messages display: fix the display of links

The links are now postprocessed to compensate for the malfunctions of
md4c. It now displays properly even with a + or - in it. It's also
checked by the unittests.

GitLab: #1903
Change-Id: I0d7f520a7a3fe06fb89107fe6b08f83325218cd4

src/app/messageparser.cpp

@@ -167,6 +167,38 @@ MessageParser::preprocessMarkdown(QString& markdown)
         markdown += block.second;
     }
 }
+const QRegularExpression linkRegex(R"((http[s]?://[^\s]*[\+\-_][^\s]*|www\.[^\s]*[\+\-_][^\s]*))");
+const QString anchorTagTemplate("<a href=\"%1\">%1</a>");
+
+QByteArray
+MessageParser::postprocessHTML(const QByteArray& html)
+{
+    // HACK : https://github.com/mity/md4c/issues/251
+    // Check for any links that haven't been properly detected by md4c.
+    // We can do this by checking for any text that starts with "http" or "www" and contains a + or
+    // a -. If we find any, we insert a <a> tag around the link.
+    // This function could become obsolete if the md4c library fixes the issue it's addressing.
+
+    QByteArray processedHtml = html;
+    QRegularExpressionMatchIterator it = linkRegex.globalMatch(html);
+
+    int offset = 0;
+    while (it.hasNext()) {
+        QRegularExpressionMatch match = it.next();
+        QString link = match.captured(0);
+        int pTagIndex = link.indexOf("</p>");
+        QString anchorTag = anchorTagTemplate.arg(pTagIndex != -1 ? link.left(pTagIndex) : link);
+        if (pTagIndex != -1) {
+            anchorTag += "</p>";
+        }
+        processedHtml.replace(match.capturedStart(0) + offset,
+                              match.capturedLength(0),
+                              anchorTag.toUtf8());
+        offset += anchorTag.toUtf8().length() - match.capturedLength(0);
+    }
+
+    return processedHtml;
+}
 
 QString
 MessageParser::markdownToHtml(const char* markdown)
@@ -179,5 +211,6 @@ MessageParser::markdownToHtml(const char* markdown)
     }
     QByteArray array;
     const int result = md_html(markdown, MD_SIZE(data_len), &htmlChunkCb, &array, md_flags, 0);
+    array = postprocessHTML(array);
     return result == 0 ? QString::fromUtf8(array) : QString();
 }

src/app/messageparser.h

@@ -60,6 +60,9 @@ private:
     // Preprocess the markdown message (e.g. handle line breaks).
     void preprocessMarkdown(QString& markdown);
 
+    // Postprocess HTML to account for issues in md4c
+    QByteArray postprocessHTML(const QByteArray& html);
+
     // Transform markdown syntax into HTML.
     QString markdownToHtml(const char* markdown);
 

tests/unittests/messageparser_unittest.cpp

@@ -107,6 +107,41 @@ TEST_F(MessageParserFixture, ALinkIsParsedCorrectly)
     // The rest of the link info is not tested here.
 }
 
+/*!
+ * WHEN  We parse a text body with a link and a + or a -.
+ * THEN  The HTML body should be returned correctly including the link even if MD4C doesn't detect it.
+ */
+TEST_F(MessageParserFixture, AComplexLinkIsParsedCorrectly)
+{
+    auto linkColor = QColor::fromRgb(0, 0, 255);
+    auto backgroundColor = QColor::fromRgb(0, 0, 255);
+
+    QSignalSpy messageParsedSpy(globalEnv.messageParser.data(), &MessageParser::messageParsed);
+    QSignalSpy linkInfoReadySpy(globalEnv.messageParser.data(), &MessageParser::linkInfoReady);
+
+    // Parse a message with a link containing a + or a -.
+    globalEnv.messageParser->parseMessage("msgId_03",
+                                          "https://review.jami.net/q/status:open+-is:wip",
+                                          false,
+                                          linkColor,
+                                          backgroundColor);
+
+    // Wait for the messageParsed signal which should be emitted once.
+    messageParsedSpy.wait();
+    EXPECT_EQ(messageParsedSpy.count(), 1);
+
+    QList<QVariant> messageParserArguments = messageParsedSpy.takeFirst();
+    EXPECT_TRUE(messageParserArguments.at(0).typeId() == qMetaTypeId<QString>());
+    EXPECT_EQ(messageParserArguments.at(0).toString(), "msgId_03");
+    EXPECT_TRUE(messageParserArguments.at(1).typeId() == qMetaTypeId<QString>());
+    EXPECT_EQ(messageParserArguments.at(1).toString(),
+              "<style>a{color:#0000ff;}</style><p><a "
+              "href=\"https://review.jami.net/q/status:open+-is:wip\">https://review.jami.net/q/"
+              "status:open+-is:wip</a></p>\n");
+
+    // The rest of the link info is not tested here.
+}
+
 /*!
  * WHEN  We parse a text body with end of line characters.
  * THEN  The HTML body should be returned correctly with the end of line characters.