/**
|
* @file domhandler
|
* @author https://github.com/fb55/domhandler
|
* @modify tanghao03 hiby
|
* @date 2018/5/17
|
*/
|
|
var ElementType = require('domelementtype');
|
|
var re_whitespace = /\s+/g;
|
var NodePrototype = require('./lib/node');
|
var ElementPrototype = require('./lib/element');
|
|
function DomHandler(callback, options, elementCB) {
|
if (typeof callback === 'object') {
|
elementCB = options;
|
options = callback;
|
callback = null;
|
} else if (typeof options === 'function') {
|
elementCB = options;
|
options = defaultOpts;
|
}
|
this._callback = callback;
|
this._options = options || defaultOpts;
|
this._elementCB = elementCB;
|
this.dom = [];
|
this._done = false;
|
this._tagStack = [];
|
this._parser = this._parser || null;
|
}
|
|
//default options
|
var defaultOpts = {
|
normalizeWhitespace: false, //Replace all whitespace with single spaces
|
withStartIndices: false, //Add startIndex properties to nodes
|
withEndIndices: false, //Add endIndex properties to nodes
|
};
|
|
DomHandler.prototype.onparserinit = function (parser) {
|
this._parser = parser;
|
};
|
|
//Resets the handler back to starting state
|
DomHandler.prototype.onreset = function () {
|
DomHandler.call(this, this._callback, this._options, this._elementCB);
|
};
|
|
//Signals the handler that parsing is done
|
DomHandler.prototype.onend = function () {
|
if (this._done) return;
|
this._done = true;
|
this._parser = null;
|
this._handleCallback(null);
|
};
|
|
DomHandler.prototype._handleCallback =
|
DomHandler.prototype.onerror = function (error) {
|
if (typeof this._callback === 'function') {
|
this._callback(error, this.dom);
|
} else {
|
if (error) throw error;
|
}
|
};
|
|
DomHandler.prototype.onclosetag = function (name, isSelfClose) {
|
//if(this._tagStack.pop().name !== name) this._handleCallback(Error('Tagname didn't match!'));
|
var elem = this._tagStack.pop();
|
elem.selfclose = !!isSelfClose;
|
if (this._options.withEndIndices && elem) {
|
elem.endIndex = this._parser.endIndex;
|
}
|
if (this._elementCB) this._elementCB(elem);
|
};
|
|
DomHandler.prototype._createDomElement = function (properties) {
|
if (!this._options.withDomLvl1) return properties;
|
|
var element;
|
if (properties.type === 'tag') {
|
element = Object.create(ElementPrototype);
|
} else {
|
element = Object.create(NodePrototype);
|
}
|
|
for (var key in properties) {
|
if (properties.hasOwnProperty(key)) {
|
element[key] = properties[key];
|
}
|
}
|
|
return element;
|
};
|
|
DomHandler.prototype._addDomElement = function (element) {
|
var parent = this._tagStack[this._tagStack.length - 1];
|
var siblings = parent ? parent.children : this.dom;
|
var previousSibling = siblings[siblings.length - 1];
|
|
// element.next = null;
|
|
if (this._options.withStartIndices) {
|
element.startIndex = this._parser.startIndex;
|
}
|
if (this._options.withEndIndices) {
|
element.endIndex = this._parser.endIndex;
|
}
|
|
// if(previousSibling){
|
// element.prev = previousSibling;
|
// previousSibling.next = element;
|
// } else {
|
// element.prev = null;
|
// }
|
|
siblings.push(element);
|
// element.parent = parent || null;
|
};
|
|
DomHandler.prototype.onopentag = function (name, attribs, singleQuoteAttribs) {
|
var properties = {
|
type: name === 'script' ? ElementType.Script : name === 'style' ? ElementType.Style : ElementType.Tag,
|
name: name,
|
attribs: attribs,
|
children: []
|
};
|
|
if (singleQuoteAttribs) {
|
properties.singleQuoteAttribs = singleQuoteAttribs;
|
}
|
|
var element = this._createDomElement(properties);
|
|
this._addDomElement(element);
|
|
this._tagStack.push(element);
|
};
|
|
DomHandler.prototype.ontext = function (data) {
|
//the ignoreWhitespace is officially dropped, but for now,
|
//it's an alias for normalizeWhitespace
|
var normalize = this._options.normalizeWhitespace || this._options.ignoreWhitespace;
|
|
var lastTag;
|
|
if (!this._tagStack.length && this.dom.length && (lastTag = this.dom[this.dom.length - 1]).type === ElementType.Text) {
|
if (normalize) {
|
lastTag.data = (lastTag.data + data).replace(re_whitespace, ' ');
|
} else {
|
lastTag.data += data;
|
}
|
} else {
|
if (
|
this._tagStack.length &&
|
(lastTag = this._tagStack[this._tagStack.length - 1]) &&
|
(lastTag = lastTag.children[lastTag.children.length - 1]) &&
|
lastTag.type === ElementType.Text
|
) {
|
if (normalize) {
|
lastTag.data = (lastTag.data + data).replace(re_whitespace, ' ');
|
} else {
|
lastTag.data += data;
|
}
|
} else {
|
if (normalize) {
|
data = data.replace(re_whitespace, ' ');
|
}
|
|
var element = this._createDomElement({
|
data: data,
|
type: ElementType.Text
|
});
|
|
this._addDomElement(element);
|
}
|
}
|
};
|
|
DomHandler.prototype.oncomment = function (data) {
|
var lastTag = this._tagStack[this._tagStack.length - 1];
|
|
if (lastTag && lastTag.type === ElementType.Comment) {
|
lastTag.data += data;
|
return;
|
}
|
|
var properties = {
|
data: data,
|
type: ElementType.Comment
|
};
|
|
var element = this._createDomElement(properties);
|
|
this._addDomElement(element);
|
this._tagStack.push(element);
|
};
|
|
DomHandler.prototype.oncdatastart = function () {
|
var properties = {
|
children: [{
|
data: '',
|
type: ElementType.Text
|
}],
|
type: ElementType.CDATA
|
};
|
|
var element = this._createDomElement(properties);
|
|
this._addDomElement(element);
|
this._tagStack.push(element);
|
};
|
|
DomHandler.prototype.oncommentend = DomHandler.prototype.oncdataend = function () {
|
this._tagStack.pop();
|
};
|
|
DomHandler.prototype.onprocessinginstruction = function (name, data) {
|
var element = this._createDomElement({
|
name: name,
|
data: data,
|
type: ElementType.Directive
|
});
|
|
this._addDomElement(element);
|
};
|
|
module.exports = DomHandler;
|
