1 files changed, 143 insertions, 0 deletions
diff --git a/tests/unit/markup.js b/tests/unit/markup.js
new file mode 100644
index 0000000..603ca81
--- /dev/null
+++ b/tests/unit/markup.js
@@ -0,0 +1,143 @@
+// -*- mode: js; js-indent-level: 4; indent-tabs-mode: nil -*-
+
+// Test cases for MessageList markup parsing
+
+const JsUnit = imports.jsUnit;
+const Pango = imports.gi.Pango;
+
+const Environment = imports.ui.environment;
+Environment.init();
+
+const Main = imports.ui.main; // unused, but needed to break dependency loop
+const MessageList = imports.ui.messageList;
+
+// Assert that @input, assumed to be markup, gets "fixed" to @output,
+// which is valid markup. If @output is null, @input is expected to
+// convert to itself
+function assertConverts(input, output) {
+    if (!output)
+        output = input;
+    let fixed = MessageList._fixMarkup(input, true);
+    JsUnit.assertEquals(output, fixed);
+
+    let parsed = false;
+    try {
+        Pango.parse_markup(fixed, -1, '');
+        parsed = true;
+    } catch (e) {}
+    JsUnit.assertEquals(true, parsed);
+}
+
+// Assert that @input, assumed to be plain text, gets escaped to @output,
+// which is valid markup.
+function assertEscapes(input, output) {
+    let fixed = MessageList._fixMarkup(input, false);
+    JsUnit.assertEquals(output, fixed);
+
+    let parsed = false;
+    try {
+        Pango.parse_markup(fixed, -1, '');
+        parsed = true;
+    } catch (e) {}
+    JsUnit.assertEquals(true, parsed);
+}
+
+
+
+// CORRECT MARKUP
+
+assertConverts('foo');
+assertEscapes('foo', 'foo');
+
+assertConverts('<b>foo</b>');
+assertEscapes('<b>foo</b>', '&lt;b&gt;foo&lt;/b&gt;');
+
+assertConverts('something <i>foo</i>');
+assertEscapes('something <i>foo</i>', 'something &lt;i&gt;foo&lt;/i&gt;');
+
+assertConverts('<u>foo</u> something');
+assertEscapes('<u>foo</u> something', '&lt;u&gt;foo&lt;/u&gt; something');
+
+assertConverts('<b>bold</b> <i>italic <u>and underlined</u></i>');
+assertEscapes('<b>bold</b> <i>italic <u>and underlined</u></i>', '&lt;b&gt;bold&lt;/b&gt; &lt;i&gt;italic &lt;u&gt;and underlined&lt;/u&gt;&lt;/i&gt;');
+
+assertConverts('this &amp; that');
+assertEscapes('this &amp; that', 'this &amp;amp; that');
+
+assertConverts('this &lt; that');
+assertEscapes('this &lt; that', 'this &amp;lt; that');
+
+assertConverts('this &lt; that &gt; the other');
+assertEscapes('this &lt; that &gt; the other', 'this &amp;lt; that &amp;gt; the other');
+
+assertConverts('this &lt;<i>that</i>&gt;');
+assertEscapes('this &lt;<i>that</i>&gt;', 'this &amp;lt;&lt;i&gt;that&lt;/i&gt;&amp;gt;');
+
+assertConverts('<b>this</b> > <i>that</i>');
+assertEscapes('<b>this</b> > <i>that</i>', '&lt;b&gt;this&lt;/b&gt; &gt; &lt;i&gt;that&lt;/i&gt;');
+
+
+
+// PARTIALLY CORRECT MARKUP
+// correct bits are kept, incorrect bits are escaped
+
+// unrecognized entity
+assertConverts('<b>smile</b> &#9786;!', '<b>smile</b> &amp;#9786;!');
+assertEscapes('<b>smile</b> &#9786;!', '&lt;b&gt;smile&lt;/b&gt; &amp;#9786;!');
+
+// stray '&'; this is really a bug, but it's easier to do it this way
+assertConverts('<b>this</b> & <i>that</i>', '<b>this</b> &amp; <i>that</i>');
+assertEscapes('<b>this</b> & <i>that</i>', '&lt;b&gt;this&lt;/b&gt; &amp; &lt;i&gt;that&lt;/i&gt;');
+
+// likewise with stray '<'
+assertConverts('this < that', 'this &lt; that');
+assertEscapes('this < that', 'this &lt; that');
+
+assertConverts('<b>this</b> < <i>that</i>', '<b>this</b> &lt; <i>that</i>');
+assertEscapes('<b>this</b> < <i>that</i>', '&lt;b&gt;this&lt;/b&gt; &lt; &lt;i&gt;that&lt;/i&gt;');
+
+assertConverts('this < that > the other', 'this &lt; that > the other');
+assertEscapes('this < that > the other', 'this &lt; that &gt; the other');
+
+assertConverts('this <<i>that</i>>', 'this &lt;<i>that</i>>');
+assertEscapes('this <<i>that</i>>', 'this &lt;&lt;i&gt;that&lt;/i&gt;&gt;');
+
+// unknown tags
+assertConverts('<unknown>tag</unknown>', '&lt;unknown>tag&lt;/unknown>');
+assertEscapes('<unknown>tag</unknown>', '&lt;unknown&gt;tag&lt;/unknown&gt;');
+
+// make sure we check beyond the first letter
+assertConverts('<bunknown>tag</bunknown>', '&lt;bunknown>tag&lt;/bunknown>');
+assertEscapes('<bunknown>tag</bunknown>', '&lt;bunknown&gt;tag&lt;/bunknown&gt;');
+
+// with mix of good and bad, we keep the good and escape the bad
+assertConverts('<i>known</i> and <unknown>tag</unknown>', '<i>known</i> and &lt;unknown>tag&lt;/unknown>');
+assertEscapes('<i>known</i> and <unknown>tag</unknown>', '&lt;i&gt;known&lt;/i&gt; and &lt;unknown&gt;tag&lt;/unknown&gt;');
+
+
+
+// FULLY INCORRECT MARKUP
+// (fall back to escaping the whole thing)
+
+// tags not matched up
+assertConverts('<b>in<i>com</i>plete', '&lt;b&gt;in&lt;i&gt;com&lt;/i&gt;plete');
+assertEscapes('<b>in<i>com</i>plete', '&lt;b&gt;in&lt;i&gt;com&lt;/i&gt;plete');
+
+assertConverts('in<i>com</i>plete</b>', 'in&lt;i&gt;com&lt;/i&gt;plete&lt;/b&gt;');
+assertEscapes('in<i>com</i>plete</b>', 'in&lt;i&gt;com&lt;/i&gt;plete&lt;/b&gt;');
+
+// we don't support attributes, and it's too complicated to try
+// to escape both start and end tags, so we just treat it as bad
+assertConverts('<b>good</b> and <b style=\'bad\'>bad</b>', '&lt;b&gt;good&lt;/b&gt; and &lt;b style=&apos;bad&apos;&gt;bad&lt;/b&gt;');
+assertEscapes('<b>good</b> and <b style=\'bad\'>bad</b>', '&lt;b&gt;good&lt;/b&gt; and &lt;b style=&apos;bad&apos;&gt;bad&lt;/b&gt;');
+
+// this is just syntactically invalid
+assertConverts('<b>unrecognized</b stuff>', '&lt;b&gt;unrecognized&lt;/b stuff&gt;');
+assertEscapes('<b>unrecognized</b stuff>', '&lt;b&gt;unrecognized&lt;/b stuff&gt;');
+
+// mismatched tags
+assertConverts('<b>mismatched</i>', '&lt;b&gt;mismatched&lt;/i&gt;');
+assertEscapes('<b>mismatched</i>', '&lt;b&gt;mismatched&lt;/i&gt;');
+
+assertConverts('<b>mismatched/unknown</bunknown>', '&lt;b&gt;mismatched/unknown&lt;/bunknown&gt;');
+assertEscapes('<b>mismatched/unknown</bunknown>', '&lt;b&gt;mismatched/unknown&lt;/bunknown&gt;');