Home
Diffusion Pillar (wip-open-project-landing) Browse

pillar/src/scripts/markdown/01_markdown-converter.js
ae38bec2180awip-open-project-landing

Last Change
Raw File

01_markdown-converter.js
View Options

"use strict";
var Markdown;
if (typeof exports === "object" && typeof require === "function") // we're in a CommonJS (e.g. Node.js) module
Markdown = exports;
else
Markdown = {};
// The following text is included for historical reasons, but should
// be taken with a pinch of salt; it's not all true anymore.
//
// Wherever possible, Showdown is a straight, line-by-line port
// of the Perl version of Markdown.
//
// This is not a normal parser design; it's basically just a
// series of string substitutions. It's hard to read and
// maintain this way, but keeping Showdown close to the original
// design makes it easier to port new features.
//
// More importantly, Showdown behaves like markdown.pl in most
// edge cases. So web applications can do client-side preview
// in Javascript, and then build identical HTML on the server.
//
// This port needs the new RegExp functionality of ECMA 262,
// 3rd Edition (i.e. Javascript 1.5). Most modern web browsers
// should do fine. Even with the new regular expression features,
// We do a lot of work to emulate Perl's regex functionality.
// The tricky changes in this file mostly have the "attacklab:"
// label. Major or self-explanatory changes don't.
//
// Smart diff tools like Araxis Merge will be able to match up
// this file with markdown.pl in a useful way. A little tweaking
// helps: in a copy of markdown.pl, replace "#" with "//" and
// replace "$text" with "text". Be sure to ignore whitespace
// and line endings.
//
//
// Usage:
//
// var text = "Markdown *rocks*.";
//
// var converter = new Markdown.Converter();
// var html = converter.makeHtml(text);
//
// alert(html);
//
// Note: move the sample code to the bottom of this
// file before uncommenting it.
//
(function () {
function identity(x) { return x; }
function returnFalse(x) { return false; }
function HookCollection() { }
HookCollection.prototype = {
chain: function (hookname, func) {
var original = this[hookname];
if (!original)
throw new Error("unknown hook " + hookname);
if (original === identity)
this[hookname] = func;
else
this[hookname] = function (text) {
var args = Array.prototype.slice.call(arguments, 0);
args[0] = original.apply(null, args);
return func.apply(null, args);
};
},
set: function (hookname, func) {
if (!this[hookname])
throw new Error("unknown hook " + hookname);
this[hookname] = func;
},
addNoop: function (hookname) {
this[hookname] = identity;
},
addFalse: function (hookname) {
this[hookname] = returnFalse;
}
};
Markdown.HookCollection = HookCollection;
// g_urls and g_titles allow arbitrary user-entered strings as keys. This
// caused an exception (and hence stopped the rendering) when the user entered
// e.g. [push] or [__proto__]. Adding a prefix to the actual key prevents this
// (since no builtin property starts with "s_"). See
// http://meta.stackexchange.com/questions/64655/strange-wmd-bug
// (granted, switching from Array() to Object() alone would have left only __proto__
// to be a problem)
function SaveHash() { }
SaveHash.prototype = {
set: function (key, value) {
this["s_" + key] = value;
},
get: function (key) {
return this["s_" + key];
}
};
Markdown.Converter = function (OPTIONS) {
var pluginHooks = this.hooks = new HookCollection();
// given a URL that was encountered by itself (without markup), should return the link text that's to be given to this link
pluginHooks.addNoop("plainLinkText");
// called with the orignal text as given to makeHtml. The result of this plugin hook is the actual markdown source that will be cooked
pluginHooks.addNoop("preConversion");
// called with the text once all normalizations have been completed (tabs to spaces, line endings, etc.), but before any conversions have
pluginHooks.addNoop("postNormalization");
// Called with the text before / after creating block elements like code blocks and lists. Note that this is called recursively
// with inner content, e.g. it's called with the full text, and then only with the content of a blockquote. The inner
// call will receive outdented text.
pluginHooks.addNoop("preBlockGamut");
pluginHooks.addNoop("postBlockGamut");
// called with the text of a single block element before / after the span-level conversions (bold, code spans, etc.) have been made
pluginHooks.addNoop("preSpanGamut");
pluginHooks.addNoop("postSpanGamut");
// called with the final cooked HTML code. The result of this plugin hook is the actual output of makeHtml
pluginHooks.addNoop("postConversion");
//
// Private state of the converter instance:
//
// Global hashes, used by various utility routines
var g_urls;
var g_titles;
var g_html_blocks;
// Used to track when we're inside an ordered or unordered list
// (see _ProcessListItems() for details):
var g_list_level;
OPTIONS = OPTIONS || {};
var asciify = identity, deasciify = identity;
if (OPTIONS.nonAsciiLetters) {
/* In JavaScript regular expressions, \w only denotes [a-zA-Z0-9_].
* That's why there's inconsistent handling e.g. with intra-word bolding
* of Japanese words. That's why we do the following if OPTIONS.nonAsciiLetters
* is true:
*
* Before doing bold and italics, we find every instance
* of a unicode word character in the Markdown source that is not
* matched by \w, and the letter "Q". We take the character's code point
* and encode it in base 51, using the "digits"
*
* A, B, ..., P, R, ..., Y, Z, a, b, ..., y, z
*
* delimiting it with "Q" on both sides. For example, the source
*
* > In Chinese, the smurfs are called ่—็ฒพ้ˆ, meaning "blue spirits".
*
* turns into
*
* > In Chinese, the smurfs are called QNIhQQMOIQQOuUQ, meaning "blue spirits".
*
* Since everything that is a letter in Unicode is now a letter (or
* several letters) in ASCII, \w and \b should always do the right thing.
*
* After the bold/italic conversion, we decode again; since "Q" was encoded
* alongside all non-ascii characters (as "QBfQ"), and the conversion
* will not generate "Q", the only instances of that letter should be our
* encoded characters. And since the conversion will not break words, the
* "Q...Q" should all still be in one piece.
*
* We're using "Q" as the delimiter because it's probably one of the
* rarest characters, and also because I can't think of any special behavior
* that would ever be triggered by this letter (to use a silly example, if we
* delimited with "H" on the left and "P" on the right, then "ฮจ" would be
* encoded as "HTTP", which may cause special behavior). The latter would not
* actually be a huge issue for bold/italic, but may be if we later use it
* in other places as well.
* */
(function () {
var lettersThatJavaScriptDoesNotKnowAndQ = /[Q\u00aa\u00b5\u00ba\u00c0-\u00d6\u00d8-\u00f6\u00f8-\u02c1\u02c6-\u02d1\u02e0-\u02e4\u02ec\u02ee\u0370-\u0374\u0376-\u0377\u037a-\u037d\u0386\u0388-\u038a\u038c\u038e-\u03a1\u03a3-\u03f5\u03f7-\u0481\u048a-\u0523\u0531-\u0556\u0559\u0561-\u0587\u05d0-\u05ea\u05f0-\u05f2\u0621-\u064a\u0660-\u0669\u066e-\u066f\u0671-\u06d3\u06d5\u06e5-\u06e6\u06ee-\u06fc\u06ff\u0710\u0712-\u072f\u074d-\u07a5\u07b1\u07c0-\u07ea\u07f4-\u07f5\u07fa\u0904-\u0939\u093d\u0950\u0958-\u0961\u0966-\u096f\u0971-\u0972\u097b-\u097f\u0985-\u098c\u098f-\u0990\u0993-\u09a8\u09aa-\u09b0\u09b2\u09b6-\u09b9\u09bd\u09ce\u09dc-\u09dd\u09df-\u09e1\u09e6-\u09f1\u0a05-\u0a0a\u0a0f-\u0a10\u0a13-\u0a28\u0a2a-\u0a30\u0a32-\u0a33\u0a35-\u0a36\u0a38-\u0a39\u0a59-\u0a5c\u0a5e\u0a66-\u0a6f\u0a72-\u0a74\u0a85-\u0a8d\u0a8f-\u0a91\u0a93-\u0aa8\u0aaa-\u0ab0\u0ab2-\u0ab3\u0ab5-\u0ab9\u0abd\u0ad0\u0ae0-\u0ae1\u0ae6-\u0aef\u0b05-\u0b0c\u0b0f-\u0b10\u0b13-\u0b28\u0b2a-\u0b30\u0b32-\u0b33\u0b35-\u0b39\u0b3d\u0b5c-\u0b5d\u0b5f-\u0b61\u0b66-\u0b6f\u0b71\u0b83\u0b85-\u0b8a\u0b8e-\u0b90\u0b92-\u0b95\u0b99-\u0b9a\u0b9c\u0b9e-\u0b9f\u0ba3-\u0ba4\u0ba8-\u0baa\u0bae-\u0bb9\u0bd0\u0be6-\u0bef\u0c05-\u0c0c\u0c0e-\u0c10\u0c12-\u0c28\u0c2a-\u0c33\u0c35-\u0c39\u0c3d\u0c58-\u0c59\u0c60-\u0c61\u0c66-\u0c6f\u0c85-\u0c8c\u0c8e-\u0c90\u0c92-\u0ca8\u0caa-\u0cb3\u0cb5-\u0cb9\u0cbd\u0cde\u0ce0-\u0ce1\u0ce6-\u0cef\u0d05-\u0d0c\u0d0e-\u0d10\u0d12-\u0d28\u0d2a-\u0d39\u0d3d\u0d60-\u0d61\u0d66-\u0d6f\u0d7a-\u0d7f\u0d85-\u0d96\u0d9a-\u0db1\u0db3-\u0dbb\u0dbd\u0dc0-\u0dc6\u0e01-\u0e30\u0e32-\u0e33\u0e40-\u0e46\u0e50-\u0e59\u0e81-\u0e82\u0e84\u0e87-\u0e88\u0e8a\u0e8d\u0e94-\u0e97\u0e99-\u0e9f\u0ea1-\u0ea3\u0ea5\u0ea7\u0eaa-\u0eab\u0ead-\u0eb0\u0eb2-\u0eb3\u0ebd\u0ec0-\u0ec4\u0ec6\u0ed0-\u0ed9\u0edc-\u0edd\u0f00\u0f20-\u0f29\u0f40-\u0f47\u0f49-\u0f6c\u0f88-\u0f8b\u1000-\u102a\u103f-\u1049\u1050-\u1055\u105a-\u105d\u1061\u1065-\u1066\u106e-\u1070\u1075-\u1081\u108e\u1090-\u1099\u10a0-\u10c5\u10d0-\u10fa\u10fc\u1100-\u1159\u115f-\u11a2\u11a8-\u11f9\u1200-\u1248\u124a-\u124d\u1250-\u1256\u1258\u125a-\u125d\u1260-\u1288\u128a-\u128d\u1290-\u12b0\u12b2-\u12b5\u12b8-\u12be\u12c0\u12c2-\u12c5\u12c8-\u12d6\u12d8-\u1310\u1312-\u1315\u1318-\u135a\u1380-\u138f\u13a0-\u13f4\u1401-\u166c\u166f-\u1676\u1681-\u169a\u16a0-\u16ea\u1700-\u170c\u170e-\u1711\u1720-\u1731\u1740-\u1751\u1760-\u176c\u176e-\u1770\u1780-\u17b3\u17d7\u17dc\u17e0-\u17e9\u1810-\u1819\u1820-\u1877\u1880-\u18a8\u18aa\u1900-\u191c\u1946-\u196d\u1970-\u1974\u1980-\u19a9\u19c1-\u19c7\u19d0-\u19d9\u1a00-\u1a16\u1b05-\u1b33\u1b45-\u1b4b\u1b50-\u1b59\u1b83-\u1ba0\u1bae-\u1bb9\u1c00-\u1c23\u1c40-\u1c49\u1c4d-\u1c7d\u1d00-\u1dbf\u1e00-\u1f15\u1f18-\u1f1d\u1f20-\u1f45\u1f48-\u1f4d\u1f50-\u1f57\u1f59\u1f5b\u1f5d\u1f5f-\u1f7d\u1f80-\u1fb4\u1fb6-\u1fbc\u1fbe\u1fc2-\u1fc4\u1fc6-\u1fcc\u1fd0-\u1fd3\u1fd6-\u1fdb\u1fe0-\u1fec\u1ff2-\u1ff4\u1ff6-\u1ffc\u203f-\u2040\u2054\u2071\u207f\u2090-\u2094\u2102\u2107\u210a-\u2113\u2115\u2119-\u211d\u2124\u2126\u2128\u212a-\u212d\u212f-\u2139\u213c-\u213f\u2145-\u2149\u214e\u2183-\u2184\u2c00-\u2c2e\u2c30-\u2c5e\u2c60-\u2c6f\u2c71-\u2c7d\u2c80-\u2ce4\u2d00-\u2d25\u2d30-\u2d65\u2d6f\u2d80-\u2d96\u2da0-\u2da6\u2da8-\u2dae\u2db0-\u2db6\u2db8-\u2dbe\u2dc0-\u2dc6\u2dc8-\u2dce\u2dd0-\u2dd6\u2dd8-\u2dde\u2e2f\u3005-\u3006\u3031-\u3035\u303b-\u303c\u3041-\u3096\u309d-\u309f\u30a1-\u30fa\u30fc-\u30ff\u3105-\u312d\u3131-\u318e\u31a0-\u31b7\u31f0-\u31ff\u3400-\u4db5\u4e00-\u9fc3\ua000-\ua48c\ua500-\ua60c\ua610-\ua62b\ua640-\ua65f\ua662-\ua66e\ua67f-\ua697\ua717-\ua71f\ua722-\ua788\ua78b-\ua78c\ua7fb-\ua801\ua803-\ua805\ua807-\ua80a\ua80c-\ua822\ua840-\ua873\ua882-\ua8b3\ua8d0-\ua8d9\ua900-\ua925\ua930-\ua946\uaa00-\uaa28\uaa40-\uaa42\uaa44-\uaa4b\uaa50-\uaa59\uac00-\ud7a3\uf900-\ufa2d\ufa30-\ufa6a\ufa70-\ufad9\ufb00-\ufb06\ufb13-\ufb17\ufb1d\ufb1f-\ufb28\ufb2a-\ufb36\ufb38-\ufb3c\ufb3e\ufb40-\ufb41\ufb43-\ufb44\ufb46-\ufbb1\ufbd3-\ufd3d\ufd50-\ufd8f\ufd92-\ufdc7\ufdf0-\ufdfb\ufe33-\ufe34\ufe4d-\ufe4f\ufe70-\ufe74\ufe76-\ufefc\uff10-\uff19\uff21-\uff3a\uff3f\uff41-\uff5a\uff66-\uffbe\uffc2-\uffc7\uffca-\uffcf\uffd2-\uffd7\uffda-\uffdc]/g;
var cp_Q = "Q".charCodeAt(0);
var cp_A = "A".charCodeAt(0);
var cp_Z = "Z".charCodeAt(0);
var dist_Za = "a".charCodeAt(0) - cp_Z - 1;
asciify = function(text) {
return text.replace(lettersThatJavaScriptDoesNotKnowAndQ, function (m) {
var c = m.charCodeAt(0);
var s = "";
var v;
while (c > 0) {
v = (c % 51) + cp_A;
if (v >= cp_Q)
v++;
if (v > cp_Z)
v += dist_Za;
s = String.fromCharCode(v) + s;
c = c / 51 | 0;
}
return "Q" + s + "Q";
})
};
deasciify = function(text) {
return text.replace(/Q([A-PR-Za-z]{1,3})Q/g, function (m, s) {
var c = 0;
var v;
for (var i = 0; i < s.length; i++) {
v = s.charCodeAt(i);
if (v > cp_Z)
v -= dist_Za;
if (v > cp_Q)
v--;
v -= cp_A;
c = (c * 51) + v;
}
return String.fromCharCode(c);
})
}
})();
}
var _DoItalicsAndBold = OPTIONS.asteriskIntraWordEmphasis ? _DoItalicsAndBold_AllowIntrawordWithAsterisk : _DoItalicsAndBoldStrict;
this.makeHtml = function (text) {
//
// Main function. The order in which other subs are called here is
// essential. Link and image substitutions need to happen before
// _EscapeSpecialCharsWithinTagAttributes(), so that any *'s or _'s in the <a>
// and <img> tags get encoded.
//
// This will only happen if makeHtml on the same converter instance is called from a plugin hook.
// Don't do that.
if (g_urls)
throw new Error("Recursive call to converter.makeHtml");
// Create the private state objects.
g_urls = new SaveHash();
g_titles = new SaveHash();
g_html_blocks = [];
g_list_level = 0;
text = pluginHooks.preConversion(text);
// attacklab: Replace ~ with ~T
// This lets us use tilde as an escape char to avoid md5 hashes
// The choice of character is arbitray; anything that isn't
// magic in Markdown will work.
text = text.replace(/~/g, "~T");
// attacklab: Replace $ with ~D
// RegExp interprets $ as a special character
// when it's in a replacement string
text = text.replace(/\$/g, "~D");
// Standardize line endings
text = text.replace(/\r\n/g, "\n"); // DOS to Unix
text = text.replace(/\r/g, "\n"); // Mac to Unix
// Make sure text begins and ends with a couple of newlines:
text = "\n\n" + text + "\n\n";
// Convert all tabs to spaces.
text = _Detab(text);
// Strip any lines consisting only of spaces and tabs.
// This makes subsequent regexen easier to write, because we can
// match consecutive blank lines with /\n+/ instead of something
// contorted like /[ \t]*\n+/ .
text = text.replace(/^[ \t]+$/mg, "");
text = pluginHooks.postNormalization(text);
// Turn block-level HTML blocks into hash entries
text = _HashHTMLBlocks(text);
// Strip link definitions, store in hashes.
text = _StripLinkDefinitions(text);
text = _RunBlockGamut(text);
text = _UnescapeSpecialChars(text);
// attacklab: Restore dollar signs
text = text.replace(/~D/g, "$$");
// attacklab: Restore tildes
text = text.replace(/~T/g, "~");
text = pluginHooks.postConversion(text);
g_html_blocks = g_titles = g_urls = null;
return text;
};
function _StripLinkDefinitions(text) {
//
// Strips link definitions from text, stores the URLs and titles in
// hash references.
//
// Link defs are in the form: ^[id]: url "optional title"
/*
text = text.replace(/
^[ ]{0,3}\[([^\[\]]+)\]: // id = $1 attacklab: g_tab_width - 1
[ \t]*
\n? // maybe *one* newline
[ \t]*
<?(\S+?)>? // url = $2
(?=\s|$) // lookahead for whitespace instead of the lookbehind removed below
[ \t]*
\n? // maybe one newline
[ \t]*
( // (potential) title = $3
(\n*) // any lines skipped = $4 attacklab: lookbehind removed
[ \t]+
["(]
(.+?) // title = $5
[")]
[ \t]*
)? // title is optional
(\n+) // subsequent newlines = $6, capturing because they must be put back if the potential title isn't an actual title
/gm, function(){...});
*/
text = text.replace(/^[ ]{0,3}\[([^\[\]]+)\]:[ \t]*\n?[ \t]*<?(\S+?)>?(?=\s|$)[ \t]*\n?[ \t]*((\n*)["(](.+?)[")][ \t]*)?(\n+)/gm,
function (wholeMatch, m1, m2, m3, m4, m5, m6) {
m1 = m1.toLowerCase();
g_urls.set(m1, _EncodeAmpsAndAngles(m2)); // Link IDs are case-insensitive
if (m4) {
// Oops, found blank lines, so it's not a title.
// Put back the parenthetical statement we stole.
return m3 + m6;
} else if (m5) {
g_titles.set(m1, m5.replace(/"/g, "&quot;"));
}
// Completely remove the definition from the text
return "";
}
);
return text;
}
function _HashHTMLBlocks(text) {
// Hashify HTML blocks:
// We only want to do this for block-level HTML tags, such as headers,
// lists, and tables. That's because we still want to wrap <p>s around
// "paragraphs" that are wrapped in non-block-level tags, such as anchors,
// phrase emphasis, and spans. The list of tags we're looking for is
// hard-coded:
var block_tags_a = "p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|math|ins|del"
var block_tags_b = "p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|math"
// First, look for nested blocks, e.g.:
// <div>
// <div>
// tags for inner block must be indented.
// </div>
// </div>
//
// The outermost tags must start at the left margin for this to match, and
// the inner nested divs must be indented.
// We need to do this before the next, more liberal match, because the next
// match will start at the first `<div>` and stop at the first `</div>`.
// attacklab: This regex can be expensive when it fails.
/*
text = text.replace(/
( // save in $1
^ // start of line (with /m)
<($block_tags_a) // start tag = $2
\b // word break
// attacklab: hack around khtml/pcre bug...
[^\r]*?\n // any number of lines, minimally matching
</\2> // the matching end tag
[ \t]* // trailing spaces/tabs
(?=\n+) // followed by a newline
) // attacklab: there are sentinel newlines at end of document
/gm,function(){...}};
*/
text = text.replace(/^(<(p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|math|ins|del)\b[^\r]*?\n<\/\2>[ \t]*(?=\n+))/gm, hashMatch);
//
// Now match more liberally, simply from `\n<tag>` to `</tag>\n`
//
/*
text = text.replace(/
( // save in $1
^ // start of line (with /m)
<($block_tags_b) // start tag = $2
\b // word break
// attacklab: hack around khtml/pcre bug...
[^\r]*? // any number of lines, minimally matching
.*</\2> // the matching end tag
[ \t]* // trailing spaces/tabs
(?=\n+) // followed by a newline
) // attacklab: there are sentinel newlines at end of document
/gm,function(){...}};
*/
text = text.replace(/^(<(p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|math)\b[^\r]*?.*<\/\2>[ \t]*(?=\n+)\n)/gm, hashMatch);
// Special case just for <hr />. It was easier to make a special case than
// to make the other regex more complicated.
/*
text = text.replace(/
\n // Starting after a blank line
[ ]{0,3}
( // save in $1
(<(hr) // start tag = $2
\b // word break
([^<>])*?
\/?>) // the matching end tag
[ \t]*
(?=\n{2,}) // followed by a blank line
)
/g,hashMatch);
*/
text = text.replace(/\n[ ]{0,3}((<(hr)\b([^<>])*?\/?>)[ \t]*(?=\n{2,}))/g, hashMatch);
// Special case for standalone HTML comments:
/*
text = text.replace(/
\n\n // Starting after a blank line
[ ]{0,3} // attacklab: g_tab_width - 1
( // save in $1
<!
(--(?:|(?:[^>-]|-[^>])(?:[^-]|-[^-])*)--) // see http://www.w3.org/TR/html-markup/syntax.html#comments and http://meta.stackexchange.com/q/95256
>
[ \t]*
(?=\n{2,}) // followed by a blank line
)
/g,hashMatch);
*/
text = text.replace(/\n\n[ ]{0,3}(<!(--(?:|(?:[^>-]|-[^>])(?:[^-]|-[^-])*)--)>[ \t]*(?=\n{2,}))/g, hashMatch);
// PHP and ASP-style processor instructions (<?...?> and <%...%>)
/*
text = text.replace(/
(?:
\n\n // Starting after a blank line
)
( // save in $1
[ ]{0,3} // attacklab: g_tab_width - 1
(?:
<([?%]) // $2
[^\r]*?
\2>
)
[ \t]*
(?=\n{2,}) // followed by a blank line
)
/g,hashMatch);
*/
text = text.replace(/(?:\n\n)([ ]{0,3}(?:<([?%])[^\r]*?\2>)[ \t]*(?=\n{2,}))/g, hashMatch);
return text;
}
function hashBlock(text) {
text = text.replace(/(^\n+|\n+$)/g, "");
// Replace the element text with a marker ("~KxK" where x is its key)
return "\n\n~K" + (g_html_blocks.push(text) - 1) + "K\n\n";
}
function hashMatch(wholeMatch, m1) {
return hashBlock(m1);
}
var blockGamutHookCallback = function (t) { return _RunBlockGamut(t); }
function _RunBlockGamut(text, doNotUnhash, doNotCreateParagraphs) {
//
// These are all the transformations that form block-level
// tags like paragraphs, headers, and list items.
//
text = pluginHooks.preBlockGamut(text, blockGamutHookCallback);
text = _DoHeaders(text);
// Do Horizontal Rules:
var replacement = "<hr />\n";
text = text.replace(/^[ ]{0,2}([ ]?\*[ ]?){3,}[ \t]*$/gm, replacement);
text = text.replace(/^[ ]{0,2}([ ]?-[ ]?){3,}[ \t]*$/gm, replacement);
text = text.replace(/^[ ]{0,2}([ ]?_[ ]?){3,}[ \t]*$/gm, replacement);
text = _DoLists(text);
text = _DoCodeBlocks(text);
text = _DoBlockQuotes(text);
text = pluginHooks.postBlockGamut(text, blockGamutHookCallback);
// We already ran _HashHTMLBlocks() before, in Markdown(), but that
// was to escape raw HTML in the original Markdown source. This time,
// we're escaping the markup we've just created, so that we don't wrap
// <p> tags around block-level tags.
text = _HashHTMLBlocks(text);
text = _FormParagraphs(text, doNotUnhash, doNotCreateParagraphs);
return text;
}
function _RunSpanGamut(text) {
//
// These are all the transformations that occur *within* block-level
// tags like paragraphs, headers, and list items.
//
text = pluginHooks.preSpanGamut(text);
text = _DoCodeSpans(text);
text = _EscapeSpecialCharsWithinTagAttributes(text);
text = _EncodeBackslashEscapes(text);
// Process anchor and image tags. Images must come first,
// because ![foo][f] looks like an anchor.
text = _DoImages(text);
text = _DoAnchors(text);
// Make links out of things like `<http://example.com/>`
// Must come after _DoAnchors(), because you can use < and >
// delimiters in inline links like [this](<url>).
text = _DoAutoLinks(text);
text = text.replace(/~P/g, "://"); // put in place to prevent autolinking; reset now
text = _EncodeAmpsAndAngles(text);
text = _DoItalicsAndBold(text);
// Do hard breaks:
text = text.replace(/ +\n/g, " <br>\n");
text = pluginHooks.postSpanGamut(text);
return text;
}
function _EscapeSpecialCharsWithinTagAttributes(text) {
//
// Within tags -- meaning between < and > -- encode [\ ` * _] so they
// don't conflict with their use in Markdown for code, italics and strong.
//
// Build a regex to find HTML tags and comments. See Friedl's
// "Mastering Regular Expressions", 2nd Ed., pp. 200-201.
// SE: changed the comment part of the regex
var regex = /(<[a-z\/!$]("[^"]*"|'[^']*'|[^'">])*>|<!(--(?:|(?:[^>-]|-[^>])(?:[^-]|-[^-])*)--)>)/gi;
text = text.replace(regex, function (wholeMatch) {
var tag = wholeMatch.replace(/(.)<\/?code>(?=.)/g, "$1`");
tag = escapeCharacters(tag, wholeMatch.charAt(1) == "!" ? "\\`*_/" : "\\`*_"); // also escape slashes in comments to prevent autolinking there -- http://meta.stackexchange.com/questions/95987
return tag;
});
return text;
}
function _DoAnchors(text) {
if (text.indexOf("[") === -1)
return text;
//
// Turn Markdown link shortcuts into XHTML <a> tags.
//
//
// First, handle reference-style links: [link text] [id]
//
/*
text = text.replace(/
( // wrap whole match in $1
\[
(
(?:
\[[^\]]*\] // allow brackets nested one level
|
[^\[] // or anything else
)*
)
\]
[ ]? // one optional space
(?:\n[ ]*)? // one optional newline followed by spaces
\[
(.*?) // id = $3
\]
)
()()()() // pad remaining backreferences
/g, writeAnchorTag);
*/
text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g, writeAnchorTag);
//
// Next, inline-style links: [link text](url "optional title")
//
/*
text = text.replace(/
( // wrap whole match in $1
\[
(
(?:
\[[^\]]*\] // allow brackets nested one level
|
[^\[\]] // or anything else
)*
)
\]
\( // literal paren
[ \t]*
() // no id, so leave $3 empty
<?( // href = $4
(?:
\([^)]*\) // allow one level of (correctly nested) parens (think MSDN)
|
[^()\s]
)*?
)>?
[ \t]*
( // $5
(['"]) // quote char = $6
(.*?) // Title = $7
\6 // matching quote
[ \t]* // ignore any spaces/tabs between closing quote and )
)? // title is optional
\)
)
/g, writeAnchorTag);
*/
text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\]\([ \t]*()<?((?:\([^)]*\)|[^()\s])*?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g, writeAnchorTag);
//
// Last, handle reference-style shortcuts: [link text]
// These must come last in case you've also got [link test][1]
// or [link test](/foo)
//
/*
text = text.replace(/
( // wrap whole match in $1
\[
([^\[\]]+) // link text = $2; can't contain '[' or ']'
\]
)
()()()()() // pad rest of backreferences
/g, writeAnchorTag);
*/
text = text.replace(/(\[([^\[\]]+)\])()()()()()/g, writeAnchorTag);
return text;
}
function writeAnchorTag(wholeMatch, m1, m2, m3, m4, m5, m6, m7) {
if (m7 == undefined) m7 = "";
var whole_match = m1;
var link_text = m2.replace(/:\/\//g, "~P"); // to prevent auto-linking withing the link. will be converted back after the auto-linker runs
var link_id = m3.toLowerCase();
var url = m4;
var title = m7;
if (url == "") {
if (link_id == "") {
// lower-case and turn embedded newlines into spaces
link_id = link_text.toLowerCase().replace(/ ?\n/g, " ");
}
url = "#" + link_id;
if (g_urls.get(link_id) != undefined) {
url = g_urls.get(link_id);
if (g_titles.get(link_id) != undefined) {
title = g_titles.get(link_id);
}
}
else {
if (whole_match.search(/\(\s*\)$/m) > -1) {
// Special case for explicit empty url
url = "";
} else {
return whole_match;
}
}
}
url = attributeSafeUrl(url);
var result = "<a href=\"" + url + "\"";
if (title != "") {
title = attributeEncode(title);
title = escapeCharacters(title, "*_");
result += " title=\"" + title + "\"";
}
result += ">" + link_text + "</a>";
return result;
}
function _DoImages(text) {
if (text.indexOf("![") === -1)
return text;
//
// Turn Markdown image shortcuts into <img> tags.
//
//
// First, handle reference-style labeled images: ![alt text][id]
//
/*
text = text.replace(/
( // wrap whole match in $1
!\[
(.*?) // alt text = $2
\]
[ ]? // one optional space
(?:\n[ ]*)? // one optional newline followed by spaces
\[
(.*?) // id = $3
\]
)
()()()() // pad rest of backreferences
/g, writeImageTag);
*/
text = text.replace(/(!\[(.*?)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g, writeImageTag);
//
// Next, handle inline images: ![alt text](url "optional title")
// Don't forget: encode * and _
/*
text = text.replace(/
( // wrap whole match in $1
!\[
(.*?) // alt text = $2
\]
\s? // One optional whitespace character
\( // literal paren
[ \t]*
() // no id, so leave $3 empty
<?(\S+?)>? // src url = $4
[ \t]*
( // $5
(['"]) // quote char = $6
(.*?) // title = $7
\6 // matching quote
[ \t]*
)? // title is optional
\)
)
/g, writeImageTag);
*/
text = text.replace(/(!\[(.*?)\]\s?\([ \t]*()<?(\S+?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g, writeImageTag);
return text;
}
function attributeEncode(text) {
// unconditionally replace angle brackets here -- what ends up in an attribute (e.g. alt or title)
// never makes sense to have verbatim HTML in it (and the sanitizer would totally break it)
return text.replace(/>/g, "&gt;").replace(/</g, "&lt;").replace(/"/g, "&quot;").replace(/'/g, "&#39;");
}
function writeImageTag(wholeMatch, m1, m2, m3, m4, m5, m6, m7) {
var whole_match = m1;
var alt_text = m2;
var link_id = m3.toLowerCase();
var url = m4;
var title = m7;
if (!title) title = "";
if (url == "") {
if (link_id == "") {
// lower-case and turn embedded newlines into spaces
link_id = alt_text.toLowerCase().replace(/ ?\n/g, " ");
}
url = "#" + link_id;
if (g_urls.get(link_id) != undefined) {
url = g_urls.get(link_id);
if (g_titles.get(link_id) != undefined) {
title = g_titles.get(link_id);
}
}
else {
return whole_match;
}
}
alt_text = escapeCharacters(attributeEncode(alt_text), "*_[]()");
url = escapeCharacters(url, "*_");
var result = "<img src=\"" + url + "\" alt=\"" + alt_text + "\"";
// attacklab: Markdown.pl adds empty title attributes to images.
// Replicate this bug.
//if (title != "") {
title = attributeEncode(title);
title = escapeCharacters(title, "*_");
result += " title=\"" + title + "\"";
//}
result += " />";
return result;
}
function _DoHeaders(text) {
// Setext-style headers:
// Header 1
// ========
//
// Header 2
// --------
//
text = text.replace(/^(.+)[ \t]*\n=+[ \t]*\n+/gm,
function (wholeMatch, m1) { return "<h1>" + _RunSpanGamut(m1) + "</h1>\n\n"; }
);
text = text.replace(/^(.+)[ \t]*\n-+[ \t]*\n+/gm,
function (matchFound, m1) { return "<h2>" + _RunSpanGamut(m1) + "</h2>\n\n"; }
);
// atx-style headers:
// # Header 1
// ## Header 2
// ## Header 2 with closing hashes ##
// ...
// ###### Header 6
//
/*
text = text.replace(/
^(\#{1,6}) // $1 = string of #'s
[ \t]*
(.+?) // $2 = Header text
[ \t]*
\#* // optional closing #'s (not counted)
\n+
/gm, function() {...});
*/
text = text.replace(/^(\#{1,6})[ \t]*(.+?)[ \t]*\#*\n+/gm,
function (wholeMatch, m1, m2) {
var h_level = m1.length;
return "<h" + h_level + ">" + _RunSpanGamut(m2) + "</h" + h_level + ">\n\n";
}
);
return text;
}
function _DoLists(text, isInsideParagraphlessListItem) {
//
// Form HTML ordered (numbered) and unordered (bulleted) lists.
//
// attacklab: add sentinel to hack around khtml/safari bug:
// http://bugs.webkit.org/show_bug.cgi?id=11231
text += "~0";
// Re-usable pattern to match any entirel ul or ol list:
/*
var whole_list = /
( // $1 = whole list
( // $2
[ ]{0,3} // attacklab: g_tab_width - 1
([*+-]|\d+[.]) // $3 = first list item marker
[ \t]+
)
[^\r]+?
( // $4
~0 // sentinel for workaround; should be $
|
\n{2,}
(?=\S)
(?! // Negative lookahead for another list item marker
[ \t]*
(?:[*+-]|\d+[.])[ \t]+
)
)
)
/g
*/
var whole_list = /^(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/gm;
if (g_list_level) {
text = text.replace(whole_list, function (wholeMatch, m1, m2) {
var list = m1;
var list_type = (m2.search(/[*+-]/g) > -1) ? "ul" : "ol";
var first_number;
if (list_type === "ol")
first_number = parseInt(m2, 10)
var result = _ProcessListItems(list, list_type, isInsideParagraphlessListItem);
// Trim any trailing whitespace, to put the closing `</$list_type>`
// up on the preceding line, to get it past the current stupid
// HTML block parser. This is a hack to work around the terrible
// hack that is the HTML block parser.
result = result.replace(/\s+$/, "");
var opening = "<" + list_type;
if (first_number && first_number !== 1)
opening += " start=\"" + first_number + "\"";
result = opening + ">" + result + "</" + list_type + ">\n";
return result;
});
} else {
whole_list = /(\n\n|^\n?)(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/g;
text = text.replace(whole_list, function (wholeMatch, m1, m2, m3) {
var runup = m1;
var list = m2;
var list_type = (m3.search(/[*+-]/g) > -1) ? "ul" : "ol";
var first_number;
if (list_type === "ol")
first_number = parseInt(m3, 10)
var result = _ProcessListItems(list, list_type);
var opening = "<" + list_type;
if (first_number && first_number !== 1)
opening += " start=\"" + first_number + "\"";
result = runup + opening + ">\n" + result + "</" + list_type + ">\n";
return result;
});
}
// attacklab: strip sentinel
text = text.replace(/~0/, "");
return text;
}
var _listItemMarkers = { ol: "\\d+[.]", ul: "[*+-]" };
function _ProcessListItems(list_str, list_type, isInsideParagraphlessListItem) {
//
// Process the contents of a single ordered or unordered list, splitting it
// into individual list items.
//
// list_type is either "ul" or "ol".
// The $g_list_level global keeps track of when we're inside a list.
// Each time we enter a list, we increment it; when we leave a list,
// we decrement. If it's zero, we're not in a list anymore.
//
// We do this because when we're not inside a list, we want to treat
// something like this:
//
// I recommend upgrading to version
// 8. Oops, now this line is treated
// as a sub-list.
//
// As a single paragraph, despite the fact that the second line starts
// with a digit-period-space sequence.
//
// Whereas when we're inside a list (or sub-list), that line will be
// treated as the start of a sub-list. What a kludge, huh? This is
// an aspect of Markdown's syntax that's hard to parse perfectly
// without resorting to mind-reading. Perhaps the solution is to
// change the syntax rules such that sub-lists must start with a
// starting cardinal number; e.g. "1." or "a.".
g_list_level++;
// trim trailing blank lines:
list_str = list_str.replace(/\n{2,}$/, "\n");
// attacklab: add sentinel to emulate \z
list_str += "~0";
// In the original attacklab showdown, list_type was not given to this function, and anything
// that matched /[*+-]|\d+[.]/ would just create the next <li>, causing this mismatch:
//
// Markdown rendered by WMD rendered by MarkdownSharp
// ------------------------------------------------------------------
// 1. first 1. first 1. first
// 2. second 2. second 2. second
// - third 3. third * third
//
// We changed this to behave identical to MarkdownSharp. This is the constructed RegEx,
// with {MARKER} being one of \d+[.] or [*+-], depending on list_type:
/*
list_str = list_str.replace(/
(^[ \t]*) // leading whitespace = $1
({MARKER}) [ \t]+ // list marker = $2
([^\r]+? // list item text = $3
(\n+)
)
(?=
(~0 | \2 ({MARKER}) [ \t]+)
)
/gm, function(){...});
*/
var marker = _listItemMarkers[list_type];
var re = new RegExp("(^[ \\t]*)(" + marker + ")[ \\t]+([^\\r]+?(\\n+))(?=(~0|\\1(" + marker + ")[ \\t]+))", "gm");
var last_item_had_a_double_newline = false;
list_str = list_str.replace(re,
function (wholeMatch, m1, m2, m3) {
var item = m3;
var leading_space = m1;
var ends_with_double_newline = /\n\n$/.test(item);
var contains_double_newline = ends_with_double_newline || item.search(/\n{2,}/) > -1;
var loose = contains_double_newline || last_item_had_a_double_newline;
item = _RunBlockGamut(_Outdent(item), /* doNotUnhash = */true, /* doNotCreateParagraphs = */ !loose);
last_item_had_a_double_newline = ends_with_double_newline;
return "<li>" + item + "</li>\n";
}
);
// attacklab: strip sentinel
list_str = list_str.replace(/~0/g, "");
g_list_level--;
return list_str;
}
function _DoCodeBlocks(text) {
//
// Process Markdown `<pre><code>` blocks.
//
/*
text = text.replace(/
(?:\n\n|^)
( // $1 = the code block -- one or more lines, starting with a space/tab
(?:
(?:[ ]{4}|\t) // Lines must start with a tab or a tab-width of spaces - attacklab: g_tab_width
.*\n+
)+
)
(\n*[ ]{0,3}[^ \t\n]|(?=~0)) // attacklab: g_tab_width
/g ,function(){...});
*/
// attacklab: sentinel workarounds for lack of \A and \Z, safari\khtml bug
text += "~0";
text = text.replace(/(?:\n\n|^\n?)((?:(?:[ ]{4}|\t).*\n+)+)(\n*[ ]{0,3}[^ \t\n]|(?=~0))/g,
function (wholeMatch, m1, m2) {
var codeblock = m1;
var nextChar = m2;
codeblock = _EncodeCode(_Outdent(codeblock));
codeblock = _Detab(codeblock);
codeblock = codeblock.replace(/^\n+/g, ""); // trim leading newlines
codeblock = codeblock.replace(/\n+$/g, ""); // trim trailing whitespace
codeblock = "<pre><code>" + codeblock + "\n</code></pre>";
return "\n\n" + codeblock + "\n\n" + nextChar;
}
);
// attacklab: strip sentinel
text = text.replace(/~0/, "");
return text;
}
function _DoCodeSpans(text) {
//
// * Backtick quotes are used for <code></code> spans.
//
// * You can use multiple backticks as the delimiters if you want to
// include literal backticks in the code span. So, this input:
//
// Just type ``foo `bar` baz`` at the prompt.
//
// Will translate to:
//
// <p>Just type <code>foo `bar` baz</code> at the prompt.</p>
//
// There's no arbitrary limit to the number of backticks you
// can use as delimters. If you need three consecutive backticks
// in your code, use four for delimiters, etc.
//
// * You can use spaces to get literal backticks at the edges:
//
// ... type `` `bar` `` ...
//
// Turns to:
//
// ... type <code>`bar`</code> ...
//
/*
text = text.replace(/
(^|[^\\`]) // Character before opening ` can't be a backslash or backtick
(`+) // $2 = Opening run of `
(?!`) // and no more backticks -- match the full run
( // $3 = The code block
[^\r]*?
[^`] // attacklab: work around lack of lookbehind
)
\2 // Matching closer
(?!`)
/gm, function(){...});
*/
text = text.replace(/(^|[^\\`])(`+)(?!`)([^\r]*?[^`])\2(?!`)/gm,
function (wholeMatch, m1, m2, m3, m4) {
var c = m3;
c = c.replace(/^([ \t]*)/g, ""); // leading whitespace
c = c.replace(/[ \t]*$/g, ""); // trailing whitespace
c = _EncodeCode(c);
c = c.replace(/:\/\//g, "~P"); // to prevent auto-linking. Not necessary in code *blocks*, but in code spans. Will be converted back after the auto-linker runs.
return m1 + "<code>" + c + "</code>";
}
);
return text;
}
function _EncodeCode(text) {
//
// Encode/escape certain characters inside Markdown code runs.
// The point is that in code, these characters are literals,
// and lose their special Markdown meanings.
//
// Encode all ampersands; HTML entities are not
// entities within a Markdown code span.
text = text.replace(/&/g, "&amp;");
// Do the angle bracket song and dance:
text = text.replace(/</g, "&lt;");
text = text.replace(/>/g, "&gt;");
// Now, escape characters that are magic in Markdown:
text = escapeCharacters(text, "\*_{}[]\\", false);
// jj the line above breaks this:
//---
//* Item
// 1. Subitem
// special char: *
//---
return text;
}
function _DoItalicsAndBoldStrict(text) {
if (text.indexOf("*") === -1 && text.indexOf("_") === - 1)
return text;
text = asciify(text);
// <strong> must go first:
// (^|[\W_]) Start with a non-letter or beginning of string. Store in \1.
// (?:(?!\1)|(?=^)) Either the next character is *not* the same as the previous,
// or we started at the end of the string (in which case the previous
// group had zero width, so we're still there). Because the next
// character is the marker, this means that if there are e.g. multiple
// underscores in a row, we can only match the left-most ones (which
// prevents foo___bar__ from getting bolded)
// (\*|_) The marker character itself, asterisk or underscore. Store in \2.
// \2 The marker again, since bold needs two.
// (?=\S) The first bolded character cannot be a space.
// ([^\r]*?\S) The actual bolded string. At least one character, and it cannot *end*
// with a space either. Note that like in many other places, [^\r] is
// just a workaround for JS' lack of single-line regexes; it's equivalent
// to a . in an /s regex, because the string cannot contain any \r (they
// are removed in the normalizing step).
// \2\2 The marker character, twice -- end of bold.
// (?!\2) Not followed by another marker character (ensuring that we match the
// rightmost two in a longer row)...
// (?=[\W_]|$) ...but by any other non-word character or the end of string.
text = text.replace(/(^|[\W_])(?:(?!\1)|(?=^))(\*|_)\2(?=\S)([^\r]*?\S)\2\2(?!\2)(?=[\W_]|$)/g,
"$1<strong>$3</strong>");
// This is almost identical to the <strong> regex, except 1) there's obviously just one marker
// character, and 2) the italicized string cannot contain the marker character.
text = text.replace(/(^|[\W_])(?:(?!\1)|(?=^))(\*|_)(?=\S)((?:(?!\2)[^\r])*?\S)\2(?!\2)(?=[\W_]|$)/g,
"$1<em>$3</em>");
return deasciify(text);
}
function _DoItalicsAndBold_AllowIntrawordWithAsterisk(text) {
if (text.indexOf("*") === -1 && text.indexOf("_") === - 1)
return text;
text = asciify(text);
// <strong> must go first:
// (?=[^\r][*_]|[*_]) Optimization only, to find potentially relevant text portions faster. Minimally slower in Chrome, but much faster in IE.
// ( Store in \1. This is the last character before the delimiter
// ^ Either we're at the start of the string (i.e. there is no last character)...
// | ... or we allow one of the following:
// (?= (lookahead; we're not capturing this, just listing legal possibilities)
// \W__ If the delimiter is __, then this last character must be non-word non-underscore (extra-word emphasis only)
// |
// (?!\*)[\W_]\*\* If the delimiter is **, then this last character can be non-word non-asterisk (extra-word emphasis)...
// |
// \w\*\*\w ...or it can be word/underscore, but only if the first bolded character is such a character as well (intra-word emphasis)
// )
// [^\r] actually capture the character (can't use `.` since it could be \n)
// )
// (\*\*|__) Store in \2: the actual delimiter
// (?!\2) not followed by the delimiter again (at most one more asterisk/underscore is allowed)
// (?=\S) the first bolded character can't be a space
// ( Store in \3: the bolded string
//
// (?:| Look at all bolded characters except for the last one. Either that's empty, meaning only a single character was bolded...
// [^\r]*? ... otherwise take arbitrary characters, minimally matching; that's all bolded characters except for the last *two*
// (?!\2) the last two characters cannot be the delimiter itself (because that would mean four underscores/asterisks in a row)
// [^\r] capture the next-to-last bolded character
// )
// (?= lookahead at the very last bolded char and what comes after
// \S_ for underscore-bolding, it can be any non-space
// |
// \w for asterisk-bolding (otherwise the previous alternative would've matched, since \w implies \S), either the last char is word/underscore...
// |
// \S\*\*(?:[\W_]|$) ... or it's any other non-space, but in that case the character *after* the delimiter may not be a word character
// )
// . actually capture the last character (can use `.` this time because the lookahead ensures \S in all cases)
// )
// (?= lookahead; list the legal possibilities for the closing delimiter and its following character
// __(?:\W|$) for underscore-bolding, the following character (if any) must be non-word non-underscore
// |
// \*\*(?:[^*]|$) for asterisk-bolding, any non-asterisk is allowed (note we already ensured above that it's not a word character if the last bolded character wasn't one)
// )
// \2 actually capture the closing delimiter (and make sure that it matches the opening one)
text = text.replace(/(?=[^\r][*_]|[*_])(^|(?=\W__|(?!\*)[\W_]\*\*|\w\*\*\w)[^\r])(\*\*|__)(?!\2)(?=\S)((?:|[^\r]*?(?!\2)[^\r])(?=\S_|\w|\S\*\*(?:[\W_]|$)).)(?=__(?:\W|$)|\*\*(?:[^*]|$))\2/g,
"$1<strong>$3</strong>");
// now <em>:
// (?=[^\r][*_]|[*_]) Optimization, see above.
// ( Store in \1. This is the last character before the delimiter
// ^ Either we're at the start of the string (i.e. there is no last character)...
// | ... or we allow one of the following:
// (?= (lookahead; we're not capturing this, just listing legal possibilities)
// \W_ If the delimiter is _, then this last character must be non-word non-underscore (extra-word emphasis only)
// |
// (?!\*) otherwise, we list two possiblities for * as the delimiter; in either case, the last characters cannot be an asterisk itself
// (?:
// [\W_]\* this last character can be non-word (extra-word emphasis)...
// |
// \D\*(?=\w)\D ...or it can be word (otherwise the first alternative would've matched), but only if
// a) the first italicized character is such a character as well (intra-word emphasis), and
// b) neither character on either side of the asterisk is a digit
// )
// )
// [^\r] actually capture the character (can't use `.` since it could be \n)
// )
// (\*|_) Store in \2: the actual delimiter
// (?!\2\2\2) not followed by more than two more instances of the delimiter
// (?=\S) the first italicized character can't be a space
// ( Store in \3: the italicized string
// (?:(?!\2)[^\r])*? arbitrary characters except for the delimiter itself, minimally matching
// (?= lookahead at the very last italicized char and what comes after
// [^\s_]_ for underscore-italicizing, it can be any non-space non-underscore
// |
// (?=\w)\D\*\D for asterisk-italicizing, either the last char is word/underscore *and* neither character on either side of the asterisk is a digit...
// |
// [^\s*]\*(?:[\W_]|$) ... or that last char is any other non-space non-asterisk, but then the character after the delimiter (if any) must be non-word
// )
// . actually capture the last character (can use `.` this time because the lookahead ensures \S in all cases)
// )
// (?= lookahead; list the legal possibilities for the closing delimiter and its following character
// _(?:\W|$) for underscore-italicizing, the following character (if any) must be non-word non-underscore
// |
// \*(?:[^*]|$) for asterisk-italicizing, any non-asterisk is allowed; all other restrictions have already been ensured in the previous lookahead
// )
// \2 actually capture the closing delimiter (and make sure that it matches the opening one)
text = text.replace(/(?=[^\r][*_]|[*_])(^|(?=\W_|(?!\*)(?:[\W_]\*|\D\*(?=\w)\D))[^\r])(\*|_)(?!\2\2\2)(?=\S)((?:(?!\2)[^\r])*?(?=[^\s_]_|(?=\w)\D\*\D|[^\s*]\*(?:[\W_]|$)).)(?=_(?:\W|$)|\*(?:[^*]|$))\2/g,
"$1<em>$3</em>");
return deasciify(text);
}
function _DoBlockQuotes(text) {
/*
text = text.replace(/
( // Wrap whole match in $1
(
^[ \t]*>[ \t]? // '>' at the start of a line
.+\n // rest of the first line
(.+\n)* // subsequent consecutive lines
\n* // blanks
)+
)
/gm, function(){...});
*/
text = text.replace(/((^[ \t]*>[ \t]?.+\n(.+\n)*\n*)+)/gm,
function (wholeMatch, m1) {
var bq = m1;
// attacklab: hack around Konqueror 3.5.4 bug:
// "----------bug".replace(/^-/g,"") == "bug"
bq = bq.replace(/^[ \t]*>[ \t]?/gm, "~0"); // trim one level of quoting
// attacklab: clean up hack
bq = bq.replace(/~0/g, "");
bq = bq.replace(/^[ \t]+$/gm, ""); // trim whitespace-only lines
bq = _RunBlockGamut(bq); // recurse
bq = bq.replace(/(^|\n)/g, "$1 ");
// These leading spaces screw with <pre> content, so we need to fix that:
bq = bq.replace(
/(\s*<pre>[^\r]+?<\/pre>)/gm,
function (wholeMatch, m1) {
var pre = m1;
// attacklab: hack around Konqueror 3.5.4 bug:
pre = pre.replace(/^ /mg, "~0");
pre = pre.replace(/~0/g, "");
return pre;
});
return hashBlock("<blockquote>\n" + bq + "\n</blockquote>");
}
);
return text;
}
function _FormParagraphs(text, doNotUnhash, doNotCreateParagraphs) {
//
// Params:
// $text - string to process with html <p> tags
//
// Strip leading and trailing lines:
text = text.replace(/^\n+/g, "");
text = text.replace(/\n+$/g, "");
var grafs = text.split(/\n{2,}/g);
var grafsOut = [];
var markerRe = /~K(\d+)K/;
//
// Wrap <p> tags.
//
var end = grafs.length;
for (var i = 0; i < end; i++) {
var str = grafs[i];
// if this is an HTML marker, copy it
if (markerRe.test(str)) {
grafsOut.push(str);
}
else if (/\S/.test(str)) {
str = _RunSpanGamut(str);
str = str.replace(/^([ \t]*)/g, doNotCreateParagraphs ? "" : "<p>");
if (!doNotCreateParagraphs)
str += "</p>"
grafsOut.push(str);
}
}
//
// Unhashify HTML blocks
//
if (!doNotUnhash) {
end = grafsOut.length;
for (var i = 0; i < end; i++) {
var foundAny = true;
while (foundAny) { // we may need several runs, since the data may be nested
foundAny = false;
grafsOut[i] = grafsOut[i].replace(/~K(\d+)K/g, function (wholeMatch, id) {
foundAny = true;
return g_html_blocks[id];
});
}
}
}
return grafsOut.join("\n\n");
}
function _EncodeAmpsAndAngles(text) {
// Smart processing for ampersands and angle brackets that need to be encoded.
// Ampersand-encoding based entirely on Nat Irons's Amputator MT plugin:
// http://bumppo.net/projects/amputator/
text = text.replace(/&(?!#?[xX]?(?:[0-9a-fA-F]+|\w+);)/g, "&amp;");
// Encode naked <'s
text = text.replace(/<(?![a-z\/?!]|~D)/gi, "&lt;");
return text;
}
function _EncodeBackslashEscapes(text) {
//
// Parameter: String.
// Returns: The string, with after processing the following backslash
// escape sequences.
//
// attacklab: The polite way to do this is with the new
// escapeCharacters() function:
//
// text = escapeCharacters(text,"\\",true);
// text = escapeCharacters(text,"`*_{}[]()>#+-.!",true);
//
// ...but we're sidestepping its use of the (slow) RegExp constructor
// as an optimization for Firefox. This function gets called a LOT.
text = text.replace(/\\(\\)/g, escapeCharacters_callback);
text = text.replace(/\\([`*_{}\[\]()>#+-.!])/g, escapeCharacters_callback);
return text;
}
var charInsideUrl = "[-A-Z0-9+&@#/%?=~_|[\\]()!:,.;]",
charEndingUrl = "[-A-Z0-9+&@#/%=~_|[\\])]",
autoLinkRegex = new RegExp("(=\"|<)?\\b(https?|ftp)(://" + charInsideUrl + "*" + charEndingUrl + ")(?=$|\\W)", "gi"),
endCharRegex = new RegExp(charEndingUrl, "i");
function handleTrailingParens(wholeMatch, lookbehind, protocol, link) {
if (lookbehind)
return wholeMatch;
if (link.charAt(link.length - 1) !== ")")
return "<" + protocol + link + ">";
var parens = link.match(/[()]/g);
var level = 0;
for (var i = 0; i < parens.length; i++) {
if (parens[i] === "(") {
if (level <= 0)
level = 1;
else
level++;
}
else {
level--;
}
}
var tail = "";
if (level < 0) {
var re = new RegExp("\\){1," + (-level) + "}$");
link = link.replace(re, function (trailingParens) {
tail = trailingParens;
return "";
});
}
if (tail) {
var lastChar = link.charAt(link.length - 1);
if (!endCharRegex.test(lastChar)) {
tail = lastChar + tail;
link = link.substr(0, link.length - 1);
}
}
return "<" + protocol + link + ">" + tail;
}
function _DoAutoLinks(text) {
// note that at this point, all other URL in the text are already hyperlinked as <a href=""></a>
// *except* for the <http://www.foo.com> case
// automatically add < and > around unadorned raw hyperlinks
// must be preceded by a non-word character (and not by =" or <) and followed by non-word/EOF character
// simulating the lookbehind in a consuming way is okay here, since a URL can neither and with a " nor
// with a <, so there is no risk of overlapping matches.
text = text.replace(autoLinkRegex, handleTrailingParens);
// autolink anything like <http://example.com>
var replacer = function (wholematch, m1) {
var url = attributeSafeUrl(m1);
return "<a href=\"" + url + "\">" + pluginHooks.plainLinkText(m1) + "</a>";
};
text = text.replace(/<((https?|ftp):[^'">\s]+)>/gi, replacer);
// Email addresses: <address@domain.foo>
/*
text = text.replace(/
<
(?:mailto:)?
(
[-.\w]+
\@
[-a-z0-9]+(\.[-a-z0-9]+)*\.[a-z]+
)
>
/gi, _DoAutoLinks_callback());
*/
/* disabling email autolinking, since we don't do that on the server, either
text = text.replace(/<(?:mailto:)?([-.\w]+\@[-a-z0-9]+(\.[-a-z0-9]+)*\.[a-z]+)>/gi,
function(wholeMatch,m1) {
return _EncodeEmailAddress( _UnescapeSpecialChars(m1) );
}
);
*/
return text;
}
function _UnescapeSpecialChars(text) {
//
// Swap back in all the special characters we've hidden.
//
text = text.replace(/~E(\d+)E/g,
function (wholeMatch, m1) {
var charCodeToReplace = parseInt(m1);
return String.fromCharCode(charCodeToReplace);
}
);
return text;
}
function _Outdent(text) {
//
// Remove one level of line-leading tabs or spaces
//
// attacklab: hack around Konqueror 3.5.4 bug:
// "----------bug".replace(/^-/g,"") == "bug"
text = text.replace(/^(\t|[ ]{1,4})/gm, "~0"); // attacklab: g_tab_width
// attacklab: clean up hack
text = text.replace(/~0/g, "")
return text;
}
function _Detab(text) {
if (!/\t/.test(text))
return text;
var spaces = [" ", " ", " ", " "],
skew = 0,
v;
return text.replace(/[\n\t]/g, function (match, offset) {
if (match === "\n") {
skew = offset + 1;
return match;
}
v = (offset - skew) % 4;
skew = offset + 1;
return spaces[v];
});
}
//
// attacklab: Utility functions
//
function attributeSafeUrl(url) {
url = attributeEncode(url);
url = escapeCharacters(url, "*_:()[]")
return url;
}
function escapeCharacters(text, charsToEscape, afterBackslash) {
// First we have to escape the escape characters so that
// we can build a character class out of them
var regexString = "([" + charsToEscape.replace(/([\[\]\\])/g, "\\$1") + "])";
if (afterBackslash) {
regexString = "\\\\" + regexString;
}
var regex = new RegExp(regexString, "g");
text = text.replace(regex, escapeCharacters_callback);
return text;
}
function escapeCharacters_callback(wholeMatch, m1) {
var charCodeToEscape = m1.charCodeAt(0);
return "~E" + charCodeToEscape + "E";
}
}; // end of the Markdown.Converter constructor
})();