generic_getLinks() decodes HTML entities

openzim · Nov 14, 2023 · 8542b7d · 8542b7d
1 parent d4a0c13
commit 8542b7d
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 6 deletions.
diff --git a/src/tools.cpp b/src/tools.cpp
@@ -357,7 +357,7 @@ std::vector<html_link> generic_getLinks(const std::string& page)
         while(*p != delimiter)
             p++;
         const std::string link(linkStart, p);
-        links.push_back(html_link(attr, link));
+        links.push_back(html_link(attr, decodeHtmlEntities(link)));
         p += 1;
     }
     return links;

diff --git a/test/tools-test.cpp b/test/tools-test.cpp
@@ -332,7 +332,6 @@ TEST(tools, getLinks)
       "{ src, https://fonts.io/css?family=OpenSans }"
     );
 
-    // Known issue - HTML entities are not decoded
     EXPECT_LINKS(
       R"(<a href="/R&amp;D">Research and development</a>
          blablabla
@@ -341,10 +340,10 @@ TEST(tools, getLinks)
          <a href="/Presidents/Dwight_&quot;Ike&quot;_Eisenhower">#34</a>
          <img src="https://example.com/getlogo?w=640&amp;h=480">
       )",
-      "{ href, /R&amp;D }"                                        "\n"
-      "{ href, ../syntax/&lt;script&gt; }"                        "\n"
-      "{ href, /Presidents/Dwight_&quot;Ike&quot;_Eisenhower }"   "\n"
-      "{ src, https://example.com/getlogo?w=640&amp;h=480 }"
+      "{ href, /R&D }"                                    "\n"
+      "{ href, ../syntax/<script> }"                      "\n"
+      "{ href, /Presidents/Dwight_\"Ike\"_Eisenhower }"   "\n"
+      "{ src, https://example.com/getlogo?w=640&h=480 }"
     );
 
     // Known issue - HTML is not parsed and therefore false links