var Tokenizer_js_1 = require('./Tokenizer.js'); var decode_js_1 = require('./entities/decode.js'); var formTags = new Set(['input', 'option', 'optgroup', 'select', 'button', 'datalist', 'textarea']); var pTag = new Set(['p']); var tableSectionTags = new Set(['thead', 'tbody']); var ddtTags = new Set(['dd', 'dt']); var rtpTags = new Set(['rt', 'rp']); var openImpliesClose = new Map([ ['tr', new Set(['tr', 'th', 'td'])], ['th', new Set(['th'])], ['td', new Set(['thead', 'th', 'td'])], ['body', new Set(['head', 'link', 'script'])], ['li', new Set(['li'])], ['p', pTag], ['h1', pTag], ['h2', pTag], ['h3', pTag], ['h4', pTag], ['h5', pTag], ['h6', pTag], ['select', formTags], ['input', formTags], ['output', formTags], ['button', formTags], ['datalist', formTags], ['textarea', formTags], ['option', new Set(['option'])], ['optgroup', new Set(['optgroup', 'option'])], ['dd', ddtTags], ['dt', ddtTags], ['address', pTag], ['article', pTag], ['aside', pTag], ['blockquote', pTag], ['details', pTag], ['div', pTag], ['dl', pTag], ['fieldset', pTag], ['figcaption', pTag], ['figure', pTag], ['footer', pTag], ['form', pTag], ['header', pTag], ['hr', pTag], ['main', pTag], ['nav', pTag], ['ol', pTag], ['pre', pTag], ['section', pTag], ['table', pTag], ['ul', pTag], ['rt', rtpTags], ['rp', rtpTags], ['tbody', tableSectionTags], ['tfoot', tableSectionTags] ]); var voidElements = new Set([ 'area', 'base', 'basefont', 'br', 'col', 'command', 'embed', 'frame', 'hr', 'img', 'input', 'isindex', 'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr' ]); var foreignContextElements = new Set(['math', 'svg']); var htmlIntegrationElements = new Set(['mi', 'mo', 'mn', 'ms', 'mtext', 'annotation-xml', 'foreignobject', 'desc', 'title']); var reNameEnd = /\s|\//; var Parser = /** @class */ (function () { function Parser(cbs, options) { if (options === void 0) { options = {}; } var _a; var _b; var _c; var _d; var _e; this.options = options; /** The start index of the last event. */ this.startIndex = 0; /** The end index of the last event. */ this.endIndex = 0; /** * Store the start index of the current open tag, * so we can update the start index for attributes. */ this.openTagStart = 0; this.tagname = ''; this.attribname = ''; this.attribvalue = ''; this.attribs = null; this.stack = []; this.foreignContext = []; this.buffers = []; this.bufferOffset = 0; /** The index of the last written buffer. Used when resuming after a `pause()`. */ this.writeIndex = 0; /** Indicates whether the parser has finished running / `.end` has been called. */ this.ended = false; this.cbs = cbs !== null && cbs !== void 0 ? cbs : {}; this.lowerCaseTagNames = (_a = options.lowerCaseTags) !== null && _a !== void 0 ? _a : !options.xmlMode; this.lowerCaseAttributeNames = (_b = options.lowerCaseAttributeNames) !== null && _b !== void 0 ? _b : !options.xmlMode; this.tokenizer = new ((_c = options.Tokenizer) !== null && _c !== void 0 ? _c : Tokenizer_js_1.default)(this.options, this); (_e = (_d = this.cbs).onparserinit) === null || _e === void 0 ? void 0 : _e.call(_d, this); } // Tokenizer event handlers /** @internal */ Parser.prototype.ontext = function (start, endIndex) { var _a; var _b; var data = this.getSlice(start, endIndex); this.endIndex = endIndex - 1; (_b = (_a = this.cbs).ontext) === null || _b === void 0 ? void 0 : _b.call(_a, data); this.startIndex = endIndex; }; /** @internal */ Parser.prototype.ontextentity = function (cp) { var _a; var _b; /* * Entities can be emitted on the character, or directly after. * We use the section start here to get accurate indices. */ var idx = this.tokenizer.getSectionStart(); this.endIndex = idx - 1; (_b = (_a = this.cbs).ontext) === null || _b === void 0 ? void 0 : _b.call(_a, (0, decode_js_1.fromCodePoint)(cp)); this.startIndex = idx; }; Parser.prototype.isVoidElement = function (name) { return !this.options.xmlMode && voidElements.has(name); }; /** @internal */ Parser.prototype.onopentagname = function (start, endIndex) { this.endIndex = endIndex; var name = this.getSlice(start, endIndex); if (this.lowerCaseTagNames) { name = name.toLowerCase(); } this.emitOpenTag(name); }; Parser.prototype.emitOpenTag = function (name) { var _a; var _b; var _c; var _d; this.openTagStart = this.startIndex; this.tagname = name; var impliesClose = !this.options.xmlMode && openImpliesClose.get(name); if (impliesClose) { while (this.stack.length > 0 && impliesClose.has(this.stack[this.stack.length - 1])) { var el = this.stack.pop(); (_b = (_a = this.cbs).onclosetag) === null || _b === void 0 ? void 0 : _b.call(_a, el, true); } } if (!this.isVoidElement(name)) { this.stack.push(name); if (foreignContextElements.has(name)) { this.foreignContext.push(true); } else if (htmlIntegrationElements.has(name)) { this.foreignContext.push(false); } } (_d = (_c = this.cbs).onopentagname) === null || _d === void 0 ? void 0 : _d.call(_c, name); if (this.cbs.onopentag) { this.attribs = {}; } }; Parser.prototype.endOpenTag = function (isImplied) { var _a; var _b; this.startIndex = this.openTagStart; if (this.attribs) { (_b = (_a = this.cbs).onopentag) === null || _b === void 0 ? void 0 : _b.call(_a, this.tagname, this.attribs, isImplied); this.attribs = null; } if (this.cbs.onclosetag && this.isVoidElement(this.tagname)) { this.cbs.onclosetag(this.tagname, true); } this.tagname = ''; }; /** @internal */ Parser.prototype.onopentagend = function (endIndex) { this.endIndex = endIndex; this.endOpenTag(false); // Set `startIndex` for next node this.startIndex = endIndex + 1; }; /** @internal */ Parser.prototype.onclosetag = function (start, endIndex) { var _a; var _b; var _c; var _d; var _e; var _f; this.endIndex = endIndex; var name = this.getSlice(start, endIndex); if (this.lowerCaseTagNames) { name = name.toLowerCase(); } if (foreignContextElements.has(name) || htmlIntegrationElements.has(name)) { this.foreignContext.pop(); } if (!this.isVoidElement(name)) { var pos = this.stack.lastIndexOf(name); if (pos !== -1) { if (this.cbs.onclosetag) { var count = this.stack.length - pos; while (count--) { // We know the stack has sufficient elements. this.cbs.onclosetag(this.stack.pop(), count !== 0); } } else { this.stack.length = pos; } } else if (!this.options.xmlMode && name === 'p') { // Implicit open before close this.emitOpenTag('p'); this.closeCurrentTag(true); } } else if (!this.options.xmlMode && name === 'br') { // We can't use `emitOpenTag` for implicit open, as `br` would be implicitly closed. (_b = (_a = this.cbs).onopentagname) === null || _b === void 0 ? void 0 : _b.call(_a, 'br'); (_d = (_c = this.cbs).onopentag) === null || _d === void 0 ? void 0 : _d.call(_c, 'br', {}, true); (_f = (_e = this.cbs).onclosetag) === null || _f === void 0 ? void 0 : _f.call(_e, 'br', false); } // Set `startIndex` for next node this.startIndex = endIndex + 1; }; /** @internal */ Parser.prototype.onselfclosingtag = function (endIndex) { this.endIndex = endIndex; if (this.options.xmlMode || this.options.recognizeSelfClosing || this.foreignContext[this.foreignContext.length - 1]) { this.closeCurrentTag(false); // Set `startIndex` for next node this.startIndex = endIndex + 1; } else { // Ignore the fact that the tag is self-closing. this.onopentagend(endIndex); } }; Parser.prototype.closeCurrentTag = function (isOpenImplied) { var _a; var _b; var name = this.tagname; this.endOpenTag(isOpenImplied); // Self-closing tags will be on the top of the stack if (this.stack[this.stack.length - 1] === name) { // If the opening tag isn't implied, the closing tag has to be implied. (_b = (_a = this.cbs).onclosetag) === null || _b === void 0 ? void 0 : _b.call(_a, name, !isOpenImplied); this.stack.pop(); } }; /** @internal */ Parser.prototype.onattribname = function (start, endIndex) { this.startIndex = start; var name = this.getSlice(start, endIndex); this.attribname = this.lowerCaseAttributeNames ? name.toLowerCase() : name; }; /** @internal */ Parser.prototype.onattribdata = function (start, endIndex) { this.attribvalue += this.getSlice(start, endIndex); }; /** @internal */ Parser.prototype.onattribentity = function (cp) { this.attribvalue += (0, decode_js_1.fromCodePoint)(cp); }; /** @internal */ Parser.prototype.onattribend = function (quote, endIndex) { var _a; var _b; this.endIndex = endIndex; (_b = (_a = this.cbs).onattribute) === null || _b === void 0 ? void 0 : _b.call( _a, this.attribname, this.attribvalue, quote === Tokenizer_js_1.QuoteType.Double ? '"' : quote === Tokenizer_js_1.QuoteType.Single ? "'" : quote === Tokenizer_js_1.QuoteType.NoValue ? undefined : null ); if (this.attribs && !Object.prototype.hasOwnProperty.call(this.attribs, this.attribname)) { this.attribs[this.attribname] = this.attribvalue; } this.attribvalue = ''; }; Parser.prototype.getInstructionName = function (value) { var idx = value.search(reNameEnd); var name = idx < 0 ? value : value.substr(0, idx); if (this.lowerCaseTagNames) { name = name.toLowerCase(); } return name; }; /** @internal */ Parser.prototype.ondeclaration = function (start, endIndex) { this.endIndex = endIndex; var value = this.getSlice(start, endIndex); if (this.cbs.onprocessinginstruction) { var name = this.getInstructionName(value); this.cbs.onprocessinginstruction('!'.concat(name), '!'.concat(value)); } // Set `startIndex` for next node this.startIndex = endIndex + 1; }; /** @internal */ Parser.prototype.onprocessinginstruction = function (start, endIndex) { this.endIndex = endIndex; var value = this.getSlice(start, endIndex); if (this.cbs.onprocessinginstruction) { var name = this.getInstructionName(value); this.cbs.onprocessinginstruction('?'.concat(name), '?'.concat(value)); } // Set `startIndex` for next node this.startIndex = endIndex + 1; }; /** @internal */ Parser.prototype.oncomment = function (start, endIndex, offset) { var _a; var _b; var _c; var _d; this.endIndex = endIndex; (_b = (_a = this.cbs).oncomment) === null || _b === void 0 ? void 0 : _b.call(_a, this.getSlice(start, endIndex - offset)); (_d = (_c = this.cbs).oncommentend) === null || _d === void 0 ? void 0 : _d.call(_c); // Set `startIndex` for next node this.startIndex = endIndex + 1; }; /** @internal */ Parser.prototype.oncdata = function (start, endIndex, offset) { var _a; var _b; var _c; var _d; var _e; var _f; var _g; var _h; var _j; var _k; this.endIndex = endIndex; var value = this.getSlice(start, endIndex - offset); if (this.options.xmlMode || this.options.recognizeCDATA) { (_b = (_a = this.cbs).oncdatastart) === null || _b === void 0 ? void 0 : _b.call(_a); (_d = (_c = this.cbs).ontext) === null || _d === void 0 ? void 0 : _d.call(_c, value); (_f = (_e = this.cbs).oncdataend) === null || _f === void 0 ? void 0 : _f.call(_e); } else { (_h = (_g = this.cbs).oncomment) === null || _h === void 0 ? void 0 : _h.call(_g, '[CDATA['.concat(value, ']]')); (_k = (_j = this.cbs).oncommentend) === null || _k === void 0 ? void 0 : _k.call(_j); } // Set `startIndex` for next node this.startIndex = endIndex + 1; }; /** @internal */ Parser.prototype.onend = function () { var _a; var _b; if (this.cbs.onclosetag) { // Set the end index for all remaining tags this.endIndex = this.startIndex; for (var i = this.stack.length; i > 0; this.cbs.onclosetag(this.stack[--i], true)) {} } (_b = (_a = this.cbs).onend) === null || _b === void 0 ? void 0 : _b.call(_a); }; /** * Resets the parser to a blank state, ready to parse a new HTML document */ Parser.prototype.reset = function () { var _a; var _b; var _c; var _d; (_b = (_a = this.cbs).onreset) === null || _b === void 0 ? void 0 : _b.call(_a); this.tokenizer.reset(); this.tagname = ''; this.attribname = ''; this.attribs = null; this.stack.length = 0; this.startIndex = 0; this.endIndex = 0; (_d = (_c = this.cbs).onparserinit) === null || _d === void 0 ? void 0 : _d.call(_c, this); this.buffers.length = 0; this.bufferOffset = 0; this.writeIndex = 0; this.ended = false; }; /** * Resets the parser, then parses a complete document and * pushes it to the handler. * * @param data Document to parse. */ Parser.prototype.parseComplete = function (data) { this.reset(); this.end(data); }; Parser.prototype.getSlice = function (start, end) { while (start - this.bufferOffset >= this.buffers[0].length) { this.shiftBuffer(); } var str = this.buffers[0].slice(start - this.bufferOffset, end - this.bufferOffset); while (end - this.bufferOffset > this.buffers[0].length) { this.shiftBuffer(); str += this.buffers[0].slice(0, end - this.bufferOffset); } return str; }; Parser.prototype.shiftBuffer = function () { this.bufferOffset += this.buffers[0].length; this.writeIndex--; this.buffers.shift(); }; /** * Parses a chunk of data and calls the corresponding callbacks. * * @param chunk Chunk to parse. */ Parser.prototype.write = function (chunk) { var _a; var _b; if (this.ended) { (_b = (_a = this.cbs).onerror) === null || _b === void 0 ? void 0 : _b.call(_a, new Error('.write() after done!')); return; } this.buffers.push(chunk); if (this.tokenizer.running) { this.tokenizer.write(chunk); this.writeIndex++; } }; /** * Parses the end of the buffer and clears the stack, calls onend. * * @param chunk Optional final chunk to parse. */ Parser.prototype.end = function (chunk) { var _a; var _b; if (this.ended) { (_b = (_a = this.cbs).onerror) === null || _b === void 0 ? void 0 : _b.call(_a, Error('.end() after done!')); return; } if (chunk) { this.write(chunk); } this.ended = true; this.tokenizer.end(); }; /** * Pauses parsing. The parser won't emit events until `resume` is called. */ Parser.prototype.pause = function () { this.tokenizer.pause(); }; /** * Resumes parsing after `pause` was called. */ Parser.prototype.resume = function () { this.tokenizer.resume(); while (this.tokenizer.running && this.writeIndex < this.buffers.length) { this.tokenizer.write(this.buffers[this.writeIndex++]); } if (this.ended) { this.tokenizer.end(); } }; /** * Alias of `write`, for backwards compatibility. * * @param chunk Chunk to parse. * @deprecated */ Parser.prototype.parseChunk = function (chunk) { this.write(chunk); }; /** * Alias of `end`, for backwards compatibility. * * @param chunk Optional final chunk to parse. * @deprecated */ Parser.prototype.done = function (chunk) { this.end(chunk); }; return Parser; })(); module.exports = Parser;