improve sohu extractor

This commit is contained in:
huohuarong 2013-08-06 10:26:57 +08:00
parent b5a6d40818
commit d5b00ee6e0

View File

@ -31,6 +31,7 @@ def _real_extract(self, url):
compiled = re.compile(pattern, re.DOTALL) compiled = re.compile(pattern, re.DOTALL)
title = self._search_regex(compiled, webpage, u'video title') title = self._search_regex(compiled, webpage, u'video title')
title = clean_html(title).split('-')[0].strip() title = clean_html(title).split('-')[0].strip()
self.to_screen('Title: %s' % title)
pattern = re.compile(r'var vid="(\d+)"') pattern = re.compile(r'var vid="(\d+)"')
result = re.search(pattern, webpage) result = re.search(pattern, webpage)
if not result: if not result:
@ -70,6 +71,7 @@ def _real_extract(self, url):
base_url_3 = 'http://allot/?prot=prot&file=clipsURL[i]&new=su[i]' base_url_3 = 'http://allot/?prot=prot&file=clipsURL[i]&new=su[i]'
files_info = [] files_info = []
for i in range(num_of_parts): for i in range(num_of_parts):
self.to_screen('Geting json infomation of part %s/%s' % (i + 1, num_of_parts))
middle_url = 'http://%s/?prot=%s&file=%s&new=%s' % (allot, prot, clipsURL[i], su[i]) middle_url = 'http://%s/?prot=%s&file=%s&new=%s' % (allot, prot, clipsURL[i], su[i])
logging.info('middle url part %d: %s' % (i, middle_url)) logging.info('middle url part %d: %s' % (i, middle_url))
middle_info = urllib2.urlopen(middle_url).read().split('|') middle_info = urllib2.urlopen(middle_url).read().split('|')