ayrisdev/crawl4ai
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge branch 'vr0.4.3b3'

Browse Source
This commit is contained in:
UncleCode
2025-01-25 21:57:29 +08:00
parent 0afc3e9e5e 09ac7ed008
commit d0586f09a9
11 changed files with 425 additions and 75 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
docs/examples/v0_4_3b2_features_demo.py
View File

@@ -297,8 +297,7 @@ async def demo_proxy_rotation():
}
except Exception as e:
print(f"Error loading proxy: {e}")
return None
return None
# Create 10 test requests to httpbin
urls = ["https://httpbin.org/ip"] * 2
@@ -314,7 +313,7 @@ async def demo_proxy_rotation():
continue
# Create new config with proxy
current_config = run_config.clone(proxy_config=proxy)
current_config = run_config.clone(proxy_config=proxy, user_agent="")
result = await crawler.arun(url=url, config=current_config)
if result.success:

195
docs/md_v2/advanced/multi-url-crawling.md
View File

@@ -5,16 +5,20 @@
## 1. Introduction
When crawling many URLs:
- **Basic**: Use `arun()` in a loop (simple but less efficient)
- **Better**: Use `arun_many()`, which efficiently handles multiple URLs with proper concurrency control
- **Best**: Customize dispatcher behavior for your specific needs (memory management, rate limits, etc.)
**Why Dispatchers?**
- **Adaptive**: Memory-based dispatchers can pause or slow down based on system resources
- **Rate-limiting**: Built-in rate limiting with exponential backoff for 429/503 responses
- **Real-time Monitoring**: Live dashboard of ongoing tasks, memory usage, and performance
- **Flexibility**: Choose between memory-adaptive or semaphore-based concurrency
---
## 2. Core Components
### 2.1 Rate Limiter
@@ -22,34 +26,116 @@ When crawling many URLs:
```python
class RateLimiter:
def __init__(
base_delay: Tuple[float, float] = (1.0, 3.0), # Random delay range between requests
max_delay: float = 60.0, # Maximum backoff delay
max_retries: int = 3, # Retries before giving up
rate_limit_codes: List[int] = [429, 503] # Status codes triggering backoff
# Random delay range between requests
base_delay: Tuple[float, float] = (1.0, 3.0),
# Maximum backoff delay
max_delay: float = 60.0,
# Retries before giving up
max_retries: int = 3,
# Status codes triggering backoff
rate_limit_codes: List[int] = [429, 503]
)
```
The RateLimiter provides:
- Random delays between requests
- Exponential backoff on rate limit responses
- Domain-specific rate limiting
- Automatic retry handling
Here’s the revised and simplified explanation of the **RateLimiter**, focusing on constructor parameters and adhering to your markdown style and mkDocs guidelines.
#### RateLimiter Constructor Parameters
The **RateLimiter** is a utility that helps manage the pace of requests to avoid overloading servers or getting blocked due to rate limits. It operates internally to delay requests and handle retries but can be configured using its constructor parameters.
**Parameters of the `RateLimiter` constructor:**
1. **`base_delay`** (`Tuple[float, float]`, default: `(1.0, 3.0)`)
  The range for a random delay (in seconds) between consecutive requests to the same domain.
- A random delay is chosen between `base_delay[0]` and `base_delay[1]` for each request.
- This prevents sending requests at a predictable frequency, reducing the chances of triggering rate limits.
**Example:**
If `base_delay = (2.0, 5.0)`, delays could be randomly chosen as `2.3s`, `4.1s`, etc.
---
2. **`max_delay`** (`float`, default: `60.0`)
  The maximum allowable delay when rate-limiting errors occur.
- When servers return rate-limit responses (e.g., 429 or 503), the delay increases exponentially with jitter.
- The `max_delay` ensures the delay doesn’t grow unreasonably high, capping it at this value.
**Example:**
For a `max_delay = 30.0`, even if backoff calculations suggest a delay of `45s`, it will cap at `30s`.
---
3. **`max_retries`** (`int`, default: `3`)
  The maximum number of retries for a request if rate-limiting errors occur.
- After encountering a rate-limit response, the `RateLimiter` retries the request up to this number of times.
- If all retries fail, the request is marked as failed, and the process continues.
**Example:**
If `max_retries = 3`, the system retries a failed request three times before giving up.
---
4. **`rate_limit_codes`** (`List[int]`, default: `[429, 503]`)
  A list of HTTP status codes that trigger the rate-limiting logic.
- These status codes indicate the server is overwhelmed or actively limiting requests.
- You can customize this list to include other codes based on specific server behavior.
**Example:**
If `rate_limit_codes = [429, 503, 504]`, the crawler will back off on these three error codes.
---
**How to Use the `RateLimiter`:**
Here’s an example of initializing and using a `RateLimiter` in your project:
```python
from crawl4ai import RateLimiter
# Create a RateLimiter with custom settings
rate_limiter = RateLimiter(
base_delay=(2.0, 4.0), # Random delay between 2-4 seconds
max_delay=30.0, # Cap delay at 30 seconds
max_retries=5, # Retry up to 5 times on rate-limiting errors
rate_limit_codes=[429, 503] # Handle these HTTP status codes
)
# RateLimiter will handle delays and retries internally
# No additional setup is required for its operation
```
The `RateLimiter` integrates seamlessly with dispatchers like `MemoryAdaptiveDispatcher` and `SemaphoreDispatcher`, ensuring requests are paced correctly without user intervention. Its internal mechanisms manage delays and retries to avoid overwhelming servers while maximizing efficiency.
### 2.2 Crawler Monitor
The CrawlerMonitor provides real-time visibility into crawling operations:
```python
from crawl4ai import CrawlerMonitor, DisplayMode
monitor = CrawlerMonitor(
max_visible_rows=15, # Maximum rows in live display
display_mode=DisplayMode.DETAILED # DETAILED or AGGREGATED view
# Maximum rows in live display
max_visible_rows=15,
# DETAILED or AGGREGATED view
display_mode=DisplayMode.DETAILED
)
```
**Display Modes**:
1. **DETAILED**: Shows individual task status, memory usage, and timing
2. **AGGREGATED**: Displays summary statistics and overall progress
---
## 3. Available Dispatchers
### 3.1 MemoryAdaptiveDispatcher (Default)
@@ -57,6 +143,8 @@ monitor = CrawlerMonitor(
Automatically manages concurrency based on system memory usage:
```python
from crawl4ai.async_dispatcher import MemoryAdaptiveDispatcher
dispatcher = MemoryAdaptiveDispatcher(
memory_threshold_percent=90.0, # Pause if memory exceeds this
check_interval=1.0, # How often to check memory
@@ -73,13 +161,37 @@ dispatcher = MemoryAdaptiveDispatcher(
)
```
**Constructor Parameters:**
1. **`memory_threshold_percent`** (`float`, default: `90.0`)
  Specifies the memory usage threshold (as a percentage). If system memory usage exceeds this value, the dispatcher pauses crawling to prevent system overload.
2. **`check_interval`** (`float`, default: `1.0`)
  The interval (in seconds) at which the dispatcher checks system memory usage.
3. **`max_session_permit`** (`int`, default: `10`)
  The maximum number of concurrent crawling tasks allowed. This ensures resource limits are respected while maintaining concurrency.
4. **`memory_wait_timeout`** (`float`, default: `300.0`)
  Optional timeout (in seconds). If memory usage exceeds `memory_threshold_percent` for longer than this duration, a `MemoryError` is raised.
5. **`rate_limiter`** (`RateLimiter`, default: `None`)
  Optional rate-limiting logic to avoid server-side blocking (e.g., for handling 429 or 503 errors). See **RateLimiter** for details.
6. **`monitor`** (`CrawlerMonitor`, default: `None`)
  Optional monitoring for real-time task tracking and performance insights. See **CrawlerMonitor** for details.
---
### 3.2 SemaphoreDispatcher
Provides simple concurrency control with a fixed limit:
```python
from crawl4ai.async_dispatcher import SemaphoreDispatcher
dispatcher = SemaphoreDispatcher(
max_session_permit=5, # Fixed concurrent tasks
max_session_permit=20, # Maximum concurrent tasks
rate_limiter=RateLimiter( # Optional rate limiting
base_delay=(0.5, 1.0),
max_delay=10.0
@@ -91,6 +203,19 @@ dispatcher = SemaphoreDispatcher(
)
```
**Constructor Parameters:**
1. **`max_session_permit`** (`int`, default: `20`)
  The maximum number of concurrent crawling tasks allowed, irrespective of semaphore slots.
2. **`rate_limiter`** (`RateLimiter`, default: `None`)
  Optional rate-limiting logic to avoid overwhelming servers. See **RateLimiter** for details.
3. **`monitor`** (`CrawlerMonitor`, default: `None`)
  Optional monitoring for tracking task progress and resource usage. See **CrawlerMonitor** for details.
---
## 4. Usage Examples
### 4.1 Batch Processing (Default)
@@ -128,6 +253,14 @@ async def crawl_batch():
print(f"Failed to crawl {result.url}: {result.error_message}")
```
**Review:**
- **Purpose:** Executes a batch crawl with all URLs processed together after crawling is complete.
- **Dispatcher:** Uses `MemoryAdaptiveDispatcher` to manage concurrency and system memory.
- **Stream:** Disabled (`stream=False`), so all results are collected at once for post-processing.
- **Best Use Case:** When you need to analyze results in bulk rather than individually during the crawl.
---
### 4.2 Streaming Mode
```python
@@ -161,6 +294,14 @@ async def crawl_streaming():
print(f"Failed to crawl {result.url}: {result.error_message}")
```
**Review:**
- **Purpose:** Enables streaming to process results as soon as they’re available.
- **Dispatcher:** Uses `MemoryAdaptiveDispatcher` for concurrency and memory management.
- **Stream:** Enabled (`stream=True`), allowing real-time processing during crawling.
- **Best Use Case:** When you need to act on results immediately, such as for real-time analytics or progressive data storage.
---
### 4.3 Semaphore-based Crawling
```python
@@ -189,6 +330,14 @@ async def crawl_with_semaphore(urls):
return results
```
**Review:**
- **Purpose:** Uses `SemaphoreDispatcher` to limit concurrency with a fixed number of slots.
- **Dispatcher:** Configured with a semaphore to control parallel crawling tasks.
- **Rate Limiter:** Prevents servers from being overwhelmed by pacing requests.
- **Best Use Case:** When you want precise control over the number of concurrent requests, independent of system memory.
---
### 4.4 Robots.txt Consideration
```python
@@ -221,11 +370,13 @@ if __name__ == "__main__":
asyncio.run(main())
```
**Key Points**:
- When `check_robots_txt=True`, each URL's robots.txt is checked before crawling
- Robots.txt files are cached for efficiency
- Failed robots.txt checks return 403 status code
- Dispatcher handles robots.txt checks automatically for each URL
**Review:**
- **Purpose:** Ensures compliance with `robots.txt` rules for ethical and legal web crawling.
- **Configuration:** Set `check_robots_txt=True` to validate each URL against `robots.txt` before crawling.
- **Dispatcher:** Handles requests with concurrency limits (`semaphore_count=3`).
- **Best Use Case:** When crawling websites that strictly enforce robots.txt policies or for responsible crawling practices.
---
## 5. Dispatch Results
@@ -255,20 +406,24 @@ for result in results:
## 6. Summary
1. **Two Dispatcher Types**:
1. **Two Dispatcher Types**:
- MemoryAdaptiveDispatcher (default): Dynamic concurrency based on memory
- SemaphoreDispatcher: Fixed concurrency limit
2. **Optional Components**:
2. **Optional Components**:
- RateLimiter: Smart request pacing and backoff
- CrawlerMonitor: Real-time progress visualization
3. **Key Benefits**:
3. **Key Benefits**:
- Automatic memory management
- Built-in rate limiting
- Live progress monitoring
- Flexible concurrency control
Choose the dispatcher that best fits your needs:
- **MemoryAdaptiveDispatcher**: For large crawls or limited resources
- **SemaphoreDispatcher**: For simple, fixed-concurrency scenarios

14
docs/md_v2/assets/styles.css
View File

@@ -95,6 +95,10 @@ strong {
}
div.highlight {
margin-bottom: 2em;
}
.terminal-card > header {
color: var(--font-color);
text-align: center;
@@ -231,6 +235,16 @@ pre {
font-size: 2em;
}
.terminal h2 {
font-size: 1.5em;
margin-bottom: 0.8em;
}
.terminal h3 {
font-size: 1.3em;
margin-bottom: 0.8em;
}
.terminal h1, .terminal h2, .terminal h3, .terminal h4, .terminal h5, .terminal h6 {
text-shadow: 0 0 0px var(--font-color), 0 0 0px var(--font-color), 0 0 0px var(--font-color);
}