Пример #1
0
 def parse_file(self, url, page_file):
     try:
         self.children = set()
         parser = PageParser(self, url)
         parser.parse_links(page_file)
     finally:
         page_file.close()
Пример #2
0
 def parse(self, url, url_id):
     print("parsing " + url + "...", file=sys.stderr)
     volume_id = self.get_volume_id(url_id)
     reader = self.open_page(url_id, volume_id)
     if reader:
         try:
             parser = PageParser(self, url)
             parser.parse_links(reader)
         finally:
             reader.close()
Пример #3
0
    def parse(self, url_id, url, volume_id):
        print("parsing " + url + "...", file=sys.stderr)
        reader = self.open_page(url_id, volume_id)
        if reader:
            try:
                parser = PageParser(self, url)
                parser.parse_links(reader)
            finally:
                reader.close()

        self.cur.execute(
            """update field
set parsed=localtimestamp
where id=%s""", (url_id, ))