Source code for


    The CheckExternalLinksBuilder class.

    :copyright: Copyright 2007-2020 by the Sphinx team, see AUTHORS.
    :license: BSD, see LICENSE for details.

import json
import queue
import re
import socket
import threading
from html.parser import HTMLParser
from os import path
from typing import Any, Dict, List, Set, Tuple
from urllib.parse import unquote, urlparse

from docutils import nodes
from docutils.nodes import Node
from requests.exceptions import HTTPError

from sphinx.application import Sphinx
from import Builder
from sphinx.locale import __
from sphinx.util import encode_uri, logging, requests
from sphinx.util.console import darkgray, darkgreen, purple, red, turquoise  # type: ignore
from sphinx.util.nodes import get_node_line

logger = logging.getLogger(__name__)

uri_re = re.compile('([a-z]+:)?//')  # matches to foo:// and // (a protocol relative URL)

    'Accept': 'text/html,application/xhtml+xml;q=0.9,*/*;q=0.8',

class AnchorCheckParser(HTMLParser):
    """Specialized HTML parser that looks for a specific anchor."""

    def __init__(self, search_anchor: str) -> None:

        self.search_anchor = search_anchor
        self.found = False

    def handle_starttag(self, tag: Any, attrs: Any) -> None:
        for key, value in attrs:
            if key in ('id', 'name') and value == self.search_anchor:
                self.found = True

def check_anchor(response: requests.requests.Response, anchor: str) -> bool:
    """Reads HTML data from a response object `response` searching for `anchor`.
    Returns True if anchor was found, False otherwise.
    parser = AnchorCheckParser(anchor)
    # Read file in chunks. If we find a matching anchor, we break
    # the loop early in hopes not to have to download the whole thing.
    for chunk in response.iter_content(chunk_size=4096, decode_unicode=True):
        if isinstance(chunk, bytes):    # requests failed to decode
            chunk = chunk.decode()      # manually try to decode it

        if parser.found:
    return parser.found

[docs]class CheckExternalLinksBuilder(Builder): """ Checks for broken external links. """ name = 'linkcheck' epilog = __('Look for any errors in the above output or in ' '%(outdir)s/output.txt') def init(self) -> None: self.to_ignore = [re.compile(x) for x in] self.anchors_ignore = [re.compile(x) for x in] self.auth = [(re.compile(pattern), auth_info) for pattern, auth_info in] self.good = set() # type: Set[str] self.broken = {} # type: Dict[str, str] self.redirected = {} # type: Dict[str, Tuple[str, int]] # set a timeout for non-responding servers socket.setdefaulttimeout(5.0) # create output file open(path.join(self.outdir, 'output.txt'), 'w').close() # create JSON output file open(path.join(self.outdir, 'output.json'), 'w').close() # create queues and worker threads self.wqueue = queue.Queue() # type: queue.Queue self.rqueue = queue.Queue() # type: queue.Queue self.workers = [] # type: List[threading.Thread] for i in range( thread = threading.Thread(target=self.check_thread, daemon=True) thread.start() self.workers.append(thread) def check_thread(self) -> None: kwargs = {} if kwargs['timeout'] = def get_request_headers() -> Dict: url = urlparse(uri) candidates = ["%s://%s" % (url.scheme, url.netloc), "%s://%s/" % (url.scheme, url.netloc), uri, "*"] for u in candidates: if u in self.config.linkcheck_request_headers: headers = dict(DEFAULT_REQUEST_HEADERS) headers.update(self.config.linkcheck_request_headers[u]) return headers return {} def check_uri() -> Tuple[str, str, int]: # split off anchor if '#' in uri: req_url, anchor = uri.split('#', 1) for rex in self.anchors_ignore: if rex.match(anchor): anchor = None break else: req_url = uri anchor = None # handle non-ASCII URIs try: req_url.encode('ascii') except UnicodeError: req_url = encode_uri(req_url) # Get auth info, if any for pattern, auth_info in self.auth: if pattern.match(uri): break else: auth_info = None # update request headers for the URL kwargs['headers'] = get_request_headers() try: if anchor and # Read the whole document and see if #anchor exists response = requests.get(req_url, stream=True,, auth=auth_info, **kwargs) response.raise_for_status() found = check_anchor(response, unquote(anchor)) if not found: raise Exception(__("Anchor '%s' not found") % anchor) else: try: # try a HEAD request first, which should be easier on # the server and the network response = requests.head(req_url, allow_redirects=True,, auth=auth_info, **kwargs) response.raise_for_status() except HTTPError: # retry with GET request if that fails, some servers # don't like HEAD requests. response = requests.get(req_url, stream=True,, auth=auth_info, **kwargs) response.raise_for_status() except HTTPError as err: if err.response.status_code == 401: # We'll take "Unauthorized" as working. return 'working', ' - unauthorized', 0 elif err.response.status_code == 503: # We'll take "Service Unavailable" as ignored. return 'ignored', str(err), 0 else: return 'broken', str(err), 0 except Exception as err: return 'broken', str(err), 0 if response.url.rstrip('/') == req_url.rstrip('/'): return 'working', '', 0 else: new_url = response.url if anchor: new_url += '#' + anchor # history contains any redirects, get last if response.history: code = response.history[-1].status_code return 'redirected', new_url, code else: return 'redirected', new_url, 0 def check(docname: str) -> Tuple[str, str, int]: # check for various conditions without bothering the network if len(uri) == 0 or uri.startswith(('#', 'mailto:', 'tel:')): return 'unchecked', '', 0 elif not uri.startswith(('http:', 'https:')): if uri_re.match(uri): # non supported URI schemes (ex. ftp) return 'unchecked', '', 0 else: srcdir = path.dirname(self.env.doc2path(docname)) if path.exists(path.join(srcdir, uri)): return 'working', '', 0 else: for rex in self.to_ignore: if rex.match(uri): return 'ignored', '', 0 else: self.broken[uri] = '' return 'broken', '', 0 elif uri in self.good: return 'working', 'old', 0 elif uri in self.broken: return 'broken', self.broken[uri], 0 elif uri in self.redirected: return 'redirected', self.redirected[uri][0], self.redirected[uri][1] for rex in self.to_ignore: if rex.match(uri): return 'ignored', '', 0 # need to actually check the URI for _ in range( status, info, code = check_uri() if status != "broken": break if status == "working": self.good.add(uri) elif status == "broken": self.broken[uri] = info elif status == "redirected": self.redirected[uri] = (info, code) return (status, info, code) while True: uri, docname, lineno = self.wqueue.get() if uri is None: break status, info, code = check(docname) self.rqueue.put((uri, docname, lineno, status, info, code)) def process_result(self, result: Tuple[str, str, int, str, str, int]) -> None: uri, docname, lineno, status, info, code = result filename = self.env.doc2path(docname, None) linkstat = dict(filename=filename, lineno=lineno, status=status, code=code, uri=uri, info=info) if status == 'unchecked': self.write_linkstat(linkstat) return if status == 'working' and info == 'old': self.write_linkstat(linkstat) return if lineno:'(line %4d) ', lineno, nonl=True) if status == 'ignored': if info:'-ignored- ') + uri + ': ' + info) else:'-ignored- ') + uri) self.write_linkstat(linkstat) elif status == 'local':'-local- ') + uri) self.write_entry('local', docname, filename, lineno, uri) self.write_linkstat(linkstat) elif status == 'working':'ok ') + uri + info) self.write_linkstat(linkstat) elif status == 'broken': if or logger.warning(__('broken link: %s (%s)'), uri, info, location=(filename, lineno)) else:'broken ') + uri + red(' - ' + info)) self.write_entry('broken', docname, filename, lineno, uri + ': ' + info) self.write_linkstat(linkstat) elif status == 'redirected': try: text, color = { 301: ('permanently', purple), 302: ('with Found', purple), 303: ('with See Other', purple), 307: ('temporarily', turquoise), 308: ('permanently', purple), }[code] except KeyError: text, color = ('with unknown code', purple) linkstat['text'] = text'redirect ') + uri + color(' - ' + text + ' to ' + info)) self.write_entry('redirected ' + text, docname, filename, lineno, uri + ' to ' + info) self.write_linkstat(linkstat) def get_target_uri(self, docname: str, typ: str = None) -> str: return '' def get_outdated_docs(self) -> Set[str]: return self.env.found_docs def prepare_writing(self, docnames: Set[str]) -> None: return def write_doc(self, docname: str, doctree: Node) -> None:'') n = 0 # reference nodes for refnode in doctree.traverse(nodes.reference): if 'refuri' not in refnode: continue uri = refnode['refuri'] lineno = get_node_line(refnode) self.wqueue.put((uri, docname, lineno), False) n += 1 # image nodes for imgnode in doctree.traverse(nodes.image): uri = imgnode['candidates'].get('?') if uri and '://' in uri: lineno = get_node_line(imgnode) self.wqueue.put((uri, docname, lineno), False) n += 1 done = 0 while done < n: self.process_result(self.rqueue.get()) done += 1 if self.broken: = 1 def write_entry(self, what: str, docname: str, filename: str, line: int, uri: str) -> None: with open(path.join(self.outdir, 'output.txt'), 'a') as output: output.write("%s:%s: [%s] %s\n" % (filename, line, what, uri)) def write_linkstat(self, data: dict) -> None: with open(path.join(self.outdir, 'output.json'), 'a') as output: output.write(json.dumps(data)) output.write('\n') def finish(self) -> None: for worker in self.workers: self.wqueue.put((None, None, None), False)
def setup(app: Sphinx) -> Dict[str, Any]: app.add_builder(CheckExternalLinksBuilder) app.add_config_value('linkcheck_ignore', [], None) app.add_config_value('linkcheck_auth', [], None) app.add_config_value('linkcheck_request_headers', {}, None) app.add_config_value('linkcheck_retries', 1, None) app.add_config_value('linkcheck_timeout', None, None, [int]) app.add_config_value('linkcheck_workers', 5, None) app.add_config_value('linkcheck_anchors', True, None) # Anchors starting with ! are ignored since they are # commonly used for dynamic pages app.add_config_value('linkcheck_anchors_ignore', ["^!"], None) return { 'version': 'builtin', 'parallel_read_safe': True, 'parallel_write_safe': True, }