ayrisdev/crawl4ai
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
fix/https-reditrect
crawl4ai/docs/md_v2/assets/llm.txt/diagrams/llms-diagram.txt
UncleCode 40640badad feat: add Script Builder to Chrome Extension and reorganize LLM context files 
This commit introduces significant enhancements to the Crawl4AI ecosystem:

  Chrome Extension - Script Builder (Alpha):
  - Add recording functionality to capture user interactions (clicks, typing, scrolling)
  - Implement smart event grouping for cleaner script generation
  - Support export to both JavaScript and C4A script formats
  - Add timeline view for visualizing and editing recorded actions
  - Include wait commands (time-based and element-based)
  - Add saved flows functionality for reusing automation scripts
  - Update UI with consistent dark terminal theme (Dank Mono font, green/pink accents)
  - Release new extension versions: v1.1.0, v1.2.0, v1.2.1

  LLM Context Builder Improvements:
  - Reorganize context files from llmtxt/ to llm.txt/ with better structure
  - Separate diagram templates from text content (diagrams/ and txt/ subdirectories)
  - Add comprehensive context files for all major Crawl4AI components
  - Improve file naming convention for better discoverability

  Documentation Updates:
  - Update apps index page to match main documentation theme
  - Standardize color scheme: "Available" tags use primary color (#50ffff)
  - Change "Coming Soon" tags to dark gray for better visual hierarchy
  - Add interactive two-column layout for extension landing page
  - Include code examples for both Schema Builder and Script Builder features

  Technical Improvements:
  - Enhance event capture mechanism with better element selection
  - Add support for contenteditable elements and complex form interactions
  - Implement proper scroll event handling for both window and element scrolling
  - Add meta key support for keyboard shortcuts
  - Improve selector generation for more reliable element targeting

  The Script Builder is released as Alpha, acknowledging potential bugs while providing
  early access to this powerful automation recording feature.
2025-06-08 22:02:12 +08:00

5913 lines
153 KiB
Plaintext
Raw Permalink Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Crawl4AI
> Open-source LLM-friendly web crawler and scraper for AI applications
Crawl4AI simplifies asynchronous web crawling and data extraction, making it accessible for large language models (LLMs) and AI applications. Built with Python and Playwright for high-performance crawling with structured data extraction.
**Key Features:**
- Asynchronous crawling with high concurrency
- Multiple extraction strategies (CSS, XPath, LLM-based)
- Built-in markdown generation with content filtering
- Docker deployment with REST API
- Session management and browser automation
- Advanced anti-detection capabilities
**Quick Links:**
- [GitHub Repository](https://github.com/unclecode/crawl4ai)
- [Documentation](https://docs.crawl4ai.com)
- [Examples](https://github.com/unclecode/crawl4ai/tree/main/docs/examples)
---
## Installation Workflows and Architecture
Visual representations of Crawl4AI installation processes, deployment options, and system interactions.
### Installation Decision Flow
```mermaid
flowchart TD
A[Start Installation] --> B{Environment Type?}
B -->|Local Development| C[Basic Python Install]
B -->|Production| D[Docker Deployment]
B -->|Research/Testing| E[Google Colab]
B -->|CI/CD Pipeline| F[Automated Setup]
C --> C1[pip install crawl4ai]
C1 --> C2[crawl4ai-setup]
C2 --> C3{Need Advanced Features?}
C3 -->|No| C4[Basic Installation Complete]
C3 -->|Text Clustering| C5[pip install crawl4ai with torch]
C3 -->|Transformers| C6[pip install crawl4ai with transformer]
C3 -->|All Features| C7[pip install crawl4ai with all]
C5 --> C8[crawl4ai-download-models]
C6 --> C8
C7 --> C8
C8 --> C9[Advanced Installation Complete]
D --> D1{Deployment Method?}
D1 -->|Pre-built Image| D2[docker pull unclecode/crawl4ai]
D1 -->|Docker Compose| D3[Clone repo + docker compose]
D1 -->|Custom Build| D4[docker buildx build]
D2 --> D5[Configure .llm.env]
D3 --> D5
D4 --> D5
D5 --> D6[docker run with ports]
D6 --> D7[Docker Deployment Complete]
E --> E1[Colab pip install]
E1 --> E2[playwright install chromium]
E2 --> E3[Test basic crawl]
E3 --> E4[Colab Setup Complete]
F --> F1[Automated pip install]
F1 --> F2[Automated setup scripts]
F2 --> F3[CI/CD Integration Complete]
C4 --> G[Verify with crawl4ai-doctor]
C9 --> G
D7 --> H[Health check via API]
E4 --> I[Run test crawl]
F3 --> G
G --> J[Installation Verified]
H --> J
I --> J
style A fill:#e1f5fe
style J fill:#c8e6c9
style C4 fill:#fff3e0
style C9 fill:#fff3e0
style D7 fill:#f3e5f5
style E4 fill:#fce4ec
style F3 fill:#e8f5e8
```
### Basic Installation Sequence
```mermaid
sequenceDiagram
participant User
participant PyPI
participant System
participant Playwright
participant Crawler
User->>PyPI: pip install crawl4ai
PyPI-->>User: Package downloaded
User->>System: crawl4ai-setup
System->>Playwright: Install browser binaries
Playwright-->>System: Chromium, Firefox installed
System-->>User: Setup complete
User->>System: crawl4ai-doctor
System->>System: Check Python version
System->>System: Verify Playwright installation
System->>System: Test browser launch
System-->>User: Diagnostics report
User->>Crawler: Basic crawl test
Crawler->>Playwright: Launch browser
Playwright-->>Crawler: Browser ready
Crawler->>Crawler: Navigate to test URL
Crawler-->>User: Success confirmation
```
### Docker Deployment Architecture
```mermaid
graph TB
subgraph "Host System"
A[Docker Engine] --> B[Crawl4AI Container]
C[.llm.env File] --> B
D[Port 11235] --> B
end
subgraph "Container Environment"
B --> E[FastAPI Server]
B --> F[Playwright Browsers]
B --> G[Python Runtime]
E --> H[/crawl Endpoint]
E --> I[/playground Interface]
E --> J[/health Monitoring]
E --> K[/metrics Prometheus]
F --> L[Chromium Browser]
F --> M[Firefox Browser]
F --> N[WebKit Browser]
end
subgraph "External Services"
O[OpenAI API] --> B
P[Anthropic API] --> B
Q[Local LLM Ollama] --> B
end
subgraph "Client Applications"
R[Python SDK] --> H
S[REST API Calls] --> H
T[Web Browser] --> I
U[Monitoring Tools] --> J
V[Prometheus] --> K
end
style B fill:#e3f2fd
style E fill:#f3e5f5
style F fill:#e8f5e8
style G fill:#fff3e0
```
### Advanced Features Installation Flow
```mermaid
stateDiagram-v2
[*] --> BasicInstall
BasicInstall --> FeatureChoice: crawl4ai installed
FeatureChoice --> TorchInstall: Need text clustering
FeatureChoice --> TransformerInstall: Need HuggingFace models
FeatureChoice --> AllInstall: Need everything
FeatureChoice --> Complete: Basic features sufficient
TorchInstall --> TorchSetup: pip install crawl4ai with torch
TransformerInstall --> TransformerSetup: pip install crawl4ai with transformer
AllInstall --> AllSetup: pip install crawl4ai with all
TorchSetup --> ModelDownload: crawl4ai-setup
TransformerSetup --> ModelDownload: crawl4ai-setup
AllSetup --> ModelDownload: crawl4ai-setup
ModelDownload --> PreDownload: crawl4ai-download-models
PreDownload --> Complete: All models cached
Complete --> Verification: crawl4ai-doctor
Verification --> [*]: Installation verified
note right of TorchInstall : PyTorch for semantic operations
note right of TransformerInstall : HuggingFace for LLM features
note right of AllInstall : Complete feature set
```
### Platform-Specific Installation Matrix
```mermaid
graph LR
subgraph "Installation Methods"
A[Python Package] --> A1[pip install]
B[Docker Image] --> B1[docker pull]
C[Source Build] --> C1[git clone + build]
D[Cloud Platform] --> D1[Colab/Kaggle]
end
subgraph "Operating Systems"
E[Linux x86_64]
F[Linux ARM64]
G[macOS Intel]
H[macOS Apple Silicon]
I[Windows x86_64]
end
subgraph "Feature Sets"
J[Basic crawling]
K[Text clustering torch]
L[LLM transformers]
M[All features]
end
A1 --> E
A1 --> F
A1 --> G
A1 --> H
A1 --> I
B1 --> E
B1 --> F
B1 --> G
B1 --> H
C1 --> E
C1 --> F
C1 --> G
C1 --> H
C1 --> I
D1 --> E
D1 --> I
E --> J
E --> K
E --> L
E --> M
F --> J
F --> K
F --> L
F --> M
G --> J
G --> K
G --> L
G --> M
H --> J
H --> K
H --> L
H --> M
I --> J
I --> K
I --> L
I --> M
style A1 fill:#e3f2fd
style B1 fill:#f3e5f5
style C1 fill:#e8f5e8
style D1 fill:#fff3e0
```
### Docker Multi-Stage Build Process
```mermaid
sequenceDiagram
participant Dev as Developer
participant Git as GitHub Repo
participant Docker as Docker Engine
participant Registry as Docker Hub
participant User as End User
Dev->>Git: Push code changes
Docker->>Git: Clone repository
Docker->>Docker: Stage 1 - Base Python image
Docker->>Docker: Stage 2 - Install dependencies
Docker->>Docker: Stage 3 - Install Playwright
Docker->>Docker: Stage 4 - Copy application code
Docker->>Docker: Stage 5 - Setup FastAPI server
Note over Docker: Multi-architecture build
Docker->>Docker: Build for linux/amd64
Docker->>Docker: Build for linux/arm64
Docker->>Registry: Push multi-arch manifest
Registry-->>Docker: Build complete
User->>Registry: docker pull unclecode/crawl4ai
Registry-->>User: Download appropriate architecture
User->>Docker: docker run with configuration
Docker->>Docker: Start container
Docker->>Docker: Initialize FastAPI server
Docker->>Docker: Setup Playwright browsers
Docker-->>User: Service ready on port 11235
```
### Installation Verification Workflow
```mermaid
flowchart TD
A[Installation Complete] --> B[Run crawl4ai-doctor]
B --> C{Python Version Check}
C -->|✓ 3.10+| D{Playwright Check}
C -->|✗ < 3.10| C1[Upgrade Python]
C1 --> D
D -->|✓ Installed| E{Browser Binaries}
D -->|✗ Missing| D1[Run crawl4ai-setup]
D1 --> E
E -->|✓ Available| F{Test Browser Launch}
E -->|✗ Missing| E1[playwright install]
E1 --> F
F -->|✓ Success| G[Test Basic Crawl]
F -->|✗ Failed| F1[Check system dependencies]
F1 --> F
G --> H{Crawl Test Result}
H -->|✓ Success| I[Installation Verified ✓]
H -->|✗ Failed| H1[Check network/permissions]
H1 --> G
I --> J[Ready for Production Use]
style I fill:#c8e6c9
style J fill:#e8f5e8
style C1 fill:#ffcdd2
style D1 fill:#fff3e0
style E1 fill:#fff3e0
style F1 fill:#ffcdd2
style H1 fill:#ffcdd2
```
### Resource Requirements by Installation Type
```mermaid
graph TD
subgraph "Basic Installation"
A1[Memory: 512MB]
A2[Disk: 2GB]
A3[CPU: 1 core]
A4[Network: Required for setup]
end
subgraph "Advanced Features torch"
B1[Memory: 2GB+]
B2[Disk: 5GB+]
B3[CPU: 2+ cores]
B4[GPU: Optional CUDA]
end
subgraph "All Features"
C1[Memory: 4GB+]
C2[Disk: 10GB+]
C3[CPU: 4+ cores]
C4[GPU: Recommended]
end
subgraph "Docker Deployment"
D1[Memory: 1GB+]
D2[Disk: 3GB+]
D3[CPU: 2+ cores]
D4[Ports: 11235]
D5[Shared Memory: 1GB]
end
style A1 fill:#e8f5e8
style B1 fill:#fff3e0
style C1 fill:#ffecb3
style D1 fill:#e3f2fd
```
**📖 Learn more:** [Installation Guide](https://docs.crawl4ai.com/core/installation/), [Docker Deployment](https://docs.crawl4ai.com/core/docker-deployment/), [System Requirements](https://docs.crawl4ai.com/core/installation/#prerequisites)
---
## Simple Crawling Workflows and Data Flow
Visual representations of basic web crawling operations, configuration patterns, and result processing workflows.
### Basic Crawling Sequence
```mermaid
sequenceDiagram
participant User
participant Crawler as AsyncWebCrawler
participant Browser as Browser Instance
participant Page as Web Page
participant Processor as Content Processor
User->>Crawler: Create with BrowserConfig
Crawler->>Browser: Launch browser instance
Browser-->>Crawler: Browser ready
User->>Crawler: arun(url, CrawlerRunConfig)
Crawler->>Browser: Create new page/context
Browser->>Page: Navigate to URL
Page-->>Browser: Page loaded
Browser->>Processor: Extract raw HTML
Processor->>Processor: Clean HTML
Processor->>Processor: Generate markdown
Processor->>Processor: Extract media/links
Processor-->>Crawler: CrawlResult created
Crawler-->>User: Return CrawlResult
Note over User,Processor: All processing happens asynchronously
```
### Crawling Configuration Flow
```mermaid
flowchart TD
A[Start Crawling] --> B{Browser Config Set?}
B -->|No| B1[Use Default BrowserConfig]
B -->|Yes| B2[Custom BrowserConfig]
B1 --> C[Launch Browser]
B2 --> C
C --> D{Crawler Run Config Set?}
D -->|No| D1[Use Default CrawlerRunConfig]
D -->|Yes| D2[Custom CrawlerRunConfig]
D1 --> E[Navigate to URL]
D2 --> E
E --> F{Page Load Success?}
F -->|No| F1[Return Error Result]
F -->|Yes| G[Apply Content Filters]
G --> G1{excluded_tags set?}
G1 -->|Yes| G2[Remove specified tags]
G1 -->|No| G3[Keep all tags]
G2 --> G4{css_selector set?}
G3 --> G4
G4 -->|Yes| G5[Extract selected elements]
G4 -->|No| G6[Process full page]
G5 --> H[Generate Markdown]
G6 --> H
H --> H1{markdown_generator set?}
H1 -->|Yes| H2[Use custom generator]
H1 -->|No| H3[Use default generator]
H2 --> I[Extract Media and Links]
H3 --> I
I --> I1{process_iframes?}
I1 -->|Yes| I2[Include iframe content]
I1 -->|No| I3[Skip iframes]
I2 --> J[Create CrawlResult]
I3 --> J
J --> K[Return Result]
style A fill:#e1f5fe
style K fill:#c8e6c9
style F1 fill:#ffcdd2
```
### CrawlResult Data Structure
```mermaid
graph TB
subgraph "CrawlResult Object"
A[CrawlResult] --> B[Basic Info]
A --> C[Content Variants]
A --> D[Extracted Data]
A --> E[Media Assets]
A --> F[Optional Outputs]
B --> B1[url: Final URL]
B --> B2[success: Boolean]
B --> B3[status_code: HTTP Status]
B --> B4[error_message: Error Details]
C --> C1[html: Raw HTML]
C --> C2[cleaned_html: Sanitized HTML]
C --> C3[markdown: MarkdownGenerationResult]
C3 --> C3A[raw_markdown: Basic conversion]
C3 --> C3B[markdown_with_citations: With references]
C3 --> C3C[fit_markdown: Filtered content]
C3 --> C3D[references_markdown: Citation list]
D --> D1[links: Internal/External]
D --> D2[media: Images/Videos/Audio]
D --> D3[metadata: Page info]
D --> D4[extracted_content: JSON data]
D --> D5[tables: Structured table data]
E --> E1[screenshot: Base64 image]
E --> E2[pdf: PDF bytes]
E --> E3[mhtml: Archive file]
E --> E4[downloaded_files: File paths]
F --> F1[session_id: Browser session]
F --> F2[ssl_certificate: Security info]
F --> F3[response_headers: HTTP headers]
F --> F4[network_requests: Traffic log]
F --> F5[console_messages: Browser logs]
end
style A fill:#e3f2fd
style C3 fill:#f3e5f5
style D5 fill:#e8f5e8
```
### Content Processing Pipeline
```mermaid
flowchart LR
subgraph "Input Sources"
A1[Web URL]
A2[Raw HTML]
A3[Local File]
end
A1 --> B[Browser Navigation]
A2 --> C[Direct Processing]
A3 --> C
B --> D[Raw HTML Capture]
C --> D
D --> E{Content Filtering}
E --> E1[Remove Scripts/Styles]
E --> E2[Apply excluded_tags]
E --> E3[Apply css_selector]
E --> E4[Remove overlay elements]
E1 --> F[Cleaned HTML]
E2 --> F
E3 --> F
E4 --> F
F --> G{Markdown Generation}
G --> G1[HTML to Markdown]
G --> G2[Apply Content Filter]
G --> G3[Generate Citations]
G1 --> H[MarkdownGenerationResult]
G2 --> H
G3 --> H
F --> I{Media Extraction}
I --> I1[Find Images]
I --> I2[Find Videos/Audio]
I --> I3[Score Relevance]
I1 --> J[Media Dictionary]
I2 --> J
I3 --> J
F --> K{Link Extraction}
K --> K1[Internal Links]
K --> K2[External Links]
K --> K3[Apply Link Filters]
K1 --> L[Links Dictionary]
K2 --> L
K3 --> L
H --> M[Final CrawlResult]
J --> M
L --> M
style D fill:#e3f2fd
style F fill:#f3e5f5
style H fill:#e8f5e8
style M fill:#c8e6c9
```
### Table Extraction Workflow
```mermaid
stateDiagram-v2
[*] --> DetectTables
DetectTables --> ScoreTables: Find table elements
ScoreTables --> EvaluateThreshold: Calculate quality scores
EvaluateThreshold --> PassThreshold: score >= table_score_threshold
EvaluateThreshold --> RejectTable: score < threshold
PassThreshold --> ExtractHeaders: Parse table structure
ExtractHeaders --> ExtractRows: Get header cells
ExtractRows --> ExtractMetadata: Get data rows
ExtractMetadata --> CreateTableObject: Get caption/summary
CreateTableObject --> AddToResult: {headers, rows, caption, summary}
AddToResult --> [*]: Table extraction complete
RejectTable --> [*]: Table skipped
note right of ScoreTables : Factors: header presence, data density, structure quality
note right of EvaluateThreshold : Threshold 1-10, higher = stricter
```
### Error Handling Decision Tree
```mermaid
flowchart TD
A[Start Crawl] --> B[Navigate to URL]
B --> C{Navigation Success?}
C -->|Network Error| C1[Set error_message: Network failure]
C -->|Timeout| C2[Set error_message: Page timeout]
C -->|Invalid URL| C3[Set error_message: Invalid URL format]
C -->|Success| D[Process Page Content]
C1 --> E[success = False]
C2 --> E
C3 --> E
D --> F{Content Processing OK?}
F -->|Parser Error| F1[Set error_message: HTML parsing failed]
F -->|Memory Error| F2[Set error_message: Insufficient memory]
F -->|Success| G[Generate Outputs]
F1 --> E
F2 --> E
G --> H{Output Generation OK?}
H -->|Markdown Error| H1[Partial success with warnings]
H -->|Extraction Error| H2[Partial success with warnings]
H -->|Success| I[success = True]
H1 --> I
H2 --> I
E --> J[Return Failed CrawlResult]
I --> K[Return Successful CrawlResult]
J --> L[User Error Handling]
K --> M[User Result Processing]
L --> L1{Check error_message}
L1 -->|Network| L2[Retry with different config]
L1 -->|Timeout| L3[Increase page_timeout]
L1 -->|Parser| L4[Try different scraping_strategy]
style E fill:#ffcdd2
style I fill:#c8e6c9
style J fill:#ffcdd2
style K fill:#c8e6c9
```
### Configuration Impact Matrix
```mermaid
graph TB
subgraph "Configuration Categories"
A[Content Processing]
B[Page Interaction]
C[Output Generation]
D[Performance]
end
subgraph "Configuration Options"
A --> A1[word_count_threshold]
A --> A2[excluded_tags]
A --> A3[css_selector]
A --> A4[exclude_external_links]
B --> B1[process_iframes]
B --> B2[remove_overlay_elements]
B --> B3[scan_full_page]
B --> B4[wait_for]
C --> C1[screenshot]
C --> C2[pdf]
C --> C3[markdown_generator]
C --> C4[table_score_threshold]
D --> D1[cache_mode]
D --> D2[verbose]
D --> D3[page_timeout]
D --> D4[semaphore_count]
end
subgraph "Result Impact"
A1 --> R1[Filters short text blocks]
A2 --> R2[Removes specified HTML tags]
A3 --> R3[Focuses on selected content]
A4 --> R4[Cleans links dictionary]
B1 --> R5[Includes iframe content]
B2 --> R6[Removes popups/modals]
B3 --> R7[Loads dynamic content]
B4 --> R8[Waits for specific elements]
C1 --> R9[Adds screenshot field]
C2 --> R10[Adds pdf field]
C3 --> R11[Custom markdown processing]
C4 --> R12[Filters table quality]
D1 --> R13[Controls caching behavior]
D2 --> R14[Detailed logging output]
D3 --> R15[Prevents timeout errors]
D4 --> R16[Limits concurrent operations]
end
style A fill:#e3f2fd
style B fill:#f3e5f5
style C fill:#e8f5e8
style D fill:#fff3e0
```
### Raw HTML and Local File Processing
```mermaid
sequenceDiagram
participant User
participant Crawler
participant Processor
participant FileSystem
Note over User,FileSystem: Raw HTML Processing
User->>Crawler: arun("raw://html_content")
Crawler->>Processor: Parse raw HTML directly
Processor->>Processor: Apply same content filters
Processor-->>Crawler: Standard CrawlResult
Crawler-->>User: Result with markdown
Note over User,FileSystem: Local File Processing
User->>Crawler: arun("file:///path/to/file.html")
Crawler->>FileSystem: Read local file
FileSystem-->>Crawler: File content
Crawler->>Processor: Process file HTML
Processor->>Processor: Apply content processing
Processor-->>Crawler: Standard CrawlResult
Crawler-->>User: Result with markdown
Note over User,FileSystem: Both return identical CrawlResult structure
```
### Comprehensive Processing Example Flow
```mermaid
flowchart TD
A[Input: example.com] --> B[Create Configurations]
B --> B1[BrowserConfig verbose=True]
B --> B2[CrawlerRunConfig with filters]
B1 --> C[Launch AsyncWebCrawler]
B2 --> C
C --> D[Navigate and Process]
D --> E{Check Success}
E -->|Failed| E1[Print Error Message]
E -->|Success| F[Extract Content Summary]
F --> F1[Get Page Title]
F --> F2[Get Content Preview]
F --> F3[Process Media Items]
F --> F4[Process Links]
F3 --> F3A[Count Images]
F3 --> F3B[Show First 3 Images]
F4 --> F4A[Count Internal Links]
F4 --> F4B[Show First 3 Links]
F1 --> G[Display Results]
F2 --> G
F3A --> G
F3B --> G
F4A --> G
F4B --> G
E1 --> H[End with Error]
G --> I[End with Success]
style E1 fill:#ffcdd2
style G fill:#c8e6c9
style H fill:#ffcdd2
style I fill:#c8e6c9
```
**📖 Learn more:** [Simple Crawling Guide](https://docs.crawl4ai.com/core/simple-crawling/), [Configuration Options](https://docs.crawl4ai.com/core/browser-crawler-config/), [Result Processing](https://docs.crawl4ai.com/core/crawler-result/), [Table Extraction](https://docs.crawl4ai.com/extraction/no-llm-strategies/)
---
## Configuration Objects and System Architecture
Visual representations of Crawl4AI's configuration system, object relationships, and data flow patterns.
### Configuration Object Relationships
```mermaid
classDiagram
class BrowserConfig {
+browser_type: str
+headless: bool
+viewport_width: int
+viewport_height: int
+proxy: str
+user_agent: str
+cookies: list
+headers: dict
+clone() BrowserConfig
+to_dict() dict
}
class CrawlerRunConfig {
+cache_mode: CacheMode
+extraction_strategy: ExtractionStrategy
+markdown_generator: MarkdownGenerator
+js_code: list
+wait_for: str
+screenshot: bool
+session_id: str
+clone() CrawlerRunConfig
+dump() dict
}
class LLMConfig {
+provider: str
+api_token: str
+base_url: str
+temperature: float
+max_tokens: int
+clone() LLMConfig
+to_dict() dict
}
class CrawlResult {
+url: str
+success: bool
+html: str
+cleaned_html: str
+markdown: MarkdownGenerationResult
+extracted_content: str
+media: dict
+links: dict
+screenshot: str
+pdf: bytes
}
class AsyncWebCrawler {
+config: BrowserConfig
+arun() CrawlResult
}
AsyncWebCrawler --> BrowserConfig : uses
AsyncWebCrawler --> CrawlerRunConfig : accepts
CrawlerRunConfig --> LLMConfig : contains
AsyncWebCrawler --> CrawlResult : returns
note for BrowserConfig "Controls browser\nenvironment and behavior"
note for CrawlerRunConfig "Controls individual\ncrawl operations"
note for LLMConfig "Configures LLM\nproviders and parameters"
note for CrawlResult "Contains all crawl\noutputs and metadata"
```
### Configuration Decision Flow
```mermaid
flowchart TD
A[Start Configuration] --> B{Use Case Type?}
B -->|Simple Web Scraping| C[Basic Config Pattern]
B -->|Data Extraction| D[Extraction Config Pattern]
B -->|Stealth Crawling| E[Stealth Config Pattern]
B -->|High Performance| F[Performance Config Pattern]
C --> C1[BrowserConfig: headless=True]
C --> C2[CrawlerRunConfig: basic options]
C1 --> C3[No LLMConfig needed]
C2 --> C3
C3 --> G[Simple Crawling Ready]
D --> D1[BrowserConfig: standard setup]
D --> D2[CrawlerRunConfig: with extraction_strategy]
D --> D3[LLMConfig: for LLM extraction]
D1 --> D4[Advanced Extraction Ready]
D2 --> D4
D3 --> D4
E --> E1[BrowserConfig: proxy + user_agent]
E --> E2[CrawlerRunConfig: simulate_user=True]
E1 --> E3[Stealth Crawling Ready]
E2 --> E3
F --> F1[BrowserConfig: lightweight]
F --> F2[CrawlerRunConfig: caching + concurrent]
F1 --> F3[High Performance Ready]
F2 --> F3
G --> H[Execute Crawl]
D4 --> H
E3 --> H
F3 --> H
H --> I[Get CrawlResult]
style A fill:#e1f5fe
style I fill:#c8e6c9
style G fill:#fff3e0
style D4 fill:#f3e5f5
style E3 fill:#ffebee
style F3 fill:#e8f5e8
```
### Configuration Lifecycle Sequence
```mermaid
sequenceDiagram
participant User
participant BrowserConfig as Browser Config
participant CrawlerConfig as Crawler Config
participant LLMConfig as LLM Config
participant Crawler as AsyncWebCrawler
participant Browser as Browser Instance
participant Result as CrawlResult
User->>BrowserConfig: Create with browser settings
User->>CrawlerConfig: Create with crawl options
User->>LLMConfig: Create with LLM provider
User->>Crawler: Initialize with BrowserConfig
Crawler->>Browser: Launch browser with config
Browser-->>Crawler: Browser ready
User->>Crawler: arun(url, CrawlerConfig)
Crawler->>Crawler: Apply CrawlerConfig settings
alt LLM Extraction Needed
Crawler->>LLMConfig: Get LLM settings
LLMConfig-->>Crawler: Provider configuration
end
Crawler->>Browser: Navigate with settings
Browser->>Browser: Apply page interactions
Browser->>Browser: Execute JavaScript if specified
Browser->>Browser: Wait for conditions
Browser-->>Crawler: Page content ready
Crawler->>Crawler: Process content per config
Crawler->>Result: Create CrawlResult
Result-->>User: Return complete result
Note over User,Result: Configuration objects control every aspect
```
### BrowserConfig Parameter Flow
```mermaid
graph TB
subgraph "BrowserConfig Parameters"
A[browser_type] --> A1[chromium/firefox/webkit]
B[headless] --> B1[true: invisible / false: visible]
C[viewport] --> C1[width x height dimensions]
D[proxy] --> D1[proxy server configuration]
E[user_agent] --> E1[browser identification string]
F[cookies] --> F1[session authentication]
G[headers] --> G1[HTTP request headers]
H[extra_args] --> H1[browser command line flags]
end
subgraph "Browser Instance"
I[Playwright Browser]
J[Browser Context]
K[Page Instance]
end
A1 --> I
B1 --> I
C1 --> J
D1 --> J
E1 --> J
F1 --> J
G1 --> J
H1 --> I
I --> J
J --> K
style I fill:#e3f2fd
style J fill:#f3e5f5
style K fill:#e8f5e8
```
### CrawlerRunConfig Category Breakdown
```mermaid
mindmap
root((CrawlerRunConfig))
Content Processing
word_count_threshold
css_selector
target_elements
excluded_tags
markdown_generator
extraction_strategy
Page Navigation
wait_until
page_timeout
wait_for
wait_for_images
delay_before_return_html
Page Interaction
js_code
scan_full_page
simulate_user
magic
remove_overlay_elements
Caching Session
cache_mode
session_id
shared_data
Media Output
screenshot
pdf
capture_mhtml
image_score_threshold
Link Filtering
exclude_external_links
exclude_domains
exclude_social_media_links
```
### LLM Provider Selection Flow
```mermaid
flowchart TD
A[Need LLM Processing?] --> B{Provider Type?}
B -->|Cloud API| C{Which Service?}
B -->|Local Model| D[Local Setup]
B -->|Custom Endpoint| E[Custom Config]
C -->|OpenAI| C1[OpenAI GPT Models]
C -->|Anthropic| C2[Claude Models]
C -->|Google| C3[Gemini Models]
C -->|Groq| C4[Fast Inference]
D --> D1[Ollama Setup]
E --> E1[Custom base_url]
C1 --> F1[LLMConfig with OpenAI settings]
C2 --> F2[LLMConfig with Anthropic settings]
C3 --> F3[LLMConfig with Google settings]
C4 --> F4[LLMConfig with Groq settings]
D1 --> F5[LLMConfig with Ollama settings]
E1 --> F6[LLMConfig with custom settings]
F1 --> G[Use in Extraction Strategy]
F2 --> G
F3 --> G
F4 --> G
F5 --> G
F6 --> G
style A fill:#e1f5fe
style G fill:#c8e6c9
```
### CrawlResult Structure and Data Flow
```mermaid
graph TB
subgraph "CrawlResult Output"
A[Basic Info]
B[HTML Content]
C[Markdown Output]
D[Extracted Data]
E[Media Files]
F[Metadata]
end
subgraph "Basic Info Details"
A --> A1[url: final URL]
A --> A2[success: boolean]
A --> A3[status_code: HTTP status]
A --> A4[error_message: if failed]
end
subgraph "HTML Content Types"
B --> B1[html: raw HTML]
B --> B2[cleaned_html: processed]
B --> B3[fit_html: filtered content]
end
subgraph "Markdown Variants"
C --> C1[raw_markdown: basic conversion]
C --> C2[markdown_with_citations: with refs]
C --> C3[fit_markdown: filtered content]
C --> C4[references_markdown: citation list]
end
subgraph "Extracted Content"
D --> D1[extracted_content: JSON string]
D --> D2[From CSS extraction]
D --> D3[From LLM extraction]
D --> D4[From XPath extraction]
end
subgraph "Media and Links"
E --> E1[images: list with scores]
E --> E2[videos: media content]
E --> E3[internal_links: same domain]
E --> E4[external_links: other domains]
end
subgraph "Generated Files"
F --> F1[screenshot: base64 PNG]
F --> F2[pdf: binary PDF data]
F --> F3[mhtml: archive format]
F --> F4[ssl_certificate: cert info]
end
style A fill:#e3f2fd
style B fill:#f3e5f5
style C fill:#e8f5e8
style D fill:#fff3e0
style E fill:#ffebee
style F fill:#f1f8e9
```
### Configuration Pattern State Machine
```mermaid
stateDiagram-v2
[*] --> ConfigCreation
ConfigCreation --> BasicConfig: Simple use case
ConfigCreation --> AdvancedConfig: Complex requirements
ConfigCreation --> TemplateConfig: Use predefined pattern
BasicConfig --> Validation: Check parameters
AdvancedConfig --> Validation: Check parameters
TemplateConfig --> Validation: Check parameters
Validation --> Invalid: Missing required fields
Validation --> Valid: All parameters correct
Invalid --> ConfigCreation: Fix and retry
Valid --> InUse: Passed to crawler
InUse --> Cloning: Need variation
InUse --> Serialization: Save configuration
InUse --> Complete: Crawl finished
Cloning --> Modified: clone() with updates
Modified --> Valid: Validate changes
Serialization --> Stored: dump() to dict
Stored --> Restoration: load() from dict
Restoration --> Valid: Recreate config object
Complete --> [*]
note right of BasicConfig : Minimal required settings
note right of AdvancedConfig : Full feature configuration
note right of TemplateConfig : Pre-built patterns
```
### Configuration Integration Architecture
```mermaid
graph TB
subgraph "User Layer"
U1[Configuration Creation]
U2[Parameter Selection]
U3[Pattern Application]
end
subgraph "Configuration Layer"
C1[BrowserConfig]
C2[CrawlerRunConfig]
C3[LLMConfig]
C4[Config Validation]
C5[Config Cloning]
end
subgraph "Crawler Engine"
E1[Browser Management]
E2[Page Navigation]
E3[Content Processing]
E4[Extraction Pipeline]
E5[Result Generation]
end
subgraph "Output Layer"
O1[CrawlResult Assembly]
O2[Data Formatting]
O3[File Generation]
O4[Metadata Collection]
end
U1 --> C1
U2 --> C2
U3 --> C3
C1 --> C4
C2 --> C4
C3 --> C4
C4 --> E1
C2 --> E2
C2 --> E3
C3 --> E4
E1 --> E2
E2 --> E3
E3 --> E4
E4 --> E5
E5 --> O1
O1 --> O2
O2 --> O3
O3 --> O4
C5 -.-> C1
C5 -.-> C2
C5 -.-> C3
style U1 fill:#e1f5fe
style C4 fill:#fff3e0
style E4 fill:#f3e5f5
style O4 fill:#c8e6c9
```
### Configuration Best Practices Flow
```mermaid
flowchart TD
A[Configuration Planning] --> B{Performance Priority?}
B -->|Speed| C[Fast Config Pattern]
B -->|Quality| D[Comprehensive Config Pattern]
B -->|Stealth| E[Stealth Config Pattern]
B -->|Balanced| F[Standard Config Pattern]
C --> C1[Enable caching]
C --> C2[Disable heavy features]
C --> C3[Use text_mode]
C1 --> G[Apply Configuration]
C2 --> G
C3 --> G
D --> D1[Enable all processing]
D --> D2[Use content filters]
D --> D3[Capture everything]
D1 --> G
D2 --> G
D3 --> G
E --> E1[Rotate user agents]
E --> E2[Use proxies]
E --> E3[Simulate human behavior]
E1 --> G
E2 --> G
E3 --> G
F --> F1[Balanced timeouts]
F --> F2[Selective processing]
F --> F3[Smart caching]
F1 --> G
F2 --> G
F3 --> G
G --> H[Test Configuration]
H --> I{Results Satisfactory?}
I -->|Yes| J[Production Ready]
I -->|No| K[Adjust Parameters]
K --> L[Clone and Modify]
L --> H
J --> M[Deploy with Confidence]
style A fill:#e1f5fe
style J fill:#c8e6c9
style M fill:#e8f5e8
```
## Advanced Configuration Workflows and Patterns
Visual representations of advanced Crawl4AI configuration strategies, proxy management, session handling, and identity-based crawling patterns.
### User Agent and Anti-Detection Strategy Flow
```mermaid
flowchart TD
A[Start Configuration] --> B{Detection Avoidance Needed?}
B -->|No| C[Standard User Agent]
B -->|Yes| D[Anti-Detection Strategy]
C --> C1[Static user_agent string]
C1 --> Z[Basic Configuration]
D --> E{User Agent Strategy}
E -->|Random| F[user_agent_mode: random]
E -->|Static Custom| G[Custom user_agent string]
E -->|Platform Specific| H[Generator Config]
F --> I[Configure Generator]
H --> I
I --> I1[Platform: windows/macos/linux]
I1 --> I2[Browser: chrome/firefox/safari]
I2 --> I3[Device: desktop/mobile/tablet]
G --> J[Behavioral Simulation]
I3 --> J
J --> K{Enable Simulation?}
K -->|Yes| L[simulate_user: True]
K -->|No| M[Standard Behavior]
L --> N[override_navigator: True]
N --> O[Configure Delays]
O --> O1[mean_delay: 1.5]
O1 --> O2[max_range: 2.0]
O2 --> P[Magic Mode]
M --> P
P --> Q{Auto-Handle Patterns?}
Q -->|Yes| R[magic: True]
Q -->|No| S[Manual Handling]
R --> T[Complete Anti-Detection Setup]
S --> T
Z --> T
style D fill:#ffeb3b
style T fill:#c8e6c9
style L fill:#ff9800
style R fill:#9c27b0
```
### Proxy Configuration and Rotation Architecture
```mermaid
graph TB
subgraph "Proxy Configuration Types"
A[Single Proxy] --> A1[ProxyConfig object]
B[Proxy String] --> B1[from_string method]
C[Environment Proxies] --> C1[from_env method]
D[Multiple Proxies] --> D1[ProxyRotationStrategy]
end
subgraph "ProxyConfig Structure"
A1 --> E[server: URL]
A1 --> F[username: auth]
A1 --> G[password: auth]
A1 --> H[ip: extracted]
end
subgraph "Rotation Strategies"
D1 --> I[round_robin]
D1 --> J[random]
D1 --> K[least_used]
D1 --> L[failure_aware]
end
subgraph "Configuration Flow"
M[CrawlerRunConfig] --> N[proxy_config]
M --> O[proxy_rotation_strategy]
N --> P[Single Proxy Usage]
O --> Q[Multi-Proxy Rotation]
end
subgraph "Runtime Behavior"
P --> R[All requests use same proxy]
Q --> S[Requests rotate through proxies]
S --> T[Health monitoring]
T --> U[Automatic failover]
end
style A1 fill:#e3f2fd
style D1 fill:#f3e5f5
style M fill:#e8f5e8
style T fill:#fff3e0
```
### Content Selection Strategy Comparison
```mermaid
sequenceDiagram
participant Browser
participant HTML as Raw HTML
participant CSS as css_selector
participant Target as target_elements
participant Processor as Content Processor
participant Output
Note over Browser,Output: css_selector Strategy
Browser->>HTML: Load complete page
HTML->>CSS: Apply css_selector
CSS->>CSS: Extract matching elements only
CSS->>Processor: Process subset HTML
Processor->>Output: Markdown + Extraction from subset
Note over Browser,Output: target_elements Strategy
Browser->>HTML: Load complete page
HTML->>Processor: Process entire page
Processor->>Target: Focus on target_elements
Target->>Target: Extract from specified elements
Processor->>Output: Full page links/media + targeted content
Note over CSS,Target: Key Difference
Note over CSS: Affects entire processing pipeline
Note over Target: Affects only content extraction
```
### Advanced wait_for Conditions Decision Tree
```mermaid
flowchart TD
A[Configure wait_for] --> B{Condition Type?}
B -->|CSS Element| C[CSS Selector Wait]
B -->|JavaScript Condition| D[JS Expression Wait]
B -->|Complex Logic| E[Custom JS Function]
B -->|No Wait| F[Default domcontentloaded]
C --> C1["wait_for: 'css:.element'"]
C1 --> C2[Element appears in DOM]
C2 --> G[Continue Processing]
D --> D1["wait_for: 'js:() => condition'"]
D1 --> D2[JavaScript returns true]
D2 --> G
E --> E1[Complex JS Function]
E1 --> E2{Multiple Conditions}
E2 -->|AND Logic| E3[All conditions true]
E2 -->|OR Logic| E4[Any condition true]
E2 -->|Custom Logic| E5[User-defined logic]
E3 --> G
E4 --> G
E5 --> G
F --> G
G --> H{Timeout Reached?}
H -->|No| I[Page Ready]
H -->|Yes| J[Timeout Error]
I --> K[Begin Content Extraction]
J --> L[Handle Error/Retry]
style C1 fill:#e8f5e8
style D1 fill:#fff3e0
style E1 fill:#ffeb3b
style I fill:#c8e6c9
style J fill:#ffcdd2
```
### Session Management Lifecycle
```mermaid
stateDiagram-v2
[*] --> SessionCreate
SessionCreate --> SessionActive: session_id provided
SessionCreate --> OneTime: no session_id
SessionActive --> BrowserLaunch: First arun() call
BrowserLaunch --> PageLoad: Navigate to URL
PageLoad --> JSExecution: Execute js_code
JSExecution --> ContentExtract: Extract content
ContentExtract --> SessionHold: Keep session alive
SessionHold --> ReuseSession: Subsequent arun() calls
ReuseSession --> JSOnlyMode: js_only=True
ReuseSession --> NewNavigation: js_only=False
JSOnlyMode --> JSExecution: Execute JS in existing page
NewNavigation --> PageLoad: Navigate to new URL
SessionHold --> SessionKill: kill_session() called
SessionHold --> SessionTimeout: Timeout reached
SessionHold --> SessionError: Error occurred
SessionKill --> SessionCleanup
SessionTimeout --> SessionCleanup
SessionError --> SessionCleanup
SessionCleanup --> [*]
OneTime --> BrowserLaunch
ContentExtract --> OneTimeCleanup: No session_id
OneTimeCleanup --> [*]
note right of SessionActive : Persistent browser context
note right of JSOnlyMode : Reuse existing page
note right of OneTime : Temporary browser instance
```
### Identity-Based Crawling Configuration Matrix
```mermaid
graph TD
subgraph "Geographic Identity"
A[Geolocation] --> A1[latitude/longitude]
A2[Timezone] --> A3[timezone_id]
A4[Locale] --> A5[language/region]
end
subgraph "Browser Identity"
B[User Agent] --> B1[Platform fingerprint]
B2[Navigator Properties] --> B3[override_navigator]
B4[Headers] --> B5[Accept-Language]
end
subgraph "Behavioral Identity"
C[Mouse Simulation] --> C1[simulate_user]
C2[Timing Patterns] --> C3[mean_delay/max_range]
C4[Interaction Patterns] --> C5[Human-like behavior]
end
subgraph "Configuration Integration"
D[CrawlerRunConfig] --> A
D --> B
D --> C
D --> E[Complete Identity Profile]
E --> F[Geographic Consistency]
E --> G[Browser Consistency]
E --> H[Behavioral Consistency]
end
F --> I[Paris, France Example]
I --> I1[locale: fr-FR]
I --> I2[timezone: Europe/Paris]
I --> I3[geolocation: 48.8566, 2.3522]
G --> J[Windows Chrome Example]
J --> J1[platform: windows]
J --> J2[browser: chrome]
J --> J3[user_agent: matching pattern]
H --> K[Human Simulation]
K --> K1[Random delays]
K --> K2[Mouse movements]
K --> K3[Navigation patterns]
style E fill:#ff9800
style I fill:#e3f2fd
style J fill:#f3e5f5
style K fill:#e8f5e8
```
### Multi-Step Crawling Sequence Flow
```mermaid
sequenceDiagram
participant User
participant Crawler
participant Session as Browser Session
participant Page1 as Login Page
participant Page2 as Dashboard
participant Page3 as Data Pages
User->>Crawler: Step 1 - Login
Crawler->>Session: Create session_id="user_session"
Session->>Page1: Navigate to login
Page1->>Page1: Execute login JS
Page1->>Page1: Wait for dashboard redirect
Page1-->>Crawler: Login complete
User->>Crawler: Step 2 - Navigate dashboard
Note over Crawler,Session: Reuse existing session
Crawler->>Session: js_only=True (no page reload)
Session->>Page2: Execute navigation JS
Page2->>Page2: Wait for data table
Page2-->>Crawler: Dashboard ready
User->>Crawler: Step 3 - Extract data pages
loop For each page 1-5
Crawler->>Session: js_only=True
Session->>Page3: Click page button
Page3->>Page3: Wait for page active
Page3->>Page3: Extract content
Page3-->>Crawler: Page data
end
User->>Crawler: Cleanup
Crawler->>Session: kill_session()
Session-->>Crawler: Session destroyed
```
### Configuration Import and Usage Patterns
```mermaid
graph LR
subgraph "Main Package Imports"
A[crawl4ai] --> A1[AsyncWebCrawler]
A --> A2[BrowserConfig]
A --> A3[CrawlerRunConfig]
A --> A4[LLMConfig]
A --> A5[CacheMode]
A --> A6[ProxyConfig]
A --> A7[GeolocationConfig]
end
subgraph "Strategy Imports"
A --> B1[JsonCssExtractionStrategy]
A --> B2[LLMExtractionStrategy]
A --> B3[DefaultMarkdownGenerator]
A --> B4[PruningContentFilter]
A --> B5[RegexChunking]
end
subgraph "Configuration Assembly"
C[Configuration Builder] --> A2
C --> A3
C --> A4
A2 --> D[Browser Environment]
A3 --> E[Crawl Behavior]
A4 --> F[LLM Integration]
E --> B1
E --> B2
E --> B3
E --> B4
E --> B5
end
subgraph "Runtime Flow"
G[Crawler Instance] --> D
G --> H[Execute Crawl]
H --> E
H --> F
H --> I[CrawlResult]
end
style A fill:#e3f2fd
style C fill:#fff3e0
style G fill:#e8f5e8
style I fill:#c8e6c9
```
### Advanced Configuration Decision Matrix
```mermaid
flowchart TD
A[Advanced Configuration Needed] --> B{Primary Use Case?}
B -->|Bot Detection Avoidance| C[Anti-Detection Setup]
B -->|Geographic Simulation| D[Identity-Based Config]
B -->|Multi-Step Workflows| E[Session Management]
B -->|Network Reliability| F[Proxy Configuration]
B -->|Content Precision| G[Selector Strategy]
C --> C1[Random User Agents]
C --> C2[Behavioral Simulation]
C --> C3[Navigator Override]
C --> C4[Magic Mode]
D --> D1[Geolocation Setup]
D --> D2[Locale Configuration]
D --> D3[Timezone Setting]
D --> D4[Browser Fingerprinting]
E --> E1[Session ID Management]
E --> E2[JS-Only Navigation]
E --> E3[Shared Data Context]
E --> E4[Session Cleanup]
F --> F1[Single Proxy]
F --> F2[Proxy Rotation]
F --> F3[Failover Strategy]
F --> F4[Health Monitoring]
G --> G1[css_selector for Subset]
G --> G2[target_elements for Focus]
G --> G3[excluded_selector for Removal]
G --> G4[Hierarchical Selection]
C1 --> H[Production Configuration]
C2 --> H
C3 --> H
C4 --> H
D1 --> H
D2 --> H
D3 --> H
D4 --> H
E1 --> H
E2 --> H
E3 --> H
E4 --> H
F1 --> H
F2 --> H
F3 --> H
F4 --> H
G1 --> H
G2 --> H
G3 --> H
G4 --> H
style H fill:#c8e6c9
style C fill:#ff9800
style D fill:#9c27b0
style E fill:#2196f3
style F fill:#4caf50
style G fill:#ff5722
```
## Advanced Features Workflows and Architecture
Visual representations of advanced crawling capabilities, session management, hooks system, and performance optimization strategies.
### File Download Workflow
```mermaid
sequenceDiagram
participant User
participant Crawler
participant Browser
participant FileSystem
participant Page
User->>Crawler: Configure downloads_path
Crawler->>Browser: Create context with download handling
Browser-->>Crawler: Context ready
Crawler->>Page: Navigate to target URL
Page-->>Crawler: Page loaded
Crawler->>Page: Execute download JavaScript
Page->>Page: Find download links (.pdf, .zip, etc.)
loop For each download link
Page->>Browser: Click download link
Browser->>FileSystem: Save file to downloads_path
FileSystem-->>Browser: File saved
Browser-->>Page: Download complete
end
Page-->>Crawler: All downloads triggered
Crawler->>FileSystem: Check downloaded files
FileSystem-->>Crawler: List of file paths
Crawler-->>User: CrawlResult with downloaded_files[]
Note over User,FileSystem: Files available in downloads_path
```
### Hooks Execution Flow
```mermaid
flowchart TD
A[Start Crawl] --> B[on_browser_created Hook]
B --> C[Browser Instance Created]
C --> D[on_page_context_created Hook]
D --> E[Page & Context Setup]
E --> F[before_goto Hook]
F --> G[Navigate to URL]
G --> H[after_goto Hook]
H --> I[Page Loaded]
I --> J[before_retrieve_html Hook]
J --> K[Extract HTML Content]
K --> L[Return CrawlResult]
subgraph "Hook Capabilities"
B1[Route Filtering]
B2[Authentication]
B3[Custom Headers]
B4[Viewport Setup]
B5[Content Manipulation]
end
D --> B1
F --> B2
F --> B3
D --> B4
J --> B5
style A fill:#e1f5fe
style L fill:#c8e6c9
style B fill:#fff3e0
style D fill:#f3e5f5
style F fill:#e8f5e8
style H fill:#fce4ec
style J fill:#fff9c4
```
### Session Management State Machine
```mermaid
stateDiagram-v2
[*] --> SessionCreated: session_id provided
SessionCreated --> PageLoaded: Initial arun()
PageLoaded --> JavaScriptExecution: js_code executed
JavaScriptExecution --> ContentUpdated: DOM modified
ContentUpdated --> NextOperation: js_only=True
NextOperation --> JavaScriptExecution: More interactions
NextOperation --> SessionMaintained: Keep session alive
NextOperation --> SessionClosed: kill_session()
SessionMaintained --> PageLoaded: Navigate to new URL
SessionMaintained --> JavaScriptExecution: Continue interactions
SessionClosed --> [*]: Session terminated
note right of SessionCreated
Browser tab created
Context preserved
end note
note right of ContentUpdated
State maintained
Cookies preserved
Local storage intact
end note
note right of SessionClosed
Clean up resources
Release browser tab
end note
```
### Lazy Loading & Dynamic Content Strategy
```mermaid
flowchart TD
A[Page Load] --> B{Content Type?}
B -->|Static Content| C[Standard Extraction]
B -->|Lazy Loaded| D[Enable scan_full_page]
B -->|Infinite Scroll| E[Custom Scroll Strategy]
B -->|Load More Button| F[JavaScript Interaction]
D --> D1[Automatic Scrolling]
D1 --> D2[Wait for Images]
D2 --> D3[Content Stabilization]
E --> E1[Detect Scroll Triggers]
E1 --> E2[Progressive Loading]
E2 --> E3[Monitor Content Changes]
F --> F1[Find Load More Button]
F1 --> F2[Click and Wait]
F2 --> F3{More Content?}
F3 -->|Yes| F1
F3 -->|No| G[Complete Extraction]
D3 --> G
E3 --> G
C --> G
G --> H[Return Enhanced Content]
subgraph "Optimization Techniques"
I[exclude_external_images]
J[image_score_threshold]
K[wait_for selectors]
L[scroll_delay tuning]
end
D --> I
E --> J
F --> K
D1 --> L
style A fill:#e1f5fe
style H fill:#c8e6c9
style D fill:#fff3e0
style E fill:#f3e5f5
style F fill:#e8f5e8
```
### Network & Console Monitoring Architecture
```mermaid
graph TB
subgraph "Browser Context"
A[Web Page] --> B[Network Requests]
A --> C[Console Messages]
A --> D[Resource Loading]
end
subgraph "Monitoring Layer"
B --> E[Request Interceptor]
C --> F[Console Listener]
D --> G[Resource Monitor]
E --> H[Request Events]
E --> I[Response Events]
E --> J[Failure Events]
F --> K[Log Messages]
F --> L[Error Messages]
F --> M[Warning Messages]
end
subgraph "Data Collection"
H --> N[Request Details]
I --> O[Response Analysis]
J --> P[Failure Tracking]
K --> Q[Debug Information]
L --> R[Error Analysis]
M --> S[Performance Insights]
end
subgraph "Output Aggregation"
N --> T[network_requests Array]
O --> T
P --> T
Q --> U[console_messages Array]
R --> U
S --> U
end
T --> V[CrawlResult]
U --> V
style V fill:#c8e6c9
style E fill:#fff3e0
style F fill:#f3e5f5
style T fill:#e8f5e8
style U fill:#fce4ec
```
### Multi-Step Workflow Sequence
```mermaid
sequenceDiagram
participant User
participant Crawler
participant Session
participant Page
participant Server
User->>Crawler: Step 1 - Initial load
Crawler->>Session: Create session_id
Session->>Page: New browser tab
Page->>Server: GET /step1
Server-->>Page: Page content
Page-->>Crawler: Content ready
Crawler-->>User: Result 1
User->>Crawler: Step 2 - Navigate (js_only=true)
Crawler->>Session: Reuse existing session
Session->>Page: Execute JavaScript
Page->>Page: Click next button
Page->>Server: Navigate to /step2
Server-->>Page: New content
Page-->>Crawler: Updated content
Crawler-->>User: Result 2
User->>Crawler: Step 3 - Form submission
Crawler->>Session: Continue session
Session->>Page: Execute form JS
Page->>Page: Fill form fields
Page->>Server: POST form data
Server-->>Page: Results page
Page-->>Crawler: Final content
Crawler-->>User: Result 3
User->>Crawler: Cleanup
Crawler->>Session: kill_session()
Session->>Page: Close tab
Session-->>Crawler: Session terminated
Note over User,Server: State preserved across steps
Note over Session: Cookies, localStorage maintained
```
### SSL Certificate Analysis Flow
```mermaid
flowchart LR
A[Enable SSL Fetch] --> B[HTTPS Connection]
B --> C[Certificate Retrieval]
C --> D[Certificate Analysis]
D --> E[Basic Info]
D --> F[Validity Check]
D --> G[Chain Verification]
D --> H[Security Assessment]
E --> E1[Issuer Details]
E --> E2[Subject Information]
E --> E3[Serial Number]
F --> F1[Not Before Date]
F --> F2[Not After Date]
F --> F3[Expiration Warning]
G --> G1[Root CA]
G --> G2[Intermediate Certs]
G --> G3[Trust Path]
H --> H1[Key Length]
H --> H2[Signature Algorithm]
H --> H3[Vulnerabilities]
subgraph "Export Formats"
I[JSON Format]
J[PEM Format]
K[DER Format]
end
E1 --> I
F1 --> I
G1 --> I
H1 --> I
I --> J
J --> K
style A fill:#e1f5fe
style D fill:#fff3e0
style I fill:#e8f5e8
style J fill:#f3e5f5
style K fill:#fce4ec
```
### Performance Optimization Decision Tree
```mermaid
flowchart TD
A[Performance Optimization] --> B{Primary Goal?}
B -->|Speed| C[Fast Crawling Mode]
B -->|Resource Usage| D[Memory Optimization]
B -->|Scale| E[Batch Processing]
B -->|Quality| F[Comprehensive Extraction]
C --> C1[text_mode=True]
C --> C2[exclude_all_images=True]
C --> C3[excluded_tags=['script','style']]
C --> C4[page_timeout=30000]
D --> D1[light_mode=True]
D --> D2[headless=True]
D --> D3[semaphore_count=3]
D --> D4[disable monitoring]
E --> E1[stream=True]
E --> E2[cache_mode=ENABLED]
E --> E3[arun_many()]
E --> E4[concurrent batches]
F --> F1[wait_for_images=True]
F --> F2[process_iframes=True]
F --> F3[capture_network=True]
F --> F4[screenshot=True]
subgraph "Trade-offs"
G[Speed vs Quality]
H[Memory vs Features]
I[Scale vs Detail]
end
C --> G
D --> H
E --> I
subgraph "Monitoring Metrics"
J[Response Time]
K[Memory Usage]
L[Success Rate]
M[Content Quality]
end
C1 --> J
D1 --> K
E1 --> L
F1 --> M
style A fill:#e1f5fe
style C fill:#e8f5e8
style D fill:#fff3e0
style E fill:#f3e5f5
style F fill:#fce4ec
```
### Advanced Page Interaction Matrix
```mermaid
graph LR
subgraph "Interaction Types"
A[Form Filling]
B[Dynamic Loading]
C[Modal Handling]
D[Scroll Interactions]
E[Button Clicks]
end
subgraph "Detection Methods"
F[CSS Selectors]
G[JavaScript Conditions]
H[Element Visibility]
I[Content Changes]
J[Network Activity]
end
subgraph "Automation Features"
K[simulate_user=True]
L[magic=True]
M[remove_overlay_elements=True]
N[override_navigator=True]
O[scan_full_page=True]
end
subgraph "Wait Strategies"
P[wait_for CSS]
Q[wait_for JS]
R[wait_for_images]
S[delay_before_return]
T[custom timeouts]
end
A --> F
A --> K
A --> P
B --> G
B --> O
B --> Q
C --> H
C --> L
C --> M
D --> I
D --> O
D --> S
E --> F
E --> K
E --> T
style A fill:#e8f5e8
style B fill:#fff3e0
style C fill:#f3e5f5
style D fill:#fce4ec
style E fill:#e1f5fe
```
### Input Source Processing Flow
```mermaid
flowchart TD
A[Input Source] --> B{Input Type?}
B -->|URL| C[Web Request]
B -->|file://| D[Local File]
B -->|raw:| E[Raw HTML]
C --> C1[HTTP/HTTPS Request]
C1 --> C2[Browser Navigation]
C2 --> C3[Page Rendering]
C3 --> F[Content Processing]
D --> D1[File System Access]
D1 --> D2[Read HTML File]
D2 --> D3[Parse Content]
D3 --> F
E --> E1[Parse Raw HTML]
E1 --> E2[Create Virtual Page]
E2 --> E3[Direct Processing]
E3 --> F
F --> G[Common Processing Pipeline]
G --> H[Markdown Generation]
G --> I[Link Extraction]
G --> J[Media Processing]
G --> K[Data Extraction]
H --> L[CrawlResult]
I --> L
J --> L
K --> L
subgraph "Processing Features"
M[Same extraction strategies]
N[Same filtering options]
O[Same output formats]
P[Consistent results]
end
F --> M
F --> N
F --> O
F --> P
style A fill:#e1f5fe
style L fill:#c8e6c9
style C fill:#e8f5e8
style D fill:#fff3e0
style E fill:#f3e5f5
```
**📖 Learn more:** [Advanced Features Guide](https://docs.crawl4ai.com/advanced/advanced-features/), [Session Management](https://docs.crawl4ai.com/advanced/session-management/), [Hooks System](https://docs.crawl4ai.com/advanced/hooks-auth/), [Performance Optimization](https://docs.crawl4ai.com/advanced/performance/)
**📖 Learn more:** [Identity-Based Crawling](https://docs.crawl4ai.com/advanced/identity-based-crawling/), [Session Management](https://docs.crawl4ai.com/advanced/session-management/), [Proxy & Security](https://docs.crawl4ai.com/advanced/proxy-security/), [Content Selection](https://docs.crawl4ai.com/core/content-selection/)
**📖 Learn more:** [Configuration Reference](https://docs.crawl4ai.com/api/parameters/), [Best Practices](https://docs.crawl4ai.com/core/browser-crawler-config/), [Advanced Configuration](https://docs.crawl4ai.com/advanced/advanced-features/)
---
## Extraction Strategy Workflows and Architecture
Visual representations of Crawl4AI's data extraction approaches, strategy selection, and processing workflows.
### Extraction Strategy Decision Tree
```mermaid
flowchart TD
A[Content to Extract] --> B{Content Type?}
B -->|Simple Patterns| C[Common Data Types]
B -->|Structured HTML| D[Predictable Structure]
B -->|Complex Content| E[Requires Reasoning]
B -->|Mixed Content| F[Multiple Data Types]
C --> C1{Pattern Type?}
C1 -->|Email, Phone, URLs| C2[Built-in Regex Patterns]
C1 -->|Custom Patterns| C3[Custom Regex Strategy]
C1 -->|LLM-Generated| C4[One-time Pattern Generation]
D --> D1{Selector Type?}
D1 -->|CSS Selectors| D2[JsonCssExtractionStrategy]
D1 -->|XPath Expressions| D3[JsonXPathExtractionStrategy]
D1 -->|Need Schema?| D4[Auto-generate Schema with LLM]
E --> E1{LLM Provider?}
E1 -->|OpenAI/Anthropic| E2[Cloud LLM Strategy]
E1 -->|Local Ollama| E3[Local LLM Strategy]
E1 -->|Cost-sensitive| E4[Hybrid: Generate Schema Once]
F --> F1[Multi-Strategy Approach]
F1 --> F2[1. Regex for Patterns]
F1 --> F3[2. CSS for Structure]
F1 --> F4[3. LLM for Complex Analysis]
C2 --> G[Fast Extraction ⚡]
C3 --> G
C4 --> H[Cached Pattern Reuse]
D2 --> I[Schema-based Extraction 🏗️]
D3 --> I
D4 --> J[Generated Schema Cache]
E2 --> K[Intelligent Parsing 🧠]
E3 --> K
E4 --> L[Hybrid Cost-Effective]
F2 --> M[Comprehensive Results 📊]
F3 --> M
F4 --> M
style G fill:#c8e6c9
style I fill:#e3f2fd
style K fill:#fff3e0
style M fill:#f3e5f5
style H fill:#e8f5e8
style J fill:#e8f5e8
style L fill:#ffecb3
```
### LLM Extraction Strategy Workflow
```mermaid
sequenceDiagram
participant User
participant Crawler
participant LLMStrategy
participant Chunker
participant LLMProvider
participant Parser
User->>Crawler: Configure LLMExtractionStrategy
User->>Crawler: arun(url, config)
Crawler->>Crawler: Navigate to URL
Crawler->>Crawler: Extract content (HTML/Markdown)
Crawler->>LLMStrategy: Process content
LLMStrategy->>LLMStrategy: Check content size
alt Content > chunk_threshold
LLMStrategy->>Chunker: Split into chunks with overlap
Chunker-->>LLMStrategy: Return chunks[]
loop For each chunk
LLMStrategy->>LLMProvider: Send chunk + schema + instruction
LLMProvider-->>LLMStrategy: Return structured JSON
end
LLMStrategy->>LLMStrategy: Merge chunk results
else Content <= threshold
LLMStrategy->>LLMProvider: Send full content + schema
LLMProvider-->>LLMStrategy: Return structured JSON
end
LLMStrategy->>Parser: Validate JSON schema
Parser-->>LLMStrategy: Validated data
LLMStrategy->>LLMStrategy: Track token usage
LLMStrategy-->>Crawler: Return extracted_content
Crawler-->>User: CrawlResult with JSON data
User->>LLMStrategy: show_usage()
LLMStrategy-->>User: Token count & estimated cost
```
### Schema-Based Extraction Architecture
```mermaid
graph TB
subgraph "Schema Definition"
A[JSON Schema] --> A1[baseSelector]
A --> A2[fields[]]
A --> A3[nested structures]
A2 --> A4[CSS/XPath selectors]
A2 --> A5[Data types: text, html, attribute]
A2 --> A6[Default values]
A3 --> A7[nested objects]
A3 --> A8[nested_list arrays]
A3 --> A9[simple lists]
end
subgraph "Extraction Engine"
B[HTML Content] --> C[Selector Engine]
C --> C1[CSS Selector Parser]
C --> C2[XPath Evaluator]
C1 --> D[Element Matcher]
C2 --> D
D --> E[Type Converter]
E --> E1[Text Extraction]
E --> E2[HTML Preservation]
E --> E3[Attribute Extraction]
E --> E4[Nested Processing]
end
subgraph "Result Processing"
F[Raw Extracted Data] --> G[Structure Builder]
G --> G1[Object Construction]
G --> G2[Array Assembly]
G --> G3[Type Validation]
G1 --> H[JSON Output]
G2 --> H
G3 --> H
end
A --> C
E --> F
H --> I[extracted_content]
style A fill:#e3f2fd
style C fill:#f3e5f5
style G fill:#e8f5e8
style H fill:#c8e6c9
```
### Automatic Schema Generation Process
```mermaid
stateDiagram-v2
[*] --> CheckCache
CheckCache --> CacheHit: Schema exists
CheckCache --> SamplePage: Schema missing
CacheHit --> LoadSchema
LoadSchema --> FastExtraction
SamplePage --> ExtractHTML: Crawl sample URL
ExtractHTML --> LLMAnalysis: Send HTML to LLM
LLMAnalysis --> GenerateSchema: Create CSS/XPath selectors
GenerateSchema --> ValidateSchema: Test generated schema
ValidateSchema --> SchemaWorks: Valid selectors
ValidateSchema --> RefineSchema: Invalid selectors
RefineSchema --> LLMAnalysis: Iterate with feedback
SchemaWorks --> CacheSchema: Save for reuse
CacheSchema --> FastExtraction: Use cached schema
FastExtraction --> [*]: No more LLM calls needed
note right of CheckCache : One-time LLM cost
note right of FastExtraction : Unlimited fast reuse
note right of CacheSchema : JSON file storage
```
### Multi-Strategy Extraction Pipeline
```mermaid
flowchart LR
A[Web Page Content] --> B[Strategy Pipeline]
subgraph B["Extraction Pipeline"]
B1[Stage 1: Regex Patterns]
B2[Stage 2: Schema-based CSS]
B3[Stage 3: LLM Analysis]
B1 --> B1a[Email addresses]
B1 --> B1b[Phone numbers]
B1 --> B1c[URLs and links]
B1 --> B1d[Currency amounts]
B2 --> B2a[Structured products]
B2 --> B2b[Article metadata]
B2 --> B2c[User reviews]
B2 --> B2d[Navigation links]
B3 --> B3a[Sentiment analysis]
B3 --> B3b[Key topics]
B3 --> B3c[Entity recognition]
B3 --> B3d[Content summary]
end
B1a --> C[Result Merger]
B1b --> C
B1c --> C
B1d --> C
B2a --> C
B2b --> C
B2c --> C
B2d --> C
B3a --> C
B3b --> C
B3c --> C
B3d --> C
C --> D[Combined JSON Output]
D --> E[Final CrawlResult]
style B1 fill:#c8e6c9
style B2 fill:#e3f2fd
style B3 fill:#fff3e0
style C fill:#f3e5f5
```
### Performance Comparison Matrix
```mermaid
graph TD
subgraph "Strategy Performance"
A[Extraction Strategy Comparison]
subgraph "Speed ⚡"
S1[Regex: ~10ms]
S2[CSS Schema: ~50ms]
S3[XPath: ~100ms]
S4[LLM: ~2-10s]
end
subgraph "Accuracy 🎯"
A1[Regex: Pattern-dependent]
A2[CSS: High for structured]
A3[XPath: Very high]
A4[LLM: Excellent for complex]
end
subgraph "Cost 💰"
C1[Regex: Free]
C2[CSS: Free]
C3[XPath: Free]
C4[LLM: $0.001-0.01 per page]
end
subgraph "Complexity 🔧"
X1[Regex: Simple patterns only]
X2[CSS: Structured HTML]
X3[XPath: Complex selectors]
X4[LLM: Any content type]
end
end
style S1 fill:#c8e6c9
style S2 fill:#e8f5e8
style S3 fill:#fff3e0
style S4 fill:#ffcdd2
style A2 fill:#e8f5e8
style A3 fill:#c8e6c9
style A4 fill:#c8e6c9
style C1 fill:#c8e6c9
style C2 fill:#c8e6c9
style C3 fill:#c8e6c9
style C4 fill:#fff3e0
style X1 fill:#ffcdd2
style X2 fill:#e8f5e8
style X3 fill:#c8e6c9
style X4 fill:#c8e6c9
```
### Regex Pattern Strategy Flow
```mermaid
flowchart TD
A[Regex Extraction] --> B{Pattern Source?}
B -->|Built-in| C[Use Predefined Patterns]
B -->|Custom| D[Define Custom Regex]
B -->|LLM-Generated| E[Generate with AI]
C --> C1[Email Pattern]
C --> C2[Phone Pattern]
C --> C3[URL Pattern]
C --> C4[Currency Pattern]
C --> C5[Date Pattern]
D --> D1[Write Custom Regex]
D --> D2[Test Pattern]
D --> D3{Pattern Works?}
D3 -->|No| D1
D3 -->|Yes| D4[Use Pattern]
E --> E1[Provide Sample Content]
E --> E2[LLM Analyzes Content]
E --> E3[Generate Optimized Regex]
E --> E4[Cache Pattern for Reuse]
C1 --> F[Pattern Matching]
C2 --> F
C3 --> F
C4 --> F
C5 --> F
D4 --> F
E4 --> F
F --> G[Extract Matches]
G --> H[Group by Pattern Type]
H --> I[JSON Output with Labels]
style C fill:#e8f5e8
style D fill:#e3f2fd
style E fill:#fff3e0
style F fill:#f3e5f5
```
### Complex Schema Structure Visualization
```mermaid
graph TB
subgraph "E-commerce Schema Example"
A[Category baseSelector] --> B[Category Fields]
A --> C[Products nested_list]
B --> B1[category_name]
B --> B2[category_id attribute]
B --> B3[category_url attribute]
C --> C1[Product baseSelector]
C1 --> C2[name text]
C1 --> C3[price text]
C1 --> C4[Details nested object]
C1 --> C5[Features list]
C1 --> C6[Reviews nested_list]
C4 --> C4a[brand text]
C4 --> C4b[model text]
C4 --> C4c[specs html]
C5 --> C5a[feature text array]
C6 --> C6a[reviewer text]
C6 --> C6b[rating attribute]
C6 --> C6c[comment text]
C6 --> C6d[date attribute]
end
subgraph "JSON Output Structure"
D[categories array] --> D1[category object]
D1 --> D2[category_name]
D1 --> D3[category_id]
D1 --> D4[products array]
D4 --> D5[product object]
D5 --> D6[name, price]
D5 --> D7[details object]
D5 --> D8[features array]
D5 --> D9[reviews array]
D7 --> D7a[brand, model, specs]
D8 --> D8a[feature strings]
D9 --> D9a[review objects]
end
A -.-> D
B1 -.-> D2
C2 -.-> D6
C4 -.-> D7
C5 -.-> D8
C6 -.-> D9
style A fill:#e3f2fd
style C fill:#f3e5f5
style C4 fill:#e8f5e8
style D fill:#fff3e0
```
### Error Handling and Fallback Strategy
```mermaid
stateDiagram-v2
[*] --> PrimaryStrategy
PrimaryStrategy --> Success: Extraction successful
PrimaryStrategy --> ValidationFailed: Invalid data
PrimaryStrategy --> ExtractionFailed: No matches found
PrimaryStrategy --> TimeoutError: LLM timeout
ValidationFailed --> FallbackStrategy: Try alternative
ExtractionFailed --> FallbackStrategy: Try alternative
TimeoutError --> FallbackStrategy: Try alternative
FallbackStrategy --> FallbackSuccess: Fallback works
FallbackStrategy --> FallbackFailed: All strategies failed
FallbackSuccess --> Success: Return results
FallbackFailed --> ErrorReport: Log failure details
Success --> [*]: Complete
ErrorReport --> [*]: Return empty results
note right of PrimaryStrategy : Try fastest/most accurate first
note right of FallbackStrategy : Use simpler but reliable method
note left of ErrorReport : Provide debugging information
```
### Token Usage and Cost Optimization
```mermaid
flowchart TD
A[LLM Extraction Request] --> B{Content Size Check}
B -->|Small < 1200 tokens| C[Single LLM Call]
B -->|Large > 1200 tokens| D[Chunking Strategy]
C --> C1[Send full content]
C1 --> C2[Parse JSON response]
C2 --> C3[Track token usage]
D --> D1[Split into chunks]
D1 --> D2[Add overlap between chunks]
D2 --> D3[Process chunks in parallel]
D3 --> D4[Chunk 1 → LLM]
D3 --> D5[Chunk 2 → LLM]
D3 --> D6[Chunk N → LLM]
D4 --> D7[Merge results]
D5 --> D7
D6 --> D7
D7 --> D8[Deduplicate data]
D8 --> D9[Aggregate token usage]
C3 --> E[Cost Calculation]
D9 --> E
E --> F[Usage Report]
F --> F1[Prompt tokens: X]
F --> F2[Completion tokens: Y]
F --> F3[Total cost: $Z]
style C fill:#c8e6c9
style D fill:#fff3e0
style E fill:#e3f2fd
style F fill:#f3e5f5
```
**📖 Learn more:** [LLM Strategies](https://docs.crawl4ai.com/extraction/llm-strategies/), [Schema-Based Extraction](https://docs.crawl4ai.com/extraction/no-llm-strategies/), [Pattern Matching](https://docs.crawl4ai.com/extraction/no-llm-strategies/#regexextractionstrategy), [Performance Optimization](https://docs.crawl4ai.com/advanced/multi-url-crawling/)
---
## Multi-URL Crawling Workflows and Architecture
Visual representations of concurrent crawling patterns, resource management, and monitoring systems for handling multiple URLs efficiently.
### Multi-URL Processing Modes
```mermaid
flowchart TD
A[Multi-URL Crawling Request] --> B{Processing Mode?}
B -->|Batch Mode| C[Collect All URLs]
B -->|Streaming Mode| D[Process URLs Individually]
C --> C1[Queue All URLs]
C1 --> C2[Execute Concurrently]
C2 --> C3[Wait for All Completion]
C3 --> C4[Return Complete Results Array]
D --> D1[Queue URLs]
D1 --> D2[Start First Batch]
D2 --> D3[Yield Results as Available]
D3 --> D4{More URLs?}
D4 -->|Yes| D5[Start Next URLs]
D4 -->|No| D6[Stream Complete]
D5 --> D3
C4 --> E[Process Results]
D6 --> E
E --> F[Success/Failure Analysis]
F --> G[End]
style C fill:#e3f2fd
style D fill:#f3e5f5
style C4 fill:#c8e6c9
style D6 fill:#c8e6c9
```
### Memory-Adaptive Dispatcher Flow
```mermaid
stateDiagram-v2
[*] --> Initializing
Initializing --> MonitoringMemory: Start dispatcher
MonitoringMemory --> CheckingMemory: Every check_interval
CheckingMemory --> MemoryOK: Memory < threshold
CheckingMemory --> MemoryHigh: Memory >= threshold
MemoryOK --> DispatchingTasks: Start new crawls
MemoryHigh --> WaitingForMemory: Pause dispatching
DispatchingTasks --> TaskRunning: Launch crawler
TaskRunning --> TaskCompleted: Crawl finished
TaskRunning --> TaskFailed: Crawl error
TaskCompleted --> MonitoringMemory: Update stats
TaskFailed --> MonitoringMemory: Update stats
WaitingForMemory --> CheckingMemory: Wait timeout
WaitingForMemory --> MonitoringMemory: Memory freed
note right of MemoryHigh: Prevents OOM crashes
note right of DispatchingTasks: Respects max_session_permit
note right of WaitingForMemory: Configurable timeout
```
### Concurrent Crawling Architecture
```mermaid
graph TB
subgraph "URL Queue Management"
A[URL Input List] --> B[URL Queue]
B --> C[Priority Scheduler]
C --> D[Batch Assignment]
end
subgraph "Dispatcher Layer"
E[Memory Adaptive Dispatcher]
F[Semaphore Dispatcher]
G[Rate Limiter]
H[Resource Monitor]
E --> I[Memory Checker]
F --> J[Concurrency Controller]
G --> K[Delay Calculator]
H --> L[System Stats]
end
subgraph "Crawler Pool"
M[Crawler Instance 1]
N[Crawler Instance 2]
O[Crawler Instance 3]
P[Crawler Instance N]
M --> Q[Browser Session 1]
N --> R[Browser Session 2]
O --> S[Browser Session 3]
P --> T[Browser Session N]
end
subgraph "Result Processing"
U[Result Collector]
V[Success Handler]
W[Error Handler]
X[Retry Queue]
Y[Final Results]
end
D --> E
D --> F
E --> M
F --> N
G --> O
H --> P
Q --> U
R --> U
S --> U
T --> U
U --> V
U --> W
W --> X
X --> B
V --> Y
style E fill:#e3f2fd
style F fill:#f3e5f5
style G fill:#e8f5e8
style H fill:#fff3e0
```
### Rate Limiting and Backoff Strategy
```mermaid
sequenceDiagram
participant C as Crawler
participant RL as Rate Limiter
participant S as Server
participant D as Dispatcher
C->>RL: Request to crawl URL
RL->>RL: Calculate delay
RL->>RL: Apply base delay (1-3s)
RL->>C: Delay applied
C->>S: HTTP Request
alt Success Response
S-->>C: 200 OK + Content
C->>RL: Report success
RL->>RL: Reset failure count
C->>D: Return successful result
else Rate Limited
S-->>C: 429 Too Many Requests
C->>RL: Report rate limit
RL->>RL: Exponential backoff
RL->>RL: Increase delay (up to max_delay)
RL->>C: Apply longer delay
C->>S: Retry request after delay
else Server Error
S-->>C: 503 Service Unavailable
C->>RL: Report server error
RL->>RL: Moderate backoff
RL->>C: Retry with backoff
else Max Retries Exceeded
RL->>C: Stop retrying
C->>D: Return failed result
end
```
### Large-Scale Crawling Workflow
```mermaid
flowchart TD
A[Load URL List 10k+ URLs] --> B[Initialize Dispatcher]
B --> C{Select Dispatcher Type}
C -->|Memory Constrained| D[Memory Adaptive]
C -->|Fixed Resources| E[Semaphore Based]
D --> F[Set Memory Threshold 70%]
E --> G[Set Concurrency Limit]
F --> H[Configure Monitoring]
G --> H
H --> I[Start Crawling Process]
I --> J[Monitor System Resources]
J --> K{Memory Usage?}
K -->|< Threshold| L[Continue Dispatching]
K -->|>= Threshold| M[Pause New Tasks]
L --> N[Process Results Stream]
M --> O[Wait for Memory]
O --> K
N --> P{Result Type?}
P -->|Success| Q[Save to Database]
P -->|Failure| R[Log Error]
Q --> S[Update Progress Counter]
R --> S
S --> T{More URLs?}
T -->|Yes| U[Get Next Batch]
T -->|No| V[Generate Final Report]
U --> L
V --> W[Analysis Complete]
style A fill:#e1f5fe
style D fill:#e8f5e8
style E fill:#f3e5f5
style V fill:#c8e6c9
style W fill:#a5d6a7
```
### Real-Time Monitoring Dashboard Flow
```mermaid
graph LR
subgraph "Data Collection"
A[Crawler Tasks] --> B[Performance Metrics]
A --> C[Memory Usage]
A --> D[Success/Failure Rates]
A --> E[Response Times]
end
subgraph "Monitor Processing"
F[CrawlerMonitor] --> G[Aggregate Statistics]
F --> H[Display Formatter]
F --> I[Update Scheduler]
end
subgraph "Display Modes"
J[DETAILED Mode]
K[AGGREGATED Mode]
J --> L[Individual Task Status]
J --> M[Task-Level Metrics]
K --> N[Summary Statistics]
K --> O[Overall Progress]
end
subgraph "Output Interface"
P[Console Display]
Q[Progress Bars]
R[Status Tables]
S[Real-time Updates]
end
B --> F
C --> F
D --> F
E --> F
G --> J
G --> K
H --> J
H --> K
I --> J
I --> K
L --> P
M --> Q
N --> R
O --> S
style F fill:#e3f2fd
style J fill:#f3e5f5
style K fill:#e8f5e8
```
### Error Handling and Recovery Pattern
```mermaid
stateDiagram-v2
[*] --> ProcessingURL
ProcessingURL --> CrawlAttempt: Start crawl
CrawlAttempt --> Success: HTTP 200
CrawlAttempt --> NetworkError: Connection failed
CrawlAttempt --> RateLimit: HTTP 429
CrawlAttempt --> ServerError: HTTP 5xx
CrawlAttempt --> Timeout: Request timeout
Success --> [*]: Return result
NetworkError --> RetryCheck: Check retry count
RateLimit --> BackoffWait: Apply exponential backoff
ServerError --> RetryCheck: Check retry count
Timeout --> RetryCheck: Check retry count
BackoffWait --> RetryCheck: After delay
RetryCheck --> CrawlAttempt: retries < max_retries
RetryCheck --> Failed: retries >= max_retries
Failed --> ErrorLog: Log failure details
ErrorLog --> [*]: Return failed result
note right of BackoffWait: Exponential backoff for rate limits
note right of RetryCheck: Configurable max_retries
note right of ErrorLog: Detailed error tracking
```
### Resource Management Timeline
```mermaid
gantt
title Multi-URL Crawling Resource Management
dateFormat X
axisFormat %s
section Memory Usage
Initialize Dispatcher :0, 1
Memory Monitoring :1, 10
Peak Usage Period :3, 7
Memory Cleanup :7, 9
section Task Execution
URL Queue Setup :0, 2
Batch 1 Processing :2, 5
Batch 2 Processing :4, 7
Batch 3 Processing :6, 9
Final Results :9, 10
section Rate Limiting
Normal Delays :2, 4
Backoff Period :4, 6
Recovery Period :6, 8
section Monitoring
System Health Check :0, 10
Progress Updates :1, 9
Performance Metrics :2, 8
```
### Concurrent Processing Performance Matrix
```mermaid
graph TD
subgraph "Input Factors"
A[Number of URLs]
B[Concurrency Level]
C[Memory Threshold]
D[Rate Limiting]
end
subgraph "Processing Characteristics"
A --> E[Low 1-100 URLs]
A --> F[Medium 100-1k URLs]
A --> G[High 1k-10k URLs]
A --> H[Very High 10k+ URLs]
B --> I[Conservative 1-5]
B --> J[Moderate 5-15]
B --> K[Aggressive 15-30]
C --> L[Strict 60-70%]
C --> M[Balanced 70-80%]
C --> N[Relaxed 80-90%]
end
subgraph "Recommended Configurations"
E --> O[Simple Semaphore]
F --> P[Memory Adaptive Basic]
G --> Q[Memory Adaptive Advanced]
H --> R[Memory Adaptive + Monitoring]
I --> O
J --> P
K --> Q
K --> R
L --> Q
M --> P
N --> O
end
style O fill:#c8e6c9
style P fill:#fff3e0
style Q fill:#ffecb3
style R fill:#ffcdd2
```
**📖 Learn more:** [Multi-URL Crawling Guide](https://docs.crawl4ai.com/advanced/multi-url-crawling/), [Dispatcher Configuration](https://docs.crawl4ai.com/advanced/crawl-dispatcher/), [Performance Optimization](https://docs.crawl4ai.com/advanced/multi-url-crawling/#performance-optimization)
---
## Deep Crawling Workflows and Architecture
Visual representations of multi-level website exploration, filtering strategies, and intelligent crawling patterns.
### Deep Crawl Strategy Overview
```mermaid
flowchart TD
A[Start Deep Crawl] --> B{Strategy Selection}
B -->|Explore All Levels| C[BFS Strategy]
B -->|Dive Deep Fast| D[DFS Strategy]
B -->|Smart Prioritization| E[Best-First Strategy]
C --> C1[Breadth-First Search]
C1 --> C2[Process all depth 0 links]
C2 --> C3[Process all depth 1 links]
C3 --> C4[Continue by depth level]
D --> D1[Depth-First Search]
D1 --> D2[Follow first link deeply]
D2 --> D3[Backtrack when max depth reached]
D3 --> D4[Continue with next branch]
E --> E1[Best-First Search]
E1 --> E2[Score all discovered URLs]
E2 --> E3[Process highest scoring URLs first]
E3 --> E4[Continuously re-prioritize queue]
C4 --> F[Apply Filters]
D4 --> F
E4 --> F
F --> G{Filter Chain Processing}
G -->|Domain Filter| G1[Check allowed/blocked domains]
G -->|URL Pattern Filter| G2[Match URL patterns]
G -->|Content Type Filter| G3[Verify content types]
G -->|SEO Filter| G4[Evaluate SEO quality]
G -->|Content Relevance| G5[Score content relevance]
G1 --> H{Passed All Filters?}
G2 --> H
G3 --> H
G4 --> H
G5 --> H
H -->|Yes| I[Add to Crawl Queue]
H -->|No| J[Discard URL]
I --> K{Processing Mode}
K -->|Streaming| L[Process Immediately]
K -->|Batch| M[Collect All Results]
L --> N[Stream Result to User]
M --> O[Return Complete Result Set]
J --> P{More URLs in Queue?}
N --> P
O --> P
P -->|Yes| Q{Within Limits?}
P -->|No| R[Deep Crawl Complete]
Q -->|Max Depth OK| S{Max Pages OK}
Q -->|Max Depth Exceeded| T[Skip Deeper URLs]
S -->|Under Limit| U[Continue Crawling]
S -->|Limit Reached| R
T --> P
U --> F
style A fill:#e1f5fe
style R fill:#c8e6c9
style C fill:#fff3e0
style D fill:#f3e5f5
style E fill:#e8f5e8
```
### Deep Crawl Strategy Comparison
```mermaid
graph TB
subgraph "BFS - Breadth-First Search"
BFS1[Level 0: Start URL]
BFS2[Level 1: All direct links]
BFS3[Level 2: All second-level links]
BFS4[Level 3: All third-level links]
BFS1 --> BFS2
BFS2 --> BFS3
BFS3 --> BFS4
BFS_NOTE[Complete each depth before going deeper<br/>Good for site mapping<br/>Memory intensive for wide sites]
end
subgraph "DFS - Depth-First Search"
DFS1[Start URL]
DFS2[First Link → Deep]
DFS3[Follow until max depth]
DFS4[Backtrack and try next]
DFS1 --> DFS2
DFS2 --> DFS3
DFS3 --> DFS4
DFS4 --> DFS2
DFS_NOTE[Go deep on first path<br/>Memory efficient<br/>May miss important pages]
end
subgraph "Best-First - Priority Queue"
BF1[Start URL]
BF2[Score all discovered links]
BF3[Process highest scoring first]
BF4[Continuously re-prioritize]
BF1 --> BF2
BF2 --> BF3
BF3 --> BF4
BF4 --> BF2
BF_NOTE[Intelligent prioritization<br/>Finds relevant content fast<br/>Recommended for most use cases]
end
style BFS1 fill:#e3f2fd
style DFS1 fill:#f3e5f5
style BF1 fill:#e8f5e8
style BFS_NOTE fill:#fff3e0
style DFS_NOTE fill:#fff3e0
style BF_NOTE fill:#fff3e0
```
### Filter Chain Processing Sequence
```mermaid
sequenceDiagram
participant URL as Discovered URL
participant Chain as Filter Chain
participant Domain as Domain Filter
participant Pattern as URL Pattern Filter
participant Content as Content Type Filter
participant SEO as SEO Filter
participant Relevance as Content Relevance Filter
participant Queue as Crawl Queue
URL->>Chain: Process URL
Chain->>Domain: Check domain rules
alt Domain Allowed
Domain-->>Chain: ✓ Pass
Chain->>Pattern: Check URL patterns
alt Pattern Matches
Pattern-->>Chain: ✓ Pass
Chain->>Content: Check content type
alt Content Type Valid
Content-->>Chain: ✓ Pass
Chain->>SEO: Evaluate SEO quality
alt SEO Score Above Threshold
SEO-->>Chain: ✓ Pass
Chain->>Relevance: Score content relevance
alt Relevance Score High
Relevance-->>Chain: ✓ Pass
Chain->>Queue: Add to crawl queue
Queue-->>URL: Queued for crawling
else Relevance Score Low
Relevance-->>Chain: ✗ Reject
Chain-->>URL: Filtered out - Low relevance
end
else SEO Score Low
SEO-->>Chain: ✗ Reject
Chain-->>URL: Filtered out - Poor SEO
end
else Invalid Content Type
Content-->>Chain: ✗ Reject
Chain-->>URL: Filtered out - Wrong content type
end
else Pattern Mismatch
Pattern-->>Chain: ✗ Reject
Chain-->>URL: Filtered out - Pattern mismatch
end
else Domain Blocked
Domain-->>Chain: ✗ Reject
Chain-->>URL: Filtered out - Blocked domain
end
```
### URL Lifecycle State Machine
```mermaid
stateDiagram-v2
[*] --> Discovered: Found on page
Discovered --> FilterPending: Enter filter chain
FilterPending --> DomainCheck: Apply domain filter
DomainCheck --> PatternCheck: Domain allowed
DomainCheck --> Rejected: Domain blocked
PatternCheck --> ContentCheck: Pattern matches
PatternCheck --> Rejected: Pattern mismatch
ContentCheck --> SEOCheck: Content type valid
ContentCheck --> Rejected: Invalid content
SEOCheck --> RelevanceCheck: SEO score sufficient
SEOCheck --> Rejected: Poor SEO score
RelevanceCheck --> Scored: Relevance score calculated
RelevanceCheck --> Rejected: Low relevance
Scored --> Queued: Added to priority queue
Queued --> Crawling: Selected for processing
Crawling --> Success: Page crawled successfully
Crawling --> Failed: Crawl failed
Success --> LinkExtraction: Extract new links
LinkExtraction --> [*]: Process complete
Failed --> [*]: Record failure
Rejected --> [*]: Log rejection reason
note right of Scored : Score determines priority<br/>in Best-First strategy
note right of Failed : Errors logged with<br/>depth and reason
```
### Streaming vs Batch Processing Architecture
```mermaid
graph TB
subgraph "Input"
A[Start URL] --> B[Deep Crawl Strategy]
end
subgraph "Crawl Engine"
B --> C[URL Discovery]
C --> D[Filter Chain]
D --> E[Priority Queue]
E --> F[Page Processor]
end
subgraph "Streaming Mode stream=True"
F --> G1[Process Page]
G1 --> H1[Extract Content]
H1 --> I1[Yield Result Immediately]
I1 --> J1[async for result]
J1 --> K1[Real-time Processing]
G1 --> L1[Extract Links]
L1 --> M1[Add to Queue]
M1 --> F
end
subgraph "Batch Mode stream=False"
F --> G2[Process Page]
G2 --> H2[Extract Content]
H2 --> I2[Store Result]
I2 --> N2[Result Collection]
G2 --> L2[Extract Links]
L2 --> M2[Add to Queue]
M2 --> O2{More URLs?}
O2 -->|Yes| F
O2 -->|No| P2[Return All Results]
P2 --> Q2[Batch Processing]
end
style I1 fill:#e8f5e8
style K1 fill:#e8f5e8
style P2 fill:#e3f2fd
style Q2 fill:#e3f2fd
```
### Advanced Scoring and Prioritization System
```mermaid
flowchart LR
subgraph "URL Discovery"
A[Page Links] --> B[Extract URLs]
B --> C[Normalize URLs]
end
subgraph "Scoring System"
C --> D[Keyword Relevance Scorer]
D --> D1[URL Text Analysis]
D --> D2[Keyword Matching]
D --> D3[Calculate Base Score]
D3 --> E[Additional Scoring Factors]
E --> E1[URL Structure weight: 0.2]
E --> E2[Link Context weight: 0.3]
E --> E3[Page Depth Penalty weight: 0.1]
E --> E4[Domain Authority weight: 0.4]
D1 --> F[Combined Score]
D2 --> F
D3 --> F
E1 --> F
E2 --> F
E3 --> F
E4 --> F
end
subgraph "Prioritization"
F --> G{Score Threshold}
G -->|Above Threshold| H[Priority Queue]
G -->|Below Threshold| I[Discard URL]
H --> J[Best-First Selection]
J --> K[Highest Score First]
K --> L[Process Page]
L --> M[Update Scores]
M --> N[Re-prioritize Queue]
N --> J
end
style F fill:#fff3e0
style H fill:#e8f5e8
style L fill:#e3f2fd
```
### Deep Crawl Performance and Limits
```mermaid
graph TD
subgraph "Crawl Constraints"
A[Max Depth: 2] --> A1[Prevents infinite crawling]
B[Max Pages: 50] --> B1[Controls resource usage]
C[Score Threshold: 0.3] --> C1[Quality filtering]
D[Domain Limits] --> D1[Scope control]
end
subgraph "Performance Monitoring"
E[Pages Crawled] --> F[Depth Distribution]
E --> G[Success Rate]
E --> H[Average Score]
E --> I[Processing Time]
F --> J[Performance Report]
G --> J
H --> J
I --> J
end
subgraph "Resource Management"
K[Memory Usage] --> L{Memory Threshold}
L -->|Under Limit| M[Continue Crawling]
L -->|Over Limit| N[Reduce Concurrency]
O[CPU Usage] --> P{CPU Threshold}
P -->|Normal| M
P -->|High| Q[Add Delays]
R[Network Load] --> S{Rate Limits}
S -->|OK| M
S -->|Exceeded| T[Throttle Requests]
end
M --> U[Optimal Performance]
N --> V[Reduced Performance]
Q --> V
T --> V
style U fill:#c8e6c9
style V fill:#fff3e0
style J fill:#e3f2fd
```
### Error Handling and Recovery Flow
```mermaid
sequenceDiagram
participant Strategy as Deep Crawl Strategy
participant Queue as Priority Queue
participant Crawler as Page Crawler
participant Error as Error Handler
participant Result as Result Collector
Strategy->>Queue: Get next URL
Queue-->>Strategy: Return highest priority URL
Strategy->>Crawler: Crawl page
alt Successful Crawl
Crawler-->>Strategy: Return page content
Strategy->>Result: Store successful result
Strategy->>Strategy: Extract new links
Strategy->>Queue: Add new URLs to queue
else Network Error
Crawler-->>Error: Network timeout/failure
Error->>Error: Log error with details
Error->>Queue: Mark URL as failed
Error-->>Strategy: Skip to next URL
else Parse Error
Crawler-->>Error: HTML parsing failed
Error->>Error: Log parse error
Error->>Result: Store failed result
Error-->>Strategy: Continue with next URL
else Rate Limit Hit
Crawler-->>Error: Rate limit exceeded
Error->>Error: Apply backoff strategy
Error->>Queue: Re-queue URL with delay
Error-->>Strategy: Wait before retry
else Depth Limit
Strategy->>Strategy: Check depth constraint
Strategy-->>Queue: Skip URL - too deep
else Page Limit
Strategy->>Strategy: Check page count
Strategy-->>Result: Stop crawling - limit reached
end
Strategy->>Queue: Request next URL
Queue-->>Strategy: More URLs available?
alt Queue Empty
Queue-->>Result: Crawl complete
else Queue Has URLs
Queue-->>Strategy: Continue crawling
end
```
**📖 Learn more:** [Deep Crawling Strategies](https://docs.crawl4ai.com/core/deep-crawling/), [Content Filtering](https://docs.crawl4ai.com/core/content-selection/), [Advanced Crawling Patterns](https://docs.crawl4ai.com/advanced/advanced-features/)
---
## Docker Deployment Architecture and Workflows
Visual representations of Crawl4AI Docker deployment, API architecture, configuration management, and service interactions.
### Docker Deployment Decision Flow
```mermaid
flowchart TD
A[Start Docker Deployment] --> B{Deployment Type?}
B -->|Quick Start| C[Pre-built Image]
B -->|Development| D[Docker Compose]
B -->|Custom Build| E[Manual Build]
B -->|Production| F[Production Setup]
C --> C1[docker pull unclecode/crawl4ai]
C1 --> C2{Need LLM Support?}
C2 -->|Yes| C3[Setup .llm.env]
C2 -->|No| C4[Basic run]
C3 --> C5[docker run with --env-file]
C4 --> C6[docker run basic]
D --> D1[git clone repository]
D1 --> D2[cp .llm.env.example .llm.env]
D2 --> D3{Build Type?}
D3 -->|Pre-built| D4[IMAGE=latest docker compose up]
D3 -->|Local Build| D5[docker compose up --build]
D3 -->|All Features| D6[INSTALL_TYPE=all docker compose up]
E --> E1[docker buildx build]
E1 --> E2{Architecture?}
E2 -->|Single| E3[--platform linux/amd64]
E2 -->|Multi| E4[--platform linux/amd64,linux/arm64]
E3 --> E5[Build complete]
E4 --> E5
F --> F1[Production configuration]
F1 --> F2[Custom config.yml]
F2 --> F3[Resource limits]
F3 --> F4[Health monitoring]
F4 --> F5[Production ready]
C5 --> G[Service running on :11235]
C6 --> G
D4 --> G
D5 --> G
D6 --> G
E5 --> H[docker run custom image]
H --> G
F5 --> I[Production deployment]
G --> J[Access playground at /playground]
G --> K[Health check at /health]
I --> L[Production monitoring]
style A fill:#e1f5fe
style G fill:#c8e6c9
style I fill:#c8e6c9
style J fill:#fff3e0
style K fill:#fff3e0
style L fill:#e8f5e8
```
### Docker Container Architecture
```mermaid
graph TB
subgraph "Host Environment"
A[Docker Engine] --> B[Crawl4AI Container]
C[.llm.env] --> B
D[Custom config.yml] --> B
E[Port 11235] --> B
F[Shared Memory 1GB+] --> B
end
subgraph "Container Services"
B --> G[FastAPI Server :8020]
B --> H[Gunicorn WSGI]
B --> I[Supervisord Process Manager]
B --> J[Redis Cache :6379]
G --> K[REST API Endpoints]
G --> L[WebSocket Connections]
G --> M[MCP Protocol]
H --> N[Worker Processes]
I --> O[Service Monitoring]
J --> P[Request Caching]
end
subgraph "Browser Management"
B --> Q[Playwright Framework]
Q --> R[Chromium Browser]
Q --> S[Firefox Browser]
Q --> T[WebKit Browser]
R --> U[Browser Pool]
S --> U
T --> U
U --> V[Page Sessions]
U --> W[Context Management]
end
subgraph "External Services"
X[OpenAI API] -.-> K
Y[Anthropic Claude] -.-> K
Z[Local Ollama] -.-> K
AA[Groq API] -.-> K
BB[Google Gemini] -.-> K
end
subgraph "Client Interactions"
CC[Python SDK] --> K
DD[REST API Calls] --> K
EE[MCP Clients] --> M
FF[Web Browser] --> G
GG[Monitoring Tools] --> K
end
style B fill:#e3f2fd
style G fill:#f3e5f5
style Q fill:#e8f5e8
style K fill:#fff3e0
```
### API Endpoints Architecture
```mermaid
graph LR
subgraph "Core Endpoints"
A[/crawl] --> A1[Single URL crawl]
A2[/crawl/stream] --> A3[Streaming multi-URL]
A4[/crawl/job] --> A5[Async job submission]
A6[/crawl/job/{id}] --> A7[Job status check]
end
subgraph "Specialized Endpoints"
B[/html] --> B1[Preprocessed HTML]
B2[/screenshot] --> B3[PNG capture]
B4[/pdf] --> B5[PDF generation]
B6[/execute_js] --> B7[JavaScript execution]
B8[/md] --> B9[Markdown extraction]
end
subgraph "Utility Endpoints"
C[/health] --> C1[Service status]
C2[/metrics] --> C3[Prometheus metrics]
C4[/schema] --> C5[API documentation]
C6[/playground] --> C7[Interactive testing]
end
subgraph "LLM Integration"
D[/llm/{url}] --> D1[Q&A over URL]
D2[/ask] --> D3[Library context search]
D4[/config/dump] --> D5[Config validation]
end
subgraph "MCP Protocol"
E[/mcp/sse] --> E1[Server-Sent Events]
E2[/mcp/ws] --> E3[WebSocket connection]
E4[/mcp/schema] --> E5[MCP tool definitions]
end
style A fill:#e3f2fd
style B fill:#f3e5f5
style C fill:#e8f5e8
style D fill:#fff3e0
style E fill:#fce4ec
```
### Request Processing Flow
```mermaid
sequenceDiagram
participant Client
participant FastAPI
participant RequestValidator
participant BrowserPool
participant Playwright
participant ExtractionEngine
participant LLMProvider
Client->>FastAPI: POST /crawl with config
FastAPI->>RequestValidator: Validate JSON structure
alt Valid Request
RequestValidator-->>FastAPI: ✓ Validated
FastAPI->>BrowserPool: Request browser instance
BrowserPool->>Playwright: Launch browser/reuse session
Playwright-->>BrowserPool: Browser ready
BrowserPool-->>FastAPI: Browser allocated
FastAPI->>Playwright: Navigate to URL
Playwright->>Playwright: Execute JS, wait conditions
Playwright-->>FastAPI: Page content ready
FastAPI->>ExtractionEngine: Process content
alt LLM Extraction
ExtractionEngine->>LLMProvider: Send content + schema
LLMProvider-->>ExtractionEngine: Structured data
else CSS Extraction
ExtractionEngine->>ExtractionEngine: Apply CSS selectors
end
ExtractionEngine-->>FastAPI: Extraction complete
FastAPI->>BrowserPool: Release browser
FastAPI-->>Client: CrawlResult response
else Invalid Request
RequestValidator-->>FastAPI: ✗ Validation error
FastAPI-->>Client: 400 Bad Request
end
```
### Configuration Management Flow
```mermaid
stateDiagram-v2
[*] --> ConfigLoading
ConfigLoading --> DefaultConfig: Load default config.yml
ConfigLoading --> CustomConfig: Custom config mounted
ConfigLoading --> EnvOverrides: Environment variables
DefaultConfig --> ConfigMerging
CustomConfig --> ConfigMerging
EnvOverrides --> ConfigMerging
ConfigMerging --> ConfigValidation
ConfigValidation --> Valid: Schema validation passes
ConfigValidation --> Invalid: Validation errors
Invalid --> ConfigError: Log errors and exit
ConfigError --> [*]
Valid --> ServiceInitialization
ServiceInitialization --> FastAPISetup
ServiceInitialization --> BrowserPoolInit
ServiceInitialization --> CacheSetup
FastAPISetup --> Running
BrowserPoolInit --> Running
CacheSetup --> Running
Running --> ConfigReload: Config change detected
ConfigReload --> ConfigValidation
Running --> [*]: Service shutdown
note right of ConfigMerging : Priority: ENV > Custom > Default
note right of ServiceInitialization : All services must initialize successfully
```
### Multi-Architecture Build Process
```mermaid
flowchart TD
A[Developer Push] --> B[GitHub Repository]
B --> C[Docker Buildx]
C --> D{Build Strategy}
D -->|Multi-arch| E[Parallel Builds]
D -->|Single-arch| F[Platform-specific Build]
E --> G[AMD64 Build]
E --> H[ARM64 Build]
F --> I[Target Platform Build]
subgraph "AMD64 Build Process"
G --> G1[Ubuntu base image]
G1 --> G2[Python 3.11 install]
G2 --> G3[System dependencies]
G3 --> G4[Crawl4AI installation]
G4 --> G5[Playwright setup]
G5 --> G6[FastAPI configuration]
G6 --> G7[AMD64 image ready]
end
subgraph "ARM64 Build Process"
H --> H1[Ubuntu ARM64 base]
H1 --> H2[Python 3.11 install]
H2 --> H3[ARM-specific deps]
H3 --> H4[Crawl4AI installation]
H4 --> H5[Playwright setup]
H5 --> H6[FastAPI configuration]
H6 --> H7[ARM64 image ready]
end
subgraph "Single Architecture"
I --> I1[Base image selection]
I1 --> I2[Platform dependencies]
I2 --> I3[Application setup]
I3 --> I4[Platform image ready]
end
G7 --> J[Multi-arch Manifest]
H7 --> J
I4 --> K[Platform Image]
J --> L[Docker Hub Registry]
K --> L
L --> M[Pull Request Auto-selects Architecture]
style A fill:#e1f5fe
style J fill:#c8e6c9
style K fill:#c8e6c9
style L fill:#f3e5f5
style M fill:#e8f5e8
```
### MCP Integration Architecture
```mermaid
graph TB
subgraph "MCP Client Applications"
A[Claude Code] --> B[MCP Protocol]
C[Cursor IDE] --> B
D[Windsurf] --> B
E[Custom MCP Client] --> B
end
subgraph "Crawl4AI MCP Server"
B --> F[MCP Endpoint Router]
F --> G[SSE Transport /mcp/sse]
F --> H[WebSocket Transport /mcp/ws]
F --> I[Schema Endpoint /mcp/schema]
G --> J[MCP Tool Handler]
H --> J
J --> K[Tool: md]
J --> L[Tool: html]
J --> M[Tool: screenshot]
J --> N[Tool: pdf]
J --> O[Tool: execute_js]
J --> P[Tool: crawl]
J --> Q[Tool: ask]
end
subgraph "Crawl4AI Core Services"
K --> R[Markdown Generator]
L --> S[HTML Preprocessor]
M --> T[Screenshot Service]
N --> U[PDF Generator]
O --> V[JavaScript Executor]
P --> W[Batch Crawler]
Q --> X[Context Search]
R --> Y[Browser Pool]
S --> Y
T --> Y
U --> Y
V --> Y
W --> Y
X --> Z[Knowledge Base]
end
subgraph "External Resources"
Y --> AA[Playwright Browsers]
Z --> BB[Library Documentation]
Z --> CC[Code Examples]
AA --> DD[Web Pages]
end
style B fill:#e3f2fd
style J fill:#f3e5f5
style Y fill:#e8f5e8
style Z fill:#fff3e0
```
### API Request/Response Flow Patterns
```mermaid
sequenceDiagram
participant Client
participant LoadBalancer
participant FastAPI
participant ConfigValidator
participant BrowserManager
participant CrawlEngine
participant ResponseBuilder
Note over Client,ResponseBuilder: Basic Crawl Request
Client->>LoadBalancer: POST /crawl
LoadBalancer->>FastAPI: Route request
FastAPI->>ConfigValidator: Validate browser_config
ConfigValidator-->>FastAPI: ✓ Valid BrowserConfig
FastAPI->>ConfigValidator: Validate crawler_config
ConfigValidator-->>FastAPI: ✓ Valid CrawlerRunConfig
FastAPI->>BrowserManager: Allocate browser
BrowserManager-->>FastAPI: Browser instance
FastAPI->>CrawlEngine: Execute crawl
Note over CrawlEngine: Page processing
CrawlEngine->>CrawlEngine: Navigate & wait
CrawlEngine->>CrawlEngine: Extract content
CrawlEngine->>CrawlEngine: Apply strategies
CrawlEngine-->>FastAPI: CrawlResult
FastAPI->>ResponseBuilder: Format response
ResponseBuilder-->>FastAPI: JSON response
FastAPI->>BrowserManager: Release browser
FastAPI-->>LoadBalancer: Response ready
LoadBalancer-->>Client: 200 OK + CrawlResult
Note over Client,ResponseBuilder: Streaming Request
Client->>FastAPI: POST /crawl/stream
FastAPI-->>Client: 200 OK (stream start)
loop For each URL
FastAPI->>CrawlEngine: Process URL
CrawlEngine-->>FastAPI: Result ready
FastAPI-->>Client: NDJSON line
end
FastAPI-->>Client: Stream completed
```
### Configuration Validation Workflow
```mermaid
flowchart TD
A[Client Request] --> B[JSON Payload]
B --> C{Pre-validation}
C -->|✓ Valid JSON| D[Extract Configurations]
C -->|✗ Invalid JSON| E[Return 400 Bad Request]
D --> F[BrowserConfig Validation]
D --> G[CrawlerRunConfig Validation]
F --> H{BrowserConfig Valid?}
G --> I{CrawlerRunConfig Valid?}
H -->|✓ Valid| J[Browser Setup]
H -->|✗ Invalid| K[Log Browser Config Errors]
I -->|✓ Valid| L[Crawler Setup]
I -->|✗ Invalid| M[Log Crawler Config Errors]
K --> N[Collect All Errors]
M --> N
N --> O[Return 422 Validation Error]
J --> P{Both Configs Valid?}
L --> P
P -->|✓ Yes| Q[Proceed to Crawling]
P -->|✗ No| O
Q --> R[Execute Crawl Pipeline]
R --> S[Return CrawlResult]
E --> T[Client Error Response]
O --> T
S --> U[Client Success Response]
style A fill:#e1f5fe
style Q fill:#c8e6c9
style S fill:#c8e6c9
style U fill:#c8e6c9
style E fill:#ffcdd2
style O fill:#ffcdd2
style T fill:#ffcdd2
```
### Production Deployment Architecture
```mermaid
graph TB
subgraph "Load Balancer Layer"
A[NGINX/HAProxy] --> B[Health Check]
A --> C[Request Routing]
A --> D[SSL Termination]
end
subgraph "Application Layer"
C --> E[Crawl4AI Instance 1]
C --> F[Crawl4AI Instance 2]
C --> G[Crawl4AI Instance N]
E --> H[FastAPI Server]
F --> I[FastAPI Server]
G --> J[FastAPI Server]
H --> K[Browser Pool 1]
I --> L[Browser Pool 2]
J --> M[Browser Pool N]
end
subgraph "Shared Services"
N[Redis Cluster] --> E
N --> F
N --> G
O[Monitoring Stack] --> P[Prometheus]
O --> Q[Grafana]
O --> R[AlertManager]
P --> E
P --> F
P --> G
end
subgraph "External Dependencies"
S[OpenAI API] -.-> H
T[Anthropic API] -.-> I
U[Local LLM Cluster] -.-> J
end
subgraph "Persistent Storage"
V[Configuration Volume] --> E
V --> F
V --> G
W[Cache Volume] --> N
X[Logs Volume] --> O
end
style A fill:#e3f2fd
style E fill:#f3e5f5
style F fill:#f3e5f5
style G fill:#f3e5f5
style N fill:#e8f5e8
style O fill:#fff3e0
```
### Docker Resource Management
```mermaid
graph TD
subgraph "Resource Allocation"
A[Host Resources] --> B[CPU Cores]
A --> C[Memory GB]
A --> D[Disk Space]
A --> E[Network Bandwidth]
B --> F[Container Limits]
C --> F
D --> F
E --> F
end
subgraph "Container Configuration"
F --> G[--cpus=4]
F --> H[--memory=8g]
F --> I[--shm-size=2g]
F --> J[Volume Mounts]
G --> K[Browser Processes]
H --> L[Browser Memory]
I --> M[Shared Memory for Browsers]
J --> N[Config & Cache Storage]
end
subgraph "Monitoring & Scaling"
O[Resource Monitor] --> P[CPU Usage %]
O --> Q[Memory Usage %]
O --> R[Request Queue Length]
P --> S{CPU > 80%?}
Q --> T{Memory > 90%?}
R --> U{Queue > 100?}
S -->|Yes| V[Scale Up]
T -->|Yes| V
U -->|Yes| V
V --> W[Add Container Instance]
W --> X[Update Load Balancer]
end
subgraph "Performance Optimization"
Y[Browser Pool Tuning] --> Z[Max Pages: 40]
Y --> AA[Idle TTL: 30min]
Y --> BB[Concurrency Limits]
Z --> CC[Memory Efficiency]
AA --> DD[Resource Cleanup]
BB --> EE[Throughput Control]
end
style A fill:#e1f5fe
style F fill:#f3e5f5
style O fill:#e8f5e8
style Y fill:#fff3e0
```
**📖 Learn more:** [Docker Deployment Guide](https://docs.crawl4ai.com/core/docker-deployment/), [API Reference](https://docs.crawl4ai.com/api/), [MCP Integration](https://docs.crawl4ai.com/core/docker-deployment/#mcp-model-context-protocol-support), [Production Configuration](https://docs.crawl4ai.com/core/docker-deployment/#production-deployment)
---
## CLI Workflows and Profile Management
Visual representations of command-line interface operations, browser profile management, and identity-based crawling workflows.
### CLI Command Flow Architecture
```mermaid
flowchart TD
A[crwl command] --> B{Command Type?}
B -->|URL Crawling| C[Parse URL & Options]
B -->|Profile Management| D[profiles subcommand]
B -->|CDP Browser| E[cdp subcommand]
B -->|Browser Control| F[browser subcommand]
B -->|Configuration| G[config subcommand]
C --> C1{Output Format?}
C1 -->|Default| C2[HTML/Markdown]
C1 -->|JSON| C3[Structured Data]
C1 -->|markdown| C4[Clean Markdown]
C1 -->|markdown-fit| C5[Filtered Content]
C --> C6{Authentication?}
C6 -->|Profile Specified| C7[Load Browser Profile]
C6 -->|No Profile| C8[Anonymous Session]
C7 --> C9[Launch with User Data]
C8 --> C10[Launch Clean Browser]
C9 --> C11[Execute Crawl]
C10 --> C11
C11 --> C12{Success?}
C12 -->|Yes| C13[Return Results]
C12 -->|No| C14[Error Handling]
D --> D1[Interactive Profile Menu]
D1 --> D2{Menu Choice?}
D2 -->|Create| D3[Open Browser for Setup]
D2 -->|List| D4[Show Existing Profiles]
D2 -->|Delete| D5[Remove Profile]
D2 -->|Use| D6[Crawl with Profile]
E --> E1[Launch CDP Browser]
E1 --> E2[Remote Debugging Active]
F --> F1{Browser Action?}
F1 -->|start| F2[Start Builtin Browser]
F1 -->|stop| F3[Stop Builtin Browser]
F1 -->|status| F4[Check Browser Status]
F1 -->|view| F5[Open Browser Window]
G --> G1{Config Action?}
G1 -->|list| G2[Show All Settings]
G1 -->|set| G3[Update Setting]
G1 -->|get| G4[Read Setting]
style A fill:#e1f5fe
style C13 fill:#c8e6c9
style C14 fill:#ffcdd2
style D3 fill:#fff3e0
style E2 fill:#f3e5f5
```
### Profile Management Workflow
```mermaid
sequenceDiagram
participant User
participant CLI
participant ProfileManager
participant Browser
participant FileSystem
User->>CLI: crwl profiles
CLI->>ProfileManager: Initialize profile manager
ProfileManager->>FileSystem: Scan for existing profiles
FileSystem-->>ProfileManager: Profile list
ProfileManager-->>CLI: Show interactive menu
CLI-->>User: Display options
Note over User: User selects "Create new profile"
User->>CLI: Create profile "linkedin-auth"
CLI->>ProfileManager: create_profile("linkedin-auth")
ProfileManager->>FileSystem: Create profile directory
ProfileManager->>Browser: Launch with new user data dir
Browser-->>User: Opens browser window
Note over User: User manually logs in to LinkedIn
User->>Browser: Navigate and authenticate
Browser->>FileSystem: Save cookies, session data
User->>CLI: Press 'q' to save profile
CLI->>ProfileManager: finalize_profile()
ProfileManager->>FileSystem: Lock profile settings
ProfileManager-->>CLI: Profile saved
CLI-->>User: Profile "linkedin-auth" created
Note over User: Later usage
User->>CLI: crwl https://linkedin.com/feed -p linkedin-auth
CLI->>ProfileManager: load_profile("linkedin-auth")
ProfileManager->>FileSystem: Read profile data
FileSystem-->>ProfileManager: User data directory
ProfileManager-->>CLI: Profile configuration
CLI->>Browser: Launch with existing profile
Browser-->>CLI: Authenticated session ready
CLI->>Browser: Navigate to target URL
Browser-->>CLI: Crawl results with auth context
CLI-->>User: Authenticated content
```
### Browser Management State Machine
```mermaid
stateDiagram-v2
[*] --> Stopped: Initial state
Stopped --> Starting: crwl browser start
Starting --> Running: Browser launched
Running --> Viewing: crwl browser view
Viewing --> Running: Close window
Running --> Stopping: crwl browser stop
Stopping --> Stopped: Cleanup complete
Running --> Restarting: crwl browser restart
Restarting --> Running: New browser instance
Stopped --> CDP_Mode: crwl cdp
CDP_Mode --> CDP_Running: Remote debugging active
CDP_Running --> CDP_Mode: Manual close
CDP_Mode --> Stopped: Exit CDP
Running --> StatusCheck: crwl browser status
StatusCheck --> Running: Return status
note right of Running : Port 9222 active\nBuiltin browser available
note right of CDP_Running : Remote debugging\nManual control enabled
note right of Viewing : Visual browser window\nDirect interaction
```
### Authentication Workflow for Protected Sites
```mermaid
flowchart TD
A[Protected Site Access Needed] --> B[Create Profile Strategy]
B --> C{Existing Profile?}
C -->|Yes| D[Test Profile Validity]
C -->|No| E[Create New Profile]
D --> D1{Profile Valid?}
D1 -->|Yes| F[Use Existing Profile]
D1 -->|No| E
E --> E1[crwl profiles]
E1 --> E2[Select Create New Profile]
E2 --> E3[Enter Profile Name]
E3 --> E4[Browser Opens for Auth]
E4 --> E5{Authentication Method?}
E5 -->|Login Form| E6[Fill Username/Password]
E5 -->|OAuth| E7[OAuth Flow]
E5 -->|2FA| E8[Handle 2FA]
E5 -->|Session Cookie| E9[Import Cookies]
E6 --> E10[Manual Login Process]
E7 --> E10
E8 --> E10
E9 --> E10
E10 --> E11[Verify Authentication]
E11 --> E12{Auth Successful?}
E12 -->|Yes| E13[Save Profile - Press q]
E12 -->|No| E10
E13 --> F
F --> G[Execute Authenticated Crawl]
G --> H[crwl URL -p profile-name]
H --> I[Load Profile Data]
I --> J[Launch Browser with Auth]
J --> K[Navigate to Protected Content]
K --> L[Extract Authenticated Data]
L --> M[Return Results]
style E4 fill:#fff3e0
style E10 fill:#e3f2fd
style F fill:#e8f5e8
style M fill:#c8e6c9
```
### CDP Browser Architecture
```mermaid
graph TB
subgraph "CLI Layer"
A[crwl cdp command] --> B[CDP Manager]
B --> C[Port Configuration]
B --> D[Profile Selection]
end
subgraph "Browser Process"
E[Chromium/Firefox] --> F[Remote Debugging]
F --> G[WebSocket Endpoint]
G --> H[ws://localhost:9222]
end
subgraph "Client Connections"
I[Manual Browser Control] --> H
J[DevTools Interface] --> H
K[External Automation] --> H
L[Crawl4AI Crawler] --> H
end
subgraph "Profile Data"
M[User Data Directory] --> E
N[Cookies & Sessions] --> M
O[Extensions] --> M
P[Browser State] --> M
end
A --> E
C --> H
D --> M
style H fill:#e3f2fd
style E fill:#f3e5f5
style M fill:#e8f5e8
```
### Configuration Management Hierarchy
```mermaid
graph TD
subgraph "Global Configuration"
A[~/.crawl4ai/config.yml] --> B[Default Settings]
B --> C[LLM Providers]
B --> D[Browser Defaults]
B --> E[Output Preferences]
end
subgraph "Profile Configuration"
F[Profile Directory] --> G[Browser State]
F --> H[Authentication Data]
F --> I[Site-Specific Settings]
end
subgraph "Command-Line Overrides"
J[-b browser_config] --> K[Runtime Browser Settings]
L[-c crawler_config] --> M[Runtime Crawler Settings]
N[-o output_format] --> O[Runtime Output Format]
end
subgraph "Configuration Files"
P[browser.yml] --> Q[Browser Config Template]
R[crawler.yml] --> S[Crawler Config Template]
T[extract.yml] --> U[Extraction Config]
end
subgraph "Resolution Order"
V[Command Line Args] --> W[Config Files]
W --> X[Profile Settings]
X --> Y[Global Defaults]
end
J --> V
L --> V
N --> V
P --> W
R --> W
T --> W
F --> X
A --> Y
style V fill:#ffcdd2
style W fill:#fff3e0
style X fill:#e3f2fd
style Y fill:#e8f5e8
```
### Identity-Based Crawling Decision Tree
```mermaid
flowchart TD
A[Target Website Assessment] --> B{Authentication Required?}
B -->|No| C[Standard Anonymous Crawl]
B -->|Yes| D{Authentication Type?}
D -->|Login Form| E[Create Login Profile]
D -->|OAuth/SSO| F[Create OAuth Profile]
D -->|API Key/Token| G[Use Headers/Config]
D -->|Session Cookies| H[Import Cookie Profile]
E --> E1[crwl profiles → Manual login]
F --> F1[crwl profiles → OAuth flow]
G --> G1[Configure headers in crawler config]
H --> H1[Import cookies to profile]
E1 --> I[Test Authentication]
F1 --> I
G1 --> I
H1 --> I
I --> J{Auth Test Success?}
J -->|Yes| K[Production Crawl Setup]
J -->|No| L[Debug Authentication]
L --> L1{Common Issues?}
L1 -->|Rate Limiting| L2[Add delays/user simulation]
L1 -->|Bot Detection| L3[Enable stealth mode]
L1 -->|Session Expired| L4[Refresh authentication]
L1 -->|CAPTCHA| L5[Manual intervention needed]
L2 --> M[Retry with Adjustments]
L3 --> M
L4 --> E1
L5 --> N[Semi-automated approach]
M --> I
N --> O[Manual auth + automated crawl]
K --> P[Automated Authenticated Crawling]
O --> P
C --> P
P --> Q[Monitor & Maintain Profiles]
style I fill:#fff3e0
style K fill:#e8f5e8
style P fill:#c8e6c9
style L fill:#ffcdd2
style N fill:#f3e5f5
```
### CLI Usage Patterns and Best Practices
```mermaid
timeline
title CLI Workflow Evolution
section Setup Phase
Installation : pip install crawl4ai
: crawl4ai-setup
Basic Test : crwl https://example.com
Config Setup : crwl config set defaults
section Profile Creation
Site Analysis : Identify auth requirements
Profile Creation : crwl profiles
Manual Login : Authenticate in browser
Profile Save : Press 'q' to save
section Development Phase
Test Crawls : crwl URL -p profile -v
Config Tuning : Adjust browser/crawler settings
Output Testing : Try different output formats
Error Handling : Debug authentication issues
section Production Phase
Automated Crawls : crwl URL -p profile -o json
Batch Processing : Multiple URLs with same profile
Monitoring : Check profile validity
Maintenance : Update profiles as needed
```
### Multi-Profile Management Strategy
```mermaid
graph LR
subgraph "Profile Categories"
A[Social Media Profiles]
B[Work/Enterprise Profiles]
C[E-commerce Profiles]
D[Research Profiles]
end
subgraph "Social Media"
A --> A1[linkedin-personal]
A --> A2[twitter-monitor]
A --> A3[facebook-research]
A --> A4[instagram-brand]
end
subgraph "Enterprise"
B --> B1[company-intranet]
B --> B2[github-enterprise]
B --> B3[confluence-docs]
B --> B4[jira-tickets]
end
subgraph "E-commerce"
C --> C1[amazon-seller]
C --> C2[shopify-admin]
C --> C3[ebay-monitor]
C --> C4[marketplace-competitor]
end
subgraph "Research"
D --> D1[academic-journals]
D --> D2[data-platforms]
D --> D3[survey-tools]
D --> D4[government-portals]
end
subgraph "Usage Patterns"
E[Daily Monitoring] --> A2
E --> B1
F[Weekly Reports] --> C3
F --> D2
G[On-Demand Research] --> D1
G --> D4
H[Competitive Analysis] --> C4
H --> A4
end
style A1 fill:#e3f2fd
style B1 fill:#f3e5f5
style C1 fill:#e8f5e8
style D1 fill:#fff3e0
```
**📖 Learn more:** [CLI Reference](https://docs.crawl4ai.com/core/cli/), [Identity-Based Crawling](https://docs.crawl4ai.com/advanced/identity-based-crawling/), [Profile Management](https://docs.crawl4ai.com/advanced/session-management/), [Authentication Strategies](https://docs.crawl4ai.com/advanced/hooks-auth/)
---
## HTTP Crawler Strategy Workflows
Visual representations of HTTP-based crawling architecture, request flows, and performance characteristics compared to browser-based strategies.
### HTTP vs Browser Strategy Decision Tree
```mermaid
flowchart TD
A[Content Crawling Need] --> B{Content Type Analysis}
B -->|Static HTML| C{JavaScript Required?}
B -->|Dynamic SPA| D[Browser Strategy Required]
B -->|API Endpoints| E[HTTP Strategy Optimal]
B -->|Mixed Content| F{Primary Content Source?}
C -->|No JS Needed| G[HTTP Strategy Recommended]
C -->|JS Required| H[Browser Strategy Required]
C -->|Unknown| I{Performance Priority?}
I -->|Speed Critical| J[Try HTTP First]
I -->|Accuracy Critical| K[Use Browser Strategy]
F -->|Mostly Static| G
F -->|Mostly Dynamic| D
G --> L{Resource Constraints?}
L -->|Memory Limited| M[HTTP Strategy - Lightweight]
L -->|CPU Limited| N[HTTP Strategy - No Browser]
L -->|Network Limited| O[HTTP Strategy - Efficient]
L -->|No Constraints| P[Either Strategy Works]
J --> Q[Test HTTP Results]
Q --> R{Content Complete?}
R -->|Yes| S[Continue with HTTP]
R -->|No| T[Switch to Browser Strategy]
D --> U[Browser Strategy Features]
H --> U
K --> U
T --> U
U --> V[JavaScript Execution]
U --> W[Screenshots/PDFs]
U --> X[Complex Interactions]
U --> Y[Session Management]
M --> Z[HTTP Strategy Benefits]
N --> Z
O --> Z
S --> Z
Z --> AA[10x Faster Processing]
Z --> BB[Lower Memory Usage]
Z --> CC[Higher Concurrency]
Z --> DD[Simpler Deployment]
style G fill:#c8e6c9
style M fill:#c8e6c9
style N fill:#c8e6c9
style O fill:#c8e6c9
style S fill:#c8e6c9
style D fill:#e3f2fd
style H fill:#e3f2fd
style K fill:#e3f2fd
style T fill:#e3f2fd
style U fill:#e3f2fd
```
### HTTP Request Lifecycle Sequence
```mermaid
sequenceDiagram
participant Client
participant HTTPStrategy as HTTP Strategy
participant Session as HTTP Session
participant Server as Target Server
participant Processor as Content Processor
Client->>HTTPStrategy: crawl(url, config)
HTTPStrategy->>HTTPStrategy: validate_url()
alt URL Type Check
HTTPStrategy->>HTTPStrategy: handle_file_url()
Note over HTTPStrategy: file:// URLs
else
HTTPStrategy->>HTTPStrategy: handle_raw_content()
Note over HTTPStrategy: raw:// content
else
HTTPStrategy->>Session: prepare_request()
Session->>Session: apply_config()
Session->>Session: set_headers()
Session->>Session: setup_auth()
Session->>Server: HTTP Request
Note over Session,Server: GET/POST/PUT with headers
alt Success Response
Server-->>Session: HTTP 200 + Content
Session-->>HTTPStrategy: response_data
else Redirect Response
Server-->>Session: HTTP 3xx + Location
Session->>Server: Follow redirect
Server-->>Session: HTTP 200 + Content
Session-->>HTTPStrategy: final_response
else Error Response
Server-->>Session: HTTP 4xx/5xx
Session-->>HTTPStrategy: error_response
end
end
HTTPStrategy->>Processor: process_content()
Processor->>Processor: clean_html()
Processor->>Processor: extract_metadata()
Processor->>Processor: generate_markdown()
Processor-->>HTTPStrategy: processed_result
HTTPStrategy-->>Client: CrawlResult
Note over Client,Processor: Fast, lightweight processing
Note over HTTPStrategy: No browser overhead
```
### HTTP Strategy Architecture
```mermaid
graph TB
subgraph "HTTP Crawler Strategy"
A[AsyncHTTPCrawlerStrategy] --> B[Session Manager]
A --> C[Request Builder]
A --> D[Response Handler]
A --> E[Error Manager]
B --> B1[Connection Pool]
B --> B2[DNS Cache]
B --> B3[SSL Context]
C --> C1[Headers Builder]
C --> C2[Auth Handler]
C --> C3[Payload Encoder]
D --> D1[Content Decoder]
D --> D2[Redirect Handler]
D --> D3[Status Validator]
E --> E1[Retry Logic]
E --> E2[Timeout Handler]
E --> E3[Exception Mapper]
end
subgraph "Content Processing"
F[Raw HTML] --> G[HTML Cleaner]
G --> H[Markdown Generator]
H --> I[Link Extractor]
I --> J[Media Extractor]
J --> K[Metadata Parser]
end
subgraph "External Resources"
L[Target Websites]
M[Local Files]
N[Raw Content]
end
subgraph "Output"
O[CrawlResult]
O --> O1[HTML Content]
O --> O2[Markdown Text]
O --> O3[Extracted Links]
O --> O4[Media References]
O --> O5[Status Information]
end
A --> F
L --> A
M --> A
N --> A
K --> O
style A fill:#e3f2fd
style B fill:#f3e5f5
style F fill:#e8f5e8
style O fill:#fff3e0
```
### Performance Comparison Flow
```mermaid
graph LR
subgraph "HTTP Strategy Performance"
A1[Request Start] --> A2[DNS Lookup: 50ms]
A2 --> A3[TCP Connect: 100ms]
A3 --> A4[HTTP Request: 200ms]
A4 --> A5[Content Download: 300ms]
A5 --> A6[Processing: 50ms]
A6 --> A7[Total: ~700ms]
end
subgraph "Browser Strategy Performance"
B1[Request Start] --> B2[Browser Launch: 2000ms]
B2 --> B3[Page Navigation: 1000ms]
B3 --> B4[JS Execution: 500ms]
B4 --> B5[Content Rendering: 300ms]
B5 --> B6[Processing: 100ms]
B6 --> B7[Total: ~3900ms]
end
subgraph "Resource Usage"
C1[HTTP Memory: ~50MB]
C2[Browser Memory: ~500MB]
C3[HTTP CPU: Low]
C4[Browser CPU: High]
C5[HTTP Concurrency: 100+]
C6[Browser Concurrency: 10-20]
end
A7 --> D[5.5x Faster]
B7 --> D
C1 --> E[10x Less Memory]
C2 --> E
C5 --> F[5x More Concurrent]
C6 --> F
style A7 fill:#c8e6c9
style B7 fill:#ffcdd2
style C1 fill:#c8e6c9
style C2 fill:#ffcdd2
style C5 fill:#c8e6c9
style C6 fill:#ffcdd2
```
### HTTP Request Types and Configuration
```mermaid
stateDiagram-v2
[*] --> HTTPConfigSetup
HTTPConfigSetup --> MethodSelection
MethodSelection --> GET: Simple data retrieval
MethodSelection --> POST: Form submission
MethodSelection --> PUT: Data upload
MethodSelection --> DELETE: Resource removal
GET --> HeaderSetup: Set Accept headers
POST --> PayloadSetup: JSON or form data
PUT --> PayloadSetup: File or data upload
DELETE --> AuthSetup: Authentication required
PayloadSetup --> JSONPayload: application/json
PayloadSetup --> FormPayload: form-data
PayloadSetup --> RawPayload: custom content
JSONPayload --> HeaderSetup
FormPayload --> HeaderSetup
RawPayload --> HeaderSetup
HeaderSetup --> AuthSetup
AuthSetup --> SSLSetup
SSLSetup --> RedirectSetup
RedirectSetup --> RequestExecution
RequestExecution --> [*]: Request complete
note right of GET : Default method for most crawling
note right of POST : API interactions, form submissions
note right of JSONPayload : Structured data transmission
note right of HeaderSetup : User-Agent, Accept, Custom headers
```
### Error Handling and Retry Workflow
```mermaid
flowchart TD
A[HTTP Request] --> B{Response Received?}
B -->|No| C[Connection Error]
B -->|Yes| D{Status Code Check}
C --> C1{Timeout Error?}
C1 -->|Yes| C2[ConnectionTimeoutError]
C1 -->|No| C3[Network Error]
D -->|2xx| E[Success Response]
D -->|3xx| F[Redirect Response]
D -->|4xx| G[Client Error]
D -->|5xx| H[Server Error]
F --> F1{Follow Redirects?}
F1 -->|Yes| F2[Follow Redirect]
F1 -->|No| F3[Return Redirect Response]
F2 --> A
G --> G1{Retry on 4xx?}
G1 -->|No| G2[HTTPStatusError]
G1 -->|Yes| I[Check Retry Count]
H --> H1{Retry on 5xx?}
H1 -->|Yes| I
H1 -->|No| H2[HTTPStatusError]
C2 --> I
C3 --> I
I --> J{Retries < Max?}
J -->|No| K[Final Error]
J -->|Yes| L[Calculate Backoff]
L --> M[Wait Backoff Time]
M --> N[Increment Retry Count]
N --> A
E --> O[Process Content]
F3 --> O
O --> P[Return CrawlResult]
G2 --> Q[Error CrawlResult]
H2 --> Q
K --> Q
style E fill:#c8e6c9
style P fill:#c8e6c9
style G2 fill:#ffcdd2
style H2 fill:#ffcdd2
style K fill:#ffcdd2
style Q fill:#ffcdd2
```
### Batch Processing Architecture
```mermaid
sequenceDiagram
participant Client
participant BatchManager as Batch Manager
participant HTTPPool as Connection Pool
participant Workers as HTTP Workers
participant Targets as Target Servers
Client->>BatchManager: batch_crawl(urls)
BatchManager->>BatchManager: create_semaphore(max_concurrent)
loop For each URL batch
BatchManager->>HTTPPool: acquire_connection()
HTTPPool->>Workers: assign_worker()
par Concurrent Processing
Workers->>Targets: HTTP Request 1
Workers->>Targets: HTTP Request 2
Workers->>Targets: HTTP Request N
end
par Response Handling
Targets-->>Workers: Response 1
Targets-->>Workers: Response 2
Targets-->>Workers: Response N
end
Workers->>HTTPPool: return_connection()
HTTPPool->>BatchManager: batch_results()
end
BatchManager->>BatchManager: aggregate_results()
BatchManager-->>Client: final_results()
Note over Workers,Targets: 20-100 concurrent connections
Note over BatchManager: Memory-efficient processing
Note over HTTPPool: Connection reuse optimization
```
### Content Type Processing Pipeline
```mermaid
graph TD
A[HTTP Response] --> B{Content-Type Detection}
B -->|text/html| C[HTML Processing]
B -->|application/json| D[JSON Processing]
B -->|text/plain| E[Text Processing]
B -->|application/xml| F[XML Processing]
B -->|Other| G[Binary Processing]
C --> C1[Parse HTML Structure]
C1 --> C2[Extract Text Content]
C2 --> C3[Generate Markdown]
C3 --> C4[Extract Links/Media]
D --> D1[Parse JSON Structure]
D1 --> D2[Extract Data Fields]
D2 --> D3[Format as Readable Text]
E --> E1[Clean Text Content]
E1 --> E2[Basic Formatting]
F --> F1[Parse XML Structure]
F1 --> F2[Extract Text Nodes]
F2 --> F3[Convert to Markdown]
G --> G1[Save Binary Content]
G1 --> G2[Generate Metadata]
C4 --> H[Content Analysis]
D3 --> H
E2 --> H
F3 --> H
G2 --> H
H --> I[Link Extraction]
H --> J[Media Detection]
H --> K[Metadata Parsing]
I --> L[CrawlResult Assembly]
J --> L
K --> L
L --> M[Final Output]
style C fill:#e8f5e8
style H fill:#fff3e0
style L fill:#e3f2fd
style M fill:#c8e6c9
```
### Integration with Processing Strategies
```mermaid
graph LR
subgraph "HTTP Strategy Core"
A[HTTP Request] --> B[Raw Content]
B --> C[Content Decoder]
end
subgraph "Processing Pipeline"
C --> D[HTML Cleaner]
D --> E[Markdown Generator]
E --> F{Content Filter?}
F -->|Yes| G[Pruning Filter]
F -->|Yes| H[BM25 Filter]
F -->|No| I[Raw Markdown]
G --> J[Fit Markdown]
H --> J
end
subgraph "Extraction Strategies"
I --> K[CSS Extraction]
J --> K
I --> L[XPath Extraction]
J --> L
I --> M[LLM Extraction]
J --> M
end
subgraph "Output Generation"
K --> N[Structured JSON]
L --> N
M --> N
I --> O[Clean Markdown]
J --> P[Filtered Content]
N --> Q[Final CrawlResult]
O --> Q
P --> Q
end
style A fill:#e3f2fd
style C fill:#f3e5f5
style E fill:#e8f5e8
style Q fill:#c8e6c9
```
**📖 Learn more:** [HTTP vs Browser Strategies](https://docs.crawl4ai.com/core/browser-crawler-config/), [Performance Optimization](https://docs.crawl4ai.com/advanced/multi-url-crawling/), [Error Handling](https://docs.crawl4ai.com/api/async-webcrawler/)
---
## URL Seeding Workflows and Architecture
Visual representations of URL discovery strategies, filtering pipelines, and smart crawling workflows.
### URL Seeding vs Deep Crawling Strategy Comparison
```mermaid
graph TB
subgraph "Deep Crawling Approach"
A1[Start URL] --> A2[Load Page]
A2 --> A3[Extract Links]
A3 --> A4{More Links?}
A4 -->|Yes| A5[Queue Next Page]
A5 --> A2
A4 -->|No| A6[Complete]
A7[⏱️ Real-time Discovery]
A8[🐌 Sequential Processing]
A9[🔍 Limited by Page Structure]
A10[💾 High Memory Usage]
end
subgraph "URL Seeding Approach"
B1[Domain Input] --> B2[Query Sitemap]
B1 --> B3[Query Common Crawl]
B2 --> B4[Merge Results]
B3 --> B4
B4 --> B5[Apply Filters]
B5 --> B6[Score Relevance]
B6 --> B7[Rank Results]
B7 --> B8[Select Top URLs]
B9[⚡ Instant Discovery]
B10[🚀 Parallel Processing]
B11[🎯 Pattern-based Filtering]
B12[💡 Smart Relevance Scoring]
end
style A1 fill:#ffecb3
style B1 fill:#e8f5e8
style A6 fill:#ffcdd2
style B8 fill:#c8e6c9
```
### URL Discovery Data Flow
```mermaid
sequenceDiagram
participant User
participant Seeder as AsyncUrlSeeder
participant SM as Sitemap
participant CC as Common Crawl
participant Filter as URL Filter
participant Scorer as BM25 Scorer
User->>Seeder: urls("example.com", config)
par Parallel Data Sources
Seeder->>SM: Fetch sitemap.xml
SM-->>Seeder: 500 URLs
and
Seeder->>CC: Query Common Crawl
CC-->>Seeder: 2000 URLs
end
Seeder->>Seeder: Merge and deduplicate
Note over Seeder: 2200 unique URLs
Seeder->>Filter: Apply pattern filter
Filter-->>Seeder: 800 matching URLs
alt extract_head=True
loop For each URL
Seeder->>Seeder: Extract <head> metadata
end
Note over Seeder: Title, description, keywords
end
alt query provided
Seeder->>Scorer: Calculate relevance scores
Scorer-->>Seeder: Scored URLs
Seeder->>Seeder: Filter by score_threshold
Note over Seeder: 200 relevant URLs
end
Seeder->>Seeder: Sort by relevance
Seeder->>Seeder: Apply max_urls limit
Seeder-->>User: Top 100 URLs ready for crawling
```
### SeedingConfig Decision Tree
```mermaid
flowchart TD
A[SeedingConfig Setup] --> B{Data Source Strategy?}
B -->|Fast & Official| C[source="sitemap"]
B -->|Comprehensive| D[source="cc"]
B -->|Maximum Coverage| E[source="sitemap+cc"]
C --> F{Need Filtering?}
D --> F
E --> F
F -->|Yes| G[Set URL Pattern]
F -->|No| H[pattern="*"]
G --> I{Pattern Examples}
I --> I1[pattern="*/blog/*"]
I --> I2[pattern="*/docs/api/*"]
I --> I3[pattern="*.pdf"]
I --> I4[pattern="*/product/*"]
H --> J{Need Metadata?}
I1 --> J
I2 --> J
I3 --> J
I4 --> J
J -->|Yes| K[extract_head=True]
J -->|No| L[extract_head=False]
K --> M{Need Validation?}
L --> M
M -->|Yes| N[live_check=True]
M -->|No| O[live_check=False]
N --> P{Need Relevance Scoring?}
O --> P
P -->|Yes| Q[Set Query + BM25]
P -->|No| R[Skip Scoring]
Q --> S[query="search terms"]
S --> T[scoring_method="bm25"]
T --> U[score_threshold=0.3]
R --> V[Performance Tuning]
U --> V
V --> W[Set max_urls]
W --> X[Set concurrency]
X --> Y[Set hits_per_sec]
Y --> Z[Configuration Complete]
style A fill:#e3f2fd
style Z fill:#c8e6c9
style K fill:#fff3e0
style N fill:#fff3e0
style Q fill:#f3e5f5
```
### BM25 Relevance Scoring Pipeline
```mermaid
graph TB
subgraph "Text Corpus Preparation"
A1[URL Collection] --> A2[Extract Metadata]
A2 --> A3[Title + Description + Keywords]
A3 --> A4[Tokenize Text]
A4 --> A5[Remove Stop Words]
A5 --> A6[Create Document Corpus]
end
subgraph "BM25 Algorithm"
B1[Query Terms] --> B2[Term Frequency Calculation]
A6 --> B2
B2 --> B3[Inverse Document Frequency]
B3 --> B4[BM25 Score Calculation]
B4 --> B5[Score = Σ(IDF × TF × K1+1)/(TF + K1×(1-b+b×|d|/avgdl))]
end
subgraph "Scoring Results"
B5 --> C1[URL Relevance Scores]
C1 --> C2{Score ≥ Threshold?}
C2 -->|Yes| C3[Include in Results]
C2 -->|No| C4[Filter Out]
C3 --> C5[Sort by Score DESC]
C5 --> C6[Return Top URLs]
end
subgraph "Example Scores"
D1["python async tutorial" → 0.85]
D2["python documentation" → 0.72]
D3["javascript guide" → 0.23]
D4["contact us page" → 0.05]
end
style B5 fill:#e3f2fd
style C6 fill:#c8e6c9
style D1 fill:#c8e6c9
style D2 fill:#c8e6c9
style D3 fill:#ffecb3
style D4 fill:#ffcdd2
```
### Multi-Domain Discovery Architecture
```mermaid
graph TB
subgraph "Input Layer"
A1[Domain List]
A2[SeedingConfig]
A3[Query Terms]
end
subgraph "Discovery Engine"
B1[AsyncUrlSeeder]
B2[Parallel Workers]
B3[Rate Limiter]
B4[Memory Manager]
end
subgraph "Data Sources"
C1[Sitemap Fetcher]
C2[Common Crawl API]
C3[Live URL Checker]
C4[Metadata Extractor]
end
subgraph "Processing Pipeline"
D1[URL Deduplication]
D2[Pattern Filtering]
D3[Relevance Scoring]
D4[Quality Assessment]
end
subgraph "Output Layer"
E1[Scored URL Lists]
E2[Domain Statistics]
E3[Performance Metrics]
E4[Cache Storage]
end
A1 --> B1
A2 --> B1
A3 --> B1
B1 --> B2
B2 --> B3
B3 --> B4
B2 --> C1
B2 --> C2
B2 --> C3
B2 --> C4
C1 --> D1
C2 --> D1
C3 --> D2
C4 --> D3
D1 --> D2
D2 --> D3
D3 --> D4
D4 --> E1
B4 --> E2
B3 --> E3
D1 --> E4
style B1 fill:#e3f2fd
style D3 fill:#f3e5f5
style E1 fill:#c8e6c9
```
### Complete Discovery-to-Crawl Pipeline
```mermaid
stateDiagram-v2
[*] --> Discovery
Discovery --> SourceSelection: Configure data sources
SourceSelection --> Sitemap: source="sitemap"
SourceSelection --> CommonCrawl: source="cc"
SourceSelection --> Both: source="sitemap+cc"
Sitemap --> URLCollection
CommonCrawl --> URLCollection
Both --> URLCollection
URLCollection --> Filtering: Apply patterns
Filtering --> MetadataExtraction: extract_head=True
Filtering --> LiveValidation: extract_head=False
MetadataExtraction --> LiveValidation: live_check=True
MetadataExtraction --> RelevanceScoring: live_check=False
LiveValidation --> RelevanceScoring
RelevanceScoring --> ResultRanking: query provided
RelevanceScoring --> ResultLimiting: no query
ResultRanking --> ResultLimiting: apply score_threshold
ResultLimiting --> URLSelection: apply max_urls
URLSelection --> CrawlPreparation: URLs ready
CrawlPreparation --> CrawlExecution: AsyncWebCrawler
CrawlExecution --> StreamProcessing: stream=True
CrawlExecution --> BatchProcessing: stream=False
StreamProcessing --> [*]
BatchProcessing --> [*]
note right of Discovery : 🔍 Smart URL Discovery
note right of URLCollection : 📚 Merge & Deduplicate
note right of RelevanceScoring : 🎯 BM25 Algorithm
note right of CrawlExecution : 🕷️ High-Performance Crawling
```
### Performance Optimization Strategies
```mermaid
graph LR
subgraph "Input Optimization"
A1[Smart Source Selection] --> A2[Sitemap First]
A2 --> A3[Add CC if Needed]
A3 --> A4[Pattern Filtering Early]
end
subgraph "Processing Optimization"
B1[Parallel Workers] --> B2[Bounded Queues]
B2 --> B3[Rate Limiting]
B3 --> B4[Memory Management]
B4 --> B5[Lazy Evaluation]
end
subgraph "Output Optimization"
C1[Relevance Threshold] --> C2[Max URL Limits]
C2 --> C3[Caching Strategy]
C3 --> C4[Streaming Results]
end
subgraph "Performance Metrics"
D1[URLs/Second: 100-1000]
D2[Memory Usage: Bounded]
D3[Network Efficiency: 95%+]
D4[Cache Hit Rate: 80%+]
end
A4 --> B1
B5 --> C1
C4 --> D1
style A2 fill:#e8f5e8
style B2 fill:#e3f2fd
style C3 fill:#f3e5f5
style D3 fill:#c8e6c9
```
### URL Discovery vs Traditional Crawling Comparison
```mermaid
graph TB
subgraph "Traditional Approach"
T1[Start URL] --> T2[Crawl Page]
T2 --> T3[Extract Links]
T3 --> T4[Queue New URLs]
T4 --> T2
T5[❌ Time: Hours/Days]
T6[❌ Resource Heavy]
T7[❌ Depth Limited]
T8[❌ Discovery Bias]
end
subgraph "URL Seeding Approach"
S1[Domain Input] --> S2[Query All Sources]
S2 --> S3[Pattern Filter]
S3 --> S4[Relevance Score]
S4 --> S5[Select Best URLs]
S5 --> S6[Ready to Crawl]
S7[✅ Time: Seconds/Minutes]
S8[✅ Resource Efficient]
S9[✅ Complete Coverage]
S10[✅ Quality Focused]
end
subgraph "Use Case Decision Matrix"
U1[Small Sites < 1000 pages] --> U2[Use Deep Crawling]
U3[Large Sites > 10000 pages] --> U4[Use URL Seeding]
U5[Unknown Structure] --> U6[Start with Seeding]
U7[Real-time Discovery] --> U8[Use Deep Crawling]
U9[Quality over Quantity] --> U10[Use URL Seeding]
end
style S6 fill:#c8e6c9
style S7 fill:#c8e6c9
style S8 fill:#c8e6c9
style S9 fill:#c8e6c9
style S10 fill:#c8e6c9
style T5 fill:#ffcdd2
style T6 fill:#ffcdd2
style T7 fill:#ffcdd2
style T8 fill:#ffcdd2
```
### Data Source Characteristics and Selection
```mermaid
graph TB
subgraph "Sitemap Source"
SM1[📋 Official URL List]
SM2[⚡ Fast Response]
SM3[📅 Recently Updated]
SM4[🎯 High Quality URLs]
SM5[❌ May Miss Some Pages]
end
subgraph "Common Crawl Source"
CC1[🌐 Comprehensive Coverage]
CC2[📚 Historical Data]
CC3[🔍 Deep Discovery]
CC4[⏳ Slower Response]
CC5[🧹 May Include Noise]
end
subgraph "Combined Strategy"
CB1[🚀 Best of Both]
CB2[📊 Maximum Coverage]
CB3[✨ Automatic Deduplication]
CB4[⚖️ Balanced Performance]
end
subgraph "Selection Guidelines"
G1[Speed Critical → Sitemap Only]
G2[Coverage Critical → Common Crawl]
G3[Best Quality → Combined]
G4[Unknown Domain → Combined]
end
style SM2 fill:#c8e6c9
style SM4 fill:#c8e6c9
style CC1 fill:#e3f2fd
style CC3 fill:#e3f2fd
style CB1 fill:#f3e5f5
style CB3 fill:#f3e5f5
```
**📖 Learn more:** [URL Seeding Guide](https://docs.crawl4ai.com/core/url-seeding/), [Performance Optimization](https://docs.crawl4ai.com/advanced/optimization/), [Multi-URL Crawling](https://docs.crawl4ai.com/advanced/multi-url-crawling/)
---
## Deep Crawling Filters & Scorers Architecture
Visual representations of advanced URL filtering, scoring strategies, and performance optimization workflows for intelligent deep crawling.
### Filter Chain Processing Pipeline
```mermaid
flowchart TD
A[URL Input] --> B{Domain Filter}
B -->|✓ Pass| C{Pattern Filter}
B -->|✗ Fail| X1[Reject: Invalid Domain]
C -->|✓ Pass| D{Content Type Filter}
C -->|✗ Fail| X2[Reject: Pattern Mismatch]
D -->|✓ Pass| E{SEO Filter}
D -->|✗ Fail| X3[Reject: Wrong Content Type]
E -->|✓ Pass| F{Content Relevance Filter}
E -->|✗ Fail| X4[Reject: Low SEO Score]
F -->|✓ Pass| G[URL Accepted]
F -->|✗ Fail| X5[Reject: Low Relevance]
G --> H[Add to Crawl Queue]
subgraph "Fast Filters"
B
C
D
end
subgraph "Slow Filters"
E
F
end
style A fill:#e3f2fd
style G fill:#c8e6c9
style H fill:#e8f5e8
style X1 fill:#ffcdd2
style X2 fill:#ffcdd2
style X3 fill:#ffcdd2
style X4 fill:#ffcdd2
style X5 fill:#ffcdd2
```
### URL Scoring System Architecture
```mermaid
graph TB
subgraph "Input URL"
A[https://python.org/tutorial/2024/ml-guide.html]
end
subgraph "Individual Scorers"
B[Keyword Relevance Scorer]
C[Path Depth Scorer]
D[Content Type Scorer]
E[Freshness Scorer]
F[Domain Authority Scorer]
end
subgraph "Scoring Process"
B --> B1[Keywords: python, tutorial, ml<br/>Score: 0.85]
C --> C1[Depth: 4 levels<br/>Optimal: 3<br/>Score: 0.75]
D --> D1[Content: HTML<br/>Score: 1.0]
E --> E1[Year: 2024<br/>Score: 1.0]
F --> F1[Domain: python.org<br/>Score: 1.0]
end
subgraph "Composite Scoring"
G[Weighted Combination]
B1 --> G
C1 --> G
D1 --> G
E1 --> G
F1 --> G
end
subgraph "Final Result"
H[Composite Score: 0.92]
I{Score > Threshold?}
J[Accept URL]
K[Reject URL]
end
A --> B
A --> C
A --> D
A --> E
A --> F
G --> H
H --> I
I -->|✓ 0.92 > 0.6| J
I -->|✗ Score too low| K
style A fill:#e3f2fd
style G fill:#fff3e0
style H fill:#e8f5e8
style J fill:#c8e6c9
style K fill:#ffcdd2
```
### Filter vs Scorer Decision Matrix
```mermaid
flowchart TD
A[URL Processing Decision] --> B{Binary Decision Needed?}
B -->|Yes - Include/Exclude| C[Use Filters]
B -->|No - Quality Rating| D[Use Scorers]
C --> C1{Filter Type Needed?}
C1 -->|Domain Control| C2[DomainFilter]
C1 -->|Pattern Matching| C3[URLPatternFilter]
C1 -->|Content Type| C4[ContentTypeFilter]
C1 -->|SEO Quality| C5[SEOFilter]
C1 -->|Content Relevance| C6[ContentRelevanceFilter]
D --> D1{Scoring Criteria?}
D1 -->|Keyword Relevance| D2[KeywordRelevanceScorer]
D1 -->|URL Structure| D3[PathDepthScorer]
D1 -->|Content Quality| D4[ContentTypeScorer]
D1 -->|Time Sensitivity| D5[FreshnessScorer]
D1 -->|Source Authority| D6[DomainAuthorityScorer]
C2 --> E[Chain Filters]
C3 --> E
C4 --> E
C5 --> E
C6 --> E
D2 --> F[Composite Scorer]
D3 --> F
D4 --> F
D5 --> F
D6 --> F
E --> G[Binary Output: Pass/Fail]
F --> H[Numeric Score: 0.0-1.0]
G --> I[Apply to URL Queue]
H --> J[Priority Ranking]
style C fill:#e8f5e8
style D fill:#fff3e0
style E fill:#f3e5f5
style F fill:#e3f2fd
style G fill:#c8e6c9
style H fill:#ffecb3
```
### Performance Optimization Strategy
```mermaid
sequenceDiagram
participant Queue as URL Queue
participant Fast as Fast Filters
participant Slow as Slow Filters
participant Score as Scorers
participant Output as Filtered URLs
Note over Queue, Output: Batch Processing (1000 URLs)
Queue->>Fast: Apply Domain Filter
Fast-->>Queue: 60% passed (600 URLs)
Queue->>Fast: Apply Pattern Filter
Fast-->>Queue: 70% passed (420 URLs)
Queue->>Fast: Apply Content Type Filter
Fast-->>Queue: 90% passed (378 URLs)
Note over Fast: Fast filters eliminate 62% of URLs
Queue->>Slow: Apply SEO Filter (378 URLs)
Slow-->>Queue: 80% passed (302 URLs)
Queue->>Slow: Apply Relevance Filter
Slow-->>Queue: 75% passed (227 URLs)
Note over Slow: Content analysis on remaining URLs
Queue->>Score: Calculate Composite Scores
Score-->>Queue: Scored and ranked
Queue->>Output: Top 100 URLs by score
Output-->>Queue: Processing complete
Note over Queue, Output: Total: 90% filtered out, 10% high-quality URLs retained
```
### Custom Filter Implementation Flow
```mermaid
stateDiagram-v2
[*] --> Planning
Planning --> IdentifyNeeds: Define filtering criteria
IdentifyNeeds --> ChooseType: Binary vs Scoring decision
ChooseType --> FilterImpl: Binary decision needed
ChooseType --> ScorerImpl: Quality rating needed
FilterImpl --> InheritURLFilter: Extend URLFilter base class
ScorerImpl --> InheritURLScorer: Extend URLScorer base class
InheritURLFilter --> ImplementApply: def apply(url) -> bool
InheritURLScorer --> ImplementScore: def _calculate_score(url) -> float
ImplementApply --> AddLogic: Add custom filtering logic
ImplementScore --> AddLogic
AddLogic --> TestFilter: Unit testing
TestFilter --> OptimizePerf: Performance optimization
OptimizePerf --> Integration: Integrate with FilterChain
Integration --> Production: Deploy to production
Production --> Monitor: Monitor performance
Monitor --> Tune: Tune parameters
Tune --> Production
note right of Planning : Consider performance impact
note right of AddLogic : Handle edge cases
note right of OptimizePerf : Cache frequently accessed data
```
### Filter Chain Optimization Patterns
```mermaid
graph TB
subgraph "Naive Approach - Poor Performance"
A1[All URLs] --> B1[Slow Filter 1]
B1 --> C1[Slow Filter 2]
C1 --> D1[Fast Filter 1]
D1 --> E1[Fast Filter 2]
E1 --> F1[Final Results]
B1 -.->|High CPU| G1[Performance Issues]
C1 -.->|Network Calls| G1
end
subgraph "Optimized Approach - High Performance"
A2[All URLs] --> B2[Fast Filter 1]
B2 --> C2[Fast Filter 2]
C2 --> D2[Batch Process]
D2 --> E2[Slow Filter 1]
E2 --> F2[Slow Filter 2]
F2 --> G2[Final Results]
D2 --> H2[Concurrent Processing]
H2 --> I2[Semaphore Control]
end
subgraph "Performance Metrics"
J[Processing Time]
K[Memory Usage]
L[CPU Utilization]
M[Network Requests]
end
G1 -.-> J
G1 -.-> K
G1 -.-> L
G1 -.-> M
G2 -.-> J
G2 -.-> K
G2 -.-> L
G2 -.-> M
style A1 fill:#ffcdd2
style G1 fill:#ffcdd2
style A2 fill:#c8e6c9
style G2 fill:#c8e6c9
style H2 fill:#e8f5e8
style I2 fill:#e8f5e8
```
### Composite Scoring Weight Distribution
```mermaid
pie title Composite Scorer Weight Distribution
"Keyword Relevance (30%)" : 30
"Domain Authority (25%)" : 25
"Content Type (20%)" : 20
"Freshness (15%)" : 15
"Path Depth (10%)" : 10
```
### Deep Crawl Integration Architecture
```mermaid
graph TD
subgraph "Deep Crawl Strategy"
A[Start URL] --> B[Extract Links]
B --> C[Apply Filter Chain]
C --> D[Calculate Scores]
D --> E[Priority Queue]
E --> F[Crawl Next URL]
F --> B
end
subgraph "Filter Chain Components"
C --> C1[Domain Filter]
C --> C2[Pattern Filter]
C --> C3[Content Filter]
C --> C4[SEO Filter]
C --> C5[Relevance Filter]
end
subgraph "Scoring Components"
D --> D1[Keyword Scorer]
D --> D2[Depth Scorer]
D --> D3[Freshness Scorer]
D --> D4[Authority Scorer]
D --> D5[Composite Score]
end
subgraph "Queue Management"
E --> E1{Score > Threshold?}
E1 -->|Yes| E2[High Priority Queue]
E1 -->|No| E3[Low Priority Queue]
E2 --> F
E3 --> G[Delayed Processing]
end
subgraph "Control Flow"
H{Max Depth Reached?}
I{Max Pages Reached?}
J[Stop Crawling]
end
F --> H
H -->|No| I
H -->|Yes| J
I -->|No| B
I -->|Yes| J
style A fill:#e3f2fd
style E2 fill:#c8e6c9
style E3 fill:#fff3e0
style J fill:#ffcdd2
```
### Filter Performance Comparison
```mermaid
xychart-beta
title "Filter Performance Comparison (1000 URLs)"
x-axis [Domain, Pattern, ContentType, SEO, Relevance]
y-axis "Processing Time (ms)" 0 --> 1000
bar [50, 80, 45, 300, 800]
```
### Scoring Algorithm Workflow
```mermaid
flowchart TD
A[Input URL] --> B[Parse URL Components]
B --> C[Extract Features]
C --> D[Domain Analysis]
C --> E[Path Analysis]
C --> F[Content Type Detection]
C --> G[Keyword Extraction]
C --> H[Freshness Detection]
D --> I[Domain Authority Score]
E --> J[Path Depth Score]
F --> K[Content Type Score]
G --> L[Keyword Relevance Score]
H --> M[Freshness Score]
I --> N[Apply Weights]
J --> N
K --> N
L --> N
M --> N
N --> O[Normalize Scores]
O --> P[Calculate Final Score]
P --> Q{Score >= Threshold?}
Q -->|Yes| R[Accept for Crawling]
Q -->|No| S[Reject URL]
R --> T[Add to Priority Queue]
S --> U[Log Rejection Reason]
style A fill:#e3f2fd
style P fill:#fff3e0
style R fill:#c8e6c9
style S fill:#ffcdd2
style T fill:#e8f5e8
```
**📖 Learn more:** [Deep Crawling Strategy](https://docs.crawl4ai.com/core/deep-crawling/), [Performance Optimization](https://docs.crawl4ai.com/advanced/performance-tuning/), [Custom Implementations](https://docs.crawl4ai.com/advanced/custom-filters/)
---
## Summary
Crawl4AI provides a comprehensive solution for web crawling and data extraction optimized for AI applications. From simple page crawling to complex multi-URL operations with advanced filtering, the library offers the flexibility and performance needed for modern data extraction workflows.
**Key Takeaways:**
- Start with basic installation and simple crawling patterns
- Use configuration objects for consistent, maintainable code
- Choose appropriate extraction strategies based on your data structure
- Leverage Docker for production deployments
- Implement advanced features like deep crawling and custom filters as needed
**Next Steps:**
- Explore the [GitHub repository](https://github.com/unclecode/crawl4ai) for latest updates
- Join the [Discord community](https://discord.gg/jP8KfhDhyN) for support
- Check out [example projects](https://github.com/unclecode/crawl4ai/tree/main/docs/examples) for inspiration
Happy crawling! 🕷️
Reference in New Issue View Git Blame Copy Permalink