urllib.py 文件源码

def install_patches():
    if six.PY3:
        # The old urllib does not exist in Py3, so delegate to urllib2 patcher
        from . import urllib2
        urllib2.install_patches()
        return

    import urllib
    import urlparse

    log.info('Instrumenting urllib methods for tracing')

    class TracedURLOpener(urllib.FancyURLopener):

        def open(self, fullurl, data=None):
            parsed_url = urlparse.urlparse(fullurl)
            host = parsed_url.hostname or None
            port = parsed_url.port or None

            span = utils.start_child_span(
                operation_name='urllib', parent=get_current_span())

            span.set_tag(ext_tags.SPAN_KIND, ext_tags.SPAN_KIND_RPC_CLIENT)

            # use span as context manager so that its finish() method is called
            with span:
                span.set_tag(ext_tags.HTTP_URL, fullurl)
                if host:
                    span.set_tag(ext_tags.PEER_HOST_IPV4, host)
                if port:
                    span.set_tag(ext_tags.PEER_PORT, port)
                # TODO add callee service name
                # TODO add headers to propagate trace
                # cannot use super here, this is an old style class
                fileobj = urllib.FancyURLopener.open(self, fullurl, data)
                if fileobj.getcode() is not None:
                    span.set_tag('http.status_code', fileobj.getcode())

            return fileobj

        def retrieve(self, url, filename=None, reporthook=None, data=None):
            raise NotImplementedError

    urllib._urlopener = TracedURLOpener()