crawlo 1.1.4__py3-none-any.whl → 1.1.6__py3-none-any.whl

crawlo/core/engine.py

@@ -1,172 +1,346 @@
-#!/usr/bin/python
-# -*- coding:UTF-8 -*-
-import asyncio
-from inspect import iscoroutine
-from typing import Optional, Generator, Callable
-
-from crawlo import Request, Item
-from crawlo.spider import Spider
-from crawlo.utils.log import get_logger
-from crawlo.exceptions import OutputError
-from crawlo.core.scheduler import Scheduler
-from crawlo.core.processor import Processor
-from crawlo.task_manager import TaskManager
-from crawlo.project import load_class
-from crawlo.downloader import DownloaderBase
-from crawlo.utils.func_tools import transform
-from crawlo.event import spider_opened, spider_error, request_scheduled
-
-
-class Engine(object):
-
-    def __init__(self, crawler):
-        self.running = False
-        self.normal = True
-        self.crawler = crawler
-        self.settings = crawler.settings
-        self.spider: Optional[Spider] = None
-        self.downloader: Optional[DownloaderBase] = None
-        self.scheduler: Optional[Scheduler] = None
-        self.processor: Optional[Processor] = None
-        self.start_requests: Optional[Generator] = None
-        self.task_manager: Optional[TaskManager] = TaskManager(self.settings.get_int('CONCURRENCY'))
-
-        self.logger = get_logger(name=self.__class__.__name__)
-
-    def _get_downloader_cls(self):
-        """获取下载器类，支持多种配置方式"""
-        # 方式1: 使用 DOWNLOADER_TYPE 简化名称（推荐）
-        downloader_type = self.settings.get('DOWNLOADER_TYPE')
-        if downloader_type:
-            try:
-                from crawlo.downloader import get_downloader_class
-                downloader_cls = get_downloader_class(downloader_type)
-                self.logger.debug(f"使用下载器类型: {downloader_type} -> {downloader_cls.__name__}")
-                return downloader_cls
-            except (ImportError, ValueError) as e:
-                self.logger.warning(f"无法使用下载器类型 '{downloader_type}': {e}，回退到默认配置")
-        
-        # 方式2: 使用 DOWNLOADER 完整类路径（兼容旧版本）
-        downloader_cls = load_class(self.settings.get('DOWNLOADER'))
-        if not issubclass(downloader_cls, DownloaderBase):
-            raise TypeError(f'Downloader {downloader_cls.__name__} is not subclass of DownloaderBase.')
-        return downloader_cls
-
-    def engine_start(self):
-        self.running = True
-        self.logger.info(
-            f"Crawlo (version {self.settings.get_float('VERSION')}) started. "
-            f"(project name : {self.settings.get('PROJECT_NAME')})"
-        )
-
-    async def start_spider(self, spider):
-        self.spider = spider
-
-        self.scheduler = Scheduler.create_instance(self.crawler)
-        if hasattr(self.scheduler, 'open'):
-            await self.scheduler.open()
-
-        downloader_cls = self._get_downloader_cls()
-        self.downloader = downloader_cls(self.crawler)
-        if hasattr(self.downloader, 'open'):
-            self.downloader.open()
-
-        self.processor = Processor(self.crawler)
-        if hasattr(self.processor, 'open'):
-            self.processor.open()
-
-        self.start_requests = iter(spider.start_requests())
-        await self._open_spider()
-
-    async def crawl(self):
-        """
-        Crawl the spider
-        """
-        while self.running:
-            if request := await self._get_next_request():
-                await self._crawl(request)
-            try:
-                start_request = next(self.start_requests)
-            except StopIteration:
-                self.start_requests = None
-            except Exception as exp:
-                # 1、发去请求的request全部运行完毕
-                # 2、调度器是否空闲
-                # 3、下载器是否空闲
-                if not await self._exit():
-                    continue
-                self.running = False
-                if self.start_requests is not None:
-                    self.logger.error(f"启动请求时发生错误: {str(exp)}")
-            else:
-                # 请求入队
-                await self.enqueue_request(start_request)
-
-        if not self.running:
-            await self.close_spider()
-
-    async def _open_spider(self):
-        asyncio.create_task(self.crawler.subscriber.notify(spider_opened))
-        crawling = asyncio.create_task(self.crawl())
-        await crawling
-
-    async def _crawl(self, request):
-        # TODO 实现并发
-        async def crawl_task():
-            outputs = await self._fetch(request)
-            # TODO 处理output
-            if outputs:
-                await self._handle_spider_output(outputs)
-
-        # 使用异步任务创建，遵守并发限制
-        await self.task_manager.create_task(crawl_task())
-
-    async def _fetch(self, request):
-        async def _successful(_response):
-            callback: Callable = request.callback or self.spider.parse
-            if _outputs := callback(_response):
-                if iscoroutine(_outputs):
-                    await _outputs
-                else:
-                    return transform(_outputs, _response)
-
-        _response = await self.downloader.fetch(request)
-        if _response is None:
-            return None
-        output = await _successful(_response)
-        return output
-
-    async def enqueue_request(self, start_request):
-        await self._schedule_request(start_request)
-
-    async def _schedule_request(self, request):
-        # TODO 去重
-        if await self.scheduler.enqueue_request(request):
-            asyncio.create_task(self.crawler.subscriber.notify(request_scheduled, request, self.crawler.spider))
-
-    async def _get_next_request(self):
-        return await self.scheduler.next_request()
-
-    async def _handle_spider_output(self, outputs):
-        async for spider_output in outputs:
-            if isinstance(spider_output, (Request, Item)):
-                await self.processor.enqueue(spider_output)
-            elif isinstance(spider_output, Exception):
-                asyncio.create_task(
-                    self.crawler.subscriber.notify(spider_error, spider_output, self.spider)
-                )
-                raise spider_output
-            else:
-                raise OutputError(f'{type(self.spider)} must return `Request` or `Item`.')
-
-    async def _exit(self):
-        if self.scheduler.idle() and self.downloader.idle() and self.task_manager.all_done() and self.processor.idle():
-            return True
-        return False
-
-    async def close_spider(self):
-        await asyncio.gather(*self.task_manager.current_task)
-        await self.scheduler.close()
-        await self.downloader.close()
-        if self.normal:
-            await self.crawler.close()
+#!/usr/bin/python
+# -*- coding:UTF-8 -*-
+import asyncio
+import time
+from inspect import iscoroutine
+from typing import Optional, Generator, Callable
+
+from crawlo import Request, Item
+from crawlo.spider import Spider
+from crawlo.utils.log import get_logger
+from crawlo.exceptions import OutputError
+from crawlo.core.scheduler import Scheduler
+from crawlo.core.processor import Processor
+from crawlo.task_manager import TaskManager
+from crawlo.project import load_class
+from crawlo.downloader import DownloaderBase
+from crawlo.utils.func_tools import transform
+from crawlo.event import spider_opened, spider_error, request_scheduled
+
+
+class Engine(object):
+
+    def __init__(self, crawler):
+        self.running = False
+        self.normal = True
+        self.crawler = crawler
+        self.settings = crawler.settings
+        self.spider: Optional[Spider] = None
+        self.downloader: Optional[DownloaderBase] = None
+        self.scheduler: Optional[Scheduler] = None
+        self.processor: Optional[Processor] = None
+        self.start_requests: Optional[Generator] = None
+        self.task_manager: Optional[TaskManager] = TaskManager(self.settings.get_int('CONCURRENCY'))
+
+        # 增强控制参数
+        self.max_queue_size = self.settings.get_int('SCHEDULER_MAX_QUEUE_SIZE', 200)
+        self.generation_batch_size = self.settings.get_int('REQUEST_GENERATION_BATCH_SIZE', 10)
+        self.generation_interval = self.settings.get_float('REQUEST_GENERATION_INTERVAL', 0.05)
+        self.backpressure_ratio = self.settings.get_float('BACKPRESSURE_RATIO', 0.8)  # 队列达到80%时启动背压
+        
+        # 状态跟踪
+        self._generation_paused = False
+        self._last_generation_time = 0
+        self._generation_stats = {
+            'total_generated': 0,
+            'backpressure_events': 0
+        }
+
+        self.logger = get_logger(name=self.__class__.__name__)
+
+    def _get_downloader_cls(self):
+        """获取下载器类，支持多种配置方式"""
+        # 方式1: 使用 DOWNLOADER_TYPE 简化名称（推荐）
+        downloader_type = self.settings.get('DOWNLOADER_TYPE')
+        if downloader_type:
+            try:
+                from crawlo.downloader import get_downloader_class
+                downloader_cls = get_downloader_class(downloader_type)
+                self.logger.debug(f"使用下载器类型: {downloader_type} -> {downloader_cls.__name__}")
+                return downloader_cls
+            except (ImportError, ValueError) as e:
+                self.logger.warning(f"无法使用下载器类型 '{downloader_type}': {e}，回退到默认配置")
+        
+        # 方式2: 使用 DOWNLOADER 完整类路径（兼容旧版本）
+        downloader_cls = load_class(self.settings.get('DOWNLOADER'))
+        if not issubclass(downloader_cls, DownloaderBase):
+            raise TypeError(f'Downloader {downloader_cls.__name__} is not subclass of DownloaderBase.')
+        return downloader_cls
+
+    def engine_start(self):
+        self.running = True
+        self.logger.info(
+            f"Crawlo (version {self.settings.get('VERSION')}) started. "
+            f"(project name : {self.settings.get('PROJECT_NAME')})"
+        )
+
+    async def start_spider(self, spider):
+        self.spider = spider
+
+        self.scheduler = Scheduler.create_instance(self.crawler)
+        if hasattr(self.scheduler, 'open'):
+            if asyncio.iscoroutinefunction(self.scheduler.open):
+                await self.scheduler.open()
+            else:
+                self.scheduler.open()
+
+        downloader_cls = self._get_downloader_cls()
+        self.downloader = downloader_cls(self.crawler)
+        if hasattr(self.downloader, 'open'):
+            if asyncio.iscoroutinefunction(self.downloader.open):
+                await self.downloader.open()
+            else:
+                self.downloader.open()
+
+        self.processor = Processor(self.crawler)
+        if hasattr(self.processor, 'open'):
+            if asyncio.iscoroutinefunction(self.processor.open):
+                await self.processor.open()
+            else:
+                self.processor.open()
+
+        self.start_requests = iter(spider.start_requests())
+        await self._open_spider()
+
+    async def crawl(self):
+        """
+        Crawl the spider
+        增强版本支持智能请求生成和背压控制
+        """
+        generation_task = None
+        
+        try:
+            # 启动请求生成任务（如果启用了受控生成）
+            if (self.start_requests and 
+                self.settings.get_bool('ENABLE_CONTROLLED_REQUEST_GENERATION', False)):
+                generation_task = asyncio.create_task(
+                    self._controlled_request_generation()
+                )
+            else:
+                # 传统方式处理启动请求
+                generation_task = asyncio.create_task(
+                    self._traditional_request_generation()
+                )
+            
+            # 主爬取循环
+            while self.running:
+                # 获取并处理请求
+                if request := await self._get_next_request():
+                    await self._crawl(request)
+                
+                # 检查退出条件
+                if await self._should_exit():
+                    break
+                
+                # 短暂休息避免忙等
+                await asyncio.sleep(0.001)
+        
+        finally:
+            # 清理生成任务
+            if generation_task and not generation_task.done():
+                generation_task.cancel()
+                try:
+                    await generation_task
+                except asyncio.CancelledError:
+                    pass
+            
+            await self.close_spider()
+
+    async def _traditional_request_generation(self):
+        """传统的请求生成方式（兼容旧版本）"""
+        while self.running:
+            try:
+                start_request = next(self.start_requests)
+                # 请求入队
+                await self.enqueue_request(start_request)
+            except StopIteration:
+                self.start_requests = None
+                break
+            except Exception as exp:
+                # 1、发去请求的request全部运行完毕
+                # 2、调度器是否空闲
+                # 3、下载器是否空闲
+                if not await self._exit():
+                    continue
+                self.running = False
+                if self.start_requests is not None:
+                    self.logger.error(f"启动请求时发生错误: {str(exp)}")
+            await asyncio.sleep(0.001)
+
+    async def _controlled_request_generation(self):
+        """受控的请求生成（增强功能）"""
+        self.logger.info("🎛️ 启动受控请求生成")
+        
+        batch = []
+        total_generated = 0
+        
+        try:
+            for request in self.start_requests:
+                batch.append(request)
+                
+                # 批量处理
+                if len(batch) >= self.generation_batch_size:
+                    generated = await self._process_generation_batch(batch)
+                    total_generated += generated
+                    batch = []
+                
+                # 背压检查
+                if await self._should_pause_generation():
+                    await self._wait_for_capacity()
+            
+            # 处理剩余请求
+            if batch:
+                generated = await self._process_generation_batch(batch)
+                total_generated += generated
+        
+        except Exception as e:
+            self.logger.error(f"❌ 请求生成失败: {e}")
+        
+        finally:
+            self.start_requests = None
+            self.logger.info(f"🎉 请求生成完成，总计: {total_generated}")
+
+    async def _process_generation_batch(self, batch) -> int:
+        """处理一批请求"""
+        generated = 0
+        
+        for request in batch:
+            if not self.running:
+                break
+            
+            # 等待队列有空间
+            while await self._is_queue_full() and self.running:
+                await asyncio.sleep(0.1)
+            
+            if self.running:
+                await self.enqueue_request(request)
+                generated += 1
+                self._generation_stats['total_generated'] += 1
+            
+            # 控制生成速度
+            if self.generation_interval > 0:
+                await asyncio.sleep(self.generation_interval)
+        
+        return generated
+
+    async def _should_pause_generation(self) -> bool:
+        """判断是否应该暂停生成"""
+        # 检查队列大小
+        if await self._is_queue_full():
+            return True
+        
+        # 检查任务管理器负载
+        if self.task_manager:
+            current_tasks = len(self.task_manager.current_task)
+            if hasattr(self.task_manager, 'semaphore'):
+                max_concurrency = getattr(self.task_manager.semaphore, '_initial_value', 8)
+                if current_tasks >= max_concurrency * self.backpressure_ratio:
+                    return True
+        
+        return False
+
+    async def _is_queue_full(self) -> bool:
+        """检查队列是否已满"""
+        if not self.scheduler:
+            return False
+        
+        queue_size = len(self.scheduler)
+        return queue_size >= self.max_queue_size * self.backpressure_ratio
+
+    async def _wait_for_capacity(self):
+        """等待系统有足够容量"""
+        self._generation_stats['backpressure_events'] += 1
+        self.logger.debug("⏸️ 触发背压，暂停请求生成")
+        
+        wait_time = 0.1
+        max_wait = 2.0
+        
+        while await self._should_pause_generation() and self.running:
+            await asyncio.sleep(wait_time)
+            wait_time = min(wait_time * 1.1, max_wait)
+
+    async def _open_spider(self):
+        asyncio.create_task(self.crawler.subscriber.notify(spider_opened))
+        crawling = asyncio.create_task(self.crawl())
+        await crawling
+
+    async def _crawl(self, request):
+        # TODO 实现并发
+        async def crawl_task():
+            outputs = await self._fetch(request)
+            # TODO 处理output
+            if outputs:
+                await self._handle_spider_output(outputs)
+
+        # 使用异步任务创建，遵守并发限制
+        await self.task_manager.create_task(crawl_task())
+
+    async def _fetch(self, request):
+        async def _successful(_response):
+            callback: Callable = request.callback or self.spider.parse
+            if _outputs := callback(_response):
+                if iscoroutine(_outputs):
+                    await _outputs
+                else:
+                    return transform(_outputs, _response)
+
+        _response = await self.downloader.fetch(request)
+        if _response is None:
+            return None
+        output = await _successful(_response)
+        return output
+
+    async def enqueue_request(self, start_request):
+        await self._schedule_request(start_request)
+
+    async def _schedule_request(self, request):
+        # TODO 去重
+        if await self.scheduler.enqueue_request(request):
+            asyncio.create_task(self.crawler.subscriber.notify(request_scheduled, request, self.crawler.spider))
+
+    async def _get_next_request(self):
+        return await self.scheduler.next_request()
+
+    async def _handle_spider_output(self, outputs):
+        async for spider_output in outputs:
+            if isinstance(spider_output, (Request, Item)):
+                await self.processor.enqueue(spider_output)
+            elif isinstance(spider_output, Exception):
+                asyncio.create_task(
+                    self.crawler.subscriber.notify(spider_error, spider_output, self.spider)
+                )
+                raise spider_output
+            else:
+                raise OutputError(f'{type(self.spider)} must return `Request` or `Item`.')
+
+    async def _exit(self):
+        if self.scheduler.idle() and self.downloader.idle() and self.task_manager.all_done() and self.processor.idle():
+            return True
+        return False
+
+    async def _should_exit(self) -> bool:
+        """检查是否应该退出（增强版本）"""
+        # 没有启动请求，且所有队列都空闲
+        if (self.start_requests is None and 
+            self.scheduler.idle() and 
+            self.downloader.idle() and 
+            self.task_manager.all_done() and 
+            self.processor.idle()):
+            return True
+        
+        return False
+
+    async def close_spider(self):
+        await asyncio.gather(*self.task_manager.current_task)
+        await self.scheduler.close()
+        await self.downloader.close()
+        if self.normal:
+            await self.crawler.close()
+    
+    def get_generation_stats(self) -> dict:
+        """获取生成统计"""
+        return {
+            **self._generation_stats,
+            'queue_size': len(self.scheduler) if self.scheduler else 0,
+            'active_tasks': len(self.task_manager.current_task) if self.task_manager else 0
+        }

crawlo/core/processor.py

@@ -1,40 +1,40 @@
-#!/usr/bin/python
-# -*- coding:UTF-8 -*-
-from asyncio import Queue
-from typing import Union, Optional
-
-from crawlo import Request, Item
-from crawlo.pipelines.pipeline_manager import PipelineManager
-
-
-class Processor(object):
-
-    def __init__(self, crawler):
-        self.crawler = crawler
-        self.queue: Queue = Queue()
-        self.pipelines: Optional[PipelineManager] = None
-
-    def open(self):
-        self.pipelines = PipelineManager.from_crawler(self.crawler)
-
-    async def process(self):
-        while not self.idle():
-            result = await self.queue.get()
-            if isinstance(result, Request):
-                await self.crawler.engine.enqueue_request(result)
-            else:
-                assert isinstance(result, Item)
-                await self._process_item(result)
-
-    async def _process_item(self, item):
-        await self.pipelines.process_item(item=item)
-
-    async def enqueue(self, output: Union[Request, Item]):
-        await self.queue.put(output)
-        await self.process()
-
-    def idle(self) -> bool:
-        return len(self) == 0
-
-    def __len__(self):
-        return self.queue.qsize()
+#!/usr/bin/python
+# -*- coding:UTF-8 -*-
+from asyncio import Queue
+from typing import Union, Optional
+
+from crawlo import Request, Item
+from crawlo.pipelines.pipeline_manager import PipelineManager
+
+
+class Processor(object):
+
+    def __init__(self, crawler):
+        self.crawler = crawler
+        self.queue: Queue = Queue()
+        self.pipelines: Optional[PipelineManager] = None
+
+    def open(self):
+        self.pipelines = PipelineManager.from_crawler(self.crawler)
+
+    async def process(self):
+        while not self.idle():
+            result = await self.queue.get()
+            if isinstance(result, Request):
+                await self.crawler.engine.enqueue_request(result)
+            else:
+                assert isinstance(result, Item)
+                await self._process_item(result)
+
+    async def _process_item(self, item):
+        await self.pipelines.process_item(item=item)
+
+    async def enqueue(self, output: Union[Request, Item]):
+        await self.queue.put(output)
+        await self.process()
+
+    def idle(self) -> bool:
+        return len(self) == 0
+
+    def __len__(self):
+        return self.queue.qsize()