parse5 3.0.3 → 5.1.0

package/lib/parser/formatting_element_list.js

@@ -1,167 +0,0 @@
-'use strict';
-
-//Const
-var NOAH_ARK_CAPACITY = 3;
-
-//List of formatting elements
-var FormattingElementList = module.exports = function (treeAdapter) {
-    this.length = 0;
-    this.entries = [];
-    this.treeAdapter = treeAdapter;
-    this.bookmark = null;
-};
-
-//Entry types
-FormattingElementList.MARKER_ENTRY = 'MARKER_ENTRY';
-FormattingElementList.ELEMENT_ENTRY = 'ELEMENT_ENTRY';
-
-//Noah Ark's condition
-//OPTIMIZATION: at first we try to find possible candidates for exclusion using
-//lightweight heuristics without thorough attributes check.
-FormattingElementList.prototype._getNoahArkConditionCandidates = function (newElement) {
-    var candidates = [];
-
-    if (this.length >= NOAH_ARK_CAPACITY) {
-        var neAttrsLength = this.treeAdapter.getAttrList(newElement).length,
-            neTagName = this.treeAdapter.getTagName(newElement),
-            neNamespaceURI = this.treeAdapter.getNamespaceURI(newElement);
-
-        for (var i = this.length - 1; i >= 0; i--) {
-            var entry = this.entries[i];
-
-            if (entry.type === FormattingElementList.MARKER_ENTRY)
-                break;
-
-            var element = entry.element,
-                elementAttrs = this.treeAdapter.getAttrList(element),
-                isCandidate = this.treeAdapter.getTagName(element) === neTagName &&
-                              this.treeAdapter.getNamespaceURI(element) === neNamespaceURI &&
-                              elementAttrs.length === neAttrsLength;
-
-            if (isCandidate)
-                candidates.push({idx: i, attrs: elementAttrs});
-        }
-    }
-
-    return candidates.length < NOAH_ARK_CAPACITY ? [] : candidates;
-};
-
-FormattingElementList.prototype._ensureNoahArkCondition = function (newElement) {
-    var candidates = this._getNoahArkConditionCandidates(newElement),
-        cLength = candidates.length;
-
-    if (cLength) {
-        var neAttrs = this.treeAdapter.getAttrList(newElement),
-            neAttrsLength = neAttrs.length,
-            neAttrsMap = Object.create(null);
-
-        //NOTE: build attrs map for the new element so we can perform fast lookups
-        for (var i = 0; i < neAttrsLength; i++) {
-            var neAttr = neAttrs[i];
-
-            neAttrsMap[neAttr.name] = neAttr.value;
-        }
-
-        for (i = 0; i < neAttrsLength; i++) {
-            for (var j = 0; j < cLength; j++) {
-                var cAttr = candidates[j].attrs[i];
-
-                if (neAttrsMap[cAttr.name] !== cAttr.value) {
-                    candidates.splice(j, 1);
-                    cLength--;
-                }
-
-                if (candidates.length < NOAH_ARK_CAPACITY)
-                    return;
-            }
-        }
-
-        //NOTE: remove bottommost candidates until Noah's Ark condition will not be met
-        for (i = cLength - 1; i >= NOAH_ARK_CAPACITY - 1; i--) {
-            this.entries.splice(candidates[i].idx, 1);
-            this.length--;
-        }
-    }
-};
-
-//Mutations
-FormattingElementList.prototype.insertMarker = function () {
-    this.entries.push({type: FormattingElementList.MARKER_ENTRY});
-    this.length++;
-};
-
-FormattingElementList.prototype.pushElement = function (element, token) {
-    this._ensureNoahArkCondition(element);
-
-    this.entries.push({
-        type: FormattingElementList.ELEMENT_ENTRY,
-        element: element,
-        token: token
-    });
-
-    this.length++;
-};
-
-FormattingElementList.prototype.insertElementAfterBookmark = function (element, token) {
-    var bookmarkIdx = this.length - 1;
-
-    for (; bookmarkIdx >= 0; bookmarkIdx--) {
-        if (this.entries[bookmarkIdx] === this.bookmark)
-            break;
-    }
-
-    this.entries.splice(bookmarkIdx + 1, 0, {
-        type: FormattingElementList.ELEMENT_ENTRY,
-        element: element,
-        token: token
-    });
-
-    this.length++;
-};
-
-FormattingElementList.prototype.removeEntry = function (entry) {
-    for (var i = this.length - 1; i >= 0; i--) {
-        if (this.entries[i] === entry) {
-            this.entries.splice(i, 1);
-            this.length--;
-            break;
-        }
-    }
-};
-
-FormattingElementList.prototype.clearToLastMarker = function () {
-    while (this.length) {
-        var entry = this.entries.pop();
-
-        this.length--;
-
-        if (entry.type === FormattingElementList.MARKER_ENTRY)
-            break;
-    }
-};
-
-//Search
-FormattingElementList.prototype.getElementEntryInScopeWithTagName = function (tagName) {
-    for (var i = this.length - 1; i >= 0; i--) {
-        var entry = this.entries[i];
-
-        if (entry.type === FormattingElementList.MARKER_ENTRY)
-            return null;
-
-        if (this.treeAdapter.getTagName(entry.element) === tagName)
-            return entry;
-    }
-
-    return null;
-};
-
-FormattingElementList.prototype.getElementEntry = function (element) {
-    for (var i = this.length - 1; i >= 0; i--) {
-        var entry = this.entries[i];
-
-        if (entry.type === FormattingElementList.ELEMENT_ENTRY && entry.element === element)
-            return entry;
-    }
-
-    return null;
-};

package/lib/parser/open_element_stack.js

@@ -1,395 +0,0 @@
-'use strict';
-
-var HTML = require('../common/html');
-
-//Aliases
-var $ = HTML.TAG_NAMES,
-    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 || tn === $.MENUITEM;
-    }
-
-    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
-var OpenElementStack = module.exports = function (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
-OpenElementStack.prototype._indexOf = function (element) {
-    var idx = -1;
-
-    for (var i = this.stackTop; i >= 0; i--) {
-        if (this.items[i] === element) {
-            idx = i;
-            break;
-        }
-    }
-    return idx;
-};
-
-//Update current element
-OpenElementStack.prototype._isInTemplate = function () {
-    return this.currentTagName === $.TEMPLATE && this.treeAdapter.getNamespaceURI(this.current) === NS.HTML;
-};
-
-OpenElementStack.prototype._updateCurrentElement = function () {
-    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
-OpenElementStack.prototype.push = function (element) {
-    this.items[++this.stackTop] = element;
-    this._updateCurrentElement();
-
-    if (this._isInTemplate())
-        this.tmplCount++;
-
-};
-
-OpenElementStack.prototype.pop = function () {
-    this.stackTop--;
-
-    if (this.tmplCount > 0 && this._isInTemplate())
-        this.tmplCount--;
-
-    this._updateCurrentElement();
-};
-
-OpenElementStack.prototype.replace = function (oldElement, newElement) {
-    var idx = this._indexOf(oldElement);
-
-    this.items[idx] = newElement;
-
-    if (idx === this.stackTop)
-        this._updateCurrentElement();
-};
-
-OpenElementStack.prototype.insertAfter = function (referenceElement, newElement) {
-    var insertionIdx = this._indexOf(referenceElement) + 1;
-
-    this.items.splice(insertionIdx, 0, newElement);
-
-    if (insertionIdx === ++this.stackTop)
-        this._updateCurrentElement();
-};
-
-OpenElementStack.prototype.popUntilTagNamePopped = function (tagName) {
-    while (this.stackTop > -1) {
-        var tn = this.currentTagName,
-            ns = this.treeAdapter.getNamespaceURI(this.current);
-
-        this.pop();
-
-        if (tn === tagName && ns === NS.HTML)
-            break;
-    }
-};
-
-OpenElementStack.prototype.popUntilElementPopped = function (element) {
-    while (this.stackTop > -1) {
-        var poppedElement = this.current;
-
-        this.pop();
-
-        if (poppedElement === element)
-            break;
-    }
-};
-
-OpenElementStack.prototype.popUntilNumberedHeaderPopped = function () {
-    while (this.stackTop > -1) {
-        var tn = this.currentTagName,
-            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;
-    }
-};
-
-OpenElementStack.prototype.popUntilTableCellPopped = function () {
-    while (this.stackTop > -1) {
-        var tn = this.currentTagName,
-            ns = this.treeAdapter.getNamespaceURI(this.current);
-
-        this.pop();
-
-        if (tn === $.TD || tn === $.TH && ns === NS.HTML)
-            break;
-    }
-};
-
-OpenElementStack.prototype.popAllUpToHtmlElement = function () {
-    //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();
-};
-
-OpenElementStack.prototype.clearBackToTableContext = function () {
-    while (this.currentTagName !== $.TABLE &&
-           this.currentTagName !== $.TEMPLATE &&
-           this.currentTagName !== $.HTML ||
-           this.treeAdapter.getNamespaceURI(this.current) !== NS.HTML)
-        this.pop();
-};
-
-OpenElementStack.prototype.clearBackToTableBodyContext = function () {
-    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();
-};
-
-OpenElementStack.prototype.clearBackToTableRowContext = function () {
-    while (this.currentTagName !== $.TR &&
-           this.currentTagName !== $.TEMPLATE &&
-           this.currentTagName !== $.HTML ||
-           this.treeAdapter.getNamespaceURI(this.current) !== NS.HTML)
-        this.pop();
-};
-
-OpenElementStack.prototype.remove = function (element) {
-    for (var i = this.stackTop; i >= 0; i--) {
-        if (this.items[i] === element) {
-            this.items.splice(i, 1);
-            this.stackTop--;
-            this._updateCurrentElement();
-            break;
-        }
-    }
-};
-
-//Search
-OpenElementStack.prototype.tryPeekProperlyNestedBodyElement = function () {
-    //Properly nested <body> element (should be second element in stack).
-    var element = this.items[1];
-
-    return element && this.treeAdapter.getTagName(element) === $.BODY ? element : null;
-};
-
-OpenElementStack.prototype.contains = function (element) {
-    return this._indexOf(element) > -1;
-};
-
-OpenElementStack.prototype.getCommonAncestor = function (element) {
-    var elementIdx = this._indexOf(element);
-
-    return --elementIdx >= 0 ? this.items[elementIdx] : null;
-};
-
-OpenElementStack.prototype.isRootHtmlElementCurrent = function () {
-    return this.stackTop === 0 && this.currentTagName === $.HTML;
-};
-
-//Element in scope
-OpenElementStack.prototype.hasInScope = function (tagName) {
-    for (var i = this.stackTop; i >= 0; i--) {
-        var tn = this.treeAdapter.getTagName(this.items[i]),
-            ns = this.treeAdapter.getNamespaceURI(this.items[i]);
-
-        if (tn === tagName && ns === NS.HTML)
-            return true;
-
-        if (isScopingElement(tn, ns))
-            return false;
-    }
-
-    return true;
-};
-
-OpenElementStack.prototype.hasNumberedHeaderInScope = function () {
-    for (var i = this.stackTop; i >= 0; i--) {
-        var tn = this.treeAdapter.getTagName(this.items[i]),
-            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;
-};
-
-OpenElementStack.prototype.hasInListItemScope = function (tagName) {
-    for (var i = this.stackTop; i >= 0; i--) {
-        var tn = this.treeAdapter.getTagName(this.items[i]),
-            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;
-};
-
-OpenElementStack.prototype.hasInButtonScope = function (tagName) {
-    for (var i = this.stackTop; i >= 0; i--) {
-        var tn = this.treeAdapter.getTagName(this.items[i]),
-            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;
-};
-
-OpenElementStack.prototype.hasInTableScope = function (tagName) {
-    for (var i = this.stackTop; i >= 0; i--) {
-        var tn = this.treeAdapter.getTagName(this.items[i]),
-            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;
-};
-
-OpenElementStack.prototype.hasTableBodyContextInTableScope = function () {
-    for (var i = this.stackTop; i >= 0; i--) {
-        var tn = this.treeAdapter.getTagName(this.items[i]),
-            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;
-};
-
-OpenElementStack.prototype.hasInSelectScope = function (tagName) {
-    for (var i = this.stackTop; i >= 0; i--) {
-        var tn = this.treeAdapter.getTagName(this.items[i]),
-            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
-OpenElementStack.prototype.generateImpliedEndTags = function () {
-    while (isImpliedEndTagRequired(this.currentTagName))
-        this.pop();
-};
-
-OpenElementStack.prototype.generateImpliedEndTagsWithExclusion = function (exclusionTagName) {
-    while (isImpliedEndTagRequired(this.currentTagName) && this.currentTagName !== exclusionTagName)
-        this.pop();
-};

package/lib/parser/parser_stream.js

@@ -1,76 +0,0 @@
-'use strict';
-
-var WritableStream = require('stream').Writable,
-    inherits = require('util').inherits,
-    Parser = require('./index');
-
-var ParserStream = module.exports = function (options) {
-    WritableStream.call(this);
-
-    this.parser = new Parser(options);
-
-    this.lastChunkWritten = false;
-    this.writeCallback = null;
-    this.pausedByScript = false;
-
-    this.document = this.parser.treeAdapter.createDocument();
-
-    this.pendingHtmlInsertions = [];
-
-    this._resume = this._resume.bind(this);
-    this._documentWrite = this._documentWrite.bind(this);
-    this._scriptHandler = this._scriptHandler.bind(this);
-
-    this.parser._bootstrap(this.document, null);
-};
-
-inherits(ParserStream, WritableStream);
-
-//WritableStream implementation
-ParserStream.prototype._write = function (chunk, encoding, callback) {
-    this.writeCallback = callback;
-    this.parser.tokenizer.write(chunk.toString('utf8'), this.lastChunkWritten);
-    this._runParsingLoop();
-};
-
-ParserStream.prototype.end = function (chunk, encoding, callback) {
-    this.lastChunkWritten = true;
-    WritableStream.prototype.end.call(this, chunk || '', encoding, callback);
-};
-
-//Scriptable parser implementation
-ParserStream.prototype._runParsingLoop = function () {
-    this.parser.runParsingLoopForCurrentChunk(this.writeCallback, this._scriptHandler);
-};
-
-ParserStream.prototype._resume = function () {
-    if (!this.pausedByScript)
-        throw new Error('Parser was already resumed');
-
-    while (this.pendingHtmlInsertions.length) {
-        var html = this.pendingHtmlInsertions.pop();
-
-        this.parser.tokenizer.insertHtmlAtCurrentPos(html);
-    }
-
-    this.pausedByScript = false;
-
-    //NOTE: keep parsing if we don't wait for the next input chunk
-    if (this.parser.tokenizer.active)
-        this._runParsingLoop();
-};
-
-ParserStream.prototype._documentWrite = function (html) {
-    if (!this.parser.stopped)
-        this.pendingHtmlInsertions.push(html);
-};
-
-ParserStream.prototype._scriptHandler = function (scriptElement) {
-    if (this.listeners('script').length) {
-        this.pausedByScript = true;
-        this.emit('script', scriptElement, this._documentWrite, this._resume);
-    }
-    else
-        this._runParsingLoop();
-};
-

package/lib/parser/plain_text_conversion_stream.js

@@ -1,20 +0,0 @@
-'use strict';
-
-var ParserStream = require('./parser_stream'),
-    inherits = require('util').inherits,
-    $ = require('../common/html').TAG_NAMES;
-
-var PlainTextConversionStream = module.exports = function (options) {
-    ParserStream.call(this, options);
-
-    // NOTE: see https://html.spec.whatwg.org/#read-text
-    this.parser._insertFakeElement($.HTML);
-    this.parser._insertFakeElement($.HEAD);
-    this.parser.openElements.pop();
-    this.parser._insertFakeElement($.BODY);
-    this.parser._insertFakeElement($.PRE);
-    this.parser.treeAdapter.insertText(this.parser.openElements.current, '\n');
-    this.parser.switchToPlaintextParsing();
-};
-
-inherits(PlainTextConversionStream, ParserStream);

package/lib/sax/dev_null_stream.js

@@ -1,14 +0,0 @@
-'use strict';
-
-var WritableStream = require('stream').Writable,
-    util = require('util');
-
-var DevNullStream = module.exports = function () {
-    WritableStream.call(this);
-};
-
-util.inherits(DevNullStream, WritableStream);
-
-DevNullStream.prototype._write = function (chunk, encoding, cb) {
-    cb();
-};