fortissimo-html/dist/dom.js

Fortissimo HTML - Flexible, Forgiving, Formatting HTML Parser

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.DomModel = exports.DomNode = exports.UnmatchedClosingTag = exports.TextElement = exports.ProcessingElement = exports.DocType = exports.DeclarationElement = exports.CommentElement = exports.CData = exports.DomElement = exports.isCommentLike = exports.CQ = exports.OQ = exports.ClosureState = void 0;
const elements_1 = require("./elements");
const characters_1 = require("./characters");
const util_1 = require("@tubular/util");
function last(array) {
    if (array && array.length > 0)
        return array[array.length - 1];
    else
        return undefined;
}
var ClosureState;
(function (ClosureState) {
    ClosureState[ClosureState["UNCLOSED"] = 0] = "UNCLOSED";
    ClosureState[ClosureState["SELF_CLOSED"] = 1] = "SELF_CLOSED";
    ClosureState[ClosureState["VOID_CLOSED"] = 2] = "VOID_CLOSED";
    ClosureState[ClosureState["EXPLICITLY_CLOSED"] = 3] = "EXPLICITLY_CLOSED";
    ClosureState[ClosureState["IMPLICITLY_CLOSED"] = 4] = "IMPLICITLY_CLOSED";
})(ClosureState = exports.ClosureState || (exports.ClosureState = {}));
// Opening quotation mark
function OQ(quote) {
    return quote.length < 2 ? quote : quote.substr(1);
}
exports.OQ = OQ;
// Closing quotation mark
function CQ(quote) {
    return quote.length < 2 ? quote : '';
}
exports.CQ = CQ;
function isCommentLike(obj) {
    return (obj instanceof CommentElement || obj instanceof DeclarationElement || obj instanceof ProcessingElement) &&
        !(obj instanceof DocType);
}
exports.isCommentLike = isCommentLike;
function stringToSelector(s) {
    const selector = {};
    const $ = /(.*)\.(.+)/.exec(s);
    if ($) {
        s = $[1];
        selector.qlass = $[2];
    }
    if (s) {
        if (s.startsWith('#'))
            selector.id = s.substr(1);
        else if (s === '*')
            selector.element = '';
        else
            selector.element = s.toLowerCase();
    }
    else
        selector.element = '';
    return selector;
}
class DomElement {
    constructor(content, line, column, terminated) {
        this.content = content;
        this.line = line;
        this.column = column;
        this.terminated = terminated;
        this.blockContext = false; // Used by formatter.ts
    }
    get depth() {
        let depth = -1;
        let node = this.parent;
        while (node) {
            depth += (node.synthetic && node.parent ? 0 : 1);
            node = node.parent;
        }
        return depth;
    }
    get syntheticDepth() {
        let depth = -1;
        let node = this.parent;
        while (node) {
            ++depth;
            node = node.parent;
        }
        return depth;
    }
    // noinspection JSUnusedGlobalSymbols
    toJSON() {
        return this.toString() + ' (' + this.depth +
            (this.line ? `; ${this.line}, ${this.column}` : '') +
            (this.parent ? '; ' + this.parent.tag : '') + ')' +
            (this.terminated ? '' : '!');
    }
}
exports.DomElement = DomElement;
class CData extends DomElement {
    toString() {
        return '<![CDATA[' + this.content + (this.terminated ? ']]>' : '');
    }
}
exports.CData = CData;
class CommentElement extends DomElement {
    toString() {
        return '<!--' + this.content + (this.terminated ? '-->' : '');
    }
}
exports.CommentElement = CommentElement;
class DeclarationElement extends DomElement {
    toString() {
        return '<!' + this.content + (this.terminated ? '>' : '');
    }
}
exports.DeclarationElement = DeclarationElement;
class DocType extends DeclarationElement {
    constructor(content, line, column, terminated) {
        super(content, line, column, terminated);
        this.type = /\bxhtml\b/i.test(content) ? 'xhtml' : 'html';
        this.variety = (/\b(frameset|strict|transitional)\b/i.exec(content.toLowerCase()) || [])[1];
        this.version = (/\bx?html[ \n\r\t\f]*([.\d]+)\b/i.exec(content) || [])[1];
        if (!this.version && /^doctype[ \n\r\t\f]+html[ \n\r\t\f]*$/i.test(content))
            this.version = '5';
    }
}
exports.DocType = DocType;
class ProcessingElement extends DomElement {
    toString() {
        return '<?' + this.content + (this.terminated ? '>' : '');
    }
}
exports.ProcessingElement = ProcessingElement;
class TextElement extends DomElement {
    constructor(content, line, column, possibleEntities) {
        super(content, line, column, true);
        this.possibleEntities = possibleEntities;
    }
    toString() {
        return this.content;
    }
}
exports.TextElement = TextElement;
class UnmatchedClosingTag extends DomElement {
    constructor(content, line, column) {
        super(content, line, column, true);
    }
    toString() {
        return this.content;
    }
}
exports.UnmatchedClosingTag = UnmatchedClosingTag;
class DomNode extends DomElement {
    constructor(tag, line = 0, column = 0, caseSensitive = false, synthetic = false) {
        super(null, line, column, true);
        this.tag = tag;
        this.attributes = [];
        this.badTerminator = null;
        this.closureState = ClosureState.UNCLOSED;
        this.endTagLine = 0;
        this.endTagColumn = 0;
        this.endTagText = '';
        this.equals = [];
        this.innerWhitespace = '';
        this.quotes = [];
        this.spacing = [];
        this.values = [];
        this.valuesLookup = {};
        this.tagLc = caseSensitive ? tag : tag.toLowerCase();
        if (synthetic)
            this.synthetic = true;
    }
    static createNode(tag) {
        const node = new DomNode(tag);
        node.setEndTag(`</${tag}>`);
        node.closureState = ClosureState.EXPLICITLY_CLOSED;
        return node;
    }
    get attributeCount() {
        return this.attributes.length;
    }
    getAttribute(nameOrIndex) {
        if ((0, util_1.isString)(nameOrIndex))
            nameOrIndex = this.attributes.indexOf(nameOrIndex);
        return [this.attributes[nameOrIndex], this.values[nameOrIndex]];
    }
    deleteAttribute(nameOrIndex) {
        if ((0, util_1.isString)(nameOrIndex))
            nameOrIndex = this.attributes.indexOf(nameOrIndex);
        if (this.attributes[nameOrIndex] === undefined)
            return false;
        delete this.valuesLookup[this.attributes[nameOrIndex]];
        this.attributes.splice(nameOrIndex, 1);
        this.values.splice(nameOrIndex, 1);
        this.equals.splice(nameOrIndex, 1);
        this.quotes.splice(nameOrIndex, 1);
        this.spacing.splice(nameOrIndex, 1);
        return true;
    }
    clearAttributes() {
        this.valuesLookup = {};
        this.attributes = [];
        this.values = [];
        this.equals = [];
        this.quotes = [];
        this.spacing = [];
    }
    // TODO: Needs more work about how raw attribute values vs. actual values will be handled.
    setAttribute(name, value = '', leadingSpace, equals, quote) {
        const index = this.attributes.indexOf(name);
        if (index < 0)
            this.addAttribute(name, value, leadingSpace, equals, quote);
        else if (value === null)
            this.deleteAttribute(index);
        else {
            this.values[index] = value;
            this.spacing[index] = leadingSpace !== null && leadingSpace !== void 0 ? leadingSpace : this.spacing[index];
            this.equals[index] = equals !== null && equals !== void 0 ? equals : this.equals[index];
            this.quotes[index] = quote !== null && quote !== void 0 ? quote : this.quotes[index];
            this.valuesLookup[name] = value;
            if (value) {
                if (!this.equals[index])
                    this.equals[index] = '=';
                if (!this.quotes[index])
                    this.quotes[index] = '"';
            }
        }
    }
    addAttribute(name, value = '', leadingSpace = ' ', equals, quote) {
        if (value === null)
            return;
        if (value === '') {
            equals = equals !== null && equals !== void 0 ? equals : '';
            quote = quote !== null && quote !== void 0 ? quote : '';
        }
        else {
            equals = equals !== null && equals !== void 0 ? equals : '=';
            quote = quote !== null && quote !== void 0 ? quote : '"';
        }
        this.attributes.push(name);
        this.values.push(value);
        this.spacing.push(leadingSpace);
        this.equals.push(equals);
        this.quotes.push(quote);
        this.valuesLookup[name] = value;
    }
    addChild(child) {
        this.children = this.children || [];
        child.parent = this;
        this.children.push(child);
    }
    remove(child) {
        if ((0, util_1.isNumber)(child))
            this.children.splice(child, 1);
        else if (child) {
            const index = this.children.indexOf(child);
            if (index >= 0)
                this.children.splice(index, 1);
        }
        else if (this.parent) {
            const index = this.parent.children.indexOf(this);
            if (index >= 0)
                this.parent.children.splice(index, 1);
        }
    }
    setEndTag(text, line = 0, column = 0) {
        this.endTagText = text;
        this.endTagLine = line;
        this.endTagColumn = column;
    }
    querySelector(selector) {
        const results = [];
        this.querySelectorImpl(selector, results, 1);
        if (results.length === 0)
            return null;
        else
            return results[0];
    }
    querySelectorAll(selector) {
        const results = [];
        this.querySelectorImpl(selector, results);
        return results;
    }
    querySelectorImpl(selector, results, limit = Number.MAX_SAFE_INTEGER) {
        if (typeof selector === 'string')
            selector = stringToSelector(selector);
        if ((!selector.element || this.tagLc === selector.element) &&
            (!selector.qlass || (this.valuesLookup.class || '').split(/\s+/).indexOf(selector.qlass) >= 0) &&
            (!selector.id || this.valuesLookup.id === selector.id))
            results.push(this);
        if (this.children) {
            for (let i = 0; i < this.children.length && results.length < limit; ++i) {
                if (this.children[i] instanceof DomNode)
                    this.children[i].querySelectorImpl(selector, results);
            }
        }
    }
    get textContent() {
        const text = [];
        if (this.children) {
            for (const child of this.children) {
                if (child instanceof CData)
                    text.push(child.content);
                else if (child instanceof TextElement)
                    text.push(child.possibleEntities ? (0, characters_1.unescapeEntities)(child.content) : child.content);
                else if (child instanceof DomNode)
                    text.push(child.textContent);
            }
        }
        return text.join('');
    }
    get innerHTML() {
        return this.toString(false);
    }
    countUnclosed() {
        let unclosed = 0;
        let implicitlyClosed = 0;
        if (!this.synthetic) {
            if (this.closureState === ClosureState.UNCLOSED)
                ++unclosed;
            else if (this.closureState === ClosureState.IMPLICITLY_CLOSED)
                ++implicitlyClosed;
        }
        if (this.children) {
            this.children.forEach(child => {
                if (child instanceof DomNode) {
                    const [childUnclosed, childImplicit] = child.countUnclosed();
                    unclosed += childUnclosed;
                    implicitlyClosed += childImplicit;
                }
            });
        }
        return [unclosed, implicitlyClosed];
    }
    toJSON() {
        const json = { tag: this.tag };
        if (this.line)
            json.line = this.line;
        if (this.column)
            json.column = this.column;
        if (this.synthetic)
            json.synthetic = true;
        if (this.badTerminator !== null)
            json.badTerminator = this.badTerminator;
        json.depth = this.depth;
        if (json.depth !== this.syntheticDepth)
            json.syntheticDepth = this.syntheticDepth;
        json.closureState = this.closureState;
        if (this.attributes.length > 0)
            json.values = this.attributes.reduce((values, attrib, index) => { values[attrib] = this.values[index]; return values; }, {});
        if (this.parent)
            json.parentTag = this.parent.tag;
        if (this.children)
            json.children = this.children;
        if (this.closureState === ClosureState.EXPLICITLY_CLOSED && this.endTagText)
            json.endTagText = `${this.endTagText} (${this.endTagLine}, ${this.endTagColumn})`;
        return json;
    }
    toString(includeSelf = true) {
        const parts = [];
        if (includeSelf && !this.synthetic) {
            parts.push('<', this.tag);
            if (this.attributes) {
                this.attributes.forEach((attrib, index) => {
                    parts.push(this.spacing[index], attrib, this.equals[index], OQ(this.quotes[index]), this.values[index], CQ(this.quotes[index]));
                });
            }
            if (this.innerWhitespace)
                parts.push(this.innerWhitespace);
            if (this.badTerminator !== null)
                parts.push(this.badTerminator);
            else if (this.closureState === ClosureState.SELF_CLOSED)
                parts.push('/>');
            else
                parts.push('>');
        }
        if (this.children)
            this.children.forEach(child => parts.push(child.toString()));
        if (includeSelf && !this.synthetic && this.closureState === ClosureState.EXPLICITLY_CLOSED && this.endTagText)
            parts.push(this.endTagText);
        return parts.join('');
    }
}
exports.DomNode = DomNode;
class DomModel {
    constructor() {
        this.root = new DomNode('/', 0, 0, false, true);
        this.currentNode = this.root;
        this.inMathOrSvg = 0;
        this.openStack = [this.root];
        this.xmlMode = false;
    }
    getRoot() {
        return this.root;
    }
    getCurrentNode() {
        return this.currentNode;
    }
    addAttribute(name, value, leadingSpace = '', equals = '=', quote = '"') {
        this.currentNode.addAttribute(name, value, leadingSpace, equals, quote);
    }
    addInnerWhitespace(whitespace) {
        if (this.currentNode)
            this.currentNode.innerWhitespace = whitespace || '';
    }
    canDoXmlMode() {
        return this.openStack.length === 1 &&
            this.root.children.length === 0 ||
            (this.root.children.length === 1 && this.root.children[0].toString().trim() === '');
    }
    getDepth() {
        return this.openStack.length - 2;
    }
    setXmlMode(mode) {
        this.xmlMode = mode;
    }
    prePush(node) {
        if (!this.xmlMode && node.tagLc in elements_1.OPEN_IMPLIES_CLOSE) {
            while (elements_1.OPEN_IMPLIES_CLOSE[node.tagLc].has(this.currentNode.tagLc)) {
                this.currentNode.closureState = ClosureState.IMPLICITLY_CLOSED;
                this.openStack.pop();
                this.updateCurrentNode();
            }
        }
    }
    addChild(child) {
        this.currentNode.addChild(child);
    }
    examineTable(table) {
        const children = table.children;
        if (!children || this.xmlMode)
            return;
        DomModel.insertRowsWhereNeeded(table);
        const sections = new Set();
        for (const elem of children) {
            if (elem instanceof DomNode) {
                if (/^(thead|tbody|tfoot)$/.test(elem.tagLc)) {
                    sections.add(elem.tagLc);
                    DomModel.insertRowsWhereNeeded(elem);
                }
            }
        }
        if (sections.size === 0)
            return;
        let section;
        for (let i = 0; i < children.length; ++i) {
            const elem = children[i];
            if (elem instanceof DomNode) {
                if (/^(thead|tbody|tfoot)$/.test(elem.tagLc))
                    section = elem.closureState === ClosureState.EXPLICITLY_CLOSED ? undefined : elem;
                else if (elem.tagLc === 'tr') {
                    const hasTh = !!elem.querySelector('th');
                    if (!section || (hasTh && (section === null || section === void 0 ? void 0 : section.tagLc) !== 'thead')) {
                        section = new DomNode(hasTh ? 'thead' : 'tbody', 0, 0, false, true);
                        section.parent = table;
                        children[i] = section;
                    }
                    section.addChild(elem);
                }
            }
        }
    }
    static insertRowsWhereNeeded(node) {
        const children = node.children;
        if (!children)
            return;
        let row;
        for (let i = 0; i < children.length; ++i) {
            const elem = children[i];
            if (elem instanceof DomNode) {
                if (elem.tagLc === 'th' || elem.tagLc === 'td') {
                    if (!row) {
                        row = new DomNode('tr', 0, 0, false, true);
                        row.parent = node;
                        children[i] = row;
                    }
                    row.addChild(elem);
                }
                else if (elem.tagLc === 'tr')
                    row = elem.closureState === ClosureState.EXPLICITLY_CLOSED ? undefined : elem;
            }
        }
    }
    push(node) {
        this.openStack.push(node);
        this.currentNode = node;
        if (node.tagLc === 'math' || node.tagLc === 'svg')
            ++this.inMathOrSvg;
    }
    pop(tagLc, endTagText = '</' + tagLc + '>', line = 0, column = 0) {
        let popped = false;
        let parseError = false;
        if (!tagLc || this.currentNode.tagLc === tagLc) {
            popped = true;
            this.openStack.pop();
            if (tagLc === null)
                this.currentNode.closureState = ClosureState.SELF_CLOSED;
            else if (tagLc === undefined)
                this.currentNode.closureState = ClosureState.VOID_CLOSED;
            else {
                this.currentNode.closureState = ClosureState.EXPLICITLY_CLOSED;
                this.currentNode.setEndTag(endTagText, line, column);
            }
            if (this.currentNode.tagLc === 'table')
                this.examineTable(this.currentNode);
        }
        if (!popped && !this.xmlMode) {
            let nodeIndex = this.openStack.map(node => node.tagLc).lastIndexOf(tagLc);
            if (nodeIndex > 0) { // No, I really don't want >= 0.
                if (elements_1.FORMATTING_ELEMENTS.has(tagLc)) {
                    for (let i = nodeIndex + 1; i < this.openStack.length; ++i) {
                        if (elements_1.MARKER_ELEMENTS.has(this.openStack[i].tagLc)) {
                            nodeIndex = -1;
                            break;
                        }
                    }
                }
                while (this.openStack.length > nodeIndex) {
                    if (!this.currentNode.closureState) {
                        if (this.openStack.length - 1 === nodeIndex) {
                            popped = true;
                            this.currentNode.closureState = ClosureState.EXPLICITLY_CLOSED;
                            this.currentNode.setEndTag(endTagText, line, column);
                        }
                        else
                            this.currentNode.closureState = ClosureState.IMPLICITLY_CLOSED;
                        if (this.currentNode.tagLc === 'table')
                            this.examineTable(this.currentNode);
                    }
                    this.openStack.pop();
                    this.updateCurrentNode();
                }
            }
        }
        if (!popped) {
            this.addChild(new UnmatchedClosingTag(endTagText, line, column));
            parseError = true;
        }
        if (this.openStack.length === 0)
            this.openStack.push(this.root);
        this.updateCurrentNode();
        this.inMathOrSvg = 0;
        this.openStack.forEach((node, index) => {
            this.inMathOrSvg += (node.tagLc === 'math' || node.tagLc === 'svg' ? 1 : 0);
            if (index > 0)
                node.parent = this.openStack[index - 1];
        });
        return !parseError;
    }
    shouldParseCData() {
        return this.xmlMode || this.inMathOrSvg > 0;
    }
    updateCurrentNode() {
        this.currentNode = last(this.openStack) || this.root;
    }
}
exports.DomModel = DomModel;
//# sourceMappingURL=dom.js.map