parse5 6.0.1 → 7.3.0

package/lib/parser/open-element-stack.js

@@ -1,482 +0,0 @@
-'use strict';
-
-const HTML = require('../common/html');
-
-//Aliases
-const $ = HTML.TAG_NAMES;
-const NS = HTML.NAMESPACES;
-
-//Element utils
-
-//OPTIMIZATION: Integer comparisons are low-cost, so we can use very fast tag name length filters here.
-//It's faster than using dictionary.
-function isImpliedEndTagRequired(tn) {
-    switch (tn.length) {
-        case 1:
-            return tn === $.P;
-
-        case 2:
-            return tn === $.RB || tn === $.RP || tn === $.RT || tn === $.DD || tn === $.DT || tn === $.LI;
-
-        case 3:
-            return tn === $.RTC;
-
-        case 6:
-            return tn === $.OPTION;
-
-        case 8:
-            return tn === $.OPTGROUP;
-    }
-
-    return false;
-}
-
-function isImpliedEndTagRequiredThoroughly(tn) {
-    switch (tn.length) {
-        case 1:
-            return tn === $.P;
-
-        case 2:
-            return (
-                tn === $.RB ||
-                tn === $.RP ||
-                tn === $.RT ||
-                tn === $.DD ||
-                tn === $.DT ||
-                tn === $.LI ||
-                tn === $.TD ||
-                tn === $.TH ||
-                tn === $.TR
-            );
-
-        case 3:
-            return tn === $.RTC;
-
-        case 5:
-            return tn === $.TBODY || tn === $.TFOOT || tn === $.THEAD;
-
-        case 6:
-            return tn === $.OPTION;
-
-        case 7:
-            return tn === $.CAPTION;
-
-        case 8:
-            return tn === $.OPTGROUP || tn === $.COLGROUP;
-    }
-
-    return false;
-}
-
-function isScopingElement(tn, ns) {
-    switch (tn.length) {
-        case 2:
-            if (tn === $.TD || tn === $.TH) {
-                return ns === NS.HTML;
-            } else if (tn === $.MI || tn === $.MO || tn === $.MN || tn === $.MS) {
-                return ns === NS.MATHML;
-            }
-
-            break;
-
-        case 4:
-            if (tn === $.HTML) {
-                return ns === NS.HTML;
-            } else if (tn === $.DESC) {
-                return ns === NS.SVG;
-            }
-
-            break;
-
-        case 5:
-            if (tn === $.TABLE) {
-                return ns === NS.HTML;
-            } else if (tn === $.MTEXT) {
-                return ns === NS.MATHML;
-            } else if (tn === $.TITLE) {
-                return ns === NS.SVG;
-            }
-
-            break;
-
-        case 6:
-            return (tn === $.APPLET || tn === $.OBJECT) && ns === NS.HTML;
-
-        case 7:
-            return (tn === $.CAPTION || tn === $.MARQUEE) && ns === NS.HTML;
-
-        case 8:
-            return tn === $.TEMPLATE && ns === NS.HTML;
-
-        case 13:
-            return tn === $.FOREIGN_OBJECT && ns === NS.SVG;
-
-        case 14:
-            return tn === $.ANNOTATION_XML && ns === NS.MATHML;
-    }
-
-    return false;
-}
-
-//Stack of open elements
-class OpenElementStack {
-    constructor(document, treeAdapter) {
-        this.stackTop = -1;
-        this.items = [];
-        this.current = document;
-        this.currentTagName = null;
-        this.currentTmplContent = null;
-        this.tmplCount = 0;
-        this.treeAdapter = treeAdapter;
-    }
-
-    //Index of element
-    _indexOf(element) {
-        let idx = -1;
-
-        for (let i = this.stackTop; i >= 0; i--) {
-            if (this.items[i] === element) {
-                idx = i;
-                break;
-            }
-        }
-        return idx;
-    }
-
-    //Update current element
-    _isInTemplate() {
-        return this.currentTagName === $.TEMPLATE && this.treeAdapter.getNamespaceURI(this.current) === NS.HTML;
-    }
-
-    _updateCurrentElement() {
-        this.current = this.items[this.stackTop];
-        this.currentTagName = this.current && this.treeAdapter.getTagName(this.current);
-
-        this.currentTmplContent = this._isInTemplate() ? this.treeAdapter.getTemplateContent(this.current) : null;
-    }
-
-    //Mutations
-    push(element) {
-        this.items[++this.stackTop] = element;
-        this._updateCurrentElement();
-
-        if (this._isInTemplate()) {
-            this.tmplCount++;
-        }
-    }
-
-    pop() {
-        this.stackTop--;
-
-        if (this.tmplCount > 0 && this._isInTemplate()) {
-            this.tmplCount--;
-        }
-
-        this._updateCurrentElement();
-    }
-
-    replace(oldElement, newElement) {
-        const idx = this._indexOf(oldElement);
-
-        this.items[idx] = newElement;
-
-        if (idx === this.stackTop) {
-            this._updateCurrentElement();
-        }
-    }
-
-    insertAfter(referenceElement, newElement) {
-        const insertionIdx = this._indexOf(referenceElement) + 1;
-
-        this.items.splice(insertionIdx, 0, newElement);
-
-        if (insertionIdx === ++this.stackTop) {
-            this._updateCurrentElement();
-        }
-    }
-
-    popUntilTagNamePopped(tagName) {
-        while (this.stackTop > -1) {
-            const tn = this.currentTagName;
-            const ns = this.treeAdapter.getNamespaceURI(this.current);
-
-            this.pop();
-
-            if (tn === tagName && ns === NS.HTML) {
-                break;
-            }
-        }
-    }
-
-    popUntilElementPopped(element) {
-        while (this.stackTop > -1) {
-            const poppedElement = this.current;
-
-            this.pop();
-
-            if (poppedElement === element) {
-                break;
-            }
-        }
-    }
-
-    popUntilNumberedHeaderPopped() {
-        while (this.stackTop > -1) {
-            const tn = this.currentTagName;
-            const ns = this.treeAdapter.getNamespaceURI(this.current);
-
-            this.pop();
-
-            if (
-                tn === $.H1 ||
-                tn === $.H2 ||
-                tn === $.H3 ||
-                tn === $.H4 ||
-                tn === $.H5 ||
-                (tn === $.H6 && ns === NS.HTML)
-            ) {
-                break;
-            }
-        }
-    }
-
-    popUntilTableCellPopped() {
-        while (this.stackTop > -1) {
-            const tn = this.currentTagName;
-            const ns = this.treeAdapter.getNamespaceURI(this.current);
-
-            this.pop();
-
-            if (tn === $.TD || (tn === $.TH && ns === NS.HTML)) {
-                break;
-            }
-        }
-    }
-
-    popAllUpToHtmlElement() {
-        //NOTE: here we assume that root <html> element is always first in the open element stack, so
-        //we perform this fast stack clean up.
-        this.stackTop = 0;
-        this._updateCurrentElement();
-    }
-
-    clearBackToTableContext() {
-        while (
-            (this.currentTagName !== $.TABLE && this.currentTagName !== $.TEMPLATE && this.currentTagName !== $.HTML) ||
-            this.treeAdapter.getNamespaceURI(this.current) !== NS.HTML
-        ) {
-            this.pop();
-        }
-    }
-
-    clearBackToTableBodyContext() {
-        while (
-            (this.currentTagName !== $.TBODY &&
-                this.currentTagName !== $.TFOOT &&
-                this.currentTagName !== $.THEAD &&
-                this.currentTagName !== $.TEMPLATE &&
-                this.currentTagName !== $.HTML) ||
-            this.treeAdapter.getNamespaceURI(this.current) !== NS.HTML
-        ) {
-            this.pop();
-        }
-    }
-
-    clearBackToTableRowContext() {
-        while (
-            (this.currentTagName !== $.TR && this.currentTagName !== $.TEMPLATE && this.currentTagName !== $.HTML) ||
-            this.treeAdapter.getNamespaceURI(this.current) !== NS.HTML
-        ) {
-            this.pop();
-        }
-    }
-
-    remove(element) {
-        for (let i = this.stackTop; i >= 0; i--) {
-            if (this.items[i] === element) {
-                this.items.splice(i, 1);
-                this.stackTop--;
-                this._updateCurrentElement();
-                break;
-            }
-        }
-    }
-
-    //Search
-    tryPeekProperlyNestedBodyElement() {
-        //Properly nested <body> element (should be second element in stack).
-        const element = this.items[1];
-
-        return element && this.treeAdapter.getTagName(element) === $.BODY ? element : null;
-    }
-
-    contains(element) {
-        return this._indexOf(element) > -1;
-    }
-
-    getCommonAncestor(element) {
-        let elementIdx = this._indexOf(element);
-
-        return --elementIdx >= 0 ? this.items[elementIdx] : null;
-    }
-
-    isRootHtmlElementCurrent() {
-        return this.stackTop === 0 && this.currentTagName === $.HTML;
-    }
-
-    //Element in scope
-    hasInScope(tagName) {
-        for (let i = this.stackTop; i >= 0; i--) {
-            const tn = this.treeAdapter.getTagName(this.items[i]);
-            const ns = this.treeAdapter.getNamespaceURI(this.items[i]);
-
-            if (tn === tagName && ns === NS.HTML) {
-                return true;
-            }
-
-            if (isScopingElement(tn, ns)) {
-                return false;
-            }
-        }
-
-        return true;
-    }
-
-    hasNumberedHeaderInScope() {
-        for (let i = this.stackTop; i >= 0; i--) {
-            const tn = this.treeAdapter.getTagName(this.items[i]);
-            const ns = this.treeAdapter.getNamespaceURI(this.items[i]);
-
-            if (
-                (tn === $.H1 || tn === $.H2 || tn === $.H3 || tn === $.H4 || tn === $.H5 || tn === $.H6) &&
-                ns === NS.HTML
-            ) {
-                return true;
-            }
-
-            if (isScopingElement(tn, ns)) {
-                return false;
-            }
-        }
-
-        return true;
-    }
-
-    hasInListItemScope(tagName) {
-        for (let i = this.stackTop; i >= 0; i--) {
-            const tn = this.treeAdapter.getTagName(this.items[i]);
-            const ns = this.treeAdapter.getNamespaceURI(this.items[i]);
-
-            if (tn === tagName && ns === NS.HTML) {
-                return true;
-            }
-
-            if (((tn === $.UL || tn === $.OL) && ns === NS.HTML) || isScopingElement(tn, ns)) {
-                return false;
-            }
-        }
-
-        return true;
-    }
-
-    hasInButtonScope(tagName) {
-        for (let i = this.stackTop; i >= 0; i--) {
-            const tn = this.treeAdapter.getTagName(this.items[i]);
-            const ns = this.treeAdapter.getNamespaceURI(this.items[i]);
-
-            if (tn === tagName && ns === NS.HTML) {
-                return true;
-            }
-
-            if ((tn === $.BUTTON && ns === NS.HTML) || isScopingElement(tn, ns)) {
-                return false;
-            }
-        }
-
-        return true;
-    }
-
-    hasInTableScope(tagName) {
-        for (let i = this.stackTop; i >= 0; i--) {
-            const tn = this.treeAdapter.getTagName(this.items[i]);
-            const ns = this.treeAdapter.getNamespaceURI(this.items[i]);
-
-            if (ns !== NS.HTML) {
-                continue;
-            }
-
-            if (tn === tagName) {
-                return true;
-            }
-
-            if (tn === $.TABLE || tn === $.TEMPLATE || tn === $.HTML) {
-                return false;
-            }
-        }
-
-        return true;
-    }
-
-    hasTableBodyContextInTableScope() {
-        for (let i = this.stackTop; i >= 0; i--) {
-            const tn = this.treeAdapter.getTagName(this.items[i]);
-            const ns = this.treeAdapter.getNamespaceURI(this.items[i]);
-
-            if (ns !== NS.HTML) {
-                continue;
-            }
-
-            if (tn === $.TBODY || tn === $.THEAD || tn === $.TFOOT) {
-                return true;
-            }
-
-            if (tn === $.TABLE || tn === $.HTML) {
-                return false;
-            }
-        }
-
-        return true;
-    }
-
-    hasInSelectScope(tagName) {
-        for (let i = this.stackTop; i >= 0; i--) {
-            const tn = this.treeAdapter.getTagName(this.items[i]);
-            const ns = this.treeAdapter.getNamespaceURI(this.items[i]);
-
-            if (ns !== NS.HTML) {
-                continue;
-            }
-
-            if (tn === tagName) {
-                return true;
-            }
-
-            if (tn !== $.OPTION && tn !== $.OPTGROUP) {
-                return false;
-            }
-        }
-
-        return true;
-    }
-
-    //Implied end tags
-    generateImpliedEndTags() {
-        while (isImpliedEndTagRequired(this.currentTagName)) {
-            this.pop();
-        }
-    }
-
-    generateImpliedEndTagsThoroughly() {
-        while (isImpliedEndTagRequiredThoroughly(this.currentTagName)) {
-            this.pop();
-        }
-    }
-
-    generateImpliedEndTagsWithExclusion(exclusionTagName) {
-        while (isImpliedEndTagRequired(this.currentTagName) && this.currentTagName !== exclusionTagName) {
-            this.pop();
-        }
-    }
-}
-
-module.exports = OpenElementStack;

package/lib/serializer/index.js

@@ -1,176 +0,0 @@
-'use strict';
-
-const defaultTreeAdapter = require('../tree-adapters/default');
-const mergeOptions = require('../utils/merge-options');
-const doctype = require('../common/doctype');
-const HTML = require('../common/html');
-
-//Aliases
-const $ = HTML.TAG_NAMES;
-const NS = HTML.NAMESPACES;
-
-//Default serializer options
-const DEFAULT_OPTIONS = {
-    treeAdapter: defaultTreeAdapter
-};
-
-//Escaping regexes
-const AMP_REGEX = /&/g;
-const NBSP_REGEX = /\u00a0/g;
-const DOUBLE_QUOTE_REGEX = /"/g;
-const LT_REGEX = /</g;
-const GT_REGEX = />/g;
-
-//Serializer
-class Serializer {
-    constructor(node, options) {
-        this.options = mergeOptions(DEFAULT_OPTIONS, options);
-        this.treeAdapter = this.options.treeAdapter;
-
-        this.html = '';
-        this.startNode = node;
-    }
-
-    //API
-    serialize() {
-        this._serializeChildNodes(this.startNode);
-
-        return this.html;
-    }
-
-    //Internals
-    _serializeChildNodes(parentNode) {
-        const childNodes = this.treeAdapter.getChildNodes(parentNode);
-
-        if (childNodes) {
-            for (let i = 0, cnLength = childNodes.length; i < cnLength; i++) {
-                const currentNode = childNodes[i];
-
-                if (this.treeAdapter.isElementNode(currentNode)) {
-                    this._serializeElement(currentNode);
-                } else if (this.treeAdapter.isTextNode(currentNode)) {
-                    this._serializeTextNode(currentNode);
-                } else if (this.treeAdapter.isCommentNode(currentNode)) {
-                    this._serializeCommentNode(currentNode);
-                } else if (this.treeAdapter.isDocumentTypeNode(currentNode)) {
-                    this._serializeDocumentTypeNode(currentNode);
-                }
-            }
-        }
-    }
-
-    _serializeElement(node) {
-        const tn = this.treeAdapter.getTagName(node);
-        const ns = this.treeAdapter.getNamespaceURI(node);
-
-        this.html += '<' + tn;
-        this._serializeAttributes(node);
-        this.html += '>';
-
-        if (
-            tn !== $.AREA &&
-            tn !== $.BASE &&
-            tn !== $.BASEFONT &&
-            tn !== $.BGSOUND &&
-            tn !== $.BR &&
-            tn !== $.COL &&
-            tn !== $.EMBED &&
-            tn !== $.FRAME &&
-            tn !== $.HR &&
-            tn !== $.IMG &&
-            tn !== $.INPUT &&
-            tn !== $.KEYGEN &&
-            tn !== $.LINK &&
-            tn !== $.META &&
-            tn !== $.PARAM &&
-            tn !== $.SOURCE &&
-            tn !== $.TRACK &&
-            tn !== $.WBR
-        ) {
-            const childNodesHolder =
-                tn === $.TEMPLATE && ns === NS.HTML ? this.treeAdapter.getTemplateContent(node) : node;
-
-            this._serializeChildNodes(childNodesHolder);
-            this.html += '</' + tn + '>';
-        }
-    }
-
-    _serializeAttributes(node) {
-        const attrs = this.treeAdapter.getAttrList(node);
-
-        for (let i = 0, attrsLength = attrs.length; i < attrsLength; i++) {
-            const attr = attrs[i];
-            const value = Serializer.escapeString(attr.value, true);
-
-            this.html += ' ';
-
-            if (!attr.namespace) {
-                this.html += attr.name;
-            } else if (attr.namespace === NS.XML) {
-                this.html += 'xml:' + attr.name;
-            } else if (attr.namespace === NS.XMLNS) {
-                if (attr.name !== 'xmlns') {
-                    this.html += 'xmlns:';
-                }
-
-                this.html += attr.name;
-            } else if (attr.namespace === NS.XLINK) {
-                this.html += 'xlink:' + attr.name;
-            } else {
-                this.html += attr.prefix + ':' + attr.name;
-            }
-
-            this.html += '="' + value + '"';
-        }
-    }
-
-    _serializeTextNode(node) {
-        const content = this.treeAdapter.getTextNodeContent(node);
-        const parent = this.treeAdapter.getParentNode(node);
-        let parentTn = void 0;
-
-        if (parent && this.treeAdapter.isElementNode(parent)) {
-            parentTn = this.treeAdapter.getTagName(parent);
-        }
-
-        if (
-            parentTn === $.STYLE ||
-            parentTn === $.SCRIPT ||
-            parentTn === $.XMP ||
-            parentTn === $.IFRAME ||
-            parentTn === $.NOEMBED ||
-            parentTn === $.NOFRAMES ||
-            parentTn === $.PLAINTEXT ||
-            parentTn === $.NOSCRIPT
-        ) {
-            this.html += content;
-        } else {
-            this.html += Serializer.escapeString(content, false);
-        }
-    }
-
-    _serializeCommentNode(node) {
-        this.html += '<!--' + this.treeAdapter.getCommentNodeContent(node) + '-->';
-    }
-
-    _serializeDocumentTypeNode(node) {
-        const name = this.treeAdapter.getDocumentTypeNodeName(node);
-
-        this.html += '<' + doctype.serializeContent(name, null, null) + '>';
-    }
-}
-
-// NOTE: used in tests and by rewriting stream
-Serializer.escapeString = function(str, attrMode) {
-    str = str.replace(AMP_REGEX, '&amp;').replace(NBSP_REGEX, '&nbsp;');
-
-    if (attrMode) {
-        str = str.replace(DOUBLE_QUOTE_REGEX, '&quot;');
-    } else {
-        str = str.replace(LT_REGEX, '&lt;').replace(GT_REGEX, '&gt;');
-    }
-
-    return str;
-};
-
-module.exports = Serializer;