// Copyright (C) 2006 Google Inc. // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // You may obtain a copy of the License at // // http://www.apache.org/licenses/LICENSE-2.0 // // Unless required by applicable law or agreed to in writing, software // distributed under the License is distributed on an "AS IS" BASIS, // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. // See the License for the specific language governing permissions and // limitations under the License. /** * @fileoverview * some functions for browser-side pretty printing of code contained in html. * *
* For a fairly comprehensive set of languages see the * README * file that came with this source. At a minimum, the lexer should work on a * number of languages including C and friends, Java, Python, Bash, SQL, HTML, * XML, CSS, Javascript, and Makefiles. It works passably on Ruby, PHP and Awk * and a subset of Perl, but, because of commenting conventions, doesn't work on * Smalltalk, Lisp-like, or CAML-like languages without an explicit lang class. *
* Usage:
} and {@code } tags in your source with
* {@code class=prettyprint.}
* You can also use the (html deprecated) {@code } tag, but the pretty
* printer needs to do more substantial DOM manipulations to support that, so
* some css styles may not be preserved.
*
} or {@code } element to specify the
* language, as in {@code }. Any class that
* starts with "lang-" followed by a file extension, specifies the file type.
* See the "lang-*.js" files in this directory for code that implements
* per-language file handlers.
*
* Change log:
* cbeust, 2006/08/22
*
* Java annotations (start with "@") are now captured as literals ("lit")
*
* @requires console
*/
// JSLint declarations
/*global console, document, navigator, setTimeout, window */
/**
* Split {@code prettyPrint} into multiple timeouts so as not to interfere with
* UI events.
* If set to {@code false}, {@code prettyPrint()} is synchronous.
*/
window['PR_SHOULD_USE_CONTINUATION'] = true;
(function () {
// Keyword lists for various languages.
// We use things that coerce to strings to make them compact when minified
// and to defeat aggressive optimizers that fold large string constants.
var FLOW_CONTROL_KEYWORDS = ["break,continue,do,else,for,if,return,while"];
var C_KEYWORDS = [FLOW_CONTROL_KEYWORDS,"auto,case,char,const,default," +
"double,enum,extern,float,goto,int,long,register,short,signed,sizeof," +
"static,struct,switch,typedef,union,unsigned,void,volatile"];
var COMMON_KEYWORDS = [C_KEYWORDS,"catch,class,delete,false,import," +
"new,operator,private,protected,public,this,throw,true,try,typeof"];
var CPP_KEYWORDS = [COMMON_KEYWORDS,"alignof,align_union,asm,axiom,bool," +
"concept,concept_map,const_cast,constexpr,decltype," +
"dynamic_cast,explicit,export,friend,inline,late_check," +
"mutable,namespace,nullptr,reinterpret_cast,static_assert,static_cast," +
"template,typeid,typename,using,virtual,where"];
var JAVA_KEYWORDS = [COMMON_KEYWORDS,
"abstract,boolean,byte,extends,final,finally,implements,import," +
"instanceof,null,native,package,strictfp,super,synchronized,throws," +
"transient"];
var CSHARP_KEYWORDS = [JAVA_KEYWORDS,
"as,base,by,checked,decimal,delegate,descending,dynamic,event," +
"fixed,foreach,from,group,implicit,in,interface,internal,into,is,lock," +
"object,out,override,orderby,params,partial,readonly,ref,sbyte,sealed," +
"stackalloc,string,select,uint,ulong,unchecked,unsafe,ushort,var"];
var COFFEE_KEYWORDS = "all,and,by,catch,class,else,extends,false,finally," +
"for,if,in,is,isnt,loop,new,no,not,null,of,off,on,or,return,super,then," +
"true,try,unless,until,when,while,yes";
var JSCRIPT_KEYWORDS = [COMMON_KEYWORDS,
"debugger,eval,export,function,get,null,set,undefined,var,with," +
"Infinity,NaN"];
var PERL_KEYWORDS = "caller,delete,die,do,dump,elsif,eval,exit,foreach,for," +
"goto,if,import,last,local,my,next,no,our,print,package,redo,require," +
"sub,undef,unless,until,use,wantarray,while,BEGIN,END";
var PYTHON_KEYWORDS = [FLOW_CONTROL_KEYWORDS, "and,as,assert,class,def,del," +
"elif,except,exec,finally,from,global,import,in,is,lambda," +
"nonlocal,not,or,pass,print,raise,try,with,yield," +
"False,True,None"];
var RUBY_KEYWORDS = [FLOW_CONTROL_KEYWORDS, "alias,and,begin,case,class," +
"def,defined,elsif,end,ensure,false,in,module,next,nil,not,or,redo," +
"rescue,retry,self,super,then,true,undef,unless,until,when,yield," +
"BEGIN,END"];
var SH_KEYWORDS = [FLOW_CONTROL_KEYWORDS, "case,done,elif,esac,eval,fi," +
"function,in,local,set,then,until"];
var ALL_KEYWORDS = [
CPP_KEYWORDS, CSHARP_KEYWORDS, JSCRIPT_KEYWORDS, PERL_KEYWORDS +
PYTHON_KEYWORDS, RUBY_KEYWORDS, SH_KEYWORDS];
var C_TYPES = /^(DIR|FILE|vector|(de|priority_)?queue|list|stack|(const_)?iterator|(multi)?(set|map)|bitset|u?(int|float)\d*)/;
// token style names. correspond to css classes
/**
* token style for a string literal
* @const
*/
var PR_STRING = 'str';
/**
* token style for a keyword
* @const
*/
var PR_KEYWORD = 'kwd';
/**
* token style for a comment
* @const
*/
var PR_COMMENT = 'com';
/**
* token style for a type
* @const
*/
var PR_TYPE = 'typ';
/**
* token style for a literal value. e.g. 1, null, true.
* @const
*/
var PR_LITERAL = 'lit';
/**
* token style for a punctuation string.
* @const
*/
var PR_PUNCTUATION = 'pun';
/**
* token style for a punctuation string.
* @const
*/
var PR_PLAIN = 'pln';
/**
* token style for an sgml tag.
* @const
*/
var PR_TAG = 'tag';
/**
* token style for a markup declaration such as a DOCTYPE.
* @const
*/
var PR_DECLARATION = 'dec';
/**
* token style for embedded source.
* @const
*/
var PR_SOURCE = 'src';
/**
* token style for an sgml attribute name.
* @const
*/
var PR_ATTRIB_NAME = 'atn';
/**
* token style for an sgml attribute value.
* @const
*/
var PR_ATTRIB_VALUE = 'atv';
/**
* A class that indicates a section of markup that is not code, e.g. to allow
* embedding of line numbers within code listings.
* @const
*/
var PR_NOCODE = 'nocode';
/**
* A set of tokens that can precede a regular expression literal in
* javascript
* http://web.archive.org/web/20070717142515/http://www.mozilla.org/js/language/js20/rationale/syntax.html
* has the full list, but I've removed ones that might be problematic when
* seen in languages that don't support regular expression literals.
*
* Specifically, I've removed any keywords that can't precede a regexp
* literal in a syntactically legal javascript program, and I've removed the
* "in" keyword since it's not a keyword in many languages, and might be used
* as a count of inches.
*
*
The link a above does not accurately describe EcmaScript rules since
* it fails to distinguish between (a=++/b/i) and (a++/b/i) but it works
* very well in practice.
*
* @private
* @const
*/
var REGEXP_PRECEDER_PATTERN = '(?:^^\\.?|[+-]|\\!|\\!=|\\!==|\\#|\\%|\\%=|&|&&|&&=|&=|\\(|\\*|\\*=|\\+=|\\,|\\-=|\\->|\\/|\\/=|:|::|\\;|<|<<|<<=|<=|=|==|===|>|>=|>>|>>=|>>>|>>>=|\\?|\\@|\\[|\\^|\\^=|\\^\\^|\\^\\^=|\\{|\\||\\|=|\\|\\||\\|\\|=|\\~|break|case|continue|delete|do|else|finally|instanceof|return|throw|try|typeof)\\s*';
// CAVEAT: this does not properly handle the case where a regular
// expression immediately follows another since a regular expression may
// have flags for case-sensitivity and the like. Having regexp tokens
// adjacent is not valid in any language I'm aware of, so I'm punting.
// TODO: maybe style special characters inside a regexp as punctuation.
/**
* Given a group of {@link RegExp}s, returns a {@code RegExp} that globally
* matches the union of the sets of strings matched by the input RegExp.
* Since it matches globally, if the input strings have a start-of-input
* anchor (/^.../), it is ignored for the purposes of unioning.
* @param {Array.} regexs non multiline, non-global regexs.
* @return {RegExp} a global regex.
*/
function combinePrefixPatterns(regexs) {
var capturedGroupIndex = 0;
var needToFoldCase = false;
var ignoreCase = false;
for (var i = 0, n = regexs.length; i < n; ++i) {
var regex = regexs[i];
if (regex.ignoreCase) {
ignoreCase = true;
} else if (/[a-z]/i.test(regex.source.replace(
/\\u[0-9a-f]{4}|\\x[0-9a-f]{2}|\\[^ux]/gi, ''))) {
needToFoldCase = true;
ignoreCase = false;
break;
}
}
var escapeCharToCodeUnit = {
'b': 8,
't': 9,
'n': 0xa,
'v': 0xb,
'f': 0xc,
'r': 0xd
};
function decodeEscape(charsetPart) {
var cc0 = charsetPart.charCodeAt(0);
if (cc0 !== 92 /* \\ */) {
return cc0;
}
var c1 = charsetPart.charAt(1);
cc0 = escapeCharToCodeUnit[c1];
if (cc0) {
return cc0;
} else if ('0' <= c1 && c1 <= '7') {
return parseInt(charsetPart.substring(1), 8);
} else if (c1 === 'u' || c1 === 'x') {
return parseInt(charsetPart.substring(2), 16);
} else {
return charsetPart.charCodeAt(1);
}
}
function encodeEscape(charCode) {
if (charCode < 0x20) {
return (charCode < 0x10 ? '\\x0' : '\\x') + charCode.toString(16);
}
var ch = String.fromCharCode(charCode);
if (ch === '\\' || ch === '-' || ch === '[' || ch === ']') {
ch = '\\' + ch;
}
return ch;
}
function caseFoldCharset(charSet) {
var charsetParts = charSet.substring(1, charSet.length - 1).match(
new RegExp(
'\\\\u[0-9A-Fa-f]{4}'
+ '|\\\\x[0-9A-Fa-f]{2}'
+ '|\\\\[0-3][0-7]{0,2}'
+ '|\\\\[0-7]{1,2}'
+ '|\\\\[\\s\\S]'
+ '|-'
+ '|[^-\\\\]',
'g'));
var groups = [];
var ranges = [];
var inverse = charsetParts[0] === '^';
for (var i = inverse ? 1 : 0, n = charsetParts.length; i < n; ++i) {
var p = charsetParts[i];
if (/\\[bdsw]/i.test(p)) { // Don't muck with named groups.
groups.push(p);
} else {
var start = decodeEscape(p);
var end;
if (i + 2 < n && '-' === charsetParts[i + 1]) {
end = decodeEscape(charsetParts[i + 2]);
i += 2;
} else {
end = start;
}
ranges.push([start, end]);
// If the range might intersect letters, then expand it.
// This case handling is too simplistic.
// It does not deal with non-latin case folding.
// It works for latin source code identifiers though.
if (!(end < 65 || start > 122)) {
if (!(end < 65 || start > 90)) {
ranges.push([Math.max(65, start) | 32, Math.min(end, 90) | 32]);
}
if (!(end < 97 || start > 122)) {
ranges.push([Math.max(97, start) & ~32, Math.min(end, 122) & ~32]);
}
}
}
}
// [[1, 10], [3, 4], [8, 12], [14, 14], [16, 16], [17, 17]]
// -> [[1, 12], [14, 14], [16, 17]]
ranges.sort(function (a, b) { return (a[0] - b[0]) || (b[1] - a[1]); });
var consolidatedRanges = [];
var lastRange = [NaN, NaN];
for (var i = 0; i < ranges.length; ++i) {
var range = ranges[i];
if (range[0] <= lastRange[1] + 1) {
lastRange[1] = Math.max(lastRange[1], range[1]);
} else {
consolidatedRanges.push(lastRange = range);
}
}
var out = ['['];
if (inverse) { out.push('^'); }
out.push.apply(out, groups);
for (var i = 0; i < consolidatedRanges.length; ++i) {
var range = consolidatedRanges[i];
out.push(encodeEscape(range[0]));
if (range[1] > range[0]) {
if (range[1] + 1 > range[0]) { out.push('-'); }
out.push(encodeEscape(range[1]));
}
}
out.push(']');
return out.join('');
}
function allowAnywhereFoldCaseAndRenumberGroups(regex) {
// Split into character sets, escape sequences, punctuation strings
// like ('(', '(?:', ')', '^'), and runs of characters that do not
// include any of the above.
var parts = regex.source.match(
new RegExp(
'(?:'
+ '\\[(?:[^\\x5C\\x5D]|\\\\[\\s\\S])*\\]' // a character set
+ '|\\\\u[A-Fa-f0-9]{4}' // a unicode escape
+ '|\\\\x[A-Fa-f0-9]{2}' // a hex escape
+ '|\\\\[0-9]+' // a back-reference or octal escape
+ '|\\\\[^ux0-9]' // other escape sequence
+ '|\\(\\?[:!=]' // start of a non-capturing group
+ '|[\\(\\)\\^]' // start/emd of a group, or line start
+ '|[^\\x5B\\x5C\\(\\)\\^]+' // run of other characters
+ ')',
'g'));
var n = parts.length;
// Maps captured group numbers to the number they will occupy in
// the output or to -1 if that has not been determined, or to
// undefined if they need not be capturing in the output.
var capturedGroups = [];
// Walk over and identify back references to build the capturedGroups
// mapping.
for (var i = 0, groupIndex = 0; i < n; ++i) {
var p = parts[i];
if (p === '(') {
// groups are 1-indexed, so max group index is count of '('
++groupIndex;
} else if ('\\' === p.charAt(0)) {
var decimalValue = +p.substring(1);
if (decimalValue && decimalValue <= groupIndex) {
capturedGroups[decimalValue] = -1;
}
}
}
// Renumber groups and reduce capturing groups to non-capturing groups
// where possible.
for (var i = 1; i < capturedGroups.length; ++i) {
if (-1 === capturedGroups[i]) {
capturedGroups[i] = ++capturedGroupIndex;
}
}
for (var i = 0, groupIndex = 0; i < n; ++i) {
var p = parts[i];
if (p === '(') {
++groupIndex;
if (capturedGroups[groupIndex] === undefined) {
parts[i] = '(?:';
}
} else if ('\\' === p.charAt(0)) {
var decimalValue = +p.substring(1);
if (decimalValue && decimalValue <= groupIndex) {
parts[i] = '\\' + capturedGroups[groupIndex];
}
}
}
// Remove any prefix anchors so that the output will match anywhere.
// ^^ really does mean an anchored match though.
for (var i = 0, groupIndex = 0; i < n; ++i) {
if ('^' === parts[i] && '^' !== parts[i + 1]) { parts[i] = ''; }
}
// Expand letters to groups to handle mixing of case-sensitive and
// case-insensitive patterns if necessary.
if (regex.ignoreCase && needToFoldCase) {
for (var i = 0; i < n; ++i) {
var p = parts[i];
var ch0 = p.charAt(0);
if (p.length >= 2 && ch0 === '[') {
parts[i] = caseFoldCharset(p);
} else if (ch0 !== '\\') {
// TODO: handle letters in numeric escapes.
parts[i] = p.replace(
/[a-zA-Z]/g,
function (ch) {
var cc = ch.charCodeAt(0);
return '[' + String.fromCharCode(cc & ~32, cc | 32) + ']';
});
}
}
}
return parts.join('');
}
var rewritten = [];
for (var i = 0, n = regexs.length; i < n; ++i) {
var regex = regexs[i];
if (regex.global || regex.multiline) { throw new Error('' + regex); }
rewritten.push(
'(?:' + allowAnywhereFoldCaseAndRenumberGroups(regex) + ')');
}
return new RegExp(rewritten.join('|'), ignoreCase ? 'gi' : 'g');
}
/**
* Split markup into a string of source code and an array mapping ranges in
* that string to the text nodes in which they appear.
*
*
* The HTML DOM structure:
*
* (Element "p"
* (Element "b"
* (Text "print ")) ; #1
* (Text "'Hello '") ; #2
* (Element "br") ; #3
* (Text " + 'World';")) ; #4
*
*
* corresponds to the HTML
* {@code
print 'Hello '
+ 'World';
}.
*
*
* It will produce the output:
*
* {
* sourceCode: "print 'Hello '\n + 'World';",
* // 1 2
* // 012345678901234 5678901234567
* spans: [0, #1, 6, #2, 14, #3, 15, #4]
* }
*
*
* where #1 is a reference to the {@code "print "} text node above, and so
* on for the other text nodes.
*
*
*
* The {@code} spans array is an array of pairs. Even elements are the start
* indices of substrings, and odd elements are the text nodes (or BR elements)
* that contain the text for those substrings.
* Substrings continue until the next index or the end of the source.
*
*
* @param {Node} node an HTML DOM subtree containing source-code.
* @return {Object} source code and the text nodes in which they occur.
*/
function extractSourceSpans(node) {
var nocode = /(?:^|\s)nocode(?:\s|$)/;
var chunks = [];
var length = 0;
var spans = [];
var k = 0;
var whitespace;
if (node.currentStyle) {
whitespace = node.currentStyle.whiteSpace;
} else if (window.getComputedStyle) {
whitespace = document.defaultView.getComputedStyle(node, null)
.getPropertyValue('white-space');
}
var isPreformatted = whitespace && 'pre' === whitespace.substring(0, 3);
function walk(node) {
switch (node.nodeType) {
case 1: // Element
if (nocode.test(node.className)) { return; }
for (var child = node.firstChild; child; child = child.nextSibling) {
walk(child);
}
var nodeName = node.nodeName;
if ('BR' === nodeName || 'LI' === nodeName) {
chunks[k] = '\n';
spans[k << 1] = length++;
spans[(k++ << 1) | 1] = node;
}
break;
case 3: case 4: // Text
var text = node.nodeValue;
if (text.length) {
if (!isPreformatted) {
text = text.replace(/[ \t\r\n]+/g, ' ');
} else {
text = text.replace(/\r\n?/g, '\n'); // Normalize newlines.
}
// TODO: handle tabs here?
chunks[k] = text;
spans[k << 1] = length;
length += text.length;
spans[(k++ << 1) | 1] = node;
}
break;
}
}
walk(node);
return {
sourceCode: chunks.join('').replace(/\n$/, ''),
spans: spans
};
}
/**
* Apply the given language handler to sourceCode and add the resulting
* decorations to out.
* @param {number} basePos the index of sourceCode within the chunk of source
* whose decorations are already present on out.
*/
function appendDecorations(basePos, sourceCode, langHandler, out) {
if (!sourceCode) { return; }
var job = {
sourceCode: sourceCode,
basePos: basePos
};
langHandler(job);
out.push.apply(out, job.decorations);
}
var notWs = /\S/;
/**
* Given an element, if it contains only one child element and any text nodes
* it contains contain only space characters, return the sole child element.
* Otherwise returns undefined.
*
* This is meant to return the CODE element in {@code
} when
* there is a single child element that contains all the non-space textual
* content, but not to return anything where there are multiple child elements
* as in {@code ...
...
} or when there
* is textual content.
*/
function childContentWrapper(element) {
var wrapper = undefined;
for (var c = element.firstChild; c; c = c.nextSibling) {
var type = c.nodeType;
wrapper = (type === 1) // Element Node
? (wrapper ? element : c)
: (type === 3) // Text Node
? (notWs.test(c.nodeValue) ? element : wrapper)
: wrapper;
}
return wrapper === element ? undefined : wrapper;
}
/** Given triples of [style, pattern, context] returns a lexing function,
* The lexing function interprets the patterns to find token boundaries and
* returns a decoration list of the form
* [index_0, style_0, index_1, style_1, ..., index_n, style_n]
* where index_n is an index into the sourceCode, and style_n is a style
* constant like PR_PLAIN. index_n-1 <= index_n, and style_n-1 applies to
* all characters in sourceCode[index_n-1:index_n].
*
* The stylePatterns is a list whose elements have the form
* [style : string, pattern : RegExp, DEPRECATED, shortcut : string].
*
* Style is a style constant like PR_PLAIN, or can be a string of the
* form 'lang-FOO', where FOO is a language extension describing the
* language of the portion of the token in $1 after pattern executes.
* E.g., if style is 'lang-lisp', and group 1 contains the text
* '(hello (world))', then that portion of the token will be passed to the
* registered lisp handler for formatting.
* The text before and after group 1 will be restyled using this decorator
* so decorators should take care that this doesn't result in infinite
* recursion. For example, the HTML lexer rule for SCRIPT elements looks
* something like ['lang-js', /<[s]cript>(.+?)<\/script>/]. This may match
* '