Ignore text fragment directive in link comparison

close #811
w3c · Oct 31, 2024 · 8bd472b · 8bd472b
1 parent 3e5b0ab
commit 8bd472b
Show file tree

Hide file tree

Showing 2 changed files with 17 additions and 7 deletions.
diff --git a/src/lib/study-backrefs.js b/src/lib/study-backrefs.js
@@ -396,7 +396,14 @@ function studyLinks(spec, links, report, edResults, trResults, htmlFragments) {
 
       // Check anchors
       const anchors = links[link].anchors || [];
-      for (const anchor of anchors) {
+      for (let anchor of anchors) {
+	if (anchor.startsWith(':~:text=')) {
+          // links using only text fragments are inherently fragile
+          recordAnomaly(spec, 'frailLinks', link + '#' + anchor);
+	  continue;
+        }
+	// We remove the fragment directive for comparison
+	anchor = anchor.split(':~:')[0];
         const baseLink = (sourceSpec.nightly?.url === link || sourceSpec.nightly?.pages?.includes(link)) ? link : sourceSpec.nightly?.url;
         const matchFullNightlyLink = matchAnchor(baseLink, anchor);
         const matchFullReleaseLink = matchAnchor((sourceSpec.release || sourceSpec.nightly).url, anchor);
@@ -426,10 +433,6 @@ function studyLinks(spec, links, report, edResults, trResults, htmlFragments) {
               // the multipage version of HTML
               recordAnomaly(spec, 'frailLinks', link + '#' + anchor);
             }
-            else if (anchor.startsWith(':~:text=')) {
-              // links using text fragments are inherently fragile
-              recordAnomaly(spec, 'frailLinks', link + '#' + anchor);
-            }
             else {
               recordAnomaly(spec, 'brokenLinks', link + '#' + anchor);
             }
@@ -445,4 +448,4 @@ function studyLinks(spec, links, report, edResults, trResults, htmlFragments) {
     });
 }
 
-export default studyBackrefs;
+export default studyBackrefs;
diff --git a/test/study-backrefs.js b/test/study-backrefs.js
@@ -67,7 +67,7 @@ describe('The links analyser', () => {
   });
 
   it('reports a broken link', async () => {
-    const ids = ['validid'];
+    const ids = ['invalidid'];
     const crawlResult = toCrawlResults([], ids);
     const report = await study(crawlResult.ed, { htmlFragments: {} });
     assertNbAnomalies(report, 1);
@@ -78,6 +78,13 @@ describe('The links analyser', () => {
     });
   });
 
+
+  it('removes fragment directives when checking for broken links', async () => {
+    const crawlResult = toCrawlResults(['validid'], ['validid:~:text=foo']);
+    const report = await study(crawlResult.ed, { htmlFragments: {} });
+    assertNbAnomalies(report, 0);
+  });
+
   /* TOTEST
     "datedUrls",
     "evolvingLinks",