add options: pedantic, gfm, sanitize.

Author: chjj

README.md

@@ -63,9 +63,31 @@ Along with implementing every markdown feature, marked also implements
 
 ## Usage
 
+``` js
+var html = marked(markdown, options);
+```
+
+### Options
+
+- __pedantic__: Conform to obscure parts of `markdown.pl` as much as possible.
+  Don't fix any of the original markdown bugs or poor behavior.
+- __gfm__: Enabled github flavored markdown (default for backward compatibility).
+- __sanitize__: Sanitize the output. Ignore an HTML that has been input.
+
+None of the above are mutually exclusive/inclusive.
+
+## Example Usage
+
 ``` js
 var marked = require('marked');
 console.log(marked('i am using __markdown__.'));
+
+// gfm
+console.log(marked('```\ni am using GFM.\n```', {
+  gfm: true,
+  pedantic: false,
+  sanitize: true
+}));
 ```
 
 You also have direct access to the lexer and parser if you so desire.

bin/marked

@@ -34,6 +34,7 @@ var help = function() {
 
 var main = function(argv) {
   var files = []
+    , options = {}
     , data = ''
     , input
     , output
@@ -64,6 +65,15 @@ var main = function(argv) {
       case '--tokens':
         tokens = true;
         break;
+      case '--gfm':
+        options.gfm = true;
+        break;
+      case '--sanitize':
+        options.sanitize = true;
+        break;
+      case '--pedantic':
+        options.pedantic = true;
+        break;
       case '-h':
       case '--help':
         return help();
@@ -97,7 +107,7 @@ var main = function(argv) {
   function write() {
     data = tokens
       ? JSON.stringify(marked.lexer(data), null, 2)
-      : marked(data);
+      : marked(data, options);
 
     if (!output) {
       process.stdout.write(data + '\n');

lib/marked.js

@@ -5,6 +5,12 @@
 
 ;(function() {
 
+/**
+ * Options
+ */
+
+var options;
+
 /**
  * Block-Level Grammar
  */
@@ -109,13 +115,15 @@ block.token = function(src, tokens, top) {
       cap = cap[0].replace(/^ {4}/gm, '');
       tokens.push({
         type: 'code',
-        text: cap.replace(/\n+$/, '')
+        text: !options.pedantic
+          ? cap.replace(/\n+$/, '')
+          : cap
       });
       continue;
     }
 
     // gfm_code
-    if (cap = block.gfm_code.exec(src)) {
+    if (options.gfm && (cap = block.gfm_code.exec(src))) {
       src = src.substring(cap[0].length);
       tokens.push({
         type: 'code',
@@ -159,6 +167,7 @@ block.token = function(src, tokens, top) {
     // blockquote
     if (cap = block.blockquote.exec(src)) {
       src = src.substring(cap[0].length);
+
       tokens.push({
         type: 'blockquote_start'
       });
@@ -173,6 +182,7 @@ block.token = function(src, tokens, top) {
       tokens.push({
         type: 'blockquote_end'
       });
+
       continue;
     }
 
@@ -206,7 +216,9 @@ block.token = function(src, tokens, top) {
         // list item contains. Hacky.
         if (~item.indexOf('\n ')) {
           space -= item.length;
-          item = item.replace(new RegExp('^ {1,' + space + '}', 'gm'), '');
+          item = !options.pedantic
+            ? item.replace(new RegExp('^ {1,' + space + '}', 'gm'), '')
+            : item.replace(/^ {1,4}/gm, '');
         }
 
         // Determine whether item is loose or not.
@@ -299,6 +311,11 @@ var inline = {
   nolink: /^!?\[((?:\[[^\]]*\]|[^\[\]])*)\]/,
   strong: /^__([^\0]+?)__(?!_)|^\*\*([^\0]+?)\*\*(?!\*)/,
   em: /^\b_((?:__|[^\0])+?)_\b|^\*((?:\*\*|[^\0])+?)\*(?!\*)/,
+
+  // pedantic strong/em
+  p_strong: /^__(?=\S)([^\0]*?\S)__(?!_)|^\*\*(?=\S)([^\0]*?\S)\*\*(?!\*)/,
+  p_em: /^\b_(?=\S)([^\0]*?\S)_\b|^\*(?=\S)([^\0]*?\S)\*(?!\*)/,
+
   code: /^(`+)([^\0]*?[^`])\1(?!`)/,
   br: /^ {2,}\n(?!\s*$)/,
   text: /^[^\0]+?(?=[\\<!\[_*`]|\w+:\/\/| {2,}\n|$)/
@@ -345,7 +362,7 @@ inline.lexer = function(src) {
     }
 
     // gfm_autolink
-    if (cap = inline.gfm_autolink.exec(src)) {
+    if (options.gfm && (cap = inline.gfm_autolink.exec(src))) {
       src = src.substring(cap[0].length);
       text = escape(cap[1]);
       href = text;
@@ -360,7 +377,9 @@ inline.lexer = function(src) {
     // tag
     if (cap = inline.tag.exec(src)) {
       src = src.substring(cap[0].length);
-      out += cap[0];
+      out += options.sanitize
+        ? escape(cap[0])
+        : cap[0];
       continue;
     }
 
@@ -561,7 +580,7 @@ var tok = function() {
         + '</li>\n';
     }
     case 'html': {
-      return !token.pre
+      return !token.pre && !options.pedantic
         ? inline.lexer(token.text)
         : token.text;
     }
@@ -644,13 +663,26 @@ function tag() {
 }
 
 /**
- * Expose
+ * Marked
  */
 
-var marked = function(src) {
+var marked = function(src, opt) {
+  options = opt || marked.defaults;
   return parse(block.lexer(src));
 };
 
+marked.defaults = {
+  gfm: true,
+  pedantic: false,
+  sanitize: false
+};
+
+options = marked.defaults;
+
+/**
+ * Expose
+ */
+
 marked.parser = parse;
 marked.lexer = block.lexer;
 

man/marked.1

@@ -22,6 +22,16 @@ is specified, read from stdin.
 .BI \-t,\ \-\-tokens
 Output a token stream instead of html.
 .TP
+.BI \-\-pedantic
+Conform to obscure parts of markdown.pl as much as possible. Don't fix original
+markdown bugs.
+.TP
+.BI \-\-gfm
+Enabled github flavored markdown.
+.TP
+.BI \-\-sanitize
+Sanitize output. Ignore any HTML input.
+.TP
 .BI \-h,\ \-\-help
 Display help information.
 .SH EXAMPLES