//
Attacklab.showdown.converter = function() {
+
+// 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.stackoverflow.com/questions/64655/strange-wmd-bug
+// (granted, switching from Array() to Object() alone would have left only __proto__
+// to be a problem)
+var SaveHash = function () {
+ this.set = function (key, value) {
+ this["s_" + key] = value;
+ }
+ this.get = function (key) {
+ return this["s_" + key];
+ }
+}
+
//
// Globals:
//
// _EscapeSpecialCharsWithinTagAttributes(), so that any *'s or _'s in the <a>
// and <img> tags get encoded.
//
+ text = html_sanitize(text, function(url) {return url;}, function(id) {return id;});
// Clear the global hashes. If we don't clear these, you get conflicts
// from other articles when generating a page which contains more than
// one article (e.g. an index page that shows the N most recent
// articles):
- g_urls = new Array();
- g_titles = new Array();
+ g_urls = new SaveHash();
+ g_titles = new SaveHash();
g_html_blocks = new Array();
// attacklab: Replace ~ with ~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]*
- (?:
- (\n*) // any lines skipped = $3 attacklab: lookbehind removed
+ ( // (potential) title = $3
+ (\n*) // any lines skipped = $4 attacklab: lookbehind removed
+ [ \t]+
["(]
- (.+?) // title = $4
+ (.+?) // title = $5
[")]
[ \t]*
)? // title is optional
/gm,
function(){...});
*/
- var text = text.replace(/^[ ]{0,3}\[(.+)\]:[ \t]*\n?[ \t]*<?(\S+?)>?[ \t]*\n?[ \t]*(?:(\n*)["(](.+?)[")][ \t]*)?(?:\n+)/gm,
- function (wholeMatch,m1,m2,m3,m4) {
+ var text = text.replace(/^[ ]{0,3}\[(.+)\]:[ \t]*\n?[ \t]*<?(\S+?)>?(?=\s|$)[ \t]*\n?[ \t]*((\n*)["(](.+?)[")][ \t]*)?(?:\n+)/gm,
+ function (wholeMatch,m1,m2,m3,m4,m5) {
m1 = m1.toLowerCase();
- g_urls[m1] = _EncodeAmpsAndAngles(m2); // Link IDs are case-insensitive
- if (m3) {
+ 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+m4;
- } else if (m4) {
- g_titles[m1] = m4.replace(/"/g,""");
+ return m3;
+ } else if (m5) {
+ g_titles.set(m1, m5.replace(/"/g,"""));
}
-
+
// Completely remove the definition from the text
return "";
}
}
var _HashHTMLBlocks = function(text) {
- // attacklab: Double up blank lines to reduce lookaround
- text = text.replace(/\n/g,"\n\n");
// Hashify HTML blocks:
// We only want to do this for block-level HTML tags, such as headers,
text = text.replace(/^(<(p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math)\b[^\r]*?.*<\/\2>[ \t]*(?=\n+)\n)/gm,hashElement);
// Special case just for <hr />. It was easier to make a special case than
- // to make the other regex more complicated.
+ // to make the other regex more complicated.
/*
text = text.replace(/
+ \n // Starting after a blank line
+ [ ]{0,3}
( // save in $1
- \n\n // Starting after a blank line
- [ ]{0,3}
(<(hr) // start tag = $2
\b // word break
- ([^<>])*? //
+ ([^<>])*? //
\/?>) // the matching end tag
[ \t]*
(?=\n{2,}) // followed by a blank line
)
/g,hashElement);
*/
- text = text.replace(/(\n[ ]{0,3}(<(hr)\b([^<>])*?\/?>)[ \t]*(?=\n{2,}))/g,hashElement);
+ text = text.replace(/\n[ ]{0,3}((<(hr)\b([^<>])*?\/?>)[ \t]*(?=\n{2,}))/g,hashElement);
// 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 $1
- \n\n // Starting after a blank line
- [ ]{0,3} // attacklab: g_tab_width - 1
<!
- (--[^\r]*?--\s*)+
+ (--(?:|(?:[^>-]|-[^>])(?:[^-]|-[^-])*)--) // see http://www.w3.org/TR/html-markup/syntax.html#comments
>
[ \t]*
(?=\n{2,}) // followed by a blank line
)
/g,hashElement);
*/
- text = text.replace(/(\n\n[ ]{0,3}<!(--[^\r]*?--\s*)+>[ \t]*(?=\n{2,}))/g,hashElement);
+ text = text.replace(/\n\n[ ]{0,3}(<!(--(?:|(?:[^>-]|-[^>])(?:[^-]|-[^-])*)--)>[ \t]*(?=\n{2,}))/g, hashElement);
// PHP and ASP-style processor instructions (<?...?> and <%...%>)
*/
text = text.replace(/(?:\n\n)([ ]{0,3}(?:<([?%])[^\r]*?\2>)[ \t]*(?=\n{2,}))/g,hashElement);
- // attacklab: Undo double lines (see comment at top of this function)
- text = text.replace(/\n\n/g,"\n");
return text;
}
var blockText = m1;
// Undo double lines
- blockText = blockText.replace(/\n\n/g,"\n");
- blockText = blockText.replace(/^\n/,"");
-
+ blockText = blockText.replace(/^\n+/,"");
+
// strip trailing blank lines
blockText = blockText.replace(/\n+$/g,"");
-
+
// Replace the element text with a marker ("~KxK" where x is its key)
blockText = "\n\n~K" + (g_html_blocks.push(blockText)-1) + "K\n\n";
-
+
return blockText;
};
-var _RunBlockGamut = function(text) {
+var _RunBlockGamut = function(text, doNotUnhash) {
//
// These are all the transformations that form block-level
// tags like paragraphs, headers, and list items.
// 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);
+ text = _FormParagraphs(text, doNotUnhash);
return text;
}
// 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.
- var regex = /(<[a-z\/!$]("[^"]*"|'[^']*'|[^'">])*>|<!(--.*?--\s*)+>)/gi;
+ // Build a regex to find HTML tags and comments. See Friedl's
+ // "Mastering Regular Expressions", 2nd Ed., pp. 200-201.
+
+ // SE: changed the comment part of the regex
+
+ var regex = /(<[a-z\/!$]("[^"]*"|'[^']*'|[^'">])*>|<!(--(?:|(?:[^>-]|-[^>])(?:[^-]|-[^-])*)--)>)/gi;
text = text.replace(regex, function(wholeMatch) {
var tag = wholeMatch.replace(/(.)<\/?code>(?=.)/g,"$1`");
/*
text = text.replace(/
- ( // wrap whole match in $1
- \[
+ ( // wrap whole match in $1
+ \[
(
(?:
\[[^\]]*\] // allow brackets nested one level
- |
- [^\[\]] // or anything else
- )
- )
+ |
+ [^\[\]] // or anything else
+ )*
+ )
\]
\( // literal paren
[ \t]*
() // no id, so leave $3 empty
- <?(.*?)>? // href = $4
+ <?( // href = $4
+ (?:
+ \([^)]*\) // allow one level of (correctly nested) parens (think MSDN)
+ |
+ [^()]
+ )*?
+ )>?
[ \t]*
( // $5
(['"]) // quote char = $6
)
/g,writeAnchorTag);
*/
- text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\]\([ \t]*()<?(.*?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g,writeAnchorTag);
+
+ text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\]\([ \t]*()<?((?:\([^)]*\)|[^()])*?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g,writeAnchorTag);
//
// Last, handle reference-style shortcuts: [link text]
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[link_id] != undefined) {
- url = g_urls[link_id];
- if (g_titles[link_id] != undefined) {
- title = g_titles[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;
}
}
- }
-
+ }
+
url = escapeCharacters(url,"*_");
var result = "<a href=\"" + url + "\"";
-
+
if (title != "") {
title = title.replace(/"/g,""");
title = escapeCharacters(title,"*_");
result += " title=\"" + title + "\"";
}
-
+
result += ">" + link_text + "</a>";
-
+
return result;
}
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[link_id] != undefined) {
- url = g_urls[link_id];
- if (g_titles[link_id] != undefined) {
- title = g_titles[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 = alt_text.replace(/"/g,""");
url = escapeCharacters(url,"*_");
- if (url.toString().indexOf('http://') != 0 && url.toString().indexOf('https://') != 0) {
- url = scriptUrl + url
- }
var result = "<img src=\"" + url + "\" alt=\"" + alt_text + "\"";
// attacklab: Markdown.pl adds empty title attributes to images.
title = escapeCharacters(title,"*_");
result += " title=\"" + title + "\"";
//}
-
+
result += " />";
-
+
return result;
}
// Setext-style headers:
// Header 1
// ========
- //
+ //
// Header 2
// --------
//
text = text.replace(/^(.+)[ \t]*\n=+[ \t]*\n+/gm,
- function(wholeMatch,m1){return hashBlock("<h1>" + _RunSpanGamut(m1) + "</h1>");});
+ function(wholeMatch,m1){return "<h1>" + _RunSpanGamut(m1) + "</h1>\n\n";});
text = text.replace(/^(.+)[ \t]*\n-+[ \t]*\n+/gm,
- function(matchFound,m1){return hashBlock("<h2>" + _RunSpanGamut(m1) + "</h2>");});
+ function(matchFound,m1){return "<h2>" + _RunSpanGamut(m1) + "</h2>\n\n";});
// atx-style headers:
// # Header 1
text = text.replace(/^(\#{1,6})[ \t]*(.+?)[ \t]*\#*\n+/gm,
function(wholeMatch,m1,m2) {
var h_level = m1.length;
- return hashBlock("<h" + h_level + ">" + _RunSpanGamut(m2) + "</h" + h_level + ">");
+ return "<h" + h_level + ">" + _RunSpanGamut(m2) + "</h" + h_level + ">\n\n";
});
return text;
var list = m1;
var list_type = (m2.search(/[*+-]/g)>-1) ? "ul" : "ol";
- // Turn double returns into triple returns, so that we can make a
- // paragraph for the last item in a list, if necessary:
- list = list.replace(/\n{2,}/g,"\n\n\n");;
- var result = _ProcessListItems(list);
-
+ var result = _ProcessListItems(list, list_type);
+
// 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
var list = m2;
var list_type = (m3.search(/[*+-]/g)>-1) ? "ul" : "ol";
- // Turn double returns into triple returns, so that we can make a
- // paragraph for the last item in a list, if necessary:
- var list = list.replace(/\n{2,}/g,"\n\n\n");;
- var result = _ProcessListItems(list);
- result = runup + "<"+list_type+">\n" + result + "</"+list_type+">\n";
+ var result = _ProcessListItems(list, list_type);
+ result = runup + "<"+list_type+">\n" + result + "</"+list_type+">\n";
return result;
});
}
return text;
}
-_ProcessListItems = function(list_str) {
+var _listItemMarkers = { ol: "\\d+[.]", ul: "[*+-]" };
+
+_ProcessListItems = function(list_str, list_type) {
//
// 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.
// attacklab: add sentinel to emulate \z
list_str += "~0";
+ // In the original attacklab WMD, 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(/
- (\n)? // leading line = $1
- (^[ \t]*) // leading whitespace = $2
- ([*+-]|\d+[.]) [ \t]+ // list marker = $3
- ([^\r]+? // list item text = $4
- (\n{1,2}))
- (?= \n* (~0 | \2 ([*+-]|\d+[.]) [ \t]+))
+ (^[ \t]*) // leading whitespace = $1
+ ({MARKER}) [ \t]+ // list marker = $2
+ ([^\r]+? // list item text = $3
+ (\n+))
+ (?= (~0 | \2 ({MARKER}) [ \t]+))
/gm, function(){...});
*/
- list_str = list_str.replace(/(\n)?(^[ \t]*)([*+-]|\d+[.])[ \t]+([^\r]+?(\n{1,2}))(?=\n*(~0|\2([*+-]|\d+[.])[ \t]+))/gm,
- function(wholeMatch,m1,m2,m3,m4){
- var item = m4;
- var leading_line = m1;
- var leading_space = m2;
-
- if (leading_line || (item.search(/\n{2,}/)>-1)) {
- item = _RunBlockGamut(_Outdent(item));
+
+ var marker = _listItemMarkers[list_type];
+ var re = new RegExp("(^[ \\t]*)(" + marker + ")[ \\t]+([^\\r]+?(\\n+))(?=(~0|\\1(" + marker + ")[ \\t]+))", "gm");
+ var last_item_had_a_double_newline = false;
+ list_str = list_str.replace(re,
+ function(wholeMatch,m1,m2,m3){
+ var item = m3;
+ var leading_space = m1;
+ var ends_with_double_newline = /\n\n$/.test(item);
+ var contains_double_newline = ends_with_double_newline || item.search(/\n{2,}/)>-1;
+
+ if (contains_double_newline || last_item_had_a_double_newline) {
+ item = _RunBlockGamut(_Outdent(item), /* doNotUnhash = */ true);
}
else {
// Recursion for sub-lists:
item = _DoLists(_Outdent(item));
item = item.replace(/\n$/,""); // chomp(item)
item = _RunSpanGamut(item);
- }
-
+ }
+ last_item_had_a_double_newline = ends_with_double_newline;
return "<li>" + item + "</li>\n";
}
);
var _DoCodeBlocks = function(text) {
//
// Process Markdown `<pre><code>` blocks.
-//
+//
/*
text = text.replace(text,
// attacklab: sentinel workarounds for lack of \A and \Z, safari\khtml bug
text += "~0";
-
+
text = text.replace(/(?:\n\n|^)((?:(?:[ ]{4}|\t).*\n+)+)(\n*[ ]{0,3}[^ \t\n]|(?=~0))/g,
function(wholeMatch,m1,m2) {
var codeblock = m1;
var nextChar = m2;
-
+
codeblock = _EncodeCode( _Outdent(codeblock));
codeblock = _Detab(codeblock);
codeblock = codeblock.replace(/^\n+/g,""); // trim leading newlines
codeblock = "<pre><code>" + codeblock + "\n</code></pre>";
- return hashBlock(codeblock) + nextChar;
+ return "\n\n" + codeblock + "\n\n" + nextChar;
}
);
var _DoCodeSpans = function(text) {
//
// * Backtick quotes are used for <code></code> spans.
-//
+//
// * You can use multiple backticks as the delimiters if you want to
// include literal backticks in the code span. So, this input:
-//
+//
// Just type ``foo `bar` baz`` at the prompt.
-//
+//
// Will translate to:
-//
+//
// <p>Just type <code>foo `bar` baz</code> at the prompt.</p>
-//
+//
// There's no arbitrary limit to the number of backticks you
// can use as delimters. If you need three consecutive backticks
// in your code, use four for delimiters, etc.
//
// * You can use spaces to get literal backticks at the edges:
-//
+//
// ... type `` `bar` `` ...
-//
+//
// Turns to:
-//
+//
// ... type <code>`bar`</code> ...
//
bq = bq.replace(/^[ \t]+$/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(
pre = pre.replace(/~0/g,"");
return pre;
});
-
+
return hashBlock("<blockquote>\n" + bq + "\n</blockquote>");
});
return text;
}
-var _FormParagraphs = function(text) {
+var _FormParagraphs = function(text, doNotUnhash) {
//
// Params:
// $text - string to process with html <p> tags
}
}
-
//
// Unhashify HTML blocks
//
- end = grafsOut.length;
- for (var i=0; i<end; i++) {
- // if this is a marker for an html block...
- while (grafsOut[i].search(/~K(\d+)K/) >= 0) {
- var blockText = g_html_blocks[RegExp.$1];
- blockText = blockText.replace(/\$/g,"$$$$"); // Escape any dollar signs
- grafsOut[i] = grafsOut[i].replace(/~K\d+K/,blockText);
- }
- }
-
+ if (!doNotUnhash) {
+ end = grafsOut.length;
+ for (var i=0; i<end; i++) {
+ // if this is a marker for an html block...
+ while (grafsOut[i].search(/~K(\d+)K/) >= 0) {
+ var blockText = g_html_blocks[RegExp.$1];
+ blockText = blockText.replace(/\$/g,"$$$$"); // Escape any dollar signs
+ grafsOut[i] = grafsOut[i].replace(/~K\d+K/,blockText);
+ }
+ }
+ }
return grafsOut.join("\n\n");
}
var _EncodeAmpsAndAngles = function(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,"&");
-
+
// Encode naked <'s
text = text.replace(/<(?![a-z\/?\$!])/gi,"<");
-
+
return text;
}
return text;
}
-var _Detab = function(text) {
-// attacklab: Detab's completely rewritten for speed.
-// In perl we could fix it by anchoring the regexp with \G.
-// In javascript we're less fortunate.
-
- // expand first n-1 tabs
- text = text.replace(/\t(?=\t)/g," "); // attacklab: g_tab_width
+var _Detab = function (text) {
+ if (!/\t/.test(text))
+ return text;
- // replace the nth with two sentinels
- text = text.replace(/\t/g,"~A~B");
+ var spaces = [" ", " ", " ", " "],
+ skew = 0,
+ v;
- // use the sentinel to anchor our regex so it doesn't explode
- text = text.replace(/~B(.+?)~A/g,
- function(wholeMatch,m1,m2) {
- var leadingText = m1;
- var numSpaces = 4 - leadingText.length % 4; // attacklab: g_tab_width
-
- // there *must* be a better way to do this:
- for (var i=0; i<numSpaces; i++) leadingText+=" ";
-
- return leadingText;
+ return text.replace(/[\n\t]/g, function (match, offset) {
+ if (match === "\n") {
+ skew = offset + 1;
+ return match;
}
- );
-
- // clean up sentinels
- text = text.replace(/~A/g," "); // attacklab: g_tab_width
- text = text.replace(/~B/g,"");
-
- return text;
+ v = (offset - skew) % 4;
+ skew = offset + 1;
+ return spaces[v];
+ });
}
-
//
// attacklab: Utility functions
//
projects / osqa.git / blobdiff