From 6e6af3cc0356f86038649f327b2f98746374b4e6 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Estev=C3=A3o=20Soares=20dos=20Santos?=
 <estevao.santos@gmail.com>
Date: Mon, 9 May 2022 03:17:43 +0100
Subject: [PATCH] improve compliance in fenced blocks

---
 src/subParsers/makehtml/githubCodeBlock.js    | 43 +++++++++++++------
 .../makehtml/extra.testsuite.commonmark.js    | 12 ++++++
 2 files changed, 43 insertions(+), 12 deletions(-)

diff --git a/src/subParsers/makehtml/githubCodeBlock.js b/src/subParsers/makehtml/githubCodeBlock.js
index 4833c11..ad09aa7 100644
--- a/src/subParsers/makehtml/githubCodeBlock.js
+++ b/src/subParsers/makehtml/githubCodeBlock.js
@@ -32,9 +32,37 @@ showdown.subParser('makehtml.githubCodeBlock', function (text, options, globals)
   startEvent = globals.converter.dispatch(startEvent);
   text = startEvent.output + '¨0';
 
-  let pattern = /(?:^|\n) {0,3}(```+|~~~+) *([^\n\t`~]*)\n([\s\S]*?)\n {0,3}\1/g;
+  //const accentRegex = /(?:^|\n) {0,3}(```+|~~~+) *([^\n\t`~]*)\n([\s\S]*?)(?:(\n {0,3}\1[`~]*)|¨0)/g;
+  const closedBlockRegex   = /^ {0,3}(```+|~~~+) *([^\n\t`~]*)\n([\s\S]*?)\n {0,3}\1[`~]*/gm;
+  const unclosedBlockRegex = /^ {0,3}(```+|~~~+) *([^\n\t`~]*)\n([\s\S]*?)¨0/gm;
+  const emptyBlockRegex    = /^ {0,3}(```+|~~~+) *([^\n\t`~]*)\n {0,3}\1[`~]*/gm;
 
-  text = text.replace(pattern, function (wholeMatch, delim, language, codeblock) {
+  text = text.replace(closedBlockRegex, function (wholeMatch, delim, language, codeblock) {
+    return parse(closedBlockRegex, wholeMatch, delim, language, codeblock);
+  });
+
+  text = text.replace(emptyBlockRegex, function (wholeMatch, delim, language) {
+    console.log('|', wholeMatch, '|>');
+    return parse(emptyBlockRegex, wholeMatch, delim, language, '');
+  });
+
+  text = text.replace(unclosedBlockRegex, function (wholeMatch, delim, language, codeblock) {
+    return parse(unclosedBlockRegex, wholeMatch, delim, language, codeblock);
+  });
+
+  // attacklab: strip sentinel
+  text = text.replace(/¨0/, '');
+
+  let afterEvent = new showdown.Event('makehtml.githubCodeBlock.onEnd', text);
+  afterEvent
+    .setOutput(text)
+    ._setGlobals(globals)
+    ._setOptions(options);
+  afterEvent = globals.converter.dispatch(afterEvent);
+  return afterEvent.output;
+
+
+  function parse (pattern, wholeMatch, delim, language, codeblock) {
     let end = (options.omitExtraWLInCodeBlocks) ? '' : '\n',
         otp,
         attributes = {
@@ -119,16 +147,7 @@ showdown.subParser('makehtml.githubCodeBlock', function (text, options, globals)
     // store the primitive text and the parsed text in a global var,
     // and then return a token
     return '\n\n¨G' + (globals.ghCodeBlocks.push({text: wholeMatch, codeblock: otp}) - 1) + 'G\n\n';
-  });
+  }
 
-  // attacklab: strip sentinel
-  text = text.replace(/¨0/, '');
 
-  let afterEvent = new showdown.Event('makehtml.githubCodeBlock.onEnd', text);
-  afterEvent
-    .setOutput(text)
-    ._setGlobals(globals)
-    ._setOptions(options);
-  afterEvent = globals.converter.dispatch(afterEvent);
-  return afterEvent.output;
 });
diff --git a/test/functional/makehtml/extra.testsuite.commonmark.js b/test/functional/makehtml/extra.testsuite.commonmark.js
index 1194f63..16edca8 100644
--- a/test/functional/makehtml/extra.testsuite.commonmark.js
+++ b/test/functional/makehtml/extra.testsuite.commonmark.js
@@ -10,6 +10,7 @@ let bootstrap = require('./makehtml.bootstrap.js'),
     }),
     assertion = bootstrap.assertion,
     testsuite = bootstrap.getJsonTestSuite('test/functional/makehtml/cases/commonmark.testsuite.json');
+const {tests} = require('commonmark-spec');
 
 describe('makeHtml() commonmark testsuite', function () {
   'use strict';
@@ -24,10 +25,21 @@ describe('makeHtml() commonmark testsuite', function () {
             case 'Setext headings_93': // spec says it cannot be lazy continuation but then proceeds to make it a lazy continuation.
             case 'Thematic breaks_43': // malformed input of test case
             case 'Thematic breaks_61': // hr inside lists does not make sense
+            case 'Fenced code blocks_146': // as of date, github doesn't support this so we don't either
               continue;
+
             case 'Setext headings_91': //it's failing because the testcase converts " to &quot; even though it's not supposed to
               testsuite[section][i].expected = testsuite[section][i].expected.replace(/&quot;/g, '"')
               break;
+
+            case 'Fenced code blocks_142': // we use different classes to mark languages in fenced code blocks
+            case 'Fenced code blocks_143': // we use different classes to mark languages in fenced code blocks
+              testsuite[section][i].expected = testsuite[section][i].expected.replace('language-ruby', 'ruby language-ruby');
+              break;
+
+            case 'Fenced code blocks_144': // we use different classes to mark languages in fenced code blocks
+              testsuite[section][i].expected = testsuite[section][i].expected.replace('language-;', '; language-;');
+              break;
           }
           it(name, assertion(testsuite[section][i], converter, true));
         }