feat: Add semantic search documentation and bump version to 0.2.0

- Add OpenRouter API configuration guide for Claude Desktop, 5ire, Gemini CLI
- Document semantic search workflow (initial_keyword + query)
- Update tool count to reflect optional semantic search tool
- Bump version to 0.2.0 for semantic search feature release

.env.example

@@ -70,6 +70,15 @@ JWT_SECRET_KEY=your_jwt_secret_key_here
 # MAX_REQUESTS_PER_MINUTE=60
 # BURST_CAPACITY=20
 
+# =============================================================================
+# SEMANTIC SEARCH SETTINGS (Optional)
+# =============================================================================
+
+# OpenRouter API Key for semantic search functionality
+# Get your API key from: https://openrouter.ai/keys
+# If not set, semantic search tool will be disabled
+OPENROUTER_API_KEY=sk-or-v1-your_openrouter_api_key_here
+
 # =============================================================================
 # USAGE INSTRUCTIONS
 # =============================================================================

.gitignore

@@ -213,3 +213,4 @@ measure_mcp_directly.py
 playwright_mcp_overhead.json
 simple_test.py
 analyze_anayasa_html.py
+CLAUDE.md

.serena/.gitignore

@@ -0,0 +1 @@
+/cache

.serena/project.yml

@@ -0,0 +1,84 @@
+# 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: []

Dockerfile

@@ -1,5 +1,5 @@
-# -------- BASE IMAGE (includes Chromium & deps) ----------------------------
-FROM mcr.microsoft.com/playwright/python:v1.53.0-noble
+# -------- BASE IMAGE ---------------------------------------------------------
+FROM python:3.12-slim
 
 # -------- Runtime setup ----------------------------------------------------
 WORKDIR /app
@@ -9,8 +9,13 @@ COPY pyproject.toml poetry.lock* requirements*.txt* ./
 
 # 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]
 
+# Cache buster - force rebuild
+ARG CACHE_BUST=202510061202
+RUN echo "Cache bust: $CACHE_BUST"
+
 # Copy application source
 COPY . .
 

README.md

@@ -4,6 +4,28 @@
 
 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.
 
+---
+
+## 🚀 5 Dakikada Başla (Remote MCP)
+
+### ✅ Kurulum Gerektirmez! Hemen Kullan!
+
+🔗 **Remote MCP Adresi:** `https://yargimcp.fastmcp.app/mcp`
+
+### Claude Desktop ile Kullanım
+
+1. **Claude Desktop'ı açın**
+2. **Settings → Connectors → Add Custom Connector**
+3. **Bilgileri girin:**
+   - **Name:** `Yargı MCP`
+   - **URL:** `https://yargimcp.fastmcp.app/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!
+
+---
+
 ![örnek](./ornek.png)
 
 🎯 **Temel Özellikler**
@@ -132,10 +154,65 @@ Yargı MCP'yi Gemini CLI ile kullanmak için:
 
 </details>
 
+---
+<details>
+<summary>🧠 <strong>Semantik Arama (Opsiyonel - OpenRouter API)</strong></summary>
+
+Yargı MCP, **semantik arama** özelliği ile kararları anlamsal olarak sıralayabilir. Bu özellik opsiyoneldir ve `OPENROUTER_API_KEY` ayarlandığında otomatik olarak etkinleşir.
+
+### 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)
+
+### Claude Desktop için Yapılandırma
+```json
+{
+  "mcpServers": {
+    "Yargı MCP": {
+      "command": "uvx",
+      "args": ["yargi-mcp"],
+      "env": {
+        "OPENROUTER_API_KEY": "sk-or-v1-xxx..."
+      }
+    }
+  }
+}
+```
+
+### 5ire için Yapılandırma
+Tool ayarlarında **Environment Variables** alanına ekleyin:
+```
+OPENROUTER_API_KEY=sk-or-v1-xxx...
+```
+
+### Gemini CLI için Yapılandırma
+```json
+{
+  "mcpServers": {
+    "yargi_mcp": {
+      "command": "uvx",
+      "args": ["yargi-mcp"],
+      "env": {
+        "OPENROUTER_API_KEY": "sk-or-v1-xxx..."
+      }
+    }
+  }
+}
+```
+
+> 💡 **Not:** `OPENROUTER_API_KEY` ayarlanmazsa semantik arama aracı görünmez, diğer 19 araç normal şekilde çalışmaya devam eder.
+
+</details>
+
 <details>
 <summary>🛠️ <strong>Kullanılabilir Araçlar (MCP Tools)</strong></summary>
 
-Bu FastMCP sunucusu **19 optimize edilmiş MCP aracı** sunar (token verimliliği için optimize edilmiş):
+Bu FastMCP sunucusu **19 temel 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*
@@ -200,7 +277,7 @@ Bu FastMCP sunucusu **19 optimize edilmiş MCP aracı** sunar (token verimliliğ
 
 **GENEL İSTATİSTİKLER:**
 - **Toplam Mahkeme/Kurum:** 13 farklı hukuki kurum (KVKK dahil)
-- **Toplam MCP Tool:** 19 optimize edilmiş arama ve belge getirme aracı  
+- **Toplam MCP Tool:** 19 temel 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)

analyze_kik_hash_generation.py

@@ -0,0 +1,239 @@
+#!/usr/bin/env python3
+
+"""
+Analyze KİK v2 hash generation by examining JavaScript code patterns
+and trying to reverse engineer the hash generation logic.
+"""
+
+import asyncio
+import json
+import hashlib
+import hmac
+import base64
+from fastmcp import Client
+from mcp_server_main import app
+
+def analyze_webpack_hash_patterns():
+    """
+    Analyze the webpack JavaScript code you provided to find hash generation patterns
+    """
+    print("🔍 Analyzing webpack hash generation patterns...")
+    
+    # From the JavaScript code, I can see several hash/ID generation patterns:
+    hash_patterns = {
+        # Webpack chunk system hashes (from the JS code)
+        "webpack_chunks": {
+            315: "d9a9486a4f5ba326",
+            531: "cd8fb385c88033ae", 
+            671: "04c48b287646627a",
+            856: "682c9a7b87351f90",
+            1017: "9de022378fc275f6",
+            # ... many more from the __webpack_require__.u function
+        },
+        
+        # Symbol generation from Zone.js
+        "zone_symbols": [
+            "__zone_symbol__",
+            "__Zone_symbol_prefix",
+            "Zone.__symbol__"
+        ],
+        
+        # Angular module federation patterns
+        "module_federation": [
+            "__webpack_modules__",
+            "__webpack_module_cache__",
+            "__webpack_require__"
+        ]
+    }
+    
+    # The target hash format
+    target_hash = "42f9bcd59e0dfbca36dec9accf5686c7a92aa97724cd8fc3550beb84b80409da"
+    print(f"🎯 Target hash: {target_hash}")
+    print(f"   Length: {len(target_hash)} characters")
+    print(f"   Format: {'SHA256' if len(target_hash) == 64 else 'Other'} (64 chars = SHA256)")
+    
+    return hash_patterns
+
+def test_webpack_style_hashing(data_dict):
+    """Test webpack-style hash generation methods"""
+    hashes = {}
+    
+    for key, value in data_dict.items():
+        test_string = str(value)
+        
+        # Try various webpack-style hash methods
+        hashes[f"webpack_md5_{key}"] = hashlib.md5(test_string.encode()).hexdigest()
+        hashes[f"webpack_sha1_{key}"] = hashlib.sha1(test_string.encode()).hexdigest()  
+        hashes[f"webpack_sha256_{key}"] = hashlib.sha256(test_string.encode()).hexdigest()
+        
+        # Try with various prefixes/suffixes (common in webpack)
+        prefixed = f"__webpack__{test_string}"
+        hashes[f"webpack_prefixed_sha256_{key}"] = hashlib.sha256(prefixed.encode()).hexdigest()
+        
+        # Try with module federation style
+        module_style = f"shell:{test_string}"
+        hashes[f"module_fed_sha256_{key}"] = hashlib.sha256(module_style.encode()).hexdigest()
+        
+        # Try JSON stringified
+        json_style = json.dumps({"id": value, "type": "decision"}, separators=(',', ':'))
+        hashes[f"json_sha256_{key}"] = hashlib.sha256(json_style.encode()).hexdigest()
+        
+        # Try with timestamp or sequence
+        with_seq = f"{test_string}_0"
+        hashes[f"seq_sha256_{key}"] = hashlib.sha256(with_seq.encode()).hexdigest()
+        
+    return hashes
+
+def test_angular_routing_hashes(data_dict):
+    """Test Angular routing/state management hash generation"""
+    hashes = {}
+    
+    for key, value in data_dict.items():
+        # Angular often uses route parameters for hash generation
+        route_style = f"/kurul-kararlari/{value}"
+        hashes[f"route_sha256_{key}"] = hashlib.sha256(route_style.encode()).hexdigest()
+        
+        # Component state style
+        state_style = f"KurulKararGoster_{value}"
+        hashes[f"state_sha256_{key}"] = hashlib.sha256(state_style.encode()).hexdigest()
+        
+        # Angular module style
+        module_style = f"kik.kurul.karar.{value}"
+        hashes[f"module_sha256_{key}"] = hashlib.sha256(module_style.encode()).hexdigest()
+        
+    return hashes
+
+def test_base64_encoding_variants(data_dict):
+    """Test various base64 and encoding variants"""
+    hashes = {}
+    
+    for key, value in data_dict.items():
+        test_string = str(value)
+        
+        # Try base64 encoding then hashing
+        b64_encoded = base64.b64encode(test_string.encode()).decode()
+        hashes[f"b64_sha256_{key}"] = hashlib.sha256(b64_encoded.encode()).hexdigest()
+        
+        # Try URL-safe base64
+        b64_url = base64.urlsafe_b64encode(test_string.encode()).decode()
+        hashes[f"b64url_sha256_{key}"] = hashlib.sha256(b64_url.encode()).hexdigest()
+        
+        # Try hex encoding
+        hex_encoded = test_string.encode().hex()
+        hashes[f"hex_sha256_{key}"] = hashlib.sha256(hex_encoded.encode()).hexdigest()
+        
+    return hashes
+
+async def test_hash_generation_comprehensive():
+    print("🔐 Comprehensive KİK document hash generation analysis...")
+    print("=" * 70)
+    
+    # First analyze the webpack patterns
+    webpack_patterns = analyze_webpack_hash_patterns()
+    
+    client = Client(app)
+    
+    async with client:
+        print("✅ MCP client connected")
+        
+        # Get sample decisions 
+        print("\n📊 Getting sample decisions for hash analysis...")
+        search_result = await client.call_tool("search_kik_v2_decisions", {
+            "decision_type": "uyusmazlik",
+            "karar_metni": "2024"
+        })
+        
+        if hasattr(search_result, 'content') and search_result.content:
+            search_data = json.loads(search_result.content[0].text)
+            decisions = search_data.get('decisions', [])
+            
+            if decisions:
+                print(f"✅ Found {len(decisions)} decisions")
+                
+                # Test with first decision
+                sample_decision = decisions[0]
+                print(f"\n📋 Sample decision for hash analysis:")
+                for key, value in sample_decision.items():
+                    print(f"   {key}: {value}")
+                
+                target_hash = "42f9bcd59e0dfbca36dec9accf5686c7a92aa97724cd8fc3550beb84b80409da"
+                print(f"\n🎯 Target hash to match: {target_hash}")
+                
+                all_hashes = {}
+                
+                # Test different hash generation methods
+                print(f"\n🔨 Testing webpack-style hashing...")
+                webpack_hashes = test_webpack_style_hashing(sample_decision)
+                all_hashes.update(webpack_hashes)
+                
+                print(f"🔨 Testing Angular routing hashes...")  
+                angular_hashes = test_angular_routing_hashes(sample_decision)
+                all_hashes.update(angular_hashes)
+                
+                print(f"🔨 Testing base64 encoding variants...")
+                b64_hashes = test_base64_encoding_variants(sample_decision)
+                all_hashes.update(b64_hashes)
+                
+                # Check for matches
+                print(f"\n🎯 Checking for hash matches...")
+                matches_found = []
+                partial_matches = []
+                
+                for hash_name, hash_value in all_hashes.items():
+                    if hash_value == target_hash:
+                        matches_found.append((hash_name, hash_value))
+                        print(f"   🎉 EXACT MATCH FOUND: {hash_name}")
+                    elif hash_value[:8] == target_hash[:8]:  # First 8 chars match
+                        partial_matches.append((hash_name, hash_value))
+                        print(f"   🔍 Partial match (first 8): {hash_name} -> {hash_value[:16]}...")
+                    elif hash_value[-8:] == target_hash[-8:]:  # Last 8 chars match
+                        partial_matches.append((hash_name, hash_value))
+                        print(f"   🔍 Partial match (last 8): {hash_name} -> ...{hash_value[-16:]}")
+                
+                if not matches_found and not partial_matches:
+                    print(f"   ❌ No matches found")
+                    print(f"\n📝 Sample generated hashes (first 10):")
+                    for i, (hash_name, hash_value) in enumerate(list(all_hashes.items())[:10]):
+                        print(f"   {hash_name}: {hash_value}")
+                
+                # Try combinations with other decisions
+                print(f"\n🔄 Testing hash combinations with multiple decisions...")
+                if len(decisions) > 1:
+                    for i, decision in enumerate(decisions[1:3]):  # Test 2 more
+                        print(f"\n   Testing decision {i+2}: {decision.get('kararNo')}")
+                        decision_hashes = test_webpack_style_hashing(decision)
+                        
+                        for hash_name, hash_value in decision_hashes.items():
+                            if hash_value == target_hash:
+                                print(f"   🎉 MATCH FOUND in decision {i+2}: {hash_name}")
+                                matches_found.append((f"decision_{i+2}_{hash_name}", hash_value))
+                
+                # Try composite hashes (combining multiple fields)
+                print(f"\n🔗 Testing composite hash generation...")
+                composite_tests = [
+                    f"{sample_decision.get('gundemMaddesiId')}_{sample_decision.get('kararNo')}",
+                    f"{sample_decision.get('kararNo')}_{sample_decision.get('kararTarihi')}",
+                    f"uyusmazlik_{sample_decision.get('gundemMaddesiId')}_{sample_decision.get('kararTarihi')}",
+                    json.dumps(sample_decision, separators=(',', ':'), sort_keys=True),
+                    f"{sample_decision.get('basvuran')}_{sample_decision.get('gundemMaddesiId')}",
+                ]
+                
+                for i, composite_str in enumerate(composite_tests):
+                    composite_hash = hashlib.sha256(composite_str.encode()).hexdigest()
+                    if composite_hash == target_hash:
+                        print(f"   🎉 COMPOSITE MATCH FOUND: test_{i} -> {composite_str[:50]}...")
+                        matches_found.append((f"composite_{i}", composite_hash))
+                        
+                print(f"\n🎯 Hash analysis completed!")
+                print(f"   Total matches found: {len(matches_found)}")
+                print(f"   Partial matches: {len(partial_matches)}")
+                
+            else:
+                print("❌ No decisions found")
+        else:
+            print("❌ Search failed")
+    
+    print("=" * 70)
+
+if __name__ == "__main__":
+    asyncio.run(test_hash_generation_comprehensive())

asgi_app.py

@@ -3,7 +3,7 @@ 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 Stripe webhook integration.
+for OAuth integration and proper middleware support.
 
 Usage:
     uvicorn asgi_app:app --host 0.0.0.0 --port 8000
@@ -12,52 +12,92 @@ Usage:
 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
+from fastapi.responses import JSONResponse, HTMLResponse, Response
 from fastapi.exception_handlers import http_exception_handler
 from starlette.middleware import Middleware
 from starlette.middleware.cors import CORSMiddleware
-from starlette.responses import Response
+from starlette.middleware.base import BaseHTTPMiddleware
 
-# Import the fully configured MCP app with all tools
-from mcp_server_main import app as mcp_server
+# Import the proper create_app function that includes all middleware
+from mcp_server_main import create_app
 
-# Import Stripe webhook router
-from stripe_webhook import router as stripe_router
+# Conditional auth-related imports (only if auth enabled)
+_auth_check = os.getenv("ENABLE_AUTH", "false").lower() == "true"
 
-# Import simplified MCP Auth HTTP adapter
-from mcp_auth_http_simple import router as mcp_auth_router
+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://accounts.yargimcp.com")
-BASE_URL = os.getenv("BASE_URL", "https://yargimcp.com")
+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 middleware
+# Configure CORS and Auth middleware
 cors_origins = os.getenv("ALLOWED_ORIGINS", "*").split(",")
-custom_middleware = [
-    Middleware(
-        CORSMiddleware,
-        allow_origins=cors_origins,
-        allow_credentials=True,
-        allow_methods=["GET", "POST", "OPTIONS"],
-        allow_headers=["Content-Type", "Authorization", "X-Request-ID"],
-    ),
-]
 
-# Create MCP Starlette sub-application (without auth wrapper)
-mcp_app = mcp_server.http_app(
-    path="/",
-    middleware=custom_middleware
-)
+# Import FastMCP Bearer Auth Provider
+from fastmcp.server.auth import BearerAuthProvider
+from fastmcp.server.auth.providers.bearer import RSAKeyPair
+
+# 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
+mcp_app = mcp_server.http_app(path="/")
+
 
 # Configure JSON encoder for proper Turkish character support
-import json
-from fastapi.responses import JSONResponse
-
 class UTF8JSONResponse(JSONResponse):
     def __init__(self, content=None, status_code=200, headers=None, **kwargs):
         if headers is None:
@@ -74,21 +114,32 @@ class UTF8JSONResponse(JSONResponse):
             separators=(",", ":"),
         ).encode("utf-8")
 
-# Create FastAPI wrapper application with MCP lifespan
+custom_middleware = [
+    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"],
+    ),
+]
+
+# Create FastAPI wrapper application
 app = FastAPI(
     title="Yargı MCP Server",
     description="MCP server for Turkish legal databases with OAuth authentication",
     version="0.1.0",
     middleware=custom_middleware,
-    lifespan=mcp_app.lifespan,  # MCP app lifespan
-    default_response_class=UTF8JSONResponse  # Use UTF-8 JSON encoder
+    default_response_class=UTF8JSONResponse,  # Use UTF-8 JSON encoder
+    redirect_slashes=False  # Disable to prevent 307 redirects on /mcp endpoint
 )
 
-# Add Stripe webhook router to FastAPI
-app.include_router(stripe_router, prefix="/api")
+# Add auth-related routers to FastAPI (only if available)
+if stripe_router:
+    app.include_router(stripe_router, prefix="/api/stripe")
 
-# Add MCP Auth HTTP adapter to FastAPI (handles OAuth endpoints)
-app.include_router(mcp_auth_router)
+if mcp_auth_router:
+    app.include_router(mcp_auth_router)
 
 # Custom 401 exception handler for MCP spec compliance
 @app.exception_handler(401)
@@ -107,157 +158,110 @@ async def custom_401_handler(request: Request, exc: HTTPException):
     
     return response
 
-# Mount MCP app as sub-application at /mcp-server to avoid path conflicts
-app.mount("/mcp-server", mcp_app)
-
-# Add custom route to handle /mcp requests and forward to mounted app
-@app.api_route("/mcp", methods=["POST", "DELETE", "OPTIONS"])
-@app.api_route("/mcp/", methods=["POST", "DELETE", "OPTIONS"])
-async def mcp_protocol_handler(request: Request):
-    """Handle MCP protocol requests by forwarding to mounted app"""
-    
-    # Handle DELETE requests for session termination
-    if request.method == "DELETE":
-        logger.info("DELETE request received for session termination")
-        # For session termination, we just return 200 OK
-        # The actual session cleanup is handled by the underlying MCP transport
-        from starlette.responses import Response
-        return Response(
-            status_code=200,
-            content="Session terminated successfully"
-        )
-    
-    # REQUIRED: Validate Bearer JWT tokens for all MCP requests
-    auth_header = request.headers.get("Authorization")
-    if not auth_header or not auth_header.startswith("Bearer "):
-        logger.error("Missing or invalid Authorization header")
-        raise HTTPException(
-            status_code=401, 
-            detail="Missing or invalid Authorization header. Bearer token required."
-        )
-    
-    token = auth_header.split(" ")[1]
-    try:
-        # Check if this is a mock token for development/testing
-        if token.startswith("mock_clerk_jwt_"):
-            logger.info(f"Using mock JWT token for development: {token[:30]}...")
-            # For mock tokens, we'll allow access with a mock user
-            request.state.user_id = "mock_user_dev"
-            request.state.session_id = "mock_session_dev"
-            request.state.token_scopes = ["read", "search"]
-            logger.info("Mock JWT token accepted for development")
-        elif token.startswith("eyJ"):
-            # This looks like a real JWT token (starts with eyJ which is base64 encoded '{"')
-            logger.info(f"Processing real JWT token: {token[:30]}...")
-            # Validate real Clerk JWT token
-            from clerk_backend_api import Clerk, models
-            import jwt
-            
-            # Decode JWT token and extract user info
-            try:
-                decoded_token = jwt.decode(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"])
-                session_id = decoded_token.get("sid", "jwt_session")
-                
-                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 signed by Clerk and contains valid user info
-                    request.state.user_id = user_id
-                    request.state.user_email = user_email
-                    request.state.session_id = session_id
-                    request.state.token_scopes = token_scopes
-                    logger.info(f"Real JWT token accepted for user: {user_id}")
-                else:
-                    logger.error(f"Missing required fields in JWT token - user_id: {bool(user_id)}, email: {bool(user_email)}")
-                    raise HTTPException(
-                        status_code=401,
-                        detail="Invalid token - missing user_id or email in claims"
-                    )
-                    
-            except Exception as e:
-                logger.error(f"JWT token decoding failed: {e}")
-                raise HTTPException(
-                    status_code=401,
-                    detail="Invalid JWT token format"
-                )
-        else:
-            # Invalid token format - doesn't start with expected patterns
-            logger.error(f"Invalid token format: {token[:30]}...")
-            raise HTTPException(
-                status_code=401,
-                detail="Invalid token format - must be a valid JWT token"
-            )
-        
-    except HTTPException:
-        # Re-raise HTTPException as-is
-        raise
-    except Exception as e:
-        logger.error(f"Bearer token validation failed: {str(e)}")
-        raise HTTPException(
-            status_code=401,
-            detail=f"Token validation failed: {str(e)}"
-        )
-    
-    # Forward the request to the mounted MCP app
-    async def receive():
-        return await request.receive()
-    
-    # Create new scope for the mounted app
-    scope = request.scope.copy()
-    scope["path"] = "/"  # Root path for mounted app
-    scope["path_info"] = "/"
-    
-    # Capture the response
-    response_parts = {"status": 200, "headers": [], "body": b""}
-    
-    async def send(message):
-        if message["type"] == "http.response.start":
-            response_parts["status"] = message["status"]
-            response_parts["headers"] = message["headers"]
-        elif message["type"] == "http.response.body":
-            response_parts["body"] += message.get("body", b"")
-    
-    # Call the mounted MCP app
-    await mcp_app(scope, receive, send)
-    
-    # Return the response
-    from starlette.responses import Response
-    
-    # Convert ASGI headers to dict
-    headers = {}
-    for name, value in response_parts["headers"]:
-        headers[name.decode()] = value.decode()
-    
-    return Response(
-        content=response_parts["body"],
-        status_code=response_parts["status"],
-        headers=headers
-    )
-
-
-# SSE transport deprecated - removed
-
-
-# FastAPI health check endpoint
+# FastAPI health check endpoint - BEFORE mounting MCP app
 @app.get("/health")
 async def health_check():
     """Health check endpoint for monitoring"""
-    return JSONResponse({
+    return {
         "status": "healthy",
         "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 JSONResponse({
+    return {
         "service": "Yargı MCP Server",
         "description": "MCP server for Turkish legal databases with OAuth authentication",
         "endpoints": {
@@ -282,25 +286,26 @@ async def root():
             "Kamu İhale Kurulu (Public Procurement Authority)",
             "Rekabet Kurumu (Competition Authority)",
             "Sayıştay (Court of Accounts)",
+            "KVKK (Personal Data Protection Authority)",
+            "BDDK (Banking Regulation and Supervision Agency)",
             "Bedesten API (Multiple courts)"
         ],
         "authentication": {
             "enabled": os.getenv("ENABLE_AUTH", "false").lower() == "true",
             "type": "OAuth 2.0 via Clerk",
-            "issuer": os.getenv("CLERK_ISSUER", "https://clerk.accounts.dev"),
+            "issuer": CLERK_ISSUER,
             "providers": ["google"],
             "flow": "authorization_code"
         }
-    })
+    }
 
-# OAuth 2.0 Authorization Server Metadata proxy (for MCP clients that can't reach Clerk directly)
-# MCP Auth Toolkit expects this to be under /mcp/.well-known/oauth-authorization-server
-@app.get("/mcp/.well-known/oauth-authorization-server")
-async def oauth_authorization_server():
-    """OAuth 2.0 Authorization Server Metadata proxy to Clerk - MCP Auth Toolkit standard location"""
-    return JSONResponse({
-        "issuer": BASE_URL,
-        "authorization_endpoint": "https://yargimcp.com/mcp-callback",
+# 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"],
@@ -314,15 +319,15 @@ async def oauth_authorization_server():
         "service_documentation": f"{BASE_URL}/mcp",
         "registration_endpoint": f"{BASE_URL}/register",
         "resource_documentation": f"{BASE_URL}/mcp"
-    })
+    }
 
-# Claude AI MCP specific endpoint format
+# 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 JSONResponse({
-        "issuer": BASE_URL,
-        "authorization_endpoint": "https://yargimcp.com/mcp-callback",
+    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"],
@@ -336,12 +341,12 @@ async def oauth_authorization_server_mcp_suffix():
         "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 JSONResponse({
+    return {
         "resource": BASE_URL,
         "authorization_servers": [
             BASE_URL
@@ -350,78 +355,13 @@ async def oauth_protected_resource_mcp_suffix():
         "bearer_methods_supported": ["header"],
         "resource_documentation": f"{BASE_URL}/mcp",
         "resource_policy_uri": f"{BASE_URL}/privacy"
-    })
-
-# Keep root level for compatibility with some MCP clients
-@app.get("/.well-known/oauth-authorization-server")
-async def oauth_authorization_server_root():
-    """OAuth 2.0 Authorization Server Metadata proxy to Clerk - root level for compatibility"""
-    return JSONResponse({
-        "issuer": BASE_URL,
-        "authorization_endpoint": "https://yargimcp.com/mcp-callback",
-        "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"
-    })
-
-# MCP endpoint info for GET requests (ChatGPT compatibility)
-@app.get("/mcp")
-async def mcp_info():
-    """MCP endpoint information for discovery"""
-    return JSONResponse({
-        "mcp_server": True,
-        "name": "Yargı MCP Server",
-        "version": "0.1.0",
-        "description": "MCP server for Turkish legal databases",
-        "protocol": "mcp/1.0",
-        "transport": ["http"],
-        "authentication_required": True,
-        "authentication": {
-            "type": "oauth2",
-            "authorization_url": "https://yargimcp.com/sign-in?redirect_url=https://api.yargimcp.com/auth/mcp-callback",
-            "token_url": f"{BASE_URL}/auth/mcp-token",
-            "scopes": ["read", "search"],
-            "provider": "clerk"
-        },
-        "endpoints": {
-            "mcp_protocol": "/mcp",
-            "discovery": "/mcp/discovery",
-            "well_known": "/.well-known/mcp",
-            "health": "/health",
-            "oauth_login": "/auth/login"
-        },
-        "capabilities": {
-            "tools": True,
-            "resources": True,
-            "prompts": False
-        },
-        "tools_count": len(mcp_server._tool_manager._tools),
-        "usage": {
-            "note": "This is an MCP server. Use POST to /mcp/ with proper MCP protocol headers.",
-            "headers_required": [
-                "Content-Type: application/json",
-                "Accept: application/json",
-                "Authorization: Bearer <token>",
-                "X-Session-ID: <session-id>"
-            ]
-        }
-    })
+    }
 
 # 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 JSONResponse({
+    return {
         "resource": BASE_URL,
         "authorization_servers": [
             BASE_URL
@@ -430,13 +370,13 @@ async def oauth_protected_resource():
         "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 JSONResponse({
+    return {
         "mcp_server": {
             "name": "Yargı MCP Server",
             "version": "0.1.0",
@@ -449,13 +389,13 @@ async def well_known_mcp():
             "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 JSONResponse({
+    return {
         "name": "Yargı MCP Server",
         "description": "MCP server for Turkish legal databases",
         "version": "0.1.0",
@@ -465,7 +405,7 @@ async def mcp_discovery():
         "authentication": {
             "type": "oauth2",
             "authorization_url": "/auth/login",
-            "token_url": "/auth/callback",
+            "token_url": "/token",
             "scopes": ["read", "search"],
             "provider": "clerk"
         },
@@ -479,7 +419,7 @@ async def mcp_discovery():
             "url": BASE_URL,
             "email": "support@yargi-mcp.dev"
         }
-    })
+    }
 
 # FastAPI status endpoint
 @app.get("/status")
@@ -492,85 +432,54 @@ async def status():
             "description": tool.description[:100] + "..." if len(tool.description) > 100 else tool.description
         })
     
-    return JSONResponse({
+    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"
-    })
+    }
 
-# Note: JWT token validation is now handled entirely by Clerk
-# All authentication flows use Clerk JWT tokens directly
-
-async def validate_clerk_session(request: Request, clerk_token: str = None) -> str:
-    """Validate Clerk session from cookies or JWT token and return user_id"""
-    logger.info(f"Validating Clerk session - token provided: {bool(clerk_token)}")
+# 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:
-        # Try to import Clerk SDK
-        from clerk_backend_api import Clerk
-        clerk = Clerk(bearer_auth=os.getenv("CLERK_SECRET_KEY"))
+        # 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:
-            logger.info("Validating Clerk JWT token from URL parameter")
             try:
-                # 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")  # Use standard JWT 'sid' claim
-                
-                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
-                    
-                    if user_id:
-                        logger.info(f"JWT token validation successful - user_id: {user_id}")
-                        return user_id
-                    else:
-                        logger.error("JWT token validation failed - no user_id in session")
-                else:
-                    logger.error("No session_id found in JWT token")
+                return "oauth_user_from_token"
             except Exception as e:
-                logger.error(f"JWT token validation failed: {str(e)}")
-                # Fall through to cookie validation
-        
+                pass
+
         # Fallback to cookie validation
-        logger.info("Attempting cookie-based session validation")
         clerk_session = request.cookies.get("__session")
         if not clerk_session:
-            logger.error("No Clerk session cookie found")
             raise HTTPException(status_code=401, detail="No Clerk session found")
-        
+
         # Validate session with Clerk
         session = clerk.sessions.verify_session(clerk_session)
-        logger.info(f"Cookie session validation successful - user_id: {session.user_id}")
         return session.user_id
         
     except ImportError:
-        # Fallback for development without Clerk SDK
-        logger.warning("Clerk SDK not available - using development fallback")
         return "dev_user_123"
     except Exception as e:
-        logger.error(f"Session validation failed: {str(e)}")
-        raise HTTPException(status_code=401, detail=f"Session validation failed: {str(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"""
-    logger.info(f"MCP OAuth callback - clerk_token provided: {bool(clerk_token)}")
     
     try:
         # Validate Clerk session with JWT token support
-        user_id = await validate_clerk_session(request, clerk_token)
-        logger.info(f"User authenticated successfully - user_id: {user_id}")
-        
-        # Use the Clerk JWT token directly (no need to generate custom token)
-        logger.info("User authenticated successfully via Clerk")
+        user_id = await validate_clerk_session_for_oauth(request, clerk_token)
         
         # Return success response
         return HTMLResponse(f"""
@@ -606,7 +515,6 @@ async def mcp_oauth_callback(request: Request, clerk_token: str = Query(None)):
         """)
         
     except HTTPException as e:
-        logger.error(f"MCP OAuth callback failed: {e.detail}")
         return HTMLResponse(f"""
         <html>
             <head>
@@ -632,7 +540,6 @@ async def mcp_oauth_callback(request: Request, clerk_token: str = Query(None)):
         </html>
         """, status_code=e.status_code)
     except Exception as e:
-        logger.error(f"Unexpected error in MCP OAuth callback: {str(e)}")
         return HTMLResponse(f"""
         <html>
             <head>
@@ -657,22 +564,26 @@ 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(request)
+        user_id = await validate_clerk_session_for_oauth(request)
         
-        return JSONResponse({
+        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}
         )
 
-# Note: Only HTTP transport supported - SSE transport deprecated
+# Mount MCP app at /mcp/ with trailing slash
+app.mount("/mcp/", mcp_app)
+
+# Set the lifespan context after mounting
+app.router.lifespan_context = mcp_app.lifespan
 
 # Export for uvicorn
 __all__ = ["app"]

bedesten_mcp_module/client.py

@@ -102,8 +102,22 @@ class BedestenApiClient:
             response_json = response.json()
             doc_response = BedestenDocumentResponse(**response_json)
             
-            # Decode base64 content
-            content_bytes = base64.b64decode(doc_response.data.content)
+            # Add null safety checks for document data
+            if not hasattr(doc_response, 'data') or doc_response.data is None:
+                raise ValueError("Document response does not contain data")
+            
+            if not hasattr(doc_response.data, 'content') or doc_response.data.content is None:
+                raise ValueError("Document data does not contain content")
+                
+            if not hasattr(doc_response.data, 'mimeType') or doc_response.data.mimeType is None:
+                raise ValueError("Document data does not contain mimeType")
+            
+            # Decode base64 content with error handling
+            try:
+                content_bytes = base64.b64decode(doc_response.data.content)
+            except Exception as e:
+                raise ValueError(f"Failed to decode base64 content: {str(e)}")
+            
             mime_type = doc_response.data.mimeType
             
             logger.info(f"BedestenApiClient: Document mime type: {mime_type}")

bedesten_mcp_module/models.py

@@ -21,7 +21,7 @@ class BedestenSearchData(BaseModel):
     pageSize: int = Field(..., description="Results per page (1-10)")
     pageNumber: int = Field(..., description="Page number (1-indexed)")
     itemTypeList: List[str] = Field(..., description="Court type filter (YARGITAYKARARI/DANISTAYKARAR/YERELHUKUK/ISTINAFHUKUK/KYB)")
-    phrase: str = Field(..., description="Search phrase (use \"exact phrase\" for precise matching)")
+    phrase: str = Field(..., description="Search phrase. Supports: 'word', \"exact phrase\", +required, -exclude, AND/OR/NOT operators. No wildcards or regex.")
     birimAdi: BirimAdiEnum = Field("ALL", description="""
         Chamber filter (optional). Abbreviated values with Turkish names:
         • Yargıtay: H1-H23 (1-23. Hukuk Dairesi), C1-C23 (1-23. Ceza Dairesi), HGK (Hukuk Genel Kurulu), CGK (Ceza Genel Kurulu), BGK (Büyük Genel Kurulu), HBK (Hukuk Daireleri Başkanlar Kurulu), CBK (Ceza Daireleri Başkanlar Kurulu)

fly-no-auth.toml

@@ -0,0 +1,46 @@
+# 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"

fly.toml

@@ -17,10 +17,16 @@ LOG_LEVEL = "info"
 [http_service]
   internal_port = 8000
   force_https = true
-  auto_stop_machines = 'stop'
+  auto_stop_machines = 'off'
   auto_start_machines = true
-  min_machines_running = 0
+  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'

kik_mcp_module/client_v2.py

@@ -0,0 +1,475 @@
+# kik_mcp_module/client_v2.py
+
+import httpx
+import logging
+import uuid
+import ssl
+import os
+from typing import Optional
+from datetime import datetime
+
+# Cryptography imports for AES-256-CBC encryption of document IDs
+try:
+    from cryptography.hazmat.primitives.ciphers import Cipher, algorithms, modes
+    from cryptography.hazmat.backends import default_backend
+    HAS_CRYPTOGRAPHY = True
+except ImportError:
+    HAS_CRYPTOGRAPHY = False
+
+from .models_v2 import (
+    KikV2DecisionType, KikV2SearchPayload, KikV2SearchPayloadDk, KikV2SearchPayloadMk,
+    KikV2RequestData, KikV2QueryRequest, KikV2KeyValuePair, 
+    KikV2SearchResponse, KikV2SearchResponseDk, KikV2SearchResponseMk,
+    KikV2SearchResult, KikV2CompactDecision, KikV2DocumentMarkdown
+)
+
+logger = logging.getLogger(__name__)
+
+class KikV2ApiClient:
+    """
+    New KIK v2 API Client for https://ekapv2.kik.gov.tr
+    
+    This client uses the modern JSON-based API endpoint that provides
+    better structured data compared to the legacy form-based API.
+    """
+    
+    BASE_URL = "https://ekapv2.kik.gov.tr"
+    
+    # Endpoint mappings for different decision types
+    ENDPOINTS = {
+        KikV2DecisionType.UYUSMAZLIK: "/b_ihalearaclari/api/KurulKararlari/GetKurulKararlari",
+        KikV2DecisionType.DUZENLEYICI: "/b_ihalearaclari/api/KurulKararlari/GetKurulKararlariDk",
+        KikV2DecisionType.MAHKEME: "/b_ihalearaclari/api/KurulKararlari/GetKurulKararlariMk"
+    }
+
+    # AES-256-CBC encryption key for document ID encryption (reverse engineered from ekapv2.kik.gov.tr Angular app)
+    # This key is used to encrypt numeric gundemMaddesiId values to 64-character hex hashes for document URLs
+    DOCUMENT_ID_ENCRYPTION_KEY = bytes([
+        236, 193, 164, 43, 12, 135, 121, 170, 4, 244, 123, 219, 82, 158, 124, 174,
+        174, 228, 219, 174, 208, 104, 174, 120, 32, 76, 250, 4, 143, 159, 211, 176
+    ])
+
+    @staticmethod
+    def encrypt_document_id(numeric_id: str) -> str:
+        """
+        Encrypt a numeric KİK gundemMaddesiId to the 64-character hex hash
+        used in document URLs.
+
+        Algorithm: AES-256-CBC with PKCS7 padding
+        Output format: IV (16 bytes hex) + Ciphertext (16 bytes hex) = 64 chars
+
+        Args:
+            numeric_id: The numeric document ID from search results (e.g., "177280")
+
+        Returns:
+            64-character hex string for use in document URL KararId parameter
+        """
+        if not HAS_CRYPTOGRAPHY:
+            raise ImportError("cryptography library required for document ID encryption")
+
+        # Generate random IV (16 bytes)
+        iv = os.urandom(16)
+
+        # Create AES-CBC cipher with the encryption key
+        cipher = Cipher(
+            algorithms.AES(KikV2ApiClient.DOCUMENT_ID_ENCRYPTION_KEY),
+            modes.CBC(iv),
+            backend=default_backend()
+        )
+        encryptor = cipher.encryptor()
+
+        # Encode plaintext and apply PKCS7 padding
+        plaintext = numeric_id.encode('utf-8')
+        block_size = 16
+        padding_len = block_size - (len(plaintext) % block_size)
+        padded_plaintext = plaintext + bytes([padding_len] * padding_len)
+
+        # Encrypt
+        ciphertext = encryptor.update(padded_plaintext) + encryptor.finalize()
+
+        # Return IV + ciphertext as lowercase hex (64 characters total)
+        return iv.hex() + ciphertext.hex()
+
+    def __init__(self, request_timeout: float = 60.0):
+        # Create SSL context with legacy server support
+        ssl_context = ssl.create_default_context()
+        ssl_context.check_hostname = False
+        ssl_context.verify_mode = ssl.CERT_NONE
+        
+        # Enable legacy server connect option for older SSL implementations (Python 3.12+)
+        if hasattr(ssl, 'OP_LEGACY_SERVER_CONNECT'):
+            ssl_context.options |= ssl.OP_LEGACY_SERVER_CONNECT
+        
+        # Set broader cipher suite support including legacy ciphers
+        ssl_context.set_ciphers('ALL:!aNULL:!eNULL:!EXPORT:!DES:!RC4:!MD5:!PSK:!SRP:!CAMELLIA')
+        
+        self.http_client = httpx.AsyncClient(
+            base_url=self.BASE_URL,
+            verify=ssl_context,
+            headers={
+                "Accept": "application/json",
+                "Accept-Language": "tr",
+                "Content-Type": "application/json",
+                "Origin": self.BASE_URL,
+                "Referer": f"{self.BASE_URL}/sorgulamalar/kurul-kararlari",
+                "Sec-Fetch-Dest": "empty",
+                "Sec-Fetch-Mode": "cors", 
+                "Sec-Fetch-Site": "same-origin",
+                "User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/139.0.0.0 Safari/537.36",
+                "api-version": "v1",
+                "sec-ch-ua": '"Not;A=Brand";v="99", "Google Chrome";v="139", "Chromium";v="139"',
+                "sec-ch-ua-mobile": "?0",
+                "sec-ch-ua-platform": '"macOS"'
+            },
+            timeout=request_timeout
+        )
+        
+        # Generate security headers (these might need to be updated based on API requirements)
+        self.security_headers = self._generate_security_headers()
+    
+    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 a random GUID for each session
+        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
+        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=="
+        }
+    
+    def _build_search_payload(self, 
+                             decision_type: KikV2DecisionType,
+                             karar_metni: str = "",
+                             karar_no: str = "",
+                             basvuran: str = "",
+                             idare_adi: str = "",
+                             baslangic_tarihi: str = "",
+                             bitis_tarihi: str = ""):
+        """Build the search payload for KIK v2 API."""
+        
+        key_value_pairs = []
+        
+        # Add non-empty search criteria
+        if karar_metni:
+            key_value_pairs.append(KikV2KeyValuePair(key="KararMetni", value=karar_metni))
+        
+        if karar_no:
+            key_value_pairs.append(KikV2KeyValuePair(key="KararNo", value=karar_no))
+            
+        if basvuran:
+            key_value_pairs.append(KikV2KeyValuePair(key="BasvuranAdi", value=basvuran))
+            
+        if idare_adi:
+            key_value_pairs.append(KikV2KeyValuePair(key="IdareAdi", value=idare_adi))
+            
+        if baslangic_tarihi:
+            key_value_pairs.append(KikV2KeyValuePair(key="BaslangicTarihi", value=baslangic_tarihi))
+            
+        if bitis_tarihi:
+            key_value_pairs.append(KikV2KeyValuePair(key="BitisTarihi", value=bitis_tarihi))
+        
+        # If no search criteria provided, use a generic search
+        if not key_value_pairs:
+            key_value_pairs.append(KikV2KeyValuePair(key="KararMetni", value=""))
+        
+        query_request = KikV2QueryRequest(keyValueOfstringanyType=key_value_pairs)
+        request_data = KikV2RequestData(keyValuePairs=query_request)
+        
+        # Return appropriate payload based on decision type
+        if decision_type == KikV2DecisionType.UYUSMAZLIK:
+            return KikV2SearchPayload(sorgulaKurulKararlari=request_data)
+        elif decision_type == KikV2DecisionType.DUZENLEYICI:
+            return KikV2SearchPayloadDk(sorgulaKurulKararlariDk=request_data)
+        elif decision_type == KikV2DecisionType.MAHKEME:
+            return KikV2SearchPayloadMk(sorgulaKurulKararlariMk=request_data)
+        else:
+            raise ValueError(f"Unsupported decision type: {decision_type}")
+    
+    async def search_decisions(self,
+                              decision_type: KikV2DecisionType = KikV2DecisionType.UYUSMAZLIK,
+                              karar_metni: str = "",
+                              karar_no: str = "",
+                              basvuran: str = "",
+                              idare_adi: str = "",
+                              baslangic_tarihi: str = "",
+                              bitis_tarihi: str = "") -> KikV2SearchResult:
+        """
+        Search KIK decisions using the v2 API.
+        
+        Args:
+            decision_type: Type of decision to search (uyusmazlik/duzenleyici/mahkeme)
+            karar_metni: Decision text search
+            karar_no: Decision number (e.g., "2025/UH.II-1801")
+            basvuran: Applicant name
+            idare_adi: Administration name
+            baslangic_tarihi: Start date (YYYY-MM-DD format)
+            bitis_tarihi: End date (YYYY-MM-DD format)
+            
+        Returns:
+            KikV2SearchResult with compact decision list
+        """
+        
+        logger.info(f"KikV2ApiClient: Searching {decision_type.value} decisions with criteria - karar_metni: '{karar_metni}', karar_no: '{karar_no}', basvuran: '{basvuran}'")
+        
+        try:
+            # Build request payload
+            payload = self._build_search_payload(
+                decision_type=decision_type,
+                karar_metni=karar_metni,
+                karar_no=karar_no,
+                basvuran=basvuran,
+                idare_adi=idare_adi,
+                baslangic_tarihi=baslangic_tarihi,
+                bitis_tarihi=bitis_tarihi
+            )
+            
+            # Update security headers for this request
+            headers = {**self.http_client.headers, **self._generate_security_headers()}
+            
+            # Get the appropriate endpoint for this decision type
+            endpoint = self.ENDPOINTS[decision_type]
+            
+            # Make API request
+            response = await self.http_client.post(
+                endpoint,
+                json=payload.model_dump(),
+                headers=headers
+            )
+            
+            response.raise_for_status()
+            response_data = response.json()
+            
+            logger.debug(f"KikV2ApiClient: Raw API response structure: {type(response_data)}")
+            
+            # Parse the API response based on decision type
+            if decision_type == KikV2DecisionType.UYUSMAZLIK:
+                api_response = KikV2SearchResponse(**response_data)
+                result_data = api_response.SorgulaKurulKararlariResponse.SorgulaKurulKararlariResult
+            elif decision_type == KikV2DecisionType.DUZENLEYICI:
+                api_response = KikV2SearchResponseDk(**response_data)
+                result_data = api_response.SorgulaKurulKararlariDkResponse.SorgulaKurulKararlariDkResult
+            elif decision_type == KikV2DecisionType.MAHKEME:
+                api_response = KikV2SearchResponseMk(**response_data)
+                result_data = api_response.SorgulaKurulKararlariMkResponse.SorgulaKurulKararlariMkResult
+            else:
+                raise ValueError(f"Unsupported decision type: {decision_type}")
+                
+            # Check for API errors
+            if result_data.hataKodu and result_data.hataKodu != "0":
+                logger.warning(f"KikV2ApiClient: API returned error - Code: {result_data.hataKodu}, Message: {result_data.hataMesaji}")
+                return KikV2SearchResult(
+                    decisions=[],
+                    total_records=0,
+                    page=1,
+                    error_code=result_data.hataKodu,
+                    error_message=result_data.hataMesaji
+                )
+            
+            # Convert to compact format
+            compact_decisions = []
+            total_count = 0
+            
+            for decision_group in result_data.KurulKararTutanakDetayListesi:
+                for decision_detail in decision_group.KurulKararTutanakDetayi:
+                    compact_decision = KikV2CompactDecision(
+                        kararNo=decision_detail.kararNo,
+                        kararTarihi=decision_detail.kararTarihi,
+                        basvuran=decision_detail.basvuran,
+                        idareAdi=decision_detail.idareAdi,
+                        basvuruKonusu=decision_detail.basvuruKonusu,
+                        gundemMaddesiId=decision_detail.gundemMaddesiId,
+                        decision_type=decision_type.value
+                    )
+                    compact_decisions.append(compact_decision)
+                    total_count += 1
+            
+            logger.info(f"KikV2ApiClient: Found {total_count} decisions")
+            
+            return KikV2SearchResult(
+                decisions=compact_decisions,
+                total_records=total_count,
+                page=1,
+                error_code="0",
+                error_message=""
+            )
+            
+        except httpx.HTTPStatusError as e:
+            logger.error(f"KikV2ApiClient: HTTP error during search: {e.response.status_code} - {e.response.text}")
+            return KikV2SearchResult(
+                decisions=[],
+                total_records=0, 
+                page=1,
+                error_code="HTTP_ERROR",
+                error_message=f"HTTP {e.response.status_code}: {e.response.text}"
+            )
+        except Exception as e:
+            logger.error(f"KikV2ApiClient: Unexpected error during search: {str(e)}")
+            return KikV2SearchResult(
+                decisions=[],
+                total_records=0,
+                page=1, 
+                error_code="UNEXPECTED_ERROR",
+                error_message=str(e)
+            )
+    
+    async def get_document_markdown(self, document_id: str) -> KikV2DocumentMarkdown:
+        """
+        Get KİK decision document content in Markdown format.
+        
+        This method uses a two-step process:
+        1. Call GetSorgulamaUrl endpoint to get the actual document URL
+        2. Use httpx to fetch the document content
+        
+        Args:
+            document_id: The gundemMaddesiId from search results
+            
+        Returns:
+            KikV2DocumentMarkdown with document content converted to Markdown
+        """
+        
+        logger.info(f"KikV2ApiClient: Getting document for ID: {document_id}")
+        
+        if not document_id or not document_id.strip():
+            return KikV2DocumentMarkdown(
+                document_id=document_id,
+                kararNo="",
+                markdown_content="",
+                source_url="",
+                error_message="Document ID is required"
+            )
+        
+        try:
+            # Step 1: Get the actual document URL using GetSorgulamaUrl endpoint
+            logger.info(f"KikV2ApiClient: Step 1 - Getting document URL for ID: {document_id}")
+            
+            # Update security headers for this request
+            headers = {**self.http_client.headers, **self._generate_security_headers()}
+            
+            # Call GetSorgulamaUrl to get the real document URL
+            url_payload = {"sorguSayfaTipi": 2}  # As shown in curl example
+            
+            url_response = await self.http_client.post(
+                "/b_ihalearaclari/api/KurulKararlari/GetSorgulamaUrl",
+                json=url_payload,
+                headers=headers
+            )
+            
+            url_response.raise_for_status()
+            url_data = url_response.json()
+            
+            # Get the base document URL from API response
+            base_document_url = url_data.get("sorgulamaUrl", "")
+            if not base_document_url:
+                return KikV2DocumentMarkdown(
+                    document_id=document_id,
+                    kararNo="",
+                    markdown_content="",
+                    source_url="",
+                    error_message="Could not get document URL from GetSorgulamaUrl API"
+                )
+            
+            # If document_id is numeric, encrypt it to get the KararId hash
+            # The web interface uses AES-256-CBC encrypted hashes for document URLs
+            karar_id = document_id
+            if document_id.isdigit():
+                try:
+                    karar_id = self.encrypt_document_id(document_id)
+                    logger.info(f"KikV2ApiClient: Encrypted numeric ID {document_id} to hash: {karar_id}")
+                except Exception as enc_error:
+                    logger.warning(f"KikV2ApiClient: Could not encrypt document ID, using as-is: {enc_error}")
+
+            # Construct full document URL with the encrypted KararId
+            document_url = f"{base_document_url}?KararId={karar_id}"
+            logger.info(f"KikV2ApiClient: Step 2 - Retrieved document URL: {document_url}")
+
+        except Exception as e:
+            logger.error(f"KikV2ApiClient: Error getting document URL for ID {document_id}: {str(e)}")
+            # Fallback to old method if GetSorgulamaUrl fails
+            # Also encrypt numeric IDs in fallback path
+            karar_id = document_id
+            if document_id.isdigit():
+                try:
+                    karar_id = self.encrypt_document_id(document_id)
+                    logger.info(f"KikV2ApiClient: Encrypted numeric ID in fallback: {karar_id}")
+                except Exception as enc_error:
+                    logger.warning(f"KikV2ApiClient: Could not encrypt in fallback: {enc_error}")
+            document_url = f"https://ekap.kik.gov.tr/EKAP/Vatandas/KurulKararGoster.aspx?KararId={karar_id}"
+            logger.info(f"KikV2ApiClient: Falling back to direct URL: {document_url}")
+        
+        try:
+            # Step 2: Use httpx to get the document content
+            logger.info(f"KikV2ApiClient: Step 2 - Using httpx to retrieve document from: {document_url}")
+
+            # Create a separate httpx client for document retrieval with HTML headers
+            doc_ssl_context = ssl.create_default_context()
+            doc_ssl_context.check_hostname = False
+            doc_ssl_context.verify_mode = ssl.CERT_NONE
+            if hasattr(ssl, 'OP_LEGACY_SERVER_CONNECT'):
+                doc_ssl_context.options |= ssl.OP_LEGACY_SERVER_CONNECT
+            doc_ssl_context.set_ciphers('ALL:!aNULL:!eNULL:!EXPORT:!DES:!RC4:!MD5:!PSK:!SRP:!CAMELLIA')
+
+            async with httpx.AsyncClient(
+                verify=doc_ssl_context,
+                headers={
+                    "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
+                    "Accept-Language": "tr,en-US;q=0.5",
+                    "User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/139.0.0.0 Safari/537.36"
+                },
+                timeout=60.0,
+                follow_redirects=True
+            ) as doc_client:
+                response = await doc_client.get(document_url)
+                response.raise_for_status()
+                html_content = response.text
+                logger.info(f"KikV2ApiClient: Retrieved content via httpx, length: {len(html_content)}")
+            
+            # Convert HTML to Markdown using MarkItDown with BytesIO
+            try:
+                from markitdown import MarkItDown
+                from io import BytesIO
+                
+                md = MarkItDown()
+                html_bytes = html_content.encode('utf-8')
+                html_stream = BytesIO(html_bytes)
+                
+                result = md.convert_stream(html_stream, file_extension=".html")
+                markdown_content = result.text_content
+                
+                return KikV2DocumentMarkdown(
+                    document_id=document_id,
+                    kararNo="",
+                    markdown_content=markdown_content,
+                    source_url=document_url,
+                    error_message=""
+                )
+                
+            except ImportError:
+                return KikV2DocumentMarkdown(
+                    document_id=document_id,
+                    kararNo="",
+                    markdown_content="MarkItDown library not available",
+                    source_url=document_url,
+                    error_message="MarkItDown library not installed"
+                )
+                
+        except Exception as e:
+            logger.error(f"KikV2ApiClient: Error retrieving document {document_id}: {str(e)}")
+            return KikV2DocumentMarkdown(
+                document_id=document_id,
+                kararNo="",
+                markdown_content="",
+                source_url=document_url,
+                error_message=str(e)
+            )
+    
+    async def close_client_session(self):
+        """Close HTTP client session."""
+        await self.http_client.aclose()
+        logger.info("KikV2ApiClient: HTTP client session closed.")

kik_mcp_module/models.py

@@ -1,74 +0,0 @@
-# kik_mcp_module/models.py
-from pydantic import BaseModel, Field, HttpUrl, computed_field, ConfigDict
-from typing import List, Optional
-from enum import Enum
-import base64 # Base64 encoding/decoding için
-
-class KikKararTipi(str, Enum):
-    """Enum for KIK (Public Procurement Authority) Decision Types."""
-    UYUSMAZLIK = "rbUyusmazlik"
-    DUZENLEYICI = "rbDuzenleyici"
-    MAHKEME = "rbMahkeme"
-
-class KikSearchRequest(BaseModel):
-    """Model for KIK Decision search criteria."""
-    karar_tipi: KikKararTipi = Field(KikKararTipi.UYUSMAZLIK, description="Type")
-    karar_no: str = Field("", description="No")
-    karar_tarihi_baslangic: str = Field("", description="Start", pattern=r"^\d{2}\.\d{2}\.\d{4}$|^$")
-    karar_tarihi_bitis: str = Field("", description="End", pattern=r"^\d{2}\.\d{2}\.\d{4}$|^$")
-    resmi_gazete_sayisi: str = Field("", description="Gazette")
-    resmi_gazete_tarihi: str = Field("", description="Date", pattern=r"^\d{2}\.\d{2}\.\d{4}$|^$")
-    basvuru_konusu_ihale: str = Field("", description="Subject")
-    basvuru_sahibi: str = Field("", description="Applicant")
-    ihaleyi_yapan_idare: str = Field("", description="Entity")
-    yil: str = Field("", description="Year")
-    karar_metni: str = Field("", description="Text")
-    page: int = Field(1, ge=1, description="Page")
-
-class KikDecisionEntry(BaseModel):
-    """Represents a single decision entry from KIK search results."""
-    preview_event_target: str = Field(..., description="Event target")
-    karar_no_str: str = Field(..., alias="kararNo", description="Decision number")
-    karar_tipi: KikKararTipi = Field(..., description="Decision type")
-    
-    karar_tarihi_str: str = Field(..., alias="kararTarihi", description="Date")
-    idare_str: str = Field("", alias="idare", description="Entity")
-    basvuru_sahibi_str: str = Field("", alias="basvuruSahibi", description="Applicant")
-    ihale_konusu_str: str = Field("", alias="ihaleKonusu", description="Subject")
-
-    @computed_field
-    @property
-    def karar_id(self) -> str:
-        """
-        A Base64 encoded unique ID for the decision, combining decision type and number.
-        Format before encoding: "{karar_tipi.value}|{karar_no_str}"
-        """
-        combined_key = f"{self.karar_tipi.value}|{self.karar_no_str}"
-        return base64.b64encode(combined_key.encode('utf-8')).decode('utf-8')
-
-    model_config = ConfigDict(populate_by_name=True)
-
-class KikSearchResult(BaseModel):
-    """Model for KIK search results."""
-    decisions: List[KikDecisionEntry]
-    total_records: int = 0
-    current_page: int = 1
-
-class KikDocumentMarkdown(BaseModel):
-    """
-    KIK decision document, with Markdown content potentially paginated.
-    """
-    retrieved_with_karar_id: Optional[str] = Field(None, description="Request ID")
-    retrieved_karar_no: Optional[str] = Field(None, description="Decision number")
-    retrieved_karar_tipi: Optional[KikKararTipi] = Field(None, description="Decision type")
-    
-    karar_id_param_from_url: Optional[str] = Field(None, alias="kararIdParam", description="Internal ID")
-    markdown_chunk: Optional[str] = Field(None, description="Content")
-    source_url: Optional[str] = Field(None, description="Source URL")
-    error_message: Optional[str] = Field(None, description="Error")
-    current_page: int = Field(1, description="Page")
-    total_pages: int = Field(1, description="Total pages")
-    is_paginated: bool = Field(False, description="Paginated")
-    full_content_char_count: Optional[int] = Field(None, description="Char count")
-
-    model_config = ConfigDict(populate_by_name=True)

kik_mcp_module/models_v2.py

@@ -0,0 +1,147 @@
+# kik_mcp_module/models_v2.py
+from pydantic import BaseModel, Field, ConfigDict
+from typing import List, Optional
+from datetime import datetime
+from enum import Enum
+
+# New KIK v2 API Models
+
+class KikV2DecisionType(str, Enum):
+    """KIK v2 Decision Types with corresponding endpoints."""
+    UYUSMAZLIK = "uyusmazlik"        # Disputes - GetKurulKararlari
+    DUZENLEYICI = "duzenleyici"      # Regulatory - GetKurulKararlariDk  
+    MAHKEME = "mahkeme"              # Court - GetKurulKararlariMk
+
+class KikV2SearchRequest(BaseModel):
+    """Model for KIK v2 API search request."""
+    KararMetni: str = Field("", description="Decision text search query")
+    KararNo: str = Field("", description="Decision number (e.g., '2025/UH.II-1801')")
+    BasvuranAdi: str = Field("", description="Applicant name")
+    IdareAdi: str = Field("", description="Administration name")
+    BaslangicTarihi: str = Field("", description="Start date (YYYY-MM-DD)")
+    BitisTarihi: str = Field("", description="End date (YYYY-MM-DD)")
+    
+class KikV2KeyValuePair(BaseModel):
+    """Key-value pair for KIK v2 API request."""
+    key: str
+    value: str
+
+class KikV2QueryRequest(BaseModel):
+    """Nested query structure for KIK v2 API."""
+    keyValueOfstringanyType: List[KikV2KeyValuePair]
+
+class KikV2RequestData(BaseModel):
+    """Main request data structure for KIK v2 API."""
+    keyValuePairs: KikV2QueryRequest
+
+# Request Payloads for different decision types
+class KikV2SearchPayload(BaseModel):
+    """Complete payload for KIK v2 API search - Uyuşmazlık (Disputes)."""
+    sorgulaKurulKararlari: KikV2RequestData
+
+class KikV2SearchPayloadDk(BaseModel):
+    """Complete payload for KIK v2 API search - Düzenleyici (Regulatory)."""
+    sorgulaKurulKararlariDk: KikV2RequestData
+
+class KikV2SearchPayloadMk(BaseModel):
+    """Complete payload for KIK v2 API search - Mahkeme (Court)."""
+    sorgulaKurulKararlariMk: KikV2RequestData
+
+# Response Models
+
+class KikV2DecisionDetail(BaseModel):
+    """Individual decision detail from KIK v2 API response."""
+    resmiGazeteMukerrerSayi: str = Field("", description="Official Gazette duplicate number")
+    itiraz: str = Field("", description="Objection")
+    yayinlanmaTarihi: str = Field("", description="Publication date")
+    idareAdi: str = Field("", description="Administration name")
+    uzmanTCKN: str = Field("", description="Expert TCKN")
+    resmiGazeteTarihi: str = Field("", description="Official Gazette date")
+    basvuruKonusu: str = Field("", description="Application subject")
+    kararTurKod: str = Field("", description="Decision type code")
+    kararTurAciklama: str = Field("", description="Decision type description")
+    karar: str = Field("", description="Decision text")
+    kararNo: str = Field("", description="Decision number")
+    resmiGazeteSayisi: str = Field("", description="Official Gazette number")
+    inceleme: str = Field("", description="Review")
+    basvuruTarihi: str = Field("", description="Application date")
+    kararNitelikKod: str = Field("", description="Decision nature code")
+    resmiGazeteMukerrer: str = Field("", description="Official Gazette duplicate")
+    basvuruSayisi: str = Field("", description="Application number")
+    basvuran: str = Field("", description="Applicant")
+    kararNitelik: str = Field("", description="Decision nature")
+    uyusmazlikKararNo: str = Field("", description="Dispute decision number")
+    kurulNo: str = Field("", description="Board number")
+    gundemMaddesiSiraNo: str = Field("", description="Agenda item sequence")
+    kararTarihi: str = Field("", description="Decision date (ISO format)")
+    dosyaBirimKodu: str = Field("", description="File unit code")
+    gundemMaddesiId: str = Field("", description="Agenda item ID")
+
+class KikV2DecisionGroup(BaseModel):
+    """Group of decision details."""
+    KurulKararTutanakDetayi: List[KikV2DecisionDetail] = Field(alias="kurulKararTutanakDetayi")
+    
+    model_config = ConfigDict(populate_by_name=True)
+
+class KikV2SearchResultData(BaseModel):
+    """Search result data structure."""
+    hataKodu: str = Field("", description="Error code")
+    hataMesaji: str = Field("", description="Error message") 
+    KurulKararTutanakDetayListesi: List[KikV2DecisionGroup]
+    
+    model_config = ConfigDict(populate_by_name=True)
+
+class KikV2SearchResultWrapper(BaseModel):
+    """Wrapper for search result."""
+    SorgulaKurulKararlariResult: KikV2SearchResultData
+
+# Base Response Models
+class KikV2SearchResponse(BaseModel):
+    """Complete KIK v2 API search response for Uyuşmazlık (Disputes)."""
+    SorgulaKurulKararlariResponse: KikV2SearchResultWrapper
+
+# Düzenleyici Kararlar (Regulatory Decisions) Response Models
+class KikV2SearchResultWrapperDk(BaseModel):
+    """Wrapper for regulatory decisions search result."""
+    SorgulaKurulKararlariDkResult: KikV2SearchResultData
+
+class KikV2SearchResponseDk(BaseModel):
+    """Complete KIK v2 API search response for Düzenleyici (Regulatory) decisions."""
+    SorgulaKurulKararlariDkResponse: KikV2SearchResultWrapperDk
+
+# Mahkeme Kararlar (Court Decisions) Response Models  
+class KikV2SearchResultWrapperMk(BaseModel):
+    """Wrapper for court decisions search result."""
+    SorgulaKurulKararlariMkResult: KikV2SearchResultData
+
+class KikV2SearchResponseMk(BaseModel):
+    """Complete KIK v2 API search response for Mahkeme (Court) decisions."""
+    SorgulaKurulKararlariMkResponse: KikV2SearchResultWrapperMk
+
+# Simplified Models for MCP Tools
+
+class KikV2CompactDecision(BaseModel):
+    """Compact decision format for MCP tool responses."""
+    kararNo: str = Field("", description="Decision number")
+    kararTarihi: str = Field("", description="Decision date")
+    basvuran: str = Field("", description="Applicant")
+    idareAdi: str = Field("", description="Administration")
+    basvuruKonusu: str = Field("", description="Application subject")
+    gundemMaddesiId: str = Field("", description="Document ID for retrieval")
+    decision_type: str = Field("", description="Decision type (uyusmazlik/duzenleyici/mahkeme)")
+    
+class KikV2SearchResult(BaseModel):
+    """Compact search results for MCP tools."""
+    decisions: List[KikV2CompactDecision]
+    total_records: int = Field(0, description="Total number of decisions found")
+    page: int = Field(1, description="Current page number")
+    error_code: str = Field("", description="API error code")
+    error_message: str = Field("", description="API error message")
+
+class KikV2DocumentMarkdown(BaseModel):
+    """Document content in Markdown format."""
+    document_id: str = Field("", description="Document ID")
+    kararNo: str = Field("", description="Decision number")
+    markdown_content: str = Field("", description="Decision content in Markdown")
+    source_url: str = Field("", description="Source URL")
+    error_message: str = Field("", description="Error message if retrieval failed")

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "yargi-mcp"
-version = "0.1.5"
+version = "0.2.0"
 description = "MCP Server For Turkish Legal Databases"
 readme = "README.md"
 requires-python = ">=3.11"
@@ -25,12 +25,12 @@ dependencies = [
     "markitdown[pdf]>=0.1.1",
     "pydantic>=2.11.4",
     "aiohttp>=3.11.18",
-    "playwright>=1.52.0",
     "fastmcp>=2.10.5",
     "pypdf>=5.5.0",
     "fastapi>=0.115.14",
-    "PyJWT>=2.8.0",
-    "tiktoken>=0.5.0",
+    "cryptography>=44.0.0",
+    "openai>=1.0.0",
+    "numpy>=1.24.0",
 ]
 
 [project.optional-dependencies]
@@ -50,6 +50,8 @@ 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]
@@ -59,7 +61,7 @@ yargi-mcp = "mcp_server_main:main"
 py-modules = ["mcp_server_main", "mcp_auth_factory", "mcp_auth_http_adapter", "asgi_app", "fastapi_app", "starlette_app", "run_asgi", "stripe_webhook"]
 
 [tool.setuptools.packages.find]
-include = ["*_mcp_module", "mcp_auth"]
+include = ["*_mcp_module", "mcp_auth", "semantic_search"]
 
 [build-system]
 requires = ["setuptools>=65.0", "wheel"]

requirements.txt

@@ -1,11 +0,0 @@
-fastmcp
-httpx
-beautifulsoup4
-markitdown[pdf]
-pydantic
-aiohttp
-playwright
-pypdf
-fastapi>=0.115.14
-uvicorn[standard]>=0.30.0
-starlette>=0.37.0

semantic_search/__init__.py

@@ -0,0 +1,7 @@
+# semantic_search/__init__.py
+
+from .embedder import OpenRouterEmbedder, is_openrouter_available
+from .vector_store import VectorStore
+from .processor import DocumentProcessor
+
+__all__ = ['OpenRouterEmbedder', 'is_openrouter_available', 'VectorStore', 'DocumentProcessor']

semantic_search/embedder.py

@@ -0,0 +1,154 @@
+# semantic_search/embedder.py
+
+import logging
+import os
+from typing import List, Optional
+import numpy as np
+
+logger = logging.getLogger(__name__)
+
+
+def is_openrouter_available() -> bool:
+    """Check if OpenRouter API key is available."""
+    return bool(os.getenv("OPENROUTER_API_KEY"))
+
+
+class OpenRouterEmbedder:
+    """
+    Embedder using OpenRouter API with Google's Gemini Embedding model.
+    Requires OPENROUTER_API_KEY environment variable.
+    """
+
+    def __init__(self):
+        """
+        Initialize OpenRouter Embedder.
+
+        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}")
+
+    def encode_query(self, query: str, task: str = "search result") -> np.ndarray:
+        """
+        Encode a search query.
+
+        Args:
+            query: The search query text
+            task: Task type for prompt template
+
+        Returns:
+            Numpy array of embeddings (3072 dimensions)
+        """
+        # Apply query prompt template
+        text = f"task: {task} | query: {query}"
+
+        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",
+                }
+            )
+
+            embedding = np.array(response.data[0].embedding, dtype=np.float32)
+
+            # L2 normalize for cosine similarity
+            norm = np.linalg.norm(embedding)
+            if norm > 0:
+                embedding = embedding / norm
+
+            logger.debug(f"Encoded query: {query[:50]}... -> shape: {embedding.shape}")
+            return embedding
+
+        except Exception as e:
+            logger.error(f"Failed to encode query: {e}")
+            raise
+
+    def encode_documents(self, documents: List[str], titles: Optional[List[str]] = None) -> np.ndarray:
+        """
+        Encode multiple documents with batch API call.
+
+        Args:
+            documents: List of document texts
+            titles: Optional list of document titles
+
+        Returns:
+            Numpy array of embeddings (N x 3072 dimensions)
+        """
+        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)
+
+        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",
+                }
+            )
+
+            # Extract embeddings in order
+            embeddings = np.array(
+                [d.embedding for d in sorted(response.data, key=lambda x: x.index)],
+                dtype=np.float32
+            )
+
+            # L2 normalize each embedding for cosine similarity
+            norms = np.linalg.norm(embeddings, axis=1, keepdims=True)
+            embeddings = embeddings / (norms + 1e-8)
+
+            logger.info(f"Encoded {len(documents)} documents -> shape: {embeddings.shape}")
+            return embeddings
+
+        except Exception as e:
+            logger.error(f"Failed to encode documents: {e}")
+            raise
+
+    def compute_similarity(self, query_embedding: np.ndarray, document_embeddings: np.ndarray) -> np.ndarray:
+        """
+        Compute cosine similarity between query and documents.
+
+        Args:
+            query_embedding: Query embedding (3072,)
+            document_embeddings: Document embeddings (N x 3072)
+
+        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)
+        similarities = np.dot(document_embeddings, query_embedding.T).squeeze()
+
+        return similarities

semantic_search/processor.py

@@ -0,0 +1,305 @@
+# semantic_search/processor.py
+
+import logging
+import re
+from typing import List, Dict, Any, Optional
+from dataclasses import dataclass
+import hashlib
+
+logger = logging.getLogger(__name__)
+
+@dataclass
+class DocumentChunk:
+    """Represents a chunk of a document."""
+    chunk_id: str
+    document_id: str
+    text: str
+    metadata: Dict[str, Any]
+    chunk_index: int
+    total_chunks: int
+
+class DocumentProcessor:
+    """
+    Processes legal documents for semantic search.
+    Handles chunking, cleaning, and metadata extraction.
+    """
+    
+    def __init__(self, 
+                 chunk_size: int = 1000,
+                 chunk_overlap: int = 200,
+                 min_chunk_size: int = 100):
+        """
+        Initialize document processor.
+        
+        Args:
+            chunk_size: Target size for each chunk in characters
+            chunk_overlap: Number of overlapping characters between chunks
+            min_chunk_size: Minimum chunk size to keep
+        """
+        self.chunk_size = chunk_size
+        self.chunk_overlap = chunk_overlap
+        self.min_chunk_size = min_chunk_size
+        
+        logger.info(f"Initialized DocumentProcessor (chunk_size={chunk_size}, overlap={chunk_overlap})")
+    
+    def process_document(self, 
+                        document_id: str,
+                        text: str,
+                        metadata: Optional[Dict[str, Any]] = None) -> List[DocumentChunk]:
+        """
+        Process a single document into chunks.
+        
+        Args:
+            document_id: Unique document identifier
+            text: Document text content
+            metadata: Optional document metadata
+            
+        Returns:
+            List of document chunks
+        """
+        if not text or len(text.strip()) < self.min_chunk_size:
+            logger.warning(f"Document {document_id} too short to process")
+            return []
+        
+        # Clean text
+        cleaned_text = self._clean_text(text)
+        
+        # Extract metadata from text if not provided
+        if metadata is None:
+            metadata = {}
+        
+        # Add extracted metadata
+        extracted_metadata = self._extract_metadata(cleaned_text)
+        metadata.update(extracted_metadata)
+        
+        # Create chunks
+        chunks = self._create_chunks(cleaned_text)
+        
+        # Create DocumentChunk objects
+        document_chunks = []
+        for i, chunk_text in enumerate(chunks):
+            chunk_id = self._generate_chunk_id(document_id, i)
+            
+            chunk = DocumentChunk(
+                chunk_id=chunk_id,
+                document_id=document_id,
+                text=chunk_text,
+                metadata={
+                    **metadata,
+                    'chunk_index': i,
+                    'total_chunks': len(chunks)
+                },
+                chunk_index=i,
+                total_chunks=len(chunks)
+            )
+            document_chunks.append(chunk)
+        
+        logger.info(f"Processed document {document_id} into {len(chunks)} chunks")
+        return document_chunks
+    
+    def _clean_text(self, text: str) -> str:
+        """
+        Clean and normalize text for processing.
+        
+        Args:
+            text: Raw text
+            
+        Returns:
+            Cleaned text
+        """
+        # Remove excessive whitespace
+        text = re.sub(r'\s+', ' ', text)
+        
+        # Remove special characters but keep Turkish characters
+        # Keep: letters, numbers, spaces, and common punctuation
+        text = re.sub(r'[^\w\s\.\,\;\:\!\?\-\(\)\"\'ÇĞIİÖŞÜçğıiöşü]', ' ', text)
+        
+        # Remove multiple spaces
+        text = re.sub(r' +', ' ', text)
+        
+        # Trim
+        text = text.strip()
+        
+        return text
+    
+    def _extract_metadata(self, text: str) -> Dict[str, Any]:
+        """
+        Extract metadata from legal document text.
+        
+        Args:
+            text: Document text
+            
+        Returns:
+            Extracted metadata
+        """
+        metadata = {}
+        
+        # Extract case numbers (Esas/Karar)
+        esas_pattern = r'E(?:sas)?[\s\.\:]*(\d{4})[\/\-](\d+)'
+        karar_pattern = r'K(?:arar)?[\s\.\:]*(\d{4})[\/\-](\d+)'
+        
+        esas_match = re.search(esas_pattern, text[:500])  # Look in first 500 chars
+        if esas_match:
+            metadata['esas_no'] = f"E.{esas_match.group(1)}/{esas_match.group(2)}"
+        
+        karar_match = re.search(karar_pattern, text[:500])
+        if karar_match:
+            metadata['karar_no'] = f"K.{karar_match.group(1)}/{karar_match.group(2)}"
+        
+        # Extract dates (DD.MM.YYYY or DD/MM/YYYY format)
+        date_pattern = r'(\d{1,2})[\.\/](\d{1,2})[\.\/](\d{4})'
+        dates = re.findall(date_pattern, text[:1000])  # Look in first 1000 chars
+        if dates:
+            # Take the first date as decision date
+            day, month, year = dates[0]
+            metadata['karar_tarihi'] = f"{year}-{month.zfill(2)}-{day.zfill(2)}"
+        
+        # Extract court/chamber name
+        chamber_patterns = [
+            r'(\d+)\.\s*Hukuk\s+Dairesi',
+            r'(\d+)\.\s*Ceza\s+Dairesi',
+            r'Hukuk\s+Genel\s+Kurulu',
+            r'Ceza\s+Genel\s+Kurulu',
+            r'(\d+)\.\s*Daire'
+        ]
+        
+        for pattern in chamber_patterns:
+            match = re.search(pattern, text[:500], re.IGNORECASE)
+            if match:
+                metadata['chamber'] = match.group(0)
+                break
+        
+        return metadata
+    
+    def _create_chunks(self, text: str) -> List[str]:
+        """
+        Create overlapping chunks from text.
+        
+        Args:
+            text: Cleaned document text
+            
+        Returns:
+            List of text chunks
+        """
+        chunks = []
+        
+        # Split by sentences for better semantic coherence
+        sentences = self._split_sentences(text)
+        
+        current_chunk = []
+        current_size = 0
+        
+        for sentence in sentences:
+            sentence_size = len(sentence)
+            
+            # If adding this sentence exceeds chunk size
+            if current_size + sentence_size > self.chunk_size and current_chunk:
+                # Save current chunk
+                chunk_text = ' '.join(current_chunk)
+                chunks.append(chunk_text)
+                
+                # Create overlap for next chunk
+                overlap_size = 0
+                overlap_sentences = []
+                
+                # Add sentences from the end until we reach overlap size
+                for sent in reversed(current_chunk):
+                    overlap_size += len(sent)
+                    overlap_sentences.insert(0, sent)
+                    if overlap_size >= self.chunk_overlap:
+                        break
+                
+                # Start new chunk with overlap
+                current_chunk = overlap_sentences
+                current_size = sum(len(s) for s in current_chunk)
+            
+            # Add sentence to current chunk
+            current_chunk.append(sentence)
+            current_size += sentence_size
+        
+        # Add final chunk if not empty
+        if current_chunk:
+            chunk_text = ' '.join(current_chunk)
+            if len(chunk_text) >= self.min_chunk_size:
+                chunks.append(chunk_text)
+        
+        return chunks
+    
+    def _split_sentences(self, text: str) -> List[str]:
+        """
+        Split text into sentences.
+        
+        Args:
+            text: Text to split
+            
+        Returns:
+            List of sentences
+        """
+        # Simple sentence splitting for Turkish text
+        # Split on period, question mark, exclamation, but not on abbreviations
+        
+        # Common Turkish abbreviations to preserve
+        abbreviations = ['Dr', 'Prof', 'Av', 'Md', 'Yrd', 'Doç', 'No', 'S', 'vs', 'vb', 'bkz']
+        
+        # Replace abbreviations temporarily
+        temp_text = text
+        replacements = {}
+        for i, abbr in enumerate(abbreviations):
+            placeholder = f"__ABBR{i}__"
+            temp_text = temp_text.replace(f"{abbr}.", placeholder)
+            replacements[placeholder] = f"{abbr}."
+        
+        # Split sentences
+        sentence_endings = re.compile(r'[.!?]+')
+        sentences = sentence_endings.split(temp_text)
+        
+        # Restore abbreviations and clean
+        cleaned_sentences = []
+        for sentence in sentences:
+            # Restore abbreviations
+            for placeholder, original in replacements.items():
+                sentence = sentence.replace(placeholder, original)
+            
+            # Clean and add if not empty
+            sentence = sentence.strip()
+            if sentence and len(sentence) > 10:  # Minimum sentence length
+                cleaned_sentences.append(sentence)
+        
+        return cleaned_sentences
+    
+    def _generate_chunk_id(self, document_id: str, chunk_index: int) -> str:
+        """
+        Generate unique chunk ID.
+        
+        Args:
+            document_id: Parent document ID
+            chunk_index: Index of chunk in document
+            
+        Returns:
+            Unique chunk ID
+        """
+        chunk_string = f"{document_id}_chunk_{chunk_index}"
+        chunk_hash = hashlib.md5(chunk_string.encode()).hexdigest()[:8]
+        return f"{document_id}_c{chunk_index}_{chunk_hash}"
+    
+    def combine_chunks(self, chunks: List[DocumentChunk]) -> str:
+        """
+        Combine chunks back into full document text.
+        
+        Args:
+            chunks: List of document chunks
+            
+        Returns:
+            Combined text
+        """
+        if not chunks:
+            return ""
+        
+        # Sort by chunk index
+        sorted_chunks = sorted(chunks, key=lambda x: x.chunk_index)
+        
+        # For overlapping chunks, we need to be careful about duplication
+        # Simple approach: just concatenate with space
+        combined = " ".join([chunk.text for chunk in sorted_chunks])
+        
+        return combined

semantic_search/vector_store.py

@@ -0,0 +1,235 @@
+# semantic_search/vector_store.py
+
+import logging
+import numpy as np
+from typing import List, Dict, Any, Tuple, Optional
+from dataclasses import dataclass
+import json
+
+logger = logging.getLogger(__name__)
+
+@dataclass
+class Document:
+    """Represents a document with its embedding and metadata."""
+    id: str
+    text: str
+    embedding: np.ndarray
+    metadata: Dict[str, Any]
+    
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary (excluding embedding for serialization)."""
+        return {
+            'id': self.id,
+            'text': self.text,
+            'metadata': self.metadata
+        }
+
+class VectorStore:
+    """
+    In-memory vector storage with similarity search capabilities.
+    Future versions can use Faiss, ChromaDB, or other vector databases.
+    """
+    
+    def __init__(self, dimension: int = 768):
+        """
+        Initialize vector store.
+        
+        Args:
+            dimension: Embedding dimension size
+        """
+        self.dimension = dimension
+        self.documents: List[Document] = []
+        self.embeddings: Optional[np.ndarray] = None
+        self.index_built = False
+        
+        logger.info(f"Initialized VectorStore with dimension: {dimension}")
+    
+    def add_documents(self, 
+                     ids: List[str],
+                     texts: List[str],
+                     embeddings: np.ndarray,
+                     metadata: Optional[List[Dict[str, Any]]] = None) -> int:
+        """
+        Add documents to the vector store.
+        
+        Args:
+            ids: Document IDs
+            texts: Document texts
+            embeddings: Document embeddings (N x dimension)
+            metadata: Optional metadata for each document
+            
+        Returns:
+            Number of documents added
+        """
+        if len(ids) != len(texts) or len(ids) != embeddings.shape[0]:
+            raise ValueError("Mismatched lengths for ids, texts, and embeddings")
+        
+        if metadata and len(metadata) != len(ids):
+            raise ValueError("Metadata length doesn't match document count")
+        
+        # Add documents
+        for i in range(len(ids)):
+            doc = Document(
+                id=ids[i],
+                text=texts[i],
+                embedding=embeddings[i],
+                metadata=metadata[i] if metadata else {}
+            )
+            self.documents.append(doc)
+        
+        # Rebuild index
+        self._build_index()
+        
+        logger.info(f"Added {len(ids)} documents to vector store. Total: {len(self.documents)}")
+        return len(ids)
+    
+    def _build_index(self):
+        """Build or rebuild the embedding index."""
+        if not self.documents:
+            self.embeddings = None
+            self.index_built = False
+            return
+        
+        # Stack all embeddings into a single array
+        self.embeddings = np.vstack([doc.embedding for doc in self.documents])
+        self.index_built = True
+        
+        logger.debug(f"Built index with shape: {self.embeddings.shape}")
+    
+    def search(self, 
+              query_embedding: np.ndarray,
+              top_k: int = 10,
+              threshold: Optional[float] = None) -> List[Tuple[Document, float]]:
+        """
+        Search for similar documents using cosine similarity.
+        
+        Args:
+            query_embedding: Query embedding vector
+            top_k: Number of results to return
+            threshold: Optional similarity threshold (0-1)
+            
+        Returns:
+            List of (Document, similarity_score) tuples
+        """
+        if not self.index_built or self.embeddings is None:
+            logger.warning("No documents in vector store")
+            return []
+        
+        # Ensure query is 2D
+        if len(query_embedding.shape) == 1:
+            query_embedding = query_embedding.reshape(1, -1)
+        
+        # Compute cosine similarities (assuming normalized embeddings)
+        similarities = np.dot(self.embeddings, query_embedding.T).squeeze()
+        
+        # Apply threshold if specified
+        if threshold is not None:
+            valid_indices = np.where(similarities >= threshold)[0]
+            if len(valid_indices) == 0:
+                logger.info(f"No documents above threshold {threshold}")
+                return []
+            similarities = similarities[valid_indices]
+            valid_docs = [self.documents[i] for i in valid_indices]
+        else:
+            valid_docs = self.documents
+        
+        # Get top-k indices
+        top_k = min(top_k, len(valid_docs))
+        if top_k == 0:
+            return []
+        
+        # Use argpartition for efficiency with large arrays
+        if len(similarities) > top_k:
+            top_indices = np.argpartition(similarities, -top_k)[-top_k:]
+            top_indices = top_indices[np.argsort(similarities[top_indices])[::-1]]
+        else:
+            top_indices = np.argsort(similarities)[::-1]
+        
+        # Create results
+        results = []
+        for idx in top_indices:
+            doc = valid_docs[idx] if threshold else self.documents[idx]
+            score = float(similarities[idx])
+            results.append((doc, score))
+        
+        logger.info(f"Search returned {len(results)} results (top_k={top_k})")
+        return results
+    
+    def hybrid_search(self,
+                     query_embedding: np.ndarray,
+                     keyword_scores: Dict[str, float],
+                     top_k: int = 10,
+                     alpha: float = 0.5) -> List[Tuple[Document, float]]:
+        """
+        Hybrid search combining vector similarity and keyword scores.
+        
+        Args:
+            query_embedding: Query embedding vector
+            keyword_scores: Document ID to keyword relevance score mapping
+            top_k: Number of results to return
+            alpha: Weight for vector similarity (1-alpha for keyword score)
+            
+        Returns:
+            List of (Document, combined_score) tuples
+        """
+        if not self.index_built:
+            logger.warning("No documents in vector store")
+            return []
+        
+        # Get vector similarities
+        vector_results = self.search(query_embedding, top_k=len(self.documents))
+        
+        # Combine scores
+        combined_scores = []
+        for doc, vector_score in vector_results:
+            keyword_score = keyword_scores.get(doc.id, 0.0)
+            # Normalize keyword score to 0-1 range if needed
+            if keyword_score > 1.0:
+                keyword_score = keyword_score / max(keyword_scores.values())
+            
+            combined_score = alpha * vector_score + (1 - alpha) * keyword_score
+            combined_scores.append((doc, combined_score))
+        
+        # Sort by combined score and return top-k
+        combined_scores.sort(key=lambda x: x[1], reverse=True)
+        results = combined_scores[:top_k]
+        
+        logger.info(f"Hybrid search returned {len(results)} results")
+        return results
+    
+    def clear(self):
+        """Clear all documents from the store."""
+        self.documents = []
+        self.embeddings = None
+        self.index_built = False
+        logger.info("Cleared vector store")
+    
+    def size(self) -> int:
+        """Get number of documents in store."""
+        return len(self.documents)
+    
+    def get_by_id(self, doc_id: str) -> Optional[Document]:
+        """Get document by ID."""
+        for doc in self.documents:
+            if doc.id == doc_id:
+                return doc
+        return None
+    
+    def get_stats(self) -> Dict[str, Any]:
+        """Get statistics about the vector store."""
+        stats = {
+            'num_documents': len(self.documents),
+            'dimension': self.dimension,
+            'index_built': self.index_built,
+            'memory_usage_mb': 0
+        }
+        
+        if self.embeddings is not None:
+            # Estimate memory usage
+            memory_bytes = self.embeddings.nbytes
+            for doc in self.documents:
+                memory_bytes += len(doc.text.encode('utf-8'))
+                memory_bytes += len(json.dumps(doc.metadata).encode('utf-8'))
+            stats['memory_usage_mb'] = memory_bytes / (1024 * 1024)
+        
+        return stats

uyusmazlik_mcp_module/client.py

@@ -1,7 +1,6 @@
 # uyusmazlik_mcp_module/client.py
 
 import httpx 
-import aiohttp 
 from bs4 import BeautifulSoup
 from typing import Dict, Any, List, Optional, Union, Tuple 
 import logging
@@ -9,7 +8,7 @@ import html
 import re
 import io
 from markitdown import MarkItDown
-from urllib.parse import urljoin, urlencode # urlencode for aiohttp form data
+from urllib.parse import urljoin
 
 from .models import (
     UyusmazlikSearchRequest,
@@ -56,17 +55,21 @@ class UyusmazlikApiClient:
     # Individual documents are fetched by their full URLs obtained from search results.
 
     def __init__(self, request_timeout: float = 30.0):
-        self.request_timeout = request_timeout # Store timeout for aiohttp and httpx
-        # Headers for aiohttp search. httpx for docs will create its own.
-        self.default_aiohttp_search_headers = {
-            "Accept": "*/*", # Mimicking browser headers provided by user
-            "Accept-Encoding": "gzip, deflate, br, zstd",
-            "Accept-Language": "tr-TR,tr;q=0.9,en-US;q=0.8,en;q=0.7",
-            "X-Requested-With": "XMLHttpRequest",
-            "Origin": self.BASE_URL,
-            "Referer": self.BASE_URL + "/",
-
-        }
+        self.request_timeout = request_timeout
+        # Create shared httpx client for all requests
+        self.http_client = httpx.AsyncClient(
+            base_url=self.BASE_URL,
+            headers={
+                "Accept": "*/*",
+                "Accept-Encoding": "gzip, deflate, br, zstd", 
+                "Accept-Language": "tr-TR,tr;q=0.9,en-US;q=0.8,en;q=0.7",
+                "X-Requested-With": "XMLHttpRequest",
+                "Origin": self.BASE_URL,
+                "Referer": self.BASE_URL + "/",
+            },
+            timeout=request_timeout,
+            verify=False
+        )
 
 
     async def search_decisions(
@@ -107,32 +110,36 @@ class UyusmazlikApiClient:
         add_to_form_data("Hepsi", params.hepsi)
         add_to_form_data("Herhangibirisi", params.herhangi_birisi)
         add_to_form_data("NotHepsi", params.not_hepsi)
-        # X-Requested-With is handled by default_aiohttp_search_headers
 
-        search_url = urljoin(self.BASE_URL, self.SEARCH_ENDPOINT)
-        # For aiohttp, data for application/x-www-form-urlencoded should be a dict or str.
-        # Using urlencode for list of tuples.
-        encoded_form_payload = urlencode(form_data_list, encoding='UTF-8') 
+        # Convert form data to dict for httpx
+        form_data_dict = {}
+        for key, value in form_data_list:
+            if key in form_data_dict:
+                # Handle multiple values (like KararSonucuList)
+                if not isinstance(form_data_dict[key], list):
+                    form_data_dict[key] = [form_data_dict[key]]
+                form_data_dict[key].append(value)
+            else:
+                form_data_dict[key] = value
 
-        logger.info(f"UyusmazlikApiClient (aiohttp): Performing search to {search_url} with form_data: {encoded_form_payload}")
+        logger.info(f"UyusmazlikApiClient (httpx): Performing search to {self.SEARCH_ENDPOINT} with form_data: {form_data_dict}")
         
-        html_content = ""
-        aiohttp_headers = self.default_aiohttp_search_headers.copy()
-        aiohttp_headers["Content-Type"] = "application/x-www-form-urlencoded; charset=UTF-8"
-
         try:
-            # Create a new session for each call for simplicity with aiohttp here
-            async with aiohttp.ClientSession(headers=aiohttp_headers) as session:
-                async with session.post(search_url, data=encoded_form_payload, timeout=self.request_timeout) as response:
-                    response.raise_for_status() # Raises ClientResponseError for 400-599
-                    html_content = await response.text(encoding='utf-8') # Ensure correct encoding
-                    logger.debug("UyusmazlikApiClient (aiohttp): Received HTML response for search.")
+            # Use shared httpx client
+            response = await self.http_client.post(
+                self.SEARCH_ENDPOINT,
+                data=form_data_dict,
+                headers={"Content-Type": "application/x-www-form-urlencoded; charset=UTF-8"}
+            )
+            response.raise_for_status()
+            html_content = response.text
+            logger.debug("UyusmazlikApiClient (httpx): Received HTML response for search.")
         
-        except aiohttp.ClientError as e:
-            logger.error(f"UyusmazlikApiClient (aiohttp): HTTP client error during search: {e}")
+        except httpx.HTTPError as e:
+            logger.error(f"UyusmazlikApiClient (httpx): HTTP client error during search: {e}")
             raise # Re-raise to be handled by the MCP tool
         except Exception as e:
-            logger.error(f"UyusmazlikApiClient (aiohttp): Error processing search request: {e}")
+            logger.error(f"UyusmazlikApiClient (httpx): Error processing search request: {e}")
             raise
 
         # --- HTML Parsing (remains the same as previous version) ---
@@ -217,7 +224,6 @@ class UyusmazlikApiClient:
         try:
             # Using a new httpx.AsyncClient instance for this GET request for simplicity
             async with httpx.AsyncClient(verify=False, timeout=self.request_timeout) as doc_fetch_client:
-
                  get_response = await doc_fetch_client.get(document_url, headers={"Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8"})
             get_response.raise_for_status()
             html_content_from_api = get_response.text
@@ -236,5 +242,9 @@ class UyusmazlikApiClient:
             raise
 
     async def close_client_session(self):
-
-        logger.info("UyusmazlikApiClient: No persistent client session from __init__ to close.")
+        """Close the shared httpx client session."""
+        if hasattr(self, 'http_client') and self.http_client:
+            await self.http_client.aclose()
+            logger.info("UyusmazlikApiClient: HTTP client session closed.")
+        else:
+            logger.info("UyusmazlikApiClient: No persistent client session from __init__ to close.")