FIX: localization of outgoing URIs in JS corrupts response (#374)

turquoiseowl · Jul 10, 2018 · d085eb2 · d085eb2
1 parent c82d747
commit d085eb2
Show file tree

Hide file tree

Showing 2 changed files with 23 additions and 1 deletion.
diff --git a/src/i18n.Tests/Tests/ResponseFilterTests.cs b/src/i18n.Tests/Tests/ResponseFilterTests.cs
@@ -50,6 +50,12 @@ public void ResponseFilter_can_patch_html_urls()
             Helper_ResponseFilter_can_patch_html_urls("fr", "<img src=\"http://example.com/content/fred.txt\"></img>", "<img src=\"http://example.com/fr/content/fred.txt\"></img>" , "http://example.com/blog/");
             Helper_ResponseFilter_can_patch_html_urls("fr", "<img src=\"http://other.com/content/fred.txt\"></img>"  , "<img src=\"http://other.com/content/fred.txt\"></img>"      , "http://example.com/blog/"); // NB: foreign site so no langtag added
 
+            // One attribute - empty url
+            Helper_ResponseFilter_can_patch_html_urls(
+                "fr",
+                "<a href=\"\"></a>",
+                "<a href=\"\"></a>");
+
             // One attribute.
             Helper_ResponseFilter_can_patch_html_urls(
                 "fr",
@@ -396,6 +402,22 @@ public void ResponseFilter_can_patch_html_urls()
                 "fr",
                 string.Format("<script src=\"{0}https://example.com/fr/123\"></script>", EarlyUrlLocalizer.IgnoreLocalizationUrlPrefix),
                 "<script src=\"https://example.com/fr/123\"></script>");
+
+
+            string strMultilineScriptWithHref = @"<script>
+  try {
+    this._baseHref="""";
+  }
+  catch (e) {
+  }
+</script>
+<div class=""page_style_a"">";
+
+            Helper_ResponseFilter_can_patch_html_urls(
+                "fr",
+                strMultilineScriptWithHref,
+                strMultilineScriptWithHref);
+
         }
     }
 }
diff --git a/src/i18n/Concrete/EarlyUrlLocalizer.cs b/src/i18n/Concrete/EarlyUrlLocalizer.cs
@@ -157,7 +157,7 @@ public string ProcessOutgoing(
         /// Regex for finding and replacing urls in html.
         /// </summary>
         public static Regex m_regexHtmlUrls = new Regex(
-            "(?<pre><(?:script|img|a|area|link|base|input|frame|iframe|form)\\b.*?(?:src|href|action)\\s*=\\s*[\"']\\s*)(?<url>.+?)(?<post>\\s*[\"'][^>]*?>)",
+            "(?<pre><(?:script|img|a|area|link|base|input|frame|iframe|form)\\b[^>]*?\\b(?:src|href|action)\\s*=\\s*[\"']\\s*)(?<url>.+?)(?<post>\\s*[\"'][^>]*?>)",
             RegexOptions.CultureInvariant | RegexOptions.IgnoreCase | RegexOptions.Singleline);
                 // The above supports most common ways for a URI to appear in HTML/XHTML.
                 // Note that if we fail to catch a URL here, it is not fatal; it only means we don't avoid a redirect