_SDK_AVAILABLE:
raise ImportError("Clerk SDK not available")
clerk = Clerk(bearer_auth=CLERK_SECRET_KEY)
# Try JWT token first (from URL parameter)
if clerk_token:
try:
return "oauth_user_from_token"
except Exception as e:
pass
# Fallback to cookie validation
clerk_session = request.cookies.get("__session")
if not clerk_session:
raise HTTPException(status_code=401, detail="No Clerk session found")
# Validate session with Clerk
session = clerk.sessions.verify_session(clerk_session)
return session.user_id
except ImportError:
return "dev_user_123"
except Exception as e:
raise HTTPException(status_code=401, detail=f"OAuth session validation failed: {str(e)}")
# MCP OAuth Callback Endpoint
@app.get("/auth/mcp-callback")
async def mcp_oauth_callback(request: Request, clerk_token: str = Query(None)):
"""Handle OAuth callback for MCP token generation"""
try:
# Validate Clerk session with JWT token support
user_id = await validate_clerk_session_for_oauth(request, clerk_token)
# Return success response
return HTMLResponse(f"""
<html>
<head>
<title>MCP Connection Successful</title>
<style>
body {{ font-family: Arial, sans-serif; text-align: center; padding: 50px; }}
.success {{ color: #28a745; }}
.token {{ background: #f8f9fa; padding: 15px; border-radius: 5px; margin: 20px 0; word-break: break-all; }}
</style>
</head>
<body>
<h1 class="success">✅ MCP Connection Successful!</h1>
<p>Your Yargı MCP integration is now active.</p>
<div class="token">
<strong>Authentication:</strong><br>
<code>Use your Clerk JWT token directly with Bearer authentication</code>
</div>
<p>You can now close this window and return to your MCP client.</p>
<script>
// Try to close the popup if opened as such
if (window.opener) {{
window.opener.postMessage({{
type: 'MCP_AUTH_SUCCESS',
token: 'use_clerk_jwt_token'
}}, '*');
setTimeout(() => window.close(), 3000);
}}
</script>
</body>
</html>
""")
except HTTPException as e:
return HTMLResponse(f"""
<html>
<head>
<title>MCP Connection Failed</title>
<style>
body {{ font-family: Arial, sans-serif; text-align: center; padding: 50px; }}
.error {{ color: #dc3545; }}
.debug {{ background: #f8f9fa; padding: 10px; margin: 20px 0; border-radius: 5px; font-family: monospace; }}
</style>
</head>
<body>
<h1 class="error">❌ MCP Connection Failed</h1>
<p>{e.detail}</p>
<div class="debug">
<strong>Debug Info:</strong><br>
Clerk Token: {'✅ Provided' if clerk_token else '❌ Missing'}<br>
Error: {e.detail}<br>
Status: {e.status_code}
</div>
<p>Please try again or contact support.</p>
<a href="https://yargimcp.com/sign-in">Return to Sign In</a>
</body>
</html>
""", status_code=e.status_code)
except Exception as e:
return HTMLResponse(f"""
<html>
<head>
<title>MCP Connection Error</title>
<style>
body {{ font-family: Arial, sans-serif; text-align: center; padding: 50px; }}
.error {{ color: #dc3545; }}
</style>
</head>
<body>
<h1 class="error">❌ Unexpected Error</h1>
<p>An unexpected error occurred during authentication.</p>
<p>Error: {str(e)}</p>
<a href="https://yargimcp.com/sign-in">Return to Sign In</a>
</body>
</html>
""", status_code=500)
# OAuth2 Token Endpoint - Now uses Clerk JWT tokens directly
@app.post("/auth/mcp-token")
async def mcp_token_endpoint(request: Request):
"""OAuth2 token endpoint for MCP clients - returns Clerk JWT token info"""
try:
# Validate Clerk session
user_id = await validate_clerk_session_for_oauth(request)
return {
"message": "Use your Clerk JWT token directly with Bearer authentication",
"token_type": "Bearer",
"scope": "yargi.read",
"user_id": user_id,
"instructions": "Include 'Authorization: Bearer YOUR_CLERK_JWT_TOKEN' in your requests"
}
except HTTPException as e:
return JSONResponse(
status_code=e.status_code,
content={"error": "invalid_request", "error_description": e.detail}
)
# Mount MCP app at /mcp/ with trailing slash
# Mount MCP app at /mcp/
app.mount("/mcp/", mcp_app)
# Set the lifespan context after mounting
app.router.lifespan_context = mcp_app.lifespan
# Export for uvicorn
__all__ = ["app"]
__all__ = ["app"]
+10 -4
View File
@@ -1,5 +1,6 @@
# bddk_mcp_module/client.py
import asyncio
import httpx
from typing import List, Optional, Dict, Any
import logging
@@ -210,14 +211,19 @@ class BddkApiClient:
# Convert to Markdown based on content type
if "pdf" in content_type:
# Handle PDF documents
# Handle PDF documents. markitdown is sync; offload to thread
# so PDF parsing doesn't block the event-loop / other requests.
pdf_stream = io.BytesIO(response.content)
result = self.markitdown.convert_stream(pdf_stream, file_extension=".pdf")
result = await asyncio.to_thread(
self.markitdown.convert_stream, pdf_stream, file_extension=".pdf"
)
markdown_content = result.text_content
else:
# Handle HTML documents
# Handle HTML documents (sync conversion offloaded to thread)
html_stream = io.BytesIO(response.content)
result = self.markitdown.convert_stream(html_stream, file_extension=".html")
result = await asyncio.to_thread(
self.markitdown.convert_stream, html_stream, file_extension=".html"
)
markdown_content = result.text_content
# Clean up the markdown content
+124 -11
View File
@@ -1,11 +1,15 @@
# bedesten_mcp_module/client.py
import httpx
import asyncio
import base64
from typing import Optional
import logging
from markitdown import MarkItDown
import io
import logging
import os
import time
from typing import Optional
import httpx
from markitdown import MarkItDown
from .models import (
BedestenSearchRequest, BedestenSearchResponse,
@@ -16,6 +20,72 @@ from .enums import get_full_birim_adi
logger = logging.getLogger(__name__)
class BedestenRateLimited(Exception):
"""Raised when the local rate-limit bucket would block longer than allowed.
Carries the suggested retry-after (seconds) so callers can surface a
structured 429-style response to the MCP client instead of silently
blocking the event-loop slot for the full bucket-pause window.
"""
def __init__(self, retry_after: float) -> None:
self.retry_after = retry_after
super().__init__(f"local bucket would block {retry_after:.1f}s")
class _TokenBucket:
"""Asyncio token bucket with explicit back-pressure.
Measured Bedesten limit (per source IP, 2026-05-08): 10 requests per
rolling 30s window with full refill — equivalent to capacity=10,
refill_rate=1 token / 3s. Even with margin, 429s still leak through
when other clients share the egress IP, so we also expose
``penalize_until`` so callers can freeze the bucket when the server
actually returns 429 (Retry-After).
"""
def __init__(self, capacity: int, refill_per_s: float) -> None:
self.capacity = float(capacity)
self.refill_per_s = float(refill_per_s)
self._tokens = float(capacity)
self._last = time.monotonic()
self._not_before = 0.0
self._lock = asyncio.Lock()
async def acquire(self, max_wait: Optional[float] = None) -> None:
"""Acquire one token. If ``max_wait`` is set and the next wait would
exceed it, raise :class:`BedestenRateLimited` immediately instead of
sleeping — keeps a single rate-limited request from holding the
worker-slot for the full bucket-pause window (up to ~30s on 429)."""
deadline = (time.monotonic() + max_wait) if max_wait is not None else None
while True:
async with self._lock:
now = time.monotonic()
if now < self._not_before:
wait_s = self._not_before - now
else:
self._tokens = min(
self.capacity,
self._tokens + (now - self._last) * self.refill_per_s,
)
self._last = now
if self._tokens >= 1.0:
self._tokens -= 1.0
return
wait_s = (1.0 - self._tokens) / self.refill_per_s
if deadline is not None:
remaining = deadline - time.monotonic()
if wait_s > remaining:
raise BedestenRateLimited(retry_after=wait_s)
await asyncio.sleep(wait_s)
def penalize_until(self, monotonic_deadline: float) -> None:
"""Pause the bucket until ``monotonic_deadline`` (drains tokens)."""
self._not_before = max(self._not_before, monotonic_deadline)
self._tokens = 0.0
self._last = time.monotonic()
class BedestenApiClient:
"""
API Client for Bedesten (bedesten.adalet.gov.tr) - Alternative legal decision search system.
@@ -25,6 +95,17 @@ class BedestenApiClient:
SEARCH_ENDPOINT = "/emsal-karar/searchDocuments"
DOCUMENT_ENDPOINT = "/emsal-karar/getDocumentContent"
# Measured limit (per source IP): 10 requests per 30s window with full
# refill (≈ 1 token / 3s steady). We default to 1-token capacity and
# 3.5s spacing (no burst, ~14% safety margin). Override via env:
# BEDESTEN_RATE_CAPACITY (default 1)
# BEDESTEN_RATE_REFILL_S (default 3.5; seconds per token)
# BEDESTEN_RATE_MAX_WAIT_S (default 8.0; max seconds to wait in the
# local bucket before returning a structured 429 to the caller)
_DEFAULT_CAPACITY = int(os.getenv("BEDESTEN_RATE_CAPACITY", "1"))
_DEFAULT_REFILL_S = float(os.getenv("BEDESTEN_RATE_REFILL_S", "3.5"))
_DEFAULT_MAX_WAIT_S = float(os.getenv("BEDESTEN_RATE_MAX_WAIT_S", "8.0"))
def __init__(self, request_timeout: float = 60.0):
self.http_client = httpx.AsyncClient(
base_url=self.BASE_URL,
@@ -42,6 +123,24 @@ class BedestenApiClient:
},
timeout=request_timeout
)
self._bucket = _TokenBucket(
capacity=self._DEFAULT_CAPACITY,
refill_per_s=1.0 / self._DEFAULT_REFILL_S,
)
def _handle_429(self, response: httpx.Response, op: str) -> None:
"""Apply back-pressure to the shared bucket based on Retry-After."""
retry_after_raw = response.headers.get("Retry-After", "")
try:
retry_after = float(retry_after_raw)
except (TypeError, ValueError):
retry_after = 30.0
# Cap penalty so a hostile/buggy server can't freeze us indefinitely.
retry_after = max(1.0, min(retry_after, 60.0))
self._bucket.penalize_until(time.monotonic() + retry_after + 0.5)
logger.warning(
f"BedestenApiClient: 429 on {op}; bucket paused {retry_after + 0.5:.1f}s"
)
async def search_documents(self, search_request: BedestenSearchRequest) -> BedestenSearchResponse:
"""
@@ -63,16 +162,19 @@ class BedestenApiClient:
if not request_dict["data"]["birimAdi"]: # Remove if empty string
del request_dict["data"]["birimAdi"]
await self._bucket.acquire(max_wait=self._DEFAULT_MAX_WAIT_S)
response = await self.http_client.post(
self.SEARCH_ENDPOINT,
self.SEARCH_ENDPOINT,
json=request_dict
)
if response.status_code == 429:
self._handle_429(response, "search")
response.raise_for_status()
response_json = response.json()
# Parse and return the response
return BedestenSearchResponse(**response_json)
except httpx.RequestError as e:
logger.error(f"BedestenApiClient: HTTP request error during search: {e}")
raise
@@ -94,10 +196,13 @@ class BedestenApiClient:
)
# Get document
await self._bucket.acquire(max_wait=self._DEFAULT_MAX_WAIT_S)
response = await self.http_client.post(
self.DOCUMENT_ENDPOINT,
json=doc_request.model_dump()
)
if response.status_code == 429:
self._handle_429(response, f"document {document_id}")
response.raise_for_status()
response_json = response.json()
doc_response = BedestenDocumentResponse(**response_json)
@@ -122,12 +227,20 @@ class BedestenApiClient:
logger.info(f"BedestenApiClient: Document mime type: {mime_type}")
# Convert to markdown based on mime type
# Convert to markdown based on mime type. markitdown is sync and
# PDF parsing in particular can block the event-loop for seconds,
# which on a single-worker uvicorn deployment stalls every other
# in-flight MCP request and new TLS handshakes. Offload to a
# thread so the event-loop stays responsive.
if mime_type == "text/html":
html_content = content_bytes.decode('utf-8')
markdown_content = self._convert_html_to_markdown(html_content)
markdown_content = await asyncio.to_thread(
self._convert_html_to_markdown, html_content
)
elif mime_type == "application/pdf":
markdown_content = self._convert_pdf_to_markdown(content_bytes)
markdown_content = await asyncio.to_thread(
self._convert_pdf_to_markdown, content_bytes
)
else:
logger.warning(f"Unsupported mime type: {mime_type}")
markdown_content = f"Unsupported content type: {mime_type}. Unable to convert to markdown."
@@ -135,7 +248,7 @@ class BedestenApiClient:
return BedestenDocumentMarkdown(
documentId=document_id,
markdown_content=markdown_content,
source_url=f"{self.BASE_URL}/document/{document_id}",
source_url=f"https://mevzuat.adalet.gov.tr/ictihat/{document_id}",
mime_type=mime_type
)
+3 -2
View File
@@ -1,7 +1,8 @@
# danistay_mcp_module/client.py
import asyncio
import httpx
from bs4 import BeautifulSoup
from bs4 import BeautifulSoup
from typing import Dict, Any, List, Optional
import logging
import html
@@ -170,7 +171,7 @@ class DanistayApiClient:
source_url=source_url
)
markdown_content = self._convert_html_to_markdown_danistay(html_content_from_api)
markdown_content = await asyncio.to_thread(self._convert_html_to_markdown_danistay, html_content_from_api)
return DanistayDocumentMarkdown(
id=id,
-66
View File
@@ -1,66 +0,0 @@
version: '3.8'
services:
yargi-mcp:
build: .
image: yargi-mcp:latest
container_name: yargi-mcp-server
ports:
- "${PORT:-8000}:8000"
environment:
- HOST=0.0.0.0
- PORT=8000
- LOG_LEVEL=${LOG_LEVEL:-info}
- ALLOWED_ORIGINS=${ALLOWED_ORIGINS:-*}
- API_TOKEN=${API_TOKEN:-}
- PYTHONUNBUFFERED=1
volumes:
# Mount logs directory
- ./logs:/app/logs
# Mount .env file if it exists
- ./.env:/app/.env:ro
restart: unless-stopped
healthcheck:
test: ["CMD", "python", "-c", "import httpx; httpx.get('http://localhost:8000/health').raise_for_status()"]
interval: 30s
timeout: 10s
retries: 3
start_period: 10s
networks:
- yargi-network
# Optional: Nginx reverse proxy
nginx:
image: nginx:alpine
container_name: yargi-nginx
ports:
- "80:80"
- "443:443"
volumes:
- ./nginx.conf:/etc/nginx/nginx.conf:ro
- ./ssl:/etc/nginx/ssl:ro
depends_on:
- yargi-mcp
networks:
- yargi-network
profiles:
- production
# Optional: Redis for caching (future enhancement)
redis:
image: redis:alpine
container_name: yargi-redis
command: redis-server --appendonly yes
volumes:
- redis-data:/data
networks:
- yargi-network
profiles:
- with-cache
networks:
yargi-network:
driver: bridge
volumes:
redis-data:
-428
View File
@@ -1,428 +0,0 @@
# Yargı MCP Server Dağıtım Rehberi
Bu rehber, Yargı MCP Server'ın ASGI web servisi olarak çeşitli dağıtım seçeneklerini kapsar.
## İçindekiler
- [Hızlı Başlangıç](#hızlı-başlangıç)
- [Yerel Geliştirme](#yerel-geliştirme)
- [Production Dağıtımı](#production-dağıtımı)
- [Cloud Dağıtımı](#cloud-dağıtımı)
- [Docker Dağıtımı](#docker-dağıtımı)
- [Güvenlik Hususları](#güvenlik-hususları)
- [İzleme](#izleme)
## Hızlı Başlangıç
### 1. Bağımlılıkları Yükleyin
```bash
# ASGI sunucusu için uvicorn yükleyin
pip install uvicorn
# Veya tüm bağımlılıklarla birlikte yükleyin
pip install -e .
pip install uvicorn
```
### 2. Sunucuyu Çalıştırın
```bash
# Temel başlatma
python run_asgi.py
# Veya doğrudan uvicorn ile
uvicorn asgi_app:app --host 0.0.0.0 --port 8000
```
Sunucu şu adreslerde kullanılabilir olacak:
- MCP Endpoint: `http://localhost:8000/mcp/`
- Sağlık Kontrolü: `http://localhost:8000/health`
- API Durumu: `http://localhost:8000/status`
## Yerel Geliştirme
### Otomatik Yeniden Yükleme ile Geliştirme Sunucusu
```bash
python run_asgi.py --reload --log-level debug
```
### FastAPI Entegrasyonunu Kullanma
Ek REST API endpoint'leri için:
```bash
uvicorn fastapi_app:app --reload
```
Bu şunları sağlar:
- `/docs` adresinde interaktif API dokümantasyonu
- `/api/tools` adresinde araç listesi
- `/api/databases` adresinde veritabanı bilgileri
### Ortam Değişkenleri
`.env.example` dosyasını temel alarak bir `.env` dosyası oluşturun:
```bash
cp .env.example .env
```
Temel değişkenler:
- `HOST`: Sunucu host adresi (varsayılan: 127.0.0.1)
- `PORT`: Sunucu portu (varsayılan: 8000)
- `ALLOWED_ORIGINS`: CORS kökenleri (virgülle ayrılmış)
- `LOG_LEVEL`: Log seviyesi (debug, info, warning, error)
## Production Dağıtımı
### 1. Uvicorn ile Çoklu Worker Kullanımı
```bash
python run_asgi.py --host 0.0.0.0 --port 8000 --workers 4
```
### 2. Gunicorn Kullanımı
```bash
pip install gunicorn
gunicorn asgi_app:app -w 4 -k uvicorn.workers.UvicornWorker --bind 0.0.0.0:8000
```
### 3. Nginx Reverse Proxy ile
1. Nginx'i yükleyin
2. Sağlanan `nginx.conf` dosyasını kullanın:
```bash
sudo cp nginx.conf /etc/nginx/sites-available/yargi-mcp
sudo ln -s /etc/nginx/sites-available/yargi-mcp /etc/nginx/sites-enabled/
sudo nginx -t
sudo systemctl reload nginx
```
### 4. Systemd Servisi
`/etc/systemd/system/yargi-mcp.service` dosyasını oluşturun:
```ini
[Unit]
Description=Yargı MCP Server
After=network.target
[Service]
Type=exec
User=www-data
WorkingDirectory=/opt/yargi-mcp
Environment="PATH=/opt/yargi-mcp/venv/bin"
ExecStart=/opt/yargi-mcp/venv/bin/uvicorn asgi_app:app --host 0.0.0.0 --port 8000 --workers 4
Restart=on-failure
RestartSec=5
[Install]
WantedBy=multi-user.target
```
Etkinleştirin ve başlatın:
```bash
sudo systemctl enable yargi-mcp
sudo systemctl start yargi-mcp
```
## Cloud Dağıtımı
### Heroku
1. `Procfile` oluşturun:
```
web: uvicorn asgi_app:app --host 0.0.0.0 --port $PORT
```
2. Dağıtın:
```bash
heroku create uygulama-isminiz
git push heroku main
```
### Railway
1. `railway.json` ekleyin:
```json
{
"build": {
"builder": "NIXPACKS"
},
"deploy": {
"startCommand": "uvicorn asgi_app:app --host 0.0.0.0 --port $PORT"
}
}
```
2. Railway CLI veya GitHub entegrasyonu ile dağıtın
### Google Cloud Run
1. Container oluşturun:
```bash
docker build -t yargi-mcp .
docker tag yargi-mcp gcr.io/PROJE_ADINIZ/yargi-mcp
docker push gcr.io/PROJE_ADINIZ/yargi-mcp
```
2. Dağıtın:
```bash
gcloud run deploy yargi-mcp \
--image gcr.io/PROJE_ADINIZ/yargi-mcp \
--platform managed \
--region us-central1 \
--allow-unauthenticated
```
### AWS Lambda (Mangum kullanarak)
1. Mangum'u yükleyin:
```bash
pip install mangum
```
2. `lambda_handler.py` oluşturun:
```python
from mangum import Mangum
from asgi_app import app
handler = Mangum(app, lifespan="off")
```
3. AWS SAM veya Serverless Framework kullanarak dağıtın
## Docker Dağıtımı
### Tek Container
```bash
# Oluşturun
docker build -t yargi-mcp .
# Çalıştırın
docker run -p 8000:8000 --env-file .env yargi-mcp
```
### Docker Compose
```bash
# Geliştirme
docker-compose up
# Nginx ile Production
docker-compose --profile production up
# Redis önbellekleme ile
docker-compose --profile with-cache up
```
### Kubernetes
Deployment YAML oluşturun:
```yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: yargi-mcp
spec:
replicas: 3
selector:
matchLabels:
app: yargi-mcp
template:
metadata:
labels:
app: yargi-mcp
spec:
containers:
- name: yargi-mcp
image: yargi-mcp:latest
ports:
- containerPort: 8000
env:
- name: HOST
value: "0.0.0.0"
- name: PORT
value: "8000"
livenessProbe:
httpGet:
path: /health
port: 8000
initialDelaySeconds: 10
periodSeconds: 30
---
apiVersion: v1
kind: Service
metadata:
name: yargi-mcp-service
spec:
selector:
app: yargi-mcp
ports:
- port: 80
targetPort: 8000
type: LoadBalancer
```
## Güvenlik Hususları
### 1. Kimlik Doğrulama
`API_TOKEN` ortam değişkenini ayarlayarak token kimlik doğrulamasını etkinleştirin:
```bash
export API_TOKEN=gizli-token-degeri
```
Ardından isteklere ekleyin:
```bash
curl -H "Authorization: Bearer gizli-token-degeri" http://localhost:8000/api/tools
```
### 2. HTTPS/SSL
Production için her zaman HTTPS kullanın:
1. SSL sertifikası edinin (Let's Encrypt vb.)
2. Nginx veya cloud sağlayıcıda yapılandırın
3. `ALLOWED_ORIGINS` değerini https:// kullanacak şekilde güncelleyin
### 3. Rate Limiting (Hız Sınırlama)
Sağlanan Nginx yapılandırması rate limiting içerir:
- API endpoint'leri: 10 istek/saniye
- MCP endpoint: 100 istek/saniye
### 4. CORS Yapılandırması
Production için belirli kaynaklara izin verin:
```bash
ALLOWED_ORIGINS=https://app.sizindomain.com,https://www.sizindomain.com
```
## İzleme
### Sağlık Kontrolleri
`/health` endpoint'ini izleyin:
```bash
curl http://localhost:8000/health
```
Yanıt:
```json
{
"status": "healthy",
"timestamp": "2024-12-26T10:00:00",
"uptime_seconds": 3600,
"tools_operational": true
}
```
### Loglama
Ortam değişkeni ile log seviyesini yapılandırın:
```bash
LOG_LEVEL=info # veya debug, warning, error
```
Loglar şuraya yazılır:
- Konsol (stdout)
- `logs/mcp_server.log` dosyası
### Metrikler (Opsiyonel)
OpenTelemetry desteği için:
```bash
pip install opentelemetry-instrumentation-fastapi
```
Ortam değişkenlerini ayarlayın:
```bash
OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4317
OTEL_SERVICE_NAME=yargi-mcp-server
```
## Sorun Giderme
### Port Zaten Kullanımda
```bash
# 8000 portunu kullanan işlemi bulun
lsof -i :8000
# İşlemi sonlandırın
kill -9 <PID>
```
### İzin Hataları
Dosya izinlerinin doğru olduğundan emin olun:
```bash
chmod +x run_asgi.py
chown -R www-data:www-data /opt/yargi-mcp
```
### Bellek Sorunları
Büyük belge işleme için worker belleğini artırın:
```bash
# systemd servisinde
Environment="PYTHONMALLOC=malloc"
LimitNOFILE=65536
```
### Zaman Aşımı Sorunları
Zaman aşımlarını ayarlayın:
1. Uvicorn: `--timeout-keep-alive 75`
2. Nginx: `proxy_read_timeout 300s;`
3. Cloud sağlayıcılar: Platform özel zaman aşımı ayarlarını kontrol edin
## Performans Ayarlama
### 1. Worker İşlemleri
- Geliştirme: 1 worker
- Production: CPU çekirdeği başına 2-4 worker
### 2. Bağlantı Havuzlama
Sunucu varsayılan olarak httpx ile bağlantı havuzlama kullanır.
### 3. Önbellekleme (Gelecek Geliştirme)
Redis önbellekleme docker-compose ile etkinleştirilebilir:
```bash
docker-compose --profile with-cache up
```
### 4. Veritabanı Zaman Aşımları
`.env` dosyasında veritabanı başına zaman aşımlarını ayarlayın:
```bash
YARGITAY_TIMEOUT=60
DANISTAY_TIMEOUT=60
ANAYASA_TIMEOUT=90
```
## Destek
Sorunlar ve sorular için:
- GitHub Issues: https://github.com/saidsurucu/yargi-mcp/issues
- Dokümantasyon: README.md dosyasına bakın
+2 -1
View File
@@ -1,5 +1,6 @@
# emsal_mcp_module/client.py
import asyncio
import httpx
# from bs4 import BeautifulSoup # Uncomment if needed for advanced HTML pre-processing
from typing import Dict, Any, List, Optional
@@ -153,7 +154,7 @@ class EmsalApiClient:
logger.warning(f"EmsalApiClient: Received empty or non-string HTML in 'data' field for Emsal ID {id}.")
return EmsalDocumentMarkdown(id=id, markdown_content=None, source_url=source_url)
markdown_content = self._clean_html_and_convert_to_markdown_emsal(html_content_from_api)
markdown_content = await asyncio.to_thread(self._clean_html_and_convert_to_markdown_emsal, html_content_from_api)
return EmsalDocumentMarkdown(
id=id,
+1 -1
View File
@@ -84,7 +84,7 @@ class EmsalApiResponseInnerData(BaseModel):
class EmsalApiResponse(BaseModel):
"""Model for the complete search response from the Emsal API."""
data: EmsalApiResponseInnerData
data: Optional[EmsalApiResponseInnerData] = None
metadata: Optional[Dict[str, Any]] = Field(None, description="Optional metadata (Meta Veri) from API, if any.")
class EmsalDocumentMarkdown(BaseModel):
-46
View File
@@ -1,46 +0,0 @@
# fly.toml app configuration file for yargi-mcp-noauth
#
# See https://fly.io/docs/reference/configuration/ for information about how to use this file.
#
app = 'yargi-mcp-free'
primary_region = 'fra'
[env]
ENABLE_AUTH = "false"
HOST = "0.0.0.0"
PORT = "8000"
LOG_LEVEL = "info"
[build]
[http_service]
internal_port = 8000
force_https = true
auto_stop_machines = 'off'
auto_start_machines = true
min_machines_running = 1
processes = ['app']
# Enable connection persistence for MCP sessions
[http_service.concurrency]
type = "connections"
hard_limit = 100
soft_limit = 80
[[vm]]
memory = '1gb'
cpu_kind = 'shared'
cpus = 1
[deploy]
strategy = "immediate"
[processes]
app = "python asgi_app.py"
[checks.http_health] # keep MCP /health live
type = "http"
interval = "30s"
timeout = "10s"
path = "/health"
-40
View File
@@ -1,40 +0,0 @@
# fly.toml app configuration file generated for yargi-mcp on 2025-06-29T00:23:47+03:00
#
# See https://fly.io/docs/reference/configuration/ for information about how to use this file.
#
app = 'yargi-mcp'
primary_region = 'fra'
[env]
ENABLE_AUTH = "true"
HOST = "0.0.0.0"
PORT = "8000"
LOG_LEVEL = "info"
[build]
[http_service]
internal_port = 8000
force_https = true
auto_stop_machines = 'off'
auto_start_machines = true
min_machines_running = 1
processes = ['app']
# Enable connection persistence for MCP sessions
[http_service.concurrency]
type = "connections"
hard_limit = 100
soft_limit = 80
[[vm]]
memory = '1gb'
cpu_kind = 'shared'
cpus = 1
[checks.http_health] # keep MCP /health live
type = "http"
interval = "30s"
timeout = "10s"
path = "/health"
+1
View File
@@ -0,0 +1 @@
# gib_mcp_module/__init__.py
+355
View File
@@ -0,0 +1,355 @@
# gib_mcp_module/client.py
import asyncio
import httpx
import io
import logging
import math
from typing import Optional, Any, Dict
from markitdown import MarkItDown
from .models import (
GibSearchRequest,
GibOzelgeSummary,
GibSearchResult,
GibDocumentMarkdown,
)
logger = logging.getLogger(__name__)
if not logger.hasHandlers():
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
)
class GibApiClient:
"""
API client for searching and retrieving GİB özelgeler (Turkish Revenue
Administration tax rulings) via the public gib.gov.tr JSON API.
The endpoint is a single POST list endpoint; document retrieval is done
by filtering the same endpoint with an exact `id`.
"""
BASE_URL = "https://gib.gov.tr/api"
LIST_PATH = "/gibportal/mevzuat/ozelge/list"
DOCUMENT_MARKDOWN_CHUNK_SIZE = 5000
# Fixed filter values required by the backend
_REQUIRED_STATUS = 2
_REQUIRED_DELETED = False
_REQUIRED_KTYPE = 99 # ktype=99 selects özelge
_SORT_FIELD = "ozelgeTarih"
_SORT_TYPE = "DESC"
def __init__(self, request_timeout: float = 60.0):
self.http_client = httpx.AsyncClient(
base_url=self.BASE_URL,
headers={
"Accept": "application/json",
"Accept-Language": "tr-TR,tr;q=0.9,en;q=0.7",
"Content-Type": "application/json",
"User-Agent": "Mozilla/5.0 (compatible; yargi-mcp/1.0; +https://github.com/saidsurucu/yargi-mcp)",
},
timeout=request_timeout,
verify=True,
follow_redirects=True,
)
@staticmethod
def _normalize_date(value: str, end_of_day: bool = False) -> Optional[str]:
"""
Accept 'YYYY-MM-DD' or full ISO 8601; always return full ISO 8601.
GİB backend rejects date-only strings.
"""
if not value:
return None
v = value.strip()
if not v:
return None
# Already ISO with time component
if "T" in v:
return v
# Simple YYYY-MM-DD - expand to start/end of day
suffix = "T23:59:59.999Z" if end_of_day else "T00:00:00.000Z"
return f"{v}{suffix}"
def _build_search_body(self, params: GibSearchRequest) -> Dict[str, Any]:
body: Dict[str, Any] = {
"status": self._REQUIRED_STATUS,
"deleted": self._REQUIRED_DELETED,
"ktype": self._REQUIRED_KTYPE,
}
keywords = params.keywords.strip()
kanun_no = params.kanunNo.strip()
# Frontend sets title/kanunNo/description to the SAME value; the backend
# ORs across them. If the caller supplies both, combine them so kanun_no
# still biases toward ruling text, while keywords remain primary.
search_term = keywords or kanun_no
if keywords and kanun_no and kanun_no not in keywords:
search_term = f"{keywords} {kanun_no}"
if search_term:
body["title"] = search_term
body["kanunNo"] = search_term
body["description"] = search_term
if params.ozelgeNo.strip():
body["ozelgeNo"] = params.ozelgeNo.strip()
if params.kanunId and params.kanunId > 0:
body["kanunIds"] = [params.kanunId]
start_iso = self._normalize_date(params.ozelgeStartDate, end_of_day=False)
end_iso = self._normalize_date(params.ozelgeEndDate, end_of_day=True)
if start_iso:
body["ozelgeStartDate"] = start_iso
if end_iso:
body["ozelgeEndDate"] = end_iso
return body
def _build_query_params(self, page_1_indexed: int, page_size: int) -> Dict[str, Any]:
# API expects 0-indexed page
zero_indexed = max(0, page_1_indexed - 1)
return {
"page": zero_indexed,
"size": page_size,
"sortFieldName": self._SORT_FIELD,
"sortType": self._SORT_TYPE,
}
@staticmethod
def _to_summary(item: Dict[str, Any]) -> Optional[GibOzelgeSummary]:
if not isinstance(item, dict):
return None
raw_id = item.get("id")
if raw_id is None:
return None
try:
ozelge_id = int(raw_id)
except (TypeError, ValueError):
return None
return GibOzelgeSummary(
id=ozelge_id,
ozelgeNo=item.get("ozelgeNo"),
ozelgeTarih=item.get("ozelgeTarih"),
title=item.get("title"),
kanunNo=item.get("kanunNo"),
kanunTitle=item.get("kanunTitle"),
siteLink=item.get("siteLink"),
)
async def search_ozelge(self, params: GibSearchRequest) -> GibSearchResult:
"""Search GİB özelgeler."""
body = self._build_search_body(params)
query = self._build_query_params(params.page, params.pageSize)
logger.info(
"GibApiClient: search page=%s size=%s body_keys=%s",
params.page, params.pageSize, sorted(body.keys()),
)
try:
resp = await self.http_client.post(self.LIST_PATH, params=query, json=body)
resp.raise_for_status()
payload = resp.json()
except httpx.HTTPStatusError as e:
logger.error("GibApiClient: HTTP %s during search", e.response.status_code)
return GibSearchResult(
ozelgeler=[],
total_results=0,
total_pages=0,
current_page=params.page,
page_size=params.pageSize,
)
except Exception as e:
logger.error("GibApiClient: search request failed: %s", e)
return GibSearchResult(
ozelgeler=[],
total_results=0,
total_pages=0,
current_page=params.page,
page_size=params.pageSize,
)
container = (payload or {}).get("resultContainer") or {}
raw_items = container.get("content") or []
summaries = []
for raw in raw_items:
summary = self._to_summary(raw)
if summary is not None:
summaries.append(summary)
total_results = container.get("totalElements") or 0
total_pages = container.get("totalPages") or 0
try:
total_results = int(total_results)
except (TypeError, ValueError):
total_results = 0
try:
total_pages = int(total_pages)
except (TypeError, ValueError):
total_pages = 0
return GibSearchResult(
ozelgeler=summaries,
total_results=total_results,
total_pages=total_pages,
current_page=params.page,
page_size=params.pageSize,
)
def _convert_html_to_markdown(self, html_content: str) -> Optional[str]:
"""Convert HTML content to Markdown using MarkItDown with BytesIO."""
if not html_content:
return None
try:
html_bytes = html_content.encode("utf-8")
html_stream = io.BytesIO(html_bytes)
md_converter = MarkItDown(enable_plugins=False)
result = md_converter.convert(html_stream)
return result.text_content
except Exception as e:
logger.error("GibApiClient: HTML→Markdown conversion failed: %s", e)
return None
@staticmethod
def _build_header_block(item: Dict[str, Any]) -> str:
"""Build a small Markdown header block summarising the ruling metadata."""
parts = []
title = item.get("title")
if title:
parts.append(f"# {title}")
meta_lines = []
if item.get("ozelgeNo"):
meta_lines.append(f"**Sayı:** {item['ozelgeNo']}")
if item.get("ozelgeTarih"):
meta_lines.append(f"**Tarih:** {item['ozelgeTarih']}")
if item.get("kanunTitle"):
kanun_no = item.get("kanunNo")
if kanun_no:
meta_lines.append(f"**Kanun:** {item['kanunTitle']} ({kanun_no})")
else:
meta_lines.append(f"**Kanun:** {item['kanunTitle']}")
if item.get("siteLink"):
meta_lines.append(f"**Kaynak:** {item['siteLink']}")
if meta_lines:
parts.append("\n".join(meta_lines))
return "\n\n".join(parts).strip()
async def get_ozelge_document(
self, ozelge_id: int, page_number: int = 1
) -> GibDocumentMarkdown:
"""Retrieve a single özelge and return its paginated Markdown form."""
logger.info(
"GibApiClient: fetching özelge id=%s page=%s", ozelge_id, page_number
)
if not isinstance(ozelge_id, int) or ozelge_id <= 0:
return GibDocumentMarkdown(
ozelge_id=ozelge_id if isinstance(ozelge_id, int) else 0,
current_page=page_number,
total_pages=0,
is_paginated=False,
error_message="ozelge_id must be a positive integer",
)
body = {
"status": self._REQUIRED_STATUS,
"deleted": self._REQUIRED_DELETED,
"ktype": self._REQUIRED_KTYPE,
"id": ozelge_id,
}
query = {"page": 0, "size": 1}
try:
resp = await self.http_client.post(self.LIST_PATH, params=query, json=body)
resp.raise_for_status()
payload = resp.json()
except httpx.HTTPStatusError as e:
msg = f"HTTP {e.response.status_code} when fetching özelge {ozelge_id}"
logger.error("GibApiClient: %s", msg)
return GibDocumentMarkdown(
ozelge_id=ozelge_id,
current_page=page_number,
total_pages=0,
is_paginated=False,
error_message=msg,
)
except Exception as e:
msg = f"Request failed: {e}"
logger.error("GibApiClient: %s", msg)
return GibDocumentMarkdown(
ozelge_id=ozelge_id,
current_page=page_number,
total_pages=0,
is_paginated=False,
error_message=msg,
)
container = (payload or {}).get("resultContainer") or {}
content = container.get("content") or []
if not content:
return GibDocumentMarkdown(
ozelge_id=ozelge_id,
current_page=page_number,
total_pages=0,
is_paginated=False,
error_message=f"Özelge {ozelge_id} not found",
)
item = content[0] if isinstance(content[0], dict) else {}
description_html = item.get("description") or ""
markdown_body = (await asyncio.to_thread(self._convert_html_to_markdown, description_html)) or ""
header_block = self._build_header_block(item)
if header_block and markdown_body:
full_markdown = f"{header_block}\n\n---\n\n{markdown_body}"
else:
full_markdown = header_block or markdown_body
if not full_markdown.strip():
return GibDocumentMarkdown(
ozelge_id=ozelge_id,
ozelge_no=item.get("ozelgeNo"),
title=item.get("title"),
ozelge_tarih=item.get("ozelgeTarih"),
kanun_title=item.get("kanunTitle"),
kanun_no=item.get("kanunNo"),
site_link=item.get("siteLink"),
current_page=page_number,
total_pages=0,
is_paginated=False,
error_message="Document body is empty",
)
total_pages = max(
1, math.ceil(len(full_markdown) / self.DOCUMENT_MARKDOWN_CHUNK_SIZE)
)
current_page_clamped = max(1, min(page_number, total_pages))
start = (current_page_clamped - 1) * self.DOCUMENT_MARKDOWN_CHUNK_SIZE
end = start + self.DOCUMENT_MARKDOWN_CHUNK_SIZE
chunk = full_markdown[start:end]
return GibDocumentMarkdown(
ozelge_id=ozelge_id,
ozelge_no=item.get("ozelgeNo"),
title=item.get("title"),
ozelge_tarih=item.get("ozelgeTarih"),
kanun_title=item.get("kanunTitle"),
kanun_no=item.get("kanunNo"),
site_link=item.get("siteLink"),
markdown_chunk=chunk,
current_page=current_page_clamped,
total_pages=total_pages,
is_paginated=total_pages > 1,
error_message=None,
)
async def close_client_session(self):
if hasattr(self, "http_client") and self.http_client and not self.http_client.is_closed:
await self.http_client.aclose()
logger.info("GibApiClient: HTTP client session closed.")
+64
View File
@@ -0,0 +1,64 @@
# gib_mcp_module/models.py
from pydantic import BaseModel, Field
from typing import List, Optional
class GibSearchRequest(BaseModel):
"""
Request model for searching GİB özelgeler (Turkish Revenue Administration tax rulings).
GİB (Gelir İdaresi Başkanlığı) publishes official tax-ruling letters
("özelge") responding to taxpayer questions on VAT, income tax,
corporate tax, stamp duty, and other tax matters. 18,000+ rulings
are searchable via the public gib.gov.tr API.
"""
keywords: str = Field("", description="Keywords searched across title, kanunNo and description (Turkish)")
ozelgeNo: str = Field("", description="Exact özelge reference number (e.g., 'E-40247694-130-15524')")
kanunNo: str = Field("", description="Law number filter, e.g. '3065' for KDV")
kanunId: int = Field(0, description="Optional numeric law ID filter (0=ignore)")
ozelgeStartDate: str = Field("", description="Start date YYYY-MM-DD or full ISO 8601")
ozelgeEndDate: str = Field("", description="End date YYYY-MM-DD or full ISO 8601")
page: int = Field(1, ge=1, description="Page number (1-indexed)")
pageSize: int = Field(10, ge=1, le=50, description="Results per page (1-50)")
class GibOzelgeSummary(BaseModel):
"""Summary of a single GİB özelge from search results (no full HTML)."""
id: int = Field(..., description="Numeric özelge ID for document retrieval")
ozelgeNo: Optional[str] = Field(None, description="Official ruling reference number")
ozelgeTarih: Optional[str] = Field(None, description="Ruling date (ISO datetime)")
title: Optional[str] = Field(None, description="Subject/title of the ruling")
kanunNo: Optional[str] = Field(None, description="Law number (e.g., '3065')")
kanunTitle: Optional[str] = Field(None, description="Law title (e.g., 'KATMA DEĞER VERGİSİ KANUNU')")
siteLink: Optional[str] = Field(None, description="Direct URL to the ruling on gib.gov.tr")
class GibSearchResult(BaseModel):
"""Response model for GİB özelge search results."""
ozelgeler: List[GibOzelgeSummary] = Field(default_factory=list, description="Matching özelge summaries")
total_results: int = Field(0, description="Total number of matching özelgeler across all pages")
total_pages: int = Field(0, description="Total number of pages for this query")
current_page: int = Field(1, description="Current page (1-indexed)")
page_size: int = Field(10, description="Results per page")
class GibDocumentMarkdown(BaseModel):
"""
GİB özelge document converted to paginated Markdown.
Long rulings are split into 5000-character chunks; request successive
pages via page_number to read the full text.
"""
ozelge_id: int = Field(..., description="Numeric özelge ID")
ozelge_no: Optional[str] = Field(None, description="Official ruling reference number")
title: Optional[str] = Field(None, description="Subject/title of the ruling")
ozelge_tarih: Optional[str] = Field(None, description="Ruling date (ISO datetime)")
kanun_title: Optional[str] = Field(None, description="Related law title")
kanun_no: Optional[str] = Field(None, description="Related law number")
site_link: Optional[str] = Field(None, description="Direct URL to the ruling on gib.gov.tr")
markdown_chunk: Optional[str] = Field(None, description="Current 5000-character Markdown chunk")
current_page: int = Field(1, description="Current page number (1-indexed)")
total_pages: int = Field(0, description="Total pages for the full Markdown content")
is_paginated: bool = Field(False, description="True if split across multiple pages")
error_message: Optional[str] = Field(None, description="Populated when retrieval failed")
+42 -10
View File
@@ -1,5 +1,7 @@
# kik_mcp_module/client_v2.py
import asyncio
import base64
import httpx
import logging
import uuid
@@ -49,6 +51,12 @@ class KikV2ApiClient:
174, 228, 219, 174, 208, 104, 174, 120, 32, 76, 250, 4, 143, 159, 211, 176
])
# AES-192-CBC key (environment.r8fact) used by the Angular HTTP interceptor to sign every
# request. The server decrypts X-Custom-Request-Ts and rejects stale timestamps with
# HTTP 401 "İstek zaman aşımına uğradı.", so these headers MUST be generated per-request
# with the current timestamp (see _generate_security_headers).
REQUEST_SIGNING_KEY = b"Qm2LtXR0aByP69vZNKef4wMJ" # UTF-8 bytes, 24 chars -> AES-192
@staticmethod
def encrypt_document_id(numeric_id: str) -> str:
"""
@@ -127,21 +135,43 @@ class KikV2ApiClient:
# Generate security headers (these might need to be updated based on API requirements)
self.security_headers = self._generate_security_headers()
def _sign_request_value(self, plaintext: str, iv: bytes) -> str:
"""AES-192-CBC encrypt a value with the request signing key, return base64 ciphertext."""
cipher = Cipher(
algorithms.AES(self.REQUEST_SIGNING_KEY),
modes.CBC(iv),
backend=default_backend()
)
encryptor = cipher.encryptor()
data = plaintext.encode("utf-8")
block_size = 16
padding_len = block_size - (len(data) % block_size)
padded = data + bytes([padding_len] * padding_len)
ciphertext = encryptor.update(padded) + encryptor.finalize()
return base64.b64encode(ciphertext).decode("ascii")
def _generate_security_headers(self) -> dict:
"""
Generate the custom security headers required by KIK v2 API.
These headers appear to be for request validation/encryption.
Generate the custom security headers required by the KIK v2 API.
Mirrors the Angular HTTP interceptor on ekapv2.kik.gov.tr: a random GUID and a
current-timestamp (epoch milliseconds) are AES-192-CBC encrypted with environment.r8fact
using a fresh random IV. The IV is sent as -Siv, the encrypted timestamp as -Ts, and the
encrypted GUID as -R8id. The server validates the decrypted timestamp's freshness, so these
MUST be regenerated on every request; stale values yield HTTP 401 "İstek zaman aşımına uğradı.".
"""
# Generate a random GUID for each session
if not HAS_CRYPTOGRAPHY:
raise ImportError("cryptography library required for KIK v2 request signing")
request_guid = str(uuid.uuid4())
# These are example values - in a real implementation, these might need
# to be calculated based on the request content or session
iv = os.urandom(16)
timestamp_ms = str(int(datetime.now().timestamp() * 1000))
return {
"X-Custom-Request-Guid": request_guid,
"X-Custom-Request-R8id": "hwnOjsN8qdgtDw70x3sKkxab0rj2bQ8Uph4+C+oU+9AMmQqRN3eMOEEeet748DOf",
"X-Custom-Request-Siv": "p2IQRTitF8z7I39nBjdAqA==",
"X-Custom-Request-Ts": "1vB3Wwrt8YQ5U6t3XAzZ+Q=="
"X-Custom-Request-R8id": self._sign_request_value(request_guid, iv),
"X-Custom-Request-Siv": base64.b64encode(iv).decode("ascii"),
"X-Custom-Request-Ts": self._sign_request_value(timestamp_ms, iv),
}
def _build_search_payload(self,
@@ -439,7 +469,9 @@ class KikV2ApiClient:
html_bytes = html_content.encode('utf-8')
html_stream = BytesIO(html_bytes)
result = md.convert_stream(html_stream, file_extension=".html")
# markitdown is sync; offload to thread so HTML parsing doesn't
# block the event-loop / other in-flight MCP requests.
result = await asyncio.to_thread(md.convert_stream, html_stream, file_extension=".html")
markdown_content = result.text_content
return KikV2DocumentMarkdown(
+2 -1
View File
@@ -1,5 +1,6 @@
# kvkk_mcp_module/client.py
import asyncio
import httpx
from bs4 import BeautifulSoup
from typing import List, Optional, Dict, Any
@@ -291,7 +292,7 @@ class KvkkApiClient:
# Convert HTML content to Markdown
full_markdown_content = None
if extracted_data["html_content"]:
full_markdown_content = self._convert_html_to_markdown(extracted_data["html_content"])
full_markdown_content = await asyncio.to_thread(self._convert_html_to_markdown, extracted_data["html_content"])
if not full_markdown_content:
return KvkkDocumentMarkdown(
-28
View File
@@ -1,28 +0,0 @@
"""
MCP Auth Toolkit - OAuth 2.1 + Authorization for Model Context Protocol Servers
Integrated with Clerk Authentication
"""
from .middleware import (
AuthContext,
FastMCPAuthWrapper,
MCPAuthMiddleware,
auth_required,
)
from .oauth import OAuthConfig, OAuthProvider
from .policy import PolicyEngine, ToolPolicy, create_default_policies
from .storage import PersistentStorage
__version__ = "0.1.0"
__all__ = [
"OAuthProvider",
"OAuthConfig",
"AuthContext",
"auth_required",
"create_default_policies",
"MCPAuthMiddleware",
"FastMCPAuthWrapper",
"PolicyEngine",
"ToolPolicy",
"PersistentStorage",
]
-73
View File
@@ -1,73 +0,0 @@
"""
Clerk OAuth configuration for MCP Auth Toolkit
"""
import os
import logging
from .oauth import OAuthConfig
logger = logging.getLogger(__name__)
def create_clerk_oauth_config() -> OAuthConfig:
"""Create OAuth configuration for Clerk integration using SDK"""
# Get Clerk configuration from environment
clerk_domain = os.getenv("CLERK_DOMAIN", "accounts.yargimcp.com")
clerk_publishable_key = os.getenv("CLERK_PUBLISHABLE_KEY")
clerk_secret_key = os.getenv("CLERK_SECRET_KEY")
if not clerk_publishable_key or not clerk_secret_key:
raise ValueError("CLERK_PUBLISHABLE_KEY and CLERK_SECRET_KEY are required")
# For Clerk with custom domains, we use our adapter endpoints
# This allows us to handle the custom domain flow properly
base_url = os.getenv("BASE_URL", "https://yargimcp.com")
config = OAuthConfig(
client_id=clerk_publishable_key,
client_secret=clerk_secret_key,
# Use our adapter endpoints instead of Clerk's direct endpoints
authorization_endpoint=f"{base_url}/authorize",
token_endpoint=f"{base_url}/token",
# Keep Clerk's JWKS for token validation
jwks_uri=f"https://{clerk_domain}/.well-known/jwks.json",
issuer=base_url, # We're the issuer for MCP tokens
scopes=["mcp:tools:read", "mcp:tools:write", "openid", "profile", "email"]
)
logger.info(f"Created Clerk OAuth config with adapter endpoints")
logger.info(f"Clerk domain: {clerk_domain}")
logger.debug(f"Authorization endpoint: {config.authorization_endpoint}")
logger.debug(f"Token endpoint: {config.token_endpoint}")
return config
def get_jwt_secret() -> str:
"""Get JWT secret for token signing"""
jwt_secret = os.getenv("JWT_SECRET_KEY")
if not jwt_secret:
raise ValueError("JWT_SECRET_KEY environment variable is required")
return jwt_secret
def create_mcp_server_config():
"""Create complete MCP server configuration for Clerk integration"""
try:
oauth_config = create_clerk_oauth_config()
jwt_secret = get_jwt_secret()
return {
"oauth_config": oauth_config,
"jwt_secret": jwt_secret,
"base_url": os.getenv("BASE_URL", "https://yargi-mcp.fly.dev"),
"auth_enabled": os.getenv("ENABLE_AUTH", "true").lower() == "true"
}
except Exception as e:
logger.error(f"Failed to create MCP server config: {e}")
raise
-315
View File
@@ -1,315 +0,0 @@
"""
MCP server middleware for OAuth authentication and authorization
"""
import functools
import logging
from collections.abc import Callable
from dataclasses import dataclass
from typing import Any, Optional
logger = logging.getLogger(__name__)
try:
from fastmcp import FastMCP
FASTMCP_AVAILABLE = True
except ImportError:
FASTMCP_AVAILABLE = False
FastMCP = None
logger.warning("FastMCP not available, some features will be disabled")
from .oauth import OAuthProvider
from .policy import PolicyEngine
@dataclass
class AuthContext:
"""Authentication context passed to MCP tools"""
user_id: str
scopes: list[str]
claims: dict[str, Any]
token: str
class MCPAuthMiddleware:
"""Authentication middleware for MCP servers"""
def __init__(self, oauth_provider: OAuthProvider, policy_engine: PolicyEngine):
self.oauth_provider = oauth_provider
self.policy_engine = policy_engine
def authenticate_request(self, authorization_header: str) -> AuthContext | None:
"""Extract and validate auth token from request"""
if not authorization_header:
logger.debug("No authorization header provided")
return None
if not authorization_header.startswith("Bearer "):
logger.debug("Authorization header does not start with 'Bearer '")
return None
token = authorization_header[7:] # Remove 'Bearer ' prefix
token_info = self.oauth_provider.introspect_token(token)
if not token_info.get("active"):
logger.warning("Token is not active")
return None
logger.debug(f"Authenticated user: {token_info.get('sub', 'unknown')}")
return AuthContext(
user_id=token_info.get("sub", "unknown"),
scopes=token_info.get("mcp_tool_scopes", []),
claims=token_info,
token=token,
)
def authorize_tool_call(
self, tool_name: str, auth_context: AuthContext
) -> tuple[bool, str | None]:
"""Check if user can call the specified tool"""
return self.policy_engine.authorize_tool_call(
tool_name=tool_name,
user_scopes=auth_context.scopes,
user_claims=auth_context.claims,
)
def auth_required(
oauth_provider: OAuthProvider,
policy_engine: PolicyEngine,
tool_name: str | None = None,
):
"""
Decorator to require authentication for MCP tool functions
Usage:
@auth_required(oauth_provider, policy_engine, "search_yargitay")
def my_tool_function(context: AuthContext, ...):
pass
"""
def decorator(func: Callable) -> Callable:
middleware = MCPAuthMiddleware(oauth_provider, policy_engine)
@functools.wraps(func)
async def wrapper(*args, **kwargs):
# Extract authorization header from kwargs
auth_header = kwargs.pop("authorization", None)
# Also check in args if it's a Request object
if not auth_header and args:
for arg in args:
if hasattr(arg, 'headers'):
auth_header = arg.headers.get("Authorization")
break
if not auth_header:
logger.warning(f"No authorization header for tool '{tool_name or func.__name__}'")
raise PermissionError("Authorization header required")
auth_context = middleware.authenticate_request(auth_header)
if not auth_context:
logger.warning(f"Authentication failed for tool '{tool_name or func.__name__}'")
raise PermissionError("Invalid or expired token")
actual_tool_name = tool_name or func.__name__
authorized, reason = middleware.authorize_tool_call(
actual_tool_name, auth_context
)
if not authorized:
logger.warning(f"Authorization failed for tool '{actual_tool_name}': {reason}")
raise PermissionError(f"Access denied: {reason}")
# Add auth context to function call
return await func(auth_context, *args, **kwargs)
return wrapper
return decorator
class FastMCPAuthWrapper:
"""Wrapper for FastMCP servers to add authentication"""
def __init__(
self,
mcp_server: "FastMCP",
oauth_provider: OAuthProvider,
policy_engine: PolicyEngine,
):
if not FASTMCP_AVAILABLE:
raise ImportError("FastMCP is required for FastMCPAuthWrapper")
self.mcp_server = mcp_server
self.middleware = MCPAuthMiddleware(oauth_provider, policy_engine)
self.oauth_provider = oauth_provider
logger.info("Initializing FastMCP authentication wrapper")
self._wrap_tools()
def _wrap_tools(self):
"""Wrap all existing tools with auth middleware"""
# Try different FastMCP tool storage locations
tool_registry = None
if hasattr(self.mcp_server, '_tools'):
tool_registry = self.mcp_server._tools
elif hasattr(self.mcp_server, 'tools'):
tool_registry = self.mcp_server.tools
elif hasattr(self.mcp_server, '_tool_registry'):
tool_registry = self.mcp_server._tool_registry
elif hasattr(self.mcp_server, '_handlers') and hasattr(self.mcp_server._handlers, 'tools'):
tool_registry = self.mcp_server._handlers.tools
if not tool_registry:
logger.warning("FastMCP server tool registry not found, tools will not be automatically wrapped")
logger.debug(f"Available server attributes: {dir(self.mcp_server)}")
return
logger.debug(f"Found tool registry with {len(tool_registry)} tools")
original_tools = dict(tool_registry)
wrapped_count = 0
for tool_name, tool_func in original_tools.items():
try:
wrapped_func = self._create_auth_wrapper(tool_name, tool_func)
tool_registry[tool_name] = wrapped_func
wrapped_count += 1
logger.debug(f"Wrapped tool: {tool_name}")
except Exception as e:
logger.error(f"Failed to wrap tool {tool_name}: {e}")
logger.info(f"Successfully wrapped {wrapped_count} tools with authentication")
def _create_auth_wrapper(self, tool_name: str, original_func: Callable) -> Callable:
"""Create auth wrapper for a specific tool"""
@functools.wraps(original_func)
async def auth_wrapper(*args, **kwargs):
# Extract authorization from various sources
auth_header = None
# Check kwargs first
auth_header = kwargs.pop("authorization", None)
# Check if first argument is a Request object
if not auth_header and args:
first_arg = args[0]
if hasattr(first_arg, 'headers'):
auth_header = first_arg.headers.get("Authorization")
if not auth_header:
logger.warning(f"No authorization header for tool '{tool_name}'")
raise PermissionError("Authorization required")
auth_context = self.middleware.authenticate_request(auth_header)
if not auth_context:
logger.warning(f"Authentication failed for tool '{tool_name}'")
raise PermissionError("Invalid token")
authorized, reason = self.middleware.authorize_tool_call(
tool_name, auth_context
)
if not authorized:
logger.warning(f"Authorization failed for tool '{tool_name}': {reason}")
raise PermissionError(f"Access denied: {reason}")
# Add auth context to kwargs
kwargs["auth_context"] = auth_context
logger.debug(f"Calling tool '{tool_name}' for user {auth_context.user_id}")
return await original_func(*args, **kwargs)
return auth_wrapper
def add_oauth_endpoints(self):
"""Add OAuth endpoints to the MCP server"""
@self.mcp_server.tool(
description="Initiate OAuth 2.1 authorization flow with PKCE",
annotations={"readOnlyHint": True, "idempotentHint": False}
)
async def oauth_authorize(redirect_uri: str, scopes: Optional[str] = None):
"""OAuth authorization endpoint"""
scope_list = scopes.split(" ") if scopes else None
auth_url, pkce = self.oauth_provider.generate_authorization_url(
redirect_uri=redirect_uri, scopes=scope_list
)
logger.info(f"Generated authorization URL for redirect_uri: {redirect_uri}")
return {
"authorization_url": auth_url,
"code_verifier": pkce.verifier, # For PKCE flow
"code_challenge": pkce.challenge,
"instructions": "Use the authorization_url to complete OAuth flow, then exchange the returned code using oauth_token tool"
}
@self.mcp_server.tool(
description="Exchange OAuth authorization code for access token",
annotations={"readOnlyHint": False, "idempotentHint": False}
)
async def oauth_token(
code: str,
state: str,
redirect_uri: str
):
"""OAuth token exchange endpoint"""
try:
result = await self.oauth_provider.exchange_code_for_token(
code=code, state=state, redirect_uri=redirect_uri
)
logger.info("Successfully exchanged authorization code for token")
return result
except Exception as e:
logger.error(f"Token exchange failed: {e}")
raise
@self.mcp_server.tool(
description="Validate and introspect OAuth access token",
annotations={"readOnlyHint": True, "idempotentHint": True}
)
async def oauth_introspect(token: str):
"""Token introspection endpoint"""
result = self.oauth_provider.introspect_token(token)
logger.debug(f"Token introspection: active={result.get('active', False)}")
return result
@self.mcp_server.tool(
description="Revoke OAuth access token",
annotations={"readOnlyHint": False, "idempotentHint": False}
)
async def oauth_revoke(token: str):
"""Token revocation endpoint"""
success = self.oauth_provider.revoke_token(token)
logger.info(f"Token revocation: success={success}")
return {"revoked": success}
@self.mcp_server.tool(
description="Get list of tools available to authenticated user",
annotations={"readOnlyHint": True, "idempotentHint": True}
)
async def oauth_user_tools(authorization: str):
"""Get user's allowed tools based on scopes"""
auth_context = self.middleware.authenticate_request(authorization)
if not auth_context:
raise PermissionError("Invalid token")
allowed_patterns = self.middleware.policy_engine.get_allowed_tools(auth_context.scopes)
return {
"user_id": auth_context.user_id,
"scopes": auth_context.scopes,
"allowed_tool_patterns": allowed_patterns,
"message": "Use these patterns to determine which tools you can access"
}
logger.info("Added OAuth endpoints: oauth_authorize, oauth_token, oauth_introspect, oauth_revoke, oauth_user_tools")
-304
View File
@@ -1,304 +0,0 @@
"""
OAuth 2.1 + PKCE implementation for MCP servers with Clerk integration
"""
import base64
import hashlib
import secrets
import time
import logging
from dataclasses import dataclass
from datetime import datetime, timedelta
from typing import Any, Optional
from urllib.parse import urlencode
import httpx
import jwt
from jwt.exceptions import PyJWTError, InvalidTokenError
from .storage import PersistentStorage
# Try to import Clerk SDK
try:
from clerk_backend_api import Clerk
CLERK_AVAILABLE = True
except ImportError:
CLERK_AVAILABLE = False
Clerk = None
logger = logging.getLogger(__name__)
@dataclass
class OAuthConfig:
"""OAuth provider configuration for Clerk"""
client_id: str
client_secret: str
authorization_endpoint: str
token_endpoint: str
jwks_uri: str | None = None
issuer: str = "mcp-auth"
scopes: list[str] = None
def __post_init__(self):
if self.scopes is None:
self.scopes = ["mcp:tools:read", "mcp:tools:write"]
class PKCEChallenge:
"""PKCE challenge/verifier pair for OAuth 2.1"""
def __init__(self):
self.verifier = (
base64.urlsafe_b64encode(secrets.token_bytes(32))
.decode("utf-8")
.rstrip("=")
)
challenge_bytes = hashlib.sha256(self.verifier.encode("utf-8")).digest()
self.challenge = (
base64.urlsafe_b64encode(challenge_bytes).decode("utf-8").rstrip("=")
)
class OAuthProvider:
"""OAuth 2.1 provider with PKCE support and Clerk integration"""
def __init__(self, config: OAuthConfig, jwt_secret: str):
self.config = config
self.jwt_secret = jwt_secret
# Use persistent storage instead of memory
self.storage = PersistentStorage()
# Initialize Clerk SDK if available
self.clerk = None
if CLERK_AVAILABLE and config.client_secret:
try:
self.clerk = Clerk(bearer_auth=config.client_secret)
logger.info("Clerk SDK initialized successfully")
except Exception as e:
logger.warning(f"Failed to initialize Clerk SDK: {e}")
logger.info("OAuth provider initialized with persistent storage")
def generate_authorization_url(
self,
redirect_uri: str,
state: str | None = None,
scopes: list[str] | None = None,
) -> tuple[str, PKCEChallenge]:
"""Generate OAuth authorization URL with PKCE for Clerk"""
pkce = PKCEChallenge()
session_id = secrets.token_urlsafe(32)
if state is None:
state = secrets.token_urlsafe(16)
if scopes is None:
scopes = self.config.scopes
# Store session data with expiration
session_data = {
"pkce_verifier": pkce.verifier,
"state": state,
"redirect_uri": redirect_uri,
"scopes": scopes,
"created_at": time.time(),
"expires_at": (datetime.utcnow() + timedelta(minutes=10)).timestamp(),
}
self.storage.set_session(session_id, session_data)
# Build Clerk OAuth URL
# Check if this is a custom domain (sign-in endpoint)
if self.config.authorization_endpoint.endswith('/sign-in'):
# For custom domains, Clerk expects redirect_url parameter
params = {
"redirect_url": redirect_uri,
"state": f"{state}:{session_id}",
}
auth_url = f"{self.config.authorization_endpoint}?{urlencode(params)}"
else:
# Standard OAuth flow with PKCE
params = {
"response_type": "code",
"client_id": self.config.client_id,
"redirect_uri": redirect_uri,
"scope": " ".join(scopes),
"state": f"{state}:{session_id}", # Combine state with session ID
"code_challenge": pkce.challenge,
"code_challenge_method": "S256",
}
auth_url = f"{self.config.authorization_endpoint}?{urlencode(params)}"
logger.info(f"Generated OAuth URL with session {session_id[:8]}...")
logger.debug(f"Auth URL: {auth_url}")
return auth_url, pkce
async def exchange_code_for_token(
self, code: str, state: str, redirect_uri: str
) -> dict[str, Any]:
"""Exchange authorization code for access token with Clerk"""
try:
original_state, session_id = state.split(":", 1)
except ValueError as e:
logger.error(f"Invalid state format: {state}")
raise ValueError("Invalid state format") from e
session = self.storage.get_session(session_id)
if not session:
logger.error(f"Session {session_id} not found")
raise ValueError("Invalid session")
# Check session expiration
if datetime.utcnow().timestamp() > session.get("expires_at", 0):
self.storage.delete_session(session_id)
logger.error(f"Session {session_id} expired")
raise ValueError("Session expired")
if session["state"] != original_state:
logger.error(f"State mismatch: expected {session['state']}, got {original_state}")
raise ValueError("State mismatch")
if session["redirect_uri"] != redirect_uri:
logger.error(f"Redirect URI mismatch: expected {session['redirect_uri']}, got {redirect_uri}")
raise ValueError("Redirect URI mismatch")
# Prepare token exchange request for Clerk
token_data = {
"grant_type": "authorization_code",
"client_id": self.config.client_id,
"client_secret": self.config.client_secret,
"code": code,
"redirect_uri": redirect_uri,
"code_verifier": session["pkce_verifier"],
}
logger.info(f"Exchanging code with Clerk for session {session_id[:8]}...")
async with httpx.AsyncClient() as client:
response = await client.post(
self.config.token_endpoint,
data=token_data,
headers={"Content-Type": "application/x-www-form-urlencoded"},
timeout=30.0,
)
if response.status_code != 200:
logger.error(f"Clerk token exchange failed: {response.status_code} - {response.text}")
raise ValueError(f"Token exchange failed: {response.text}")
token_response = response.json()
logger.info("Successfully exchanged code for Clerk token")
# Create MCP-scoped JWT token
access_token = self._create_mcp_token(
session["scopes"], token_response.get("access_token"), session_id
)
# Store token for introspection
token_id = secrets.token_urlsafe(16)
token_data = {
"access_token": access_token,
"scopes": session["scopes"],
"created_at": time.time(),
"expires_at": (datetime.utcnow() + timedelta(hours=1)).timestamp(),
"session_id": session_id,
"clerk_token": token_response.get("access_token"),
}
self.storage.set_token(token_id, token_data)
# Clean up session
self.storage.delete_session(session_id)
return {
"access_token": access_token,
"token_type": "bearer",
"expires_in": 3600,
"scope": " ".join(session["scopes"]),
}
def validate_pkce(self, code_verifier: str, code_challenge: str) -> bool:
"""Validate PKCE code challenge (RFC 7636)"""
# S256 method
verifier_hash = hashlib.sha256(code_verifier.encode()).digest()
expected_challenge = base64.urlsafe_b64encode(verifier_hash).decode().rstrip('=')
return expected_challenge == code_challenge
def _create_mcp_token(
self, scopes: list[str], upstream_token: str, session_id: str
) -> str:
"""Create MCP-scoped JWT token with Clerk token embedded"""
now = int(time.time())
payload = {
"iss": self.config.issuer,
"sub": session_id,
"aud": "mcp-server",
"iat": now,
"exp": now + 3600, # 1 hour expiration
"mcp_tool_scopes": scopes,
"upstream_token": upstream_token,
"clerk_integration": True,
}
return jwt.encode(payload, self.jwt_secret, algorithm="HS256")
def introspect_token(self, token: str) -> dict[str, Any]:
"""Introspect and validate MCP token"""
try:
payload = jwt.decode(token, self.jwt_secret, algorithms=["HS256"])
# Check if token is expired
if payload.get("exp", 0) < time.time():
return {"active": False, "error": "token_expired"}
return {
"active": True,
"sub": payload.get("sub"),
"aud": payload.get("aud"),
"iss": payload.get("iss"),
"exp": payload.get("exp"),
"iat": payload.get("iat"),
"mcp_tool_scopes": payload.get("mcp_tool_scopes", []),
"upstream_token": payload.get("upstream_token"),
"clerk_integration": payload.get("clerk_integration", False),
}
except PyJWTError as e:
logger.warning(f"Token validation failed: {e}")
return {"active": False, "error": "invalid_token"}
def revoke_token(self, token: str) -> bool:
"""Revoke a token"""
try:
payload = jwt.decode(token, self.jwt_secret, algorithms=["HS256"])
session_id = payload.get("sub")
# Remove all tokens associated with this session
all_tokens = self.storage.get_tokens()
tokens_to_remove = [
token_id
for token_id, token_data in all_tokens.items()
if token_data.get("session_id") == session_id
]
for token_id in tokens_to_remove:
self.storage.delete_token(token_id)
logger.info(f"Revoked {len(tokens_to_remove)} tokens for session {session_id}")
return True
except InvalidTokenError as e:
logger.warning(f"Token revocation failed: {e}")
return False
def cleanup_expired_sessions(self):
"""Clean up expired sessions and tokens"""
# This is now handled automatically by persistent storage
self.storage.cleanup_expired_sessions()
logger.debug("Cleanup completed via persistent storage")
-201
View File
@@ -1,201 +0,0 @@
"""
Authorization policy engine for MCP tools
"""
import re
import logging
from dataclasses import dataclass
from enum import Enum
from typing import Any
logger = logging.getLogger(__name__)
class PolicyAction(Enum):
ALLOW = "allow"
DENY = "deny"
@dataclass
class ToolPolicy:
"""Policy rule for MCP tool access"""
tool_pattern: str # regex pattern for tool names
required_scopes: list[str]
action: PolicyAction = PolicyAction.ALLOW
conditions: dict[str, Any] | None = None
def matches_tool(self, tool_name: str) -> bool:
"""Check if the policy applies to given tool"""
return bool(re.match(self.tool_pattern, tool_name))
def evaluate_scopes(self, user_scopes: list[str]) -> bool:
"""Check if user has required scopes"""
return all(scope in user_scopes for scope in self.required_scopes)
class PolicyEngine:
"""Authorization policy engine for Turkish legal database tools"""
def __init__(self):
self.policies: list[ToolPolicy] = []
self.default_action = PolicyAction.DENY
def add_policy(self, policy: ToolPolicy):
"""Add a policy rule"""
self.policies.append(policy)
logger.debug(f"Added policy: {policy.tool_pattern} -> {policy.required_scopes}")
def add_tool_scope_policy(
self,
tool_pattern: str,
required_scopes: str | list[str],
action: PolicyAction = PolicyAction.ALLOW,
):
"""Convenience method to add tool-scope policy"""
if isinstance(required_scopes, str):
required_scopes = [required_scopes]
policy = ToolPolicy(
tool_pattern=tool_pattern, required_scopes=required_scopes, action=action
)
self.add_policy(policy)
def authorize_tool_call(
self,
tool_name: str,
user_scopes: list[str],
user_claims: dict[str, Any] | None = None,
) -> tuple[bool, str | None]:
"""
Authorize a tool call
Returns:
(authorized: bool, reason: Optional[str])
"""
logger.debug(f"Authorizing tool '{tool_name}' for user with scopes: {user_scopes}")
matching_policies = [
policy for policy in self.policies if policy.matches_tool(tool_name)
]
if not matching_policies:
if self.default_action == PolicyAction.ALLOW:
logger.debug(f"No policies found for '{tool_name}', allowing by default")
return True, None
else:
logger.warning(f"No policies found for '{tool_name}', denying by default")
return False, f"No policy found for tool '{tool_name}', default deny"
# Check for explicit deny policies first
for policy in matching_policies:
if policy.action == PolicyAction.DENY:
if policy.evaluate_scopes(user_scopes):
logger.warning(f"Explicit deny policy matched for '{tool_name}'")
return False, f"Explicit deny policy for tool '{tool_name}'"
# Check allow policies
allow_policies = [
p for p in matching_policies if p.action == PolicyAction.ALLOW
]
if not allow_policies:
logger.warning(f"No allow policies found for '{tool_name}'")
return False, f"No allow policies found for tool '{tool_name}'"
for policy in allow_policies:
if policy.evaluate_scopes(user_scopes):
if self._evaluate_conditions(policy.conditions, user_claims):
logger.debug(f"Authorization granted for '{tool_name}'")
return True, None
logger.warning(f"Insufficient scopes for '{tool_name}'. Required: {[p.required_scopes for p in allow_policies]}, User has: {user_scopes}")
return False, f"Insufficient scopes for tool '{tool_name}'"
def _evaluate_conditions(
self,
conditions: dict[str, Any] | None,
user_claims: dict[str, Any] | None,
) -> bool:
"""Evaluate additional policy conditions"""
if not conditions:
return True
if not user_claims:
logger.debug("No user claims provided, conditions evaluation failed")
return False
for key, expected_value in conditions.items():
user_value = user_claims.get(key)
if isinstance(expected_value, list):
if user_value not in expected_value:
logger.debug(f"Condition failed: {key} = {user_value} not in {expected_value}")
return False
elif user_value != expected_value:
logger.debug(f"Condition failed: {key} = {user_value} != {expected_value}")
return False
return True
def get_allowed_tools(self, user_scopes: list[str]) -> list[str]:
"""Get list of tool patterns user is allowed to call"""
allowed_tools = []
for policy in self.policies:
if policy.action == PolicyAction.ALLOW and policy.evaluate_scopes(
user_scopes
):
allowed_tools.append(policy.tool_pattern)
return allowed_tools
def create_turkish_legal_policies() -> PolicyEngine:
"""Create policy set for Turkish legal database MCP server"""
engine = PolicyEngine()
# Administrative tools (full access)
engine.add_tool_scope_policy(".*", ["mcp:tools:admin"])
# Search tools - require read access
engine.add_tool_scope_policy("search.*", ["mcp:tools:read"])
# Fetch/get document tools - require read access
engine.add_tool_scope_policy("get_.*", ["mcp:tools:read"])
engine.add_tool_scope_policy("fetch.*", ["mcp:tools:read"])
# Specific Turkish legal database tools
engine.add_tool_scope_policy("search_yargitay.*", ["mcp:tools:read"])
engine.add_tool_scope_policy("search_danistay.*", ["mcp:tools:read"])
engine.add_tool_scope_policy("search_anayasa.*", ["mcp:tools:read"])
engine.add_tool_scope_policy("search_rekabet.*", ["mcp:tools:read"])
engine.add_tool_scope_policy("search_kik.*", ["mcp:tools:read"])
engine.add_tool_scope_policy("search_emsal.*", ["mcp:tools:read"])
engine.add_tool_scope_policy("search_uyusmazlik.*", ["mcp:tools:read"])
engine.add_tool_scope_policy("search_sayistay.*", ["mcp:tools:read"])
engine.add_tool_scope_policy("search_.*_bedesten", ["mcp:tools:read"])
engine.add_tool_scope_policy("search_yerel_hukuk.*", ["mcp:tools:read"])
engine.add_tool_scope_policy("search_istinaf_hukuk.*", ["mcp:tools:read"])
engine.add_tool_scope_policy("search_kyb.*", ["mcp:tools:read"])
# Document retrieval tools
engine.add_tool_scope_policy("get_.*_document.*", ["mcp:tools:read"])
engine.add_tool_scope_policy("get_.*_markdown", ["mcp:tools:read"])
# Write operations (if any future tools need them)
engine.add_tool_scope_policy("create_.*", ["mcp:tools:write"])
engine.add_tool_scope_policy("update_.*", ["mcp:tools:write"])
engine.add_tool_scope_policy("delete_.*", ["mcp:tools:write"])
logger.info("Created Turkish legal database policy engine")
return engine
def create_default_policies() -> PolicyEngine:
"""Create a default policy set for MCP servers (backwards compatibility)"""
return create_turkish_legal_policies()
-112
View File
@@ -1,112 +0,0 @@
"""
Persistent storage for OAuth sessions and tokens
"""
import json
import os
import tempfile
import logging
from datetime import datetime
from typing import Dict, Any, Optional
logger = logging.getLogger(__name__)
class PersistentStorage:
"""File-based persistent storage for OAuth data"""
def __init__(self, storage_dir: str = None):
if storage_dir is None:
# Use system temp directory or environment variable
storage_dir = os.environ.get('TEMP', tempfile.gettempdir())
self.storage_dir = os.path.join(storage_dir, 'mcp_oauth_storage')
os.makedirs(self.storage_dir, exist_ok=True)
self.sessions_file = os.path.join(self.storage_dir, 'oauth_sessions.json')
self.tokens_file = os.path.join(self.storage_dir, 'oauth_tokens.json')
logger.info(f"Persistent OAuth storage initialized at: {self.storage_dir}")
def _load_json(self, filepath: str) -> Dict:
"""Load JSON data from file"""
try:
if os.path.exists(filepath):
with open(filepath, 'r', encoding='utf-8') as f:
return json.load(f)
except Exception as e:
logger.error(f"Error loading {filepath}: {e}")
return {}
def _save_json(self, filepath: str, data: Dict):
"""Save JSON data to file"""
try:
with open(filepath, 'w', encoding='utf-8') as f:
json.dump(data, f, indent=2, default=str)
except Exception as e:
logger.error(f"Error saving {filepath}: {e}")
def get_sessions(self) -> Dict[str, Dict[str, Any]]:
"""Get all OAuth sessions"""
data = self._load_json(self.sessions_file)
# Clean expired sessions
now = datetime.utcnow().timestamp()
valid_sessions = {k: v for k, v in data.items()
if v.get('expires_at', 0) > now}
if len(valid_sessions) != len(data):
self._save_json(self.sessions_file, valid_sessions)
return valid_sessions
def set_session(self, session_id: str, data: Dict[str, Any]):
"""Set OAuth session data"""
sessions = self.get_sessions()
sessions[session_id] = data
self._save_json(self.sessions_file, sessions)
def get_session(self, session_id: str) -> Optional[Dict[str, Any]]:
"""Get specific OAuth session data"""
sessions = self.get_sessions()
return sessions.get(session_id)
def delete_session(self, session_id: str):
"""Delete OAuth session"""
sessions = self.get_sessions()
if session_id in sessions:
del sessions[session_id]
self._save_json(self.sessions_file, sessions)
def get_tokens(self) -> Dict[str, Dict[str, Any]]:
"""Get all OAuth tokens"""
data = self._load_json(self.tokens_file)
# Clean expired tokens
now = datetime.utcnow().timestamp()
valid_tokens = {k: v for k, v in data.items()
if v.get('expires_at', 0) > now}
if len(valid_tokens) != len(data):
self._save_json(self.tokens_file, valid_tokens)
return valid_tokens
def set_token(self, token_id: str, token_data: Dict[str, Any]):
"""Set OAuth token data"""
tokens = self.get_tokens()
tokens[token_id] = token_data
self._save_json(self.tokens_file, tokens)
def get_token(self, token_id: str) -> Optional[Dict[str, Any]]:
"""Get specific OAuth token data"""
tokens = self.get_tokens()
return tokens.get(token_id)
def delete_token(self, token_id: str):
"""Delete OAuth token"""
tokens = self.get_tokens()
if token_id in tokens:
del tokens[token_id]
self._save_json(self.tokens_file, tokens)
def cleanup_expired_sessions(self):
"""Clean up expired sessions and tokens"""
# This is handled automatically in get_sessions() and get_tokens()
sessions = self.get_sessions()
tokens = self.get_tokens()
logger.debug(f"Cleanup: {len(sessions)} active sessions, {len(tokens)} active tokens")
-193
View File
@@ -1,193 +0,0 @@
"""
Factory for creating FastMCP app with MCP Auth Toolkit integration
"""
import logging
import os
from typing import Optional
logger = logging.getLogger(__name__)
try:
from fastmcp import FastMCP
FASTMCP_AVAILABLE = True
except ImportError:
FASTMCP_AVAILABLE = False
FastMCP = None
from mcp_auth import (
OAuthProvider,
PolicyEngine,
FastMCPAuthWrapper,
create_default_policies
)
from mcp_auth.clerk_config import create_mcp_server_config
def create_auth_enabled_app(app_name: str = "Yargı MCP Server") -> FastMCP:
"""Create FastMCP app with authentication enabled"""
if not FASTMCP_AVAILABLE:
raise ImportError("FastMCP is required for authenticated MCP server")
logger.info("Creating FastMCP app with MCP Auth Toolkit integration")
# Create base FastMCP app
app = FastMCP(app_name)
# Check if authentication is enabled
auth_enabled = os.getenv("ENABLE_AUTH", "true").lower() == "true"
if not auth_enabled:
logger.info("Authentication disabled, returning basic FastMCP app")
return app
try:
# Get configuration
logger.info("Getting MCP server configuration...")
config = create_mcp_server_config()
logger.info("Configuration loaded successfully")
# Create OAuth provider with Clerk config
logger.info("Creating OAuth provider...")
oauth_provider = OAuthProvider(
config=config["oauth_config"],
jwt_secret=config["jwt_secret"]
)
logger.info("OAuth provider created successfully")
# Create policy engine for Turkish legal database
policy_engine = create_default_policies()
# Store auth components for later wrapping (after tools are defined)
app._oauth_provider = oauth_provider
app._policy_engine = policy_engine
app._auth_config = config
# Add OAuth endpoints immediately
@app.tool(
description="Initiate OAuth 2.1 authorization flow with PKCE",
annotations={"readOnlyHint": True, "idempotentHint": False}
)
async def oauth_authorize(redirect_uri: str, scopes: str = None):
"""OAuth authorization endpoint"""
scope_list = scopes.split(" ") if scopes else ["mcp:tools:read", "mcp:tools:write"]
auth_url, pkce = oauth_provider.generate_authorization_url(
redirect_uri=redirect_uri, scopes=scope_list
)
logger.info(f"Generated authorization URL for redirect_uri: {redirect_uri}")
return {
"authorization_url": auth_url,
"code_verifier": pkce.verifier,
"code_challenge": pkce.challenge,
"instructions": "Use the authorization_url to complete OAuth flow, then exchange the returned code using oauth_token tool"
}
@app.tool(
description="Exchange OAuth authorization code for access token",
annotations={"readOnlyHint": False, "idempotentHint": False}
)
async def oauth_token(code: str, state: str, redirect_uri: str):
"""OAuth token exchange endpoint"""
try:
result = await oauth_provider.exchange_code_for_token(
code=code, state=state, redirect_uri=redirect_uri
)
logger.info("Successfully exchanged authorization code for token")
return result
except Exception as e:
logger.error(f"Token exchange failed: {e}")
raise
@app.tool(
description="Validate and introspect OAuth access token",
annotations={"readOnlyHint": True, "idempotentHint": True}
)
async def oauth_introspect(token: str):
"""Token introspection endpoint"""
result = oauth_provider.introspect_token(token)
logger.debug(f"Token introspection: active={result.get('active', False)}")
return result
@app.tool(
description="Revoke OAuth access token",
annotations={"readOnlyHint": False, "idempotentHint": False}
)
async def oauth_revoke(token: str):
"""Token revocation endpoint"""
success = oauth_provider.revoke_token(token)
logger.info(f"Token revocation: success={success}")
return {"revoked": success}
logger.info("Successfully created authenticated FastMCP app")
except Exception as e:
logger.error(f"Failed to create authenticated app: {e}")
logger.info("Falling back to non-authenticated FastMCP app")
# Return basic app if auth setup fails
return app
return app
def create_app() -> FastMCP:
"""Create FastMCP app (backwards compatible with mcp_factory.py)"""
return create_auth_enabled_app()
def get_auth_wrapper(app: FastMCP) -> Optional[FastMCPAuthWrapper]:
"""Get auth wrapper from app if available"""
return getattr(app, '_auth_wrapper', None)
def get_oauth_provider(app: FastMCP) -> Optional[OAuthProvider]:
"""Get OAuth provider from app if available"""
return getattr(app, '_oauth_provider', None)
def get_policy_engine(app: FastMCP) -> Optional[PolicyEngine]:
"""Get policy engine from app if available"""
return getattr(app, '_policy_engine', None)
def is_auth_enabled(app: FastMCP) -> bool:
"""Check if authentication is enabled for the app"""
return hasattr(app, '_oauth_provider') or hasattr(app, '_auth_wrapper')
def enable_tool_authentication(app: FastMCP):
"""Enable authentication on all existing tools (call after tools are defined)"""
if not is_auth_enabled(app):
logger.debug("Authentication not enabled, skipping tool authentication")
return
oauth_provider = get_oauth_provider(app)
policy_engine = get_policy_engine(app)
if not oauth_provider or not policy_engine:
logger.warning("OAuth provider or policy engine not available")
return
try:
# Create auth wrapper and wrap tools
auth_wrapper = FastMCPAuthWrapper(
mcp_server=app,
oauth_provider=oauth_provider,
policy_engine=policy_engine
)
# Store wrapper for reference
app._auth_wrapper = auth_wrapper
logger.info("Tool authentication enabled successfully")
except Exception as e:
logger.error(f"Failed to enable tool authentication: {e}")
def cleanup_auth_sessions(app: FastMCP):
"""Clean up expired auth sessions and tokens"""
oauth_provider = get_oauth_provider(app)
if oauth_provider:
oauth_provider.cleanup_expired_sessions()
logger.debug("Cleaned up expired OAuth sessions")
-383
View File
@@ -1,383 +0,0 @@
"""
HTTP adapter for MCP Auth Toolkit OAuth endpoints
Exposes MCP OAuth tools as HTTP endpoints for Claude.ai integration
"""
import os
import logging
import secrets
import time
from typing import Optional
from urllib.parse import urlencode, quote
from datetime import datetime, timedelta
from fastapi import APIRouter, Request, Query, HTTPException
from fastapi.responses import RedirectResponse, JSONResponse
# Try to import Clerk SDK
try:
from clerk_backend_api import Clerk
CLERK_AVAILABLE = True
except ImportError as e:
CLERK_AVAILABLE = False
Clerk = None
logger = logging.getLogger(__name__)
router = APIRouter()
# OAuth configuration
BASE_URL = os.getenv("BASE_URL", "https://yargimcp.com")
@router.get("/.well-known/oauth-authorization-server")
async def get_oauth_metadata():
"""OAuth 2.0 Authorization Server Metadata (RFC 8414)"""
return JSONResponse({
"issuer": BASE_URL,
"authorization_endpoint": f"{BASE_URL}/authorize",
"token_endpoint": f"{BASE_URL}/token",
"registration_endpoint": f"{BASE_URL}/register",
"response_types_supported": ["code"],
"grant_types_supported": ["authorization_code", "refresh_token"],
"code_challenge_methods_supported": ["S256"],
"token_endpoint_auth_methods_supported": ["none"],
"scopes_supported": ["mcp:tools:read", "mcp:tools:write", "openid", "profile", "email"],
"service_documentation": f"{BASE_URL}/mcp/"
})
@router.get("/.well-known/oauth-protected-resource")
async def get_protected_resource_metadata():
"""OAuth Protected Resource Metadata (RFC 9728)"""
return JSONResponse({
"resource": BASE_URL,
"authorization_servers": [BASE_URL],
"bearer_methods_supported": ["header"],
"scopes_supported": ["mcp:tools:read", "mcp:tools:write"],
"resource_documentation": f"{BASE_URL}/docs"
})
@router.get("/authorize")
async def authorize_endpoint(
response_type: str = Query(...),
client_id: str = Query(...),
redirect_uri: str = Query(...),
code_challenge: str = Query(...),
code_challenge_method: str = Query("S256"),
state: Optional[str] = Query(None),
scope: Optional[str] = Query(None)
):
"""OAuth 2.1 Authorization Endpoint - Uses Clerk SDK for custom domains"""
logger.info(f"OAuth authorize request - client_id: {client_id}, redirect_uri: {redirect_uri}")
if not CLERK_AVAILABLE:
logger.error("Clerk SDK not available")
raise HTTPException(status_code=500, detail="Clerk SDK not available")
# Store OAuth session for later validation
try:
from mcp_server_main import app as mcp_app
from mcp_auth_factory import get_oauth_provider
oauth_provider = get_oauth_provider(mcp_app)
if not oauth_provider:
raise HTTPException(status_code=500, detail="OAuth provider not configured")
# Generate session and store PKCE
session_id = secrets.token_urlsafe(32)
if state is None:
state = secrets.token_urlsafe(16)
# Create PKCE challenge
from mcp_auth.oauth import PKCEChallenge
pkce = PKCEChallenge()
# Store session data
session_data = {
"pkce_verifier": pkce.verifier,
"pkce_challenge": code_challenge, # Store the client's challenge
"state": state,
"redirect_uri": redirect_uri,
"client_id": client_id,
"scopes": scope.split(" ") if scope else ["mcp:tools:read", "mcp:tools:write"],
"created_at": time.time(),
"expires_at": (datetime.utcnow() + timedelta(minutes=10)).timestamp(),
}
oauth_provider.storage.set_session(session_id, session_data)
# For Clerk with custom domains, we need to use their hosted sign-in page
# We'll pass our callback URL and session info in the state
callback_url = f"{BASE_URL}/auth/callback"
# Encode session info in state for retrieval after Clerk auth
combined_state = f"{state}:{session_id}"
# Use Clerk's sign-in URL with proper parameters
clerk_domain = os.getenv("CLERK_DOMAIN", "accounts.yargimcp.com")
sign_in_params = {
"redirect_url": f"{callback_url}?state={quote(combined_state)}",
}
sign_in_url = f"https://{clerk_domain}/sign-in?{urlencode(sign_in_params)}"
logger.info(f"Redirecting to Clerk sign-in: {sign_in_url}")
return RedirectResponse(url=sign_in_url)
except Exception as e:
logger.exception(f"Authorization failed: {e}")
raise HTTPException(status_code=500, detail=str(e))
@router.get("/auth/callback")
async def oauth_callback(
request: Request,
state: Optional[str] = Query(None),
clerk_token: Optional[str] = Query(None)
):
"""Handle OAuth callback from Clerk - supports both JWT token and cookie auth"""
logger.info(f"OAuth callback received - state: {state}")
logger.info(f"Query params: {dict(request.query_params)}")
logger.info(f"Cookies: {dict(request.cookies)}")
logger.info(f"Clerk JWT token provided: {bool(clerk_token)}")
# Support both JWT token (for cross-domain) and cookie auth (for subdomain)
try:
if not state:
logger.error("No state parameter provided")
return JSONResponse(
status_code=400,
content={"error": "invalid_request", "error_description": "Missing state parameter"}
)
# Parse state to get original state and session ID
try:
if ":" in state:
original_state, session_id = state.rsplit(":", 1)
else:
original_state = state
session_id = state # Fallback
except ValueError:
logger.error(f"Invalid state format: {state}")
return JSONResponse(
status_code=400,
content={"error": "invalid_request", "error_description": "Invalid state format"}
)
# Get OAuth provider
from mcp_server_main import app as mcp_app
from mcp_auth_factory import get_oauth_provider
oauth_provider = get_oauth_provider(mcp_app)
if not oauth_provider:
raise HTTPException(status_code=500, detail="OAuth provider not configured")
# Get stored session
oauth_session = oauth_provider.storage.get_session(session_id)
if not oauth_session:
logger.error(f"OAuth session not found for ID: {session_id}")
return JSONResponse(
status_code=400,
content={"error": "invalid_request", "error_description": "OAuth session expired or not found"}
)
# Check if we have a JWT token (for cross-domain auth)
user_authenticated = False
auth_method = "none"
if clerk_token:
logger.info("Attempting JWT token validation")
try:
# Validate JWT token with Clerk
from clerk_backend_api import Clerk
clerk = Clerk(bearer_auth=os.getenv("CLERK_SECRET_KEY"))
# Extract session_id from JWT token and verify with Clerk
import jwt
decoded_token = jwt.decode(clerk_token, options={"verify_signature": False})
session_id = decoded_token.get("sid") or decoded_token.get("session_id")
if session_id:
# Verify with Clerk using session_id
session = clerk.sessions.verify(session_id=session_id, token=clerk_token)
user_id = session.user_id if session else None
else:
user_id = None
if user_id:
logger.info(f"JWT token validation successful - user_id: {user_id}")
user_authenticated = True
auth_method = "jwt_token"
# Store user info in session for token exchange
oauth_session["user_id"] = user_id
oauth_session["auth_method"] = "jwt_token"
else:
logger.error("JWT token validation failed - no user_id in claims")
except Exception as e:
logger.error(f"JWT token validation failed: {str(e)}")
# Fall through to cookie validation
# If no JWT token or validation failed, check cookies
if not user_authenticated:
logger.info("Checking for Clerk session cookies")
# Check for Clerk session cookies (for subdomain auth)
clerk_session_cookie = request.cookies.get("__session")
if clerk_session_cookie:
logger.info("Found Clerk session cookie, assuming authenticated")
user_authenticated = True
auth_method = "cookie"
oauth_session["auth_method"] = "cookie"
else:
logger.info("No Clerk session cookie found")
# For custom domains, we'll also trust that Clerk redirected here
if not user_authenticated:
logger.info("Trusting Clerk redirect for custom domain flow")
user_authenticated = True
auth_method = "trusted_redirect"
oauth_session["auth_method"] = "trusted_redirect"
logger.info(f"User authenticated: {user_authenticated}, method: {auth_method}")
# Generate simple authorization code for custom domain flow
auth_code = f"clerk_custom_{session_id}_{int(time.time())}"
# Store the code mapping for token exchange
code_data = {
"session_id": session_id,
"clerk_authenticated": user_authenticated,
"auth_method": auth_method,
"custom_domain_flow": True,
"created_at": time.time(),
"expires_at": (datetime.utcnow() + timedelta(minutes=5)).timestamp(),
}
if "user_id" in oauth_session:
code_data["user_id"] = oauth_session["user_id"]
oauth_provider.storage.set_session(f"code_{auth_code}", code_data)
# Build redirect URL back to Claude
redirect_params = {
"code": auth_code,
"state": original_state
}
redirect_url = f"{oauth_session['redirect_uri']}?{urlencode(redirect_params)}"
logger.info(f"Redirecting back to Claude: {redirect_url}")
return RedirectResponse(url=redirect_url)
except Exception as e:
logger.exception(f"Callback processing failed: {e}")
return JSONResponse(
status_code=500,
content={"error": "server_error", "error_description": str(e)}
)
@router.post("/register")
async def register_client(request: Request):
"""Dynamic Client Registration (RFC 7591)"""
data = await request.json()
logger.info(f"Client registration request: {data}")
# Simple dynamic registration - accept any client
client_id = f"mcp-client-{os.urandom(8).hex()}"
return JSONResponse({
"client_id": client_id,
"client_secret": None, # Public client
"redirect_uris": data.get("redirect_uris", []),
"grant_types": ["authorization_code", "refresh_token"],
"response_types": ["code"],
"client_name": data.get("client_name", "MCP Client"),
"token_endpoint_auth_method": "none",
"client_id_issued_at": int(datetime.now().timestamp())
})
@router.post("/token")
async def token_endpoint(request: Request):
"""OAuth 2.1 Token Endpoint"""
# Parse form data
form_data = await request.form()
grant_type = form_data.get("grant_type")
code = form_data.get("code")
redirect_uri = form_data.get("redirect_uri")
client_id = form_data.get("client_id")
code_verifier = form_data.get("code_verifier")
logger.info(f"Token exchange - grant_type: {grant_type}, code: {code[:20] if code else 'None'}...")
if grant_type != "authorization_code":
return JSONResponse(
status_code=400,
content={"error": "unsupported_grant_type"}
)
try:
# OAuth token exchange - validate code and return Clerk JWT
# This supports proper OAuth flow while using Clerk JWT tokens
if not code or not redirect_uri:
logger.error("Missing required parameters: code or redirect_uri")
return JSONResponse(
status_code=400,
content={"error": "invalid_request", "error_description": "Missing code or redirect_uri"}
)
# Validate OAuth code with Clerk
if CLERK_AVAILABLE:
try:
clerk = Clerk(bearer_auth=os.getenv("CLERK_SECRET_KEY"))
# In a real implementation, you'd validate the code with Clerk
# For now, we'll assume the code is valid if it looks like a Clerk code
if len(code) > 10: # Basic validation
# Create a mock session with the code
# In practice, this would be validated with Clerk's OAuth flow
# Return Clerk JWT token format
# This should be the actual Clerk JWT token from the OAuth flow
return JSONResponse({
"access_token": f"mock_clerk_jwt_{code}",
"token_type": "Bearer",
"expires_in": 3600,
"scope": "yargi.read yargi.search"
})
else:
logger.error(f"Invalid code format: {code}")
return JSONResponse(
status_code=400,
content={"error": "invalid_grant", "error_description": "Invalid authorization code"}
)
except Exception as e:
logger.error(f"Clerk validation failed: {e}")
return JSONResponse(
status_code=400,
content={"error": "invalid_grant", "error_description": "Authorization code validation failed"}
)
else:
logger.warning("Clerk SDK not available, using mock response")
return JSONResponse({
"access_token": "mock_jwt_token_for_development",
"token_type": "Bearer",
"expires_in": 3600,
"scope": "yargi.read yargi.search"
})
except Exception as e:
logger.exception(f"Token exchange failed: {e}")
return JSONResponse(
status_code=500,
content={"error": "server_error", "error_description": str(e)}
)
-522
View File
@@ -1,522 +0,0 @@
"""
Simplified MCP OAuth HTTP adapter - only Clerk JWT based authentication
Uses Redis for authorization code storage to support multi-machine deployment
"""
import os
import logging
from typing import Optional
from urllib.parse import urlencode, quote
from fastapi import APIRouter, Request, Query, HTTPException
from fastapi.responses import RedirectResponse, JSONResponse
# Import Redis session store
from redis_session_store import get_redis_store
# Try to import Clerk SDK
try:
from clerk_backend_api import Clerk
CLERK_AVAILABLE = True
except ImportError:
CLERK_AVAILABLE = False
Clerk = None
logger = logging.getLogger(__name__)
router = APIRouter()
# OAuth configuration
BASE_URL = os.getenv("BASE_URL", "https://api.yargimcp.com")
CLERK_DOMAIN = os.getenv("CLERK_DOMAIN", "accounts.yargimcp.com")
# Initialize Redis store
redis_store = None
def get_redis_session_store():
"""Get Redis store instance with lazy initialization."""
global redis_store
if redis_store is None:
try:
import concurrent.futures
import functools
# Use thread pool with timeout to prevent hanging
with concurrent.futures.ThreadPoolExecutor(max_workers=1) as executor:
future = executor.submit(get_redis_store)
try:
# 5 second timeout for Redis initialization
redis_store = future.result(timeout=5.0)
if redis_store:
logger.info("Redis session store initialized for OAuth handler")
else:
logger.warning("Redis store initialization returned None")
except concurrent.futures.TimeoutError:
logger.error("Redis initialization timed out after 5 seconds")
redis_store = None
future.cancel() # Try to cancel the hanging operation
except Exception as e:
logger.error(f"Failed to initialize Redis store: {e}")
redis_store = None
if redis_store is None:
# Fall back to in-memory storage with warning
logger.warning("Falling back to in-memory storage - multi-machine deployment will not work")
return redis_store
@router.get("/.well-known/oauth-authorization-server")
async def get_oauth_metadata():
"""OAuth 2.0 Authorization Server Metadata (RFC 8414)"""
return JSONResponse({
"issuer": BASE_URL,
"authorization_endpoint": "https://yargimcp.com/mcp-callback",
"token_endpoint": f"{BASE_URL}/token",
"registration_endpoint": f"{BASE_URL}/register",
"response_types_supported": ["code"],
"grant_types_supported": ["authorization_code"],
"code_challenge_methods_supported": ["S256"],
"token_endpoint_auth_methods_supported": ["none"],
"scopes_supported": ["read", "search", "openid", "profile", "email"],
"service_documentation": f"{BASE_URL}/mcp/"
})
@router.get("/auth/login")
async def oauth_authorize(
request: Request,
client_id: str = Query(...),
redirect_uri: str = Query(...),
response_type: str = Query("code"),
scope: Optional[str] = Query("read search"),
state: Optional[str] = Query(None),
code_challenge: Optional[str] = Query(None),
code_challenge_method: Optional[str] = Query(None)
):
"""OAuth 2.1 Authorization Endpoint - redirects to Clerk"""
logger.info(f"OAuth authorize request - client_id: {client_id}")
logger.info(f"Redirect URI: {redirect_uri}")
logger.info(f"State: {state}")
logger.info(f"PKCE Challenge: {bool(code_challenge)}")
try:
# Build callback URL with all necessary parameters
callback_url = f"{BASE_URL}/auth/callback"
callback_params = {
"client_id": client_id,
"redirect_uri": redirect_uri,
"state": state or "",
"scope": scope or "read search"
}
# Add PKCE parameters if present
if code_challenge:
callback_params["code_challenge"] = code_challenge
callback_params["code_challenge_method"] = code_challenge_method or "S256"
# Encode callback URL as redirect_url for Clerk
callback_with_params = f"{callback_url}?{urlencode(callback_params)}"
# Build Clerk sign-in URL - use yargimcp.com frontend for JWT token generation
clerk_params = {
"redirect_url": callback_with_params
}
# Use frontend sign-in page that handles JWT token generation
clerk_signin_url = f"https://yargimcp.com/sign-in?{urlencode(clerk_params)}"
logger.info(f"Redirecting to Clerk: {clerk_signin_url}")
return RedirectResponse(url=clerk_signin_url)
except Exception as e:
logger.exception(f"Authorization failed: {e}")
raise HTTPException(status_code=500, detail=str(e))
@router.get("/auth/callback")
async def oauth_callback(
request: Request,
client_id: str = Query(...),
redirect_uri: str = Query(...),
state: Optional[str] = Query(None),
scope: Optional[str] = Query("read search"),
code_challenge: Optional[str] = Query(None),
code_challenge_method: Optional[str] = Query(None),
clerk_token: Optional[str] = Query(None)
):
"""OAuth callback from Clerk - generates authorization code"""
logger.info(f"OAuth callback - client_id: {client_id}")
logger.info(f"Clerk token provided: {bool(clerk_token)}")
try:
# Validate user with Clerk and generate real JWT token
user_authenticated = False
user_id = None
session_id = None
real_jwt_token = None
if clerk_token and CLERK_AVAILABLE:
try:
# Extract user info from JWT token (no Clerk session verification needed)
import jwt
decoded_token = jwt.decode(clerk_token, options={"verify_signature": False})
user_id = decoded_token.get("user_id") or decoded_token.get("sub")
user_email = decoded_token.get("email")
token_scopes = decoded_token.get("scopes", ["read", "search"])
logger.info(f"JWT token claims - user_id: {user_id}, email: {user_email}, scopes: {token_scopes}")
if user_id and user_email:
# JWT token is already signed by Clerk and contains valid user info
user_authenticated = True
logger.info(f"User authenticated via JWT token - user_id: {user_id}")
# Use the JWT token directly as the real token (it's already from Clerk template)
real_jwt_token = clerk_token
logger.info("Using Clerk JWT token directly (already real token)")
else:
logger.error(f"Missing required fields in JWT token - user_id: {bool(user_id)}, email: {bool(user_email)}")
except Exception as e:
logger.error(f"JWT validation failed: {e}")
# Fallback to cookie validation
if not user_authenticated:
clerk_session = request.cookies.get("__session")
if clerk_session:
user_authenticated = True
logger.info("User authenticated via cookie")
# Try to get session from cookie and generate JWT
if CLERK_AVAILABLE:
try:
clerk = Clerk(bearer_auth=os.getenv("CLERK_SECRET_KEY"))
# Note: sessions.verify_session is deprecated, but we'll try
# In practice, you'd need to extract session_id from cookie
logger.info("Cookie authentication - JWT generation not implemented yet")
except Exception as e:
logger.warning(f"Failed to generate JWT from cookie: {e}")
# Only generate authorization code if we have a real JWT token
if user_authenticated and real_jwt_token:
# Generate authorization code
auth_code = f"clerk_auth_{os.urandom(16).hex()}"
# Prepare code data
import time
code_data = {
"user_id": user_id,
"session_id": session_id,
"real_jwt_token": real_jwt_token,
"user_authenticated": user_authenticated,
"client_id": client_id,
"redirect_uri": redirect_uri,
"scope": scope or "read search"
}
# Try to store in Redis, fall back to in-memory if Redis unavailable
store = get_redis_session_store()
if store:
# Store in Redis with automatic expiration
success = store.set_oauth_code(auth_code, code_data)
if success:
logger.info(f"Stored authorization code {auth_code[:10]}... in Redis with real JWT token")
else:
logger.error(f"Failed to store authorization code in Redis, falling back to in-memory")
# Fall back to in-memory storage
if not hasattr(oauth_callback, '_code_storage'):
oauth_callback._code_storage = {}
oauth_callback._code_storage[auth_code] = code_data
else:
# Fall back to in-memory storage
logger.warning("Redis not available, using in-memory storage")
if not hasattr(oauth_callback, '_code_storage'):
oauth_callback._code_storage = {}
oauth_callback._code_storage[auth_code] = code_data
logger.info(f"Stored authorization code in memory (fallback)")
# Redirect back to client with authorization code
redirect_params = {
"code": auth_code,
"state": state or ""
}
final_redirect_url = f"{redirect_uri}?{urlencode(redirect_params)}"
logger.info(f"Redirecting back to client: {final_redirect_url}")
return RedirectResponse(url=final_redirect_url)
else:
# No JWT token yet - redirect back to sign-in page to wait for authentication
logger.info("No JWT token provided - redirecting back to sign-in to complete authentication")
# Keep the same redirect URL so the flow continues
sign_in_params = {
"redirect_url": f"{request.url._url}" # Current callback URL with all params
}
sign_in_url = f"https://yargimcp.com/sign-in?{urlencode(sign_in_params)}"
logger.info(f"Redirecting back to sign-in: {sign_in_url}")
return RedirectResponse(url=sign_in_url)
except Exception as e:
logger.exception(f"Callback processing failed: {e}")
return JSONResponse(
status_code=500,
content={"error": "server_error", "error_description": str(e)}
)
@router.post("/auth/register")
async def register_client(request: Request):
"""Dynamic Client Registration (RFC 7591)"""
data = await request.json()
logger.info(f"Client registration request: {data}")
# Simple dynamic registration - accept any client
client_id = f"mcp-client-{os.urandom(8).hex()}"
return JSONResponse({
"client_id": client_id,
"client_secret": None, # Public client
"redirect_uris": data.get("redirect_uris", []),
"grant_types": ["authorization_code"],
"response_types": ["code"],
"client_name": data.get("client_name", "MCP Client"),
"token_endpoint_auth_method": "none"
})
@router.post("/auth/callback")
async def oauth_callback_post(request: Request):
"""OAuth callback POST endpoint for token exchange"""
# Parse form data (standard OAuth token exchange format)
form_data = await request.form()
grant_type = form_data.get("grant_type")
code = form_data.get("code")
redirect_uri = form_data.get("redirect_uri")
client_id = form_data.get("client_id")
code_verifier = form_data.get("code_verifier")
logger.info(f"OAuth callback POST - grant_type: {grant_type}")
logger.info(f"Code: {code[:20] if code else 'None'}...")
logger.info(f"Client ID: {client_id}")
logger.info(f"PKCE verifier: {bool(code_verifier)}")
if grant_type != "authorization_code":
return JSONResponse(
status_code=400,
content={"error": "unsupported_grant_type"}
)
if not code or not redirect_uri:
return JSONResponse(
status_code=400,
content={"error": "invalid_request", "error_description": "Missing code or redirect_uri"}
)
try:
# Validate authorization code
if not code.startswith("clerk_auth_"):
return JSONResponse(
status_code=400,
content={"error": "invalid_grant", "error_description": "Invalid authorization code"}
)
# Retrieve stored JWT token using authorization code from Redis or in-memory fallback
stored_code_data = None
# Try to get from Redis first, then fall back to in-memory
store = get_redis_session_store()
if store:
stored_code_data = store.get_oauth_code(code, delete_after_use=True)
if stored_code_data:
logger.info(f"Retrieved authorization code {code[:10]}... from Redis")
else:
logger.warning(f"Authorization code {code[:10]}... not found in Redis")
# Fall back to in-memory storage if Redis unavailable or code not found
if not stored_code_data and hasattr(oauth_callback, '_code_storage'):
stored_code_data = oauth_callback._code_storage.get(code)
if stored_code_data:
# Clean up in-memory storage
oauth_callback._code_storage.pop(code, None)
logger.info(f"Retrieved authorization code {code[:10]}... from in-memory storage")
if not stored_code_data:
logger.error(f"No stored data found for authorization code: {code}")
return JSONResponse(
status_code=400,
content={"error": "invalid_grant", "error_description": "Authorization code not found or expired"}
)
# Note: Redis TTL handles expiration automatically, but check for manual expiration for in-memory fallback
import time
expires_at = stored_code_data.get("expires_at", 0)
if expires_at and time.time() > expires_at:
logger.error(f"Authorization code expired: {code}")
return JSONResponse(
status_code=400,
content={"error": "invalid_grant", "error_description": "Authorization code expired"}
)
# Get the real JWT token
real_jwt_token = stored_code_data.get("real_jwt_token")
if real_jwt_token:
logger.info("Returning real Clerk JWT token")
# Note: Code already deleted from Redis, clean up in-memory fallback if used
if hasattr(oauth_callback, '_code_storage'):
oauth_callback._code_storage.pop(code, None)
return JSONResponse({
"access_token": real_jwt_token,
"token_type": "Bearer",
"expires_in": 3600,
"scope": "read search"
})
else:
logger.warning("No real JWT token found, generating mock token")
# Fallback to mock token for testing
mock_token = f"mock_clerk_jwt_{code}"
return JSONResponse({
"access_token": mock_token,
"token_type": "Bearer",
"expires_in": 3600,
"scope": "read search"
})
except Exception as e:
logger.exception(f"OAuth callback POST failed: {e}")
return JSONResponse(
status_code=500,
content={"error": "server_error", "error_description": str(e)}
)
@router.post("/register")
async def register_client(request: Request):
"""Dynamic Client Registration (RFC 7591)"""
data = await request.json()
logger.info(f"Client registration request: {data}")
# Simple dynamic registration - accept any client
client_id = f"mcp-client-{os.urandom(8).hex()}"
return JSONResponse({
"client_id": client_id,
"client_secret": None, # Public client
"redirect_uris": data.get("redirect_uris", []),
"grant_types": ["authorization_code"],
"response_types": ["code"],
"client_name": data.get("client_name", "MCP Client"),
"token_endpoint_auth_method": "none"
})
@router.post("/token")
async def token_endpoint(request: Request):
"""OAuth 2.1 Token Endpoint - exchanges code for Clerk JWT"""
# Parse form data
form_data = await request.form()
grant_type = form_data.get("grant_type")
code = form_data.get("code")
redirect_uri = form_data.get("redirect_uri")
client_id = form_data.get("client_id")
code_verifier = form_data.get("code_verifier")
logger.info(f"Token exchange - grant_type: {grant_type}")
logger.info(f"Code: {code[:20] if code else 'None'}...")
if grant_type != "authorization_code":
return JSONResponse(
status_code=400,
content={"error": "unsupported_grant_type"}
)
if not code or not redirect_uri:
return JSONResponse(
status_code=400,
content={"error": "invalid_request", "error_description": "Missing code or redirect_uri"}
)
try:
# Validate authorization code
if not code.startswith("clerk_auth_"):
return JSONResponse(
status_code=400,
content={"error": "invalid_grant", "error_description": "Invalid authorization code"}
)
# Retrieve stored JWT token using authorization code from Redis or in-memory fallback
stored_code_data = None
# Try to get from Redis first, then fall back to in-memory
store = get_redis_session_store()
if store:
stored_code_data = store.get_oauth_code(code, delete_after_use=True)
if stored_code_data:
logger.info(f"Retrieved authorization code {code[:10]}... from Redis (/token endpoint)")
else:
logger.warning(f"Authorization code {code[:10]}... not found in Redis (/token endpoint)")
# Fall back to in-memory storage if Redis unavailable or code not found
if not stored_code_data and hasattr(oauth_callback, '_code_storage'):
stored_code_data = oauth_callback._code_storage.get(code)
if stored_code_data:
# Clean up in-memory storage
oauth_callback._code_storage.pop(code, None)
logger.info(f"Retrieved authorization code {code[:10]}... from in-memory storage (/token endpoint)")
if not stored_code_data:
logger.error(f"No stored data found for authorization code: {code}")
return JSONResponse(
status_code=400,
content={"error": "invalid_grant", "error_description": "Authorization code not found or expired"}
)
# Note: Redis TTL handles expiration automatically, but check for manual expiration for in-memory fallback
import time
expires_at = stored_code_data.get("expires_at", 0)
if expires_at and time.time() > expires_at:
logger.error(f"Authorization code expired: {code}")
return JSONResponse(
status_code=400,
content={"error": "invalid_grant", "error_description": "Authorization code expired"}
)
# Get the real JWT token
real_jwt_token = stored_code_data.get("real_jwt_token")
if real_jwt_token:
logger.info("Returning real Clerk JWT token from /token endpoint")
# Note: Code already deleted from Redis, clean up in-memory fallback if used
if hasattr(oauth_callback, '_code_storage'):
oauth_callback._code_storage.pop(code, None)
return JSONResponse({
"access_token": real_jwt_token,
"token_type": "Bearer",
"expires_in": 3600,
"scope": "read search"
})
else:
logger.warning("No real JWT token found in /token endpoint, generating mock token")
# Fallback to mock token for testing
mock_token = f"mock_clerk_jwt_{code}"
return JSONResponse({
"access_token": mock_token,
"token_type": "Bearer",
"expires_in": 3600,
"scope": "read search"
})
except Exception as e:
logger.exception(f"Token exchange failed: {e}")
return JSONResponse(
status_code=500,
content={"error": "server_error", "error_description": str(e)}
)
+485 -202
View File
File diff suppressed because it is too large Load Diff
+57
View File
@@ -0,0 +1,57 @@
"""
Migration stub for the deprecated Yargı MCP endpoint.
Exposes a single tool that informs the MCP client the server has moved
and the user must update their configuration.
Entrypoint variable `app` is a FastMCP instance so it works with
Dokploy's FastMCP build pipeline (`fastmcp inspect`, `fastmcp run`).
Run with:
fastmcp run migration_app.py:app --transport http --port 8000
"""
from starlette.responses import JSONResponse
from fastmcp import FastMCP
NEW_URL = "https://yargimcp.surucu.dev/mcp"
app = FastMCP(
name="Yargı MCP (taşındı / moved)",
instructions=(
f"Bu Yargı MCP endpoint'i kullanımdan kaldırıldı. "
f"Sunucu yeni adrese taşındı: {NEW_URL}. "
f"Kullanıcıya MCP istemci ayarlarındaki sunucu URL'sini güncellemesini söyle."
),
)
@app.tool(
description=(
"DEPRECATED ENDPOINT — Yargı MCP sunucusu yeni adrese taşındı. "
"Bu endpoint'teki eski araçlar (Yargıtay, Danıştay, Anayasa Mahkemesi, "
"KVKK, BDDK, Rekabet Kurumu, Sayıştay, KİK, Emsal, Uyuşmazlık, GİB, "
"Sigorta Tahkim, Bedesten vb.) artık burada çalışmıyor. "
"Bu aracı çağırarak yeni sunucu adresini ve migrasyon talimatını al, "
"ardından kullanıcıya MCP istemci (Claude Desktop, 5ire, ChatGPT vb.) "
"ayarlarındaki sunucu URL'sini güncellemesi gerektiğini açıkça söyle."
)
)
def migration_notice() -> dict:
"""Yargı MCP sunucusunun yeni adresini ve migrasyon talimatını döner."""
return {
"status": "moved",
"new_url": NEW_URL,
"message": (
f"Yargı MCP sunucusu yeni adrese taşındı: {NEW_URL}\n\n"
f"Lütfen MCP istemcinin (Claude Desktop, 5ire, ChatGPT vb.) "
f"ayarlarındaki sunucu URL'sini yukarıdaki yeni adresle güncelleyin. "
f"Mevcut endpoint artık kullanım dışıdır ve sadece bu uyarıyı döner."
),
}
@app.custom_route("/health", methods=["GET"])
async def health(request):
"""Health check endpoint for monitoring services."""
return JSONResponse({"status": "deprecated", "new_url": NEW_URL})
-94
View File
@@ -1,94 +0,0 @@
events {
worker_connections 1024;
}
http {
upstream yargi_mcp {
server yargi-mcp:8000;
}
# Rate limiting
limit_req_zone $binary_remote_addr zone=api_limit:10m rate=10r/s;
limit_req_zone $binary_remote_addr zone=mcp_limit:10m rate=100r/s;
server {
listen 80;
server_name localhost;
# Redirect HTTP to HTTPS in production
# return 301 https://$server_name$request_uri;
# Security headers
add_header X-Content-Type-Options nosniff;
add_header X-Frame-Options DENY;
add_header X-XSS-Protection "1; mode=block";
add_header Referrer-Policy "strict-origin-when-cross-origin";
# API endpoints
location /api/ {
limit_req zone=api_limit burst=20 nodelay;
proxy_pass http://yargi_mcp;
proxy_set_header Host $host;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header X-Forwarded-Proto $scheme;
# Timeouts
proxy_connect_timeout 60s;
proxy_send_timeout 60s;
proxy_read_timeout 60s;
}
# MCP endpoint (higher rate limit)
location /mcp-server/mcp/ {
limit_req zone=mcp_limit burst=50 nodelay;
proxy_pass http://yargi_mcp;
proxy_set_header Host $host;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header X-Forwarded-Proto $scheme;
# WebSocket support
proxy_http_version 1.1;
proxy_set_header Upgrade $http_upgrade;
proxy_set_header Connection "upgrade";
# Longer timeouts for MCP operations
proxy_connect_timeout 300s;
proxy_send_timeout 300s;
proxy_read_timeout 300s;
}
# Health check (no rate limit)
location /health {
proxy_pass http://yargi_mcp;
proxy_set_header Host $host;
}
# Root and other paths
location / {
limit_req zone=api_limit burst=10 nodelay;
proxy_pass http://yargi_mcp;
proxy_set_header Host $host;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header X-Forwarded-Proto $scheme;
}
}
# SSL configuration (uncomment for production)
# server {
# listen 443 ssl http2;
# server_name your-domain.com;
#
# ssl_certificate /etc/nginx/ssl/cert.pem;
# ssl_certificate_key /etc/nginx/ssl/key.pem;
# ssl_protocols TLSv1.2 TLSv1.3;
# ssl_ciphers HIGH:!aNULL:!MD5;
#
# # Include all location blocks from above
# }
}
+3 -10
View File
@@ -1,6 +1,6 @@
[project]
name = "yargi-mcp"
version = "0.2.0"
version = "0.2.1"
description = "MCP Server For Turkish Legal Databases"
readme = "README.md"
requires-python = ">=3.11"
@@ -46,22 +46,15 @@ production = [
"gunicorn>=22.0.0",
"uvicorn[standard]>=0.30.0",
]
saas = [
"clerk-backend-api>=3.0.0",
"stripe>=9.1.0",
"upstash-redis>=1.1.0",
"tiktoken>=0.5.0",
"PyJWT>=2.8.0",
]
[project.scripts]
yargi-mcp = "mcp_server_main:main"
[tool.setuptools]
py-modules = ["mcp_server_main", "mcp_auth_factory", "mcp_auth_http_adapter", "asgi_app", "fastapi_app", "starlette_app", "run_asgi", "stripe_webhook"]
py-modules = ["mcp_server_main", "asgi_app"]
[tool.setuptools.packages.find]
include = ["*_mcp_module", "mcp_auth", "semantic_search"]
include = ["*_mcp_module", "semantic_search"]
[build-system]
requires = ["setuptools>=65.0", "wheel"]
+20 -23
View File
@@ -1,5 +1,6 @@
# rekabet_mcp_module/client.py
import asyncio
import httpx
from bs4 import BeautifulSoup
from typing import List, Optional, Tuple, Dict, Any
@@ -141,30 +142,30 @@ class RekabetKurumuApiClient:
# Row 1: Publication Date, Decision Number, Related Cases Link
td_elements_r1 = rows[0].find_all("td")
pub_date = td_elements_r1[0].get_text(strip=True) if len(td_elements_r1) > 0 else None
dec_num = td_elements_r1[1].get_text(strip=True) if len(td_elements_r1) > 1 else None
pub_date = td_elements_r1[0].get_text(strip=True) if len(td_elements_r1) > 0 else ""
dec_num = td_elements_r1[1].get_text(strip=True) if len(td_elements_r1) > 1 else ""
related_cases_link_tag = td_elements_r1[2].find("a", href=True) if len(td_elements_r1) > 2 else None
related_cases_url_str: Optional[str] = None
karar_id_from_related: Optional[str] = None
related_cases_url_str: str = ""
karar_id_from_related: str = ""
if related_cases_link_tag and related_cases_link_tag.has_attr('href'):
related_cases_url_str = urljoin(self.BASE_URL, related_cases_link_tag['href'])
qs_related = parse_qs(urlparse(related_cases_link_tag['href']).query)
if 'kararId' in qs_related and qs_related['kararId']:
karar_id_from_related = qs_related['kararId'][0]
# Row 2: Decision Date, Decision Type
td_elements_r2 = rows[1].find_all("td")
dec_date = td_elements_r2[0].get_text(strip=True) if len(td_elements_r2) > 0 else None
dec_type_text = td_elements_r2[1].get_text(strip=True) if len(td_elements_r2) > 1 else None
dec_date = td_elements_r2[0].get_text(strip=True) if len(td_elements_r2) > 0 else ""
dec_type_text = td_elements_r2[1].get_text(strip=True) if len(td_elements_r2) > 1 else ""
# Row 3: Title and Main Decision Link
title_cell = rows[2].find("td", colspan="5")
decision_link_tag = title_cell.find("a", href=True) if title_cell else None
title_text: Optional[str] = None
decision_landing_url_str: Optional[str] = None
karar_id_from_main_link: Optional[str] = None
title_text: str = ""
decision_landing_url_str: str = ""
karar_id_from_main_link: str = ""
if decision_link_tag and decision_link_tag.has_attr('href'):
title_text = decision_link_tag.get_text(strip=True)
@@ -178,23 +179,19 @@ class RekabetKurumuApiClient:
logger.warning(f"Table {idx+1} decision link has unexpected format: {href_val}")
else:
logger.warning(f"Table {idx+1} could not find title/decision link tag.")
current_karar_id = karar_id_from_main_link or karar_id_from_related
if not current_karar_id:
if not current_karar_id:
logger.warning(f"Table {idx+1} Karar ID not found. Skipping. Title (if any): {title_text}")
continue
# Convert string URLs to HttpUrl for the model
final_decision_url = HttpUrl(decision_landing_url_str) if decision_landing_url_str else None
final_related_cases_url = HttpUrl(related_cases_url_str) if related_cases_url_str else None
processed_decisions.append(RekabetDecisionSummary(
publication_date=pub_date, decision_number=dec_num, decision_date=dec_date,
decision_type_text=dec_type_text, title=title_text,
decision_url=final_decision_url,
karar_id=current_karar_id,
related_cases_url=final_related_cases_url
decision_type_text=dec_type_text, title=title_text,
decision_url=decision_landing_url_str,
karar_id=current_karar_id,
related_cases_url=related_cases_url_str
))
logger.debug(f"Table {idx+1} parsed successfully: Karar ID '{current_karar_id}', Title '{title_text[:50] if title_text else 'N/A'}...'")
@@ -357,7 +354,7 @@ class RekabetKurumuApiClient:
total_pdf_pages = total_pdf_pages_from_extraction
if single_page_pdf_bytes:
markdown_for_requested_page = self._convert_pdf_bytes_to_markdown(single_page_pdf_bytes, str(pdf_url_to_report or full_landing_page_url))
markdown_for_requested_page = await asyncio.to_thread(self._convert_pdf_bytes_to_markdown, single_page_pdf_bytes, str(pdf_url_to_report or full_landing_page_url))
if not markdown_for_requested_page:
error_message = (error_message or "") + f"; Could not convert page {page_number} of PDF to Markdown."
elif total_pdf_pages > 0 :
-119
View File
@@ -1,119 +0,0 @@
#!/usr/bin/env python3
"""
Standalone ASGI server runner for Yargı MCP
This script provides a simple way to run the Yargı MCP server
as a web service using uvicorn.
Usage:
python run_asgi.py
python run_asgi.py --host 0.0.0.0 --port 8080
python run_asgi.py --reload # For development
"""
import os
import sys
import argparse
import logging
from pathlib import Path
# Add project root to Python path
sys.path.insert(0, str(Path(__file__).parent))
try:
import uvicorn
except ImportError:
print("Error: uvicorn is not installed.")
print("Please install it with: pip install uvicorn")
sys.exit(1)
# Configure logging
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
)
def main():
parser = argparse.ArgumentParser(
description="Run Yargı MCP server as an ASGI web service"
)
parser.add_argument(
"--host",
type=str,
default=os.getenv("HOST", "127.0.0.1"),
help="Host to bind to (default: 127.0.0.1)"
)
parser.add_argument(
"--port",
type=int,
default=int(os.getenv("PORT", "8000")),
help="Port to bind to (default: 8000)"
)
parser.add_argument(
"--reload",
action="store_true",
help="Enable auto-reload for development"
)
parser.add_argument(
"--transport",
choices=["http", "sse"],
default="http",
help="Transport type (default: http)"
)
parser.add_argument(
"--log-level",
choices=["debug", "info", "warning", "error"],
default=os.getenv("LOG_LEVEL", "info").lower(),
help="Log level (default: info)"
)
parser.add_argument(
"--workers",
type=int,
default=1,
help="Number of worker processes (default: 1)"
)
args = parser.parse_args()
# Select app based on transport
app_name = "asgi_app:app" if args.transport == "http" else "asgi_app:sse_app"
# Configure uvicorn
config = {
"app": app_name,
"host": args.host,
"port": args.port,
"log_level": args.log_level,
"reload": args.reload,
"access_log": True,
}
# Add workers only if not in reload mode
if not args.reload and args.workers > 1:
config["workers"] = args.workers
# Print startup information
print(f"Starting Yargı MCP server...")
print(f"Host: {args.host}")
print(f"Port: {args.port}")
print(f"Transport: {args.transport}")
print(f"Log level: {args.log_level}")
if args.reload:
print("Auto-reload: enabled")
else:
print(f"Workers: {args.workers}")
print(f"\nServer will be available at: http://{args.host}:{args.port}")
print(f"MCP endpoint: http://{args.host}:{args.port}/mcp/")
print(f"Health check: http://{args.host}:{args.port}/health")
print(f"API status: http://{args.host}:{args.port}/status")
print("\nPress CTRL+C to stop the server\n")
# Run uvicorn
try:
uvicorn.run(**config)
except KeyboardInterrupt:
print("\nShutting down server...")
sys.exit(0)
if __name__ == "__main__":
main()
+31 -4
View File
@@ -1,5 +1,6 @@
# sayistay_mcp_module/client.py
import asyncio
import httpx
import re
from bs4 import BeautifulSoup
@@ -42,11 +43,17 @@ class SayistayApiClient:
"""
BASE_URL = "https://www.sayistay.gov.tr"
# Search endpoints for each decision type
GENEL_KURUL_ENDPOINT = "/KararlarGenelKurul/DataTablesList"
TEMYIZ_KURULU_ENDPOINT = "/KararlarTemyiz/DataTablesList"
TEMYIZ_KURULU_ENDPOINT = "/KararlarTemyiz/DataTablesList"
DAIRE_ENDPOINT = "/KararlarDaire/DataTablesList"
# Marker present in the upstream WAF block page (also returns HTTP 418).
# Verified 2026-05-03 against real Chrome — the block targets POSTs to
# the DataTablesList endpoints regardless of headers/cookies/CSRF, so
# we surface a specific error instead of the generic "I'm a teapot".
_WAF_BLOCK_MARKER = "Bilgi Güvenliği Politikaları Gereği Kısıtlanmıştır"
# Page endpoints for session initialization and document access
GENEL_KURUL_PAGE = "/KararlarGenelKurul"
@@ -141,6 +148,23 @@ class SayistayApiClient:
return enum_value
def _raise_if_waf_blocked(self, response: httpx.Response, endpoint_label: str) -> None:
"""
Sayıştay's upstream WAF returns HTTP 418 with a Turkish HTML block
page for POSTs to the DataTablesList endpoints. This affects every
client (verified with real Chrome on 2026-05-03), so there is no
client-side workaround. Detect it and raise a clear error.
"""
if response.status_code == 418 or self._WAF_BLOCK_MARKER in response.text:
raise RuntimeError(
f"Sayıştay upstream WAF blocked the {endpoint_label} request "
f"(HTTP {response.status_code} from {response.request.url}). "
"This is a server-side restriction at sayistay.gov.tr — affects "
"all clients including a real browser — and cannot be worked "
"around from yargi-mcp. Try again later or contact Sayıştay if "
"the block persists."
)
def _build_datatables_params(self, start: int, length: int, draw: int = 1) -> List[Tuple[str, str]]:
"""Build standard DataTables parameters for all endpoints."""
params = [
@@ -384,9 +408,10 @@ class SayistayApiClient:
data=encoded_data,
headers=headers
)
self._raise_if_waf_blocked(response, "Genel Kurul")
response.raise_for_status()
response_json = response.json()
# Parse response
decisions = []
for item in response_json.get('data', []):
@@ -443,6 +468,7 @@ class SayistayApiClient:
data=encoded_data,
headers=headers
)
self._raise_if_waf_blocked(response, "Temyiz Kurulu")
response.raise_for_status()
response_json = response.json()
@@ -502,6 +528,7 @@ class SayistayApiClient:
data=encoded_data,
headers=headers
)
self._raise_if_waf_blocked(response, "Daire")
response.raise_for_status()
response_json = response.json()
@@ -631,7 +658,7 @@ class SayistayApiClient:
)
# Convert HTML to Markdown using existing method
markdown_content = self._convert_html_to_markdown(html_content)
markdown_content = await asyncio.to_thread(self._convert_html_to_markdown, html_content)
if markdown_content and "Error converting HTML content" not in markdown_content:
logger.info(f"Successfully retrieved and converted document {decision_id} to Markdown")
+18 -2
View File
@@ -1,7 +1,23 @@
# semantic_search/__init__.py
from .embedder import OpenRouterEmbedder, is_openrouter_available
from .embedder import (
OpenRouterEmbedder,
LocalEmbedder,
get_embedder,
is_openrouter_available,
is_local_embedding_configured,
is_semantic_search_available,
)
from .vector_store import VectorStore
from .processor import DocumentProcessor
__all__ = ['OpenRouterEmbedder', 'is_openrouter_available', 'VectorStore', 'DocumentProcessor']
__all__ = [
'OpenRouterEmbedder',
'LocalEmbedder',
'get_embedder',
'is_openrouter_available',
'is_local_embedding_configured',
'is_semantic_search_available',
'VectorStore',
'DocumentProcessor',
]
+247 -53
View File
@@ -2,72 +2,129 @@
import logging
import os
from typing import List, Optional
from typing import Dict, List, Optional
import numpy as np
logger = logging.getLogger(__name__)
# OpenRouter defaults (preserve backward compatibility)
DEFAULT_MODEL = "google/gemini-embedding-001"
DEFAULT_DIMENSION = 3072
# Local provider defaults — Ollama with nomic-embed-text out of the box.
# Override via LOCAL_EMBEDDING_BASE_URL / LOCAL_EMBEDDING_MODEL /
# LOCAL_EMBEDDING_DIMENSION when using a different server or model.
# For Turkish, intfloat/multilingual-e5-large (1024 dims, prompt_style=e5)
# served via HuggingFace TEI is the recommended setup — see README.
LOCAL_DEFAULT_BASE_URL = "http://localhost:11434/v1"
LOCAL_DEFAULT_MODEL = "nomic-embed-text"
LOCAL_DEFAULT_DIMENSION = 768
# Prompt-template styles. Embedding models are trained with specific
# prefixes — using the wrong style silently degrades retrieval quality.
# - "gemini": "task: {task} | query: {text}" / "title: {title} | text: {text}"
# (matches google/gemini-embedding-001, the OpenRouter default)
# - "e5": "query: {text}" / "passage: {text}"
# (matches intfloat/multilingual-e5-* models — best for Turkish)
# - "raw": no prefix; pass text through as-is
PROMPT_STYLES = ("gemini", "e5", "raw")
DEFAULT_PROMPT_STYLE = "gemini"
def _format_query(prompt_style: str, query: str, task: str) -> str:
if prompt_style == "e5":
return f"query: {query}"
if prompt_style == "raw":
return query
# gemini (default)
return f"task: {task} | query: {query}"
def _format_document(prompt_style: str, doc: str, title: str) -> str:
if prompt_style == "e5":
return f"passage: {doc}"
if prompt_style == "raw":
return doc
# gemini (default)
return f"title: {title} | text: {doc}"
def _resolve_prompt_style(explicit: Optional[str], default: str) -> str:
style = (explicit or os.getenv("EMBEDDING_PROMPT_STYLE") or default).strip().lower()
if style not in PROMPT_STYLES:
raise ValueError(
f"Unknown EMBEDDING_PROMPT_STYLE {style!r}; expected one of {PROMPT_STYLES}"
)
return style
def is_openrouter_available() -> bool:
"""Check if OpenRouter API key is available."""
return bool(os.getenv("OPENROUTER_API_KEY"))
class OpenRouterEmbedder:
def is_local_embedding_configured() -> bool:
"""Check if the user opted into a local embedding endpoint."""
return os.getenv("EMBEDDING_PROVIDER", "").strip().lower() == "local"
def is_semantic_search_available() -> bool:
"""Returns True if any embedding provider is configured."""
return is_local_embedding_configured() or is_openrouter_available()
def _coerce_dimension(value, env_name: str, default: int) -> int:
"""Parse a dimension value (int or str) with clear error messages."""
if value is None:
return default
try:
parsed = int(value)
except (TypeError, ValueError) as e:
raise ValueError(
f"{env_name} must be an integer, got {value!r}"
) from e
if parsed <= 0:
raise ValueError(f"Embedding dimension must be positive, got {parsed}")
return parsed
class _BaseOpenAICompatibleEmbedder:
"""
Embedder using OpenRouter API with Google's Gemini Embedding model.
Requires OPENROUTER_API_KEY environment variable.
Shared encode/similarity logic for embedders backed by the OpenAI Python
SDK. Subclasses configure ``client``, ``model``, ``dimension``, and
optionally ``_extra_headers`` (e.g. OpenRouter ranking headers).
"""
def __init__(self):
"""
Initialize OpenRouter Embedder.
# Subclasses may override; sent on every embeddings.create call when set.
_extra_headers: Dict[str, str] = {}
Raises:
ValueError: If OPENROUTER_API_KEY is not set
ImportError: If openai package is not installed
"""
api_key = os.getenv("OPENROUTER_API_KEY")
if not api_key:
raise ValueError("OPENROUTER_API_KEY environment variable is not set")
try:
from openai import OpenAI
except ImportError:
raise ImportError("openai package is required. Install with: pip install openai")
self.client = OpenAI(
base_url="https://openrouter.ai/api/v1",
api_key=api_key,
)
self.model = "google/gemini-embedding-001"
self.dimension = 3072
logger.info(f"OpenRouter Embedder initialized with model: {self.model}")
# Set by subclasses
client = None
model: str = ""
dimension: int = 0
prompt_style: str = DEFAULT_PROMPT_STYLE
def encode_query(self, query: str, task: str = "search result") -> np.ndarray:
"""
Encode a search query.
Encode a search query. Prefix is selected by ``self.prompt_style``.
Args:
query: The search query text
task: Task type for prompt template
task: Task hint used by the gemini-style prefix; ignored for
e5/raw styles.
Returns:
Numpy array of embeddings (3072 dimensions)
Numpy array of embeddings (``self.dimension`` elements).
"""
# Apply query prompt template
text = f"task: {task} | query: {query}"
text = _format_query(self.prompt_style, query, task)
try:
response = self.client.embeddings.create(
model=self.model,
input=text,
encoding_format="float",
extra_headers={
"HTTP-Referer": "https://yargimcp.com",
"X-Title": "Yargi MCP Server",
}
extra_headers=self._extra_headers or None,
)
embedding = np.array(response.data[0].embedding, dtype=np.float32)
@@ -86,40 +143,34 @@ class OpenRouterEmbedder:
def encode_documents(self, documents: List[str], titles: Optional[List[str]] = None) -> np.ndarray:
"""
Encode multiple documents with batch API call.
Encode multiple documents with a batch API call.
Args:
documents: List of document texts
titles: Optional list of document titles
Returns:
Numpy array of embeddings (N x 3072 dimensions)
Numpy array of embeddings (N x ``self.dimension``).
"""
if not documents:
return np.array([])
# Apply document prompt template
texts = []
for i, doc in enumerate(documents):
title = titles[i] if titles and i < len(titles) else "none"
text = f"title: {title} | text: {doc}"
texts.append(text)
texts.append(_format_document(self.prompt_style, doc, title))
try:
response = self.client.embeddings.create(
model=self.model,
input=texts,
encoding_format="float",
extra_headers={
"HTTP-Referer": "https://yargimcp.com",
"X-Title": "Yargi MCP Server",
}
extra_headers=self._extra_headers or None,
)
# Extract embeddings in order
embeddings = np.array(
[d.embedding for d in sorted(response.data, key=lambda x: x.index)],
dtype=np.float32
dtype=np.float32,
)
# L2 normalize each embedding for cosine similarity
@@ -138,17 +189,160 @@ class OpenRouterEmbedder:
Compute cosine similarity between query and documents.
Args:
query_embedding: Query embedding (3072,)
document_embeddings: Document embeddings (N x 3072)
query_embedding: Query embedding (``self.dimension``,)
document_embeddings: Document embeddings (N x ``self.dimension``)
Returns:
Similarity scores (N,)
"""
# Ensure query is 2D for matrix multiplication
if len(query_embedding.shape) == 1:
query_embedding = query_embedding.reshape(1, -1)
# Compute cosine similarity (embeddings are already normalized)
# Embeddings are already L2-normalized.
similarities = np.dot(document_embeddings, query_embedding.T).squeeze()
return similarities
class OpenRouterEmbedder(_BaseOpenAICompatibleEmbedder):
"""
Embedder using OpenRouter's embedding API.
The model and dimension are configurable so users can pick any OpenRouter
embedding model (e.g. when one becomes paid). Configuration precedence:
explicit constructor args > environment variables > defaults.
Environment variables:
OPENROUTER_API_KEY (required): OpenRouter credential
OPENROUTER_EMBEDDING_MODEL (optional): override the embedding model id
OPENROUTER_EMBEDDING_DIMENSION (optional): override the vector size
Defaults preserve backward compatibility: ``google/gemini-embedding-001``
at 3072 dimensions.
"""
_extra_headers = {
"HTTP-Referer": "https://yargimcp.com",
"X-Title": "Yargi MCP Server",
}
def __init__(
self,
model: Optional[str] = None,
dimension: Optional[int] = None,
prompt_style: Optional[str] = None,
):
api_key = os.getenv("OPENROUTER_API_KEY")
if not api_key:
raise ValueError("OPENROUTER_API_KEY environment variable is not set")
try:
from openai import OpenAI
except ImportError:
raise ImportError("openai package is required. Install with: pip install openai")
self.client = OpenAI(
base_url="https://openrouter.ai/api/v1",
api_key=api_key,
)
self.model = model or os.getenv("OPENROUTER_EMBEDDING_MODEL") or DEFAULT_MODEL
self.dimension = _coerce_dimension(
dimension if dimension is not None else os.getenv("OPENROUTER_EMBEDDING_DIMENSION"),
"OPENROUTER_EMBEDDING_DIMENSION",
DEFAULT_DIMENSION,
)
# Default to gemini-style prefix for OpenRouter — matches the default
# google/gemini-embedding-001 model. Override via constructor or
# EMBEDDING_PROMPT_STYLE env var when picking a different model.
self.prompt_style = _resolve_prompt_style(prompt_style, "gemini")
logger.info(
f"OpenRouter Embedder initialized with model: {self.model} "
f"(dimension={self.dimension}, prompt_style={self.prompt_style})"
)
class LocalEmbedder(_BaseOpenAICompatibleEmbedder):
"""
Embedder for a local OpenAI-compatible embedding server Ollama,
llama.cpp, vLLM, LM Studio, etc. Zero new Python dependencies; just
point the existing OpenAI SDK at a local base URL.
Environment variables:
EMBEDDING_PROVIDER=local (selects this provider)
LOCAL_EMBEDDING_BASE_URL (default: http://localhost:11434/v1)
LOCAL_EMBEDDING_MODEL (default: nomic-embed-text)
LOCAL_EMBEDDING_DIMENSION (default: 768)
LOCAL_EMBEDDING_API_KEY (optional; ignored by most local servers)
Setup (Ollama):
$ ollama serve
$ ollama pull nomic-embed-text # or bge-m3 for better Turkish
The dimension MUST match the model's actual output size (e.g. 768 for
nomic-embed-text, 1024 for bge-m3, 1024 for mxbai-embed-large).
"""
def __init__(
self,
base_url: Optional[str] = None,
model: Optional[str] = None,
dimension: Optional[int] = None,
api_key: Optional[str] = None,
prompt_style: Optional[str] = None,
):
try:
from openai import OpenAI
except ImportError:
raise ImportError("openai package is required. Install with: pip install openai")
self.base_url = (
base_url
or os.getenv("LOCAL_EMBEDDING_BASE_URL")
or LOCAL_DEFAULT_BASE_URL
)
# Most local servers don't validate the key — use a placeholder so
# the OpenAI SDK doesn't error on the missing-key check.
effective_key = (
api_key
or os.getenv("LOCAL_EMBEDDING_API_KEY")
or "no-key-needed"
)
self.client = OpenAI(base_url=self.base_url, api_key=effective_key)
self.model = model or os.getenv("LOCAL_EMBEDDING_MODEL") or LOCAL_DEFAULT_MODEL
self.dimension = _coerce_dimension(
dimension if dimension is not None else os.getenv("LOCAL_EMBEDDING_DIMENSION"),
"LOCAL_EMBEDDING_DIMENSION",
LOCAL_DEFAULT_DIMENSION,
)
# Default to e5 prefix for local — the recommended Turkish setup
# (multilingual-e5-large). Override via EMBEDDING_PROMPT_STYLE when
# using a different model family (e.g. nomic, bge).
self.prompt_style = _resolve_prompt_style(prompt_style, "e5")
logger.info(
f"Local Embedder initialized: model={self.model} "
f"base_url={self.base_url} dimension={self.dimension} "
f"prompt_style={self.prompt_style}"
)
def get_embedder():
"""
Factory that picks the embedder based on EMBEDDING_PROVIDER.
- ``EMBEDDING_PROVIDER=local`` -> ``LocalEmbedder``
- otherwise -> ``OpenRouterEmbedder`` (requires OPENROUTER_API_KEY)
Raises:
ValueError: If no provider is configured (neither local nor OpenRouter).
"""
if is_local_embedding_configured():
return LocalEmbedder()
if is_openrouter_available():
return OpenRouterEmbedder()
raise ValueError(
"No embedding provider configured. Set OPENROUTER_API_KEY for hosted "
"embeddings, or EMBEDDING_PROVIDER=local (with LOCAL_EMBEDDING_* "
"env vars) for a local OpenAI-compatible server like Ollama."
)
+21
View File
@@ -0,0 +1,21 @@
# sigorta_tahkim_mcp_module/__init__.py
from .client import SigortaTahkimApiClient
from .models import (
SigortaTahkimSearchRequest,
SigortaTahkimDecisionSummary,
SigortaTahkimSearchResult,
SigortaTahkimDocumentMarkdown,
SigortaTahkimSearchWithinMatch,
SigortaTahkimSearchWithinResult
)
__all__ = [
"SigortaTahkimApiClient",
"SigortaTahkimSearchRequest",
"SigortaTahkimDecisionSummary",
"SigortaTahkimSearchResult",
"SigortaTahkimDocumentMarkdown",
"SigortaTahkimSearchWithinMatch",
"SigortaTahkimSearchWithinResult"
]
+345
View File
@@ -0,0 +1,345 @@
# sigorta_tahkim_mcp_module/client.py
import asyncio
import httpx
from typing import Optional
import logging
import os
import re
import io
import math
from markitdown import MarkItDown
from .models import (
SigortaTahkimSearchRequest,
SigortaTahkimDecisionSummary,
SigortaTahkimSearchResult,
SigortaTahkimDocumentMarkdown,
SigortaTahkimSearchWithinMatch,
SigortaTahkimSearchWithinResult
)
logger = logging.getLogger(__name__)
if not logger.hasHandlers():
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
)
# Turkish-specific lowercase: İ→i, I→ı (Python's str.lower() doesn't handle these)
_TR_UPPER = str.maketrans("İIÇĞÖŞÜ", "iıçğöşü")
def _turkish_lower(text: str) -> str:
"""Lowercase with Turkish İ/I handling."""
return text.translate(_TR_UPPER).lower()
class SigortaTahkimApiClient:
"""
API client for searching and retrieving Sigorta Tahkim Komisyonu
(Insurance Arbitration Commission) decisions using Tavily Search API
for discovery and direct PDF download for content retrieval.
The commission publishes quarterly PDF journals ("Hakem Karar Dergisi")
containing arbitration decisions. There are 64 issues spanning 2010-2025.
"""
TAVILY_API_URL = "https://api.tavily.com/search"
BASE_URL = "https://www.sigortatahkim.org"
PDF_BASE_URL = "https://www.sigortatahkim.org/content/CmsFiles/"
DOCUMENT_MARKDOWN_CHUNK_SIZE = 5000
def __init__(self, request_timeout: float = 60.0):
"""Initialize the Sigorta Tahkim API client."""
self.tavily_api_key = os.getenv("TAVILY_API_KEY")
if not self.tavily_api_key:
self.tavily_api_key = "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"
logger.info("Using fallback Tavily API token (development token)")
else:
logger.info("Using Tavily API key from environment variable")
self.http_client = httpx.AsyncClient(
headers={
"User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36"
},
timeout=httpx.Timeout(request_timeout)
)
self.markitdown = MarkItDown()
async def close_client_session(self):
"""Close the HTTP client session."""
await self.http_client.aclose()
logger.info("SigortaTahkimApiClient: HTTP client session closed.")
def _get_pdf_filename(self, issue_number: int) -> str:
"""Get the PDF filename for a given journal issue number."""
if issue_number == 4:
return "karardergisisayi4.pdf"
elif 57 <= issue_number <= 61:
return f"revizekd{issue_number}.pdf"
else:
return f"karardrgs{issue_number}.pdf"
def _extract_issue_number(self, url: str) -> Optional[str]:
"""Extract journal issue number from a sigortatahkim.org URL."""
# Pattern: karardrgs{N}.pdf
match = re.search(r'karardrgs(\d+)\.pdf', url, re.IGNORECASE)
if match:
return match.group(1)
# Pattern: revizekd{N}.pdf
match = re.search(r'revizekd(\d+)\.pdf', url, re.IGNORECASE)
if match:
return match.group(1)
# Pattern: karardergisisayi{N}.pdf
match = re.search(r'karardergisisayi(\d+)\.pdf', url, re.IGNORECASE)
if match:
return match.group(1)
# Pattern: sayı or sayi in URL path with number
match = re.search(r'say[ıi]\s*[-:]?\s*(\d+)', url, re.IGNORECASE)
if match:
return match.group(1)
return None
async def search_decisions(
self,
request: SigortaTahkimSearchRequest
) -> SigortaTahkimSearchResult:
"""
Search for Sigorta Tahkim Komisyonu decisions using Tavily API.
Args:
request: Search request parameters
Returns:
SigortaTahkimSearchResult with matching decisions
"""
try:
headers = {
"Content-Type": "application/json",
"Authorization": f"Bearer {self.tavily_api_key}"
}
payload = {
"query": request.keywords,
"country": "turkey",
"include_domains": ["sigortatahkim.org"],
"max_results": request.pageSize,
"search_depth": "advanced"
}
if request.page > 1:
logger.warning(f"Tavily API doesn't support pagination. Page {request.page} requested.")
response = await self.http_client.post(
self.TAVILY_API_URL,
json=payload,
headers=headers
)
response.raise_for_status()
data = response.json()
logger.info(f"Tavily returned {len(data.get('results', []))} results for Sigorta Tahkim")
decisions = []
for result in data.get("results", []):
url = result.get("url", "")
title = result.get("title", "").strip()
content = result.get("content", "")[:500]
issue_num = self._extract_issue_number(url)
doc_id = issue_num if issue_num else url
decision = SigortaTahkimDecisionSummary(
title=title,
document_id=doc_id,
content=content,
url=url
)
decisions.append(decision)
return SigortaTahkimSearchResult(
decisions=decisions,
total_results=len(data.get("results", [])),
page=request.page,
pageSize=request.pageSize
)
except httpx.HTTPStatusError as e:
logger.error(f"HTTP error searching Sigorta Tahkim decisions: {e}")
if e.response.status_code == 401:
raise Exception("Tavily API authentication failed. Check API key.")
raise Exception(f"Failed to search Sigorta Tahkim decisions: {str(e)}")
except Exception as e:
logger.error(f"Error searching Sigorta Tahkim decisions: {e}")
raise Exception(f"Failed to search Sigorta Tahkim decisions: {str(e)}")
# Regex pattern to split decisions within a journal issue
DECISION_HEADER_PATTERN = re.compile(
r'(\d{2}\.\d{2}\.\d{4}\s+Tarih\s+ve\s+K-\d{4}/\d+\s+Sayılı\s+Hakem\s+Kararı)'
)
# Minimum body length to distinguish real decisions from TOC entries
MIN_DECISION_BODY_LENGTH = 1000
async def _download_and_convert_pdf(self, issue_number: str) -> tuple[str, str]:
"""
Download a journal issue PDF and convert to markdown.
Returns:
Tuple of (markdown_content, pdf_url)
"""
issue_num = int(issue_number)
filename = self._get_pdf_filename(issue_num)
pdf_url = f"{self.PDF_BASE_URL}{filename}"
logger.info(f"Downloading Sigorta Tahkim PDF: {pdf_url}")
response = await self.http_client.get(pdf_url, follow_redirects=True)
response.raise_for_status()
pdf_stream = io.BytesIO(response.content)
# markitdown is sync; offload to thread so PDF parsing doesn't block
# the event-loop / other in-flight MCP requests.
result = await asyncio.to_thread(
self.markitdown.convert_stream, pdf_stream, file_extension=".pdf"
)
return result.text_content.strip(), pdf_url
def _split_into_decisions(self, markdown_content: str) -> list[tuple[str, str]]:
"""
Split markdown content into individual decisions.
Returns:
List of (header, body) tuples for decisions with substantial content.
"""
parts = self.DECISION_HEADER_PATTERN.split(markdown_content)
decisions = []
for i in range(1, len(parts) - 1, 2):
header = parts[i].strip()
body = parts[i + 1].strip() if i + 1 < len(parts) else ""
if len(body) >= self.MIN_DECISION_BODY_LENGTH:
decisions.append((header, body))
return decisions
async def get_document_markdown(
self,
issue_number: str,
page_number: int = 1
) -> SigortaTahkimDocumentMarkdown:
"""
Retrieve a Sigorta Tahkim journal issue PDF and convert to Markdown.
Args:
issue_number: Journal issue number (e.g., '64')
page_number: Page number for paginated content (1-indexed)
Returns:
SigortaTahkimDocumentMarkdown with paginated content
"""
try:
markdown_content, pdf_url = await self._download_and_convert_pdf(issue_number)
total_length = len(markdown_content)
total_pages = max(1, math.ceil(total_length / self.DOCUMENT_MARKDOWN_CHUNK_SIZE))
start_idx = (page_number - 1) * self.DOCUMENT_MARKDOWN_CHUNK_SIZE
end_idx = start_idx + self.DOCUMENT_MARKDOWN_CHUNK_SIZE
page_content = markdown_content[start_idx:end_idx]
return SigortaTahkimDocumentMarkdown(
document_id=issue_number,
markdown_content=page_content,
page_number=page_number,
total_pages=total_pages,
source_url=pdf_url
)
except ValueError:
raise Exception(f"Invalid issue number: {issue_number}. Must be a number (e.g., '64').")
except httpx.HTTPStatusError as e:
logger.error(f"HTTP error fetching Sigorta Tahkim issue {issue_number}: {e}")
raise Exception(f"Failed to fetch journal issue {issue_number}: {str(e)}")
except Exception as e:
logger.error(f"Error processing Sigorta Tahkim issue {issue_number}: {e}")
raise Exception(f"Failed to process journal issue {issue_number}: {str(e)}")
async def search_within_issue(
self,
issue_number: str,
keyword: str,
max_results: int = 10
) -> SigortaTahkimSearchWithinResult:
"""
Search for a keyword within a specific journal issue's decisions.
Downloads the PDF, splits into individual decisions, and returns
matching decisions sorted by relevance (match count).
Args:
issue_number: Journal issue number (e.g., '64')
keyword: Search keyword or phrase in Turkish
max_results: Maximum matching decisions to return
Returns:
SigortaTahkimSearchWithinResult with matching decisions
"""
try:
markdown_content, _ = await self._download_and_convert_pdf(issue_number)
decisions = self._split_into_decisions(markdown_content)
logger.info(
f"Searching '{keyword}' within issue {issue_number}: "
f"{len(decisions)} decisions found"
)
keyword_lower = _turkish_lower(keyword)
matches = []
for header, body in decisions:
body_lower = _turkish_lower(body)
count = body_lower.count(keyword_lower)
if count == 0:
continue
# Extract excerpt around the first match
first_pos = body_lower.find(keyword_lower)
excerpt_start = max(0, first_pos - 200)
excerpt_end = min(len(body), first_pos + len(keyword) + 200)
excerpt = body[excerpt_start:excerpt_end].strip()
if excerpt_start > 0:
excerpt = "..." + excerpt
if excerpt_end < len(body):
excerpt = excerpt + "..."
matches.append(SigortaTahkimSearchWithinMatch(
decision_header=header,
relevance_score=count,
excerpt=excerpt,
body_length=len(body)
))
# Sort by relevance (highest match count first)
matches.sort(key=lambda m: m.relevance_score, reverse=True)
matches = matches[:max_results]
return SigortaTahkimSearchWithinResult(
issue_number=issue_number,
keyword=keyword,
total_decisions=len(decisions),
matching_decisions=len(matches),
matches=matches
)
except ValueError:
raise Exception(f"Invalid issue number: {issue_number}. Must be a number (e.g., '64').")
except httpx.HTTPStatusError as e:
logger.error(f"HTTP error in search_within issue {issue_number}: {e}")
raise Exception(f"Failed to fetch journal issue {issue_number}: {str(e)}")
except Exception as e:
logger.error(f"Error in search_within issue {issue_number}: {e}")
raise Exception(f"Failed to search within issue {issue_number}: {str(e)}")
+59
View File
@@ -0,0 +1,59 @@
# sigorta_tahkim_mcp_module/models.py
from pydantic import BaseModel, Field
from typing import List
class SigortaTahkimSearchRequest(BaseModel):
"""Request model for searching Sigorta Tahkim Komisyonu decisions via Tavily API."""
keywords: str = Field(..., description="Search keywords in Turkish")
page: int = Field(1, ge=1, description="Page number (1-indexed)")
pageSize: int = Field(10, ge=1, le=50, description="Results per page (1-50)")
class SigortaTahkimDecisionSummary(BaseModel):
"""Summary of a Sigorta Tahkim decision from search results."""
title: str = Field(..., description="Decision title or journal issue info")
document_id: str = Field(..., description="Journal issue number (e.g., '64')")
content: str = Field(..., description="Decision summary/excerpt")
url: str = Field("", description="Source URL")
class SigortaTahkimSearchResult(BaseModel):
"""Response model for Sigorta Tahkim decision search results."""
decisions: List[SigortaTahkimDecisionSummary] = Field(
default_factory=list,
description="List of matching decisions"
)
total_results: int = Field(0, description="Total number of results")
page: int = Field(1, description="Current page number")
pageSize: int = Field(10, description="Results per page")
class SigortaTahkimDocumentMarkdown(BaseModel):
"""Sigorta Tahkim journal issue converted to Markdown format."""
document_id: str = Field(..., description="Journal issue number")
markdown_content: str = Field("", description="Document content in Markdown")
page_number: int = Field(1, description="Current page number")
total_pages: int = Field(1, description="Total number of pages")
source_url: str = Field("", description="PDF source URL")
class SigortaTahkimSearchWithinMatch(BaseModel):
"""A single matching decision from search within a journal issue."""
decision_header: str = Field(..., description="Decision header (date and K-number)")
relevance_score: int = Field(0, description="Number of keyword matches")
excerpt: str = Field("", description="Matching excerpt with context")
body_length: int = Field(0, description="Full decision body length in chars")
class SigortaTahkimSearchWithinResult(BaseModel):
"""Response model for search within a journal issue."""
issue_number: str = Field(..., description="Journal issue number searched")
keyword: str = Field("", description="Search keyword used")
total_decisions: int = Field(0, description="Total decisions in issue")
matching_decisions: int = Field(0, description="Number of matching decisions")
matches: List[SigortaTahkimSearchWithinMatch] = Field(
default_factory=list,
description="List of matching decisions sorted by relevance"
)
-159
View File
@@ -1,159 +0,0 @@
"""
Starlette integration example for Yargı MCP Server
This module demonstrates how to integrate the Yargı MCP server
with a Starlette application, including authentication middleware
and custom routing.
Usage:
uvicorn starlette_app:app --host 0.0.0.0 --port 8000
"""
import os
from starlette.applications import Starlette
from starlette.routing import Mount, Route
from starlette.requests import Request
from starlette.responses import JSONResponse, PlainTextResponse, RedirectResponse
from starlette.middleware import Middleware
from starlette.middleware.cors import CORSMiddleware
from starlette.middleware.authentication import AuthenticationMiddleware
from starlette.authentication import (
AuthenticationBackend, AuthCredentials, SimpleUser, AuthenticationError
)
# Import the main MCP app
from mcp_server_main import app as mcp_server
# Simple token authentication backend
class TokenAuthBackend(AuthenticationBackend):
async def authenticate(self, request):
auth_header = request.headers.get("Authorization")
expected_token = os.getenv("API_TOKEN")
# Skip auth for health check and public endpoints
if request.url.path in ["/health", "/", "/login"]:
return None
if not expected_token:
# No token configured, allow all
return AuthCredentials(["authenticated"]), SimpleUser("anonymous")
if not auth_header:
raise AuthenticationError("Authorization header required")
try:
scheme, token = auth_header.split()
if scheme.lower() != "bearer":
raise AuthenticationError("Invalid authentication scheme")
if token != expected_token:
raise AuthenticationError("Invalid token")
return AuthCredentials(["authenticated"]), SimpleUser("user")
except ValueError:
raise AuthenticationError("Invalid authorization header format")
# Homepage
async def homepage(request: Request):
return JSONResponse({
"service": "Yargı MCP Server",
"version": "0.1.0",
"endpoints": {
"mcp": "/mcp-server/mcp/",
"api": "/api/",
"health": "/health"
}
})
# API info endpoint
async def api_info(request: Request):
if not request.user.is_authenticated:
return JSONResponse({"error": "Authentication required"}, status_code=401)
return JSONResponse({
"authenticated_as": request.user.display_name,
"available_tools": len(mcp_server._tool_manager._tools),
"databases": [
"Yargıtay", "Danıştay", "Emsal", "Uyuşmazlık",
"Anayasa", "KIK", "Rekabet", "Bedesten"
]
})
# Health check
async def health_check(request: Request):
return JSONResponse({
"status": "healthy",
"service": "Yargı MCP Server"
})
# Login example (returns token for demo)
async def login(request: Request):
token = os.getenv("API_TOKEN", "demo-token")
return JSONResponse({
"message": "Use this token in Authorization header",
"example": f"Authorization: Bearer {token}",
"note": "Set API_TOKEN environment variable to change token"
})
# Create MCP ASGI app
mcp_app = mcp_server.http_app(path='/mcp')
# Configure middleware
middleware = [
Middleware(
CORSMiddleware,
allow_origins=os.getenv("ALLOWED_ORIGINS", "*").split(","),
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
),
Middleware(AuthenticationMiddleware, backend=TokenAuthBackend()),
]
# Create routes
routes = [
Route("/", homepage),
Route("/health", health_check),
Route("/login", login),
Route("/api/info", api_info),
Mount("/mcp-server", app=mcp_app),
]
# Create Starlette app
app = Starlette(
routes=routes,
middleware=middleware,
lifespan=mcp_app.lifespan
)
# Nested mount example
def create_nested_app():
"""Example of nested mounting for complex routing structures"""
# Create inner app with MCP
inner_app = Starlette(
routes=[Mount("/services", app=mcp_app)],
middleware=middleware
)
# Create outer app
outer_app = Starlette(
routes=[
Route("/", homepage),
Mount("/v1", app=inner_app),
],
lifespan=mcp_app.lifespan
)
# MCP would be available at /v1/services/mcp/
return outer_app
# Export both apps
nested_app = create_nested_app()
if __name__ == "__main__":
import uvicorn
print("Starting Starlette app with authentication...")
print("Set API_TOKEN environment variable to enable authentication")
print("Example: API_TOKEN=secret-token python starlette_app.py")
uvicorn.run(app, host="0.0.0.0", port=8000)
-25
View File
@@ -1,25 +0,0 @@
import os, stripe
from clerk_backend_api import Clerk # Clerk backend SDK
from fastapi import APIRouter, Request, HTTPException
router = APIRouter()
stripe.api_key = os.getenv("STRIPE_SECRET")
clerk = Clerk(bearer_auth=os.getenv("CLERK_SECRET_KEY"))
@router.post("/stripe/webhook")
async def stripe_hook(req: Request):
payload, sig = await req.body(), req.headers["stripe-signature"]
try:
event = stripe.Webhook.construct_event( # Stripe-recommended verify
payload, sig, os.getenv("STRIPE_WEBHOOK_SECRET"))
except stripe.error.SignatureVerificationError:
raise HTTPException(400, "Bad sig")
if event["type"] == "customer.subscription.updated":
item = event["data"]["object"]["items"]["data"][0]
plan = item["price"]["nickname"] # "Pro", "Enterprise"…
userID = event["data"]["object"]["metadata"]["clerk_user_id"]
clerk.users.update_user_metadata( # merge into unsafe_metadata
userID, unsafe_metadata={"plan": plan})
return {"ok": True}
Generated
+2 -201
View File
@@ -369,22 +369,6 @@ wheels = [
{ url = "https://files.pythonhosted.org/packages/0a/4c/925909008ed5a988ccbb72dcc897407e5d6d3bd72410d69e051fc0c14647/charset_normalizer-3.4.4-py3-none-any.whl", hash = "sha256:7a32c560861a02ff789ad905a2fe94e3f840803362c84fecf1851cb4cf3dc37f", size = 53402, upload-time = "2025-10-14T04:42:31.76Z" },
]
[[package]]
name = "clerk-backend-api"
version = "4.1.2"
source = { registry = "https://pypi.org/simple" }
dependencies = [
{ name = "cryptography" },
{ name = "httpcore" },
{ name = "httpx" },
{ name = "pydantic" },
{ name = "pyjwt" },
]
sdist = { url = "https://files.pythonhosted.org/packages/4f/c6/0a56ce9e2e6a7ea4cf3b5dc2b03a61c300565347e9f8b72883fe7ddf9316/clerk_backend_api-4.1.2.tar.gz", hash = "sha256:758fa0f05a50e399466efa360b7f1a3df3ad67e513fd57b24d14b491eaeaca22", size = 208867, upload-time = "2025-12-03T13:35:34.513Z" }
wheels = [
{ url = "https://files.pythonhosted.org/packages/88/d8/e938c31ee3a70a428f4d3c17492fd7df2ce2eb870d6eff4d41170a42f8b2/clerk_backend_api-4.1.2-py3-none-any.whl", hash = "sha256:032c7be7bf5b0b220f1b2b1654fee2591309a99f54c9ff029d55ca54a640ad59", size = 424710, upload-time = "2025-12-03T13:35:33.36Z" },
]
[[package]]
name = "click"
version = "8.3.1"
@@ -1813,98 +1797,6 @@ wheels = [
{ url = "https://files.pythonhosted.org/packages/c1/b1/3baf80dc6d2b7bc27a95a67752d0208e410351e3feb4eb78de5f77454d8d/referencing-0.36.2-py3-none-any.whl", hash = "sha256:e8699adbbf8b5c7de96d8ffa0eb5c158b3beafce084968e2ea8bb08c6794dcd0", size = 26775, upload-time = "2025-01-25T08:48:14.241Z" },
]
[[package]]
name = "regex"
version = "2025.11.3"
source = { registry = "https://pypi.org/simple" }
sdist = { url = "https://files.pythonhosted.org/packages/cc/a9/546676f25e573a4cf00fe8e119b78a37b6a8fe2dc95cda877b30889c9c45/regex-2025.11.3.tar.gz", hash = "sha256:1fedc720f9bb2494ce31a58a1631f9c82df6a09b49c19517ea5cc280b4541e01", size = 414669, upload-time = "2025-11-03T21:34:22.089Z" }
wheels = [
{ url = "https://files.pythonhosted.org/packages/f7/90/4fb5056e5f03a7048abd2b11f598d464f0c167de4f2a51aa868c376b8c70/regex-2025.11.3-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:eadade04221641516fa25139273505a1c19f9bf97589a05bc4cfcd8b4a618031", size = 488081, upload-time = "2025-11-03T21:31:11.946Z" },
{ url = "https://files.pythonhosted.org/packages/85/23/63e481293fac8b069d84fba0299b6666df720d875110efd0338406b5d360/regex-2025.11.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:feff9e54ec0dd3833d659257f5c3f5322a12eee58ffa360984b716f8b92983f4", size = 290554, upload-time = "2025-11-03T21:31:13.387Z" },
{ url = "https://files.pythonhosted.org/packages/2b/9d/b101d0262ea293a0066b4522dfb722eb6a8785a8c3e084396a5f2c431a46/regex-2025.11.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:3b30bc921d50365775c09a7ed446359e5c0179e9e2512beec4a60cbcef6ddd50", size = 288407, upload-time = "2025-11-03T21:31:14.809Z" },
{ url = "https://files.pythonhosted.org/packages/0c/64/79241c8209d5b7e00577ec9dca35cd493cc6be35b7d147eda367d6179f6d/regex-2025.11.3-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f99be08cfead2020c7ca6e396c13543baea32343b7a9a5780c462e323bd8872f", size = 793418, upload-time = "2025-11-03T21:31:16.556Z" },
{ url = "https://files.pythonhosted.org/packages/3d/e2/23cd5d3573901ce8f9757c92ca4db4d09600b865919b6d3e7f69f03b1afd/regex-2025.11.3-cp311-cp311-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:6dd329a1b61c0ee95ba95385fb0c07ea0d3fe1a21e1349fa2bec272636217118", size = 860448, upload-time = "2025-11-03T21:31:18.12Z" },
{ url = "https://files.pythonhosted.org/packages/2a/4c/aecf31beeaa416d0ae4ecb852148d38db35391aac19c687b5d56aedf3a8b/regex-2025.11.3-cp311-cp311-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:4c5238d32f3c5269d9e87be0cf096437b7622b6920f5eac4fd202468aaeb34d2", size = 907139, upload-time = "2025-11-03T21:31:20.753Z" },
{ url = "https://files.pythonhosted.org/packages/61/22/b8cb00df7d2b5e0875f60628594d44dba283e951b1ae17c12f99e332cc0a/regex-2025.11.3-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:10483eefbfb0adb18ee9474498c9a32fcf4e594fbca0543bb94c48bac6183e2e", size = 800439, upload-time = "2025-11-03T21:31:22.069Z" },
{ url = "https://files.pythonhosted.org/packages/02/a8/c4b20330a5cdc7a8eb265f9ce593f389a6a88a0c5f280cf4d978f33966bc/regex-2025.11.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:78c2d02bb6e1da0720eedc0bad578049cad3f71050ef8cd065ecc87691bed2b0", size = 782965, upload-time = "2025-11-03T21:31:23.598Z" },
{ url = "https://files.pythonhosted.org/packages/b4/4c/ae3e52988ae74af4b04d2af32fee4e8077f26e51b62ec2d12d246876bea2/regex-2025.11.3-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:e6b49cd2aad93a1790ce9cffb18964f6d3a4b0b3dbdbd5de094b65296fce6e58", size = 854398, upload-time = "2025-11-03T21:31:25.008Z" },
{ url = "https://files.pythonhosted.org/packages/06/d1/a8b9cf45874eda14b2e275157ce3b304c87e10fb38d9fc26a6e14eb18227/regex-2025.11.3-cp311-cp311-musllinux_1_2_s390x.whl", hash = "sha256:885b26aa3ee56433b630502dc3d36ba78d186a00cc535d3806e6bfd9ed3c70ab", size = 845897, upload-time = "2025-11-03T21:31:26.427Z" },
{ url = "https://files.pythonhosted.org/packages/ea/fe/1830eb0236be93d9b145e0bd8ab499f31602fe0999b1f19e99955aa8fe20/regex-2025.11.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:ddd76a9f58e6a00f8772e72cff8ebcff78e022be95edf018766707c730593e1e", size = 788906, upload-time = "2025-11-03T21:31:28.078Z" },
{ url = "https://files.pythonhosted.org/packages/66/47/dc2577c1f95f188c1e13e2e69d8825a5ac582ac709942f8a03af42ed6e93/regex-2025.11.3-cp311-cp311-win32.whl", hash = "sha256:3e816cc9aac1cd3cc9a4ec4d860f06d40f994b5c7b4d03b93345f44e08cc68bf", size = 265812, upload-time = "2025-11-03T21:31:29.72Z" },
{ url = "https://files.pythonhosted.org/packages/50/1e/15f08b2f82a9bbb510621ec9042547b54d11e83cb620643ebb54e4eb7d71/regex-2025.11.3-cp311-cp311-win_amd64.whl", hash = "sha256:087511f5c8b7dfbe3a03f5d5ad0c2a33861b1fc387f21f6f60825a44865a385a", size = 277737, upload-time = "2025-11-03T21:31:31.422Z" },
{ url = "https://files.pythonhosted.org/packages/f4/fc/6500eb39f5f76c5e47a398df82e6b535a5e345f839581012a418b16f9cc3/regex-2025.11.3-cp311-cp311-win_arm64.whl", hash = "sha256:1ff0d190c7f68ae7769cd0313fe45820ba07ffebfddfaa89cc1eb70827ba0ddc", size = 270290, upload-time = "2025-11-03T21:31:33.041Z" },
{ url = "https://files.pythonhosted.org/packages/e8/74/18f04cb53e58e3fb107439699bd8375cf5a835eec81084e0bddbd122e4c2/regex-2025.11.3-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:bc8ab71e2e31b16e40868a40a69007bc305e1109bd4658eb6cad007e0bf67c41", size = 489312, upload-time = "2025-11-03T21:31:34.343Z" },
{ url = "https://files.pythonhosted.org/packages/78/3f/37fcdd0d2b1e78909108a876580485ea37c91e1acf66d3bb8e736348f441/regex-2025.11.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:22b29dda7e1f7062a52359fca6e58e548e28c6686f205e780b02ad8ef710de36", size = 291256, upload-time = "2025-11-03T21:31:35.675Z" },
{ url = "https://files.pythonhosted.org/packages/bf/26/0a575f58eb23b7ebd67a45fccbc02ac030b737b896b7e7a909ffe43ffd6a/regex-2025.11.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:3a91e4a29938bc1a082cc28fdea44be420bf2bebe2665343029723892eb073e1", size = 288921, upload-time = "2025-11-03T21:31:37.07Z" },
{ url = "https://files.pythonhosted.org/packages/ea/98/6a8dff667d1af907150432cf5abc05a17ccd32c72a3615410d5365ac167a/regex-2025.11.3-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:08b884f4226602ad40c5d55f52bf91a9df30f513864e0054bad40c0e9cf1afb7", size = 798568, upload-time = "2025-11-03T21:31:38.784Z" },
{ url = "https://files.pythonhosted.org/packages/64/15/92c1db4fa4e12733dd5a526c2dd2b6edcbfe13257e135fc0f6c57f34c173/regex-2025.11.3-cp312-cp312-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:3e0b11b2b2433d1c39c7c7a30e3f3d0aeeea44c2a8d0bae28f6b95f639927a69", size = 864165, upload-time = "2025-11-03T21:31:40.559Z" },
{ url = "https://files.pythonhosted.org/packages/f9/e7/3ad7da8cdee1ce66c7cd37ab5ab05c463a86ffeb52b1a25fe7bd9293b36c/regex-2025.11.3-cp312-cp312-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:87eb52a81ef58c7ba4d45c3ca74e12aa4b4e77816f72ca25258a85b3ea96cb48", size = 912182, upload-time = "2025-11-03T21:31:42.002Z" },
{ url = "https://files.pythonhosted.org/packages/84/bd/9ce9f629fcb714ffc2c3faf62b6766ecb7a585e1e885eb699bcf130a5209/regex-2025.11.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a12ab1f5c29b4e93db518f5e3872116b7e9b1646c9f9f426f777b50d44a09e8c", size = 803501, upload-time = "2025-11-03T21:31:43.815Z" },
{ url = "https://files.pythonhosted.org/packages/7c/0f/8dc2e4349d8e877283e6edd6c12bdcebc20f03744e86f197ab6e4492bf08/regex-2025.11.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:7521684c8c7c4f6e88e35ec89680ee1aa8358d3f09d27dfbdf62c446f5d4c695", size = 787842, upload-time = "2025-11-03T21:31:45.353Z" },
{ url = "https://files.pythonhosted.org/packages/f9/73/cff02702960bc185164d5619c0c62a2f598a6abff6695d391b096237d4ab/regex-2025.11.3-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:7fe6e5440584e94cc4b3f5f4d98a25e29ca12dccf8873679a635638349831b98", size = 858519, upload-time = "2025-11-03T21:31:46.814Z" },
{ url = "https://files.pythonhosted.org/packages/61/83/0e8d1ae71e15bc1dc36231c90b46ee35f9d52fab2e226b0e039e7ea9c10a/regex-2025.11.3-cp312-cp312-musllinux_1_2_s390x.whl", hash = "sha256:8e026094aa12b43f4fd74576714e987803a315c76edb6b098b9809db5de58f74", size = 850611, upload-time = "2025-11-03T21:31:48.289Z" },
{ url = "https://files.pythonhosted.org/packages/c8/f5/70a5cdd781dcfaa12556f2955bf170cd603cb1c96a1827479f8faea2df97/regex-2025.11.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:435bbad13e57eb5606a68443af62bed3556de2f46deb9f7d4237bc2f1c9fb3a0", size = 789759, upload-time = "2025-11-03T21:31:49.759Z" },
{ url = "https://files.pythonhosted.org/packages/59/9b/7c29be7903c318488983e7d97abcf8ebd3830e4c956c4c540005fcfb0462/regex-2025.11.3-cp312-cp312-win32.whl", hash = "sha256:3839967cf4dc4b985e1570fd8d91078f0c519f30491c60f9ac42a8db039be204", size = 266194, upload-time = "2025-11-03T21:31:51.53Z" },
{ url = "https://files.pythonhosted.org/packages/1a/67/3b92df89f179d7c367be654ab5626ae311cb28f7d5c237b6bb976cd5fbbb/regex-2025.11.3-cp312-cp312-win_amd64.whl", hash = "sha256:e721d1b46e25c481dc5ded6f4b3f66c897c58d2e8cfdf77bbced84339108b0b9", size = 277069, upload-time = "2025-11-03T21:31:53.151Z" },
{ url = "https://files.pythonhosted.org/packages/d7/55/85ba4c066fe5094d35b249c3ce8df0ba623cfd35afb22d6764f23a52a1c5/regex-2025.11.3-cp312-cp312-win_arm64.whl", hash = "sha256:64350685ff08b1d3a6fff33f45a9ca183dc1d58bbfe4981604e70ec9801bbc26", size = 270330, upload-time = "2025-11-03T21:31:54.514Z" },
{ url = "https://files.pythonhosted.org/packages/e1/a7/dda24ebd49da46a197436ad96378f17df30ceb40e52e859fc42cac45b850/regex-2025.11.3-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:c1e448051717a334891f2b9a620fe36776ebf3dd8ec46a0b877c8ae69575feb4", size = 489081, upload-time = "2025-11-03T21:31:55.9Z" },
{ url = "https://files.pythonhosted.org/packages/19/22/af2dc751aacf88089836aa088a1a11c4f21a04707eb1b0478e8e8fb32847/regex-2025.11.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:9b5aca4d5dfd7fbfbfbdaf44850fcc7709a01146a797536a8f84952e940cca76", size = 291123, upload-time = "2025-11-03T21:31:57.758Z" },
{ url = "https://files.pythonhosted.org/packages/a3/88/1a3ea5672f4b0a84802ee9891b86743438e7c04eb0b8f8c4e16a42375327/regex-2025.11.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:04d2765516395cf7dda331a244a3282c0f5ae96075f728629287dfa6f76ba70a", size = 288814, upload-time = "2025-11-03T21:32:01.12Z" },
{ url = "https://files.pythonhosted.org/packages/fb/8c/f5987895bf42b8ddeea1b315c9fedcfe07cadee28b9c98cf50d00adcb14d/regex-2025.11.3-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5d9903ca42bfeec4cebedba8022a7c97ad2aab22e09573ce9976ba01b65e4361", size = 798592, upload-time = "2025-11-03T21:32:03.006Z" },
{ url = "https://files.pythonhosted.org/packages/99/2a/6591ebeede78203fa77ee46a1c36649e02df9eaa77a033d1ccdf2fcd5d4e/regex-2025.11.3-cp313-cp313-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:639431bdc89d6429f6721625e8129413980ccd62e9d3f496be618a41d205f160", size = 864122, upload-time = "2025-11-03T21:32:04.553Z" },
{ url = "https://files.pythonhosted.org/packages/94/d6/be32a87cf28cf8ed064ff281cfbd49aefd90242a83e4b08b5a86b38e8eb4/regex-2025.11.3-cp313-cp313-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:f117efad42068f9715677c8523ed2be1518116d1c49b1dd17987716695181efe", size = 912272, upload-time = "2025-11-03T21:32:06.148Z" },
{ url = "https://files.pythonhosted.org/packages/62/11/9bcef2d1445665b180ac7f230406ad80671f0fc2a6ffb93493b5dd8cd64c/regex-2025.11.3-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:4aecb6f461316adf9f1f0f6a4a1a3d79e045f9b71ec76055a791affa3b285850", size = 803497, upload-time = "2025-11-03T21:32:08.162Z" },
{ url = "https://files.pythonhosted.org/packages/e5/a7/da0dc273d57f560399aa16d8a68ae7f9b57679476fc7ace46501d455fe84/regex-2025.11.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:3b3a5f320136873cc5561098dfab677eea139521cb9a9e8db98b7e64aef44cbc", size = 787892, upload-time = "2025-11-03T21:32:09.769Z" },
{ url = "https://files.pythonhosted.org/packages/da/4b/732a0c5a9736a0b8d6d720d4945a2f1e6f38f87f48f3173559f53e8d5d82/regex-2025.11.3-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:75fa6f0056e7efb1f42a1c34e58be24072cb9e61a601340cc1196ae92326a4f9", size = 858462, upload-time = "2025-11-03T21:32:11.769Z" },
{ url = "https://files.pythonhosted.org/packages/0c/f5/a2a03df27dc4c2d0c769220f5110ba8c4084b0bfa9ab0f9b4fcfa3d2b0fc/regex-2025.11.3-cp313-cp313-musllinux_1_2_s390x.whl", hash = "sha256:dbe6095001465294f13f1adcd3311e50dd84e5a71525f20a10bd16689c61ce0b", size = 850528, upload-time = "2025-11-03T21:32:13.906Z" },
{ url = "https://files.pythonhosted.org/packages/d6/09/e1cd5bee3841c7f6eb37d95ca91cdee7100b8f88b81e41c2ef426910891a/regex-2025.11.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:454d9b4ae7881afbc25015b8627c16d88a597479b9dea82b8c6e7e2e07240dc7", size = 789866, upload-time = "2025-11-03T21:32:15.748Z" },
{ url = "https://files.pythonhosted.org/packages/eb/51/702f5ea74e2a9c13d855a6a85b7f80c30f9e72a95493260193c07f3f8d74/regex-2025.11.3-cp313-cp313-win32.whl", hash = "sha256:28ba4d69171fc6e9896337d4fc63a43660002b7da53fc15ac992abcf3410917c", size = 266189, upload-time = "2025-11-03T21:32:17.493Z" },
{ url = "https://files.pythonhosted.org/packages/8b/00/6e29bb314e271a743170e53649db0fdb8e8ff0b64b4f425f5602f4eb9014/regex-2025.11.3-cp313-cp313-win_amd64.whl", hash = "sha256:bac4200befe50c670c405dc33af26dad5a3b6b255dd6c000d92fe4629f9ed6a5", size = 277054, upload-time = "2025-11-03T21:32:19.042Z" },
{ url = "https://files.pythonhosted.org/packages/25/f1/b156ff9f2ec9ac441710764dda95e4edaf5f36aca48246d1eea3f1fd96ec/regex-2025.11.3-cp313-cp313-win_arm64.whl", hash = "sha256:2292cd5a90dab247f9abe892ac584cb24f0f54680c73fcb4a7493c66c2bf2467", size = 270325, upload-time = "2025-11-03T21:32:21.338Z" },
{ url = "https://files.pythonhosted.org/packages/20/28/fd0c63357caefe5680b8ea052131acbd7f456893b69cc2a90cc3e0dc90d4/regex-2025.11.3-cp313-cp313t-macosx_10_13_universal2.whl", hash = "sha256:1eb1ebf6822b756c723e09f5186473d93236c06c579d2cc0671a722d2ab14281", size = 491984, upload-time = "2025-11-03T21:32:23.466Z" },
{ url = "https://files.pythonhosted.org/packages/df/ec/7014c15626ab46b902b3bcc4b28a7bae46d8f281fc7ea9c95e22fcaaa917/regex-2025.11.3-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:1e00ec2970aab10dc5db34af535f21fcf32b4a31d99e34963419636e2f85ae39", size = 292673, upload-time = "2025-11-03T21:32:25.034Z" },
{ url = "https://files.pythonhosted.org/packages/23/ab/3b952ff7239f20d05f1f99e9e20188513905f218c81d52fb5e78d2bf7634/regex-2025.11.3-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:a4cb042b615245d5ff9b3794f56be4138b5adc35a4166014d31d1814744148c7", size = 291029, upload-time = "2025-11-03T21:32:26.528Z" },
{ url = "https://files.pythonhosted.org/packages/21/7e/3dc2749fc684f455f162dcafb8a187b559e2614f3826877d3844a131f37b/regex-2025.11.3-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:44f264d4bf02f3176467d90b294d59bf1db9fe53c141ff772f27a8b456b2a9ed", size = 807437, upload-time = "2025-11-03T21:32:28.363Z" },
{ url = "https://files.pythonhosted.org/packages/1b/0b/d529a85ab349c6a25d1ca783235b6e3eedf187247eab536797021f7126c6/regex-2025.11.3-cp313-cp313t-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:7be0277469bf3bd7a34a9c57c1b6a724532a0d235cd0dc4e7f4316f982c28b19", size = 873368, upload-time = "2025-11-03T21:32:30.4Z" },
{ url = "https://files.pythonhosted.org/packages/7d/18/2d868155f8c9e3e9d8f9e10c64e9a9f496bb8f7e037a88a8bed26b435af6/regex-2025.11.3-cp313-cp313t-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:0d31e08426ff4b5b650f68839f5af51a92a5b51abd8554a60c2fbc7c71f25d0b", size = 914921, upload-time = "2025-11-03T21:32:32.123Z" },
{ url = "https://files.pythonhosted.org/packages/2d/71/9d72ff0f354fa783fe2ba913c8734c3b433b86406117a8db4ea2bf1c7a2f/regex-2025.11.3-cp313-cp313t-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:e43586ce5bd28f9f285a6e729466841368c4a0353f6fd08d4ce4630843d3648a", size = 812708, upload-time = "2025-11-03T21:32:34.305Z" },
{ url = "https://files.pythonhosted.org/packages/e7/19/ce4bf7f5575c97f82b6e804ffb5c4e940c62609ab2a0d9538d47a7fdf7d4/regex-2025.11.3-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:0f9397d561a4c16829d4e6ff75202c1c08b68a3bdbfe29dbfcdb31c9830907c6", size = 795472, upload-time = "2025-11-03T21:32:36.364Z" },
{ url = "https://files.pythonhosted.org/packages/03/86/fd1063a176ffb7b2315f9a1b08d17b18118b28d9df163132615b835a26ee/regex-2025.11.3-cp313-cp313t-musllinux_1_2_ppc64le.whl", hash = "sha256:dd16e78eb18ffdb25ee33a0682d17912e8cc8a770e885aeee95020046128f1ce", size = 868341, upload-time = "2025-11-03T21:32:38.042Z" },
{ url = "https://files.pythonhosted.org/packages/12/43/103fb2e9811205e7386366501bc866a164a0430c79dd59eac886a2822950/regex-2025.11.3-cp313-cp313t-musllinux_1_2_s390x.whl", hash = "sha256:ffcca5b9efe948ba0661e9df0fa50d2bc4b097c70b9810212d6b62f05d83b2dd", size = 854666, upload-time = "2025-11-03T21:32:40.079Z" },
{ url = "https://files.pythonhosted.org/packages/7d/22/e392e53f3869b75804762c7c848bd2dd2abf2b70fb0e526f58724638bd35/regex-2025.11.3-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:c56b4d162ca2b43318ac671c65bd4d563e841a694ac70e1a976ac38fcf4ca1d2", size = 799473, upload-time = "2025-11-03T21:32:42.148Z" },
{ url = "https://files.pythonhosted.org/packages/4f/f9/8bd6b656592f925b6845fcbb4d57603a3ac2fb2373344ffa1ed70aa6820a/regex-2025.11.3-cp313-cp313t-win32.whl", hash = "sha256:9ddc42e68114e161e51e272f667d640f97e84a2b9ef14b7477c53aac20c2d59a", size = 268792, upload-time = "2025-11-03T21:32:44.13Z" },
{ url = "https://files.pythonhosted.org/packages/e5/87/0e7d603467775ff65cd2aeabf1b5b50cc1c3708556a8b849a2fa4dd1542b/regex-2025.11.3-cp313-cp313t-win_amd64.whl", hash = "sha256:7a7c7fdf755032ffdd72c77e3d8096bdcb0eb92e89e17571a196f03d88b11b3c", size = 280214, upload-time = "2025-11-03T21:32:45.853Z" },
{ url = "https://files.pythonhosted.org/packages/8d/d0/2afc6f8e94e2b64bfb738a7c2b6387ac1699f09f032d363ed9447fd2bb57/regex-2025.11.3-cp313-cp313t-win_arm64.whl", hash = "sha256:df9eb838c44f570283712e7cff14c16329a9f0fb19ca492d21d4b7528ee6821e", size = 271469, upload-time = "2025-11-03T21:32:48.026Z" },
{ url = "https://files.pythonhosted.org/packages/31/e9/f6e13de7e0983837f7b6d238ad9458800a874bf37c264f7923e63409944c/regex-2025.11.3-cp314-cp314-macosx_10_13_universal2.whl", hash = "sha256:9697a52e57576c83139d7c6f213d64485d3df5bf84807c35fa409e6c970801c6", size = 489089, upload-time = "2025-11-03T21:32:50.027Z" },
{ url = "https://files.pythonhosted.org/packages/a3/5c/261f4a262f1fa65141c1b74b255988bd2fa020cc599e53b080667d591cfc/regex-2025.11.3-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:e18bc3f73bd41243c9b38a6d9f2366cd0e0137a9aebe2d8ff76c5b67d4c0a3f4", size = 291059, upload-time = "2025-11-03T21:32:51.682Z" },
{ url = "https://files.pythonhosted.org/packages/8e/57/f14eeb7f072b0e9a5a090d1712741fd8f214ec193dba773cf5410108bb7d/regex-2025.11.3-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:61a08bcb0ec14ff4e0ed2044aad948d0659604f824cbd50b55e30b0ec6f09c73", size = 288900, upload-time = "2025-11-03T21:32:53.569Z" },
{ url = "https://files.pythonhosted.org/packages/3c/6b/1d650c45e99a9b327586739d926a1cd4e94666b1bd4af90428b36af66dc7/regex-2025.11.3-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:c9c30003b9347c24bcc210958c5d167b9e4f9be786cb380a7d32f14f9b84674f", size = 799010, upload-time = "2025-11-03T21:32:55.222Z" },
{ url = "https://files.pythonhosted.org/packages/99/ee/d66dcbc6b628ce4e3f7f0cbbb84603aa2fc0ffc878babc857726b8aab2e9/regex-2025.11.3-cp314-cp314-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:4e1e592789704459900728d88d41a46fe3969b82ab62945560a31732ffc19a6d", size = 864893, upload-time = "2025-11-03T21:32:57.239Z" },
{ url = "https://files.pythonhosted.org/packages/bf/2d/f238229f1caba7ac87a6c4153d79947fb0261415827ae0f77c304260c7d3/regex-2025.11.3-cp314-cp314-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:6538241f45eb5a25aa575dbba1069ad786f68a4f2773a29a2bd3dd1f9de787be", size = 911522, upload-time = "2025-11-03T21:32:59.274Z" },
{ url = "https://files.pythonhosted.org/packages/bd/3d/22a4eaba214a917c80e04f6025d26143690f0419511e0116508e24b11c9b/regex-2025.11.3-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:bce22519c989bb72a7e6b36a199384c53db7722fe669ba891da75907fe3587db", size = 803272, upload-time = "2025-11-03T21:33:01.393Z" },
{ url = "https://files.pythonhosted.org/packages/84/b1/03188f634a409353a84b5ef49754b97dbcc0c0f6fd6c8ede505a8960a0a4/regex-2025.11.3-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:66d559b21d3640203ab9075797a55165d79017520685fb407b9234d72ab63c62", size = 787958, upload-time = "2025-11-03T21:33:03.379Z" },
{ url = "https://files.pythonhosted.org/packages/99/6a/27d072f7fbf6fadd59c64d210305e1ff865cc3b78b526fd147db768c553b/regex-2025.11.3-cp314-cp314-musllinux_1_2_ppc64le.whl", hash = "sha256:669dcfb2e38f9e8c69507bace46f4889e3abbfd9b0c29719202883c0a603598f", size = 859289, upload-time = "2025-11-03T21:33:05.374Z" },
{ url = "https://files.pythonhosted.org/packages/9a/70/1b3878f648e0b6abe023172dacb02157e685564853cc363d9961bcccde4e/regex-2025.11.3-cp314-cp314-musllinux_1_2_s390x.whl", hash = "sha256:32f74f35ff0f25a5021373ac61442edcb150731fbaa28286bbc8bb1582c89d02", size = 850026, upload-time = "2025-11-03T21:33:07.131Z" },
{ url = "https://files.pythonhosted.org/packages/dd/d5/68e25559b526b8baab8e66839304ede68ff6727237a47727d240006bd0ff/regex-2025.11.3-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:e6c7a21dffba883234baefe91bc3388e629779582038f75d2a5be918e250f0ed", size = 789499, upload-time = "2025-11-03T21:33:09.141Z" },
{ url = "https://files.pythonhosted.org/packages/fc/df/43971264857140a350910d4e33df725e8c94dd9dee8d2e4729fa0d63d49e/regex-2025.11.3-cp314-cp314-win32.whl", hash = "sha256:795ea137b1d809eb6836b43748b12634291c0ed55ad50a7d72d21edf1cd565c4", size = 271604, upload-time = "2025-11-03T21:33:10.9Z" },
{ url = "https://files.pythonhosted.org/packages/01/6f/9711b57dc6894a55faf80a4c1b5aa4f8649805cb9c7aef46f7d27e2b9206/regex-2025.11.3-cp314-cp314-win_amd64.whl", hash = "sha256:9f95fbaa0ee1610ec0fc6b26668e9917a582ba80c52cc6d9ada15e30aa9ab9ad", size = 280320, upload-time = "2025-11-03T21:33:12.572Z" },
{ url = "https://files.pythonhosted.org/packages/f1/7e/f6eaa207d4377481f5e1775cdeb5a443b5a59b392d0065f3417d31d80f87/regex-2025.11.3-cp314-cp314-win_arm64.whl", hash = "sha256:dfec44d532be4c07088c3de2876130ff0fbeeacaa89a137decbbb5f665855a0f", size = 273372, upload-time = "2025-11-03T21:33:14.219Z" },
{ url = "https://files.pythonhosted.org/packages/c3/06/49b198550ee0f5e4184271cee87ba4dfd9692c91ec55289e6282f0f86ccf/regex-2025.11.3-cp314-cp314t-macosx_10_13_universal2.whl", hash = "sha256:ba0d8a5d7f04f73ee7d01d974d47c5834f8a1b0224390e4fe7c12a3a92a78ecc", size = 491985, upload-time = "2025-11-03T21:33:16.555Z" },
{ url = "https://files.pythonhosted.org/packages/ce/bf/abdafade008f0b1c9da10d934034cb670432d6cf6cbe38bbb53a1cfd6cf8/regex-2025.11.3-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:442d86cf1cfe4faabf97db7d901ef58347efd004934da045c745e7b5bd57ac49", size = 292669, upload-time = "2025-11-03T21:33:18.32Z" },
{ url = "https://files.pythonhosted.org/packages/f9/ef/0c357bb8edbd2ad8e273fcb9e1761bc37b8acbc6e1be050bebd6475f19c1/regex-2025.11.3-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:fd0a5e563c756de210bb964789b5abe4f114dacae9104a47e1a649b910361536", size = 291030, upload-time = "2025-11-03T21:33:20.048Z" },
{ url = "https://files.pythonhosted.org/packages/79/06/edbb67257596649b8fb088d6aeacbcb248ac195714b18a65e018bf4c0b50/regex-2025.11.3-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:bf3490bcbb985a1ae97b2ce9ad1c0f06a852d5b19dde9b07bdf25bf224248c95", size = 807674, upload-time = "2025-11-03T21:33:21.797Z" },
{ url = "https://files.pythonhosted.org/packages/f4/d9/ad4deccfce0ea336296bd087f1a191543bb99ee1c53093dcd4c64d951d00/regex-2025.11.3-cp314-cp314t-manylinux2014_ppc64le.manylinux_2_17_ppc64le.manylinux_2_28_ppc64le.whl", hash = "sha256:3809988f0a8b8c9dcc0f92478d6501fac7200b9ec56aecf0ec21f4a2ec4b6009", size = 873451, upload-time = "2025-11-03T21:33:23.741Z" },
{ url = "https://files.pythonhosted.org/packages/13/75/a55a4724c56ef13e3e04acaab29df26582f6978c000ac9cd6810ad1f341f/regex-2025.11.3-cp314-cp314t-manylinux2014_s390x.manylinux_2_17_s390x.manylinux_2_28_s390x.whl", hash = "sha256:f4ff94e58e84aedb9c9fce66d4ef9f27a190285b451420f297c9a09f2b9abee9", size = 914980, upload-time = "2025-11-03T21:33:25.999Z" },
{ url = "https://files.pythonhosted.org/packages/67/1e/a1657ee15bd9116f70d4a530c736983eed997b361e20ecd8f5ca3759d5c5/regex-2025.11.3-cp314-cp314t-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7eb542fd347ce61e1321b0a6b945d5701528dca0cd9759c2e3bb8bd57e47964d", size = 812852, upload-time = "2025-11-03T21:33:27.852Z" },
{ url = "https://files.pythonhosted.org/packages/b8/6f/f7516dde5506a588a561d296b2d0044839de06035bb486b326065b4c101e/regex-2025.11.3-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:d6c2d5919075a1f2e413c00b056ea0c2f065b3f5fe83c3d07d325ab92dce51d6", size = 795566, upload-time = "2025-11-03T21:33:32.364Z" },
{ url = "https://files.pythonhosted.org/packages/d9/dd/3d10b9e170cc16fb34cb2cef91513cf3df65f440b3366030631b2984a264/regex-2025.11.3-cp314-cp314t-musllinux_1_2_ppc64le.whl", hash = "sha256:3f8bf11a4827cc7ce5a53d4ef6cddd5ad25595d3c1435ef08f76825851343154", size = 868463, upload-time = "2025-11-03T21:33:34.459Z" },
{ url = "https://files.pythonhosted.org/packages/f5/8e/935e6beff1695aa9085ff83195daccd72acc82c81793df480f34569330de/regex-2025.11.3-cp314-cp314t-musllinux_1_2_s390x.whl", hash = "sha256:22c12d837298651e5550ac1d964e4ff57c3f56965fc1812c90c9fb2028eaf267", size = 854694, upload-time = "2025-11-03T21:33:36.793Z" },
{ url = "https://files.pythonhosted.org/packages/92/12/10650181a040978b2f5720a6a74d44f841371a3d984c2083fc1752e4acf6/regex-2025.11.3-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:62ba394a3dda9ad41c7c780f60f6e4a70988741415ae96f6d1bf6c239cf01379", size = 799691, upload-time = "2025-11-03T21:33:39.079Z" },
{ url = "https://files.pythonhosted.org/packages/67/90/8f37138181c9a7690e7e4cb388debbd389342db3c7381d636d2875940752/regex-2025.11.3-cp314-cp314t-win32.whl", hash = "sha256:4bf146dca15cdd53224a1bf46d628bd7590e4a07fbb69e720d561aea43a32b38", size = 274583, upload-time = "2025-11-03T21:33:41.302Z" },
{ url = "https://files.pythonhosted.org/packages/8f/cd/867f5ec442d56beb56f5f854f40abcfc75e11d10b11fdb1869dd39c63aaf/regex-2025.11.3-cp314-cp314t-win_amd64.whl", hash = "sha256:adad1a1bcf1c9e76346e091d22d23ac54ef28e1365117d99521631078dfec9de", size = 284286, upload-time = "2025-11-03T21:33:43.324Z" },
{ url = "https://files.pythonhosted.org/packages/20/31/32c0c4610cbc070362bf1d2e4ea86d1ea29014d400a6d6c2486fcfd57766/regex-2025.11.3-cp314-cp314t-win_arm64.whl", hash = "sha256:c54f768482cef41e219720013cd05933b6f971d9562544d691c68699bf2b6801", size = 274741, upload-time = "2025-11-03T21:33:45.557Z" },
]
[[package]]
name = "requests"
version = "2.32.5"
@@ -2106,19 +1998,6 @@ wheels = [
{ url = "https://files.pythonhosted.org/packages/d9/52/1064f510b141bd54025f9b55105e26d1fa970b9be67ad766380a3c9b74b0/starlette-0.50.0-py3-none-any.whl", hash = "sha256:9e5391843ec9b6e472eed1365a78c8098cfceb7a74bfd4d6b1c0c0095efb3bca", size = 74033, upload-time = "2025-11-01T15:25:25.461Z" },
]
[[package]]
name = "stripe"
version = "14.0.1"
source = { registry = "https://pypi.org/simple" }
dependencies = [
{ name = "requests" },
{ name = "typing-extensions" },
]
sdist = { url = "https://files.pythonhosted.org/packages/2b/49/08df0acc094587f4d76c2ab31ebbecb8a37312ab558cddaa6a4c2ff19579/stripe-14.0.1.tar.gz", hash = "sha256:f2d56345bf5d41c1f21f814b00174a3173a0b5eb4e8fc46a8f779e3d7a2efc6e", size = 1362960, upload-time = "2025-11-22T01:07:48.862Z" }
wheels = [
{ url = "https://files.pythonhosted.org/packages/d3/88/0db878a84d333a188714f4ade57c9ae765a14a0b81862eb133ad7864711c/stripe-14.0.1-py3-none-any.whl", hash = "sha256:ff25c5e5f085beaa98b6b9c2c729d22ad99068196cbd83fdf82669fd08311b76", size = 1970603, upload-time = "2025-11-22T01:07:47.309Z" },
]
[[package]]
name = "sympy"
version = "1.14.0"
@@ -2131,60 +2010,6 @@ wheels = [
{ url = "https://files.pythonhosted.org/packages/a2/09/77d55d46fd61b4a135c444fc97158ef34a095e5681d0a6c10b75bf356191/sympy-1.14.0-py3-none-any.whl", hash = "sha256:e091cc3e99d2141a0ba2847328f5479b05d94a6635cb96148ccb3f34671bd8f5", size = 6299353, upload-time = "2025-04-27T18:04:59.103Z" },
]
[[package]]
name = "tiktoken"
version = "0.12.0"
source = { registry = "https://pypi.org/simple" }
dependencies = [
{ name = "regex" },
{ name = "requests" },
]
sdist = { url = "https://files.pythonhosted.org/packages/7d/ab/4d017d0f76ec3171d469d80fc03dfbb4e48a4bcaddaa831b31d526f05edc/tiktoken-0.12.0.tar.gz", hash = "sha256:b18ba7ee2b093863978fcb14f74b3707cdc8d4d4d3836853ce7ec60772139931", size = 37806, upload-time = "2025-10-06T20:22:45.419Z" }
wheels = [
{ url = "https://files.pythonhosted.org/packages/de/46/21ea696b21f1d6d1efec8639c204bdf20fde8bafb351e1355c72c5d7de52/tiktoken-0.12.0-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:6e227c7f96925003487c33b1b32265fad2fbcec2b7cf4817afb76d416f40f6bb", size = 1051565, upload-time = "2025-10-06T20:21:44.566Z" },
{ url = "https://files.pythonhosted.org/packages/c9/d9/35c5d2d9e22bb2a5f74ba48266fb56c63d76ae6f66e02feb628671c0283e/tiktoken-0.12.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:c06cf0fcc24c2cb2adb5e185c7082a82cba29c17575e828518c2f11a01f445aa", size = 995284, upload-time = "2025-10-06T20:21:45.622Z" },
{ url = "https://files.pythonhosted.org/packages/01/84/961106c37b8e49b9fdcf33fe007bb3a8fdcc380c528b20cc7fbba80578b8/tiktoken-0.12.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:f18f249b041851954217e9fd8e5c00b024ab2315ffda5ed77665a05fa91f42dc", size = 1129201, upload-time = "2025-10-06T20:21:47.074Z" },
{ url = "https://files.pythonhosted.org/packages/6a/d0/3d9275198e067f8b65076a68894bb52fd253875f3644f0a321a720277b8a/tiktoken-0.12.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:47a5bc270b8c3db00bb46ece01ef34ad050e364b51d406b6f9730b64ac28eded", size = 1152444, upload-time = "2025-10-06T20:21:48.139Z" },
{ url = "https://files.pythonhosted.org/packages/78/db/a58e09687c1698a7c592e1038e01c206569b86a0377828d51635561f8ebf/tiktoken-0.12.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:508fa71810c0efdcd1b898fda574889ee62852989f7c1667414736bcb2b9a4bd", size = 1195080, upload-time = "2025-10-06T20:21:49.246Z" },
{ url = "https://files.pythonhosted.org/packages/9e/1b/a9e4d2bf91d515c0f74afc526fd773a812232dd6cda33ebea7f531202325/tiktoken-0.12.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:a1af81a6c44f008cba48494089dd98cccb8b313f55e961a52f5b222d1e507967", size = 1255240, upload-time = "2025-10-06T20:21:50.274Z" },
{ url = "https://files.pythonhosted.org/packages/9d/15/963819345f1b1fb0809070a79e9dd96938d4ca41297367d471733e79c76c/tiktoken-0.12.0-cp311-cp311-win_amd64.whl", hash = "sha256:3e68e3e593637b53e56f7237be560f7a394451cb8c11079755e80ae64b9e6def", size = 879422, upload-time = "2025-10-06T20:21:51.734Z" },
{ url = "https://files.pythonhosted.org/packages/a4/85/be65d39d6b647c79800fd9d29241d081d4eeb06271f383bb87200d74cf76/tiktoken-0.12.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:b97f74aca0d78a1ff21b8cd9e9925714c15a9236d6ceacf5c7327c117e6e21e8", size = 1050728, upload-time = "2025-10-06T20:21:52.756Z" },
{ url = "https://files.pythonhosted.org/packages/4a/42/6573e9129bc55c9bf7300b3a35bef2c6b9117018acca0dc760ac2d93dffe/tiktoken-0.12.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:2b90f5ad190a4bb7c3eb30c5fa32e1e182ca1ca79f05e49b448438c3e225a49b", size = 994049, upload-time = "2025-10-06T20:21:53.782Z" },
{ url = "https://files.pythonhosted.org/packages/66/c5/ed88504d2f4a5fd6856990b230b56d85a777feab84e6129af0822f5d0f70/tiktoken-0.12.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:65b26c7a780e2139e73acc193e5c63ac754021f160df919add909c1492c0fb37", size = 1129008, upload-time = "2025-10-06T20:21:54.832Z" },
{ url = "https://files.pythonhosted.org/packages/f4/90/3dae6cc5436137ebd38944d396b5849e167896fc2073da643a49f372dc4f/tiktoken-0.12.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:edde1ec917dfd21c1f2f8046b86348b0f54a2c0547f68149d8600859598769ad", size = 1152665, upload-time = "2025-10-06T20:21:56.129Z" },
{ url = "https://files.pythonhosted.org/packages/a3/fe/26df24ce53ffde419a42f5f53d755b995c9318908288c17ec3f3448313a3/tiktoken-0.12.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:35a2f8ddd3824608b3d650a000c1ef71f730d0c56486845705a8248da00f9fe5", size = 1194230, upload-time = "2025-10-06T20:21:57.546Z" },
{ url = "https://files.pythonhosted.org/packages/20/cc/b064cae1a0e9fac84b0d2c46b89f4e57051a5f41324e385d10225a984c24/tiktoken-0.12.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:83d16643edb7fa2c99eff2ab7733508aae1eebb03d5dfc46f5565862810f24e3", size = 1254688, upload-time = "2025-10-06T20:21:58.619Z" },
{ url = "https://files.pythonhosted.org/packages/81/10/b8523105c590c5b8349f2587e2fdfe51a69544bd5a76295fc20f2374f470/tiktoken-0.12.0-cp312-cp312-win_amd64.whl", hash = "sha256:ffc5288f34a8bc02e1ea7047b8d041104791d2ddbf42d1e5fa07822cbffe16bd", size = 878694, upload-time = "2025-10-06T20:21:59.876Z" },
{ url = "https://files.pythonhosted.org/packages/00/61/441588ee21e6b5cdf59d6870f86beb9789e532ee9718c251b391b70c68d6/tiktoken-0.12.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:775c2c55de2310cc1bc9a3ad8826761cbdc87770e586fd7b6da7d4589e13dab3", size = 1050802, upload-time = "2025-10-06T20:22:00.96Z" },
{ url = "https://files.pythonhosted.org/packages/1f/05/dcf94486d5c5c8d34496abe271ac76c5b785507c8eae71b3708f1ad9b45a/tiktoken-0.12.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:a01b12f69052fbe4b080a2cfb867c4de12c704b56178edf1d1d7b273561db160", size = 993995, upload-time = "2025-10-06T20:22:02.788Z" },
{ url = "https://files.pythonhosted.org/packages/a0/70/5163fe5359b943f8db9946b62f19be2305de8c3d78a16f629d4165e2f40e/tiktoken-0.12.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:01d99484dc93b129cd0964f9d34eee953f2737301f18b3c7257bf368d7615baa", size = 1128948, upload-time = "2025-10-06T20:22:03.814Z" },
{ url = "https://files.pythonhosted.org/packages/0c/da/c028aa0babf77315e1cef357d4d768800c5f8a6de04d0eac0f377cb619fa/tiktoken-0.12.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:4a1a4fcd021f022bfc81904a911d3df0f6543b9e7627b51411da75ff2fe7a1be", size = 1151986, upload-time = "2025-10-06T20:22:05.173Z" },
{ url = "https://files.pythonhosted.org/packages/a0/5a/886b108b766aa53e295f7216b509be95eb7d60b166049ce2c58416b25f2a/tiktoken-0.12.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:981a81e39812d57031efdc9ec59fa32b2a5a5524d20d4776574c4b4bd2e9014a", size = 1194222, upload-time = "2025-10-06T20:22:06.265Z" },
{ url = "https://files.pythonhosted.org/packages/f4/f8/4db272048397636ac7a078d22773dd2795b1becee7bc4922fe6207288d57/tiktoken-0.12.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:9baf52f84a3f42eef3ff4e754a0db79a13a27921b457ca9832cf944c6be4f8f3", size = 1255097, upload-time = "2025-10-06T20:22:07.403Z" },
{ url = "https://files.pythonhosted.org/packages/8e/32/45d02e2e0ea2be3a9ed22afc47d93741247e75018aac967b713b2941f8ea/tiktoken-0.12.0-cp313-cp313-win_amd64.whl", hash = "sha256:b8a0cd0c789a61f31bf44851defbd609e8dd1e2c8589c614cc1060940ef1f697", size = 879117, upload-time = "2025-10-06T20:22:08.418Z" },
{ url = "https://files.pythonhosted.org/packages/ce/76/994fc868f88e016e6d05b0da5ac24582a14c47893f4474c3e9744283f1d5/tiktoken-0.12.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:d5f89ea5680066b68bcb797ae85219c72916c922ef0fcdd3480c7d2315ffff16", size = 1050309, upload-time = "2025-10-06T20:22:10.939Z" },
{ url = "https://files.pythonhosted.org/packages/f6/b8/57ef1456504c43a849821920d582a738a461b76a047f352f18c0b26c6516/tiktoken-0.12.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:b4e7ed1c6a7a8a60a3230965bdedba8cc58f68926b835e519341413370e0399a", size = 993712, upload-time = "2025-10-06T20:22:12.115Z" },
{ url = "https://files.pythonhosted.org/packages/72/90/13da56f664286ffbae9dbcfadcc625439142675845baa62715e49b87b68b/tiktoken-0.12.0-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:fc530a28591a2d74bce821d10b418b26a094bf33839e69042a6e86ddb7a7fb27", size = 1128725, upload-time = "2025-10-06T20:22:13.541Z" },
{ url = "https://files.pythonhosted.org/packages/05/df/4f80030d44682235bdaecd7346c90f67ae87ec8f3df4a3442cb53834f7e4/tiktoken-0.12.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:06a9f4f49884139013b138920a4c393aa6556b2f8f536345f11819389c703ebb", size = 1151875, upload-time = "2025-10-06T20:22:14.559Z" },
{ url = "https://files.pythonhosted.org/packages/22/1f/ae535223a8c4ef4c0c1192e3f9b82da660be9eb66b9279e95c99288e9dab/tiktoken-0.12.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:04f0e6a985d95913cabc96a741c5ffec525a2c72e9df086ff17ebe35985c800e", size = 1194451, upload-time = "2025-10-06T20:22:15.545Z" },
{ url = "https://files.pythonhosted.org/packages/78/a7/f8ead382fce0243cb625c4f266e66c27f65ae65ee9e77f59ea1653b6d730/tiktoken-0.12.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:0ee8f9ae00c41770b5f9b0bb1235474768884ae157de3beb5439ca0fd70f3e25", size = 1253794, upload-time = "2025-10-06T20:22:16.624Z" },
{ url = "https://files.pythonhosted.org/packages/93/e0/6cc82a562bc6365785a3ff0af27a2a092d57c47d7a81d9e2295d8c36f011/tiktoken-0.12.0-cp313-cp313t-win_amd64.whl", hash = "sha256:dc2dd125a62cb2b3d858484d6c614d136b5b848976794edfb63688d539b8b93f", size = 878777, upload-time = "2025-10-06T20:22:18.036Z" },
{ url = "https://files.pythonhosted.org/packages/72/05/3abc1db5d2c9aadc4d2c76fa5640134e475e58d9fbb82b5c535dc0de9b01/tiktoken-0.12.0-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:a90388128df3b3abeb2bfd1895b0681412a8d7dc644142519e6f0a97c2111646", size = 1050188, upload-time = "2025-10-06T20:22:19.563Z" },
{ url = "https://files.pythonhosted.org/packages/e3/7b/50c2f060412202d6c95f32b20755c7a6273543b125c0985d6fa9465105af/tiktoken-0.12.0-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:da900aa0ad52247d8794e307d6446bd3cdea8e192769b56276695d34d2c9aa88", size = 993978, upload-time = "2025-10-06T20:22:20.702Z" },
{ url = "https://files.pythonhosted.org/packages/14/27/bf795595a2b897e271771cd31cb847d479073497344c637966bdf2853da1/tiktoken-0.12.0-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:285ba9d73ea0d6171e7f9407039a290ca77efcdb026be7769dccc01d2c8d7fff", size = 1129271, upload-time = "2025-10-06T20:22:22.06Z" },
{ url = "https://files.pythonhosted.org/packages/f5/de/9341a6d7a8f1b448573bbf3425fa57669ac58258a667eb48a25dfe916d70/tiktoken-0.12.0-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:d186a5c60c6a0213f04a7a802264083dea1bbde92a2d4c7069e1a56630aef830", size = 1151216, upload-time = "2025-10-06T20:22:23.085Z" },
{ url = "https://files.pythonhosted.org/packages/75/0d/881866647b8d1be4d67cb24e50d0c26f9f807f994aa1510cb9ba2fe5f612/tiktoken-0.12.0-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:604831189bd05480f2b885ecd2d1986dc7686f609de48208ebbbddeea071fc0b", size = 1194860, upload-time = "2025-10-06T20:22:24.602Z" },
{ url = "https://files.pythonhosted.org/packages/b3/1e/b651ec3059474dab649b8d5b69f5c65cd8fcd8918568c1935bd4136c9392/tiktoken-0.12.0-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:8f317e8530bb3a222547b85a58583238c8f74fd7a7408305f9f63246d1a0958b", size = 1254567, upload-time = "2025-10-06T20:22:25.671Z" },
{ url = "https://files.pythonhosted.org/packages/80/57/ce64fd16ac390fafde001268c364d559447ba09b509181b2808622420eec/tiktoken-0.12.0-cp314-cp314-win_amd64.whl", hash = "sha256:399c3dd672a6406719d84442299a490420b458c44d3ae65516302a99675888f3", size = 921067, upload-time = "2025-10-06T20:22:26.753Z" },
{ url = "https://files.pythonhosted.org/packages/ac/a4/72eed53e8976a099539cdd5eb36f241987212c29629d0a52c305173e0a68/tiktoken-0.12.0-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:c2c714c72bc00a38ca969dae79e8266ddec999c7ceccd603cc4f0d04ccd76365", size = 1050473, upload-time = "2025-10-06T20:22:27.775Z" },
{ url = "https://files.pythonhosted.org/packages/e6/d7/0110b8f54c008466b19672c615f2168896b83706a6611ba6e47313dbc6e9/tiktoken-0.12.0-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:cbb9a3ba275165a2cb0f9a83f5d7025afe6b9d0ab01a22b50f0e74fee2ad253e", size = 993855, upload-time = "2025-10-06T20:22:28.799Z" },
{ url = "https://files.pythonhosted.org/packages/5f/77/4f268c41a3957c418b084dd576ea2fad2e95da0d8e1ab705372892c2ca22/tiktoken-0.12.0-cp314-cp314t-manylinux_2_28_aarch64.whl", hash = "sha256:dfdfaa5ffff8993a3af94d1125870b1d27aed7cb97aa7eb8c1cefdbc87dbee63", size = 1129022, upload-time = "2025-10-06T20:22:29.981Z" },
{ url = "https://files.pythonhosted.org/packages/4e/2b/fc46c90fe5028bd094cd6ee25a7db321cb91d45dc87531e2bdbb26b4867a/tiktoken-0.12.0-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:584c3ad3d0c74f5269906eb8a659c8bfc6144a52895d9261cdaf90a0ae5f4de0", size = 1150736, upload-time = "2025-10-06T20:22:30.996Z" },
{ url = "https://files.pythonhosted.org/packages/28/c0/3c7a39ff68022ddfd7d93f3337ad90389a342f761c4d71de99a3ccc57857/tiktoken-0.12.0-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:54c891b416a0e36b8e2045b12b33dd66fb34a4fe7965565f1b482da50da3e86a", size = 1194908, upload-time = "2025-10-06T20:22:32.073Z" },
{ url = "https://files.pythonhosted.org/packages/ab/0d/c1ad6f4016a3968c048545f5d9b8ffebf577774b2ede3e2e352553b685fe/tiktoken-0.12.0-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:5edb8743b88d5be814b1a8a8854494719080c28faaa1ccbef02e87354fe71ef0", size = 1253706, upload-time = "2025-10-06T20:22:33.385Z" },
{ url = "https://files.pythonhosted.org/packages/af/df/c7891ef9d2712ad774777271d39fdef63941ffba0a9d59b7ad1fd2765e57/tiktoken-0.12.0-cp314-cp314t-win_amd64.whl", hash = "sha256:f61c0aea5565ac82e2ec50a05e02a6c44734e91b51c10510b084ea1b8e633a71", size = 920667, upload-time = "2025-10-06T20:22:34.444Z" },
]
[[package]]
name = "tqdm"
version = "4.67.1"
@@ -2218,18 +2043,6 @@ wheels = [
{ url = "https://files.pythonhosted.org/packages/dc/9b/47798a6c91d8bdb567fe2698fe81e0c6b7cb7ef4d13da4114b41d239f65d/typing_inspection-0.4.2-py3-none-any.whl", hash = "sha256:4ed1cacbdc298c220f1bd249ed5287caa16f34d44ef4e9c3d0cbad5b521545e7", size = 14611, upload-time = "2025-10-01T02:14:40.154Z" },
]
[[package]]
name = "upstash-redis"
version = "1.5.0"
source = { registry = "https://pypi.org/simple" }
dependencies = [
{ name = "httpx" },
]
sdist = { url = "https://files.pythonhosted.org/packages/a8/62/bc53c35fbf4e2b774ab0eb02f3908cfe89b6636e87cdc40b264a4fc1dcce/upstash_redis-1.5.0.tar.gz", hash = "sha256:1917d4d009ca803815092892d92c7da9138b4ada6b353974fb74caf063c6d2a3", size = 39356, upload-time = "2025-10-22T10:15:34.608Z" }
wheels = [
{ url = "https://files.pythonhosted.org/packages/5a/87/d24541a1d9c29033e74aa05b5d8b4857feff79344ebd8fca410eb4683795/upstash_redis-1.5.0-py3-none-any.whl", hash = "sha256:e08de1f74d3fb48a81b383c00398cc9336c43b65b82e6d9266312143970800b9", size = 41088, upload-time = "2025-10-22T10:15:33.363Z" },
]
[[package]]
name = "urllib3"
version = "2.5.0"
@@ -2432,7 +2245,7 @@ wheels = [
[[package]]
name = "yargi-mcp"
version = "0.1.9"
version = "0.2.0"
source = { editable = "." }
dependencies = [
{ name = "aiohttp" },
@@ -2461,19 +2274,11 @@ production = [
{ name = "gunicorn" },
{ name = "uvicorn", extra = ["standard"] },
]
saas = [
{ name = "clerk-backend-api" },
{ name = "pyjwt" },
{ name = "stripe" },
{ name = "tiktoken" },
{ name = "upstash-redis" },
]
[package.metadata]
requires-dist = [
{ name = "aiohttp", specifier = ">=3.11.18" },
{ name = "beautifulsoup4", specifier = ">=4.13.4" },
{ name = "clerk-backend-api", marker = "extra == 'saas'", specifier = ">=3.0.0" },
{ name = "cryptography", specifier = ">=44.0.0" },
{ name = "fastapi", specifier = ">=0.115.14" },
{ name = "fastapi", marker = "extra == 'api'", specifier = ">=0.115.0" },
@@ -2484,17 +2289,13 @@ requires-dist = [
{ name = "numpy", specifier = ">=1.24.0" },
{ name = "openai", specifier = ">=1.0.0" },
{ name = "pydantic", specifier = ">=2.11.4" },
{ name = "pyjwt", marker = "extra == 'saas'", specifier = ">=2.8.0" },
{ name = "pypdf", specifier = ">=5.5.0" },
{ name = "starlette", marker = "extra == 'asgi'", specifier = ">=0.37.0" },
{ name = "stripe", marker = "extra == 'saas'", specifier = ">=9.1.0" },
{ name = "tiktoken", marker = "extra == 'saas'", specifier = ">=0.5.0" },
{ name = "upstash-redis", marker = "extra == 'saas'", specifier = ">=1.1.0" },
{ name = "uvicorn", extras = ["standard"], marker = "extra == 'api'", specifier = ">=0.30.0" },
{ name = "uvicorn", extras = ["standard"], marker = "extra == 'asgi'", specifier = ">=0.30.0" },
{ name = "uvicorn", extras = ["standard"], marker = "extra == 'production'", specifier = ">=0.30.0" },
]
provides-extras = ["asgi", "api", "production", "saas"]
provides-extras = ["asgi", "api", "production"]
[[package]]
name = "yarl"
+4 -3
View File
@@ -1,8 +1,9 @@
# uyusmazlik_mcp_module/client.py
import httpx
import asyncio
import httpx
from bs4 import BeautifulSoup
from typing import Dict, Any, List, Optional, Union, Tuple
from typing import Dict, Any, List, Optional, Union, Tuple
import logging
import html
import re
@@ -232,7 +233,7 @@ class UyusmazlikApiClient:
logger.warning(f"UyusmazlikApiClient: Received empty or non-string HTML from URL {document_url}.")
return UyusmazlikDocumentMarkdown(source_url=document_url, markdown_content=None)
markdown_content = self._convert_html_to_markdown_uyusmazlik(html_content_from_api)
markdown_content = await asyncio.to_thread(self._convert_html_to_markdown_uyusmazlik, html_content_from_api)
return UyusmazlikDocumentMarkdown(source_url=document_url, markdown_content=markdown_content)
except httpx.RequestError as e:
logger.error(f"UyusmazlikApiClient (httpx for docs): HTTP error fetching Uyuşmazlık document from {document_url}: {e}")
+2 -1
View File
@@ -1,5 +1,6 @@
# yargitay_mcp_module/client.py
import asyncio
import httpx
from bs4 import BeautifulSoup # Still needed for pre-processing HTML before markitdown
from typing import Dict, Any, List, Optional
@@ -159,7 +160,7 @@ class YargitayOfficialApiClient:
logger.error(f"YargitayOfficialApiClient: 'data' field in API response is not a string or not found (ID: {id}).")
raise ValueError("Expected HTML content not found in API response's 'data' field.")
markdown_content = self._convert_html_to_markdown(html_content_from_api)
markdown_content = await asyncio.to_thread(self._convert_html_to_markdown, html_content_from_api)
return YargitayDocumentMarkdown(
id=id,