Merge pull request #31 from filiph/fragment-fix

Fix checking of fragments with non-ASCII chars
filiph · Nov 16, 2018 · d50922f · d50922f
2 parents 52807c4 + c1826bd
commit d50922f
Show file tree

Hide file tree

Showing 5 changed files with 53 additions and 2 deletions.
diff --git a/lib/src/destination.dart b/lib/src/destination.dart
@@ -2,6 +2,8 @@ library linkcheck.destination;
 
 import 'dart:io' show ContentType, HttpClientResponse, RedirectInfo;
 
+import 'package:linkcheck/src/parsers/html.dart';
+
 /// RegExp for detecting URI scheme, such as `http:`, `mailto:`, etc.
 final _scheme = new RegExp(r"$(\w[\w\-]*\w):");
 
@@ -222,7 +224,7 @@ class Destination {
   bool satisfiesFragment(String fragment) {
     if (fragment == null || fragment == '') return true;
     if (anchors == null) return false;
-    return anchors.contains(Uri.decodeComponent(fragment));
+    return anchors.contains(normalizeAnchor(fragment));
   }
 
   Map<String, Object> toMap() => {

diff --git a/lib/src/parsers/html.dart b/lib/src/parsers/html.dart
@@ -67,6 +67,25 @@ Link extractLink(
   return new Link(origin, destination, destinationUri.fragment);
 }
 
+/// Takes an anchor (`id` or `name` attribute of an HTML element, or
+/// a fragment of a link) and normalizes it.
+///
+/// Anchors that can be percent-decoded, will. ("Hr%C3%A1%C4%8Dek" will
+/// become "Hráček".) Others will be kept the same. ("Hráček" will stay
+/// "Hráček".)
+String normalizeAnchor(String anchor) {
+  String decoded;
+  try {
+    decoded = Uri.decodeComponent(anchor);
+  } on ArgumentError {
+    // TODO: Report or handle ids and attributes that are not
+    //       percent-decodable (they were not percent-encoded and they
+    //       contain an invalid character.
+    decoded = anchor;
+  }
+  return decoded;
+}
+
 FetchResults parseHtml(String content, Uri uri, Destination current,
     DestinationResult checked, bool ignoreLinks) {
   var doc = parse(content, generateSpans: true, sourceUrl: uri.toString());
@@ -79,7 +98,7 @@ FetchResults parseHtml(String content, Uri uri, Destination current,
   var anchors = doc
       .querySelectorAll("body [id], body [name]")
       .map((element) => element.attributes["id"] ?? element.attributes["name"])
-      .map((fragment) => Uri.decodeComponent(fragment))
+      .map(normalizeAnchor)
       .toList();
   checked.anchors = anchors;
 

diff --git a/test/case13/index.html b/test/case13/index.html
@@ -0,0 +1,10 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <title></title>
+</head>
+<body>
+    <a href="with-non-percent-encoded-anchor.html">APL</a>
+</body>
+</html>
diff --git a/test/case13/with-non-percent-encoded-anchor.html b/test/case13/with-non-percent-encoded-anchor.html
@@ -0,0 +1,10 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <title></title>
+</head>
+<body>
+    <h1 id="cite_ref-Berättar_104-0">RMS Berättar</h1>
+</body>
+</html>
diff --git a/test/e2e_test.dart b/test/e2e_test.dart
@@ -186,6 +186,16 @@ void main() {
         await server.destroy();
       }
     });
+
+    test("fragment checking works with non-percent-encoded anchors", () async {
+      var server = await Dhttpd.start(path: getServingPath(13), port: port);
+      try {
+        int result = await run([":$port"], out);
+        expect(result, 0);
+      } finally {
+        await server.destroy();
+      }
+    });
   }, tags: ["integration"]);
 }