parser.java

gcc的组建

                  attrValue = buffer.toString();
                  break;

                // read unquoted attribute.
                case NUMTOKEN :
                  value = next;
                  optional(WS);

                  // Check maybe the opening quote is missing.
                  next = getTokenAhead();
                  if (bQUOTING.get(next.kind))
                    {
                      hTag = next;
                      error("The value without opening quote is closed with '" +
                            next.getImage() + "'"
                           );
                    }
                  attrValue = value.getImage();
                  break;

                default :
                  break attributeReading;
              }
            attributes.addAttribute(name.getImage(), attrValue);
            optional(WS);
          }
        else // The '=' is missing: attribute without value.
          {
            noValueAttribute(element, name.getImage());
          }
      }
  }

  /**
   * Return string, corresponding the given named entity.
   * The name is passed with the preceeding &, but without
   * the ending semicolon.
   */
  protected String resolveNamedEntity(final String a_tag)
  {
    // Discard &
    if (!a_tag.startsWith("&"))
      throw new AssertionError("Named entity " + a_tag +
                               " must start witn '&'."
                              );

    String tag = a_tag.substring(1);

    try
      {
        Entity entity = dtd.getEntity(tag);
        if (entity != null)
          return entity.getString();

        entity = dtd.getEntity(tag.toLowerCase());

        if (entity != null)
          {
            error("The name of this entity should be in lowercase", a_tag);
            return entity.getString();
          }
      }
    catch (IndexOutOfBoundsException ibx)
      {
        /* The error will be reported. */
      }

    error("Unknown named entity", a_tag);
    return a_tag;
  }

  /**
   * Return char, corresponding the given numeric entity.
   * The name is passed with the preceeding &#, but without
   * the ending semicolon.
   */
  protected char resolveNumericEntity(final String a_tag)
  {
    // Discard &#
    if (!a_tag.startsWith("&#"))
      throw new AssertionError("Numeric entity " + a_tag +
                               " must start witn '&#'."
                              );

    String tag = a_tag.substring(2);

    try
      {
        // Determine the encoding type:
        char cx = tag.charAt(0);
        if (cx == 'x' || cx == 'X') // Hexadecimal &#Xnnn;

          return (char) Integer.parseInt(tag.substring(1), 16);

        return (char) Integer.parseInt(tag);
      }

    /* The error will be reported. */
    catch (NumberFormatException nex)
      {
      }
    catch (IndexOutOfBoundsException ix)
      {
      }

    error("Invalid numeric entity", a_tag);
    return '?';
  }

  /**
   * Reset all fields into the intial default state, preparing the
   * parset for parsing the next document.
   */
  protected void restart()
  {
    documentTags.clear();
    titleHandled = false;
    titleOpen = false;
    buffer.setLength(0);
    title.setLength(0);
    validator.restart();
  }

  /**
   * The method is called when the HTML opening tag ((like <table>)
   * is found or if the parser concludes that the one should be present
   * in the current position. The method is called immediately before
   * calling the handleStartTag.
   * @param The tag
   */
  protected void startTag(TagElement tag)
                   throws ChangedCharSetException
  {
  }

  /**
   * Handle a complete element, when the tag content is already present in the
   * buffer and both starting and heading tags behind. This is called
   * in the case when the tag text must not be parsed for the nested
   * elements (elements STYLE and SCRIPT).
   */
  private void _handleCompleteElement(TagElement tag)
  {
    _handleStartTag(tag);

    // Suppress inclusion of the SCRIPT ans STYLE texts into the title.
    HTML.Tag h = tag.getHTMLTag();
    if (h == HTML.Tag.SCRIPT || h == HTML.Tag.STYLE)
      {
        boolean tmp = titleOpen;
        titleOpen = false;
        _handleText();
        titleOpen = tmp;
      }
    else
      _handleText();

    _handleEndTag(tag);
  }

  /**
   * A hooks for operations, preceeding call to handleEmptyTag().
   * Handle the tag with no content, like <br>. As no any
   * nested tags are expected, the tag validator is not involved.
   * @param The tag being handled.
   */
  private void _handleEmptyTag(TagElement tag)
  {
    try
      {
        validator.validateTag(tag, attributes);
        handleEmptyTag(tag);
      }
    catch (ChangedCharSetException ex)
      {
        error("Changed charset exception:", ex.getMessage());
      }
  }

  /**
   * A hooks for operations, preceeding call to handleEndTag().
   * The method is called when the HTML closing tag
   * is found. Calls handleTitle after closing the 'title' tag.
   * @param The tag
   */
  private void _handleEndTag(TagElement tag)
  {
    validator.closeTag(tag);
    _handleEndTag_remaining(tag);
  }

  /**
   * Actions that are also required if the closing action was
   * initiated by the tag validator.
   * Package-private to avoid an accessor method.
   */
  void _handleEndTag_remaining(TagElement tag)
  {
    HTML.Tag h = tag.getHTMLTag();

    handleEndTag(tag);
    endTag(tag.fictional());

    if (h.isPreformatted())
      preformatted--;
    if (preformatted < 0)
      preformatted = 0;

    if (h == HTML.Tag.TITLE)
      {
        titleOpen = false;
        titleHandled = true;

        char[] a = new char[ title.length() ];
        title.getChars(0, a.length, a, 0);
        handleTitle(a);
      }
  }

  /**
   * A hooks for operations, preceeding call to handleStartTag().
   * The method is called when the HTML opening tag ((like &lt;table&gt;)
   * is found.
   * Package-private to avoid an accessor method.
   * @param The tag
   */
  void _handleStartTag(TagElement tag)
  {
    validator.openTag(tag, attributes);
    startingTag(tag);
    handleStartTag(tag);

    HTML.Tag h = tag.getHTMLTag();

    if (h.isPreformatted())
      preformatted++;

    if (h == HTML.Tag.TITLE)
      {
        if (titleHandled)
          error("Repetetive <TITLE> tag");
        titleOpen = true;
        titleHandled = false;
      }
  }

  /**
   * Resume parsing after heavy errors in HTML tag structure.
   * @throws ParseException
   */
  private void forciblyCloseTheTag()
                            throws ParseException
  {
    int closeAt = 0;
    buffer.setLength(0);

    ahead: 
    for (int i = 1; i < 100; i++)
      {
        t = getTokenAhead(i - 1);
        if (t.kind == EOF || t.kind == BEGIN)
          break ahead;
        if (t.kind == END)
          {
            /* Closing '>' found. */
            closeAt = i;
            break ahead;
          }
      }
    if (closeAt > 0)
      {
        buffer.append("Ignoring '");
        for (int i = 1; i <= closeAt; i++)
          {
            t = getNextToken();
            append(t);
          }
        buffer.append('\'');
        error(buffer.toString());
      }
  }

  /**
   * Handle comment in string buffer. You can avoid allocating a char
   * array each time by processing your comment directly here.
   */
  private void handleComment()
  {
    char[] a = new char[ buffer.length() ];
    buffer.getChars(0, a.length, a, 0);
    handleComment(a);
  }

  private TagElement makeTagElement(String name, boolean isSupposed)
  {
    Element e = (Element) dtd.elementHash.get(name.toLowerCase());
    if (e == null)
      {
        error("Unknown tag <" + name + ">");
        e = dtd.getElement(name);
        e.name = name.toUpperCase();
        e.index = -1;
      }

    if (!documentTags.contains(e.name))
      {
        markFirstTime(e);
        documentTags.add(e.name);
      }

    return makeTag(e, isSupposed);
  }

  /**
   * Read till the given token, resolving entities. Consume the given
   * token without adding it to buffer.
   * @param till The token to read till
   * @throws ParseException
   */
  private void readTillTokenE(int till)
                       throws ParseException
  {
    buffer.setLength(0);
    read: 
    while (true)
      {
        t = getNextToken();
        if (t.kind == Constants.ENTITY)
          {
            resolveAndAppendEntity(t);
          }
        else if (t.kind == EOF)
          {
            error("unexpected eof", t);
            break read;
          }
        else if (t.kind == till)
          break read;
        else if (t.kind == WS)
          {
            // Processing whitespace in accordance with CDATA rules:
            String s = t.getImage();
            char c;
            for (int i = 0; i < s.length(); i++)
              {
                c = s.charAt(i);
                if (c == '\r')
                  buffer.append(' '); // CR replaced by space
                else if (c == '\n')
                  ; // LF ignored
                else if (c == '\t')
                  buffer.append(' '); // Tab replaced by space
                else
                  buffer.append(c);
              }
          }
        else
          append(t);
      }
  }

  /**
   * Resolve the entity and append it to the end of buffer.
   * @param entity
   */
  private void resolveAndAppendEntity(Token entity)
  {
    switch (entity.category)
      {
        case ENTITY_NAMED :
          buffer.append(resolveNamedEntity(entity.getImage()));
          break;

        case ENTITY_NUMERIC :
          buffer.append(resolveNumericEntity(entity.getImage()));
          break;

        default :
          throw new AssertionError("Invalid entity category " +
                                   entity.category
                                  );
      }
  }

  /**
   * Handle the remaining of HTML tags. This is a common end for
   * TAG, SCRIPT and STYLE.
   * @param closing True for closing tags ( &lt;/TAG&gt; ).
   * @param name Name of element
   * @param start Token where element has started
   * @throws ParseException
   */
  private void restOfTag(boolean closing, Token name, Token start)
                  throws ParseException
  {
    boolean end = false;
    Token next;

    optional(WS);

    readAttributes(name.getImage());

    optional(WS);

    next = getTokenAhead();
    if (next.kind == END)
      {
        mustBe(END);
        end = true;
      }

    hTag = new Token(start, next);

    attributes.setResolveParent(defaulter.getDefaultParameters(name.getImage()));

    if (!end)
      {
        // The tag body contains errors. If additionally the tag
        // name is not valid, this construction is treated as text.
        if (dtd.elementHash.get(name.getImage().toLowerCase()) == null &&
            backupMode
           )
          {
            error("Errors in tag body and unknown tag name. " +
                  "Treating the tag as a text."
                 );
            reset();

            hTag = mustBe(BEGIN);
            buffer.setLength(0);
            buffer.append(hTag.getImage());
            CDATA(false);
            return;
          }
        else
          {
            error("Forcibly closing invalid parameter list");
            forciblyCloseTheTag();
          }
      }

    if (closing)
      {
        endTag(false);
        _handleEndTag(makeTagElement(name.getImage(), false));
      }
    else
      {
        TagElement te = makeTagElement(name.getImage(), false);
        if (te.getElement().type == DTDConstants.EMPTY)
          _handleEmptyTag(te);
        else
          _handleStartTag(te);
      }
  }

  /**
   * This should fire additional actions in response to the
   * ChangedCharSetException.  The current implementation
   * does nothing.
   * @param tag
   */
  private void startingTag(TagElement tag)
  {
    try
      {
        startTag(tag);
      }
    catch (ChangedCharSetException cax)
      {
        error("Invalid change of charset");
      }
  }

  private void ws_error()
  {
    error("Whitespace here is not permitted");
  }
}