A simple spider frame written by Python, which needs Python3.5+
- Support two crawlling mode: multi-threading mode and asyncio mode (using aiohttp)
Support distributed crawlliing mode (not finished)- Define some utility functions and classes, for example: UrlFilter, make_random_useragent, etc
- Fewer lines of code, easyer to read, understand and expand
- utilities module: define utilities functions and classes for spider
- instances module: define classes of fetcher, parser, saver for multi-threading spider
- concurrent module: define WebSpiderFrame of multi-threading mode and asyncio mode
- distributed module: define WebSpiderFrame of distributed mode (not finished)
-
procedure of multi-threading spider
①: Fetcher gets url from UrlQueue, and makes request based on this url
②: Put the result of ① to HtmlQueue, and make Parser can get it
③: Parser gets item from HtmlQueue, and parses it to get new urls and saved items
④: Put the new urls to UrlQueue, and make Fetcher can get it
⑤: Put the saved items to ItemQueue, and make Saver can get it
⑥: Saver gets item from ItemQueue, and saves it to filesystem or database -
procedure of asyncio spider
Similar with multi-threading spider. The only difference is using "coroutine" instead of "multi-threads"
Installation: you'd better use the first method
(1)Copy the "spider" directory to your project directory, and import spider
(2)Install spider to your python system using python3 setup.py install
Getting multi-threading spider started: make a demo crawling Douban Movies
(1)Import spider import spider
(2)Make a new class of MovieFetcher based on spider.Fetcher, and rewrite functions of url_fetch
class MovieFetcher(spider.Fetcher):
def __init__(self, max_repeat=3, sleep_time=0):
spider.Fetcher.__init__(self, max_repeat=max_repeat, sleep_time=sleep_time)
self.session = requests.Session()
return
def url_fetch(self, url, keys, repeat):
resp = self.session.get(url, allow_redirects=False, verify=False, timeout=5)
if resp.status_code == 200:
return 1, resp.text
resp.raise_for_status()
(3)Make a new class of MovieParser based on spider.Parser, and rewrite functions of htm_parse
class MovieParser(spider.Parser):
def htm_parse(self, priority, url, keys, deep, content):
url_list, save_list = [], []
soup = BeautifulSoup(content, "html5lib")
# decide how to parse the content of url, based on 'keys'
if keys[0] == "index":
# parse the index page and get new urls
# get the new movie urls of detail page
div_movies = soup.find_all("a", class_="nbg", title=True)
url_list.extend([(item.get("href"), ("detail", keys[1]), 0) for item in div_movies])
# get next index page
next_page = soup.find("span", class_="next")
if next_page:
next_page_a = next_page.find("a")
if next_page_a:
url_list.append((next_page_a.get("href"), ("index", keys[1]), 1))
else:
# parse the detail page and get saved items
content = soup.find("div", id="content")
......
save_list.append(movie)
return 1, url_list, save_list
(4)Create the multi-theading spider, set the start url and start this spider using 20 threads
# initial the WebSpider
dou_spider = spider.WebSpider(MovieFetcher(), MovieParser(max_deep=-1), spider.Saver(), spider.UrlFilter())
# set the start url
dou_spider.set_start_url("https://movie.douban.com/tag/%E7%88%B1%E6%83%85", ("index",), priority=1)
# start this spider and wait for finishing
dou_spider.start_work_and_wait_done(fetcher_num=20)
Getting asyncio spider started: make a demo crawling zhushou.360
# get loop
loop = asyncio.get_event_loop()
# initial fetcher / parser / saver, you also can rewrite this three class
fetcher = spider.FetcherAsync(max_repeat=3, sleep_time=0, loop=loop)
parser = spider.ParserAsync(max_deep=1)
saver = spider.SaverAsync(save_pipe=open("out_spider_thread.txt", "w"))
# initial the WebSpiderAsync
web_spider_async = spider.WebSpiderAsync(fetcher, parser, saver, url_filter=spider.UrlFilter(), loop=loop)
# set the start url
web_spider_async.set_start_url("http://zhushou.360.cn/")
# start this spider and wait for finishing
web_spider_async.start_work_and_wait_done(fetcher_num=20)
- demos_yundama
- demos_nbastats
- demos_doubanmovies
- demos_dangdang