def __init__(self, conf_file):
self.conf_file = conf_file
self.parser = ConfigParser.ConfigParser()
try:
self.parser.read(conf_file)
except Exception as e:
LOG.warn("parse config file %s failed." % (conf_file))
raise
def fetch_page(self, url):
page = None
try:
fd = urllib.urlopen(url)
page = fd.read()
fd.close()
return page
except IOError as e:
LOG.warn("open url:%s failed." % (url))
return page
def _aging_check(self):
while True:
se_del_list = []
for key, stat in self.stats.iteritems():
timer_count = self.stats[key][self.SE_STATS_TIMER]
if timer_count == 0:
se_del_list.append(key)
continue
timer_count = timer_count - 1
self.stats[key][self.SE_STATS_TIMER] = timer_count
for l in se_del_list:
self._aging_timer_fire(key)
self.stats.pop(key)
LOG.warn("search engine %s aging!" % (key))
time.sleep(self.SE_AGING_TIMER_INTERVAL)
def _search_negative_word(self, user, link):
page = self.fetch_page(link)
if not page:
return
try:
self.html_parser.reset_parser()
self.html_parser.feed(page)
except:
LOG.warn("parse link:%s failed." % (link))
return
url = link
if self.html_parser.redirect_url:
page = self.fetch_page(self.html_parser.redirect_url)
url = self.html_parser.redirect_url
for negative_word in user['negative_word']:
self._search_one_negative_word(user, page, negative_word, url)