recipe-224043.py 文件源码

def getlinkinfos(self):
        # File reading is done in __init__() routine.  Store parser in
        # local variable to indicate success of parsing.

        # If no parser was stored, fail.
        if not self.parser: return []

        rawlinks = self.parser.getlinks()
        base = urlparse.urljoin(self.url, self.parser.getbase() or "")
        infos = []
        for rawlink in rawlinks:
            t = urlparse.urlparse(rawlink)
            # DON'T DISCARD THE FRAGMENT! Instead, include
            # it in the tuples which are returned. See Checker.dopage().
            fragment = t[-1]
            t = t[:-1] + ('',)
            rawlink = urlparse.urlunparse(t)
            link = urlparse.urljoin(base, rawlink)
            infos.append((link, rawlink, fragment))     

        return infos