linkcheck: Store the original (unquoted) anchor (#12206)

sphinx/builders/linkcheck.py

@@ -13,7 +13,7 @@ from os import path
 from queue import PriorityQueue, Queue
 from threading import Thread
 from typing import TYPE_CHECKING, NamedTuple, cast
-from urllib.parse import unquote, urlparse, urlsplit, urlunparse
+from urllib.parse import quote, unquote, urlparse, urlsplit, urlunparse
 
 from docutils import nodes
 from requests.exceptions import ConnectionError, HTTPError, SSLError, TooManyRedirects
@@ -409,6 +409,7 @@ class HyperlinkAvailabilityCheckWorker(Thread):
                     if rex.match(req_url):
                         anchor = ''
                         break
+            anchor = unquote(anchor)
 
         # handle non-ASCII URIs
         try:
@@ -446,7 +447,7 @@ class HyperlinkAvailabilityCheckWorker(Thread):
                 ) as response:
                     if (self.check_anchors and response.ok and anchor
                             and not contains_anchor(response, anchor)):
-                        raise Exception(__(f'Anchor {anchor!r} not found'))
+                        raise Exception(__(f'Anchor {quote(anchor)!r} not found'))
 
                 # Copy data we need from the (closed) response
                 status_code = response.status_code
@@ -592,7 +593,7 @@ def _get_request_headers(
 
 def contains_anchor(response: Response, anchor: str) -> bool:
     """Determine if an anchor is contained within an HTTP response."""
-    parser = AnchorCheckParser(unquote(anchor))
+    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):

tests/roots/test-linkcheck-anchors-ignore-for-url/index.rst

@@ -1,5 +1,6 @@
 * `Example valid url, no anchor <http://localhost:7777/valid>`_
 * `Example valid url, valid anchor <http://localhost:7777/valid#valid-anchor>`_
+* `Example valid url, valid quotable anchor <http://localhost:7777/valid#py:module::urllib.parse>`_
 * `Example valid url, invalid anchor <http://localhost:7777/valid#invalid-anchor>`_
 * `Example ignored url, no anchor <http://localhost:7777/ignored>`_
 * `Example ignored url, invalid anchor <http://localhost:7777/ignored#invalid-anchor>`_

tests/test_builders/test_build_linkcheck.py

@@ -295,7 +295,7 @@ def test_anchors_ignored_for_url(app):
 
     attrs = ('filename', 'lineno', 'status', 'code', 'uri', 'info')
     data = [json.loads(x) for x in content.splitlines()]
-    assert len(data) == 7
+    assert len(data) == 8
     assert all(all(attr in row for attr in attrs) for row in data)
 
     # rows may be unsorted due to network latency or
@@ -304,6 +304,7 @@ def test_anchors_ignored_for_url(app):
 
     assert rows[f'http://{address}/valid']['status'] == 'working'
     assert rows[f'http://{address}/valid#valid-anchor']['status'] == 'working'
+    assert rows['http://localhost:7777/valid#py:module::urllib.parse']['status'] == 'broken'
     assert rows[f'http://{address}/valid#invalid-anchor'] == {
         'status': 'broken',
         'info': "Anchor 'invalid-anchor' not found",