From a7eacd620011dbed3422a62e8c6f3fce9d1bf7df Mon Sep 17 00:00:00 2001
From: Chris Mayo <aklhfex@gmail.com>
Date: Mon, 27 Jul 2020 19:22:32 +0100
Subject: [PATCH 1/2] Add a test for a page with links to anchors

Query and fragment URL parts for filesystem URLs are ignored, therefore
test over http.
---
 tests/checker/data/http_anchor.html        |  4 ++++
 tests/checker/data/http_anchor.html.result | 17 +++++++++++++++++
 tests/checker/test_anchor.py               | 11 +++++++++++
 3 files changed, 32 insertions(+)
 create mode 100644 tests/checker/data/http_anchor.html
 create mode 100644 tests/checker/data/http_anchor.html.result

diff --git a/tests/checker/data/http_anchor.html b/tests/checker/data/http_anchor.html
new file mode 100644
index 00000000..8ba01be3
--- /dev/null
+++ b/tests/checker/data/http_anchor.html
@@ -0,0 +1,4 @@
+<p id="good">
+<a href="http_anchor.html">a</a>
+<a href="http_anchor.html#good">a_good</a>
+<a href="http_anchor.html#bad">a_bad</a>
diff --git a/tests/checker/data/http_anchor.html.result b/tests/checker/data/http_anchor.html.result
new file mode 100644
index 00000000..b9ebf0c7
--- /dev/null
+++ b/tests/checker/data/http_anchor.html.result
@@ -0,0 +1,17 @@
+url http://localhost:%(port)d/%(datadir)s/http_anchor.html
+cache key http://localhost:%(port)d/%(datadir)s/http_anchor.html
+real url http://localhost:%(port)d/%(datadir)s/http_anchor.html
+valid
+
+url http_anchor.html#bad
+cache key http://localhost:%(port)d/%(datadir)s/http_anchor.html#bad
+real url http://localhost:%(port)d/%(datadir)s/http_anchor.html#bad
+name a_bad
+warning Anchor `bad' not found. Available anchors: `good'.
+valid
+
+url http_anchor.html#good
+cache key http://localhost:%(port)d/%(datadir)s/http_anchor.html#good
+real url http://localhost:%(port)d/%(datadir)s/http_anchor.html#good
+name a_good
+valid
diff --git a/tests/checker/test_anchor.py b/tests/checker/test_anchor.py
index 8b980a15..7848e9e1 100644
--- a/tests/checker/test_anchor.py
+++ b/tests/checker/test_anchor.py
@@ -17,6 +17,7 @@
 Test html anchor parsing and checking.
 """
 from . import LinkCheckTest
+from .httpserver import HttpServerTest
 
 
 class TestAnchor(LinkCheckTest):
@@ -38,3 +39,13 @@ class TestAnchor(LinkCheckTest):
             "valid",
         ]
         self.direct(urlanchor, resultlines, confargs=confargs)
+
+
+class TestHttpAnchor(HttpServerTest):
+    """
+    Test checking of HTML pages containing links to anchors served over http.
+    """
+
+    def test_anchor_html(self):
+        confargs = dict(enabledplugins=["AnchorCheck"], recursionlevel=1)
+        self.file_test("http_anchor.html", confargs=confargs)

From 0912e8a2c1ccc0e1a901f27f4176ef9e0644f06a Mon Sep 17 00:00:00 2001
From: Chris Mayo <aklhfex@gmail.com>
Date: Mon, 27 Jul 2020 19:25:30 +0100
Subject: [PATCH 2/2] Don't strip the URL fragment from cache key if using
 AnchorCheck

Else once one URL for a page has been checked, URLs with different
fragments are skipped and not passed to AnchorCheck.

eaa538c ("don't check one url multiple times", 2016-11-09)
---
 linkcheck/checker/urlbase.py | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/linkcheck/checker/urlbase.py b/linkcheck/checker/urlbase.py
index daa3135c..500d7aba 100644
--- a/linkcheck/checker/urlbase.py
+++ b/linkcheck/checker/urlbase.py
@@ -363,11 +363,13 @@ class UrlBase:
 
     def set_cache_url(self):
         """Set the URL to be used for caching."""
-        # remove anchor from cached target url since we assume
-        # URLs with different anchors to have the same content
-        self.cache_url = urlutil.urlunsplit(self.urlparts[:4] + [''])
-        if self.cache_url is not None:
-            assert isinstance(self.cache_url, str), repr(self.cache_url)
+        if "AnchorCheck" in self.aggregate.config["enabledplugins"]:
+            self.cache_url = self.url
+        else:
+            # remove anchor from cached target url since we assume
+            # URLs with different anchors to have the same content
+            self.cache_url = urlutil.urlunsplit(self.urlparts[:4] + [''])
+        log.debug(LOG_CHECK, "cache_url '%s'", self.cache_url)
 
     def check_syntax(self):
         """