import urllib, re, sys p_key = re.compile ('http: \ / \ / lib \ .aldebaran \ .ru \ / getpage \ / 1 \ /.*? "') p_span = re.compile ('<span class = h> (. *?) <\ / span>') p_s = re.compile ('\'; s \ + = \ '') p_p = re.compile ('(<p>. *? <\ / p>)') p_url_name = re.compile ('^ (http: \ / \ / lib.aldebaran.ru \ / author \ /.*? __)') def getpage (url): try: key = p_key.findall (urllib.urlopen (url) .read ()) [0] [- 33: -1] except: return none urllib.urlopen ("http://lib.aldebaran.ru/getpage/1/" + key) .read () page = urllib.urlopen ("http://lib.aldebaran.ru/getpage/2/" + key) .read () page = p_span.sub ('', page) page = '' .join (p_p.findall (page)) page = p_s.sub ('', page) return page url = sys.argv [1] url = p_url_name.findall (url) [0] i = 1 while true: page = getpage (url + str (i) + '.html') i + = 1 if page! = None: print page else: sys.exit (0)
Source: https://habr.com/ru/post/21082/
All Articles