Skip to content

Instantly share code, notes, and snippets.

@yellow1912

yellow1912/CSSUtilities.js

Created September 25, 2020 16:47
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 1 You must be signed in to fork a gist
  • Save yellow1912/c9dbbab97497ec42489be55e8abe73c7 to your computer and use it in GitHub Desktop.
Save yellow1912/c9dbbab97497ec42489be55e8abe73c7 to your computer and use it in GitHub Desktop.
Download ZIP
Updated Selector.js and CSSUtilities to work in latest Chrome (v85)
Raw
CSSUtilities.js
// IMPORTANT: you need to import selector here
// import {Selector} from "@/Selector";
/*******************************************************************************
CSS0.99.1B :: CSSUtilities
-------------------------------------------------------------------------------
Copyright (c) 2010 James Edwards (brothercake) <cake@brothercake.com>
BSD License See license.txt for licensing information
Info/Docs http://www.brothercake.com/site/resources/scripts/cssutilities/
-------------------------------------------------------------------------------
Credits and thanks:
-------------------------------------------------------------------------------
Henrik Lindqvist [Selector.js] http://llamalab.com/js/selector/
Dean Edwards [technical help] http://dean.edwards.name/
Stuart Langridge [technical help] http://www.kryogenix.org/
-------------------------------------------------------------------------------
*******************************************************************************/
export function CSSUtilities() {}
(function()
{
//-- internal self-reference constant --//
//this is of course not a constant, it's a variable, but IE doesn't support const
//so we use var with the uppercase convention for values that don't change
//this hook can also be used by the code compression routine, to declare Z=this
//which can then go on to be replaced for instances of both THIS and this
var THIS = this;
//set the asynchronous ready flag to false
this._allready = false;
//set the supported flag to false by default
this.supported = false;
//do some feature detection
if(
//this generally excludes pre-DOM1 browsers
typeof document.getElementById == 'undefined'
//this excludes Opera 8
|| typeof document.styleSheets == 'undefined'
//this excludes IE 5.5
|| typeof document.nodeType == 'undefined'
//exit on failure
){ return; }
//if we're still here, set the supported flag to true
this.supported = true;
//-- start local variable declarations --//
//-- nb. the breaks in string values are to avoid unwanted compression --//
//we need a flag to signify when initialization is in progress
//so that we can detect and deal with overlapping asynchronous init calls
var busyinit = false,
//-- errors and warnings --//
//-- error constants (for data log messages) --//
ERROR_NETWORK_OR_SECURITY = 'Network Failure or Security Violation',
ERROR_NETWORK = 'Network Failure',
ERROR_SECURITY = 'Security Violation',
ERROR_UNSPECIFIED = 'Unspecified Error',
ERROR_NOT_CSS = 'Data is not CSS',
//-- warning message constants (for data log messages) --//
MESSAGE_UNKNOWN = 'unknown',
MESSAGE_OK = 'OK',
MESSAGE_DUPLICATE = 'Discarded Duplicate',
MESSAGE_DISABLED = 'Stylesheet is disabled',
MESSAGE_UNSUPPORTED_TYPE = 'Unsupported node type',
//-- fatal error constants (for console messages) --//
FATAL_ERROR_PREFIX = 'CSSUtilities ' + '(Fatal Error):',
FATAL_ERROR_INVALID_MODE = FATAL_ERROR_PREFIX + ' The specified mode is not valid',
FATAL_ERROR_INVALID_ASYNC = FATAL_ERROR_PREFIX + ' The specified async setting is not valid',
FATAL_ERROR_NOT_DOCUMENT = FATAL_ERROR_PREFIX + ' The specified document is not a Document',
FATAL_ERROR_NOT_ABSOLUTE_URI = FATAL_ERROR_PREFIX + ' The specified base is not an absolute URL',
FATAL_ERROR_INVALID_WATCH = FATAL_ERROR_PREFIX + ' The specified watch setting is not valid',
FATAL_ERROR_INVALID_ATTRS = FATAL_ERROR_PREFIX + ' The specified attributes setting is not valid',
FATAL_ERROR_NOT_SAPI = FATAL_ERROR_PREFIX + ' The specified api settings are not valid',
FATAL_ERROR_INVALID_SAPI = FATAL_ERROR_PREFIX + ' Your Selectors API is not returning the right data',
FATAL_ERROR_MISSING_SAPI = FATAL_ERROR_PREFIX + ' The Selectors API is missing',
FATAL_ERROR_NO_XHR = FATAL_ERROR_PREFIX + ' Unable to communicate with the network',
//-- error message variables (for console messages) --//
errorMessageNoElement = 'CSSUtilities.%method has an invalid Element reference or ID',
errorMessageNoSelector = 'CSSUtilities.%method requires a valid Selector reference',
errorMessageMultipleSelector = 'CSSUtilities.%method can only process one Selector at a time',
errorMessageInvalidSSID = 'CSSUtilities.%method has an invalid Stylesheet ID',
errorMessageNotAfterAPI = FATAL_ERROR_PREFIX + ' You cannot define "%var" after "api",' + ' it must be defined first',
//-- css data constants --//
//media defaults and special values
MEDIA_ALL = 'all',
MEDIA_SCREEN = 'screen',
MEDIA_NONE = 'none',
MEDIA_CURRENT = 'current',
//all valid CSS media types, used as part of the view media detection
//this list includes all standard media types from CSS2, plus the new "reader" type proposed in CSS3
MEDIA_TYPES_LIST = 'aural,braille,embossed,handheld,print,projection,reader,screen,speech,tty,tv',
//alphabetical dictionary of inheritable CSS properties
//we only actually need the keys, so we can go typeof INHERITED_PROPS[key] != 'undefined'
//but the values are used for dev reference, so that we know the status of each property:
// '2' CSS2 properties
// '3' CSS3 properties
// '?' CSS3 properties with "inheritable" marked as "yes or ?"
// 'm' mozilla extensions
// 'o' opera extensions
// 'w' webkit extensions
// 'e' explorer extensions
//nb. this list is complete with respect to standard properties, mozilla and opera extensions;
//but there may be inheritable webkit and explorer extensions not listed here
//because I couldn't find a definitive list of explorer props
//and the apple reference doesn't say whether properties are inheritable
INHERITED_PROPS = {
'azimuth': '2',
'border-collapse': '2',
'border-spacing': '2',
'caption-side': '2',
'color': '2',
'cursor': '2',
'direction': '2',
'elevation': '2',
'empty-cells': '2',
'fit': '3',
'fit-position': '3',
'font': '2',
'font-family': '2',
'font-size': '2',
'font-size-adjust': '2',
'font-stretch': '2',
'font-style': '2',
'font-variant': '2',
'font-weight': '2',
'hanging-punctuation': '3',
'hyphenate-after': '3',
'hyphenate-before': '3',
'hyphenate-character': '3',
'hyphenate-lines': '3',
'hyphenate-resource': '3',
'hyphens': '3',
'image-resolution': '3',
'letter-spacing': '2',
'line-height': '2',
'line-stacking': '3',
'line-stacking-ruby': '3',
'line-stacking-shift': '3',
'line-stacking-strategy': '3',
'list-style': '2',
'list-style-image': '2',
'list-style-position': '2',
'list-style-type': '2',
'marquee-direction': '3',
'orphans': '2',
'overflow-style': '3',
'page': '2',
'page-break-inside': '2',
'pitch': '2',
'pitch-range': '2',
'presentation-level': '3',
'punctuation-trim': '3',
'quotes': '2',
'richness': '2',
'ruby-align': '3',
'ruby-overhang': '3',
'ruby-position': '3',
'speak': '2',
'speak-header': '2',
'speak-numeral': '2',
'speak-punctuation': '2',
'speech-rate': '2',
'stress': '2',
'text-align': '2',
'text-align-last': '3',
'text-emphasis': '3',
'text-height': '3',
'text-indent': '2',
'text-justify': '3',
'text-outline': '3',
'text-replace': '?',
'text-shadow': '3',
'text-transform': '2',
'text-wrap': '3',
'visibility': '2',
'voice-balance': '3',
'voice-family': '2',
'voice-rate': '3',
'voice-pitch': '3',
'voice-pitch-range': '3',
'voice-stress': '3',
'voice-volume': '3',
'volume': '2',
'white-space': '2',
'white-space-collapse': '3',
'widows': '2',
'word-break': '3',
'word-spacing': '2',
'word-wrap': '3',
//the mozilla extensions are all proprietary properties
'-moz-force-broken-image-icon': 'm',
'-moz-image-region': 'm',
'-moz-stack-sizing': 'm',
'-moz-user-input': 'm',
'-x-system-font': 'm',
//the opera extensions are all draft implementations of CSS3 properties
'-xv-voice-balance': 'o',
'-xv-voice-pitch': 'o',
'-xv-voice-pitch-range': 'o',
'-xv-voice-rate': 'o',
'-xv-voice-stress': 'o',
'-xv-voice-volume': 'o',
//the explorer extensions are all draft implementations of CSS3 properties
'-ms-text-align-last': 'e',
'-ms-text-justify': 'e',
'-ms-word-break': 'e',
'-ms-word-wrap': 'e'
},
//lists of longhand CSS properties indexed by their defining shorthand properties
//nb. the reason this is a group object rather than individual constants
//is so that we can refer to the longhand property lists by the name of the shorthand property
LONGHAND_PROPERTIES = {
//CSS2 shorthand properties
'margin': ['margin-top','margin-right','margin-bottom','margin-left'],
'padding': ['padding-top','padding-right','padding-bottom','padding-left'],
'outline': ['outline-width','outline-style','outline-color'],
//(the "border" shorthand defines a set of further shorthands
// so we have to break those down individually as well)
'border': ['border-width','border-style','border-color','border-top','border-right','border-bottom','border-left','border-top-width','border-right-width','border-bottom-width','border-left-width','border-top-style','border-right-style','border-bottom-style','border-left-style','border-top-color','border-right-color','border-bottom-color','border-left-color'],
'border-width': ['border-top-width','border-right-width','border-bottom-width','border-left-width'],
'border-style': ['border-top-style','border-right-style','border-bottom-style','border-left-style'],
'border-color': ['border-top-color','border-right-color','border-bottom-color','border-left-color'],
'border-top': ['border-top-width','border-top-style','border-top-color'],
'border-right': ['border-right-width','border-right-style','border-right-color'],
'border-bottom': ['border-bottom-width','border-bottom-style','border-bottom-color'],
'border-left': ['border-left-width','border-left-style','border-left-color'],
'list-style': ['list-style-type','list-style-image','list-style-position'],
'font': ['font-weight','font-style','font-variant','font-size','line-height','font-family'],
//(the "background" shorthand defines five properties in CSS2 [the first five]
// plus three additional properties in CSS3 [the last three])
'background': ['background-color','background-image','background-repeat','background-attachment','background-position','background-size','background-clip','background-origin'],
//CSS3 shorthand properties
'line-stacking': ['line-stacking-strategy','line-stacking-ruby','line-stacking-shift'],
'column-rule': ['column-rule-width','column-rule-style','column-rule-color'],
'columns': ['column-width','column-count'],
'pause': ['pause-before','pause-after'],
'rest': ['rest-before','rest-after'],
'cue': ['cue-before','cue-after'],
'mark': ['mark-before','mark-after'],
'transition': ['transition-property','transition-duration','transition-timing-function','transition-delay'],
'animation': ['animation-name','animation-duration','animation-timing-function','animation-delay','animation-iteration-count','animation-direction'],
//(I hate this module, it has no place in CSS, and the draft spec doesn't even make sense)
'target': ['target-name','target-new','target-position']
},
//-- regular expression constants --//
//-- nb. the use or omission of brackets is significant in each case --//
//-- for match() returns and backreferences, so don't edit them carelessly! --//
//replace a single pseudo-element, which we have to manually filter from some selectors
//to avoid them being incorrectly returned by querySelectorAll
//for simplicity we're allowing single or double-colon syntax for all of them
//even though the spec only requires support for single-colon syntax
//for pseudo-elements that were defined in CSS2 (first-letter, first-line, before and after)
//this is non-greedy because it's used to check one selector at a time, not multiple
//comma-delimited selectors, and a single selector is only allowed to contain one pseudo-element
REGEX_SINGLE_PSEUDO_ELEMENT = /[:]{1,2}(?:first\-(letter|line)|before|after|selection|value|choices|repeat\-(item|index)|outside|alternate|(line\-)?marker|slot\([_a-z0-9\-\+\.\\]*\))/i,
//match pseudo-elements, used as part of specificity calculation
//I know this code is almost an exact duplication of the single pseudo-element regex
//that we could avoid by saving the common code as a string and then using the RegExp constructor
//but I don't like using RegExp - it behaves differently than regex literals, and leaks memory,
//and I've done extensive testing with these ones now, it would make me too nervous to change them
//(because I did change them, and it made me nervous, so I changed them back!)
//anyway it's only 142 bytes, so it's not gonna break the bank to repeat it once :-P
REGEX_PSEUDO_ELEMENTS = /([:]{1,2}(?:first\-(letter|line)|before|after|selection|value|choices|repeat\-(item|index)|outside|alternate|(line\-)?marker|slot\([_a-z0-9\-\+\.\\]*\)))/ig,
//match all pseudo-classes except :not(), used as part of specificity calculation:
//for the purposes of specificity calculation, selectors inside :not() conditions
//are counted irrespective of the negation, however :not() itself is not counted as a pseudo-class
//this will also let through some fake permutations, like "nth-letter" and "only-of-child",
//but that doesn't really matter .. it might be fun for someone to fool with, you never know!
REGEX_PSEUDO_CLASSES_EXCEPT_NOT = /([:](?:(link|visited|active|hover|focus|lang|root|empty|target|enabled|disabled|checked|default|valid|invalid|required|optional)|((in|out\-of)\-range)|(read\-(only|write))|(first|last|only|nth)(\-last)?\-(child|of\-type))(?:\([_a-z0-9\-\+\.\\]*\))?)/ig,
//match attribute selectors (including substring-matching), used as part of specificity calculation
REGEX_ATTR_SELECTORS = /(\[\s*[_a-z0-9-:\.\|\\]+\s*(?:[~\|\*\^\$]?=\s*[\"\'][^\"\']*[\"\'])?\s*\])/ig,
//match ID selectors, used as part of specificity calculation
REGEX_ID_SELECTORS = /(#[a-z]+[_a-z0-9-:\\]*)/ig,
//match class selectors, used as part of specificity calculation
REGEX_CLASS_SELECTORS = /(\.[_a-z]+[_a-z0-9-:\\]*)/ig,
//an !important rule, for which we have to check an end-substring
//so we can't get confused by a "content" property with "!important" in its text
//which is admittedly unlikely, but entirely possible nonetheless
IMPORTANT_RULE = /\!\s*important\s*$/i,
//-- other data constants --//
//the library version string is used as a user-agent header for ajax requests
//abstracting it here is useful for its own sake, but also in case
//we ever have additional future needs for this particular data
LIBRARY_VERSION_STRING = 'CSSUtilities/.99',
//shortcuts for primitive values and data types
//these may not seem like shortcuts because they're longer than the originals
//but they give the compressor something to hook onto,
//and end up making a huge difference in the long run
//in any case, I've become quite fond of the abstraction...
TYPE_UNDEFINED = 'undefined',
TYPE_OBJECT = 'object',
TYPE_STRING = 'string',
TYPE_FUNCTION = 'function',
TYPE_BOOLEAN = 'boolean',
BOOLEAN_TRUE = true,
BOOLEAN_FALSE = false,
NULL_VALUE = null,
//...and there's a few internals that benefit from it too :-)
MODE_BROWSER = 'browser',
MODE_AUTHOR = 'author',
STATUS_ACTIVE = 'active',
STATUS_CANCELLED = 'cancelled',
STATUS_INACTIVE = 'inactive',
//detect some browsers for a few little tweaks where feature detection isn't appropriate
//ie. where it's not a case of, if feature a is available,
//it's a case of, browser x has bug y which requires specific fix z
//the webkit condition catches generic webkit UAs such as Google Chrome
//and the Adobe AIR runtime; the safari definition is split like this
//so that it doesn't get compressed (that single space is crucial!)
SAFARI = navigator.vendor == 'Apple Computer,' + ' Inc.',
SAFARI3 = SAFARI && /version\/3/i.test(navigator.appVersion),
KONQUEROR = navigator.vendor == 'KDE',
WEBKIT = /applewebkit/i.test(navigator.userAgent),
IEXPLORER = typeof document.uniqueID != TYPE_UNDEFINED,
OPERA = typeof window.opera != TYPE_UNDEFINED,
//timer speed for the routine that watches for stylesheets being
//enabled and disabled, and then re-initializes automatically
//and also the routine that waits for the script to have initialized
//when calling a public method using an asynchronous callback
//nothing we do with timers needs to be super-fast,
//just fast enough not to keep you hanging around when something happens!
//and since the watch timer might be running all the time
//we don't want it to demand anything like significant CPU time
WATCHER_SPEED_TIMER = 200,
//-- privileged environment variables --//
//what kind of rules data to collect, either "browser" centric or "author" defined
//if you choose browser-centric then the data comes from the document.styleSheets collection,
//which means that what we get is subject to each browser's implementation: it only includes
//stuff that the browser understands, and properties will be normalized in different ways
//for example, firefox normalizes colors to rgb, splits out background properties,
//adds a whole bunch of default -moz properties; opera normalizes colors to hex,
//also splits out background properties; safari adds a bunch of -webkit properties
//IE doesn't do much (apart from extensive capitalization, but we change that back)
//but it does do a few bits, like splitting out margin properties
//if you choose browser-agnostic then the data is loaded and parsed as plain text:
//all values are returned exactly as you defined them in the stylesheet
//with no normalization, the same spacing [except line breaks and tabs] and quote marks you used, etc.
//and all browsers return an identical collection of rules and properties
//irrespective of whether they actually implement those rules and properties
//this of course means that queries may return data which doesn't actually apply to that browser
//NOTE: in IE "author" mode, any CSS inside <style> elements has to be retrieved by making
// a synchronous request for the host page itself, as the only way to get raw text
// to work with rather than pre-normalized CSS; therefore the content of <style> blocks
// WILL NOT be included IF those <style> blocks were JS generated, OR the host-page request fails
// the same thing is true for the data in style attributes in author mode, unless "attributes" is false
//to change mode mid session, change this value through define() then call init() to re-initialize
//the default is browser because I think that's what people will intuitively expect
//and because author makes a bunch of http request, which may
//slow down the overall loading speed of the page (although hopefully not too much,
//given that all the data it's requesting will generally already be in the browser's cache,
//except of course where this is being run on a page they're viewing for the first time anyway!)
mode = MODE_BROWSER,
//whether to get data asynchronously
async = BOOLEAN_FALSE,
//context document
page = document,
//base for qualifying stylesheet hrefs, can't be defined until init
//because it refers to page, which may change between now and then
base,
//whether to watch for changes in the disabled state of stylesheets
//and re-initialize automatically in response
//we set this to false by default because it uses setInterval
//and we don't want that running unecessarily, when users may not
//even think about it and just leave it at the default setting
watch = BOOLEAN_FALSE,
//whether to include data from style attributes
attributes = BOOLEAN_TRUE,
//selectors API backup when native implementation is not available
//we bundle Selector.js by default, or others can be used via function argument
//null means we don't yet know whether native implementation is available
//if it is this will be set to false, unless it's already been set to true
//if set to true then all browsers will use the backup function
//qsa is a reference to a wrapper for the custom backup function, if specified
api = NULL_VALUE, qsa = NULL_VALUE, apidefined = BOOLEAN_FALSE;
//-- public method for modifying privileged variables --//
this.define = function(varname, value1, value2)
{
//validate definitions before saving their values
//for definable values that are invalid, throw a fatal error
//for non-definable values just silently fail
switch(varname)
{
//"mode"
case 'mode' :
//must be "author" or "browser"
if(typeof value1 != TYPE_STRING || !/^(author|browser)$/i.test(value1))
{
throw(new Error(FATAL_ERROR_INVALID_MODE));
}
//save the value
mode = value1;
//if mode is now "browser" check that "watch" is not null
if(mode == MODE_BROWSER && watch === NULL_VALUE)
{
throw(new Error(FATAL_ERROR_INVALID_WATCH));
}
break;
//"async"
case 'async' :
//must be a boolean
if(typeof value1 != TYPE_BOOLEAN)
{
throw(new Error(FATAL_ERROR_INVALID_ASYNC));
}
//save the value
async = value1;
break;
//"page"
case 'page' :
//if "api" already has a user-definition,
//throw the error that says this must be defined first
if(apidefined == BOOLEAN_TRUE)
{
throw(new Error(errorMessageNotAfterAPI.replace('%var', 'page')));
}
//must be a #document node
if(typeof value1.nodeType == TYPE_UNDEFINED || value1.nodeType != 9)
{
throw(new Error(FATAL_ERROR_NOT_DOCUMENT));
}
//save the value
page = value1;
break;
//"base"
case 'base' :
//if "api" already has a user-definition,
//throw the error that says this must be defined first
if(apidefined == BOOLEAN_TRUE)
{
throw(new Error(errorMessageNotAfterAPI.replace('%var', 'base')));
}
//must be an absolute URI string
if(typeof value1 != TYPE_STRING || !/^(((ht|f)tp[s]?)\:)/i.test(value1))
{
throw(new Error(FATAL_ERROR_NOT_ABSOLUTE_URI));
}
//save the value
base = value1;
break;
//"attributes"
case 'attributes' :
//must be a boolean
if(typeof value1 != TYPE_BOOLEAN)
{
throw(new Error(FATAL_ERROR_INVALID_ATTRS));
}
//save the value
attributes = value1;
break;
//"watch"
case 'watch' :
//must be true or false in browser mode
//must be true, false or null in author mode
if(!(typeof value1 == TYPE_BOOLEAN || (mode == MODE_AUTHOR && value1 == NULL_VALUE)))
{
throw(new Error(FATAL_ERROR_INVALID_WATCH));
}
//save the value
watch = value1;
break;
//"api"
case 'api' :
//must be a boolean
if(typeof value1 != TYPE_BOOLEAN)
{
throw(new Error(FATAL_ERROR_NOT_SAPI));
}
//and its function if specified must be a function
if(typeof value2 != TYPE_UNDEFINED)
{
if(typeof value2 != TYPE_FUNCTION)
{
throw(new Error(FATAL_ERROR_NOT_SAPI));
}
}
//the api property is false if we have native support, or null if we don't
//so if this gets set to [or left as] false that means use native where supported
//or if it gets set to true that means use the backup function for everyone
if(value1 == BOOLEAN_FALSE)
{
api = typeof page.querySelectorAll == TYPE_UNDEFINED;
}
else
{
api = BOOLEAN_TRUE;
}
//indicate that this variable has been used-defined
//so that we can prevent the later definition of "page" or "base"
//both of which need to have been defined first
//so that the custom api method can be tested in the correct context
apidefined = BOOLEAN_TRUE;
//if we have a function definition, pre-test it to make sure
//that it returns the correct data structure (an array or nodelist)
//and if not throw a fatal error
if(typeof value2 == TYPE_FUNCTION)
{
var nodes = value2('*', page);
if(typeof nodes != TYPE_OBJECT || nodes == NULL_VALUE || typeof nodes.length == TYPE_UNDEFINED)
{
throw(new Error(FATAL_ERROR_INVALID_SAPI));
}
//save the value if we're good
qsa = value2;
}
break;
}
};
//-- public methods --//
//initialize the library to re/build/refresh the internal data cache
this.init = function(callback)
{
//if initialization is already in progress, ignore this call
//we need to do this to prevent overlapping asynchronous calls
if(busyinit === BOOLEAN_TRUE)
{
return;
}
//set the busy flag to signify that initialization is in progess
busyinit = BOOLEAN_TRUE;
//clear the global ready flag
this._allready = BOOLEAN_FALSE;
//if querySelectorAll is supported, and api has not been set to true, set it to false
//false means we can use the native method if supported; true means use the backup for all
api = (typeof page.querySelectorAll != TYPE_UNDEFINED && api !== BOOLEAN_TRUE)
? BOOLEAN_FALSE
: BOOLEAN_TRUE;
//if base hasn't been defined, do so now
if(base == NULL_VALUE) { base = page.location.href; }
//detect XML (including HTML in XHTML mode)
//opera doens't support document.xmlVersion so it has a different test
this._isXML = (OPERA && page.documentElement.namespaceURI != NULL_VALUE)
|| (!OPERA && (typeof page.xmlVersion != TYPE_UNDEFINED && page.xmlVersion != NULL_VALUE));
//get the current page view media
this._viewmedia = getViewMedia();
//nullify and delete the rules data and debug stylesheets array
//this is only pertinent when re-initialising mid-session
//but it must be done to avoid retaining properties that are gone since last time
this._cssRules = undefined;
// delete this._cssRules;
this._stylesheets = undefined;
// delete this._stylesheets;
//define the local callback
function localcallback()
{
//clear the busy init flag
busyinit = BOOLEAN_FALSE;
//set the global ready flag
CSSUtilities._allready = BOOLEAN_TRUE;
//if callback is a function, save the reference and call it
if(typeof callback == TYPE_FUNCTION)
{
CSSUtilities.initcallback = callback;
CSSUtilities.initcallback();
}
//otherwise nullify and delete any existing reference
else
{
CSSUtilities.initcallback = NULL_VALUE;
delete CSSUtilities.initcallback;
}
//***DEV
//function dbSize(obj){function stringify(obj){var str = '';for(var i in obj){if(!obj.hasOwnProperty(i)) { continue; }if(typeof obj[i] == TYPE_STRING){str += obj[i];}else{str += stringify(obj[i]);}}return str;}return stringify(obj).length;}
//document.title = (typeof time != TYPE_UNDEFINED ? '(' + time + 'ms) | ' : '') + 'mode=' + mode + ' | async=' + async + ' | dbsize=' + (Math.round(dbSize(THIS._cssRules) / 10) / 100) + 'KiB | ' + new Date().toUTCString();
}
//create the new cssRules and stylesheets arrays
THIS._cssRules = [];
THIS._stylesheets = [];
//start a new ssidcounter for assigning stylesheet IDs
THIS.ssidcounter = 0;
//[re]create the core rules data array according to mode
//passing in the local callback reference for its completion
//all the other methods will work off this data so that
//we don't have to query the stylesheets collection more than once per session
//(unless it's dynamically changed, after which we have to re-init())
if(mode == MODE_AUTHOR)
{
createAuthorRulesData(localcallback);
}
else
{
createBrowserRulesData(localcallback);
}
};
//***DEV data object dump
//this.dumpDataObject = function(){
// var str='\n\n\n'
// + '--[ raw da'+'ta dump ]------------------------------------------------------------------------------\n\n'
// + '';
// for(var i=0; i<THIS._cssRules.length; i++){str+=i+':\n';for(var j in THIS._cssRules[i]){str += '\t"' + j + '": ';if(typeof THIS._cssRules[i][j] == TYPE_STRING) { str += '"'; }str += THIS._cssRules[i][j];if(typeof THIS._cssRules[i][j] == TYPE_STRING) { str += '"'; }str += '\n';}}return str;
//};
//public getCSSStyleSheets returns all the stylesheets
//that the class was able to retrieve data from
this.getCSSStyleSheets = function()
{
//process the arguments to create an indexed object
var args = processArguments(arguments, []);
//process the onfinished argument,
//which sets it to null if undefined or not a function, for ease of reference
args.onfinished = processOnfinishedArgument(args.onfinished);
//if the script hasn't been initialized, do so now
lateinit();
//create an internal wrapper for the rest of this function
//so that we can call it differently for callback or return
//then wrap the whole thing in a returned call to the
//dispatchReturn method, which abstracts that process,
return dispatchReturn(args.onfinished, function()
{
//return the debug stylesheets array
return THIS._stylesheets;
});
};
//public getCSSRules return all the style rules
//that apply to an element within a specified media
this.getCSSRules = function()
{
//process the arguments to create an indexed object
var args = processArguments(arguments, ['element','media','accept','altstates']);
//process the element argument
//this creates an element reference from an ID
//and/or throws an error if the element us undefined, null, or not an element
args.element = processElementArgument(args.element, 'getCSSRules');
//process the media argument
//this sets default if undefined, empty or null
//then splits it into a trimmmed array
args.media = processMediaArgument(args.media);
//process the accept argument
//this parameter allows us to limit how much data is retrieved
//for the sake of efficiency, so that we only have to keep in memory
//- and more to the point, only have to process -
//the properties that are needed by the calling operation
//the options are the same as the complete set returned by this.getCSSRules:
//"selector", "css", "media", "owner", "href", "ssid",
// "properties", "specificity", "index", "inheritance", "altstate";
// or "*" to return all values
//which is the default value if the accept argument is undefined, empty, null or "null"
//once processed it will either be the string value "*"
//or an object of properties, eg. selector and css
//this is so we can easily go typeof accept[key] against a property name
//rather than having to test for substrings each time
args.accept = processAcceptArgument(args.accept);
//if altstates is null, make it false
//the altstates parameter specifies whether to include selectors as being applicable
//that apply to interactive states of the element which it isn't necessarily in at the moment
//and which are all defined in pseudo-classes such as :visited and :hover
if(args.altstates == NULL_VALUE) { args.altstates = BOOLEAN_FALSE; }
//process the onfinished argument,
//which sets it to null if undefined or not a function, for ease of reference
args.onfinished = processOnfinishedArgument(args.onfinished);
//if the script hasn't been initialized, do so now
lateinit();
//create an internal wrapper for the rest of this function
//so that we can call it differently for callback or return
//then wrap the whole thing in a returned call to the
//dispatchReturn method, which abstracts that process,
return dispatchReturn(args.onfinished, function()
{
//get and return the rule objects for the specified media
//passing on the accept and altstates arguments directly for processing in getCSSRules
return getCSSRules(args.element, args.media, args.accept, args.altstates);
});
};
//public getCSSStyleSheetRules return all the style rules
//that we have in the dataset, within a specified media
//the final argument is the oncomplte callback, which is called onfinished
//just for internal grouping of callbacks (all public methods have onfinished)
this.getCSSStyleSheetRules = function()
{
//process the arguments to create an indexed object
var args = processArguments(arguments, ['media','accept','ssid']);
//process the media argument
//this sets default if undefined, empty or null
//then splits it into a trimmmed array
args.media = processMediaArgument(args.media);
//process the accept argument
args.accept = processAcceptArgument(args.accept);
//if accept contains "properties" but not "css"
//we need css to get properties, so add it to the accept object
//but then set a flag so that we know to delete it again at the end
if(typeof args.accept.properties != TYPE_UNDEFINED && typeof args.accept.css == TYPE_UNDEFINED)
{
args.accept.css = '';
var deletecss = BOOLEAN_TRUE;
}
//if the ssid argument is null, default to -1
//we need to check these strictly to avoid automatic type detection
//that could could confuse false with zero
if(args.ssid === NULL_VALUE) { args.ssid = -1; }
//process the onfinished argument,
//which sets it to null if undefined or not a function, for ease of reference
args.onfinished = processOnfinishedArgument(args.onfinished);
//if the script hasn't been initialized, do so now
lateinit();
//if ssid is specified but invalid (ie. there's no stylesheet with that ssid), throw an error
if(args.ssid !== -1 && arrayContains(this._stylesheets, args.ssid, 'ssid') == NULL_VALUE)
{
throw(new Error(errorMessageInvalidSSID.replace('%method', 'getCSSStyleSheetRules')));
}
//create an internal wrapper for the rest of this function
//so that we can call it differently for callback or return
//then wrap the whole thing in a returned call to the
//dispatchReturn method, which abstracts that process,
return dispatchReturn(args.onfinished, function()
{
//create a rules array
var rules = [];
//now iterate through the main rules array to build the data
for(var i=0; i<THIS._cssRules.length; i++)
{
//create the object for this rule
var cssrule = {};
//if the media of this rule matches the input criteria
if(mediaMatches(args.media, THIS._cssRules[i]))
{
//if the ssid argument is not -1
//then don't include this rule if its ssid doesn't match
//again, we use strict tests to avoid any ambiguity
//caused by automatic type conversion
if(args.ssid !== -1 && THIS._cssRules[i].ssid !== args.ssid) { continue; }
//for each property in this member, if it's defined in accept,
//or accept is "*", add it to the rule object
for(var j in THIS._cssRules[i])
{
if(!THIS._cssRules[i].hasOwnProperty(j)) { continue; }
if(args.accept == '*' || typeof args.accept[j] != TYPE_UNDEFINED)
{
cssrule[j] = THIS._cssRules[i][j];
}
}
//add the index property, if defined in accept or accept is "*"
if(args.accept == '*' || typeof args.accept.index != TYPE_UNDEFINED)
{
cssrule.index = i;
}
//add this rule to the array
rules.push(cssrule);
}
}
//if accept is an object containing "properties", or its the string "*"
//pass the rules array to the addSortedProperties object
//but with a false dosort argument so that it returns all the properties equally
//rather than identifying and marking all the cancelled and inactive ones
if(typeof args.accept.properties != TYPE_UNDEFINED || args.accept === '*')
{
rules = addSortedProperties(rules, BOOLEAN_FALSE);
}
//if the deletecss flag is defined, we need to
//nullify and delete all the css properties
if(typeof deletecss != TYPE_UNDEFINED)
{
for(var i=0; i<rules.length; i++)
{
rules[i].css = NULL_VALUE;
delete rules[i].css;
}
}
//return the final rules array
return rules;
});
};
//public getCSSProperties method returns all the style properties
//that are actually defined for this element in stylesheets
//(rather than computed values or style properties)
this.getCSSProperties = function()
{
//process the arguments to create an indexed object
var args = processArguments(arguments, ['element','media']);
//process the element argument
//this creates an element reference from an ID
//and/or throws an error if the element us undefined, null, or not an element
args.element = processElementArgument(args.element, 'getCSSProperties');
//process the media argument
//this sets default if undefined, empty or null
//then splits it into a trimmmed array
args.media = processMediaArgument(args.media);
//process the onfinished argument,
//which sets it to null if undefined or not a function, for ease of reference
args.onfinished = processOnfinishedArgument(args.onfinished);
//if the script hasn't been initialized, do so now
lateinit();
//create an internal wrapper for the rest of this function
//so that we can call it differently for callback or return
//then wrap the whole thing in a returned call to the
//dispatchReturn method, which abstracts that process,
return dispatchReturn(args.onfinished, function()
{
//create the properties object
var properties = {};
//get all the rules that apply to this element and media
//but we only need the "properties" from each rule
//and for altstates to be false so we only get the properties that actually apply now
var rules = getCSSRules(args.element, args.media, 'properties', BOOLEAN_FALSE);
//if there are no rules returned, return null for failure
if(rules.length == 0) { return NULL_VALUE; }
//iterate through the returned collection, then the
//properties object in each member, and add each property value
//in turn to the final properties object, if its status is "active"
for(var i=0; i<rules.length; i++)
{
for(var j in rules[i].properties)
{
if(!rules[i].properties.hasOwnProperty(j)
|| rules[i].properties[j].status != STATUS_ACTIVE) { continue; }
properties[j] = rules[i].properties[j].value;
}
}
//return the final properties object
return properties;
});
};
//public getCSSSelectors returns all the selectors that apply to an element
//the second argument specifies whether to return only direct selectors (true)
//or selectors which apply indirectly through inheritance (false)
this.getCSSSelectors = function()
{
//process the arguments to create an indexed object
var args = processArguments(arguments, ['element','media','directonly']);
//process the element argument
//this creates an element reference from an ID
//and/or throws an error if the element us undefined, null, or not an element
args.element = processElementArgument(args.element, 'getCSSSelectors');
//process the media argument
//this sets default if undefined, empty or null
//then splits it into a trimmmed array
args.media = processMediaArgument(args.media);
//if the directonly flag is null, default to true
if(args.directonly == NULL_VALUE) { args.directonly = BOOLEAN_TRUE; }
//process the onfinished argument,
//which sets it to null if undefined or not a function, for ease of reference
args.onfinished = processOnfinishedArgument(args.onfinished);
//if the script hasn't been initialized, do so now
lateinit();
//create an internal wrapper for the rest of this function
//so that we can call it differently for callback or return
//then wrap the whole thing in a returned call to the
//dispatchReturn method, which abstracts that process,
return dispatchReturn(args.onfinished, function()
{
//create an array of all selectors
var allselectors = [];
//now get all the rules that apply to this element and media
//we only need the "selector" property from each rule
//pass true as the altstates argument so that we get selectors
//for every state, not just those that apply by default
var rules = getCSSRules(args.element, args.media, 'selector', BOOLEAN_TRUE);
//then iterate through the rules collection and send the selector text from each one
//to the parseSelectorText method, which will split the selector into individual selectors
//and then add each one to the allselectors array, if it hasn't already been listed
for(var i=0; i<rules.length; i++)
{
var selectors = parseSelectorText(rules[i].selector);
for(var j=0; j<selectors.length; j++)
{
if(arrayContains(allselectors, selectors[j]) == NULL_VALUE)
{
allselectors.push(selectors[j]);
}
}
}
//if we want inherited as well as direct selectors
if(args.directonly == BOOLEAN_FALSE)
{
//we have to iterate through it and remove any selectors
//which don't apply to the element or to any of its ancestors
//nb. we remove any state-dependent pseudo-classes from the test selector each time
//so that we include rules that apply to non-default states like :hover
var node = args.element,
ancestors = [node];
while(node.parentNode)
{
ancestors.push(node.parentNode);
node = node.parentNode;
}
for(var i=0; i<allselectors.length; i++)
{
var applies = BOOLEAN_FALSE;
for(var j=0; j<ancestors.length; j++)
{
var nodes = getElementsBySelector(allselectors[i].replace(REGEX_PSEUDO_CLASSES_EXCEPT_NOT, ''));
if(nodes.length > 0)
{
for(var k=0; k<nodes.length; k++)
{
if(nodes[k] == ancestors[j])
{
applies = BOOLEAN_TRUE;
break;
}
}
}
if(applies == BOOLEAN_TRUE) { break; }
}
//if it doens't apply remove it from the array
if(applies == BOOLEAN_FALSE)
{
allselectors.splice(i, 1);
i --;
}
}
//then return the allselectors array and we're done
return allselectors;
}
//if we only want direct selectors,
//proceed to create the final output selectors array
var selectors = [];
//iterate through the allselectors array
//and evaluate each one against the original element
//and if it matches, add it to the final selectors array
//nb. we remove any state-dependent pseudo-classes from the test selector each time
//so that we include rules that apply to non-default states like :hover
for(var i=0; i<allselectors.length; i++)
{
var nodes = getElementsBySelector(allselectors[i].replace(REGEX_PSEUDO_CLASSES_EXCEPT_NOT, ''));
if(nodes.length > 0)
{
for(var j=0; j<nodes.length; j++)
{
if(nodes[j] == args.element)
{
selectors.push(allselectors[i]);
break;
}
}
}
}
//return the selectors array
return selectors;
});
};
//public getCSSSelectorSpecificity returns the specificity of a selector
//either as it applies to an element, or in abstraction
this.getCSSSelectorSpecificity = function()
{
//process the arguments to create an indexed object
var args = processArguments(arguments, ['selector','element']);
//if selector is undefined, null, empty, or not a string, throw the no selector error
if(typeof args.selector != TYPE_STRING
|| args.selector == NULL_VALUE || trim(args.selector) == '')
{
throw(new Error(errorMessageNoSelector.replace('%method', 'getCSSSelectorSpecificity')));
}
//if selector is a string containing any commas, throw the multiple selector error
else if(args.selector.indexOf(',') != -1)
{
throw(new Error(errorMessageMultipleSelector.replace('%method', 'getCSSSelectorSpecificity')));
}
//if the element is null, leave it like that
//if the element is not null, process the element argument
//this creates an element reference from an ID
//and throws an error if the reference is not there, or not an element
//but the condition allows it to be null if it was originally undefined or null
if(args.element != NULL_VALUE)
{
args.element = processElementArgument(args.element, 'getCSSSelectorSpecificity');
}
//process the onfinished argument,
//which sets it to null if undefined or not a function, for ease of reference
args.onfinished = processOnfinishedArgument(args.onfinished);
//if the script hasn't been initialized, do so now
lateinit();
//create an internal wrapper for the rest of this function
//so that we can call it differently for callback or return
//then wrap the whole thing in a returned call to the
//dispatchReturn method, which abstracts that process,
return dispatchReturn(args.onfinished, function()
{
//trim the selector
args.selector = trim(args.selector);
//if the element argument is in use, work out whether this selector applies to it
//we remove pseudo-classes while checking this so that we get those that apply
//even if only to a different pseudo-state; but because they're not permanently removed
//we'll still end up with a specificity value that includes any pseudo-class the selector has
if(args.element != NULL_VALUE)
{
var applies = BOOLEAN_FALSE,
nodes = getElementsBySelector(args.selector.replace(REGEX_PSEUDO_CLASSES_EXCEPT_NOT, ''));
if(nodes.length > 0)
{
for(var j=0; j<nodes.length; j++)
{
if(nodes[j] == args.element)
{
applies = BOOLEAN_TRUE;
break;
}
}
}
//if the selector doesn't apply
if(applies == BOOLEAN_FALSE)
{
//see if the selector applies to any of the elements ancestors
//and if it does then return zero specificity
var node = args.element, ancestors = [node];
while(node.parentNode)
{
ancestors.push(node);
node = node.parentNode;
}
for(var i=0; i<ancestors.length; i++)
{
if(nodes.length > 0)
{
for(var j=0; j<nodes.length; j++)
{
if(nodes[j] == ancestors[i])
{
applies = BOOLEAN_TRUE;
break;
}
}
}
if(applies == BOOLEAN_TRUE) { break; }
}
if(applies == BOOLEAN_TRUE)
{
return [0,0,0,0];
}
//if it doesn't even apply to an ancestor, return null
else { return NULL_VALUE; }
}
}
//if the element argument is not in use, or the selector does apply to it,
//get and return the specificity of this selector
return getSelectorSpecificity(args.selector);
});
};
//-- shared private method that finalizes the public css methods --//
//dispatch return method deals with the onfinished argument of each public method
//by calling the applicable inner wrapper and then returning or calling the callback
function dispatchReturn(onfinished, wrapper)
{
//if the ready flag is not true and a callback is not defined
//return undefined to signal that init is incomplete
//putting this after the lateinit call ensures that it will
//only affect asynchronous uses without the internal callback, when the data is not ready
//it can then be caught by users by testing this method's return value against undefined
if(THIS._allready !== BOOLEAN_TRUE && onfinished == NULL_VALUE)
{
return;
}
//or if the ready flag is true and no callback is set
//just call and return the internal wrapper
if(onfinished == NULL_VALUE)
{
return wrapper();
}
//otherwise create a timer to wait for the ready property
//then call the internal wrapper and send its return through the callback
else
{
//if init has happened before then the data will already be ready
//so if that's the case just do the calback straight away
if(THIS._allready === BOOLEAN_TRUE)
{
onfinished(wrapper());
}
//otherwise start the timer to wait for it
//this runs at the same speed as the watch timer though it could
//actually afford to be slower, it's fine at this speed too,
//and means we can save on having to have a separate value
//it also never times out; I suppose it could, but it doesn't really
//matter if it does end up running forever - it's hardly a major overhead!
else
{
var readywaiter = window.setInterval(function()
{
if(THIS._allready === BOOLEAN_TRUE)
{
window.clearInterval(readywaiter);
onfinished(wrapper());
}
}, WATCHER_SPEED_TIMER);
}
}
}
//-- private methods for building the author-mode rules data array --//
//create the data array of css style rules from all document stylesheets
//this all has to be structured in such a way that it works
//with either synchronous or asynchronous network data retrieval
function createAuthorRulesData(oncomplete)
{
//what we're going to do is load each stylesheet and parse it as text
//so that we don't have to deal with each browser's normalizing behavior
//we get a rules collection that is exactly as author defined
//and the same in every browser, irrespective of its own support
//so to begin with, we need an array of stylesheet URIs
//which we can then load and get the text from
//or for style blocks we can just grab the text straight away
THIS._cssdata = [];
//if this is IE and the page contains any style element
//get the raw text inside the page's <style> nodes
//passing the continuation method as its callback
//this works by making a request for the host page and
//then parsing the HTML to get the contents of <style> elements
//we only need to do this in "author" mode,
//in IE, if the page contains any <style> elements
//because their innerHTML/nodeData returns normalized style information
//instead of just the plain text inside the style node, as with other browsers
if(IEXPLORER && getTheseElements('style').length > 0)
{
getRawStyleData(continuation);
}
//otherwise just call the continuation method directly
else { continuation(); }
//continuation method after getting the raw style data
function continuation()
{
//get the collection of all nodes that refer to stylesheets
var stylenodes = getStyleSheetNodes();
//now iterate through the collection of style nodes
//and process each one to get the initial data we need
for(var i=0; i<stylenodes.length; i++)
{
processStyleNode(
stylenodes[i],
stylenodes[i].nodeType == 7
? stylenodes[i].target.toLowerCase()
: stylenodes[i].nodeName.toLowerCase()
);
}
//the next few methods are called in order, each progressively
//by the callback from the previous one, so they can work asynchronously
//1. wrapper for recurring instances of the stage2 check and request method
function stage1_checkStyleSheetForDataWrapper(cdcount)
{
//pass this cssdata to the stage2 check and request method
stage2_checkStyleSheetForData(THIS._cssdata[cdcount],
//and when that completes
function(cssdata)
{
//save back the modified object
THIS._cssdata[cdcount] = cssdata;
//if we have more to process, recur
if(cdcount + 1 < THIS._cssdata.length)
{
stage1_checkStyleSheetForDataWrapper(cdcount + 1);
}
//otherwise kick off the stage 3 check for imports process
else
{
stage3_checkStyleSheetForImportsWrapper();
}
});
}
//2. check to see if we have the css text for a stylesheet
//and make a request for it if we don't
function stage2_checkStyleSheetForData(cssdata, oncomplete)
{
//if the href is null then this is a style block
//in which case we already have the text,
//so just pass the unmodified cssdata back through the oncomplete callback
if(cssdata.href == NULL_VALUE) { oncomplete(cssdata); }
//otherwise request the stylesheet, which sends back a data object
//containing the text of a successful request, or an error message
else
{
requestStyleSheet(cssdata.href, function(dataobject)
{
//if we have an error message, record it to the cssdata object
//and nullify the stylenode reference
//then if the error is an http error or "not css", set the media to "none"
if(dataobject.message != NULL_VALUE)
{
cssdata.message = dataobject.message;
cssdata.stylenode = NULL_VALUE;
if(cssdata.message == ERROR_NOT_CSS || /^[1-9]{1,3}/.test(cssdata.message))
{
cssdata.media = MEDIA_NONE;
}
}
//otherwise strain and save the text, leaving valid imports in place for now
//because we'll need to extract them later to load them in turn
else
{
cssdata.text = strainStyleSheet(dataobject.text, BOOLEAN_FALSE);
}
//call oncomplete with the modified cssdata object
oncomplete(cssdata);
});
}
}
//3. wrapper for calling recurring instances of the stage5 check and request import method
function stage3_checkStyleSheetForImportsWrapper()
{
//4. check a stylesheet for imports
function stage4_checkStyleSheetForImports(cdindex)
{
//local callback for this method
function localcallback(increment)
{
//if there are any more stylesheets to check, recur on the next one
//increasing the iterator as specified in case we need to re-check
//the index that we just checked, because it's now a nested import
//which is added before its parent in the array
//and that's why we have a .tested flag, to avoid testing the same
//stylesheet more than once if we encounter it more than once
if(cdindex + 1 < THIS._cssdata.length)
{
stage4_checkStyleSheetForImports(cdindex + increment);
}
//otherwise go on to the final stage7 data processor
else
{
stage7_processStyleSheetsData();
}
}
//if we've already tested this stylesheet
//just call the local callback with a standard increment
if(THIS._cssdata[cdindex].tested == BOOLEAN_TRUE)
{
localcallback(1);
}
//else proceed
else
{
//look for imports in this stylesheet text
var importsheets = extractImportStatements(THIS._cssdata[cdindex].text);
//mark that we've tested this stylesheet
THIS._cssdata[cdindex].tested = BOOLEAN_TRUE;
//assign its ssid; we waited until now to do this
//so that we can assign them in order of occurence
THIS._cssdata[cdindex].ssid = THIS.ssidcounter++;
//if we have any imports
if(importsheets.length > 0)
{
//5. wrapper for passing each import
//to the stage6 check and request import method
function stage5_checkImportForDataWrapper(impindex)
{
//pass this import data to the stage6 check and request import method
stage6_checkImportForData(
cdindex,
impindex,
importsheets[impindex],
//and when that completes
function(cssdata, cdnumber, impnumber)
{
//add the object to the master array before its parent
//which means that imports go before their parent,
//and sibling imports maintain their relative order;
//this is the correct source order for specificity
//and can make a huge difference when it comes to rule sorting
THIS._cssdata.splice((cdindex + impnumber), 0, cssdata);
//if there are any more imports to check, recur on the next one
if(impnumber + 1 < importsheets.length)
{
stage5_checkImportForDataWrapper(impnumber + 1);
}
//otherwise call the local callback
//with a zero increment value so that
//the iterator loops back on itself
//to check the stylesheet we just added
else
{
localcallback(0);
}
});
}
//pass the first import index to stage5 to kick it off
stage5_checkImportForDataWrapper(0);
}
//otherwise call the local callback straight away
//with a standard increment value
else
{
localcallback(1);
}
}
}
//pass the first index to stage4 to kick it off
stage4_checkStyleSheetForImports(0);
}
//6. check and request an import stylesheet
function stage6_checkImportForData(cdcount, impcount, importobj, oncomplete)
{
//create a data object for the stylesheet and assign it an ssid
var cssdata = {
'tested' : BOOLEAN_FALSE,
'owner' : '@import',
'media' : MEDIA_ALL,
'stylenode' : THIS._cssdata[cdcount].stylenode,
'href' : importobj.href,
'text' : ''
};
//save the media for this import
cssdata.media = importobj.media;
//get the context from the nearest ancestral non-import
var n = cdcount, parent = THIS._cssdata[n];
while(parent.href == cssdata.href)
{
parent = THIS._cssdata[n--];
}
context = parent.media;
//if the import itself has its own media specified, save that
//to the xmedia property, otherwise use the default value "all"
//(I originally had this as media|context, but then I figured,
// the context is not the import's media, is it?)
cssdata.xmedia = importobj.hasownmedia ? importobj.media : MEDIA_ALL;
//then contextualize the import's media
cssdata.media = contextualizeMediaTypes(context, cssdata.media);
//local finalizing function
function finalizedata(fdata)
{
//if we have an error message, record it to the cssdata object
//and nullify the stylenode reference
//then if the error is an http error or "not css", set the media to "none"
if(fdata.message != NULL_VALUE)
{
cssdata.message = fdata.message;
cssdata.stylenode = NULL_VALUE;
if(cssdata.message == ERROR_NOT_CSS || /^[1-9]{1,3}.*$/.test(cssdata.message))
{
cssdata.media = MEDIA_NONE;
}
}
//otherwise strain and save the text, leaving valid imports in place for now
//because we'll need to check for imports when its iteration comes round
//in case an imported stylesheet also contains imports
//the import statements will get removed when we do
//the final text processing operation to create the rules objects
else
{
cssdata.text = strainStyleSheet(fdata.text, BOOLEAN_FALSE);
}
//pass the data object to the oncomplete callback
oncomplete(cssdata, cdcount, impcount);
}
//if the cssdata object already has this stylesheet href,
//mark it as a duplicate and don't make a request for it
//this is particularly important to prevent infinite recursion
//for example, where two stylesheets import each other
if(arrayContains(THIS._cssdata, importobj.href, 'href') != NULL_VALUE)
{
//** how do we disable this duplicate when
//** we don't have its stylesheet reference?
//call the finalize function
//with a data object containing the redundent duplication message
//in the same format as would be returned by the request method
finalizedata({
'text' : '',
'message' : MESSAGE_DUPLICATE
});
}
//otherwise request the stylesheet, which sends back a data object
//containing the text of a successful request, or an error message
else
{
requestStyleSheet(importobj.href, finalizedata);
}
}
//7. process the cssdata object to create the final cssRules array
function stage7_processStyleSheetsData()
{
//now we're read start building the final cssRules data array
//we'll have an array of objects each of which contains the raw stylesheet text
//so what we have to do is parse each one to extract the style information
for(var i=0; i<THIS._cssdata.length; i++)
{
//check if we already have this stylesheet data
//which we can only do for stylesheets that have an href
//and will only be caused by duplication stylesheet includes
//we already dealt with duplicate imports at the request stage
//so this step is to find, mark and disable any duplicate top-level includes
//we first have to check though that we're not comparing an existing duplicate
//otherwise we could end up marking both instances as duplicates!
//we use the arrayContains method to do this, which had to be modified so that
//if it finds a match, it doesn't just return true, it returns the match it found,
//which is a _cssdata object that we can check to see if it's already a duplicate
if(THIS._cssdata[i].href)
{
var arycontains = arrayContains(THIS._stylesheets, THIS._cssdata[i].href, 'href');
if(arycontains != NULL_VALUE && arycontains.message != MESSAGE_DUPLICATE)
{
//add duplication data to the debug stylesheets array
THIS._stylesheets.push({
'ssid' : THIS._cssdata[i].ssid,
'href' : THIS._cssdata[i].href,
'owner' : THIS._cssdata[i].owner,
'media' : THIS._cssdata[i].media,
'xmedia' : THIS._cssdata[i].xmedia,
'stylenode' : NULL_VALUE,
'rules' : 0,
'message' : MESSAGE_DUPLICATE
});
//however if there are duplicate stylesheets,
//subsequently disabling one of them (ie. through stylesheet switching)
//will cause the other one's rules to take effect, which the script
//won't be aware of because its rules are not included
//so since we don't have the option of just including duplicates
//because of the possibility of infinite recursion,
//what we'll do instead is just forcibly disable any duplicates we find
if(THIS._cssdata[i].stylenode)
{
THIS._cssdata[i].stylenode.disabled = BOOLEAN_TRUE;
}
//continue to the next stylesheet
continue;
}
}
//if this data object has a message property then something went wrong
//either a security error, or an http error while loading it
//so just ignore it and continue to the next one
if(typeof THIS._cssdata[i].message != TYPE_UNDEFINED)
{
//record an entry to the debug stylesheets array
THIS._stylesheets.push({
'ssid' : THIS._cssdata[i].ssid,
'href' : THIS._cssdata[i].href,
'owner' : THIS._cssdata[i].owner,
'media' : THIS._cssdata[i].media,
'xmedia' : THIS._cssdata[i].xmedia,
'stylenode' : THIS._cssdata[i].stylenode,
'rules' : 0,
'message' : THIS._cssdata[i].message
});
//continue to the next stylesheet
continue;
}
//send the data object to the css text parsing method
//in which we'll actually build the cssRules data array
//and return the number of rules we found, for the debug stylesheets array
var rulecount = parseStyleSheetCSSText(THIS._cssdata[i]);
//record an entry to the debug stylesheets array
THIS._stylesheets.push({
'ssid' : THIS._cssdata[i].ssid,
'href' : THIS._cssdata[i].href,
'owner' : THIS._cssdata[i].owner,
'media' : THIS._cssdata[i].media,
'xmedia' : THIS._cssdata[i].xmedia,
'stylenode' : THIS._cssdata[i].stylenode,
'rules' : rulecount,
'message' : MESSAGE_OK
});
}
//and once we get here we can delete the cssdata object
//first setting it to null to ensure garbage collection in IE
THIS._cssdata = NULL_VALUE;
delete THIS._cssdata;
//do a final sort of the _stylesheets array, putting it in ssid order
THIS._stylesheets.sort(function(a, b) { return a.ssid - b.ssid; });
//nullify and delete the ssidcounter
THIS.ssidcounter = NULL_VALUE;
delete THIS.ssidcounter;
//if the watch setting is strictly true, watch for changes in disabled state
if(watch === BOOLEAN_TRUE) { watchStyleSheets(); }
//call the oncomplete callback, if defined
if(typeof oncomplete == TYPE_FUNCTION)
{
oncomplete();
}
}
//0. if we have no data in _cssdata, jump straight to the final stage7
//otherwise pass the first index to stage1 to kick everything off
if(THIS._cssdata.length == 0)
{
stage7_processStyleSheetsData();
}
else
{
stage1_checkStyleSheetForDataWrapper(0);
}
}
}
//get the raw text inside the page's <style> nodes
function getRawStyleData(oncomplete)
{
//make a request for the page specified in the base property
ajaxload(async, base,
//if the load succeeds
function(code)
{
//re-get the "style" elements collection
var stylenodes = getTheseElements('style');
//save a copy of the html and split it by opening style tags
var styleblocks = code.split(/<([^:]+:)?style[^>]*>/i);
//delete the first member, which will be
//everything up to the first style element
styleblocks.splice(0, 1);
//now iterate and split each member by closing style tags
//throw away the second member and keep the first
//to leave us with just the css data inside it
//in an array indexed by the element's position in the "style" collection
//which we'll then save as a __css property of the source node,
//just as we did for styleattrs, so that DOM changes aren't a problem
for(var j=0; j<styleblocks.length; j++)
{
stylenodes[j].__css = styleblocks[j].split(/<\/([^:]+:)?style>/i)[0];
}
//we don't need the styleblocks array anymore
//so we can delete it to save a bit of memory
//Every Little Helps :) (tm)
styleblocks = undefined;
// delete styleblocks;
//call the callback
oncomplete();
},
//there's no known reason why we should get failure
//but if we do, the lack of .__css properties will be
//handled where it occurs, so all we have to do here
//is call the oncomplete callback as if nothing happens
function()
{
oncomplete();
});
}
//get the collection of all nodes that refer to stylesheets
function getStyleSheetNodes()
{
//create the array
var stylenodes = [];
//begin by looking for xml-stylesheet processing instructions
var kids = page.childNodes;
for(var i=0; i<kids.length; i++)
{
if(kids[i].nodeType == 7 && kids[i].target.toLowerCase() == 'xml-stylesheet')
{
stylenodes.push(kids[i]);
}
}
//then get a raw collection of all elements in all namespaces
//or just all nodes if we're not in any form of XML, or the NS method fails
var allnodes = getTheseElements('*');
//then iterate through that to look for stylesheet nodes
//link elements are included from any namespace according to their rel attribute,
//as are style elements from any namespace (isn't there an svg:style for example?)
for(var i=0; i<allnodes.length; i++)
{
if((/link/i.test(allnodes[i].nodeName)
&& /stylesheet/i.test(allnodes[i].getAttribute('rel')))
|| /style/i.test(allnodes[i].nodeName))
{
stylenodes.push(allnodes[i]);
}
}
//return the final style nodes array
return stylenodes;
}
//process a style node to get the preliminary data we need
function processStyleNode(node, owner)
{
//create a data object for this stylesheet and assign it an ssid
var cssdata = {
'tested' : BOOLEAN_FALSE,
'owner' : owner,
'media' : MEDIA_ALL,
'stylenode' : node,
'href' : NULL_VALUE,
'text' : ''
};
//if the owner is an xml-stylesheet, parse its data property
//to create a dictionary of its pseudo attributes
//we use match here instead of just splitting by delimiting whitespace,
//because the values themselves might contain whitespace
//nb. xml-stylesheet regexes must be case-insensitive, to match implementations,
//in which all aspects of an xml-stylesheet PI are case-sensitive
if(/xml\-stylesheet/.test(owner))
{
var pseudoattrs = {},
nodedata = node.data.match(/([a-z]+=[\'\"][^\'\"]*[\'\"])/gm);
for(var i=0; i<nodedata.length; i++)
{
nodedata[i] = nodedata[i].split('=');
pseudoattrs[nodedata[i][0]] = nodedata[i][1].substr(1, nodedata[i][1].length - 2);
}
}
//except for safari or konqueror with nodes other than xml-stylesheet, because it always returns false
//(so this is the exact opposite of what IE does!)
//nb. though in fact, this /still/ always returns false unless the stylesheet has been
//pre-disabled, by having a default disabled=disabled attribute, or by
//having its .disabled property set to false through scripting
//if the node does have a disabled attribute, then node.sheet is null
if((SAFARI || KONQUEROR || WEBKIT) && !/xml\-stylesheet/.test(owner))
{
var isdisabled = node.disabled;
}
//get the stylesheet's disabled state from the node's styleSheet/sheet property
//we have to do this for IE because the node itself always returns not disabled
//for others we only have to use it for xml-stylesheet,
//but we may as well use it all the time to save a bit of code forking
//nb. although we know this condition is IE vs. others, an object test is better
//just in case any unknown devices have the same object reference
else
{
//get the style node reference, and if its null set isdisabled to true
//(which will happen in firefox for an alternate stylesheet that's
// missing its title attr and hence can never be switched on)
//otherwise set it according to the style node's disabled state
var disnode = node[typeof node.styleSheet != TYPE_UNDEFINED ? 'styleSheet' : 'sheet'],
isdisabled = disnode == NULL_VALUE ? BOOLEAN_TRUE : disnode.disabled;
}
//in opera, the stylesheet returns as disabled if it applies to media
//other than the one currently in force, such as "print" when the view is "screen"
//so what we'll do is temporarily switch the media to whatever the current view is
//then read its disabled property again, to find out if it's
//really disabled or just applicable to a non-view media
//this happens so fast that no visual change should be apparent
if(OPERA && isdisabled && node.sheet.media.length > 0)
{
var realmedia = node.sheet.media.mediaText;
node.sheet.media.mediaText = THIS._viewmedia;
isdisabled = node.sheet.disabled;
node.sheet.media.mediaText = realmedia;
}
//we don't want data from disabled stylesheets in the data set,
//unless this is author mode and the watch setting is null
//so if it is (and it's not) add a message to the objeect to that effect,
//the presence of which will then prevent it from being proceessed any further
//however don't stop here just yet because we want the href and media as well
if(watch !== NULL_VALUE && isdisabled)
{
cssdata.message = MESSAGE_DISABLED;
}
//if the owner is an xml-stylesheet instruction
//get the href from the pseudoattrs object and qualify it
if(/xml\-stylesheet/.test(owner))
{
cssdata.href = qualifyHREF(pseudoattrs.href, base);
}
//or if the owner node is a link element get the href from its href attribute
//we're using getAttribute so that we get the literal attribute value,
//rather than a qualified value, so that we can qualify it ourselves
//using the value for "base"; the second argument is IE proprietary
//and the value 2 tells getAttribute to return the literal attribute value
//rather than an "interpolated" value, ie. a property value, which is what
//IE normally returns for attributes, rather than returning attribute values:
//http://reference.sitepoint.com/javascript/Element/getAttribute
else if(/link/i.test(owner))
{
cssdata.href = qualifyHREF(node.getAttribute('href', 2), base);
}
//otherwise it will be a style element
else
{
//if it's not disabled or watch is null
if(!isdisabled || watch === NULL_VALUE)
{
//if this is IE, the nodeData is actually CSS to IE, not just plain text like everything else
//so it gets formatted like CSS (capitalized property names and type selectors, some value normalizing)
//what we have to do then is refer to the raw page HTML we saved earlier
//and grab the css text from that, not from this style element
if(IEXPLORER)
{
//if we have no __css property for this node, record a network error
//because that would mean we were unable to get the HTML during init
if(typeof node.__css == TYPE_UNDEFINED)
{
cssdata.message = ERROR_NETWORK;
}
//otherwise grab that data and pass it to the strain method
//to do some initial pre-parsing, but leaving valid imports in place for now
else
{
cssdata.text += strainStyleSheet(node.__css, BOOLEAN_FALSE);
}
}
//for any other browser
else
{
//save the text content, doing some initial pre-parsing as we go
//but leaving valid imports in place for now
//we used to get the CSS data from the style node as simply "node.firstChild.nodeValue"
//but running that on a stylesheet created as an <html:style> node
//within CodeBurner's results.xhtml page, the style node returned zero childNodes
//(because adding rules using insertRule doesn't create any;
// if I modified the code to add them as text nodes then they did show up
// - and still worked as CSS - but that's doesn't help us here)
//so we need to be aware of that potential situation and handle it acccordingly
//furthermore, when running it on a normal XHTML page, it missed rules
//inside HTML comments or CDATA sections, because the main data is no longer
//the first child text node, it's the node data of the second child comment or cdata-section node
//but we don't necessarily want to just put them back, because they might not apply -
//we need to match their inclusion against whether browsers actually apply them
//and after cross-browser testing in applicable modes, we find the following:
//
// /-------------------------------------------------------\
// | Browser | HTML mode | XHTML mode |
// | | Comments | CDATA | Comments | CDATA |
// |-------------------------------------------------------|
// | Opera | YES | NO | NO | YES |
// | Firefox | YES | NO | NO | YES |
// | Safari | YES | NO | YES | YES |
// | IE | YES | NO | -- | -- |
// \-------------------------------------------------------/
//
//so, taking all of that into account, we made a modification to strainStyleSheet
//that pre-removes the contents of CDATA or Comment sections,
//according to mode and browser using the rules above,
//before parsing what's left as the complete CSS stylesheet
//then we modified the code here to assemble a content string from all child nodes
//including the comment or CDATA markers themselves
//(so that the strain method would have some deletion hooks to work with)
//using innerHTML proved unreliable because konqueror was entity-encoding it
//and firefox was missing data in some situations - even though it was situations
//where we didn't need it anyway, it still made me slightly uncomfortable
//nb. the "CDATA" token is case-sensitive
//"cdata" is invalid and causes an XML parsing exception
for(var nodedata='', kids=node.childNodes, i=0; i<kids.length; i++)
{
switch(kids[i].nodeType)
{
case 3 : nodedata += kids[i].nodeValue; break;
case 4 : nodedata += '<![CDATA[' + kids[i].nodeValue + ']]>'; break;
case 8 : nodedata += '<!--' + kids[i].nodeValue + '-->'; break;
}
}
cssdata.text += strainStyleSheet(nodedata, BOOLEAN_FALSE);
}
}
}
//get the media from its media attribute / pseudoattribute,
//or set to default if it doesn't have one defined
if(/xml\-stylesheet/.test(owner))
{
cssdata.media = typeof pseudoattrs.media != TYPE_UNDEFINED
? pseudoattrs.media : MEDIA_ALL;
}
else
{
cssdata.media = node.getAttribute('media')
? node.getAttribute('media') : MEDIA_ALL;
}
//then normalize the spacing between each delimited value
//(the ',' + ' ' thing is to avoid compression)
cssdata.media = cssdata.media.split(/,\s*/).join(',' + ' ');
//and copy it to the xmedia property
cssdata.xmedia = cssdata.media;
//add this data object to the array
THIS._cssdata.push(cssdata);
}
//make a request for a stylesheet
function requestStyleSheet(href, oncomplete)
{
//create a request data object to return
var requestdata = {
'text' : '',
'message' : NULL_VALUE
};
//make a request for the stylesheet, passing the async value
ajaxload(async, href,
//if it succeeds, save the text content to the data object
//doing some initial pre-parsing as we do it
function(responsetext, contype)
{
//if we have a content type and it's not text/css
//(local or chrome requests, with a status of 0,
// also return a null or empty content type,
// so if we don't have one we'll just have to assume it's okay)
if((typeof contype == TYPE_STRING && trim(contype) != '')
&& !/^(text\/css)/.test(contype))
{
//if the content type is text/html
//scan it to see if it is in fact a server error page
//and if it is see if we can extract the status error from its <title>
//and if that succeeds record it to the data message property
if(/^(text\/html)/.test(contype))
{
var matches = responsetext.match(/<title>([1-9]{1,3}[^<]+)<\/title>/i);
if(matches && matches.length >= 2)
{
requestdata.message = matches[1];
}
}
//if we haven't got a message, just set the "data is not css" message
if(requestdata.message == NULL_VALUE)
{
requestdata.message = ERROR_NOT_CSS;
}
}
//otherwise if we're good (or we have to assume we're good)
//save the response text to the requestdata object
else
{
requestdata.text = responsetext;
}
//call oncomplete with the request data
oncomplete(requestdata);
},
//if the request fails for any reason
function(statustext)
{
//save the status text message to the requestdata object
requestdata.message = statustext;
//call oncomplete with the request data
oncomplete(requestdata);
});
}
//extract @import statements from the raw text of a stylesheet
function extractImportStatements(styletext)
{
//create the array for storing any we find
var importsheets = [];
//define regexes for matching import statements
//which may include media definitions and queries
var importreg = /@import\s*(?:url\s*\(\s*[\'\"]?|[\'\"])([^\'\"\)]+)(?:[\'\"]|[\'\"]?\s*\))([^;]*)(;|$)/ig;
//look for @import statements in the text
var matches = styletext.match(importreg);
//if we have matches
if(matches)
{
for(var i=0; i<matches.length; i++)
{
//extract and qualify the href
var href = qualifyHREF(matches[i].replace(importreg, '$1'), base);
//extract the media, from which we may have to trim
//leading space and residual close bracket, or set default
var media = trim(trim(matches[i].replace(importreg, '$2')).replace(/^\)/, ''));
//set the hasownmedia flag to true, indicating that
//the import statement had a media flag as part of it
//this is used to differentiate the original media from a
//contextual media in the final data set
var hasownmedia = BOOLEAN_TRUE;
//then if it didn't in fact one, set this flag to false,
//and the media itself to the default value "all"
if(!media)
{
hasownmedia = BOOLEAN_FALSE;
media = MEDIA_ALL;
}
//add to the importsheets array
importsheets.push({
'href' : href,
'media' : media,
'hasownmedia' : hasownmedia
});
}
}
//return the final array
return importsheets;
}
//remove unwanted information from stylesheet text
function strainStyleSheet(styletext, importsheets)
{
//css comments
//this must come first because it expects multi-line data
//and without that it will fail to remove complex comments
//that contain lots of intermediate stars and slashes
styletext = styletext.replace(/(\/\*([^*]|(\*+([^*/])))*\*+\/)/gm, '');
//in HTML mode, delete anything inside and including CDATA markers
//in XHTML mode but not webkit browsers, delete anything inside and including HTML comments
//we do this because no supported browsers implement such rules (except for webkit as noted)
//for more about this see processStyleNode()
//(near the end - the notes preceding the last call to strainStyleSheet())
//nb. the "CDATA" token is case-sensitive
//"cdata" is invalid and causes an XML parsing exception
if(!THIS._isXML)
{
styletext = styletext.replace(/(<\!\[CDATA\[([^\]]|(\]+([^>])))*\]+>)/gm, '');
}
if(THIS._isXML && (!(SAFARI || KONQUEROR || WEBKIT)))
{
styletext = styletext.replace(/(<\!\-\-([^\-]|(\-+([^>])))*\-+>)/gm, '');
}
//convert one or more contiguous tab to a single space
//and any resulting multiple spaces to single spaces
//we don't want to just remove them in case they're used in syntax
//eg. the media query "screen and (color)" would otherwise be
//converted to "screenand(color)", which would evaluate to "none"
//of course this means we'll be left with whitespace in the stylesheet css text
//but we do plenty of trimming when we parse it into properties
//so that won't matter in the final analysis, it just means that
//the css property may contain a space between each property:value; pair
styletext = styletext.replace(/[\t]+/g, ' ').replace(/[ ][ ]/g, ' ');
//line-breaks
styletext = styletext.replace(/[\r\n]/g, '');
//@charset and @namespace statements
styletext = styletext.replace(/@(charset|namespace)[^;]+;/igm, '');
//if the importsheets flag is defined and true, remove @import statements
if(typeof importsheets != TYPE_UNDEFINED && importsheets == BOOLEAN_TRUE)
{
styletext = styletext.replace(/@import[^;]+;/igm, '');
}
//otherwise we need to check and remove any imports which are invalid
//for imports to be valid they must be at the start of the stylesheet
//so any which come after a block of normal CSS don't apply and get deleted
else if(/@import[^;]+;/i.test(styletext))
{
//add split markers around each import statement and split by those markers
//(IE doesn't support backreferences after split(), so we had to do this
// to get data split by import statements without losing the statements themselves)
var cssdata = styletext.replace(/(@import[^;]+;)/igm, '{SPLIT}$1{SPLIT}').split('{SPLIT}');
//then clear the original text variable and iterate through the fragments
//so we can re-compile the text with valid code only
for(var styletext = '', nomoreimports = NULL_VALUE, i=0; i<cssdata.length; i++)
{
//don't include any fragments that are pure whitespace
//because they'll throw our calculations off, beside being useless
if(trim(cssdata[i]) == '') { continue; }
//if this fragment is an import statement
if(/@import[^;]+;/i.test(cssdata[i]))
{
//if it's the first we've seen, indicate that they're still valid
if(nomoreimports === NULL_VALUE) { nomoreimports = BOOLEAN_FALSE; }
//if they're still valid, add it to the compiled text
if(nomoreimports === BOOLEAN_FALSE) { styletext += cssdata[i]; }
}
//or if this fragment is not an import statement
//indicate that they're no longer valid
//then add it to the compiled text
else
{
nomoreimports = BOOLEAN_TRUE;
styletext += cssdata[i];
}
}
}
//@font-face and @page declarations
styletext = styletext.replace(/@(font\-face|page)[^\}]+\}/igm, '');
//remove any remaining HTML comments or CDATA markers
//(just the markers themselves, not their contents -
// we've already deleted the ones that aren't valid)
//nb. the "CDATA" token is case-sensitive
//"cdata" is invalid and causes an XML parsing exception
styletext = styletext.replace(/(<\!\-\-)|(\-\->)|(<\!\[CDATA\[)|(\]\]>)/gm, '');
//trim and return the result
return trim(styletext);
}
//parse the css text of a style sheet
function parseStyleSheetCSSText(cssdata)
{
//do token replacements for generated content
//to avoid confusion in case it contains any brace characters
var gencontent = cssdata.text.match(/(content:[^;]+;)/igm);
if(gencontent)
{
for(var g=0; g<gencontent.length; g++)
{
cssdata.text = cssdata.text.replace(gencontent[g], '[G' + g + ']');
}
}
//strain and trim (including imports), then split the text by closing-brace characters
cssdata.text = trim(strainStyleSheet(cssdata.text, BOOLEAN_TRUE)).split('}');
//iterate through the resulting lines
//trim and split each one again, by the opening brace character
for(var i=0; i<cssdata.text.length; i++)
{
cssdata.text[i] = trim(cssdata.text[i]).split('{');
for(var j=0; j<cssdata.text[i].length; j++)
{
cssdata.text[i][j] = trim(cssdata.text[i][j]);
}
}
//then iterate again through the resulting matrix
//to convert any generated content tokens back to their real value
if(gencontent)
{
for(var g=0; g<gencontent.length; g++)
{
for(var i=0; i<cssdata.text.length; i++)
{
for(var j=0; j<cssdata.text[i].length; j++)
{
cssdata.text[i][j] = cssdata.text[i][j].replace('[G' + g + ']', gencontent[g]);
}
}
}
}
//count the number of rules we find so that we can
//return that data back up to the debug stylesheets array
var rulecount = 0;
//we identify changes in media by an odd number of members in a group of text lines
//beginning with the default media values from the stylesheet
var currentmedia = cssdata.media,
currentxmedia = cssdata.xmedia;
//we identify changes in owner the same way
var currentowner = cssdata.owner;
//iterate through the text lines
for(var i=0; i<cssdata.text.length; i++)
{
//if we have two members this is just a normal rule pair
if(cssdata.text[i].length == 2)
{
//add this rule data to the cssRules array
THIS._cssRules.push({
'selector' : cssdata.text[i][0],
'css' : cssdata.text[i][1],
'media' : currentmedia,
'xmedia' : currentxmedia,
'owner' : currentowner.toLowerCase(),
'href' : cssdata.href,
'ssid' : cssdata.ssid
});
//add to the rulecount
rulecount++;
}
//if we have three then the first is an opening @media declaration
else if(cssdata.text[i].length == 3)
{
//set the new owner and media
currentmedia = trim(cssdata.text[i][0].replace('@media', '')).toLowerCase();
currentowner = '@media';
//copy to currentxmedia to save the original value
currentxmedia = currentmedia;
//then adjust the media type if necessary to take account of the context
//we only need to check as far as the containing style sheet
//because its media has already been contextualized
var context = cssdata.media;
if(context == '') { context = MEDIA_ALL; }
currentmedia = contextualizeMediaTypes(context, currentmedia);
//add this rule to the cssRules array
THIS._cssRules.push({
'selector' : cssdata.text[i][1],
'css' : cssdata.text[i][2],
'media' : currentmedia,
'xmedia' : currentxmedia,
'owner' : currentowner,
'href' : cssdata.href,
'ssid' : cssdata.ssid
});
//add to the rulecount
rulecount++;
}
//if we have one then it's closing an @media declaration
//or it's an empty member after a stylesheets' final closing brace
//either way, the context should now be the parent stylesheet
else if(cssdata.text[i].length == 1)
{
currentmedia = cssdata.media;
currentxmedia = cssdata.xmedia;
currentowner = cssdata.owner;
}
}
//return the number of rules we found
return rulecount;
}
//-- private methods for building the browser-mode rules data array --//
//create the data array of css style rules from all document stylesheets
function createBrowserRulesData(oncomplete)
{
//get the collection of stylesheets from the specified context document
//and convert it to an array so that we can splice it if necessary
var stylesheets = arrayifize(page.styleSheets);
//iterate through the collection and parse each stylesheet
for(var i=0; i<stylesheets.length; i++)
{
//assign an index to this stylesheet
stylesheets[i].__ssid = THIS.ssidcounter++;
//if the stylesheets has an ownerNode property then this is a standard implementation
//which we can parse using DOM 2 CSS properties and collections
if(typeof stylesheets[i].ownerNode != TYPE_UNDEFINED)
{
//if this is safari 3 we cannot process xml-stylesheet,
//because it doesn't support the pseudo-attributes we need
if(SAFARI3 && /xml\-stylesheet/.test(stylesheets[i].ownerNode.nodeName))
{
//add error data to the debug stylesheets array
THIS._stylesheets.push({
'ssid' : stylesheets[i].__ssid,
'href' : qualifyHREF(stylesheets[i].href, base),
'owner' : 'xml-stylesheet',
'media' : MESSAGE_UNKNOWN,
'stylesheet' : NULL_VALUE,
'rules' : 0,
'message' : MESSAGE_UNSUPPORTED_TYPE
});
//remove it from this array
stylesheets.splice(i, 1);
i --;
}
//otherwise try to parse the stylesheet as normal
else
{
try
{
parseStyleSheet(
stylesheets[i],
stylesheets[i].ownerNode.nodeName.toLowerCase(),
//the trim probably isn't necessary, but better safe than sorry!
stylesheets[i].media.length > 0 ? trim(stylesheets[i].media.mediaText) : MEDIA_ALL
);
}
//if we fail just delete this stylesheet and carry on
//the only known occurence is this exception is when
//using the script within the chrome scope in Firefox
//where a couple of extension stylesheets turned out
//to have a null ownerNode, for reasons unknown
catch(err)
{
//add error data to the debug stylesheets array
//since this is so unlikely to happen, i'm being stingy
//with code and just making all the data "unknown"
//of course this will make it harder for users
//to track down the offending stylesheet,
//but hey - that's what trial and error is for :-P
//nb. the reason that the message is "unspecified error"
//is that I don't actually know what causes it
//so "unspecified" is the most I can tell you!
THIS._stylesheets.push({
'ssid' : stylesheets[i].__ssid,
'href' : MESSAGE_UNKNOWN,
'owner' : MESSAGE_UNKNOWN,
'media' : MESSAGE_UNKNOWN,
'stylesheet' : NULL_VALUE,
'rules' : 0,
'message' : ERROR_UNSPECIFIED
});
//remove it from this array
stylesheets.splice(i, 1);
i --;
}
}
}
//or if it has an owningElement property then this is IE's implementation
//which we have to parse using IE's proprietary nastiness
//well, I say "nastiness", it's not really all that bad,
//and at least it doesn't do all the aggressive value normalization that firefox does
//it's just the way it normalizes all selectors and property names to uppercase
//which is annoying and ugly and pisses me off
//it's bad enough that it does that for HTML tag names,
//but at least it has the excuse of that being the canonical form
else if(typeof stylesheets[i].owningElement != TYPE_UNDEFINED)
{
parseIEStyleSheet(
stylesheets[i],
stylesheets[i].owningElement.nodeName.toLowerCase(),
//the trim probably isn't necessary, but better safe than sorry!
trim(stylesheets[i].media) != '' ? trim(stylesheets[i].media) : MEDIA_ALL
);
}
}
//do a final sort of the _stylesheets array, putting it in ssid order
//this is for IE's benefit which generates the ssids in occurence order
//but builds the array in rule order; so this sort will bring it
//into line with other browsers so they all output the same occurence order
THIS._stylesheets.sort(function(a, b) { return a.ssid - b.ssid; });
//nullify and delete the ssidcounter
THIS.ssidcounter = undefined;
// delete THIS.ssidcounter;
//if the watch setting is true, watch for changes in disabled state
if(watch == BOOLEAN_TRUE) { watchStyleSheets(); }
//call the oncomplete callback, if defined
if(typeof oncomplete == TYPE_FUNCTION)
{
oncomplete();
}
}
//parse an individual stylesheet using DOM 2 CSS
function parseStyleSheet(sheet, owner, media)
{
//check if we already have this stylesheet data
//which we can only do for stylesheets that have an href
//and will only be caused by duplicate stylesheet includes
//we also have to check that the href is not the same as base
//which can happen in firefox 2 for <style> elements, rather than href being null
if(sheet.href && sheet.href != base)
{
var testhref = qualifyHREF(sheet.href, base);
if(arrayContains(THIS._stylesheets, testhref, 'href') != NULL_VALUE)
{
//add duplication data to the debug stylesheets array
THIS._stylesheets.push({
'ssid' : sheet.__ssid,
'href' : testhref,
'owner' : owner,
'media' : media,
'stylesheet' : NULL_VALUE,
'rules' : 0,
'message' : MESSAGE_DUPLICATE
});
//however if there are duplicate stylesheets,
//subsequently disabling one of them (ie. through stylesheet switching)
//will cause the other one's rules to take effect, which the script
//won't be aware of because its rules are not included
//so since we don't have the option of just including duplicates
//because of the possibility of infinite recursion,
//what we'll do instead is just forcibly disable any duplicates we find
//(although infinite recursion will be prevented by the browser in this case,
//it can still happen in IE because of the way we get the data
//and so we have to do it here so that the data set is x-browser consistent)
sheet.disabled = BOOLEAN_TRUE;
//and we're done here
return;
}
}
//try to get the cssRules collection from this stylesheet
//and pass it along with the other input references
//to the standard rules collection parsing method
try
{
//is this stylesheet disabled
var isdisabled = sheet.disabled;
//in opera, the stylesheet returns as disabled if it applies to media
//other than the one currently in force, such as "print" when the view is "screen"
//so what we'll do is temporarily switch the media to whatever the current view is
//then read its disabled property again, to find out if it's
//really disabled or just applicable to a non-view media
//this happens so fast that no visual change should be apparent
if(OPERA && isdisabled && sheet.media.length > 0)
{
var realmedia = sheet.media.mediaText;
sheet.media.mediaText = THIS._viewmedia;
isdisabled = sheet.disabled;
sheet.media.mediaText = realmedia;
}
//if the stylesheet is not disabled
if(!isdisabled)
{
//count the number of ordinary rules
for(var rules=sheet.cssRules, rulecount=0, i=0; i<rules.length; i++)
{
if(rules.item(i).type == 1)
{
rulecount++;
}
else if(rules.item(i).type == 4)
{
for(var subrules=rules.item(i).cssRules, j=0; j<subrules.length; j++)
{
//no need to recur further than this,
//because nested @media are not allowed
if(subrules.item(j).type == 1) { rulecount++; }
}
}
}
//set the ok message for the debug stylesheet array
var message = MESSAGE_OK;
}
//if it is disabled, set the rulecount to zero
//and the message to "stylesheet is disabled"
else
{
rulecount = 0;
message = MESSAGE_DISABLED;
}
//add data to the debug stylesheets array
//checking that the href is not the same as base
//which can happen in firefox 2 for <style> elements,
//rather than href being null (which is what we want to record)
THIS._stylesheets.push({
'ssid' : sheet.__ssid,
'href' : (sheet.href && sheet.href != base)
? qualifyHREF(sheet.href, base)
: NULL_VALUE,
'owner' : owner,
'media' : media,
'stylesheet' : sheet,
'rules' : rulecount,
'message' : message
});
//if it's not disabled, parse the stylesheet
if(!isdisabled)
{
parseCSSRules(sheet, sheet.cssRules, owner, media);
}
}
//it will only fail because of a security error
//if the stylesheet is on a different domain
catch(err)
{
//add data to the debug stylesheets array
//checking that the href is not the same as base
//which can happen in firefox 2 for <style> elements,
//rather than href being null (which is what we want to record)
THIS._stylesheets.push({
'ssid' : sheet.__ssid,
'href' : (sheet.href && sheet.href != base)
? qualifyHREF(sheet.href, base)
: NULL_VALUE,
'owner' : owner,
'media' : media,
'stylesheet' : NULL_VALUE,
'rules' : 0,
'message' : ERROR_SECURITY
});
}
}
//parse a collection of css rules using DOM 2 CSS
//this creates an array that's in order of occurence, where rules in
//import stylesheets that are declared at the top occur before
//rules inside the stylesheet, and rules in @media block occur where they occur
function parseCSSRules(sheet, rules, owner, media)
{
//iterate through the rules collection
//we're actually going to iterate separately three times
//for @imports, rules, and media blocks
//and do them separately so that we can control the order of execution
//if don't do that then we'll may end up passing the wrong arguments back through
//eg. an import will call it before the rules within a stylesheet
//and we'll end up recording those rules with the import's media
//we have to do the imports first so that the overall rule indices remain accurate
//to their source order, which may be crucial when it comes to sorting
for(i=0; i<rules.length; i++)
{
//save a shortcut reference
rule = rules.item(i);
//if it's an @import
if(rule.type == 3)
{
//get the media from the rule itself
//or if that's empty get it from the stylesheet's parent stylesheet
//or if that's empty set it to default
media = rule.media.mediaText;
if(media == '')
{
//this exception handling is for safari, where if the parent
//stylesheet is an @import then it doesn't have a media object
//the media object belongs to its ownerRule object instead
try { media = rule.parentStyleSheet.media.mediaText; }
catch(err) { media = rule.parentStyleSheet.ownerRule.media.mediaText; }
}
if(media == '') { media = MEDIA_ALL; }
//convert the value to lower case
media = media.toLowerCase();
//then adjust the media type if necessary to take account of the context
//and so on for as many parent contexts as there are
var parent = rule.parentStyleSheet;
while(parent)
{
//this exception handling is for safari, where if the parent
//stylesheet is an @import then it doesn't have a media object
//the media object belongs to its ownerRule object instead
try { var context = parent.media.mediaText; }
catch(err) { context = parent.ownerRule.media.mediaText; }
if(context == '') { context = MEDIA_ALL; }
media = contextualizeMediaTypes(context, media);
//this condition is to catch a difference in parent stylesheet reference
//when the stylesheet object is an @import in safari
if((SAFARI || KONQUEROR || WEBKIT) && !parent.parentStyleSheet && parent.ownerRule)
{
parent = parent.ownerRule.parentStyleSheet;
}
else
{
parent = parent.parentStyleSheet;
}
}
//get the nested stylesheet reference from this rule
//assign it an ssid, and then pass it back to the parseStyleSheet method
//along with an owner name and media to save with the rule
rule.styleSheet.__ssid = THIS.ssidcounter++;
parseStyleSheet(rule.styleSheet, '@import', media);
}
}
//now iterate again and look for normal rules
for(var i=0; i<rules.length; i++)
{
//save a shortcut reference
var rule = rules.item(i);
//if this is a regular style rule
if(rule.type == 1)
{
//if owner is "link", "style" or "xml-stylesheet" then this is a normal rule
//(as opposed to being inside a @media or @import for which media was passed in as an argument)
//so get the media from the rule's parent stylesheet
//or if that's empty set it to default
//nb. we're splitting the media-getting routines into the rule-type conditions
//rather than doing them all at once in this condition
//because safari returns null for rule.parentRule
if(/^(xml\-stylesheet|link|(([a-z]+:)?style))$/i.test(owner))
{
media = rule.parentStyleSheet.media.mediaText;
if(media == '') { media = MEDIA_ALL; }
//convert the value to lower case
media = media.toLowerCase();
}
//we may have to parse the style.cssText to add !important to any !important rules
//because in Opera (maybe others) the "!important" token itself is not included in the cssText
//however we can get it from the getPropertyPriority method for each individual property
//though it is quite a lot of farting about, it has to be done
//we can't do it any later than now because we won't have the cssRule reference any more;
//so ... begin by getting the cssText and stripping any line breaks
var styletext = rule.style.cssText.replace(/[\r\n]/g, ' ');
//split the cssText by semi-colons
styletext = styletext.split(';');
//then iterate through the results and splt each one by its first colon
for(var c=0; c<styletext.length; c++)
{
//this convoluted funkiness is because IE failed to return
//any matches when a first-colon regex was used, so instead we split
//by all colons and then join together any excess matches
//even though IE doesn't actually run this particular bit of code anyway,
//it was simpler just to use the same code pattern everywhere we need to do this
styletext[c] = styletext[c].split(':');
while(styletext[c].length > 2) { styletext[c][1] += ':' + styletext[c].pop(); }
//if we only have one match, the most likely reason
//is that the semi-colon split operation latched onto
//a semi-colon that wasn't a property delimeter
//probably it was inside generated content
//and hence the value we're splitting doesn't contain a colon at all
//so if that happens we'll just have to remove it and continue
//it would happen sooner or later anyway for such a rule, so it's no ultimate loss
if(styletext[c].length == 1)
{
styletext.splice(c, 1);
c--;
continue;
}
//add !important to the property definition if applicable
//and if it doesn't already have that flag
//(we need to check for an end-substring so that we don't get confused
// by a content property with "!important" in its text
// (which is possible, however improbable!))
if(rule.style.getPropertyPriority(trim(styletext[c][0])) == 'important'
&& !IMPORTANT_RULE.test(styletext[c][1]))
{
styletext[c][1] += ' !important';
}
//join it back together
styletext[c] = styletext[c].join(':');
}
//and join the whole thing back together
styletext = styletext.join(';');
//add this rule object to the array
//qualifying the href if it's not null
THIS._cssRules.push({
'selector' : rule.selectorText,
'css' : styletext,
'media' : media,
'owner' : owner,
'href' : sheet.href == NULL_VALUE ? NULL_VALUE : qualifyHREF(sheet.href, base),
'ssid' : sheet.__ssid
});
}
}
//now iterate again and look for @media blocks
for(i=0; i<rules.length; i++)
{
//save a shortcut reference
rule = rules.item(i);
//if it's an @media
if(rule.type == 4)
{
//get the media from the rule itself
//or if that empty get it from the parent stylesheet
// which is possible - Firefox and IE support "@media { ... }"
// treating it is "@media all"; however Opera and Safari reject
// such rules, not including them in the cssRules collection at all
//or if that's empty set it to default
media = rule.media.mediaText;
if(media == '') { media = rule.parentStyleSheet.media.mediaText; }
if(media == '') { media = MEDIA_ALL; }
//convert the value to lower case
media = media.toLowerCase();
//then adjust the media type if necessary to take account of the context
//and so on for as many parent contexts as there are
var parent = rule.parentStyleSheet;
while(parent)
{
//this exception handling is for safari, where if the parent
//stylesheet is an @import then it doesn't have a media object
//the media object belongs to its ownerRule object instead
try { var context = parent.media.mediaText; }
catch(err) { context = parent.ownerRule.media.mediaText; }
if(context == '') { context = MEDIA_ALL; }
media = contextualizeMediaTypes(context, media);
//this condition is to catch a difference in parent stylesheet reference
//when the stylesheet object is an @import in safari
if((SAFARI || KONQUEROR || WEBKIT) && !parent.parentStyleSheet && parent.ownerRule)
{
parent = parent.ownerRule.parentStyleSheet;
}
else
{
parent = parent.parentStyleSheet;
}
}
//get the nested rules collection from this rule
//and then pass it back to the parseCSSRules method
//along with this stylesheet reference
//and an owner name and media to save with the rule
parseCSSRules(sheet, rule.cssRules, '@media', media);
}
}
//for any other kind of rule we're not interested
}
//parse an individual stylesheet for IE
function parseIEStyleSheet(sheet, owner, media, href)
{
//check if we already have this stylesheet data
//which we can only do for stylesheets that have an href
//and will only be caused by duplication stylesheet includes
//this is particularly necessary because it prevents infinite recursion
//for example, where two stylesheets import each other
if(sheet.href)
{
var testhref = qualifyHREF(sheet.href, base);
if(arrayContains(THIS._stylesheets, testhref, 'href') != NULL_VALUE)
{
//add duplication data to the debug stylesheets array
THIS._stylesheets.push({
'ssid' : sheet.__ssid,
'href' : testhref,
'owner' : owner,
'media' : media,
'stylesheet' : NULL_VALUE,
'rules' : 0,
'message' : MESSAGE_DUPLICATE
});
//however if there are duplicate stylesheets,
//subsequently disabling one of them (ie. through stylesheet switching)
//will cause the other one's rules to take effect, which the script
//won't be aware of because its rules are not included
//so since we don't have the option of just including duplicates
//because of the possibility of infinite recursion,
//what we'll do instead is just forcibly disable any duplicates we find
sheet.disabled = BOOLEAN_TRUE;
//and we're done here
return;
}
}
//create a temporary array for storing any imports we find
//that we'll then pass back to this method at the end
//we could just pass them back as soon as we find them
//but storing them and doing it at the end ensures that
//the original order is preserved for the debug stylesheets array
var importsheets = [];
//is this stylesheet disabled
var isdisabled = sheet.disabled;
//if the stylesheet is not disabled
if(!isdisabled)
{
//first of all go through the imports collection
for(var i=0; i<sheet.imports.length; i++)
{
//IE doesn't support @import with a media type
//but they do still show up in the imports collection
//even though they're not effective
//so we have to check for this by testing the
//length of the import's rules collection
//and if its zero just don't include the stylesheet
if(sheet.imports[i].rules.length == 0) { continue; }
//get the media from the owner stylesheet
//or if that's empty set it to default
media = trim(sheet.media);
if(media == '') { media = MEDIA_ALL; }
//get and qualify the href from the import
//nb. we need to use a different variable name here
//so that we don't affect the input href argument
var ihref = qualifyHREF(sheet.imports[i].href, base);
//add the impored stylesheet to the imports array
//along with an owner reference, the effective media, and its href
importsheets.push({
'sheet' : sheet.imports[i],
'owner' : '@import',
'media' : media,
'href' : ihref
});
}
}
//if we have any imports, assign each one an ssid,
//and then pass it back to this method
//we do this before checking the other rules in this stylesheet
//so that the overall rule indices accurately reflect source order
if(importsheets.length > 0)
{
for(var i=0; i<importsheets.length; i++)
{
importsheets[i].sheet.__ssid = THIS.ssidcounter++;
parseIEStyleSheet(
importsheets[i].sheet,
importsheets[i].owner,
importsheets[i].media,
importsheets[i].href
);
}
}
//now we have to make a manul rules collection from the stylesheet's cssText
//otherwise we won't be able to get info about @media blocks
//because they're just not included in the rules collection at all
try
{
//if the stylesheet is not disabled
if(!isdisabled)
{
//first copy the text
var styletext = sheet.cssText;
//remove any @import, @charset and @namespace statements
styletext = styletext.replace(/@(import|charset|namespace)[^;]+;/igm, '');
//remove any @font-face and @page declarations
styletext = styletext.replace(/@(font\-face|page)[^\}]+\}/igm, '');
//do token replacements for generated content
//to avoid confusion in case it contains any brace characters
//this regex is different from the one we used in author mode
//because IE6 normalizes generated content to have no quotes around it
//and all versions put their expressions at the end of the group
//but since we still have line breaks at this point,
//we can use the end of line marker to delimit the match expression
//however we don't have line-breaks within a block of CSS
//so just in case the ordering is not consistent
//we'll make semi-colon an alternative delimiter
var gencontent = styletext.match(/(content\s*:\s*[\'\"].*(;|$))/igm);
if(gencontent)
{
for(var g=0; g<gencontent.length; g++)
{
styletext = styletext.replace(gencontent[g], '[G' + g + ']');
}
}
//then trim and split the result by closing-brace characters
styletext = trim(styletext).split('}');
//iterate through the resulting lines
//trim and split each one again, by the opening brace character
for(var i=0; i<styletext.length; i++)
{
styletext[i] = trim(styletext[i]).split('{');
for(var j=0; j<styletext[i].length; j++)
{
styletext[i][j] = trim(styletext[i][j]);
}
}
//then iterate again through the resulting matrix
//to convert any generated content tokens back to their real value
if(gencontent)
{
for(var g=0; g<gencontent.length; g++)
{
for(var i=0; i<styletext.length; i++)
{
for(var j=0; j<styletext[i].length; j++)
{
styletext[i][j] = styletext[i][j].replace('[G' + g + ']', gencontent[g]);
}
}
}
}
//if href is undefined, get and qualify the href from the stylesheet,
//unless it's null or empty, in which case its null
//nb. we only really need to test empty,
//but I feel safer testing null as well
if(typeof href == TYPE_UNDEFINED)
{
href = (sheet.href == NULL_VALUE || sheet.href == '')
? NULL_VALUE
: qualifyHREF(sheet.href, base);
}
//now build a rules collection
var rules = [];
//we identify changes in media by an odd number of members in a group of text lines
//beginning with the default media value from the stylesheet
//if the value is empty then set it to default
var currentmedia = sheet.media;
if(currentmedia == '') { currentmedia = MEDIA_ALL; }
//we identify changes in owner the same way
var currentowner = owner;
//iterate through the text lines
for(var i=0; i<styletext.length; i++)
{
//if we have two members this is just a normal rule pair
if(styletext[i].length == 2)
{
rules.push({
'selector' : styletext[i][0],
'css' : styletext[i][1],
'media' : currentmedia,
'owner' : currentowner,
'href' : href,
'ssid' : sheet.__ssid
});
}
//if we have three then the first is an opening @media declaration
//which we can edit to get the actual media value
else if(styletext[i].length == 3)
{
currentmedia = trim(styletext[i][0].replace('@media', '')).toLowerCase();
currentowner = '@media';
//adjust the media type if necessary to take account of the context
var context = sheet.media;
if(context == '') { context = MEDIA_ALL; }
currentmedia = contextualizeMediaTypes(context, currentmedia);
rules.push({
'selector' : styletext[i][1],
'css' : styletext[i][2],
'media' : currentmedia,
'owner' : currentowner,
'href' : href,
'ssid' : sheet.__ssid
});
}
//if we have one then it's closing an @media declaration
else if(styletext[i].length == 1)
{
currentmedia = sheet.media;
if(currentmedia == '') { currentmedia = MEDIA_ALL; }
currentowner = owner;
}
}
//set the message for the debug array to "ok"
var message = MESSAGE_OK;
}
//if it is disabled, set the message to "stylesheet is disabled"
else
{
message = MESSAGE_DISABLED;
}
//if the stylesheet is not disabled
//pass this custom rules collection to parseIECSSRules
//which returns the final rule count
if(!isdisabled)
{
var rulecount = parseIECSSRules(rules);
}
//add data to the debug stylesheets array
THIS._stylesheets.push({
'ssid' : sheet.__ssid,
'href' : sheet.href
? qualifyHREF(sheet.href, base)
: NULL_VALUE,
'owner' : owner,
'media' : media,
'stylesheet' : sheet,
'rules' : !isdisabled ? rulecount : 0,
'message' : message
});
}
//it will only fail because of a security error
//if the stylesheet is on a different domain
catch(err)
{
//add data to the debug stylesheets array
THIS._stylesheets.push({
'ssid' : sheet.__ssid,
'href' : sheet.href
? qualifyHREF(sheet.href, base)
: NULL_VALUE,
'owner' : owner,
'media' : media,
'stylesheet' : NULL_VALUE,
'rules' : 0,
'message' : ERROR_SECURITY
});
}
}
//parse a collection of css rules using IE proprietary nastiness
//this creates an array in order of occurence, where rules in
//import stylesheets that are declared at the top occur before
//rules inside the stylesheet, and rules in media block occur where they occur
function parseIECSSRules(rules)
{
//store the number of rules that actually get added
//we used to do this by counting the pre-cursor "rules" array
//but not all of them end up in the final cssRules array,
//which this count only includes
var rulecount = 0;
//iterate through the rules collection
for(var i=0; i<rules.length; i++)
{
//if the selector is empty, unknown, or contains an unknown pseudo-node,
//just ignore it and continue on to the next rule
//(IE has a cutely-obvious way of indicating unknown syntax in css :))
if(rules[i].selector == ''
|| rules[i].selector == 'UNKNOWN'
|| rules[i].selector.indexOf(':unknown') != -1) { continue; }
//add this rule object to the array, and along the way
//parse the selectorText string to lowercase all the element type selectors
//and parse the cssText string to lowercase all the property names
THIS._cssRules.push({
'selector' : lowercaseIETypeSelectors(rules[i].selector),
'css' : lowercaseIEPropertyNames(rules[i].css),
'media' : rules[i].media,
'owner' : rules[i].owner,
'href' : rules[i].href,
'ssid' : rules[i].ssid
});
//increase the rule count
rulecount ++;
}
//return the final rule count
return rulecount;
}
//-- other private css-related methods --/
//getCSSRules return all the rule objects that apply to an element
//each of these is a custom object compiled with the most useful information
//rather than actually being a cssRule object, so that the results are consistent cross browser
function getCSSRules(element, media, accept, altstates, rules, inherited, ancestors)
{
//no processing of the first two arguments happens here
//we require them to have been pre-processed before calling this
//process the accept argument (if necessary)
//it will never be undefined at this point, but it may still be a string
//(when called by most public methods) or already an object
//(when called from a recursion of this method, or from its public equivalent)
accept = processAcceptArgument(accept);
//if rules is undefined or null, define a new array
if(typeof rules == TYPE_UNDEFINED || rules == NULL_VALUE) { rules = []; }
//if inherited is undefined or null, set it to false
if(typeof inherited == TYPE_UNDEFINED || inherited == NULL_VALUE) { inherited = BOOLEAN_FALSE; }
//if ancestors is undefined or null, create an array beginning with this element
//we'll maintain an array of all the elements we check
//which we'll need when it comes to determine the selector specificity
//so that we can reject those that don't apply
if(typeof ancestors == TYPE_UNDEFINED || ancestors == NULL_VALUE)
{
ancestors = [element];
//if the element has a style attribute, and the "attributes" setting is true,
//add a temporary member to the end of the cssRules array containing its data
//this means its index value will place it as the very last rule to be defined
//which in source-order specificity terms is exactly right
//however in IE getAttribute('style') returns an object not a string
//which we can deal with in the createStyleAttributeRule method
//but it means that it will always pass this test, even when
//the element has no style attribute; if that happens, or if the
//style attribute is empty, the create method will return null
//so we save the return value and test it before adding it to the array
if(attributes == BOOLEAN_TRUE && element.getAttribute('style'))
{
var stylerule = createStyleAttributeRule(element);
if(stylerule != NULL_VALUE)
{
THIS._cssRules.push(stylerule);
}
}
}
//otherwise add this element to the existing array
else
{
ancestors.push(element);
}
//now iterate through all the stored cssRules
for(var i=0; i<THIS._cssRules.length; i++)
{
//if the rule doesn't match the input media criteria, we don't want it
if(!mediaMatches(media, THIS._cssRules[i])) { continue; }
//if this is a temporary style attribute rule
//and the element argument is the original input element reference
//create a nodes array containing just the input element
//so that it passes straight through as a single found match
//and set the altstate variable for it to false
//because style attributes apply to all states
//this will make sure that its properties are correctly sorted
if(THIS._cssRules[i].owner == '@style' && element == ancestors[0])
{
var nodes = [element],
altstate = BOOLEAN_FALSE;
}
//otherwise, see if this selector matches any elements
else
{
//if the altstates argument is true, do a pre-search for this element
//with the original selector, not edited for pseudo-classes
//so we can compare to know whether it matched without editing, or only with editing
if(altstates == BOOLEAN_TRUE)
{
for(var altstate = BOOLEAN_TRUE,
allnodes = getElementsBySelector(THIS._cssRules[i].selector),
j=0; j<allnodes.length; j++)
{
if(allnodes[j] == element)
{
altstate = BOOLEAN_FALSE;
break;
}
}
}
//or if altstates is false then all matches will be by default, not alt states
else
{
altstate = BOOLEAN_FALSE;
}
//now do the real search, and if the altstates argument is true,
//remove pseudo-classes except not() from the test selector;
//otherwise pass it in unedited
nodes = getElementsBySelector(
altstates == BOOLEAN_TRUE
? THIS._cssRules[i].selector.replace(REGEX_PSEUDO_CLASSES_EXCEPT_NOT, '')
: THIS._cssRules[i].selector
);
}
//if we have matches, search them looking for the input element reference
if(nodes.length > 0)
{
for(var j=0; j<nodes.length; j++)
{
//and if we find it add the specified rule to the rules array
if(nodes[j] == element)
{
//create an inheritance path array, which must be a copy not a reference
//otherwise when we come to copy it into the output,
//it will be the same (its final state) every time
//we'll also do it in reverse so that
//we get an array that's in top-down DOM order
var inheritance = [];
if(inherited)
{
for(var a=ancestors.length-1; a>0; a--)
{
inheritance.push(ancestors[a]);
}
}
//create a rule object to add to the array, starting with
//the selectors, css, index in the rule array,
//or altstates is true and the selector only applies to an altstate
//a placeholder for its specificity (which we'll work out
//once we've got all the rules), its current inheritance path
//and a flag to indicate whether it actually applies to the element in its current state
//all of which we need irrespective of whether they're included
//in the accept dictionary; so for any that are not,
//we'll delete them at the end once all the other processing is done
var ruleobj = {
'selector' : THIS._cssRules[i].selector,
'css' : THIS._cssRules[i].css,
'index' : i,
'specificity' : [0,0,0,0],
'inheritance' : inheritance,
'altstate' : altstate
};
//then add the properties which are optional and only included
//if they're listed in the accept dictionary, or it's "*";
//apart from "properties" which we deal with later
for(var optionalprops = ['media','xmedia','owner','ssid','href'],
k=0; k<optionalprops.length; k++)
{
if(accept == '*' || typeof accept[optionalprops[k]] != TYPE_UNDEFINED)
{
//we need this condition because the "xmedia"
//property is only available in author mode
if(typeof THIS._cssRules[i][optionalprops[k]] != TYPE_UNDEFINED)
{
ruleobj[optionalprops[k]] = THIS._cssRules[i][optionalprops[k]];
}
}
}
//add this rule object to the rules array
rules.push(ruleobj);
//we'll only find it once!
break;
}
}
}
}
//if this element has a parent element, recur on that
//so that we get inherited rules as well
if(element.parentNode && element.parentNode.nodeType == 1)
{
return getCSSRules(element.parentNode, media, accept, altstates, rules, BOOLEAN_TRUE, ancestors);
}
//otherwise we can proceed to process the finished rules array
else
{
//if we have no rules we can just return the empty array and we're done
//as well as saving some process, this also means that we can save adding
//a condition to the bit at the bottom which removes any added @style rule
//which we'd otherwise have to check against (THIS._cssRules.length > 0)
if(rules.length == 0) { return rules; }
//first we have to sort it so it's in primary order of specificity
//with a secondary order by source index, and a tertiary order by inheritance chain depth
//so begin by iterating through the rules to get their specificities
for(var i=0; i<rules.length; i++)
{
//extract the selector from this rule and split into individuals
var selectors = rules[i].selector.split(',');
//now we need to pass each selector to the getSelectorSpecificity method
//(unless it's inherited in which case it has zero specificity)
//so we get an array of all the specificities that apply to the overall selector
for(var specs=[], j=0; j<selectors.length; j++)
{
specs.push(
rules[i].inheritance.length > 0
? [0,0,0,0]
: getSelectorSpecificity(selectors[j])
);
}
//now inverse sort the resulting array, so that the largest is first
specs.sort(function(a, b)
{
if(a[0] !== b[0]) { return b[0] - a[0]; }
if(a[1] !== b[1]) { return b[1] - a[1]; }
if(a[2] !== b[2]) { return b[2] - a[2]; }
return b[3] - a[3];
});
//then save the highest value, because obviously that's the one
//that determines the specificity of the rule's properties to this element
rules[i].specificity = specs[0];
}
//now we have the specificity data we can sort the rules array by it
rules.sort(function(a, b)
{
//if the specificity values are the same
if(a.specificity.toString() === b.specificity.toString())
{
//if the source index values are also the same,
//sort on inverse inheritance depth
//so that the highest elements come before the deepest elements
if(a.index === b.index)
{
return b.inheritance.length - a.inheritance.length;
}
//otherwise sort on source index
return a.index - b.index;
}
//otherwise sort on specificity
if(a.specificity[0] !== b.specificity[0])
{
return a.specificity[0] - b.specificity[0];
}
if(a.specificity[1] !== b.specificity[1])
{
return a.specificity[1] - b.specificity[1];
}
if(a.specificity[2] !== b.specificity[2])
{
return a.specificity[2] - b.specificity[2];
}
return a.specificity[3] - b.specificity[3];
});
//if "properties" is included in the accept dictionary, or its "*"
//we need to add and then sort the individual properties for the rules
//"sort" in this case means to organise the data so that each member
//is a further object with value and status properties, rather than just a value
//ie. each has the form {"color":{"value":"red","status":"active"} rather than just {"color":"red"}
//and the "status" value can be "active", "cancelled" or "inactive" (for altstate rules)
//this replaces what used to be separate properties,xproperties,nproperties objects
//the second argument instructs the method to do that sort,
//which if it's false will only return an unsorted properties object,
//which setting is used by the getCSSStyleSheetRules method
if(accept === "*" || typeof accept.properties != TYPE_UNDEFINED)
{
rules = addSortedProperties(rules, BOOLEAN_TRUE);
}
//if accept is not "*", then for each of the options
//"selector","css","index","specificity", "inheritance" and "altstate",
//if the accept dictionary didn't include it, delete it from each rule object
//first nullifying it to ensure garbage collection in IE
if(accept !== "*")
{
for(var i=0; i<rules.length; i++)
{
for(var props=['selector','css','index','specificity','inheritance','altstate'],
j=0; j<props.length; j++)
{
if(typeof accept[props[j]] == TYPE_UNDEFINED)
{
rules[i][props[j]] = undefined;
// rules[i][props[j]] = NULL_VALUE;
// delete rules[i][props[j]];
}
}
}
}
//if we added a temporary rule object
//for the element's style attribute, remove it again
if(THIS._cssRules[THIS._cssRules.length - 1].owner == '@style')
{
THIS._cssRules.splice(THIS._cssRules.length - 1, 1);
}
//and return the final rules array
return rules;
}
}
//create a rule object for the cssRules array
//that represents an element's style attribute
function createStyleAttributeRule(element)
{
//if this is internet explorer we can't get the css text from the style attribute
//because the style attributes return a style object, not a string,
//even when you use the incorrectly-documented second argument
//[see: http://reference.sitepoint.com/javascript/Element/getAttribute#compatibilitysection]
//but we can extract it from the element's outerHTML using match() (lol)
//and then pass it to the lowercaseIEPropertyNames method to normalize the case
if(IEXPLORER)
{
//just in case this is XML and not HTML, we'll wrap it in try..catch
//though it's very unlikely that if this is XML it will have a style attribute
//I suppose it might be SVG, but then IE doens't support SVG. VML maybe?
//whatever .. it's as well to be safe rather than allow a possible exception
try
{
//here we can rely on the fact that IE normalizes the outerHTML
//such that attributes are always a lowercase name and double-quoted
//and any quotes inside values like url() are removed
//remembering to split by the first closing angle bracket, so that we only get
//the html of the element's opening tag, not everything inside it as well!
//if the element doesn't have a style attribute the match will be null
//so in that case set styletext to an empty string
var matches = element.outerHTML.split('>')[0].match(/.*style\s*=\s*\"([^\"]*)\".*/im);
var styletext = matches ? lowercaseIEPropertyNames(matches[1]) : '';
}
//in the unlikely event it fails for the reason noted, we just return an empty string
catch(err) { styletext = ''; }
}
//for any other browser we can just grab it from the style attribute as text
else
{
styletext = element.getAttribute('style');
}
//if the styletext is empty, return null
if(styletext == '')
{
return NULL_VALUE;
}
//else create a rule object
var ruleobj =
{
//the empty string selector signifies a style attribute, for user reference,
//and particularly so that the the getSelectorSpecificity method can identify it
'selector': '',
'css' : styletext,
'media': MEDIA_ALL,
'owner' : '@style',
'href' : NULL_VALUE,
//the infinity value means that in a numerical sort it will always come out at the end
//though tbh that isn't something I planned, it just worked out nicely that way :)
//it had to be a number for consistency with other values, but any number I choose
//might conflict with an actual stylesheet, even a really high number
//an alternative I considered was to use -1, which would be safe from conflict,
//however that value already has a special meaning - it's a valid argument to
//getCSSStyleSheetRules, which means "no specific stylesheet"
'ssid' : Infinity
};
//add xmedia only in author mode
if(mode == MODE_AUTHOR)
{
ruleobj.xmedia = ruleobj.media;
}
//return the finished rule object
return ruleobj;
}
//extract the individual properties for a set of rules from each rule's css text
//then sort them according to which ones are precedent and which cancelled out
//to add status flags - "active","cancelled" or "inactive" - to each property
//this methods assumes that the rules set is already in order of specificity
//and that each one has information about its inheritance; as created by getCSSRules
function addSortedProperties(rules, dosort)
{
//run through the rules
for(var i=0; i<rules.length; i++)
{
//parse the css text to create a properties object
rules[i].properties = parsePropertyText(
rules[i].css,
{},
(typeof rules[i].inheritance != TYPE_UNDEFINED
&& rules[i].inheritance.length > 0)
);
//create a re-organised copy in which each property
//is now a sub-object with "value" and "status" properties
//then mark each status as "active" by default
rules[i].allproperties = {};
for(var j in rules[i].properties)
{
rules[i].allproperties[j] = {
'value' : rules[i].properties[j].property,
'status' : STATUS_ACTIVE
};
}
//if the dosort argument is true
if(dosort)
{
//if the altstate flag for this rule is true
//change all the status flags to "inactive"
if(rules[i].altstate == BOOLEAN_TRUE)
{
for(var ip in rules[i].allproperties)
{
if(!rules[i].allproperties.hasOwnProperty(ip)) { continue; }
rules[i].allproperties[ip].status = STATUS_INACTIVE;
}
}
//otherwise iterate from the start of the current collection up to now
//in order to find and mark the cancelled properties
else
{
for(var j=0; j<i; j++)
{
//for each of the properties in the earlier object
for(var jp in rules[j].allproperties)
{
if(!rules[j].allproperties.hasOwnProperty(jp)
|| rules[j].allproperties[jp].status != STATUS_ACTIVE) { continue; }
//for each of the properties in this object
for(var ip in rules[i].allproperties)
{
if(!rules[i].allproperties.hasOwnProperty(ip)
|| rules[i].allproperties[ip].status != STATUS_ACTIVE) { continue; }
//if we encounter this current properties in the earlier
//properties objects, change its status to "cancelled"
//we do this for both inheritable and non-inheritable properties;
//or rather: we do it for inheritable properties, or non-inheritable
//properties where the previous rule is not inherited;
//but we don't need to do that additional testing
//because only non-inherited properties will be included
//in the output of inherited rules
if(jp == ip)
{
rules[j].allproperties[jp].status = STATUS_CANCELLED;
//there will only be one
break;
}
}
}
}
}
}
}
//if the dosort argument is true
if(dosort)
{
//now we need to run through the cancelled properties
//to deal any with !important rules, so...
for(var i=0; i<rules.length; i++)
{
//we're not interested in inherited properties, because they have no specificity
//and therefore !important rules do not take precedence
if(rules[i].inheritance.length > 0) { continue; }
//run through the cancelled properties to look for !important
//(we need to check for an end-substring so that we don't get confused
// by a content property with "!important" in its text),
for(var j in rules[i].allproperties)
{
if(!rules[i].allproperties.hasOwnProperty(j)
|| rules[i].allproperties[j].status != STATUS_CANCELLED) { continue; }
//and if we find one
if(IMPORTANT_RULE.test(rules[i].allproperties[j].value))
{
//it will take precedence
var precedence = i;
//then run through every subsequent object
//to look for later properties with the same name,
for(var x=i+1; x<rules.length; x++)
{
//and if we find a cancelled one that also has !important
if(typeof rules[x].allproperties[j] != TYPE_UNDEFINED
&& rules[x].allproperties[j].status == STATUS_CANCELLED
&& IMPORTANT_RULE.test(rules[x].allproperties[j].value))
{
//then it will over-take precedence
precedence = x;
}
}
//then iterate from here to find the property identified as having precedence
for(var x=i; x<rules.length; x++)
{
//and once we've found the property with precedence,
//change its status back to "active"
if(x == precedence)
{
rules[x].allproperties[j].status = STATUS_ACTIVE;
//and any earlier active instances of that property to which we did the same thing
//then need their status changed back from "active" to "cancelled"
for(var y=0; y<x; y++)
{
if(typeof rules[y].allproperties[j] != TYPE_UNDEFINED
&& rules[y].allproperties[j].status == STATUS_ACTIVE)
{
rules[y].allproperties[j].status = STATUS_CANCELLED;
}
}
}
//and for any other instance of that property that's already "active"
//update its status to "cancelled"
else
{
if(typeof rules[x].allproperties[j] != TYPE_UNDEFINED
&& rules[x].allproperties[j].status == STATUS_ACTIVE)
{
rules[x].allproperties[j].status = STATUS_CANCELLED;
//there will only be one
break;
}
}
}
}
}
}
//now we need to deal with the interaction of shorthand and longhand properties
//basically, if a shorthand property definition has precedence, then any earlier
//longhand definitions of the same type are cancelled out; however the reverse
//is not true - if a longhand property takes precedence, then any earlier shorthand
//properties remain listed as having precedence, because some of their properties still will
//for eg. if a set of properties (in order, across the stack) goes:
//"margin-left -> margin" then the "margin-left" is cancelled out; but if it goes:
//"margin -> margin-left" then the "margin" is not cancelled out;
//so...
for(var i=0; i<rules.length; i++)
{
//for each active value in the properties object
for(var j in rules[i].allproperties)
{
if(!rules[i].allproperties.hasOwnProperty(j)
|| rules[i].allproperties[j].status != STATUS_ACTIVE) { continue; }
//if this is a shorthand property definition
//which we know simply by whether we have a longhands list for it
if(typeof LONGHAND_PROPERTIES[j] != TYPE_UNDEFINED)
{
//run through every rule up to and including this one
for(var x=0; x<=i; x++)
{
//run through the list of longhand properties that this shorthand property defines
//and if we find an active one in the rule's properties, change its status to "cancelled"
//we don't need to check whether the property is inhertiable for inherited rules
//because non-inheritable properties won't even be listed (obviously)
for(var p=0; p<LONGHAND_PROPERTIES[j].length; p++)
{
var longhand = LONGHAND_PROPERTIES[j][p];
if(typeof rules[x].allproperties[longhand] != TYPE_UNDEFINED
&& rules[x].allproperties[longhand].status == STATUS_ACTIVE)
{
//if we're examining this rule, only cancel out longhand properties
//that are defined earlier in the rule than the shorthand property we're inspecting
//so to make that disctinction we need to work out the index of
//the shorthand and longhand properties in question, if it's this rule
//nb. this presumes that for..in iterates in order of the property being added
//which is a correct presumption in all supported browsers
if(x == i)
{
var n = 0;
for(var q in rules[x].allproperties)
{
if(!rules[x].allproperties.hasOwnProperty(q)) { continue; }
if(q == j) { var jindex = n; }
if(q == longhand) { var pindex = n; }
n++;
}
}
//so go ahead and cancel the rule, if:
// we're not examining this rule;
// or the longhand property is earlier than the shorthand property,
//and: the rule does not have !important
//(earlier checks will have already determined which important rule
// takes precedence, if there's more than one; we just have to
// make sure here that we don't mark it as cancelled, if applicable)
if((x < i || pindex < jindex)
&& !IMPORTANT_RULE.test(rules[x].allproperties[longhand].value))
{
rules[x].allproperties[longhand].status = STATUS_CANCELLED;
}
}
}
}
}
}
}
}
//iterate one more time to do final maintenance on the modified rules array
for(var i=0; i<rules.length; i++)
{
//do a final sweep through the properties objects
//converting any to null that have no members
for(var j in rules[i])
{
if(!rules[i].hasOwnProperty(j)) { continue; }
//we have to spell-out these conditions, not use a regex
//so that they get converted by the compression routine
//otherwise we'll end up with a condition that
//no longer matches the post-compression property names
if((j == 'properties' || j == 'allproperties') && howmany(rules[i][j]) == 0)
{
rules[i][j] = NULL_VALUE;
}
}
//copy allproperties to properties then delete it
rules[i].properties = rules[i].allproperties;
rules[i].allproperties = undefined;
// delete rules[i].allproperties;
//then if the dosort argument is false
//run through the properties and rationalize each property
//so that it goes back to being simple name/value pairs
//and we lose the status property, because none of them have been checked
if(!dosort)
{
for(var k in rules[i].properties)
{
if(!rules[i].properties.hasOwnProperty(k)) { continue; }
rules[i].properties[k] = rules[i].properties[k].value;
}
}
}
//return the modified rules array
return rules;
}
//get the specificity of a selector, using the rules defined in the CSS3 spec
//http://www.w3.org/TR/css3-selectors/#specificity
function getSelectorSpecificity(selector)
{
//create an object for storing the scores,
//ordered by specificity class: [style, id, class, type]
var scores = [0,0,0,0];
//if the selector is an empty string, this indicates a style attribute
//so add 1 to the style score and return the scores array straight away
if(selector === '')
{
scores[0] += 1;
return scores;
}
//create an edited versions of the input selector
//that's stripped of all attribute selectors
//which we can use to avoid confusion with attribute values
//that look like other selectors, for example [href="index.html"]
//might otherwise be confused with a class selector ".html"
var editedselector = selector.replace(REGEX_ATTR_SELECTORS, '');
//look for ID selectors, which have the highest specificity category
//use the selector that's been stripped of attribute selectors,
//to avoid confusion with attribute values containing # symbols
//and we should also check for valid characters and a valid ID pattern
//nb. although "." is allowed in an ID value,
//we'd never be able to test it with an ID selector
//because it will just be interpreted as an ID.class selector
//matches from this regex will also include any pseudo-class or pseudo-elements
//that immediately follow the ID selector, but that's doesn't matter
var matches = editedselector.match(REGEX_ID_SELECTORS);
//add the number of matches (if any) to the id score
if(matches) { scores[1] += matches.length; }
//look for class selectors, in almost exactly the same way
//and with the same caveats as an ID selector, except that
//the valid syntax and pattern is slightly different
var matches = editedselector.match(REGEX_CLASS_SELECTORS);
//add the number of matches (if any) to the class score
if(matches) { scores[2] += matches.length; }
//look for attribute selectors in the unedited selector,
//these are the easiest to detect because there's
//no possibility of confusing them with anything else
matches = selector.match(REGEX_ATTR_SELECTORS);
//add the number of matches (if any) to the class score
if(matches) { scores[2] += matches.length; }
//look for any pseudo-class - except :not, which isn't counted
//use the selector that's been stripped of attribute selectors
//since there's a limited number of pseudos, we can test for each one specifically
//nb. this will let through some fake permutations, like
//"only-child" or "first-last-of-type", but I don't think that's worth worrying about
var matches = editedselector.match(REGEX_PSEUDO_CLASSES_EXCEPT_NOT);
//add the number of matches (if any) to the class score
if(matches) { scores[2] += matches.length; }
//look for element type selectors, which is by far the hardest to do
//because it's so easily confused for other types of selector
//because it has no distinguishing tokens of its own, only the lack of them
//so to begin with we'll use the selector that's been stripped of attribute selectors
//then remove all pseudo-classes except :not(), and all pseudo-elements,
//(but remove the actual word ":not", because XML element names are allowed to
// contain colons and it would otherwise look like an element called ":not")
//remove any namespace prefix (at the start of the selector, or inside a :not bracket)
//and remove any ID or class selectors
//then finally (if there's anything left!) check for valid tag name characters
var typeonlyselector = editedselector.replace(REGEX_PSEUDO_CLASSES_EXCEPT_NOT, '')
.replace(REGEX_PSEUDO_ELEMENTS, '')
.replace(/(:not)/ig, '')
.replace(/(^|\()([_a-z0-9-\.\\]+\|)/ig, '$1')
.replace(REGEX_ID_SELECTORS, '')
.replace(REGEX_CLASS_SELECTORS, '');
var matches = typeonlyselector.match(/([_a-z0-9-:\\]+)/ig);
//add the number of matches (if any) to the type score
if(matches) { scores[3] += matches.length; }
//and last but not least, look for pseudo-elements
//use the selector that's been stripped of attribute selectors
//then we can identify them easily and specifically, given such a limited range
var matches = editedselector.match(REGEX_PSEUDO_ELEMENTS);
//add the number of matches (if any) to the type score
if(matches) { scores[3] += matches.length; }
//return the final scores array
return scores;
}
//parse a selectorText string for IE to lowercase the element type selectors
function lowercaseIETypeSelectors(selectorstring)
{
//temporarily tokenize attribute selectors so that
//any all-uppercase attribute values are not affected
var attrselectors = selectorstring.match(/(^|[^\(])(\[[^\]]+\])($|[^\)])/ig);
if(attrselectors)
{
for(var a=0; a<attrselectors.length; a++)
{
//we only want the data in $2, but we don't have that much
//control over match(), so we'll have to edit them manually
if(attrselectors[a].charAt(0) != '[')
{
attrselectors[a] = attrselectors[a].substr(1, attrselectors[a].length - 1);
}
if(attrselectors[a].charAt(attrselectors[a].length -1) != ']')
{
attrselectors[a] = attrselectors[a].substr(0, attrselectors[a].length - 1);
}
//then convert the selector into a token
selectorstring = selectorstring.replace(attrselectors[a], '{a' + a + '}');
}
}
//do the same thing with class and ID selectors
var cidselectors = selectorstring.match(/([#\.][a-z]+[_a-z0-9-:\\]*)/ig);
if(cidselectors)
{
for(var c=0; c<cidselectors.length; c++)
{
selectorstring = selectorstring.replace(cidselectors[c], '{c' + c + '}');
}
}
//now convert any pure uppercase element names to lowercase
selectorstring = selectorstring.replace(/([A-Z1-6]+)/g, function(a) { return a.toLowerCase(); });
//then reconvert the attribute selector and
//class/id selector tokens back to their original selectors
if(attrselectors)
{
for(a=0; a<attrselectors.length; a++)
{
selectorstring = selectorstring.replace('{a' + a + '}', attrselectors[a]);
}
}
if(cidselectors)
{
for(c=0; c<cidselectors.length; c++)
{
selectorstring = selectorstring.replace('{c' + c + '}', cidselectors[c]);
}
}
//and return the edited selector
return selectorstring;
}
//parse a cssText string for IE to lowercase the property names
function lowercaseIEPropertyNames(styletext)
{
//split the css text by semi-colons
styletext = styletext.split(';');
//iterate through the resulting properties
for(var j=0; j<styletext.length; j++)
{
//split the property by its first colon to get the name and value
//this convoluted funkiness is because IE failed to return
//any matches when a first-colon regex was used, so instead we split
//by all colons and then join together any excess matches
var matches = styletext[j].split(':');
while(matches.length > 2) { matches[1] += ':' + matches.pop(); }
//if we only have one match, the most likely reason
//is that the semi-colon split operation latched onto
//a semi-colon that wasn't a property delimeter
//probably it was inside generated content
//and hence the value we're splitting doesn't contain a colon at all
//so if that happens we'll just have to remove it and continue
if(matches.length == 1)
{
styletext.splice(j, 1);
j--;
continue;
}
//trim them both and convert the property name to lower case
matches[0] = trim(matches[0]).toLowerCase();
matches[1] = trim(matches[1]);
//then join them back together with consistent spacing
styletext[j] = matches[0] + ': ' + matches[1];
}
//join them all back together with consistent spacing and return the reformated string
//nb. the smei-colon and space are separated out so that we can universally compress "; " to ";"
return styletext.join(';' + ' ');
}
//parse the selector text of an individual style rule
//to create an array of individual selectors
function parseSelectorText(selectorstring)
{
//create an array of selectors to return
var selectors = [];
//if the selectorstring is empty just return the empty array
if(selectorstring == '') { return selectors; }
//split the selector by commas
selectorstring = selectorstring.split(',');
//iterate through and add each trimmed selector to the array
for(var j=0; j<selectorstring.length; j++)
{
selectors.push(trim(selectorstring[j]));
}
//return the selectors array
return selectors;
}
//parse the properties text of an individual style rule
//to create object members in property/value pairs
function parsePropertyText(styletext, properties, inherited)
{
//if the css text is empty return the object unchanged
if(styletext == '') { return properties; }
//parse all line-breaks out of the string
styletext = styletext.replace(/[\r\n]/gm, '');
//split the css text by semi-colon to get individual property definitions
styletext = styletext.split(';');
//now iterate through that array and create an object member for each one
//counting the number of properties as we go so we can create a length property
for(var i=0; i<styletext.length; i++)
{
//trim this value and continue if it's empty
styletext[i] = trim(styletext[i]);
if(styletext[i] == '') { continue; }
//then split the property by its colon to get the name and value
//this convoluted funkiness is for IE's benefit, which failed to return
//any matches when a first-colon regex was used, so instead we split
//by all colons and then join together any excess matches
var matches = styletext[i].split(':');
while(matches.length > 2) { matches[1] += ':' + matches.pop(); }
//if we only have one match, the most likely reason
//is that the semi-colon split operation latched onto
//a semi-colon that wasn't a property delimeter
//probably it was inside generated content
//and hence the value we're splitting doesn't contain a colon at all
//so if that happens we'll just have to ignore it and continue
if(matches.length == 1) { continue; }
//trim the first match and save it to a key variable
var key = trim(matches[0]);
//if the rule is not inherited, or the property is inheritable
if(!inherited || (inherited && typeof INHERITED_PROPS[key] != TYPE_UNDEFINED))
{
//don't overwrite an existing rule if it has !important
//unless it was an inherited rule, or
//unless this rule also has !important
//(we need to check for an end-substring so that we don't get confused
// by a content property with "!important" in its text)
if(typeof properties[key] == TYPE_UNDEFINED
|| !IMPORTANT_RULE.test(properties[key].property)
|| properties[key].inherited == BOOLEAN_TRUE
|| IMPORTANT_RULE.test(matches[1]))
{
//save this value to the properties array
//along with an inherited flag
properties[key] = {
'property' : trim(matches[1]),
'inherited' : inherited
};
}
}
}
//return the modified properties object
return properties;
}
//compare two media strings to see if the context affects the test type
//and if so adjust the value of the test string accordingly
//this is for cases where, for example, a <link media="screen"> contains "@media all"
//and therefore the actual media that applies to those rules is "screen" not "all"
//or where the context actually precludes the test type, for example <link media="print">
//containing "@media screen", in which case we return "none"
function contextualizeMediaTypes(context, types)
{
//convert the context string to an object
var contextTypes = lexiconize(context, ',');
//we're not currently testing for media query context
//so remove and save any media query from the types string
//so that we don't get confused with them
var mediaQuery = /([ \t]and.*$)/i,
matches = types.match(mediaQuery),
typesQuery = (matches ? matches[0] : '');
//we have to remove any "only" token so that we can do comparisons
//against the media-types string that remains, but this means
//that any such token will be lost permanently from the reported media
//that's basically okay though because it doesn't affect how they evaluate
types = types.replace(mediaQuery, '').replace(/only[ \t]+/i, '');
//if the contextTypes object contains "all" then any inner types are okay
//so just rejoin the types string to its saved media query and return it unmodified
if(typeof contextTypes['all'] != TYPE_UNDEFINED)
{
return types + typesQuery;
}
//otherwise we have to to check each type against the context
//so first split and trim the types string into individual media types
types = types.split(',');
for(var i=0; i<types.length; i++)
{
types[i] = trim(types[i]);
}
//then iterate through the types
for(var i=0; i<types.length; i++)
{
//firstly, if the media type is "all"
//then all the context media types apply here
//so add them all individually to the types array
//making sure we avoid duplication and not including media queries
if(types[i] == MEDIA_ALL)
{
for(var m in contextTypes)
{
if(!contextTypes.hasOwnProperty(m)) { continue; }
if(arrayContains(types, m) == NULL_VALUE && !/[\(\)]/.test(m))
{
types.push(m);
}
}
}
//then if the context does not contain this media type
//then this media type either doesn't apply in context (so delete it from the array)
//or it's "all" in which case we've already added the individual context types (so do nothing)
if(typeof contextTypes[types[i]] == TYPE_UNDEFINED)
{
types.splice(i, 1);
i --;
}
}
//if the types array is now empty then
//the context has precluded the test type entirely
//so if that's the case add the value "none" to the array
//so that it ends up getting returned as "none [+ media query]"
if(types.length == 0)
{
types.push(MEDIA_NONE);
}
//rejoin the types array, add back any media query, and return it
//nb. the comma and space are separated like this to avoid compression
return types.join(',' + ' ') + typesQuery;
}
//compare a media types array argument with the media types of a css rules
//to return true of false by whther the rule matches the argument criteria
function mediaMatches(media, rule)
{
//get the rule's media property and convert it to an object
//so that we can easily test it as typeof mediaTypes[media]
var mediaTypes = lexiconize(rule.media, ',');
//begin by assuming no match
var matches = BOOLEAN_FALSE;
//then iterate through the input array of media types to look for at least one match
//which is all we need for the rule to match the media criteria
for(var j=0; j<media.length; j++)
{
//we have a match if this rule includes at least one of the specified media
//or the specified media is "all" and the rule's media isn't "none"
//or the rule's media contains "all" and the specified media isn't "none"
//(this allows us to get collections of rules that don't apply to any media,
// as well as the collections that apply to one or more)
if(typeof mediaTypes[media[j]] != TYPE_UNDEFINED
|| (typeof mediaTypes[MEDIA_NONE] == TYPE_UNDEFINED && media[j] == MEDIA_ALL)
|| (typeof mediaTypes[MEDIA_ALL] != TYPE_UNDEFINED) && media[j] != MEDIA_NONE)
{
matches = BOOLEAN_TRUE;
break;
}
}
//return the result
return matches;
}
//detect the current view media
//this will only work on HTML pages, but it's better than nothing!
function getViewMedia()
{
//set screen as default in case we can't get the value
var viewmedia = MEDIA_SCREEN;
//wrap all this in exception handling
//in case we're in an environment where it fails
try
{
//begin by creating a test node
//it's ID will match an ID selector in the test stylesheet we create
//with a display value that's different from the one we'll define there
//since we're setting it in .style it will have enough specificity
//to override the stylesheet value, even though that has !important
//we have to insert this node before the body's first child, we can't append to body
//because it's not yet closed, and trying to append to an unclosed element
//throws an error in IE7/8 (KB927917)
var pagebody = page.getElementsByTagName('body').item(0);
var testmedianode = pagebody.insertBefore(page.createElement('span'), pagebody.firstChild);
testmedianode.id = 'cssutilitiestest'+'medianode';
testmedianode.style.display = 'inline';
//create the array of all CSS media types; the list is a delimited string which is split here
//cos that worked out to less code than declaring an array of separate strings :)
var mediaTypes = MEDIA_TYPES_LIST.split(',');
//we see a failure condition in Safari 3 and Konqueror, where it
//applies the stylesheet rule all the time, irrespective of applicable media,
//and this manifests in it returning whichever type is first in the array
//so to catch that we'll seed the array with a fake value we can check against
mediaTypes.splice(0, 0, 'fake');
//create a stylesheet and add a rule for a test node we're going to create
//we do this differently depending on whether we're IE or everyone else
//because the techniques we're using are mutually exclusive to those browsers
//we'll define a display style that's different from the test node's default
//using !important to give it the highest possible specificity,
//and a convoluted library-specific ID value, to avoid conflict
//with any existing styles and elements on the test document
if(IEXPLORER)
{
//create a new stylesheet and add the rule using its proprietary technique
var viewmediastylesheet = page.createStyleSheet();
viewmediastylesheet.addRule('#cssutilitiestest'+'medianode', 'display:block !important;');
}
else
{
//create a style node and add the rule as a child text node
var viewmedianode = page.getElementsByTagName('head').item(0).appendChild(page.createElement('style'));
viewmedianode.setAttribute('type', 'text/css');
viewmedianode.appendChild(page.createTextNode('#cssutilitiestest'+'medianode{display:block !important;}'));
//now get a reference to the stylesheet we just created from the document.styleSheets collection
//using the stylesheet rather than the original stylenode makes it work in Webkit
var viewmediastylesheet = page.styleSheets[page.styleSheets.length - 1];
}
//now iterate through the CSS media types to test each one in turn
for(var i=0; i<mediaTypes.length; i++)
{
//set the stylesheet media to this media type
if(IEXPLORER) { viewmediastylesheet.media = mediaTypes[i]; }
else { viewmediastylesheet.media.mediaText = mediaTypes[i]; }
//if the test node is now taking on the display property
//that was specified in the test stylesheet
//then we know that the specified media applies to the current view :-)
if((IEXPLORER && testmedianode.currentStyle.display == 'block')
|| (!IEXPLORER && page.defaultView.getComputedStyle(testmedianode, '').getPropertyValue('display') == 'block'))
{
viewmedia = mediaTypes[i];
break;
}
}
//remove the test node and stylesheet
testmedianode.parentNode.removeChild(testmedianode);
if(IEXPLORER) { viewmedianode = viewmediastylesheet.owningElement; }
viewmedianode.parentNode.removeChild(viewmedianode);
}
//silently fail [and set media back to default]
catch(err) { viewmedia = MEDIA_SCREEN; }
//if the returned value is "fake" then it's been caught by our seeding trap
//so if that happens we'll just have to reset back to the default again
if(viewmedia == 'fake') { viewmedia = MEDIA_SCREEN; }
//return the final value
return viewmedia;
}
//watch a collection of stylesheets for their disabled state to change
//and when one of them does, re-initialize the script
//we used to have a few different approaches for different browsers, but we couldn't rely on them
//for example - in opera and firefox we can watch the disabled state of each stylesheet's ownerNode
//by extending the Object watch() method to the ownerNode elements, and that's virtually instantaneous
//but it won't cover view-menu stylesheet switching, which toggles the disabled state
//of the stylesheet itself, not its owner node; and we can't guarantee that any other browser
//won't be implementing some extension or browser feature that works in the same way
//so we can't take a chance of having innacurate data, we'll just have to use the timer
//so in fact only Safari defaults to a different method, and that's only because the timer method
//won't work, because disabled stylesheets don't show up in the stylesheets collection at all
//of course once we've safely implemented the timer method, we could still use the other techniques
//as well, but what's the point, just for a slightly faster response, it doesn't warrant the additional code
function watchStyleSheets()
{
//so begin by creating an optimized data set
//so we've got less to look at in each timer iteration
//and thereby minimize the timer's CPU load
for(var livesheets=[], i=0; i<THIS._stylesheets.length; i++)
{
//for author mode we use the stylenode.styleSheet/sheet references
//(or just the stylenode for safari);
//or for browser mode we use the stylesheet references directly
var stylesheet = mode == MODE_AUTHOR
? (THIS._stylesheets[i].stylenode == BOOLEAN_FALSE || THIS._stylesheets[i].stylenode == NULL_VALUE)
? NULL_VALUE
: (SAFARI || KONQUEROR || WEBKIT)
? THIS._stylesheets[i].stylenode
//nb. although we know this condition is IE vs. others, an object test is better
: THIS._stylesheets[i].stylenode[typeof THIS._stylesheets[i].stylenode.styleSheet != TYPE_UNDEFINED ? 'styleSheet' : 'sheet']
: THIS._stylesheets[i].stylesheet;
//ignore null stylesheets, such as duplicates,
//or those we couldn't read for security reasons
if(stylesheet == NULL_VALUE) { continue; }
//save the stylesheet reference, and its current disabled state
livesheets.push({
'stylesheet' : stylesheet,
'disabled' : stylesheet.disabled
});
}
//for safari and konqueror in browser mode,
//disabled stylesheets don't show up in the document stylesheets collection at all
//so what we do to detect the change is assign an ident to the owner node of each one
//and then create a timer that assigns new idents to any stylesheet node that doesn't have one
//and thereby it knows that something's changed because a new ident will appear
//or an existing ident will disappear, due to stylesheets appearing and disappearing
//from the document stylesheets collection.
//actually in safari, a change in the disabled state of a stylesheet causes its ownerNode to fire a
//DOMSubtreeModified event in response to the addition or removal of that node from the DOM
//so we culd detect that change, and use it as a springboard for re-initialization
//but that would mean code forking for safari and konqueror, which is arguably a waste of code
//since this method works for both; the only real difference is that the subtree modified
//technique would be faster (virtually instant), but again, it's not really worth it for the extra code
if((SAFARI || KONQUEROR || WEBKIT) && mode == MODE_BROWSER)
{
//function that iterates through the stylesheets collection
//and assign an ident to any owner node that doesn't already have one
function createStyleSheetIdents()
{
for(var idents=[], dsheets=page.styleSheets, i=0; i<dsheets.length; i++)
{
if(typeof dsheets[i].ownerNode.__ident == TYPE_UNDEFINED)
{
dsheets[i].ownerNode.__ident = new Date().getTime() + '' + Math.round(Math.random() * 10000);
}
idents.push(dsheets[i].ownerNode.__ident);
}
return idents;
}
//assign the initial control idents then start the watch timer
var controlidents = createStyleSheetIdents(),
watcher = window.setInterval(function()
{
//assign new idents as applicable
var currentidents = createStyleSheetIdents();
//if either of the ident arrays has changed
if(controlidents.join() != currentidents.join())
{
//clear the timer then re-initialize, calling any stored init callback
window.clearInterval(watcher);
THIS.init(typeof THIS.initcallback != TYPE_UNDEFINED ? THIS.initcallback : NULL_VALUE);
}
}, WATCHER_SPEED_TIMER);
}
//for everybody else we'll use a timer to watch the disabled property
//of all the valid stylesheets or stylenodes in the debug stylesheets array
else
{
//start a watch timer, saving a reference we can stop it with
//I actually did find a complete set of techniques for watching changes
//in the disabled state of stylesheets without resorting to setInterval
//however they all worked off the state of the ownerElement, but browsers'
//built-in stylesheets switching (eg. the Style menu in Opera)
//work off the disabled property of the stylesheet itself, not its owner node
//so all of that is unuseable, and a timer it has to be :-(
//I did consider adding it as well, just so that we get faster responses
//but what's the point? it's just a whole bunch more code to little real benefit
var watcher = window.setInterval(function()
{
//iterate through the optimized stylesheets set
for(var i=0; i<livesheets.length; i++)
{
//if the current disabled state differs from the stored state
if(livesheets[i].stylesheet.disabled != livesheets[i].disabled)
{
//clear the timer then re-initialize, calling any stored init callback
window.clearInterval(watcher);
THIS.init(typeof THIS.initcallback != TYPE_UNDEFINED ? THIS.initcallback : NULL_VALUE);
//set a flag to break the iteration
var dobreak = BOOLEAN_TRUE;
}
//this break is crucial, to prevent multiple changes
//from firing multiple re-initializations
//it's basically doing the same job as the buffer
//used for other browsers, but in a different way
if(typeof dobreak != TYPE_UNDEFINED) { break; }
}
}, WATCHER_SPEED_TIMER);
}
}
//-- private general utility methods --//
//wrapper for Selectors APIs
function getElementsBySelector(selector)
{
//manually filter out pseudo-elements, which of course never directly
//match an element reference, but some query implementations let them through
//eg. DOMAssistant in IE, and the native implementation in Safari 3
if(REGEX_SINGLE_PSEUDO_ELEMENT.test(selector))
{
var selectors = selector.split(',');
for(var i=0; i<selectors.length; i++)
{
if(REGEX_SINGLE_PSEUDO_ELEMENT.test(selectors[i]))
{
selectors.splice(i, 1);
i--;
}
}
selector = selectors.join(',');
}
//if the selector is now empty return zero matches
if(trim(selector) == '') { return []; }
//nb. whichever selectors api method we use
//we're going to silently fail around calls to it
//so that we don't get errors from invalid selectors
//the only place where we do catch and throw an explicit error
//is where we can establish that a library we need just isn't there
//if the api property is false then we've already established
//that the browser has native support for querySelectorAll
//and it hasn't been set to true to specify that the
//fallback library function should be used for everyone
if(api == BOOLEAN_FALSE)
{
try { return page.querySelectorAll(selector); }
catch(err) { return []; }
}
//otherwise, if we have a qsa function
//defined [and pre-tested], then call and return that
if(typeof qsa == TYPE_FUNCTION)
{
try { return qsa(selector, page); }
catch(err) { return []; }
}
//otherwise use the fallback library function
try
{
//if the library isn't there at all, throw the missing sapi error
if(typeof Selector != TYPE_FUNCTION)
{
throw(new Error(FATAL_ERROR_MISSING_SAPI));
}
//in IE6, Selector sometimes returns undefined instead of an array
//for selectors containing some pseudo-classes, like :hover
//I haven't clearly established why (it isn't the colon causing regex failure
//because other pseudos like :active don't display the same behavior),
//but anyway, to prevent unduev errors reaching the end user or developer,
//we'll catch that situation and return an empty array instead
//oddly enough though, this doesn't seem to prevent it from matching
//the selectors it returns undefined for correctly against their element
//so it must be a different test that causes it, not the one that matches it!?
var r = Selector(selector, page);
return typeof r == TYPE_UNDEFINED ? [] : r;
}
catch(err)
{
//if the error is the missing sapi error we threw for the missing library
//then allow it to be output to the console
if(err.message == FATAL_ERROR_MISSING_SAPI) { throw(err); }
//otherwise silently fail as normal
return [];
}
//we return an empty array after a query method silent exception
//which will usually be because it doesn't like the selector
//either it doesn't understand it, or the selector is wrong or mangled in some way
}
//get an array of elements specified by tagname or "*"
function getTheseElements(tagname)
{
//get a collection of the specified elements
//and convert it to an array so we can use array methods on it
//use a namespaced method if appropriate so that for any given tagname
//we look in all namespaces, allowing us to support things like "svg:style"
//as well as explicitly-namespaced HTML elements like "html:link"
try
{
var thesenodes = arrayifize(
THIS._isXML
? page.getElementsByTagNameNS('*', tagname)
: page.getElementsByTagName(tagname)
);
}
//just in case anything goes wrong, create an empty array
//this is just for paranoid safety - there's no known reason why it shoud fail
catch(err) { thesenodes = []; }
//we may need to filter out comments, because IE includes them in its "*" collection!
//so in fact what we'll do is just exclude anything that's not an element
//in case there's anything else it stupidly adds that I don't know about
//[i spoke too soon...] if the page contains any custom elements,
//IE also adds their closing tag as an additional node to the collection!!
//so we'll have to filter those out as well by checking for "/" at the start of the tag name
//interestingly enough, all browsers except Opera convert custom element tag names
//to the canonical uppercase form in HTML; fwiw I think Opera is right - they're not HTML elements
//so there's no reason to implement the canonical form; but other browsers probably
//aren't thinking that deeply, they're probably just doing it for every element they encounter
//also of interest is the fact that IE[6] doesn't apply style attribute properties
//to a custom element's rendering, while all other browsers do; but I'm not going to allow for that
//when returning property collections in browser mode ... maybe in the next version :-P
//I mean ffs, I only noticed by accident; it's too late in the day to think about custom elements
//except to the extent that I have done here, but that's only to ensure that the array indices
//in the styleattrs array remain accurate with regard to matching indices in the "*" collection
if(tagname == '*')
{
for(var i=0; i<thesenodes.length; i++)
{
if(thesenodes[i].nodeType != 1 || thesenodes[i].tagName.charAt(0) == '/')
{
thesenodes.splice(i--, 1);
}
}
}
//return the final array
return thesenodes;
}
//late initialize method, used internally to initialize the script
//if a public method is called without it having been pre-initialized
function lateinit()
{
//if the script hasn't been initialized, do it now
if(typeof THIS._cssRules == TYPE_UNDEFINED)
{
THIS.init();
}
}
//process a method's arguments array to create an indexed object
//this makes it possible for users to omit any number of intermediate
//optional arguments, and have the onfinished callback be the last one
function processArguments(argsary, keys)
{
//add onfinished to the keys
//it's always the last argument to every public method
//so we add it here to save listing it each time
//and to maintain control over its existence
keys.push('onfinished');
//create the empty indexed arguments object
//then iterate through the specified keys array
for(var args={}, i=0; i<keys.length; i++)
{
//as soon as we encounter a function within the arguments array
//nullify all subsequent arguments then index the function to args.onfinished
//we put onfinished at the end to preserve the order of arguments
//which is ultimately for consistency and therefore easier comprehension
//nb. should we ever require any of the public methods to take a function argument
//other than the callback, we will need to redesign this code to accomodate
if(typeof argsary[i] == TYPE_FUNCTION)
{
for(var j=i; j<keys.length; j++)
{
args[keys[j]] = NULL_VALUE;
}
args.onfinished = argsary[i];
//and we're done
break;
}
//otherwise if this argument is undefined
//create a null value at the corresponding index in the arguments object
//all the other arg processing methods will then identify null
//the same as undefined, and so put the appropriate default value in its place
//(or throw an exception over it, if that's the appropriate response)
else if(typeof argsary[i] == TYPE_UNDEFINED)
{
args[keys[i]] = NULL_VALUE;
}
//if the argument is defined then add it
//to the args object using the applicable index
else
{
args[keys[i]] = argsary[i];
}
}
//return the finished arguments object
return args;
}
//process a method's element argument
function processElementArgument(argelement, calledby)
{
//if the element argument is a string beginning with "#"
//then look for an element with that ID within the context document
//if there isn't one then element will be set to null
//which will make it fail the reference test that follows
//nb. we use the leading # partly so we have future scope
//for accepting other kinds of reference string,
//and partly for consistency with CSS reference syntax,
//but mostly just because I think it looks cooler :-D
if(typeof argelement == TYPE_STRING && argelement.charAt(0) == '#')
{
argelement = page.getElementById(argelement.substr(1, argelement.length - 1));
}
//if element is undefined, null, or not an element, throw an error
if(typeof argelement == TYPE_UNDEFINED || argelement == NULL_VALUE
|| typeof argelement.nodeType == TYPE_UNDEFINED || argelement.nodeType != 1)
{
throw(new Error(errorMessageNoElement.replace('%method', calledby)));
}
//return the element reference
return argelement;
}
//process a method's media argument
function processMediaArgument(argmedia)
{
//if media is undefined, empty or null, default to "screen"
if(typeof argmedia == TYPE_UNDEFINED || argmedia == '' || argmedia == NULL_VALUE)
{
argmedia = MEDIA_SCREEN;
}
//if argmedia is or contains the discreet value "*", this is the universal shortcut
//meaning all media, including no media, which translates to "all,none"
//so just set the argmedia value to that
if(/(^|,)\*(,|$)/.test(argmedia))
{
argmedia = 'all,none';
}
//split the media argument into an array and trim the members
argmedia = argmedia.split(',');
for(var i=0; i<argmedia.length; i++)
{
argmedia[i] = trim(argmedia[i]);
//if the value is "current" then set it to the current view media
if(argmedia[i] == MEDIA_CURRENT)
{
argmedia[i] = THIS._viewmedia;
}
}
//return the processed argument
return argmedia;
}
//process a method's accept argument
function processAcceptArgument(argaccept)
{
//if accept is undefined, empty, null or "null"
//default to "*" which means "everything"
if(typeof argaccept == TYPE_UNDEFINED
|| argaccept == '' || argaccept == NULL_VALUE || argaccept == 'null')
{
argaccept = '*';
}
//then if accept is still a string, process it into an object
//so that we can easily test for values using typeof accept[key]
//or if it's completely or contains "*" then make it just "*"
//we have this condition for if it's already been processed
//because this can get called by recursive instances of getCSSRules
//but we only actually need to process the argument
//on the first instance before it's already been processed
else if(typeof argaccept == TYPE_STRING)
{
//trim the value
argaccept = trim(argaccept);
//if the value is "*" just leave it
if(argaccept !== '*')
{
//if the string contains "*" among other delimited values,
//make it just "*" on its own
//we test for delimiters so that, for example,
//"css*" is invalid rather than equating to "css,*"
if(/(,\s*\*|\*\s*,)/.test(argaccept))
{
argaccept = '*';
}
//otherwise convert the string into an object
else
{
argaccept = lexiconize(argaccept, ',');
}
}
}
//return the processed argument or existing object
return argaccept;
}
//process a method's onfinished callback argument
function processOnfinishedArgument(argonfinished)
{
//if onfinished is [undefined or null or] not a function, set it to null
if(typeof argonfinished != TYPE_FUNCTION)
{
argonfinished = NULL_VALUE;
}
//return the processed argument
return argonfinished;
}
//trim leading and trailing whitespace from a string
function trim(str)
{
return str.replace(/^\s+|\s+$/g, '');
}
//check if an array contains a member
//this can check simple arrays for whether they contain a member
//(and in which case returns the member it found)
//or it can check arrays of objects for whether they contain
//a member with a property name matching the input key
//(and in which case returns the member's parent object)
//if there's no match it returns null
function arrayContains(ary, member, key)
{
for(var i=0; i<ary.length; i++)
{
if(typeof key != TYPE_UNDEFINED && typeof ary[i] == TYPE_OBJECT)
{
if(ary[i][key] == member)
{
return ary[i];
}
}
else if(ary[i] == member)
{
return ary[i];
}
}
return NULL_VALUE;
}
//convert a DOM NodeList to an array
//cool name huh :-D
function arrayifize(nodelist)
{
for(var ary=[], i=0; i<nodelist.length; i++)
{
ary.push(nodelist[i]);
}
return ary;
}
//convert a delimited string into a dictionary object of empty properties
//so that we can easily test for a value with typeof obj[key]
function lexiconize(str, delimiter)
{
//create the empty object
var obj = {};
//split the string by its delimiter
str = str.split(delimiter);
//if the final member contains media queries, split them off too
//and add them to the array, deleting the whole final member itself
//there's no need to look for the "only" token here because
//it will be gone from any media-types string before it gets here
var tmp = str[str.length - 1];
if(/[ \t]and/i.test(tmp))
{
tmp = tmp.split(/[ \t]and/i);
str.splice(str.length-1, 1);
for(var i=0; i<tmp.length; i++)
{
str.push(tmp[i]);
}
}
//iterate through the array and save each trimmed member to the object
//we only need the object keys, so the values can be anything
for(var i=0; i<str.length; i++)
{
obj[trim(str[i])] = '';
}
//return the object
return obj;
}
//count the members of an object
function howmany(obj)
{
var n = 0;
for(var i in obj)
{
if(!obj.hasOwnProperty(i)) { continue; }
n ++;
}
return n;
}
//qualify an HREF to form a complete URI
function qualifyHREF(href, context)
{
//if the href is undefined return an empty string
//the only time this is known to have an impact is in XHTML
//when an xml-stylesheet PI has its href pseudo-attribute
//in mixed or uppercase, resulting in no pseudoattrs.href property
//and ultimately in an xml-stylesheet PI marked as "data is not CSS"
//(which matches what implementations actually do:
// they can't load the stylesheet, because it has no known "href"!)
//more generally, this will indirectly catch failure to retrieve a stylesheet's href
//for any exceptional reason that hasn't been accounted for
if(typeof href == TYPE_UNDEFINED) { return ''; }
//extract the protocol, host and path
//and create a location object with the data
//nb. the first replacement here goes "/"+"/" instead of "//"
//so that it's safe to do a global removal of //.*
//when removing comments to minimize the script
var parts = context.replace('/'+'/', '/').split('/');
var loc = {
'protocol' : parts[0],
'host' : parts[1]
};
parts.splice(0, 2);
loc.pathname = '/' + parts.join('/');
//build a base URI from the protocol plus host (which includes port if applicable)
//nb. the "/"+"/" is for the same minimizing reason
var uri = loc.protocol + '/'+'/' + loc.host;
//if the input path is relative-from-here
//just delete the ./ token to make it relative
if(/^(\.\/)([^\/]?)/.test(href))
{
href = href.replace(/^(\.\/)([^\/]?)/, '$2');
}
//if the input href is already qualified, copy it unchanged
if(/(^([a-z]+)\:\/\/)/.test(href))
{
uri = href;
}
//or if the input href begins with a leading slash, then it's base relative
//so just add the input href to the base URI
else if(href.substr(0, 1) == '/')
{
uri += href;
}
//or if it's an up-reference we need to compute the path
else if(/^((\.\.\/)+)([^\/].*$)/.test(href))
{
//get the last part of the path, minus up-references
var lastpath = href.match(/^((\.\.\/)+)([^\/].*$)/);
lastpath = lastpath[lastpath.length - 1];
//count the number of up-references
var references = href.split('../').length - 1;
//get the path parts and delete the last one (this page or directory)
var parts = loc.pathname.split('/');
parts = parts.splice(0, parts.length - 1);
//for each of the up-references, delete the last part of the path
for(var i=0; i<references; i++)
{
parts = parts.splice(0, parts.length - 1);
}
//now rebuild the path
var path = '';
for(i=0; i<parts.length; i++)
{
if(parts[i] != '')
{
path += '/' + parts[i];
}
}
path += '/';
//and add the last part of the path
path += lastpath;
//then add the path and input href to the base URI
uri += path;
}
//otherwise it's a relative path,
else
{
//calculate the path to this directory
path = '';
parts = loc.pathname.split('/');
parts = parts.splice(0, parts.length - 1);
for(var i=0; i<parts.length; i++)
{
if(parts[i] != '')
{
path += '/' + parts[i];
}
}
path += '/';
//then add the path and input href to the base URI
uri += path + href;
}
//return the final uri
return uri;
}
//load CSS or HTML data via XHR, either asynchronously or synchronously
//according to the value of the async flag argument
//nb. we only use this method to retrieve data as text, never for active scripting,
//because calling eval() on network data is just too dangerous
function ajaxload(async, uri, oncomplete, onfail)
{
//try to create a request object
//arranging the two conditions this way is for IE7/8's benefit
//so that it works with any combination of ActiveX or Native XHR settings,
//as long as one or the other is enabled; but if both are enabled
//it prefers ActiveX, which means it still works with local files
//(Native XHR in IE7/8 is blocked and throws "access is denied",
// but ActiveX is permitted if the user allows it [default is to prompt])
var requestobject = NULL_VALUE;
if(typeof window.ActiveXObject != TYPE_UNDEFINED)
{
try { requestobject = new ActiveXObject('Microsoft.XMLHTTP'); }
catch(err) { requestobject = NULL_VALUE; }
}
if(requestobject == NULL_VALUE && typeof window.XMLHttpRequest != TYPE_UNDEFINED)
{
try { requestobject = new XMLHttpRequest(); }
catch(err) { requestobject = NULL_VALUE; }
}
//if we failed to initiate a request then we can't do anything else
//so we'll have to throw a fatal error and stop
if(requestobject == NULL_VALUE)
{
throw(new Error(FATAL_ERROR_NO_XHR));
}
//add a timestamp to the URI to prevent caching,
//there are more subtle ways of trying to ensure a fresh response
//but this approach has always proven the most reliable
//we should give this a reasonably unique key
//just in case the URI already has cgi parameters
//#uri += (uri.indexOf('?') == -1 ? '?' : '&') + 'cssutilstamp=' + new Date().getTime();
//open the request using the specified async flag
requestobject.open('GET', uri, async);
//try to set a request header that defines a library user-agent string
//this is something I like to do with programatic networking like ajax
//so that site owners can identify such requests in their server logs
//it's mostly for fun - like a little electronic hello :-)
//and as a kind of side-door viral marketing! but it also provides
//a means of filtering or otherwise handling such requests specially
//ultimately though it's not essential, so silently fail [eg. if the method is unsupported]
try { requestobject.setRequestHeader('User-Agent', LIBRARY_VERSION_STRING); }catch(err){}
//private response function
function ajaxresponse(requestobject)
{
//if the status is okay, pass the trimmed response text to oncomplete
//along with the Content-Type response header from this request
//nb. we allow the status code 0 so that file:// addresses are supported
//unfortunately though such requests also return an empty content-type
//which means that we have to just accept all response text
//without the ability to check whether its MIME type is text/css
if(/(0|200|304)/.test(requestobject.status.toString()))
{
oncomplete(
requestobject.responseText,
requestobject.getResponseHeader('Content-Type')
);
}
//otherwise pass the trimmed status text to onfail
else
{
onfail(trim(requestobject.statusText), requestobject.status);
}
}
//if the async flag is true we're making an asynchronous request
if(async == BOOLEAN_TRUE)
{
//create a readystatechange handler
requestobject.onreadystatechange = function()
{
//when the request completes
if(requestobject.readyState == 4)
{
//***DEV SEMI-RANDOMIZED LATENCY
//setTimeout(function() {
//call the response function with the request object
ajaxresponse(requestobject);
//***DEV SEMI-RANDOMIZED LATENCY
//}, (33 * Math.random()));
}
};
//make the request, with exception handling
//to catch failures such as no network connection
try
{
requestobject.send(NULL_VALUE);
}
//if just that individual request failed
//pass a network or security error message to onfail
//we have to be ambiguous here because this exception handler alone
//is not enough to establish whether the request failed
//(eg. no internet connection), or whether we tried to make a request
//that violated the same-origin policy; although most of those
//are caught by other exception handlers, it can still happen here too
catch(err)
{
onfail(ERROR_NETWORK_OR_SECURITY);
}
}
//otherwise it's false so we're making a synchronous request
else
{
//we have to make sure that the response callback is outside the try..catch construct
//otherwise any errors within it, or anything it calls, will execute the catch block
//## so any coding errors made by the user inside their init callback will execute it ##
//which can lead to any genuine errors being obscured by unrelated errors fired from here
//(errors which only happen because code is being forced to execute out of context)
//so we need to make sure that the catch block doesn't call anything else
//so that if it is called out of context, nothing else happens as a result
//we only need to do this for the synchronous case,
//because asynchronous calls don't display the same behavior
//(not 100% sure why, but it's undoubtedly something to do with execution scope)
var success;
//make the request inside the try construct
//and set the success flag to true
try
{
requestobject.send(NULL_VALUE);
success = BOOLEAN_TRUE;
}
//if we have failure set the success flag to false
catch(err)
{
success = BOOLEAN_FALSE;
}
//if success is true call the response function with the request object
if(success === BOOLEAN_TRUE)
{
ajaxresponse(requestobject);
}
//otherwise pass the network or security error message to onfail
else
{
onfail(ERROR_NETWORK_OR_SECURITY);
}
}
}
}).apply(CSSUtilities);
Raw
Selector.js
/*
Selector.js, an JavaScript implementation of a CSS3 Query Selector
Copyright 2009 Henrik Lindqvist <henrik.lindqvist@llamalab.com>
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.
*/
/**
* A JavaScript implementation of a CSS3 Query Selector.
* <p>For more information, please visit
* <a href="http://www.w3.org/TR/css3-selectors/" target="_blank">http://www.w3.org/TR/css3-selectors/</a>.</p>
* @class Selector
* @version 0.6
* @author Henrik Lindqvist &lt;<a href="mailto:henrik.lindqvist@llamalab.com">henrik.lindqvist@llamalab.com</a>&gt;
*/
new function () {
/**
* Pre-compiles an Selector query.
* <p>When creating a <code>new</code> instance of the <code>Selector</code>, the query are
* pre-compiled for later execution with a call to {@link exec}.</p>
* <p>The constructor can also be called as a function, without the <code>new</code> keyword. Then
* the query is executed against the optional context element, otherwise current document.</p>
* <h4>Example:</h4>
* <pre>
* // Constructor syntax:
* new Selector('div > p').exec(document).forEach(function (e) {
* e.style.color = 'red';
* });
* // Or, shorthand syntax:
* Selector('div > p', document).forEach(function (e) {
* e.style.color = 'red';
* });
* </pre>
* @constructor Selector
* @param {string} pattern - selector pattern.
* @param {optional Object} context - document or context element.
*/
function Selector(p, c) {
if (!(this instanceof Selector)) return new Selector(p).exec(c);
//if (!qsa)
this.exec = cache[p] || (cache[p] = new compile(p));
this.pattern = p;
}
/**
* Ensure document order for group (comma) selector result.
* <p>Group (comma) queries pre-compiled when when enabled (<code>true</code>) are guaranteed to be
* in document order. Performance will be affected.</p>
* @property {static boolean} documentOrder
*/
//Selector.documentOrder = true;
Selector.prototype = {
constructor: Selector,
/**
* Execute a selector query.
* <h4>Example:</h4>
* <pre>
* new Selector('div > p').exec(document).forEach(function (e) {
* e.style.color = 'red';
* });
* </pre>
* @function {Array} exec
* @param {optional Object} context - document or context element, otherwise current document.
* @returns Non-live <code>Array</code> with matching elements.
*/
exec: function exec(c) {
var pe = this.patchElement,
pa = this.patchArray,
p = this.pattern,
r = pe ? map.call((c || d).querySelectorAll(p), pe, this) : slice.call((c || d).querySelectorAll(p));
return pa ? pa.call(this, r) : r;
},
/**
* Returns a string representing the query source pattern.
* @function {string} toString
* @returns source pattern.
*/
toString: function toString() {
return this.pattern;
},
/**
* Returns a string representing the source code of the Selector.
* @function {string} toSource
* @returns source code.
*/
toSource: function toSource() {
return 'new Selector("' + this.pattern + '")';
}
/**
* Hook for patching result Element&rsquo;s.
* <p>When using the {@link Selector} within you own framework you can add this function to
* extend the resulting <code>Element</code>&rsquo;s before they are returned by {@link exec}.
* <p>This function is not defined by default, since calling it unneccesarily affects performance.</p>
* @function {Element} patchElement
* @param {Element} e - the result element.
* @returns the patched <code>Element</code>.
* @see patchArray
*/
//patchElement : function (e) { return e; },
/**
* Hook for patching result Array.
* <p>When using the {@link Selector} within you own framework you can add this function to
* extend the resulting <code>Array</code> before it&rsquo;sthey are returned by {@link exec}.
* <p>This function is not defined by default, since calling it unneccesarily affects performance.</p>
* @function {Array} patchArray
* @param {Array} a - the result array.
* @returns the patched <code>Array</code>.
* @see patchElement
*/
//patchArray : function (a) { return a; }
};
window.Selector = Selector;
// ------------------------------------------ Private ------------------------------------------- //
function $(s) {
var a = arguments;
return s.replace(/\$(\d)/g, function (m, i) {
return a[i];
});
}
var d = document,
de = d.documentElement,
qsa = !!d.querySelectorAll,
bcn = !!d.getElementsByClassName,
cnl = !!de.children,
cnlt = cnl && de.children.tags && !wk,
ec = !!de.contains,
cdp = !!de.compareDocumentPosition,
cbp = d.createRange && d.createRange().compareBoundaryPoints,
si = typeof de.sourceIndex == 'number',
cache = {},
cmp = {
'=': 'if($1($2=="$3")){$5}',
'^=': 'if($1((x=$2)&&!x.indexOf("$3"))){$5}',
'*=': 'if($1((x=$2)&&x.indexOf("$3")!=-1)){$5}',
'$=': 'if($1((x=$2)&&x.indexOf("$3",x.length-$4)!=-1)){$5}',
'~=': 'if($1((x=$2)&&(y=x.indexOf("$3"))!=-1&&(x.charCodeAt(y-1)||32)==32&&(x.charCodeAt(y+$4)||32)==32)){$5}',
'|=': 'if($1((x=$2)&&(x=="$3"||!x.indexOf("$3-")))){$5}'
},
/*
cmp = {
'=': 'if($1($2=="$3")){$5}',
'^=': 'if($1((x=$2)&&!x.indexOf("$3"))){$5}',
'*=': 'if($1((x=$2)&&x.indexOf("$3")!=-1)){$5}',
'$=': 'if($1/$3$/.test($2)){$5}',
'~=': 'if($1/(^|\\s)$3(\\s|$)/.test($2)){$5}',
'|=': 'if($1/^$3(-|$)/.test($2)){$5}'
},
*/
slice = Array.prototype.slice,
map = Array.prototype.map || function (fn, tp) {
var i = this.length,
r = new Array(i);
while (--i >= 0) {
r[i] = fn.call(tp, this[i], i, this);
}return r;
};
Selector.guid = 0;
Selector.nthIndex = function (LLi, c, r, tp, tv) {
var p = c.parentNode,
ci = 'LLi#' + tv,
pl = 'LLi$' + tv;
if (!p) return Number.NaN;
if (!c[ci] || c.LLi != LLi) {
for (var n = p.firstChild, i = 0; n; n = n.nextSibling) {
if (n[tp] == tv) {
n[ci] = ++i;
n.LLi = LLi;
}
}
p[pl] = i;
}
return r ? 1 + p[pl] - c[ci] : c[ci];
};
Selector.srcIndex = function (h, n) {
var i = -1,
x;
do {
if (n.nodeType === 1) {
i++;
if (x = h[n.LLn]) return x + i;
}
if (x = n.previousSibling) {
do {
n = x;
} while (x = x.lastChild);
} else n = n.parentNode;
} while (n);
return i;
};
/*
Selector.srcIndex = function (h, n) {
var i = 0, x;
do {
if (x = n.previousSibling) {
n = x;
if (n.getElementsByTagName) {
if (x = h[n.LLn]) return x + i + 1;
i += n.getElementsByTagName('*').length + 1;
}
}
else if (n = n.parentNode) i++;
} while (n);
return i;
}
*/
function compile(qp) {
this.dup = this.srt = this.idx = this.i = this.nqp = 0;
var _ref4 = this;
var _ret = function (qp, nqp, i, $, srt, type, cdp, cbp, dup, idx, me, Function) {
var js = '';
do {
i = nqp = 0;
js += $('n=c;$1q:do{$2}while(false);', srt ? 's=0;' : '', type(qp, $(srt ? 'for(x=r.length;s<x;z=s+((x-s)/2)|0,($1)?s=z+1:x=z);if(s<r.length)r.splice(s++,0,$2);else r[s++]=$2;' : 'r[s++]=$2;', cdp ? 'r[z].compareDocumentPosition(n)&4' : cbp ? 'a.selectNode(r[z])||a.collapse(1)||b.selectNode(n)||b.collapse(1)||a.compareBoundaryPoints(1,b)<0' : 'h[r[z].LLn]<h[n.LLn]', 'pe?pe.call(this,n):n'), 0, '*'));
} while (qp = nqp);
js = $('var r=[],s=0,n,x,y,z,d=c?c.ownerDocument||c.document||c:c=document,pe=this.patchElement,pa=this.patchArray$1$2$3;$4return pa?pa.call(this,r):r;', dup > 0 ? ',h={}' : '', idx ? me ? ',LLi=d.LLi||(d.LLi=++Selector.guid)' : ',LLi=++Selector.guid' : '', srt && cbp && !cdp ? ',a=d.createRange(),b=d.createRange()' : '', js);
//console.log(js);
return {
v: new Function('c', js)
};
}('qp' in _ref4 ? _ref4.qp : typeof qp !== "undefined" ? qp : undefined, 'nqp' in _ref4 ? _ref4.nqp : typeof nqp !== "undefined" ? nqp : undefined, 'i' in _ref4 ? _ref4.i : typeof i !== "undefined" ? i : undefined, '$' in _ref4 ? _ref4.$ : typeof $ !== "undefined" ? $ : undefined, 'srt' in _ref4 ? _ref4.srt : typeof srt !== "undefined" ? srt : undefined, 'type' in _ref4 ? _ref4.type : typeof type !== "undefined" ? type : undefined, 'cdp' in _ref4 ? _ref4.cdp : typeof cdp !== "undefined" ? cdp : undefined, 'cbp' in _ref4 ? _ref4.cbp : typeof cbp !== "undefined" ? cbp : undefined, 'dup' in _ref4 ? _ref4.dup : typeof dup !== "undefined" ? dup : undefined, 'idx' in _ref4 ? _ref4.idx : typeof idx !== "undefined" ? idx : undefined, 'me' in _ref4 ? _ref4.me : typeof me !== "undefined" ? me : undefined, 'Function' in _ref4 ? _ref4.Function : typeof Function !== "undefined" ? Function : undefined);
if ((typeof _ret === 'undefined' ? 'undefined' : _typeof(_ret)) === "object") return _ret.v;
}
compile.prototype = {
type: function type(qp, js, n, s, c) {
var _ref5 = this;
var _ret2 = function (qp, n, c, dup, js, pred, cnlt, $, i, cnl, ie, s, String) {
var m = /^\s*([\w-]+|\*)?(.*)/.exec(qp),
t = m[1] || '*';
if (!n && c == ' ' && !dup) dup = 1;
js = pred(m[2], js, n, t, c);
switch (c) {
case '>':
return {
v: cnlt && t != '*' ? $('for(var n$1=n.children.tags("$2"),i$1=0;n=n$1[i$1++];){$3}', ++i, t, js) : $(cnl ? 'for(var n$1=n.children,i$1=0;n=n$1[i$1++];)$2{$3}' : 'for(n=n.firstChild;n;n=n.nextSibling)$2{$3}', ++i, t != '*' ? 'if(n.nodeName==="' + t.toUpperCase() + '")' : !cnl || ie ? 'if(n.nodeType===1)' : '', js)
};
case '+':
return {
v: $('while(n=n.nextSibling)if(n.node$1){$2break}else if(n.nodeType===1)break;', t == '*' ? 'Type===1' : 'Name==="' + t.toUpperCase() + '"', js)
};
case '~':
return {
v: $('while(n=n.nextSibling)if(n.node$1){$3}else if(n.node$2)break;', t == '*' ? 'Type===1' : 'Name==="' + t.toUpperCase() + '"', s == '*' ? 'Type===1' : 'Name==="' + s.toUpperCase() + '"', js)
};
default:
return {
v: (typeof js === 'undefined' ? 'undefined' : _typeof(js)) == 'object' ? String(js) // handled by pred
: n ? t == '*' ? js : $('if(n.nodeName!="$1"){$2}', t.toUpperCase(), js) : $('for(var n$1=n.getElementsByTagName("$2"),i$1=0;n=n$1[i$1++];)$3{$4}', ++i, t, ie && t == '*' ? 'if(n.nodeType===1)' : '', js)
};
}
}('qp' in _ref5 ? _ref5.qp : typeof qp !== "undefined" ? qp : undefined, 'n' in _ref5 ? _ref5.n : typeof n !== "undefined" ? n : undefined, 'c' in _ref5 ? _ref5.c : typeof c !== "undefined" ? c : undefined, 'dup' in _ref5 ? _ref5.dup : typeof dup !== "undefined" ? dup : undefined, 'js' in _ref5 ? _ref5.js : typeof js !== "undefined" ? js : undefined, 'pred' in _ref5 ? _ref5.pred : typeof pred !== "undefined" ? pred : undefined, 'cnlt' in _ref5 ? _ref5.cnlt : typeof cnlt !== "undefined" ? cnlt : undefined, '$' in _ref5 ? _ref5.$ : typeof $ !== "undefined" ? $ : undefined, 'i' in _ref5 ? _ref5.i : typeof i !== "undefined" ? i : undefined, 'cnl' in _ref5 ? _ref5.cnl : typeof cnl !== "undefined" ? cnl : undefined, 'ie' in _ref5 ? _ref5.ie : typeof ie !== "undefined" ? ie : undefined, 's' in _ref5 ? _ref5.s : typeof s !== "undefined" ? s : undefined, 'String' in _ref5 ? _ref5.String : typeof String !== "undefined" ? String : undefined);
if ((typeof _ret2 === 'undefined' ? 'undefined' : _typeof(_ret2)) === "object") return _ret2.v;
},
pred: function (_pred) {
function pred(_x, _x2, _x3, _x4, _x5) {
return _pred.apply(this, arguments);
}
pred.toString = function () {
return _pred.toString();
};
return pred;
}(function (qp, js, n, t, c) {
var _ref6 = this;
var _ret3 = function (qp, n, type, js, t, nqp, dup, srt, Selector, Error, $, cdp, cbp, si, pred, uniq, c, ie, bcn, mz, Object, i, cmp, ab, am, flo, idx) {
var m = /^([#\.])([\w-]+)(.*)/.exec(qp) || /^(\[)\s*([\w-]+)\s*(?:([~|^$*]?=)\s*(?:(['"])(.*?)\4|([\w-]+)))?\s*\](.*)/.exec(qp) || /^:(first|last|only)-(?:(child)|of-type)(.*)/.exec(qp) || /^:(nth)-(?:(last)-)?(?:(child)|of-type)\(\s*(?:(odd|even)|(-|\d*)n([+-]\d+)?|([1-9]\d*))\s*\)(.*)/.exec(qp) || /^:(active|checked|(?:dis|en)abled|empty|focus|link|root|target)(.*)/.exec(qp) || /^:(lang)\(\s*(['"])?(.*?)\2\s*\)(.*)/.exec(qp) || !n && /^:(not)\(\s*(.*)\s*\)(.*)/.exec(qp),
x = 0;
if (!m) {
if (!n && (m = /^\s*([+>~,\s])\s*(\S.*)/.exec(qp))) {
if (m[1] != ',') return {
v: type(m[2], js, n, t, m[1])
};
nqp = m[2];
dup = 2;
srt = Selector.documentOrder;
} else if (/\S/.test(qp)) throw new Error('Illegal query near: ' + qp);
return {
v: dup < 1 ? js : $('if(!h[x=n.LLn||(n.LLn=++Selector.guid)]){h[x]=$1;$2}', !srt || cdp || cbp ? 'true' : si ? 'n.sourceIndex' : 'Selector.srcIndex(h,n)', js)
};
}
if (!n && m[1] == '#' && dup != 2) dup = -1;
js = pred(m[m.length - 1], js, n, t, 1);
switch (m[1]) {
case '#':
return {
v: uniq(js, n, t, c, ie, 'n.id', '"' + m[2] + '"', 'd.getElementById("' + m[2] + '")')
};
case '.':
return {
v: bcn && !n && (!c || c == ' ') && (t == '*' || !mz) ? Object($('for(var n$1=n.getElementsByClassName("$2"),i$1=0;n=n$1[i$1++];)$3{$4}', ++i, m[2], t == '*' ? '' : 'if(n.nodeName==="' + t.toUpperCase() + '")', js)) : $(cmp['~='], n ? '!' : '', 'n.className', x = m[2], x.length, js)
};
case '[':
return {
v: (x = m[3]) ? $(cmp[x], n ? '!' : '', ie ? (x = m[2].toLowerCase()) == 'style' ? 'style.cssText.toLowerCase()' : ab[x] ? 'n.' + x + '&&"' + x + '"' : 'n.getAttribute("' + (am[x] || x) + '",2)' : 'n.getAttribute("' + m[2] + '")', x = m[5] || m[6], x.length, js) : $(ie ? 'if($1((x=n.getAttributeNode("$2"))&&x.specified)){$3}' : 'if($1n.hasAttribute("$2")){$3}', n ? '!' : '', m[2], js)
};
case 'active':
case 'focus':
return {
v: uniq(js, n, t, c, 0, 'n', 'd.activeElement')
};
case 'checked':
return {
v: $('if($1(n.checked||n.selected)){$2}', n ? '!' : '', js)
};
case 'disabled':
x = 1;
case 'enabled':
return {
v: $('if(n.disabled===$1$2){$3}', !!(x ^ n), ie ? '&&((x=n.nodeName)==="BUTTON"||x==="INPUT"||x==="OPTION"||x==="OPTGROUP"||x==="SELECT"||x==="TEXTAREA"' : '', js)
};
case 'empty':
return {
v: $('for(x=n.firstChild;x&&x.nodeType>3;x=x.nextSibling);if($1x){$2}', n ? '' : '!', js)
};
case 'first':
return {
v: flo(js, n, m[2], 'previous')
};
case 'lang':
return {
v: $(cmp['|='], n ? '!' : '', 'n.lang', x = m[3], x.length, js)
};
case 'last':
return {
v: flo(js, n, m[2], 'next')
};
case 'link':
return {
v: $('if($1(n.nodeName==="A"&&n.href)){$2}', n ? '!' : '', js)
};
case 'nth':
var a = m[4] ? 2 : m[5] == '-' ? -1 : m[7] ? 0 : m[5] ? m[5] - 0 : 1,
b = m[4] == 'odd' ? 1 : (m[6] || m[7]) - 0 || 0;
if (a == 1) return {
v: js
};
if (a == 0 && b == 1) return {
v: flo(js, n, m[3], m[2] ? 'next' : 'previous')
};
if (a == b) b = 0;
if (b < 0) b = a + b;
idx = 1;
return {
v: $('if($1(Selector.nthIndex(LLi,n,$2,"node$3",$4)$5)){$6}', n ? '!' : '', !!m[2], m[3] ? 'Type' : 'Name', m[3] ? '1' : 'n.nodeName', a < 0 ? '<=' + b : a ? '%' + a + '===' + b : '===' + b, js)
};
case 'not':
return {
v: type(m[2], js, 1, '*')
};
case 'only':
return {
v: flo(js, n, m[2])
};
case 'root':
return {
v: uniq(js, n, t, c, 0, 'n', 'd.documentElement')
};
case 'target':
x = '(d.defaultView||d.parentWindow||window).location.hash.substr(1)';
return {
v: uniq(js, n, t, c, ie, 'n.id', x, 'd.getElementById(' + x + ')')
};
}
}('qp' in _ref6 ? _ref6.qp : typeof qp !== "undefined" ? qp : undefined, 'n' in _ref6 ? _ref6.n : typeof n !== "undefined" ? n : undefined, 'type' in _ref6 ? _ref6.type : typeof type !== "undefined" ? type : undefined, 'js' in _ref6 ? _ref6.js : typeof js !== "undefined" ? js : undefined, 't' in _ref6 ? _ref6.t : typeof t !== "undefined" ? t : undefined, 'nqp' in _ref6 ? _ref6.nqp : typeof nqp !== "undefined" ? nqp : undefined, 'dup' in _ref6 ? _ref6.dup : typeof dup !== "undefined" ? dup : undefined, 'srt' in _ref6 ? _ref6.srt : typeof srt !== "undefined" ? srt : undefined, 'Selector' in _ref6 ? _ref6.Selector : typeof Selector !== "undefined" ? Selector : undefined, 'Error' in _ref6 ? _ref6.Error : typeof Error !== "undefined" ? Error : undefined, '$' in _ref6 ? _ref6.$ : typeof $ !== "undefined" ? $ : undefined, 'cdp' in _ref6 ? _ref6.cdp : typeof cdp !== "undefined" ? cdp : undefined, 'cbp' in _ref6 ? _ref6.cbp : typeof cbp !== "undefined" ? cbp : undefined, 'si' in _ref6 ? _ref6.si : typeof si !== "undefined" ? si : undefined, 'pred' in _ref6 ? _ref6.pred : typeof pred !== "undefined" ? pred : undefined, 'uniq' in _ref6 ? _ref6.uniq : typeof uniq !== "undefined" ? uniq : undefined, 'c' in _ref6 ? _ref6.c : typeof c !== "undefined" ? c : undefined, 'ie' in _ref6 ? _ref6.ie : typeof ie !== "undefined" ? ie : undefined, 'bcn' in _ref6 ? _ref6.bcn : typeof bcn !== "undefined" ? bcn : undefined, 'mz' in _ref6 ? _ref6.mz : typeof mz !== "undefined" ? mz : undefined, 'Object' in _ref6 ? _ref6.Object : typeof Object !== "undefined" ? Object : undefined, 'i' in _ref6 ? _ref6.i : typeof i !== "undefined" ? i : undefined, 'cmp' in _ref6 ? _ref6.cmp : typeof cmp !== "undefined" ? cmp : undefined, 'ab' in _ref6 ? _ref6.ab : typeof ab !== "undefined" ? ab : undefined, 'am' in _ref6 ? _ref6.am : typeof am !== "undefined" ? am : undefined, 'flo' in _ref6 ? _ref6.flo : typeof flo !== "undefined" ? flo : undefined, 'idx' in _ref6 ? _ref6.idx : typeof idx !== "undefined" ? idx : undefined);
if ((typeof _ret3 === 'undefined' ? 'undefined' : _typeof(_ret3)) === "object") return _ret3.v;
}),
uniq: function uniq(js, n, t, c, d, p, v, w) {
return n || c && c != ' ' || d ? $(n ? 'if($1!==$2){$3}' : 'if($1===$2){$3break q}', p, v, js) : Object($(ec ? 'if((x=$1)===n||!n.contains||n.contains(x))$2' : cdp ? 'if((x=$1)===n||!n.compareDocumentPosition||n.compareDocumentPosition(x)&16)$2' : 'for(x=y=$1;y;y=y.parentNode)if(y===n)$2', w || v, t == '*' ? '{n=x;' + js + 'break q}' : '{if((n=x).nodeName==="' + t.toUpperCase() + '"){' + js + '}break q}'));
},
flo: function flo(js, n, t, s) {
return $(s ? 'for($2x=n.$1Sibling;x&&x.node$3;x=x.$1Sibling);if($4x){$5}' : 'for($2(x=n.parentNode)&&(x=x.firstChild);x&&(x.node$3||x===n);x=x.nextSibling);if($4x){$5}', s, t ? '' : 'y=n.nodeName,', t ? 'Type!==1' : 'Name!==y', n ? '' : '!', js);
}
};
}();
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment