Updated markdown regexes to better handle underscores and links

2 files changed, 42 insertions, 33 deletions

diff --git a/web/react/utils/markdown.jsx b/web/react/utils/markdown.jsx
index 4fb4c6089..e34f3d00a 100644
--- a/web/react/utils/markdown.jsx
+++ b/web/react/utils/markdown.jsx
@@ -10,9 +10,21 @@ class MattermostInlineLexer extends marked.InlineLexer {
     constructor(links, options) {
         super(links, options);
 
-        // modified version of the regex that doesn't break up words in snake_case
-        // the original is /^[\s\S]+?(?=[\\<!\[_*`]| {2,}\n|$)/
-        this.rules.text = /^[\s\S]+?(?=__|\b_|[\\<!\[*`]| {2,}\n|$)/;
+        this.rules = Object.assign({}, this.rules);
+
+        // modified version of the regex that doesn't break up words in snake_case,
+        // allows for links starting with www, and allows links succounded by parentheses
+        // the original is /^[\s\S]+?(?=[\\<!\[_*`~]|https?:\/\/| {2,}\n|$)/
+        this.rules.text = /^[\s\S]+?(?=[^\w\/]_|[\\<!\[*`~]|https?:\/\/|www\.|\(| {2,}\n|$)/;
+
+        // modified version of the regex that allows links starting with www and those surrounded
+        // by parentheses
+        // the original is /^(https?:\/\/[^\s<]+[^<.,:;"')\]\s])/
+        this.rules.url = /^(\(?(?:https?:\/\/|www\.)[^\s<.][^\s<]*[^<.,:;"'\]\s])/;
+
+        // modified version of the regex that allows <links> starting with www.
+        // the original is /^<([^ >]+(@|:\/)[^ >]+)>/
+        this.rules.autolink = /^<((?:[^ >]+(@|:\/)|www\.)[^ >]+)>/;
     }
 }
 
@@ -56,8 +68,20 @@ class MattermostMarkdownRenderer extends marked.Renderer {
 
     link(href, title, text) {
         let outHref = href;
+        let outText = text;
+        let prefix = '';
+        let suffix = '';
+
+        // some links like https://en.wikipedia.org/wiki/Rendering_(computer_graphics) contain brackets
+        // and we try our best to differentiate those from ones just wrapped in brackets when autolinking
+        if (outHref.startsWith('(') && outHref.endsWith(')') && text === outHref) {
+            prefix = '(';
+            suffix = ')';
+            outText = text.substring(1, text.length - 1);
+            outHref = outHref.substring(1, outHref.length - 1);
+        }
 
-        if (!(/^(mailto|https?|ftp)/.test(outHref))) {
+        if (!(/[a-z+.-]+:/i).test(outHref)) {
             outHref = `http://${outHref}`;
         }
 
@@ -72,26 +96,17 @@ class MattermostMarkdownRenderer extends marked.Renderer {
             output += ' target="_blank">';
         }
 
-        output += text + '</a>';
+        output += outText + '</a>';
 
-        return output;
+        return prefix + output + suffix;
     }
 
     paragraph(text) {
-        let outText = text;
-
-        // required so markdown does not strip '_' from @user_names
-        outText = TextFormatting.doFormatMentions(text);
-
-        if (!('emoticons' in this.options) || this.options.emoticon) {
-            outText = TextFormatting.doFormatEmoticons(outText);
-        }
-
         if (this.formattingOptions.singleline) {
-            return `<p class="markdown__paragraph-inline">${outText}</p>`;
+            return `<p class="markdown__paragraph-inline">${text}</p>`;
         }
 
-        return super.paragraph(outText);
+        return super.paragraph(text);
     }
 
     table(header, body) {
@@ -106,7 +121,8 @@ class MattermostMarkdownRenderer extends marked.Renderer {
 export function format(text, options) {
     const markdownOptions = {
         renderer: new MattermostMarkdownRenderer(null, options),
-        sanitize: true
+        sanitize: true,
+        gfm: true
     };
 
     const tokens = marked.lexer(text, markdownOptions);
diff --git a/web/react/utils/text_formatting.jsx b/web/react/utils/text_formatting.jsx
index 9c87d2693..2de858a17 100644
--- a/web/react/utils/text_formatting.jsx
+++ b/web/react/utils/text_formatting.jsx
@@ -43,7 +43,7 @@ export function doFormatText(text, options) {
 
     // replace important words and phrases with tokens
     output = autolinkAtMentions(output, tokens);
-    output = autolinkUrls(output, tokens);
+    output = autolinkEmails(output, tokens);
     output = autolinkHashtags(output, tokens);
 
     if (!('emoticons' in options) || options.emoticon) {
@@ -93,28 +93,21 @@ export function sanitizeHtml(text) {
     return output;
 }
 
-// Convert URLs into tokens
-function autolinkUrls(text, tokens) {
-    function replaceUrlWithToken(autolinker, match) {
+// Convert emails into tokens
+function autolinkEmails(text, tokens) {
+    function replaceEmailWithToken(autolinker, match) {
         const linkText = match.getMatchedText();
         let url = linkText;
 
         if (match.getType() === 'email') {
             url = `mailto:${url}`;
-        } else if (!(/^(mailto|https?|ftp)/.test(url))) {
-            url = `http://${url}`;
         }
 
         const index = tokens.size;
-        const alias = `MM_LINK${index}`;
-
-        var target = 'target="_blank"';
-        if (url.lastIndexOf(Utils.getTeamURLFromAddressBar(), 0) === 0) {
-            target = '';
-        }
+        const alias = `MM_EMAIL${index}`;
 
         tokens.set(alias, {
-            value: `<a class="theme" ${target} href="${url}">${linkText}</a>`,
+            value: `<a class="theme" href="${url}">${linkText}</a>`,
             originalText: linkText
         });
 
@@ -123,12 +116,12 @@ function autolinkUrls(text, tokens) {
 
     // we can't just use a static autolinker because we need to set replaceFn
     const autolinker = new Autolinker({
-        urls: true,
+        urls: false,
         email: true,
         phone: false,
         twitter: false,
         hashtag: false,
-        replaceFn: replaceUrlWithToken
+        replaceFn: replaceEmailWithToken
     });
 
     return autolinker.link(text);