**File: `crawl4ai/async_crawler_strategy.py`**
**New Parameters and Attributes Added**
- **`text_mode` (boolean)**: Enables text-only mode, disables images, JavaScript, and GPU-related features for faster, minimal rendering.
- **`light_mode` (boolean)**: Optimizes the browser by disabling unnecessary background processes and features for efficiency.
- **`viewport_width` and `viewport_height`**: Dynamically adjusts based on `text_mode` mode (default values: 800x600 for `text_mode`, 1920x1080 otherwise).
- **`extra_args`**: Adds browser-specific flags for `text_mode` mode.
- **`adjust_viewport_to_content`**: Dynamically adjusts the viewport to the content size for accurate rendering.
**Browser Context Adjustments**
- Added **`viewport` adjustments**: Dynamically computed based on `text_mode` or custom configuration.
- Enhanced support for `light_mode` and `text_mode` by adding specific browser arguments to reduce resource consumption.
**Dynamic Content Handling**
- **Full Page Scan Feature**:
- Scrolls through the entire page while dynamically detecting content changes.
- Ensures scrolling stops when no new dynamic content is loaded.
**Session Management**
- Added **`create_session`** method:
- Creates a new browser session and assigns a unique ID.
- Supports persistent and non-persistent contexts with full compatibility for cookies, headers, and proxies.
**Improved Content Loading and Adjustment**
- **`adjust_viewport_to_content`**:
- Automatically adjusts viewport to match content dimensions.
- Includes scaling via Chrome DevTools Protocol (CDP).
- Enhanced content loading:
- Waits for images to load and ensures network activity is idle before proceeding.
**Error Handling and Logging**
- Improved error handling and detailed logging for:
- Viewport adjustment (`adjust_viewport_to_content`).
- Full page scanning (`scan_full_page`).
- Dynamic content loading.
**Refactoring and Cleanup**
- Removed hardcoded viewport dimensions in multiple places, replaced with dynamic values (`self.viewport_width`, `self.viewport_height`).
- Removed commented-out and unused code for better readability.
- Added default value for `delay_before_return_html` parameter.
**Optimizations**
- Reduced resource usage in `light_mode` by disabling unnecessary browser features such as extensions, background timers, and sync.
- Improved compatibility for different browser types (`chrome`, `firefox`, `webkit`).
---
**File: `docs/examples/quickstart_async.py`**
**Schema Adjustment**
- Changed schema reference for `LLMExtractionStrategy`:
- **Old**: `OpenAIModelFee.schema()`
- **New**: `OpenAIModelFee.model_json_schema()`
- This likely ensures better compatibility with the `OpenAIModelFee` class and its JSON schema.
**Documentation Comments Updated**
- Improved extraction instruction for schema-based LLM strategies.
---
**New Features Added**
1. **Text-Only Mode**:
- Focuses on minimal resource usage by disabling non-essential browser features.
2. **Light Mode**:
- Optimizes browser for performance by disabling background tasks and unnecessary services.
3. **Full Page Scanning**:
- Ensures the entire content of a page is crawled, including dynamic elements loaded during scrolling.
4. **Dynamic Viewport Adjustment**:
- Automatically resizes the viewport to match content dimensions, improving compatibility and rendering accuracy.
5. **Session Management**:
- Simplifies session handling with better support for persistent and non-persistent contexts.
---
**Bug Fixes**
- Fixed potential viewport mismatches by ensuring consistent use of `self.viewport_width` and `self.viewport_height` throughout the code.
- Improved robustness of dynamic content loading to avoid timeouts and failed evaluations.