From b34025654132ce4f1f5803f514be235bf452b5cb Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tobias=20Ho=CC=88=C3=9Fl?= <tobias@hoessl.eu>
Date: Thu, 11 May 2017 23:01:47 +0200
Subject: [PATCH] Detect broken HTML after diff (Fixed #3229)

---
 openslides/motions/static/js/motions/diff.js | 26 +++++++++++++++++++-
 tests/karma/motions/diff.service.test.js     | 13 ++++++++++
 2 files changed, 38 insertions(+), 1 deletion(-)

diff --git a/openslides/motions/static/js/motions/diff.js b/openslides/motions/static/js/motions/diff.js
index 4e436c320..42767c9f0 100644
--- a/openslides/motions/static/js/motions/diff.js
+++ b/openslides/motions/static/js/motions/diff.js
@@ -918,6 +918,16 @@ angular.module('OpenSlidesApp.motions.diff', ['OpenSlidesApp.motions.lineNumberi
                 return true;
             }
 
+            // Opening tags, followed by </del> or </ins>, indicate broken HTML (if it's not a <ins> / <del>)
+            var brokenRegexp = /<(\w+)[^>]*><\/(ins|del)>/gi,
+                result;
+            while ((result = brokenRegexp.exec(html)) !== null) {
+                if (result[1].toLowerCase() !== 'ins' && result[1].toLowerCase() !== 'del') {
+                    return true;
+                }
+            }
+
+
             // If other HTML tags are contained within INS/DEL (e.g. "<ins>Test</p></ins>"), let's better be cautious
             // The "!!(found=...)"-construction is only used to make jshint happy :)
             var findDel = /<del>(.*?)<\/del>/gi,
@@ -1090,13 +1100,27 @@ angular.module('OpenSlidesApp.motions.diff', ['OpenSlidesApp.motions.lineNumberi
                 return out;
             });
 
+            // Remove <del> tags that only delete line numbers
             diffUnnormalized = diffUnnormalized.replace(
                 /<del>((<BR CLASS="OS-LINE-BREAK">)?<span[^>]+OS-LINE-NUMBER[^>]+?>\s*<\/span>)<\/del>/gi,
                 function(found,tag) {
-                    return tag.toLowerCase().replace(/> <\/span/gi, ">&nbsp;</span");
+                    return tag;
                 }
             );
 
+            // Lowercase line number markup
+            diffUnnormalized = diffUnnormalized.replace(
+                /<BR CLASS="OS-LINE-BREAK">/gi,
+                '<br class="os-line-break">'
+            );
+            diffUnnormalized = diffUnnormalized.replace(
+                /<span[^>]+OS-LINE-NUMBER[^>]+?>\s*<\/span>/gi,
+                function(found) {
+                    return found.toLowerCase().replace(/> <\/span/gi, ">&nbsp;</span");
+                }
+            );
+
+
             if (diffUnnormalized.substr(0, workaroundPrepend.length) === workaroundPrepend) {
                 diffUnnormalized = diffUnnormalized.substring(workaroundPrepend.length);
             }
diff --git a/tests/karma/motions/diff.service.test.js b/tests/karma/motions/diff.service.test.js
index 4fadbc178..46c7c3ec6 100644
--- a/tests/karma/motions/diff.service.test.js
+++ b/tests/karma/motions/diff.service.test.js
@@ -503,6 +503,19 @@ describe('linenumbering', function () {
           "<div>" + noMarkup(4) + "Go on</div>"
       );
     });
+
+    it('detects broken HTML and lowercases class names', function () {
+        var before = "<p><span class=\"os-line-number line-number-3\" data-line-number=\"3\" contenteditable=\"false\">&nbsp;</span>holen, da rief sie alle sieben herbei und sprach:</p>\n\n<p><span class=\"os-line-number line-number-4\" data-line-number=\"4\" contenteditable=\"false\">&nbsp;</span><span style=\"color: #000000;\">\"Liebe Kinder, ich will hinaus in den Wald, seid auf der Hut vor dem Wolf! Wenn er <br class=\"os-line-break\"><span class=\"os-line-number line-number-5\" data-line-number=\"5\" contenteditable=\"false\">&nbsp;</span>hereinkommt, frisst er euch alle mit Haut und Haar. Der Bösewicht verstellt sich oft, aber <br class=\"os-line-break\"><span class=\"os-line-number line-number-6\" data-line-number=\"6\" contenteditable=\"false\">&nbsp;</span>an der rauen Stimme und an seinen schwarzen Füßen werdet ihr ihn schon erkennen.\"</span></p>\n\n<p><span class=\"os-line-number line-number-7\" data-line-number=\"7\" contenteditable=\"false\">&nbsp;</span>Die Geißlein sagten: \" Liebe Mutter, wir wollen uns schon in acht nehmen, du kannst ohne </p>",
+            after = "<p>holen, da rief sie alle sieben herbei und sprach:</p>\n\n<p><span style=\"color: #000000;\">Hello</span></p>\n\n<p><span style=\"color: #000000;\">World</span></p>\n\n<p><span style=\"color: #000000;\">Ya</span></p>\n\n<p>Die Geißlein sagten: \" Liebe Mutter, wir wollen uns schon in acht nehmen, du kannst ohne</p>";
+        var diff = diffService.diff(before, after);
+        expect(diff).toBe("<P class=\"delete\"><SPAN class=\"os-line-number line-number-3\" data-line-number=\"3\" contenteditable=\"false\"> </SPAN>holen, da rief sie alle sieben herbei und sprach:</P><DEL>\n\n</DEL>" +
+            "<P class=\"delete\"><SPAN class=\"os-line-number line-number-4\" data-line-number=\"4\" contenteditable=\"false\"> </SPAN><SPAN style=\"color: #000000;\">\"Liebe Kinder, ich will hinaus in den Wald, seid auf der Hut vor dem Wolf! Wenn er <BR class=\"os-line-break\"><SPAN class=\"os-line-number line-number-5\" data-line-number=\"5\" contenteditable=\"false\"> </SPAN>hereinkommt, frisst er euch alle mit Haut und Haar. Der Bösewicht verstellt sich oft, aber <BR class=\"os-line-break\"><SPAN class=\"os-line-number line-number-6\" data-line-number=\"6\" contenteditable=\"false\"> </SPAN>an der rauen Stimme und an seinen schwarzen Füßen werdet ihr ihn schon erkennen.\"</SPAN></P><DEL>\n\n</DEL><P class=\"delete\"><SPAN class=\"os-line-number line-number-7\" data-line-number=\"7\" contenteditable=\"false\"> </SPAN>Die Geißlein sagten: \" Liebe Mutter, wir wollen uns schon in acht nehmen, du kannst ohne </P>" +
+            "<P class=\"insert\">holen, da rief sie alle sieben herbei und sprach:</P><INS>\n\n</INS>" +
+            "<P class=\"insert\"><SPAN style=\"color: #000000;\">Hello</SPAN></P><INS>\n\n</INS>" +
+            "<P class=\"insert\"><SPAN style=\"color: #000000;\">World</SPAN></P><INS>\n\n</INS>" +
+            "<P class=\"insert\"><SPAN style=\"color: #000000;\">Ya</SPAN></P><INS>\n\n</INS>" +
+            "<P class=\"insert\">Die Geißlein sagten: \" Liebe Mutter, wir wollen uns schon in acht nehmen, du kannst ohne</P>");
+    });
   });
 
   describe('addCSSClassToFirstTag function', function () {