From 684150d36dce827a664dc56c6c74f538aae907a4 Mon Sep 17 00:00:00 2001
From: Hannah Wolfe <erisds@gmail.com>
Date: Sun, 1 Sep 2013 16:32:54 +0100
Subject: [PATCH] GFM Auto-linking fixes

closes #577

- auto-linking was interfering with things that were already a link
- now checks to ensure we are not in a HTML or markdown link already
- added tonnes of new tests
---
 .../vendor/showdown/extensions/github.js      |  31 +--
 core/test/unit/client_showdown_int_spec.js    | 147 ++++++++++++--
 core/test/unit/shared_gfm_spec.js             | 185 +++++++++++++++++-
 3 files changed, 328 insertions(+), 35 deletions(-)

diff --git a/core/shared/vendor/showdown/extensions/github.js b/core/shared/vendor/showdown/extensions/github.js
index e8d81f564a..6c2ad2dd50 100644
--- a/core/shared/vendor/showdown/extensions/github.js
+++ b/core/shared/vendor/showdown/extensions/github.js
@@ -63,29 +63,34 @@
                     }
 
                     // filter out def urls
-                    text = text.replace(/^ *\[([^\]]+)\]: *<?([^\s>]+)>?(?: +["(]([^\n]+)[")])? *(?:\n+|$)/gim,
+                    // from Marked https://github.com/chjj/marked/blob/master/lib/marked.js#L24
+                    text = text.replace(/^ *\[([^\]]+)\]: *<?([^\s>]+)>?(?: +["(]([^\n]+)[")])? *(?:\n+|$)/gmi,
                         function (x) {
                             var hash = hashId();
                             extractions[hash] = x;
                             return "{gfm-js-extract-ref-url-" + hash + "}";
                         });
 
-                    // taken from https://gist.github.com/jorilallo/1283095#L158
-                    text = text.replace(/https?\:\/\/[^"\s\<\>]*[^.,;'">\:\s\<\>\)\]\!]/g, function (wholeMatch, matchIndex) {
-                        var left = text.slice(0, matchIndex), right = text.slice(matchIndex),
-                            href;
-                        if (left.match(/<[^>]+$/) && right.match(/^[^>]*>/)) {
-                            return wholeMatch;
-                        }
-                        href = wholeMatch.replace(/^http:\/\/github.com\//, "https://github.com/");
-                        return "<a href='" + href + "'>" + wholeMatch + "</a>";
-                    });
+                    // match a URL
+                    // adapted from https://gist.github.com/jorilallo/1283095#L158
+                    // and http://blog.stevenlevithan.com/archives/mimic-lookbehind-javascript
+                    text = text.replace(/(\]\(|\]|\[|<a[^\>]*?\>)?https?\:\/\/[^"\s\<\>]*[^.,;'">\:\s\<\>\)\]\!]/gmi,
+                        function (wholeMatch, lookBehind, matchIndex) {
+                            // Check we are not inside an HTML tag
+                            var left = text.slice(0, matchIndex), right = text.slice(matchIndex);
+                            if ((left.match(/<[^>]+$/) && right.match(/^[^>]*>/)) || lookBehind) {
+                                return wholeMatch;
+                            }
+                            // If we have a matching lookBehind, this is a failure, else wrap the match in <a> tag
+                            return lookBehind ? wholeMatch : "<a href='" + wholeMatch + "'>" + wholeMatch + "</a>";
+                        });
 
-                    text = text.replace(/[a-z0-9_\-+=.]+@[a-z0-9\-]+(\.[a-z0-9-]+)+/ig, function (wholeMatch) {
+                    // match emil
+                    text = text.replace(/[a-z0-9_\-+=.]+@[a-z0-9\-]+(\.[a-z0-9-]+)+/gmi, function (wholeMatch) {
                         return "<a href='mailto:" + wholeMatch + "'>" + wholeMatch + "</a>";
                     });
 
-                    text = text.replace(/\{gfm-js-extract-ref-url-([0-9]+)\}/gm, function (x, y) {
+                    text = text.replace(/\{gfm-js-extract-ref-url-([0-9]+)\}/gi, function (x, y) {
                         return "\n\n" + extractions[y];
                     });
 
diff --git a/core/test/unit/client_showdown_int_spec.js b/core/test/unit/client_showdown_int_spec.js
index b826514945..ae4cd4773a 100644
--- a/core/test/unit/client_showdown_int_spec.js
+++ b/core/test/unit/client_showdown_int_spec.js
@@ -136,12 +136,43 @@ describe("Showdown client side converter", function () {
         });
     });
 
-    it("should auto-link URL", function () {
+    it("should auto-link URL in text with markdown syntax", function () {
         var testPhrases = [
-                {input: "http://google.co.uk", output: /^<p><a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a><\/p>$/},
+                {
+                    input: "http://google.co.uk",
+                    output: /^<p><a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a><\/p>$/
+                },
                 {
                     input: "https://atest.com/fizz/buzz?baz=fizzbuzz",
                     output: /^<p><a href=\'https:\/\/atest.com\/fizz\/buzz\?baz=fizzbuzz\'>https:\/\/atest.com\/fizz\/buzz\?baz=fizzbuzz<\/a><\/p>$/
+                },
+                {
+                    input: "Some [ text (http://www.google.co.uk) some other text",
+                    output: /^<p>Some \[ text \(<a href=\'http:\/\/www.google.co.uk\'>http:\/\/www.google.co.uk<\/a>\) some other text<\/p>$/
+                },
+                {
+                    input: ">http://google.co.uk",
+                    output: /^<blockquote>\n  <p><a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a><\/p>\n<\/blockquote>$/
+                },
+                {
+                    input: "> http://google.co.uk",
+                    output: /^<blockquote>\n  <p><a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a><\/p>\n<\/blockquote>$/
+                },
+                {
+                    input: "<>>> http://google.co.uk",
+                    output: /^<p>&lt;>>> <a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a><\/p>$/
+                },
+                {
+                    input: "<strong>http://google.co.uk",
+                    output: /^<p><strong><a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a><\/p>$/
+                },
+                {
+                    input: "# http://google.co.uk",
+                    output: /^<h1 id="ahrefhttpgooglecoukhttpgooglecouka"><a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a><\/h1>$/
+                },
+                {
+                    input: "* http://google.co.uk",
+                    output: /^<ul>\n<li><a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a><\/li>\n<\/ul>$/
                 }
             ],
             processedMarkup;
@@ -159,7 +190,6 @@ describe("Showdown client side converter", function () {
         processedMarkup.should.match(testPhrase.output);
     });
 
-    // Fails
     it("should convert reference format URL", function () {
         var testPhrases = [
                 {
@@ -169,6 +199,10 @@ describe("Showdown client side converter", function () {
                 {
                     input: "[Google][1]\n\n[1]: http://google.co.uk \"some text\"",
                     output: /^<p><a href="http:\/\/google.co.uk" title="some text">Google<\/a><\/p>$/
+                },
+                {
+                    input: "[http://google.co.uk]: http://google.co.uk\n\n[Hello][http://google.co.uk]",
+                    output: /^<p><a href="http:\/\/google.co.uk">Hello<\/a><\/p>$/
                 }
             ],
             processedMarkup;
@@ -188,6 +222,10 @@ describe("Showdown client side converter", function () {
                 {
                     input: "![Google][1]\n\n[1]: http://dsurl.stuff/something.jpg \"some text\"",
                     output: /^<section.*?<img.*?src="http:\/\/dsurl.stuff\/something.jpg"\/>.*?<\/section>$/
+                },
+                {
+                    input: "[http://www.google.co.uk]: http://www.google.co.uk\n\n![Hello][http://www.google.co.uk]",
+                    output: /^<section.*?<img.*?src="http:\/\/www.google.co.uk"\/>.*?<\/section>$/
                 }
             ],
             processedMarkup;
@@ -198,19 +236,70 @@ describe("Showdown client side converter", function () {
         });
     });
 
-    it("should NOT auto-link reference URL", function () {
-        var testPhrase = {input: "[1]: http://google.co.uk", output: /^$/},
-            processedMarkup = converter.makeHtml(testPhrase.input);
+    it("should NOT auto-link URL in HTML", function () {
+        var testPhrases = [
+                {
+                    input: '<img src="http://placekitten.com/50">',
+                    output: /^<p><img src=\"http:\/\/placekitten.com\/50\"><\/p>$/
+                },
+                {
+                    input: '<img src="http://placekitten.com/50" />',
+                    output: /^<p><img src=\"http:\/\/placekitten.com\/50\" \/><\/p>$/
+                },
+                {
+                    input: '<script type="text/javascript" src="http://google.co.uk"></script>',
+                    output: /^<script type=\"text\/javascript\" src=\"http:\/\/google.co.uk\"><\/script>$/
+                },
+                {
+                    input: '<a href="http://facebook.com">http://google.co.uk</a>',
+                    output: /^<p><a href=\"http:\/\/facebook.com\">http:\/\/google.co.uk<\/a><\/p>$/
+                },
+                {
+                    input: '<a href="http://facebook.com">test</a> http://google.co.uk',
+                    output: /^<p><a href=\"http:\/\/facebook.com\">test<\/a> <a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a><\/p>$/
+                }
+            ],
+            processedMarkup;
 
-        processedMarkup.should.match(testPhrase.output);
+        testPhrases.forEach(function (testPhrase) {
+            processedMarkup = converter.makeHtml(testPhrase.input);
+            processedMarkup.should.match(testPhrase.output);
+        });
     });
 
+    it("should not display anything for reference URL", function () {
+        var testPhrases = [
+                {
+                    input: "[1]: http://www.google.co.uk",
+                    output: /^$/
+                },
+                {
+                    input: "[http://www.google.co.uk]: http://www.google.co.uk",
+                    output: /^$/
+                },
+                {
+                    input: "[1]: http://dsurl.stuff/something.jpg",
+                    output: /^$/
+                },
+                {
+                    input: "[1]:http://www.google.co.uk",
+                    output: /^$/
+                },
+                {
+                    input: " [1]:http://www.google.co.uk",
+                    output: /^$/
+                },
+                {
+                    input: "",
+                    output: /^$/
+                }
+            ],
+            processedMarkup;
 
-    it("should NOT auto-link image reference URL", function () {
-        var testPhrase = {input: "[1]: http://dsurl.stuff/something.jpg", output: /^$/},
+        testPhrases.forEach(function (testPhrase) {
             processedMarkup = converter.makeHtml(testPhrase.input);
-
-        processedMarkup.should.match(testPhrase.output);
+            processedMarkup.should.match(testPhrase.output);
+        });
     });
 
     it("should show placeholder for image markdown", function () {
@@ -276,8 +365,41 @@ describe("Showdown client side converter", function () {
         });
     });
 
-    it("should NOT auto-link image URL", function () {
+//    it("should NOT auto-link URL in image", function () {
+//        var testPhrases = [
+//                {
+//                    input: "![http://google.co.uk/kitten.jpg](http://google.co.uk/kitten.jpg)",
+//                    output: /^<section.*?((?!<a href=\'http:\/\/google.co.uk\/kitten.jpg\').)*<\/section>$/
+//                },
+//                {
+//                    input:  "![image stuff](http://dsurl.stuff/something)",
+//                    output: /^<section.*?((?!<a href=\'http:\/\/dsurl.stuff\/something\').)*<\/section>$/
+
+//    });
+//
+
+    it("should correctly output link and image markdown without autolinks", function () {
         var testPhrases = [
+                {
+                    input: "[1](http://google.co.uk)",
+                    output: /^<p><a href="http:\/\/google.co.uk">1<\/a><\/p>$/
+                },
+                {
+                    input: "  [1](http://google.co.uk)",
+                    output: /^<p><a href="http:\/\/google.co.uk">1<\/a><\/p>$/
+                },
+                {
+                    input: "[http://google.co.uk](http://google.co.uk)",
+                    output: /^<p><a href="http:\/\/google.co.uk">http:\/\/google.co.uk<\/a><\/p>$/
+                },
+                {
+                    input: "[http://google.co.uk][id]\n\n[id]: http://google.co.uk",
+                    output: /^<p><a href="http:\/\/google.co.uk">http:\/\/google.co.uk<\/a><\/p>$/
+                },
+                {
+                    input: "![http://google.co.uk/kitten.jpg](http://google.co.uk/kitten.jpg)",
+                    output: /^<section.*?((?!<a href=\'http:\/\/google.co.uk\/kitten.jpg\').)*<\/section>$/
+                },
                 {
                     input:  "![image stuff](http://dsurl.stuff/something)",
                     output: /^<section.*?((?!<a href=\'http:\/\/dsurl.stuff\/something\').)*<\/section>$/
@@ -290,5 +412,4 @@ describe("Showdown client side converter", function () {
             processedMarkup.should.match(testPhrase.output);
         });
     });
-
 });
\ No newline at end of file
diff --git a/core/test/unit/shared_gfm_spec.js b/core/test/unit/shared_gfm_spec.js
index 43d7b4b10b..5ad02922ff 100644
--- a/core/test/unit/shared_gfm_spec.js
+++ b/core/test/unit/shared_gfm_spec.js
@@ -46,12 +46,67 @@ describe("Github showdown extensions", function () {
         processedMarkup.should.match(testPhrase.output);
     });
 
-    it("should auto-link URL", function () {
+    it("should auto-link URL in text with markdown syntax", function () {
         var testPhrases = [
-                {input: "http://google.co.uk", output: /^<a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>$/},
+                {
+                    input: "http://google.co.uk",
+                    output: /^<a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>$/
+                },
                 {
                     input: "https://atest.com/fizz/buzz?baz=fizzbuzz",
                     output: /^<a href=\'https:\/\/atest.com\/fizz\/buzz\?baz=fizzbuzz\'>https:\/\/atest.com\/fizz\/buzz\?baz=fizzbuzz<\/a>$/
+                },
+                {
+                    input: "Some text http://www.google.co.uk some other text",
+                    output: /^Some text <a href=\'http:\/\/www.google.co.uk\'>http:\/\/www.google.co.uk<\/a> some other text$/
+                },
+                {
+                    input: "Some [ text http://www.google.co.uk some other text",
+                    output: /^Some \[ text <a href=\'http:\/\/www.google.co.uk\'>http:\/\/www.google.co.uk<\/a> some other text$/
+                },
+                {
+                    input: "Some [ text (http://www.google.co.uk) some other text",
+                    output: /^Some \[ text \(<a href=\'http:\/\/www.google.co.uk\'>http:\/\/www.google.co.uk<\/a>\) some other text$/
+                },
+                {
+                    input: "  http://google.co.uk  ",
+                    output: /^  <a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>  $/
+                },
+                {
+                    input: ">http://google.co.uk",
+                    output: /^><a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>$/
+                },
+                {
+                    input: "> http://google.co.uk",
+                    output: /^> <a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>$/
+                },
+                {
+                    input: "<>>> http://google.co.uk",
+                    output: /^<>>> <a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>$/
+                },
+                {
+                    input: "<>>>http://google.co.uk",
+                    output: /^<>>><a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>$/
+                },
+                {
+                    input: "<some text>>>http://google.co.uk",
+                    output: /^<some text>>><a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>$/
+                },
+                {
+                    input: "<strong>http://google.co.uk",
+                    output: /^<strong><a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>$/
+                },
+                {
+                    input: "# http://google.co.uk",
+                    output: /^# <a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>$/
+                },
+                {
+                    input: "#http://google.co.uk",
+                    output: /^#<a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>$/
+                },
+                {
+                    input: "* http://google.co.uk",
+                    output: /^\* <a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>$/
                 }
             ],
             processedMarkup;
@@ -69,17 +124,129 @@ describe("Github showdown extensions", function () {
         processedMarkup.should.match(testPhrase.output);
     });
 
+    it("should NOT auto-link URL in HTML", function () {
+        var testPhrases = [
+                {
+                    input: '<img src="http://placekitten.com/50">',
+                    output: /^<img src=\"http:\/\/placekitten.com\/50\">$/
+                },
+                {
+                    input: '<img src="http://placekitten.com/50" />',
+                    output: /^<img src=\"http:\/\/placekitten.com\/50\" \/>$/
+                },
+                {
+                    input: '<script type="text/javascript" src="http://google.co.uk"></script>',
+                    output: /^<script type=\"text\/javascript\" src=\"http:\/\/google.co.uk\"><\/script>$/
+                },
+                {
+                    input: '<a href="http://facebook.com">http://google.co.uk</a>',
+                    output: /^<a href=\"http:\/\/facebook.com\">http:\/\/google.co.uk<\/a>$/
+                },
+                {
+                    input: '<a href="http://facebook.com">test</a> http://google.co.uk',
+                    output: /^<a href=\"http:\/\/facebook.com\">test<\/a> <a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>$/
+                }
+            ],
+            processedMarkup;
+
+        testPhrases.forEach(function (testPhrase) {
+            processedMarkup = _ConvertPhrase(testPhrase.input);
+            processedMarkup.should.match(testPhrase.output);
+        });
+    });
+
     it("should NOT auto-link reference URL", function () {
-        var testPhrase = {input: "[1]: http://google.co.uk", output: /^\n\n\[1\]: http:\/\/google.co.uk$/},
-            processedMarkup = _ConvertPhrase(testPhrase.input);
+        var testPhrases = [
+                {
+                    input: "[1]: http://www.google.co.uk",
+                    output: /^\n\n\[1\]: http:\/\/www.google.co.uk$/
+                },
+                {
+                    input: "[http://www.google.co.uk]: http://www.google.co.uk",
+                    output: /^\n\n\[http:\/\/www.google.co.uk]: http:\/\/www.google.co.uk$/
+                },
+                {
+                    input: "[1]: http://dsurl.stuff/something.jpg",
+                    output: /^\n\n\[1\]: http:\/\/dsurl.stuff\/something.jpg$/
+                },
+                {
+                    input: "[1]:http://www.google.co.uk",
+                    output: /^\n\n\[1\]:http:\/\/www.google.co.uk$/
+                },
+                {
+                    input: " [1]:http://www.google.co.uk",
+                    output: /^\n\n \[1\]:http:\/\/www.google.co.uk$/
+                },
+                {
+                    input: "[http://www.google.co.uk]: http://www.google.co.uk",
+                    output: /^\n\n\[http:\/\/www.google.co.uk\]: http:\/\/www.google.co.uk$/
+                }
+            ],
+            processedMarkup;
 
-        processedMarkup.should.match(testPhrase.output);
+        testPhrases.forEach(function (testPhrase) {
+            processedMarkup = _ConvertPhrase(testPhrase.input);
+            processedMarkup.should.match(testPhrase.output);
+        });
     });
 
-    it("should NOT auto-link image URL", function () {
-        var testPhrase = {input: "[1]: http://dsurl.stuff/something.jpg", output: /^\n\n\[1\]: http:\/\/dsurl.stuff\/something.jpg$/},
-            processedMarkup = _ConvertPhrase(testPhrase.input);
+    it("should NOT auto-link URL in link or image markdown", function () {
+        var testPhrases = [
+                {
+                    input: "[1](http://google.co.uk)",
+                    output: /^\[1\]\(http:\/\/google.co.uk\)$/
+                },
+                {
+                    input: "  [1](http://google.co.uk)",
+                    output: /^  \[1\]\(http:\/\/google.co.uk\)$/
+                },
+                {
+                    input: "[http://google.co.uk](http://google.co.uk)",
+                    output: /^\[http:\/\/google.co.uk\]\(http:\/\/google.co.uk\)$/
+                },
+                {
+                    input: "[http://google.co.uk][id]",
+                    output: /^\[http:\/\/google.co.uk\]\[id\]$/
+                },
+                {
+                    input: "![1](http://google.co.uk/kitten.jpg)",
+                    output: /^!\[1\]\(http:\/\/google.co.uk\/kitten.jpg\)$/
+                },
+                {
+                    input: "  ![1](http://google.co.uk/kitten.jpg)",
+                    output: /^  !\[1\]\(http:\/\/google.co.uk\/kitten.jpg\)$/
+                },
+                {
+                    input: "![http://google.co.uk/kitten.jpg](http://google.co.uk/kitten.jpg)",
+                    output: /^!\[http:\/\/google.co.uk\/kitten.jpg\]\(http:\/\/google.co.uk\/kitten.jpg\)$/
+                }
+            ],
+            processedMarkup;
 
-        processedMarkup.should.match(testPhrase.output);
+        testPhrases.forEach(function (testPhrase) {
+            processedMarkup = _ConvertPhrase(testPhrase.input);
+            processedMarkup.should.match(testPhrase.output);
+        });
     });
+
+     // behaviour if you add a gap between [] and ()
+    it("should auto-link if markdown is invalid", function () {
+        var testPhrases = [
+                {
+                    input: "[1] (http://google.co.uk)",
+                    output: /^\[1\] \(<a href=\'http:\/\/google.co.uk\'>http:\/\/google.co.uk<\/a>\)$/
+                },
+                {
+                    input: "![1] (http://google.co.uk/kitten.jpg)",
+                    output: /^!\[1\] \(<a href=\'http:\/\/google.co.uk\/kitten.jpg\'>http:\/\/google.co.uk\/kitten.jpg<\/a>\)$/
+                }
+            ],
+            processedMarkup;
+
+        testPhrases.forEach(function (testPhrase) {
+            processedMarkup = _ConvertPhrase(testPhrase.input);
+            processedMarkup.should.match(testPhrase.output);
+        });
+    });
+
 });
\ No newline at end of file