1 files changed, 86 insertions, 38 deletions

diff --git a/js/lib/blocks.js b/js/lib/blocks.js
index c6e5d75..bd00b1a 100644
--- a/js/lib/blocks.js
+++ b/js/lib/blocks.js
@@ -1,35 +1,66 @@
 var Node = require('./node');
 
 var C_GREATERTHAN = 62;
+var C_NEWLINE = 10;
 var C_SPACE = 32;
 var C_OPEN_BRACKET = 91;
 
 var InlineParser = require('./inlines');
+
 var unescapeString = new InlineParser().unescapeString;
 
+var BLOCKTAGNAME = '(?:article|header|aside|hgroup|iframe|blockquote|hr|body|li|map|button|object|canvas|ol|caption|output|col|p|colgroup|pre|dd|progress|div|section|dl|table|td|dt|tbody|embed|textarea|fieldset|tfoot|figcaption|th|figure|thead|footer|footer|tr|form|ul|h1|h2|h3|h4|h5|h6|video|script|style)';
+
+var HTMLBLOCKOPEN = "<(?:" + BLOCKTAGNAME + "[\\s/>]" + "|" +
+        "/" + BLOCKTAGNAME + "[\\s>]" + "|" + "[?!])";
+
+var reHtmlBlockOpen = new RegExp('^' + HTMLBLOCKOPEN, 'i');
+
+var reHrule = /^(?:(?:\* *){3,}|(?:_ *){3,}|(?:- *){3,}) *$/;
+
+var reMaybeSpecial = /^[ #`~*+_=<>0-9-]/;
+
+var reNonSpace = /[^ \t\n]/;
+
+var reBulletListMarker = /^[*+-]( +|$)/;
+
+var reOrderedListMarker = /^(\d+)([.)])( +|$)/;
+
+var reATXHeaderMarker = /^#{1,6}(?: +|$)/;
+
+var reCodeFence = /^`{3,}(?!.*`)|^~{3,}(?!.*~)/;
+
+var reClosingCodeFence = /^(?:`{3,}|~{3,})(?= *$)/;
+
+var reSetextHeaderLine = /^(?:=+|-+) *$/;
+
+var reLineEnding = /\r\n|\n|\r/;
+
 // Returns true if string contains only space characters.
 var isBlank = function(s) {
     "use strict";
-    return /^\s*$/.test(s);
+    return !(reNonSpace.test(s));
 };
 
+var tabSpaces = ['    ', '   ', '  ', ' '];
+
 // Convert tabs to spaces on each line using a 4-space tab stop.
 var detabLine = function(text) {
     "use strict";
-    if (text.indexOf('\u0000') !== -1) {
-        // replace NUL for security
-        text = text.replace(/\0/g, '\uFFFD');
-    }
-    if (text.indexOf('\t') === -1) {
-        return text;
-    } else {
-        var lastStop = 0;
-        return text.replace(/\t/g, function(match, offset) {
-            var result = '    '.slice((offset - lastStop) % 4);
-            lastStop = offset + 1;
-            return result;
-        });
+
+    var start = 0;
+    var offset;
+    var lastStop = 0;
+
+    while ((offset = text.indexOf('\t', start)) !== -1) {
+        var numspaces = (offset - lastStop) % 4;
+        var spaces = tabSpaces[numspaces];
+        text = text.slice(0, offset) + spaces + text.slice(offset + 1);
+        lastStop = offset + numspaces;
+        start = lastStop;
     }
+
+    return text;
 };
 
 // Attempt to match a regex in string s at offset offset.
@@ -44,13 +75,15 @@ var matchAt = function(re, s, offset) {
     }
 };
 
-var BLOCKTAGNAME = '(?:article|header|aside|hgroup|iframe|blockquote|hr|body|li|map|button|object|canvas|ol|caption|output|col|p|colgroup|pre|dd|progress|div|section|dl|table|td|dt|tbody|embed|textarea|fieldset|tfoot|figcaption|th|figure|thead|footer|footer|tr|form|ul|h1|h2|h3|h4|h5|h6|video|script|style)';
-var HTMLBLOCKOPEN = "<(?:" + BLOCKTAGNAME + "[\\s/>]" + "|" +
-        "/" + BLOCKTAGNAME + "[\\s>]" + "|" + "[?!])";
-var reHtmlBlockOpen = new RegExp('^' + HTMLBLOCKOPEN, 'i');
-
-var reHrule = /^(?:(?:\* *){3,}|(?:_ *){3,}|(?:- *){3,}) *$/;
-
+// destructively trip final blank lines in an array of strings
+var stripFinalBlankLines = function(lns) {
+    "use strict";
+    var i = lns.length - 1;
+    while (!reNonSpace.test(lns[i])) {
+        lns.pop();
+        i--;
+    }
+};
 
 // DOC PARSER
 
@@ -160,12 +193,12 @@ var parseListMarker = function(ln, offset) {
     if (rest.match(reHrule)) {
         return null;
     }
-    if ((match = rest.match(/^[*+-]( +|$)/))) {
+    if ((match = rest.match(reBulletListMarker))) {
         spaces_after_marker = match[1].length;
         data.type = 'Bullet';
         data.bullet_char = match[0][0];
 
-    } else if ((match = rest.match(/^(\d+)([.)])( +|$)/))) {
+    } else if ((match = rest.match(reOrderedListMarker))) {
         spaces_after_marker = match[3].length;
         data.type = 'Ordered';
         data.start = parseInt(match[1]);
@@ -214,6 +247,11 @@ var incorporateLine = function(ln, line_number) {
     var container = this.doc;
     var oldtip = this.tip;
 
+    // replace NUL characters for security
+    if (ln.indexOf('\u0000') !== -1) {
+        ln = ln.replace(/\0/g, '\uFFFD');
+    }
+
     // Convert tabs to spaces:
     ln = detabLine(ln);
 
@@ -226,7 +264,7 @@ var incorporateLine = function(ln, line_number) {
         }
         container = container.lastChild;
 
-        match = matchAt(/[^ ]/, ln, offset);
+        match = matchAt(reNonSpace, ln, offset);
         if (match === -1) {
             first_nonspace = ln.length;
             blank = true;
@@ -319,13 +357,11 @@ var incorporateLine = function(ln, line_number) {
     // want to close unmatched blocks.  So we store this closure for
     // use later, when we have more information.
     var closeUnmatchedBlocks = function(mythis) {
-        var already_done = false;
         // finalize any blocks not matched
-        while (!already_done && oldtip !== last_matched_container) {
+        while (oldtip !== last_matched_container) {
             mythis.finalize(oldtip, line_number - 1);
             oldtip = oldtip.parent;
         }
-        already_done = true;
     };
 
     // Check to see if we've hit 2nd blank line; if so break out of list:
@@ -339,9 +375,9 @@ var incorporateLine = function(ln, line_number) {
            container.t !== 'IndentedCode' &&
            container.t !== 'HtmlBlock' &&
            // this is a little performance optimization:
-           matchAt(/^[ #`~*+_=<>0-9-]/, ln, offset) !== -1) {
+           matchAt(reMaybeSpecial, ln, offset) !== -1) {
 
-        match = matchAt(/[^ ]/, ln, offset);
+        match = matchAt(reNonSpace, ln, offset);
         if (match === -1) {
             first_nonspace = ln.length;
             blank = true;
@@ -371,7 +407,7 @@ var incorporateLine = function(ln, line_number) {
             closeUnmatchedBlocks(this);
             container = this.addChild('BlockQuote', line_number, offset);
 
-        } else if ((match = ln.slice(first_nonspace).match(/^#{1,6}(?: +|$)/))) {
+        } else if ((match = ln.slice(first_nonspace).match(reATXHeaderMarker))) {
             // ATX header
             offset = first_nonspace + match[0].length;
             closeUnmatchedBlocks(this);
@@ -382,7 +418,7 @@ var incorporateLine = function(ln, line_number) {
                 [ln.slice(offset).replace(/^ *#+ *$/, '').replace(/ +#+ *$/, '')];
             break;
 
-        } else if ((match = ln.slice(first_nonspace).match(/^`{3,}(?!.*`)|^~{3,}(?!.*~)/))) {
+        } else if ((match = ln.slice(first_nonspace).match(reCodeFence))) {
             // fenced code block
             var fence_length = match[0].length;
             closeUnmatchedBlocks(this);
@@ -402,7 +438,7 @@ var incorporateLine = function(ln, line_number) {
 
         } else if (container.t === 'Paragraph' &&
                    container.strings.length === 1 &&
-                   ((match = ln.slice(first_nonspace).match(/^(?:=+|-+) *$/)))) {
+                   ((match = ln.slice(first_nonspace).match(reSetextHeaderLine)))) {
             // setext header line
             closeUnmatchedBlocks(this);
             container.t = 'Header'; // convert Paragraph to SetextHeader
@@ -447,7 +483,7 @@ var incorporateLine = function(ln, line_number) {
     // What remains at the offset is a text line.  Add the text to the
     // appropriate container.
 
-    match = matchAt(/[^ ]/, ln, offset);
+    match = matchAt(reNonSpace, ln, offset);
     if (match === -1) {
         first_nonspace = ln.length;
         blank = true;
@@ -500,7 +536,7 @@ var incorporateLine = function(ln, line_number) {
             // check for closing code fence:
             match = (indent <= 3 &&
                      ln.charAt(first_nonspace) === container.fence_char &&
-                     ln.slice(first_nonspace).match(/^(?:`{3,}|~{3,})(?= *$)/));
+                     ln.slice(first_nonspace).match(reClosingCodeFence));
             if (match && match[0].length >= container.fence_length) {
                 // don't add closing fence to container; instead, close it:
                 this.finalize(container, line_number);
@@ -569,7 +605,8 @@ var finalize = function(block, line_number) {
         break;
 
     case 'IndentedCode':
-        block.literal = block.strings.join('\n').replace(/(\n *)*$/, '\n');
+        stripFinalBlankLines(block.strings);
+        block.literal = block.strings.join('\n') + '\n';
         block.t = 'CodeBlock';
         break;
 
@@ -644,21 +681,31 @@ var parse = function(input) {
     this.doc = Document();
     this.tip = this.doc;
     this.refmap = {};
-    var lines = input.replace(/\n$/, '').split(/\r\n|\n|\r/);
+    if (this.options.time) { console.time("preparing input"); }
+    var lines = input.split(reLineEnding);
     var len = lines.length;
+    if (input.charCodeAt(input.length - 1) === C_NEWLINE) {
+        // ignore last blank line created by final newline
+        len -= 1;
+    }
+    if (this.options.time) { console.timeEnd("preparing input"); }
+    if (this.options.time) { console.time("block parsing"); }
     for (var i = 0; i < len; i++) {
         this.incorporateLine(lines[i], i + 1);
     }
     while (this.tip) {
         this.finalize(this.tip, len);
     }
+    if (this.options.time) { console.timeEnd("block parsing"); }
+    if (this.options.time) { console.time("inline parsing"); }
     this.processInlines(this.doc);
+    if (this.options.time) { console.timeEnd("inline parsing"); }
     return this.doc;
 };
 
 
 // The DocParser object.
-function DocParser(){
+function DocParser(options){
     "use strict";
     return {
         doc: Document(),
@@ -672,7 +719,8 @@ function DocParser(){
         incorporateLine: incorporateLine,
         finalize: finalize,
         processInlines: processInlines,
-        parse: parse
+        parse: parse,
+        options: options || {}
     };
 }