validator/js/engine/htmlparser.js - main - githubmirror/amphtml

validator/js/engine/htmlparser.js
1 113 строк34 KB
Esther Kim
Use Object for validator jsdoc (#39136)
13 июн 2023, 20:46
13 июн 2023, 20:46142af5d
/**
 * Credits:
 *   Copyright 2006-2008, The Google Caja project, licensed under the
 *   Apache License (http://code.google.com/p/google-caja/).
 *   Copyright 2009, The Closure Library Authors, licensed under the
 *   Apache License.
 */

/**
 * @fileoverview A Html SAX parser.
 *
 * Examples of usage of the `goog.string.html.HtmlParser`:
 * <pre>
 *   const handler = new MyCustomHtmlVisitorHandlerThatExtendsHtmlSaxHandler();
 *   const parser = new goog.string.html.HtmlParser();
 *   parser.parse(handler, '<html><a href="google.com">link found!</a></html>');
 * </pre>
 */

goog.module('amp.htmlparser');
const parserInterface = goog.require('amp.htmlparser.interface');

/**
 * Some tags have no end tags as per HTML5 spec. These were extracted
 * from the single page spec by looking for "no end tag" with CTRL+F.
 * @type {Object<string, ?>}
 * @private
 */
const ElementsWithNoEndElements = {
  'BASE': 0,
  'LINK': 0,
  'META': 0,
  'HR': 0,
  'BR': 0,
  'WBR': 0,
  'IMG': 0,
  'EMBED': 0,
  'PARAM': 0,
  'SOURCE': 0,
  'TRACK': 0,
  'AREA': 0,
  'COL': 0,
  'INPUT': 0,
  'KEYGEN': 0,
};

/**
 * Set of HTML tags which should never trigger an implied open of a <head>
 * or <body> element.
 * @type {Object<string,?>}
 * @private
 */
const HtmlStructureElements = {
  // See https://www.w3.org/TR/html5/document-metadata.html
  '!DOCTYPE': 0,
  'HTML': 0,
  'HEAD': 0,
  'BODY': 0,
};

/**
 * The set of HTML tags which are legal in the HTML document <head> and
 * the 'HEAD' tag itself.
 * @type {Object<string,?>}
 * @private
 */
const HeadElements = {
  'HEAD': 0,
  // See https://www.w3.org/TR/html5/document-metadata.html
  'TITLE': 0,
  'BASE': 0,
  'LINK': 0,
  'META': 0,
  'STYLE': 0,
  // Also legal in the document <head>, though not per spec.
  'NOSCRIPT': 0,
  'SCRIPT': 0,
};

/**
 * The set of HTML tags whose presence will implicitly close a <p> element.
 * For example '<p>foo<h1>bar</h1>' should parse the same as
 * '<p>foo</p><h1>bar</h1>'. See https://www.w3.org/TR/html-markup/p.html
 * @type {Object<string,?>}
 * @private
 */
const ElementsWhichClosePTag = {
  'ADDRESS': 0,
  'ARTICLE': 0,
  'ASIDE': 0,
  'BLOCKQUOTE': 0,
  'DIR': 0,
  'DL': 0,
  'FIELDSET': 0,
  'FOOTER': 0,
  'FORM': 0,
  'H1': 0,
  'H2': 0,
  'H3': 0,
  'H4': 0,
  'H5': 0,
  'H6': 0,
  'HEADER': 0,
  'HR': 0,
  'MENU': 0,
  'NAV': 0,
  'OL': 0,
  'P': 0,
  'PRE': 0,
  'SECTION': 0,
  'TABLE': 0,
  'UL': 0,
};

/**
 * @enum {number}
 */
const TagRegion = {
  PRE_DOCTYPE: 0,
  PRE_HTML: 1,
  PRE_HEAD: 2,
  IN_HEAD: 3,
  PRE_BODY: 4,  // After closing head tag, but before open body tag.
  IN_BODY: 5,
  IN_SVG: 6,
  // We don't track the region after the closing body tag.
};

/**
 * Computes the byte length, rather than character length, of a utf8 string.
 * https://en.wikipedia.org/wiki/UTF-8
 * @param {string} utf8Str
 * @return {number}
 */
const byteLength = function byteLength(utf8Str) {
  // To figure out which characters are multi-byte we can abuse
  // encodeURIComponent which will escape those specific characters.
  const multiByteEscapedChars = encodeURIComponent(utf8Str).match(/%[89ABab]/g);
  if (multiByteEscapedChars === null) {
    return utf8Str.length;
  } else {
    return utf8Str.length + multiByteEscapedChars.length;
  }
};
exports.byteLength = byteLength;

/**
 * This abstraction keeps track of which tags have been opened / closed as we
 * traverse the tags in the document. Closing tags is tricky:
 * - Some tags have no end tag per spec. For example, there is no </img> tag per
 *   spec. Since we are making startTag/endTag calls, we manufacture endTag
 *   calls for these immediately after the startTag.
 * - We assume all end tags are optional and we pop tags off our stack as we
 *   encounter parent closing tags. This part differs slightly from the behavior
 *   per spec: instead of closing an <option> tag when a following <option> tag
 *   is seen, we close it when the parent closing tag (in practice <select>) is
 *   encountered
 * @private
 */
class TagNameStack {
  /**
   * Creates an empty instance.
   * @param {!parserInterface.HtmlSaxHandler|
   *     !parserInterface.HtmlSaxHandlerWithLocation} handler The
   *         HtmlSaxHandler that will receive the events.
   */
  constructor(handler) {
    /**
     * The current tag name and its parents.
     * @type {!Array<string>}
     * @private
     */
    this.stack_ = [];

    /**
     * The current tag name and its parents.
     * @type {!parserInterface.HtmlSaxHandler|
     *     !parserInterface.HtmlSaxHandlerWithLocation} handler The
     *         HtmlSaxHandler that will receive the events.
     * @private
     */
    this.handler_ = handler;

    /**
     * @type {number}
     * @private
     */
    this.region_ = TagRegion.PRE_DOCTYPE;

    /**
     * Tracks when the start <head> tag has been encountered or manufactured.
     * @type {boolean}
     * @private
     */
    this.isHeadStarted_ = false;

    /**
     * Tracks when the start <body> tag has been encountered or manufactured.
     * @type {boolean}
     * @private
     */
    this.isBodyStarted_ = false;

    /**
     * Keeps track of the attributes from all body tags encountered within the
     * document.
     * @type {!Array<!Object>}
     * @private
     */
    this.effectiveBodyAttribs_ = [];
  }

  /**
   * Returns the attributes from all body tags within the document.
   * @return {!Array<!Object>}
   */
  effectiveBodyAttribs() {
    return this.effectiveBodyAttribs_;
  }

  /**
   * Enter a tag, opening a scope for child tags. Entering a tag can close the
   * previous tag or enter other tags (such as opening a <body> tag when
   * encountering a tag not allowed outside the body.
   * @param {!parserInterface.ParsedHtmlTag} tag
   */
  startTag(tag) {
    // We only report the first body for each document - either
    // a manufactured one, or the first one encountered. However,
    // we collect all attributes in this.effectiveBodyAttribs_.
    if (tag.upperName() === 'BODY') {
      this.effectiveBodyAttribs_ =
          this.effectiveBodyAttribs_.concat(tag.attrs().slice());
    }

    // This section deals with manufacturing <head>, </head>, and <body> tags
    // if the document has left them out or placed them in the wrong location.
    switch (this.region_) {
      case TagRegion.PRE_DOCTYPE:
        if (tag.upperName() === '!DOCTYPE') {
          this.region_ = TagRegion.PRE_HTML;
        } else if (tag.upperName() === 'HTML') {
          this.region_ = TagRegion.PRE_HEAD;
        } else if (tag.upperName() === 'HEAD') {
          this.region_ = TagRegion.IN_HEAD;
        } else if (tag.upperName() === 'BODY') {
          this.region_ = TagRegion.IN_BODY;
        } else if (!HtmlStructureElements.hasOwnProperty(tag.upperName())) {
          if (HeadElements.hasOwnProperty(tag.upperName())) {
            this.startTag(new parserInterface.ParsedHtmlTag('HEAD'));
          } else {
            if (this.handler_.markManufacturedBody) {
              this.handler_.markManufacturedBody();
            }
            this.startTag(new parserInterface.ParsedHtmlTag('BODY'));
          }
        }
        break;
      case TagRegion.PRE_HTML:
        // Stray DOCTYPE/HTML tags are ignored, not emitted twice.
        if (tag.upperName() === '!DOCTYPE') {
          return;
        } else if (tag.upperName() === 'HTML') {
          this.region_ = TagRegion.PRE_HEAD;
        } else if (tag.upperName() === 'HEAD') {
          this.region_ = TagRegion.IN_HEAD;
        } else if (tag.upperName() === 'BODY') {
          this.region_ = TagRegion.IN_BODY;
        } else if (!HtmlStructureElements.hasOwnProperty(tag.upperName())) {
          if (HeadElements.hasOwnProperty(tag.upperName())) {
            this.startTag(new parserInterface.ParsedHtmlTag('HEAD'));
          } else {
            if (this.handler_.markManufacturedBody) {
              this.handler_.markManufacturedBody();
            }
            this.startTag(new parserInterface.ParsedHtmlTag('BODY'));
          }
        }
        break;
      case TagRegion.PRE_HEAD:
        // Stray DOCTYPE/HTML tags are ignored, not emitted twice.
        if (tag.upperName() === '!DOCTYPE' || tag.upperName() === 'HTML') {
          return;
        } else if (tag.upperName() === 'HEAD') {
          this.region_ = TagRegion.IN_HEAD;
        } else if (tag.upperName() === 'BODY') {
          this.region_ = TagRegion.IN_BODY;
        } else if (!HtmlStructureElements.hasOwnProperty(tag.upperName())) {
          if (HeadElements.hasOwnProperty(tag.upperName())) {
            this.startTag(new parserInterface.ParsedHtmlTag('HEAD'));
          } else {
            if (this.handler_.markManufacturedBody) {
              this.handler_.markManufacturedBody();
            }
            this.startTag(new parserInterface.ParsedHtmlTag('BODY'));
          }
        }
        break;
      case TagRegion.IN_HEAD:
        // Stray DOCTYPE/HTML/HEAD tags are ignored, not emitted twice.
        if (tag.upperName() === '!DOCTYPE' || tag.upperName() === 'HTML' ||
            tag.upperName() === 'HEAD') {
          return;
        } else if (!HeadElements.hasOwnProperty(tag.upperName())) {
          this.endTag(new parserInterface.ParsedHtmlTag('HEAD'));
          if (tag.upperName() !== 'BODY') {
            if (this.handler_.markManufacturedBody) {
              this.handler_.markManufacturedBody();
            }
            this.startTag(new parserInterface.ParsedHtmlTag('BODY'));
          } else {
            this.region_ = TagRegion.IN_BODY;
          }
        }
        break;
      case TagRegion.PRE_BODY:
        // Stray DOCTYPE/HTML/HEAD tags are ignored, not emitted twice.
        if (tag.upperName() === '!DOCTYPE' || tag.upperName() === 'HTML' ||
            tag.upperName() === 'HEAD') {
          return;
        } else if (tag.upperName() !== 'BODY') {
          if (this.handler_.markManufacturedBody) {
            this.handler_.markManufacturedBody();
          }
          this.startTag(new parserInterface.ParsedHtmlTag('BODY'));
        } else {
          this.region_ = TagRegion.IN_BODY;
        }
        break;
      case TagRegion.IN_BODY:
        // Stray DOCTYPE/HTML/HEAD tags are ignored, not emitted twice.
        if (tag.upperName() === '!DOCTYPE' || tag.upperName() === 'HTML' ||
            tag.upperName() === 'HEAD') {
          return;
        } else if (tag.upperName() === 'BODY') {
          // We only report the first body for each document - either
          // a manufactured one, or the first one encountered.
          return;
        }
        if (tag.upperName() === 'SVG') {
          this.region_ = TagRegion.IN_SVG;
          break;
        }
        // Check implicit tag closing due to opening tags.
        if (this.stack_.length > 0) {
          const parentTagName = this.stack_[this.stack_.length - 1];
          // <p> tags can be implicitly closed by certain other start tags.
          // See https://www.w3.org/TR/html-markup/p.html
          if (parentTagName === 'P' &&
              ElementsWhichClosePTag.hasOwnProperty(tag.upperName())) {
            this.endTag(new parserInterface.ParsedHtmlTag('P'));
            // <dd> and <dt> tags can be implicitly closed by other <dd> and
            // <dt> tags. See https://www.w3.org/TR/html-markup/dd.html
          } else if (
              (tag.upperName() == 'DD' || tag.upperName() == 'DT') &&
              (parentTagName == 'DD' || parentTagName == 'DT')) {
            this.endTag(new parserInterface.ParsedHtmlTag(parentTagName));
            // <li> tags can be implicitly closed by other <li> tags.
            // See https://www.w3.org/TR/html-markup/li.html
          } else if (tag.upperName() == 'LI' && parentTagName == 'LI') {
            this.endTag(new parserInterface.ParsedHtmlTag('LI'));
          }
        }
        break;
      case TagRegion.IN_SVG:
        if (this.handler_.startTag) {
          this.handler_.startTag(tag);
        }
        this.stack_.push(tag.upperName());
        return;
      default:
        break;
    }

    if (this.handler_.startTag) {
      this.handler_.startTag(tag);
    }
    if (ElementsWithNoEndElements.hasOwnProperty(tag.upperName())) {
      if (this.handler_.endTag) {
        // Ignore attributes in end tags.
        this.handler_.endTag(
            new parserInterface.ParsedHtmlTag(tag.upperName()));
      }
    } else {
      this.stack_.push(tag.upperName());
    }
  }

  /**
   * Callback for pcdata. Some text nodes can trigger the start of the body
   * region.
   * @param {string} text
   */
  pcdata(text) {
    if (SPACE_RE_.test(text)) {
      // Only ASCII whitespace; this can be ignored for validator's purposes.
    } else if (CPP_SPACE_RE_.test(text)) {
      // Non-ASCII whitespace; if this occurs outside <body>, output a
      // manufactured-body error. Do not create implicit tags, in order to match
      // the behavior of the buggy C++ parser. (It just so happens this is also
      // good UX, since the subsequent validation errors caused by the implicit
      // tags are unhelpful.)
      switch (this.region_) {
        // Fallthroughs intentional.
        case TagRegion.PRE_DOCTYPE:
        case TagRegion.PRE_HTML:
        case TagRegion.PRE_HEAD:
        case TagRegion.IN_HEAD:
        case TagRegion.PRE_BODY:
          if (this.handler_.markManufacturedBody) {
            this.handler_.markManufacturedBody();
          }
      }
    } else {
      // Non-whitespace text; if this occurs outside <body>, output a
      // manufactured-body error and create the necessary implicit tags.
      switch (this.region_) {
        // Fallthroughs intentional.
        case TagRegion.PRE_DOCTYPE:  // doctype is not manufactured
        case TagRegion.PRE_HTML:
          this.startTag(new parserInterface.ParsedHtmlTag('HTML'));
        case TagRegion.PRE_HEAD:
          this.startTag(new parserInterface.ParsedHtmlTag('HEAD'));
        case TagRegion.IN_HEAD:
          this.endTag(new parserInterface.ParsedHtmlTag('HEAD'));
        case TagRegion.PRE_BODY:
          if (this.handler_.markManufacturedBody) {
            this.handler_.markManufacturedBody();
          }
          this.startTag(new parserInterface.ParsedHtmlTag('BODY'));
      }
    }
    this.handler_.pcdata(text);
  }

  /**
   * Upon exiting a tag, validation for the current matcher is triggered,
   * e.g. for checking that the tag had some specified number of children.
   * @param {!parserInterface.ParsedHtmlTag} tag
   */
  endTag(tag) {
    if (this.region_ == TagRegion.IN_HEAD && tag.upperName() === 'HEAD') {
      this.region_ = TagRegion.PRE_BODY;
    }

    // We ignore close body tags (</body) and instead insert them when their
    // outer scope is closed (/html). This is closer to how a browser parser
    // works. The idea here is if other tags are found after the <body>,
    // (ex: <div>) which are only allowed in the <body>, we will effectively
    // move them into the body section.
    if (tag.upperName() === 'BODY') {
      return;
    }

    // We look for tag.upperName() from the end. If we can find it, we pop
    // everything from thereon off the stack. If we can't find it,
    // we don't bother with closing the tag, since it doesn't have
    // a matching open tag, though in practice the HtmlParser class
    // will have already manufactured a start tag.
    for (let idx = this.stack_.length - 1; idx >= 0; idx--) {
      if (this.stack_[idx] === tag.upperName()) {
        while (this.stack_.length > idx) {
          if (this.stack_[this.stack_.length - 1] === 'SVG') {
            this.region_ = TagRegion.IN_BODY;
          }
          if (this.handler_.endTag) {
            this.handler_.endTag(
                new parserInterface.ParsedHtmlTag(this.stack_.pop()));
          }
        }
        return;
      }
    }
  }

  /**
   * This method is called when we're done with the
   * document. Normally, the parser should actually close the tags,
   * but just in case it doesn't this easy-enough method will take care of it.
   */
  exitRemainingTags() {
    while (this.stack_.length > 0) {
      if (this.handler_.endTag) {
        this.handler_.endTag(
            new parserInterface.ParsedHtmlTag(this.stack_.pop()));
      }
    }
  }
}


/**
 * An Html parser: `parse` takes a string and calls methods on
 * `HtmlSaxHandler` while it is visiting it.
 */
const HtmlParser = class {
  constructor() {}

  /**
   * Given a SAX-like `HtmlSaxHandler` parses a
   * `htmlText` and lets the `handler` know the structure while
   * visiting the nodes. If the provided handler is an implementation of
   * `htmlparser.HtmlSaxHandlerWithLocation`, then its
   * `setDocLocator` method will get called prior to
   * `startDoc`, and the `getLine` / `getCol` methods will
   * reflect the current line / column while a SAX callback (e.g.,
   * `startTag`) is active.
   *
   * @param {!parserInterface.HtmlSaxHandler|
   *     !parserInterface.HtmlSaxHandlerWithLocation} handler The
   *         HtmlSaxHandler that will receive the events.
   * @param {string} htmlText The html text.
   */
  parse(handler, htmlText) {
    let htmlUpper = null;
    let inTag = false;   // True iff we're currently processing a tag.
    const attribs = [];  // Accumulates attribute names and values.
    let tagName;         // The name of the tag currently being processed.
    let eflags;          // The element flags for the current tag.
    let openTag;         // True if the current tag is an open tag.
    const tagStack = new TagNameStack(handler);

    // Only provide location information if the handler implements the
    // setDocLocator method.
    let locator = null;
    if (handler instanceof parserInterface.HtmlSaxHandlerWithLocation) {
      locator = new DocLocatorImpl(htmlText);
      handler.setDocLocator(locator);
    }

    // Lets the handler know that we are starting to parse the document.
    handler.startDoc();

    // Consumes tokens from the htmlText and stops once all tokens are
    // processed.
    while (htmlText) {
      const regex = inTag ? INSIDE_TAG_TOKEN_ : OUTSIDE_TAG_TOKEN_;
      // Gets the next token
      const m = htmlText.match(regex);
      if (locator) {
        locator.advancePos(m[0]);
      }
      // And removes it from the string
      htmlText = htmlText.substring(m[0].length);

      // TODO(goto): cleanup this code breaking it into separate methods.
      if (inTag) {
        if (m[1]) {  // Attribute.
          // SetAttribute with uppercase names doesn't work on IE6.
          const attribName = parserInterface.toLowerCase(m[1]);
          // Use empty string as value for valueless attribs, so
          //   <input type=checkbox checked>
          // gets attributes ['type', 'checkbox', 'checked', '']
          let decodedValue = '';
          if (m[2]) {
            let encodedValue = m[3];
            switch (encodedValue.charCodeAt(0)) {  // Strip quotes.
              case 34:                             // double quote "
              case 39:                             // single quote '
                encodedValue =
                    encodedValue.substring(1, encodedValue.length - 1);
                break;
            }
            decodedValue =
                this.unescapeEntities_(this.stripNULs_(encodedValue));
          }
          attribs.push(attribName, decodedValue);
        } else if (m[4]) {
          if (eflags !== void 0) {  // False if not in allowlist.
            if (openTag) {
              tagStack.startTag(new parserInterface.ParsedHtmlTag(
                  /** @type {string} */ (tagName), attribs));
            } else {
              tagStack.endTag(new parserInterface.ParsedHtmlTag(
                  /** @type {string} */ (tagName)));
            }
          }

          if (openTag && (eflags & (EFlags.CDATA | EFlags.RCDATA))) {
            if (htmlUpper === null) {
              htmlUpper = parserInterface.toUpperCase(htmlText);
            } else {
              htmlUpper =
                  htmlUpper.substring(htmlUpper.length - htmlText.length);
            }
            let dataEnd = htmlUpper.indexOf('</' + tagName);
            if (dataEnd < 0) {
              dataEnd = htmlText.length;
            }
            if (eflags & EFlags.CDATA) {
              if (handler.cdata) {
                handler.cdata(htmlText.substring(0, dataEnd));
              }
            } else if (handler.rcdata) {
              handler.rcdata(
                  this.normalizeRCData_(htmlText.substring(0, dataEnd)));
            }
            if (locator) {
              locator.advancePos(htmlText.substring(0, dataEnd));
            }
            htmlText = htmlText.substring(dataEnd);
          }

          tagName = eflags = openTag = void 0;
          attribs.length = 0;
          if (locator) {
            locator.snapshotPos();
          }
          inTag = false;
        }
      } else {
        if (m[1]) {  // Entity.
          tagStack.pcdata(m[0]);
        } else if (m[3]) {  // Tag.
          openTag = !m[2];
          if (locator) {
            locator.snapshotPos();
          }
          inTag = true;
          tagName = parserInterface.toUpperCase(m[3]);
          eflags = Elements.hasOwnProperty(tagName) ? Elements[tagName] :
                                                      EFlags.UNKNOWN_OR_CUSTOM;
        } else if (m[4]) {  // Text.
          if (locator) {
            locator.snapshotPos();
          }
          tagStack.pcdata(m[4]);
        } else if (m[5]) {  // Cruft.
          switch (m[5]) {
            case '<':
              tagStack.pcdata('&lt;');
              break;
            case '>':
              tagStack.pcdata('&gt;');
              break;
            default:
              tagStack.pcdata('&amp;');
              break;
          }
        } else {
        }
      }
    }

    if (!inTag && locator) {
      locator.snapshotPos();
    }
    // Lets the handler know that we are done parsing the document.
    tagStack.exitRemainingTags();
    handler.effectiveBodyTag(tagStack.effectiveBodyAttribs());
    handler.endDoc();
  }

  /**
   * Decodes an HTML entity.
   *
   * @param {string} entity The full entity (including the & and the ;).
   * @return {string} A single unicode code-point as a string.
   * @private
   */
  lookupEntity_(entity) {
    // TODO(goto): use {amp.htmlparserDecode} instead ?
    // TODO(goto): &pi; is different from &Pi;
    const name =
        parserInterface.toLowerCase(entity.substring(1, entity.length - 1));
    if (Entities.hasOwnProperty(name)) {
      return Entities[name];
    }
    let m = name.match(DECIMAL_ESCAPE_RE_);
    if (m) {
      return String.fromCharCode(parseInt(m[1], 10));
    } else if (!!(m = name.match(HEX_ESCAPE_RE_))) {
      return String.fromCharCode(parseInt(m[1], 16));
    }
    // If unable to decode, return the name.
    return name;
  }

  /**
   * Removes null characters on the string.
   * @param {string} s The string to have the null characters removed.
   * @return {string} A string without null characters.
   * @private
   */
  stripNULs_(s) {
    return s.replace(NULL_RE_, '');
  }

  /**
   * The plain text of a chunk of HTML CDATA which possibly containing.
   *
   * TODO(goto): use `goog.string.unescapeEntities` instead ?
   * @param {string} s A chunk of HTML CDATA.  It must not start or end inside
   *   an HTML entity.
   * @return {string} The unescaped entities.
   * @private
   */
  unescapeEntities_(s) {
    return s.replace(ENTITY_RE_, goog.bind(this.lookupEntity_, this));
  }

  /**
   * Escape entities in RCDATA that can be escaped without changing the meaning.
   * @param {string} rcdata The RCDATA string we want to normalize.
   * @return {string} A normalized version of RCDATA.
   * @private
   */
  normalizeRCData_(rcdata) {
    return rcdata.replace(LOOSE_AMP_RE_, '&amp;$1')
        .replace(LT_RE, '&lt;')
        .replace(GT_RE, '&gt;');
  }
};
exports.HtmlParser = HtmlParser;

/**
 * HTML entities that are encoded/decoded.
 * TODO(goto): use `goog.string.htmlEncode` instead.
 * @type {!Object<string, string>}
 */
const Entities = {
  'colon': ':',
  'lt': '<',
  'gt': '>',
  'amp': '&',
  'nbsp': '\u00a0',
  'quot': '"',
  'apos': '\'',
};
exports.Entities = Entities;

/**
 * The html eflags, used internally on the parser.
 * @enum {number}
 */
const EFlags = {
  OPTIONAL_ENDTAG: 1,
  EMPTY: 2,
  CDATA: 4,
  RCDATA: 8,
  UNSAFE: 16,
  FOLDABLE: 32,
  UNKNOWN_OR_CUSTOM: 64,
};
exports.EFlags = EFlags;

/**
 * A map of element to a bitmap of flags it has, used internally on the parser.
 * @type {Object<string,?>}
 */
const Elements = {
  'A': 0,
  'ABBR': 0,
  'ACRONYM': 0,
  'ADDRESS': 0,
  'APPLET': EFlags.UNSAFE,
  'AREA': EFlags.EMPTY,
  'B': 0,
  'BASE': EFlags.EMPTY | EFlags.UNSAFE,
  'BASEFONT': EFlags.EMPTY | EFlags.UNSAFE,
  'BDO': 0,
  'BIG': 0,
  'BLOCKQUOTE': 0,
  'BODY': EFlags.OPTIONAL_ENDTAG | EFlags.UNSAFE | EFlags.FOLDABLE,
  'BR': EFlags.EMPTY,
  'BUTTON': 0,
  'CANVAS': 0,
  'CAPTION': 0,
  'CENTER': 0,
  'CITE': 0,
  'CODE': 0,
  'COL': EFlags.EMPTY,
  'COLGROUP': EFlags.OPTIONAL_ENDTAG,
  'DD': EFlags.OPTIONAL_ENDTAG,
  'DEL': 0,
  'DFN': 0,
  'DIR': 0,
  'DIV': 0,
  'DL': 0,
  'DT': EFlags.OPTIONAL_ENDTAG,
  'EM': 0,
  'FIELDSET': 0,
  'FONT': 0,
  'FORM': 0,
  'FRAME': EFlags.EMPTY | EFlags.UNSAFE,
  'FRAMESET': EFlags.UNSAFE,
  'H1': 0,
  'H2': 0,
  'H3': 0,
  'H4': 0,
  'H5': 0,
  'H6': 0,
  'HEAD': EFlags.OPTIONAL_ENDTAG | EFlags.UNSAFE | EFlags.FOLDABLE,
  'HR': EFlags.EMPTY,
  'HTML': EFlags.OPTIONAL_ENDTAG | EFlags.UNSAFE | EFlags.FOLDABLE,
  'I': 0,
  'IFRAME': EFlags.UNSAFE | EFlags.CDATA,
  'IMG': EFlags.EMPTY,
  'INPUT': EFlags.EMPTY,
  'INS': 0,
  'ISINDEX': EFlags.EMPTY | EFlags.UNSAFE,
  'KBD': 0,
  'LABEL': 0,
  'LEGEND': 0,
  'LI': EFlags.OPTIONAL_ENDTAG,
  'LINK': EFlags.EMPTY | EFlags.UNSAFE,
  'MAP': 0,
  'MENU': 0,
  'META': EFlags.EMPTY | EFlags.UNSAFE,
  'NOFRAMES': EFlags.UNSAFE | EFlags.CDATA,
  // TODO(johannes): This used to read:
  // 'NOSCRIPT': EFlags.UNSAFE |
  //  EFlags.CDATA,
  //
  // It appears that the effect of that is that anything inside is
  // then considered cdata, so <noscript><style>foo</noscript></noscript>
  // never sees a style start tag / end tag event. But we must
  // recognize such style tags and they're also allowed by HTML, e.g. see:
  // https://developer.mozilla.org/en-US/docs/Web/HTML/Element/noscript
  // On a broader note this also means we may be missing other start/end
  // tag events inside elements marked as CDATA which our parser
  // should better reject. Yikes.
  'NOSCRIPT': EFlags.UNSAFE,
  'OBJECT': EFlags.UNSAFE,
  'OL': 0,
  'OPTGROUP': 0,
  'OPTION': EFlags.OPTIONAL_ENDTAG,
  'P': EFlags.OPTIONAL_ENDTAG,
  'PARAM': EFlags.EMPTY | EFlags.UNSAFE,
  'PRE': 0,
  'Q': 0,
  'S': 0,
  'SAMP': 0,
  'SCRIPT': EFlags.UNSAFE | EFlags.CDATA,
  'SELECT': 0,
  'SMALL': 0,
  'SPAN': 0,
  'STRIKE': 0,
  'STRONG': 0,
  'STYLE': EFlags.UNSAFE | EFlags.CDATA,
  'SUB': 0,
  'SUP': 0,
  'TABLE': 0,
  'TBODY': EFlags.OPTIONAL_ENDTAG,
  'TD': EFlags.OPTIONAL_ENDTAG,
  'TEXTAREA': EFlags.RCDATA,
  'TFOOT': EFlags.OPTIONAL_ENDTAG,
  'TH': EFlags.OPTIONAL_ENDTAG,
  'THEAD': EFlags.OPTIONAL_ENDTAG,
  'TITLE': EFlags.RCDATA | EFlags.UNSAFE,
  'TR': EFlags.OPTIONAL_ENDTAG,
  'TT': 0,
  'U': 0,
  'UL': 0,
  'VAR': 0,
};
exports.Elements = Elements;

/**
 * Regular expression that matches loose &s.
 * @type {RegExp}
 * @private
 */
const LOOSE_AMP_RE_ = /&([^a-z#]|#(?:[^0-9x]|x(?:[^0-9a-f]|$)|$)|$)/gi;


/**
 * Regular expression that matches <.
 * @type {RegExp}
 * @package
 */
const LT_RE = /</g;


/**
 * Regular expression that matches >.
 * @type {RegExp}
 * @package
 */
const GT_RE = />/g;


/**
 * Regular expression that matches null characters.
 * @type {RegExp}
 * @private
 */
const NULL_RE_ = /\0/g;


/**
 * Regular expression that matches entities.
 * @type {RegExp}
 * @private
 */
const ENTITY_RE_ = /&(#\d+|#x[0-9A-Fa-f]+|\w+);/g;

/**
 * Regular expression that matches strings composed of all space characters, as
 * defined in https://infra.spec.whatwg.org/#ascii-whitespace,
// and in the various HTML parsing rules at
// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inhtml.
 *
 * Note: Do not USE \s to match whitespace as this includes many other
 * characters that HTML parsing does not consider whitespace.
 * @type {!RegExp}
 * @private
 */
const SPACE_RE_ = /^[ \f\n\r\t]*$/;

/**
 * Regular expression that matches the characters considered whitespace by the
 * C++ HTML parser.
 *
 * @type {!RegExp}
 * @private
 */
const CPP_SPACE_RE_ =
    /^[ \f\n\r\t\v\u00a0\u1680\u2000-\u200a\u2028\u2029\u202f\u205f\u3000]*$/;

/**
 * Regular expression that matches decimal numbers.
 * @type {RegExp}
 * @private
 */
const DECIMAL_ESCAPE_RE_ = /^#(\d+)$/;


/**
 * Regular expression that matches hexadecimal numbers.
 * @type {RegExp}
 * @private
 */
const HEX_ESCAPE_RE_ = /^#x([0-9A-Fa-f]+)$/;


/**
 * Regular expression that matches the next token to be processed.
 * @type {RegExp}
 * @private
 */
const INSIDE_TAG_TOKEN_ = new RegExp(
    // Don't capture space. In this case, we don't use \s because it includes a
    // nonbreaking space which gets included as an attribute in our validation.
    '^[ \\t\\n\\f\\r\\v]*(?:' +
        // Capture an attribute name in group 1, and value in group 3.
        // We capture the fact that there was an attribute in group 2, since
        // interpreters are inconsistent in whether a group that matches nothing
        // is null, undefined, or the empty string.
        ('(?:' +
         // Allow attribute names to start with /, avoiding assigning the / in
         // close-tag syntax */>.
         '([^\\t\\r\\n /=>][^\\t\\r\\n =>]*|' +  // e.g. "href"
         '[^\\t\\r\\n =>]+[^ >]|' +              // e.g. "/asdfs/asd"
         '\/+(?!>))' +                           // e.g. "/"
         // Optionally followed by:
         ('(' +
          '\\s*=\\s*' +
          ('(' +
           // A double quoted string.
           '\"[^\"]*\"' +
           // A single quoted string.
           '|\'[^\']*\'' +
           // The positive lookahead is used to make sure that in
           // <foo bar= baz=boo>, the value for bar is blank, not "baz=boo".
           // Note that <foo bar=baz=boo zik=zak>, the value for bar is
           // "baz=boo" and the value for zip is "zak".
           '|(?=[a-z][a-z-]*\\s+=)' +
           // An unquoted value that is not an attribute name.
           // We know it is not an attribute name because the previous
           // zero-width match would've eliminated that possibility.
           '|[^>\\s]*' +
           ')') +
          ')') +
         '?' +
         ')') +
        // End of tag captured in group 3.
        '|(/?>)' +
        // Don't capture cruft
        '|[^a-z\\s>]+)',
    'i');


/**
 * Regular expression that matches the next token to be processed when we are
 * outside a tag.
 * @type {RegExp}
 * @private
 */
const OUTSIDE_TAG_TOKEN_ = new RegExp(
    '^(?:' +
        // Entity captured in group 1.
        '&(\\#[0-9]+|\\#[x][0-9a-f]+|\\w+);' +
        // Comments not captured.
        '|<[!]--[\\s\\S]*?(?:--[!]?>|$)' +
        // '/' captured in group 2 for close tags, and name captured in group 3.
        // The first character of a tag (after possibly '/') can be A-Z, a-z,
        // '!' or '?'. The remaining characters are more easily expressed as a
        // negative set of: '\0', ' ', '\n', '\r', '\t', '\f', '\v', '>', or
        // '/'.
        '|<(/)?([a-z!\\?][^\\0 \\n\\r\\t\\f\\v>/]*)' +
        // Text captured in group 4.
        '|([^<&>]+)' +
        // Cruft captured in group 5.
        '|([<&>]))',
    'i');


/**
 * An implementation of the `parserInterface.DocLocator` parserInterface
 * for use within the `HtmlParser`.
 */
const DocLocatorImpl = class extends parserInterface.DocLocator {
  /**
   * @this {DocLocatorImpl}
   * @param {string} htmlText text of the entire HTML document to be processed.
   */
  constructor(htmlText) {
    super();
    // Precomputes a mapping from positions within htmlText to line /
    // column numbers. TODO(johannes): This uses a fair amount of
    // space and we can probably do better, but it's also quite simple
    // so here we are for now.
    this.lineByPos_ = [];
    this.colByPos_ = [];
    let currentLine = 1;
    let currentCol = 0;
    for (let i = 0; i < htmlText.length; ++i) {
      this.lineByPos_[i] = currentLine;
      this.colByPos_[i] = currentCol;
      if (htmlText.charAt(i) == '\n') {
        ++currentLine;
        currentCol = 0;
      } else {
        ++currentCol;
      }
    }
    this.docByteSize_ = byteLength(htmlText);

    // The current position in the htmlText.
    this.pos_ = 0;
    // The previous position in the htmlText - we need this to know where a
    // tag or attribute etc. started.
    this.previousPos_ = 0;

    // This gets computed from the maps above and the previousPos in
    // snapshotPos, and it's what client code of the DocLocator will
    // see.
    this.line_ = 1;
    this.col_ = 0;
  }


  /**
   * Advances the internal position by the characters in {code tokenText}.
   * This method is to be called only from within the parser.
   * @param {string} tokenText The token text which we examine to advance the
   *   line / column location within the doc.
   */
  advancePos(tokenText) {
    this.previousPos_ = this.pos_;
    this.pos_ += tokenText.length;
  }

  /**
   * Snapshots the previous internal position so that getLine / getCol will
   * return it. These snapshots happen as the parser enter / exits a tag.
   * This method is to be called only from within the parser.
   */
  snapshotPos() {
    if (this.previousPos_ < this.lineByPos_.length) {
      this.line_ = this.lineByPos_[this.previousPos_];
      this.col_ = this.colByPos_[this.previousPos_];
    }
  }

  /**
   * @inheritDoc
   * @this {DocLocatorImpl}
   */
  getLine() {
    return this.line_;
  }

  /**
   * @inheritDoc
   * @this {DocLocatorImpl}
   */
  getCol() {
    return this.col_;
  }

  /**
   * @inheritDoc
   * @this {DocLocatorImpl}
   */
  getDocByteSize() {
    return this.docByteSize_;
  }
};

/**
 * This function gets eliminated by closure compiler. It's purpose in life
 * is to work around a bug wherein the compiler renames the object keys
 * for objects never accessed using an array ([]) operator. We need the keys
 * to remain unchanged for these objects.
 */
function unusedHtmlParser() {
  console./*OK*/ log(ElementsWithNoEndElements['']);
  console./*OK*/ log(HtmlStructureElements['']);
  console./*OK*/ log(HeadElements['']);
  console./*OK*/ log(ElementsWhichClosePTag['']);
}