src/java.desktop/share/classes/javax/swing/text/html/parser/Parser.java - master - niceSOFT/openjdk21

src/java.desktop/share/classes/javax/swing/text/html/parser/Parser.java
2 463 строки78 KB
Prasanta Sadhukhan
8296660: Swing HTML table with omitted closing tags misparsed
01 дек 2022, 10:57
01 дек 2022, 10:57c612512
/*
 * Copyright (c) 1998, 2022, Oracle and/or its affiliates. All rights reserved.
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 *
 * This code is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License version 2 only, as
 * published by the Free Software Foundation.  Oracle designates this
 * particular file as subject to the "Classpath" exception as provided
 * by Oracle in the LICENSE file that accompanied this code.
 *
 * This code is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
 * version 2 for more details (a copy is included in the LICENSE file that
 * accompanied this code).
 *
 * You should have received a copy of the GNU General Public License version
 * 2 along with this work; if not, write to the Free Software Foundation,
 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
 *
 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
 * or visit www.oracle.com if you need additional information or have any
 * questions.
 */

package javax.swing.text.html.parser;

import javax.swing.text.SimpleAttributeSet;
import javax.swing.text.html.HTML;
import javax.swing.text.ChangedCharSetException;
import java.io.*;
import java.util.Hashtable;
import java.util.Properties;
import java.util.Vector;
import java.util.Enumeration;
import java.net.URL;

/**
 * A simple DTD-driven HTML parser. The parser reads an
 * HTML file from an InputStream and calls various methods
 * (which should be overridden in a subclass) when tags and
 * data are encountered.
 * <p>
 * Unfortunately there are many badly implemented HTML parsers
 * out there, and as a result there are many badly formatted
 * HTML files. This parser attempts to parse most HTML files.
 * This means that the implementation sometimes deviates from
 * the SGML specification in favor of HTML.
 * <p>
 * The parser treats \r and \r\n as \n. Newlines after starttags
 * and before end tags are ignored just as specified in the SGML/HTML
 * specification.
 * <p>
 * The html spec does not specify how spaces are to be coalesced very well.
 * Specifically, the following scenarios are not discussed (note that a
 * space should be used here, but I am using &amp;nbsp to force the space to
 * be displayed):
 * <p>
 * '&lt;b&gt;blah&nbsp;&lt;i&gt;&nbsp;&lt;strike&gt;&nbsp;foo' which can be treated as:
 * '&lt;b&gt;blah&nbsp;&lt;i&gt;&lt;strike&gt;foo'
 * <p>as well as:
 * '&lt;p&gt;&lt;a href="xx"&gt;&nbsp;&lt;em&gt;Using&lt;/em&gt;&lt;/a&gt;&lt;/p&gt;'
 * which appears to be treated as:
 * '&lt;p&gt;&lt;a href="xx"&gt;&lt;em&gt;Using&lt;/em&gt;&lt;/a&gt;&lt;/p&gt;'
 * <p>
 * If <code>strict</code> is false, when a tag that breaks flow,
 * (<code>TagElement.breaksFlows</code>) or trailing whitespace is
 * encountered, all whitespace will be ignored until a non whitespace
 * character is encountered. This appears to give behavior closer to
 * the popular browsers.
 *
 * @see DTD
 * @see TagElement
 * @see SimpleAttributeSet
 * @author Arthur van Hoff
 * @author Sunita Mani
 */
public
class Parser implements DTDConstants {

    private char[] text = new char[1024];
    private int textpos = 0;
    private TagElement last;
    private boolean space;

    private char[] str = new char[128];
    private int strpos = 0;

    /**
     * The dtd.
     */
    protected DTD dtd = null;

    private int ch;
    private int ln;
    private Reader in;

    private Element recent;
    private TagStack stack;
    private boolean skipTag = false;
    private TagElement lastFormSent = null;
    private SimpleAttributeSet attributes = new SimpleAttributeSet();

    // State for <html>, <head> and <body>.  Since people like to slap
    // together HTML documents without thinking, occasionally they
    // have multiple instances of these tags.  These booleans track
    // the first sightings of these tags so they can be safely ignored
    // by the parser if repeated.
    private boolean seenHtml = false;
    private boolean seenHead = false;
    private boolean seenBody = false;

    /**
     * The html spec does not specify how spaces are coalesced very well.
     * If strict == false, ignoreSpace is used to try and mimic the behavior
     * of the popular browsers.
     * <p>
     * The problematic scenarios are:
     * '&lt;b>blah &lt;i> &lt;strike> foo' which can be treated as:
     * '&lt;b>blah &lt;i>&lt;strike>foo'
     * as well as:
     * '&lt;p>&lt;a href="xx"> &lt;em>Using&lt;/em>&lt;/a>&lt;/p>'
     * which appears to be treated as:
     * '&lt;p>&lt;a href="xx">&lt;em>Using&lt;/em>&lt;/a>&lt;/p>'
     * <p>
     * When a tag that breaks flow, or trailing whitespace is encountered
     * ignoreSpace is set to true. From then on, all whitespace will be
     * ignored.
     * ignoreSpace will be set back to false the first time a
     * non whitespace character is encountered. This appears to give
     * behavior closer to the popular browsers.
     */
    private boolean ignoreSpace;

    /**
     * This flag determines whether or not the Parser will be strict
     * in enforcing SGML compatibility.  If false, it will be lenient
     * with certain common classes of erroneous HTML constructs.
     * Strict or not, in either case an error will be recorded.
     *
     */
    protected boolean strict = false;


    /** Number of \r\n's encountered. */
    private int crlfCount;
    /** Number of \r's encountered. A \r\n will not increment this. */
    private int crCount;
    /** Number of \n's encountered. A \r\n will not increment this. */
    private int lfCount;

    //
    // To correctly identify the start of a tag/comment/text we need two
    // ivars. Two are needed as handleText isn't invoked until the tag
    // after the text has been parsed, that is the parser parses the text,
    // then a tag, then invokes handleText followed by handleStart.
    //
    /** The start position of the current block. Block is overloaded here,
     * it really means the current start position for the current comment,
     * tag, text. Use getBlockStartPosition to access this. */
    private int currentBlockStartPos;
    /** Start position of the last block. */
    private int lastBlockStartPos;

    /**
     * array for mapping numeric references in range
     * 130-159 to displayable Unicode characters.
     */
    private static final char[] cp1252Map = {
        8218,  // &#130;
        402,   // &#131;
        8222,  // &#132;
        8230,  // &#133;
        8224,  // &#134;
        8225,  // &#135;
        710,   // &#136;
        8240,  // &#137;
        352,   // &#138;
        8249,  // &#139;
        338,   // &#140;
        141,   // &#141;
        142,   // &#142;
        143,   // &#143;
        144,   // &#144;
        8216,  // &#145;
        8217,  // &#146;
        8220,  // &#147;
        8221,  // &#148;
        8226,  // &#149;
        8211,  // &#150;
        8212,  // &#151;
        732,   // &#152;
        8482,  // &#153;
        353,   // &#154;
        8250,  // &#155;
        339,   // &#156;
        157,   // &#157;
        158,   // &#158;
        376    // &#159;
    };

    /**
     * Creates parser with the specified {@code dtd}.
     *
     * @param dtd the dtd.
     */
    public Parser(DTD dtd) {
        this.dtd = dtd;
    }


    /**
     * Returns the line number of the line currently being parsed.
     *
     * @return the line number of the line currently being parsed
     */
    protected int getCurrentLine() {
        return ln;
    }

    /**
     * Returns the start position of the current block. Block is
     * overloaded here, it really means the current start position for
     * the current comment tag, text, block.... This is provided for
     * subclassers that wish to know the start of the current block when
     * called with one of the handleXXX methods.
     *
     * @return the start position of the current block
     */
    int getBlockStartPosition() {
        return Math.max(0, lastBlockStartPos - 1);
    }

    /**
     * Makes a TagElement.
     *
     * @param elem       the element storing the tag definition
     * @param fictional  the value of the flag "{@code fictional}" to be set for the tag
     *
     * @return the created {@code TagElement}
     */
    protected TagElement makeTag(Element elem, boolean fictional) {
        return new TagElement(elem, fictional);
    }

    /**
     * Makes a TagElement.
     *
     * @param elem  the element storing the tag definition
     *
     * @return the created {@code TagElement}
     */
    protected TagElement makeTag(Element elem) {
        return makeTag(elem, false);
    }

    /**
     * Returns attributes for the current tag.
     *
     * @return {@code SimpleAttributeSet} containing the attributes
     */
    protected SimpleAttributeSet getAttributes() {
        return attributes;
    }

    /**
     * Removes the current attributes.
     */
    protected void flushAttributes() {
        attributes.removeAttributes(attributes);
    }

    /**
     * Called when PCDATA is encountered.
     *
     * @param text  the section text
     */
    protected void handleText(char[] text) {
    }

    /**
     * Called when an HTML title tag is encountered.
     *
     * @param text  the title text
     */
    protected void handleTitle(char[] text) {
        // default behavior is to call handleText. Subclasses
        // can override if necessary.
        handleText(text);
    }

    /**
     * Called when an HTML comment is encountered.
     *
     * @param text  the comment being handled
     */
    protected void handleComment(char[] text) {
    }

    /**
     * Called when the content terminates without closing the HTML comment.
     */
    protected void handleEOFInComment() {
        // We've reached EOF.  Our recovery strategy is to
        // see if we have more than one line in the comment;
        // if so, we pretend that the comment was an unterminated
        // single line comment, and reparse the lines after the
        // first line as normal HTML content.

        int commentEndPos = strIndexOf('\n');
        if (commentEndPos >= 0) {
            handleComment(getChars(0, commentEndPos));
            try {
                in.close();
                in = new CharArrayReader(getChars(commentEndPos + 1));
                ch = '>';
            } catch (IOException e) {
                error("ioexception");
            }

            resetStrBuffer();
        } else {
            // no newline, so signal an error
            error("eof.comment");
        }
    }

    /**
     * Called when an empty tag is encountered.
     *
     * @param tag  the tag being handled
     * @throws ChangedCharSetException if the document charset was changed
     */
    protected void handleEmptyTag(TagElement tag) throws ChangedCharSetException {
    }

    /**
     * Called when a start tag is encountered.
     *
     * @param tag  the tag being handled
     */
    protected void handleStartTag(TagElement tag) {
    }

    /**
     * Called when an end tag is encountered.
     *
     * @param tag  the tag being handled
     */
    protected void handleEndTag(TagElement tag) {
    }

    /**
     * An error has occurred.
     *
     * @param ln   the number of line containing the error
     * @param msg  the error message
     */
    protected void handleError(int ln, String msg) {
        /*
        Thread.dumpStack();
        System.out.println("**** " + stack);
        System.out.println("line " + ln + ": error: " + msg);
        System.out.println();
        */
    }

    /**
     * Output text.
     */
    void handleText(TagElement tag) {
        if (tag.breaksFlow()) {
            space = false;
            if (!strict) {
                ignoreSpace = true;
            }
        }
        if (textpos == 0) {
            if ((!space) || (stack == null) || last.breaksFlow() ||
                !stack.advance(dtd.pcdata)) {
                last = tag;
                space = false;
                lastBlockStartPos = currentBlockStartPos;
                return;
            }
        }
        if (space) {
            if (!ignoreSpace) {
                // enlarge buffer if needed
                if (textpos + 1 > text.length) {
                    char[] newtext = new char[text.length + 200];
                    System.arraycopy(text, 0, newtext, 0, text.length);
                    text = newtext;
                }

                // output pending space
                text[textpos++] = ' ';
                if (!strict && !tag.getElement().isEmpty()) {
                    ignoreSpace = true;
                }
            }
            space = false;
        }
        char[] newtext = new char[textpos];
        System.arraycopy(text, 0, newtext, 0, textpos);
        // Handles cases of bad html where the title tag
        // was getting lost when we did error recovery.
        if (tag.getElement().getName().equals("title")) {
            handleTitle(newtext);
        } else {
            handleText(newtext);
        }
        lastBlockStartPos = currentBlockStartPos;
        textpos = 0;
        last = tag;
        space = false;
    }

    /**
     * Invokes the error handler.
     *
     * @param err   the error type
     * @param arg1  the 1st error message argument
     * @param arg2  the 2nd error message argument
     * @param arg3  the 3rd error message argument
     */
    protected void error(String err, String arg1, String arg2,
        String arg3) {
        handleError(ln, err + " " + arg1 + " " + arg2 + " " + arg3);
    }

    /**
     * Invokes the error handler with the 3rd error message argument "?".
     *
     * @param err   the error type
     * @param arg1  the 1st error message argument
     * @param arg2  the 2nd error message argument
     */
    protected void error(String err, String arg1, String arg2) {
        error(err, arg1, arg2, "?");
    }

    /**
     * Invokes the error handler with the 2nd and 3rd error message argument "?".
     *
     * @param err   the error type
     * @param arg1  the 1st error message argument
     */
    protected void error(String err, String arg1) {
        error(err, arg1, "?", "?");
    }

    /**
     * Invokes the error handler with the 1st, 2nd and 3rd error message argument "?".
     *
     * @param err   the error type
     */
    protected void error(String err) {
        error(err, "?", "?", "?");
    }


    /**
     * Handle a start tag. The new tag is pushed
     * onto the tag stack. The attribute list is
     * checked for required attributes.
     *
     * @param tag  the tag
     * @throws ChangedCharSetException if the document charset was changed
     */
    protected void startTag(TagElement tag) throws ChangedCharSetException {
        Element elem = tag.getElement();

        // If the tag is an empty tag and texpos != 0
        // this implies that there is text before the
        // start tag that needs to be processed before
        // handling the tag.
        //
        if (!elem.isEmpty() ||
                    ((last != null) && !last.breaksFlow()) ||
                    (textpos != 0)) {
            handleText(tag);
        } else {
            // this variable gets updated in handleText().
            // Since in this case we do not call handleText()
            // we need to update it here.
            //
            last = tag;
            // Note that we should really check last.breakFlows before
            // assuming this should be false.
            space = false;
        }
        lastBlockStartPos = currentBlockStartPos;

        // check required attributes
        for (AttributeList a = elem.atts ; a != null ; a = a.next) {
            if ((a.modifier == REQUIRED) &&
                ((attributes.isEmpty()) ||
                 ((!attributes.isDefined(a.name)) &&
                  (!attributes.isDefined(HTML.getAttributeKey(a.name)))))) {
                error("req.att ", a.getName(), elem.getName());
            }
        }

        if (elem.isEmpty()) {
            handleEmptyTag(tag);
            /*
        } else if (elem.getName().equals("form")) {
            handleStartTag(tag);
            */
        } else {
            recent = elem;
            stack = new TagStack(tag, stack);
            handleStartTag(tag);
        }
    }

    /**
     * Handle an end tag. The end tag is popped
     * from the tag stack.
     *
     * @param omitted  {@code true} if the tag is no actually present in the
     *                 document, but is supposed by the parser
     */
    protected void endTag(boolean omitted) {
        handleText(stack.tag);

        if (omitted && !stack.elem.omitEnd()) {
            error("end.missing", stack.elem.getName());
        } else if (!stack.terminate()) {
            error("end.unexpected", stack.elem.getName());
        }

        // handle the tag
        handleEndTag(stack.tag);
        stack = stack.next;
        recent = (stack != null) ? stack.elem : null;
    }


    boolean ignoreElement(Element elem) {

        String stackElement = stack.elem.getName();
        String elemName = elem.getName();
        /* We ignore all elements that are not valid in the context of
           a table except <td>, <th> (these we handle in
           legalElementContext()) and #pcdata.  We also ignore the
           <font> tag in the context of <ul> and <ol> We additionally
           ignore the <meta> and the <style> tag if the body tag has
           been seen. **/
        if ((elemName.equals("html") && seenHtml) ||
            (elemName.equals("head") && seenHead) ||
            (elemName.equals("body") && seenBody)) {
            return true;
        }
        if (elemName.equals("dt") || elemName.equals("dd")) {
            TagStack s = stack;
            while (s != null && !s.elem.getName().equals("dl")) {
                s = s.next;
            }
            if (s == null) {
                return true;
            }
        }

        if (((stackElement.equals("table")) &&
             (!elemName.equals("#pcdata")) && (!elemName.equals("input"))) ||
            ((elemName.equals("font")) &&
             (stackElement.equals("ul") || stackElement.equals("ol"))) ||
            (elemName.equals("meta") && stack != null) ||
            (elemName.equals("style") && seenBody) ||
            (stackElement.equals("table") && elemName.equals("a"))) {
            return true;
        }
        return false;
    }


    /**
     * Marks the first time a tag has been seen in a document
     *
     * @param elem  the element represented by the tag
     */

    protected void markFirstTime(Element elem) {
        String elemName = elem.getName();
        if (elemName.equals("html")) {
            seenHtml = true;
        } else if (elemName.equals("head")) {
            seenHead = true;
        } else if (elemName.equals("body")) {
            if (buf.length == 1) {
                // Refer to note in definition of buf for details on this.
                char[] newBuf = new char[256];

                newBuf[0] = buf[0];
                buf = newBuf;
            }
            seenBody = true;
        }
    }

    /**
     * Create a legal content for an element.
     */
    boolean legalElementContext(Element elem) throws ChangedCharSetException {

        // System.out.println("-- legalContext -- " + elem);

        // Deal with the empty stack
        if (stack == null) {
            // System.out.println("-- stack is empty");
            if (elem != dtd.html) {
                // System.out.println("-- pushing html");
                startTag(makeTag(dtd.html, true));
                return legalElementContext(elem);
            }
            return true;
        }

        // Is it allowed in the current context
        if (stack.advance(elem)) {
            // System.out.println("-- legal context");
            markFirstTime(elem);
            return true;
        }
        boolean insertTag = false;

        // The use of all error recovery strategies are contingent
        // on the value of the strict property.
        //
        // These are commonly occurring errors.  if insertTag is true,
        // then we want to adopt an error recovery strategy that
        // involves attempting to insert an additional tag to
        // legalize the context.  The two errors addressed here
        // are:
        // 1) when a <td> or <th> is seen soon after a <table> tag.
        //    In this case we insert a <tr>.
        // 2) when any other tag apart from a <tr> is seen
        //    in the context of a <tr>.  In this case we would
        //    like to add a <td>.  If a <tr> is seen within a
        //    <tr> context, then we will close out the current
        //    <tr>.
        //
        // This insertion strategy is handled later in the method.
        // The reason for checking this now, is that in other cases
        // we would like to apply other error recovery strategies for example
        // ignoring tags.
        //
        // In certain cases it is better to ignore a tag than try to
        // fix the situation.  So the first test is to see if this
        // is what we need to do.
        //
        String stackElemName = stack.elem.getName();
        String elemName = elem.getName();


        if (!strict &&
            ((stackElemName.equals("table") && elemName.equals("td")) ||
             (stackElemName.equals("table") && elemName.equals("th")) ||
             (stackElemName.equals("tr") && !elemName.equals("tr")))){
             insertTag = true;
        }


        if (!strict && !insertTag && (stack.elem.getName() != elem.getName() ||
                                      elem.getName().equals("body"))) {
            if (skipTag = ignoreElement(elem)) {
                error("tag.ignore", elem.getName());
                return skipTag;
            }
        }

        // Check for anything after the start of the table besides tr, td, th
        // or caption, and if those aren't there, insert the <tr> and call
        // legalElementContext again.
        if (!strict && stackElemName.equals("table") &&
            !elemName.equals("tr") && !elemName.equals("td") &&
            !elemName.equals("th") && !elemName.equals("caption")) {
            Element e = dtd.getElement("tr");
            TagElement t = makeTag(e, true);
            legalTagContext(t);
            startTag(t);
            error("start.missing", elem.getName());
            return legalElementContext(elem);
        }

        // They try to find a legal context by checking if the current
        // tag is valid in an enclosing context.  If so
        // close out the tags by outputting end tags and then
        // insert the current tag.  If the tags that are
        // being closed out do not have an optional end tag
        // specification in the DTD then an html error is
        // reported.
        //
        if (!insertTag && stack.terminate() && (!strict || stack.elem.omitEnd())) {
            for (TagStack s = stack.next ; s != null ; s = s.next) {
                if (s.advance(elem)) {
                    while (stack != s) {
                        endTag(true);
                    }
                    return true;
                }
                if (!s.terminate() || (strict && !s.elem.omitEnd())) {
                    break;
                } else if (s.terminate() && !s.elem.omitEnd()) {
                    // Since the current tag is not valid in current context
                    // as otherwise s.advance(elem) would have returned true
                    // so check if the stack is to be terminated
                    // in which case return false
                    // but not if the closing tag is optional like tr,th,td
                    return false;
                }
            }
        }

        // Check if we know what tag is expected next.
        // If so insert the tag.  Report an error if the
        // tag does not have its start tag spec in the DTD as optional.
        //
        Element next = stack.first();
        if (next != null && (!strict || next.omitStart()) &&
           !(next==dtd.head && elem==dtd.pcdata) ) {
            // System.out.println("-- omitting start tag: " + next);
            TagElement t = makeTag(next, true);
            legalTagContext(t);
            startTag(t);
            if (!next.omitStart()) {
                error("start.missing", elem.getName());
            }
            return legalElementContext(elem);
        }


        // Traverse the list of expected elements and determine if adding
        // any of these elements would make for a legal context.
        //

        if (!strict) {
            ContentModel content = stack.contentModel();
            Vector<Element> elemVec = new Vector<Element>();
            if (content != null) {
                content.getElements(elemVec);
                for (Element e : elemVec) {
                    // Ensure that this element has not been included as
                    // part of the exclusions in the DTD.
                    //
                    if (stack.excluded(e.getIndex())) {
                        continue;
                    }

                    boolean reqAtts = false;

                    for (AttributeList a = e.getAttributes(); a != null ; a = a.next) {
                        if (a.modifier == REQUIRED) {
                            reqAtts = true;
                            break;
                        }
                    }
                    // Ensure that no tag that has required attributes
                    // gets inserted.
                    //
                    if (reqAtts) {
                        continue;
                    }

                    ContentModel m = e.getContent();
                    if (m != null && m.first(elem)) {
                        // System.out.println("-- adding a legal tag: " + e);
                        TagElement t = makeTag(e, true);
                        legalTagContext(t);
                        startTag(t);
                        error("start.missing", e.getName());
                        return legalElementContext(elem);
                    }
                }
            }
        }

        // Check if the stack can be terminated.  If so add the appropriate
        // end tag.  Report an error if the tag being ended does not have its
        // end tag spec in the DTD as optional.
        //
        if (stack.terminate() && (stack.elem != dtd.body) && (!strict || stack.elem.omitEnd())) {
            // System.out.println("-- omitting end tag: " + stack.elem);
            if (!stack.elem.omitEnd()) {
                error("end.missing", elem.getName());
            }

            endTag(true);
            return legalElementContext(elem);
        }

        // At this point we know that something is screwed up.
        return false;
    }

    /**
     * Create a legal context for a tag.
     */
    void legalTagContext(TagElement tag) throws ChangedCharSetException {
        if (legalElementContext(tag.getElement())) {
            markFirstTime(tag.getElement());
            return;
        }

        // Avoid putting a block tag in a flow tag.
        if (tag.breaksFlow() && (stack != null) && !stack.tag.breaksFlow()) {
            endTag(true);
            legalTagContext(tag);
            return;
        }

        // Avoid putting something weird in the head of the document.
        for (TagStack s = stack ; s != null ; s = s.next) {
            if (s.tag.getElement() == dtd.head) {
                while (stack != s) {
                    endTag(true);
                }
                endTag(true);
                legalTagContext(tag);
                return;
            }
        }

        // Everything failed
        error("tag.unexpected", tag.getElement().getName());
    }

    /**
     * Error context. Something went wrong, make sure we are in
     * the document's body context
     */
    void errorContext() throws ChangedCharSetException {
        for (; (stack != null) && (stack.tag.getElement() != dtd.body) ; stack = stack.next) {
            handleEndTag(stack.tag);
        }
        if (stack == null) {
            legalElementContext(dtd.body);
            startTag(makeTag(dtd.body, true));
        }
    }

    /**
     * Add a char to the string buffer.
     */
    void addString(int c) {
        if (strpos  == str.length) {
            char[] newstr = new char[str.length + 128];
            System.arraycopy(str, 0, newstr, 0, str.length);
            str = newstr;
        }
        str[strpos++] = (char)c;
    }

    /**
     * Get the string that's been accumulated.
     */
    String getString(int pos) {
        char[] newStr = new char[strpos - pos];
        System.arraycopy(str, pos, newStr, 0, strpos - pos);
        strpos = pos;
        return new String(newStr);
    }

    char[] getChars(int pos) {
        char[] newStr = new char[strpos - pos];
        System.arraycopy(str, pos, newStr, 0, strpos - pos);
        strpos = pos;
        return newStr;
    }

    char[] getChars(int pos, int endPos) {
        char[] newStr = new char[endPos - pos];
        System.arraycopy(str, pos, newStr, 0, endPos - pos);
        // REMIND: it's not clear whether this version should set strpos or not
        // strpos = pos;
        return newStr;
    }

    void resetStrBuffer() {
        strpos = 0;
    }

    int strIndexOf(char target) {
        for (int i = 0; i < strpos; i++) {
            if (str[i] == target) {
                return i;
            }
        }

        return -1;
    }

    /**
     * Skip space.
     * [5] 297:5
     */
    void skipSpace() throws IOException {
        while (true) {
            switch (ch) {
              case '\n':
                ln++;
                ch = readCh();
                lfCount++;
                break;

              case '\r':
                ln++;
                if ((ch = readCh()) == '\n') {
                    ch = readCh();
                    crlfCount++;
                }
                else {
                    crCount++;
                }
                break;
              case ' ':
              case '\t':
                ch = readCh();
                break;

              default:
                return;
            }
        }
    }

    /**
     * Parse identifier. Uppercase characters are folded
     * to lowercase when lower is true. Returns falsed if
     * no identifier is found. [55] 346:17
     */
    boolean parseIdentifier(boolean lower) throws IOException {
        switch (ch) {
          case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
          case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
          case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
          case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
          case 'Y': case 'Z':
            if (lower) {
                ch = 'a' + (ch - 'A');
            }
            break;

          case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
          case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
          case 'm': case 'n': case 'o': case 'p': case 'q': case 'r':
          case 's': case 't': case 'u': case 'v': case 'w': case 'x':
          case 'y': case 'z':
            break;

          default:
            return false;
        }

        while (true) {
            addString(ch);

            switch (ch = readCh()) {
              case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
              case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
              case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
              case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
              case 'Y': case 'Z':
                if (lower) {
                    ch = 'a' + (ch - 'A');
                }
                break;

              case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
              case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
              case 'm': case 'n': case 'o': case 'p': case 'q': case 'r':
              case 's': case 't': case 'u': case 'v': case 'w': case 'x':
              case 'y': case 'z':

              case '0': case '1': case '2': case '3': case '4':
              case '5': case '6': case '7': case '8': case '9':

              case '.': case '-':

              case '_': // not officially allowed
                break;

              default:
                return true;
            }
        }
    }

    /**
     * Parse an entity reference. [59] 350:17
     */
    private char[] parseEntityReference() throws IOException {
        int pos = strpos;

        if ((ch = readCh()) == '#') {
            int n = 0;
            ch = readCh();
            if ((ch >= '0') && (ch <= '9') ||
                    ch == 'x' || ch == 'X') {

                if ((ch >= '0') && (ch <= '9')) {
                    // parse decimal reference
                    while ((ch >= '0') && (ch <= '9')) {
                        n = (n * 10) + ch - '0';
                        ch = readCh();
                    }
                } else {
                    // parse hexadecimal reference
                    ch = readCh();
                    char lch = (char) Character.toLowerCase(ch);
                    while ((lch >= '0') && (lch <= '9') ||
                            (lch >= 'a') && (lch <= 'f')) {
                        if (lch >= '0' && lch <= '9') {
                            n = (n * 16) + lch - '0';
                        } else {
                            n = (n * 16) + lch - 'a' + 10;
                        }
                        ch = readCh();
                        lch = (char) Character.toLowerCase(ch);
                    }
                }
                switch (ch) {
                    case '\n':
                        ln++;
                        ch = readCh();
                        lfCount++;
                        break;

                    case '\r':
                        ln++;
                        if ((ch = readCh()) == '\n') {
                            ch = readCh();
                            crlfCount++;
                        }
                        else {
                            crCount++;
                        }
                        break;

                    case ';':
                        ch = readCh();
                        break;
                }
                char[] data = mapNumericReference(n);
                return data;
            }
            addString('#');
            if (!parseIdentifier(false)) {
                error("ident.expected");
                strpos = pos;
                char[] data = {'&', '#'};
                return data;
            }
        } else if (!parseIdentifier(false)) {
            char[] data = {'&'};
            return data;
        }

        boolean semicolon = false;

        switch (ch) {
          case '\n':
            ln++;
            ch = readCh();
            lfCount++;
            break;

          case '\r':
            ln++;
            if ((ch = readCh()) == '\n') {
                ch = readCh();
                crlfCount++;
            }
            else {
                crCount++;
            }
            break;

          case ';':
            semicolon = true;

            ch = readCh();
            break;
        }

        String nm = getString(pos);
        Entity ent = dtd.getEntity(nm);

        // entities are case sensitive - however if strict
        // is false then we will try to make a match by
        // converting the string to all lowercase.
        //
        if (!strict && (ent == null)) {
            ent = dtd.getEntity(nm.toLowerCase());
        }
        if ((ent == null) || !ent.isGeneral()) {

            if (nm.length() == 0) {
                error("invalid.entref", nm);
                return new char[0];
            }
            /* given that there is not a match restore the entity reference */
            String str = "&" + nm + (semicolon ? ";" : "");

            char[] b = new char[str.length()];
            str.getChars(0, b.length, b, 0);
            return b;
        }
        return ent.getData();
    }

    /**
     * Converts numeric character reference to char array.
     *
     * Normally the code in a reference should be always converted
     * to the Unicode character with the same code, but due to
     * wide usage of Cp1252 charset most browsers map numeric references
     * in the range 130-159 (which are control chars in Unicode set)
     * to displayable characters with other codes.
     *
     * @param c the code of numeric character reference.
     * @return a char array corresponding to the reference code.
     */
    private char[] mapNumericReference(int c) {
        char[] data;
        if (c >= 0xffff) { // outside unicode BMP.
            try {
                data = Character.toChars(c);
            } catch (IllegalArgumentException e) {
                data = new char[0];
            }
        } else {
            data = new char[1];
            data[0] = (c < 130 || c > 159) ? (char) c : cp1252Map[c - 130];
        }
        return data;
    }

    /**
     * Parse a comment. [92] 391:7
     */
    void parseComment() throws IOException {

        while (true) {
            int c = ch;
            switch (c) {
              case '-':
                  /** Presuming that the start string of a comment "<!--" has
                      already been parsed, the '-' character is valid only as
                      part of a comment termination and further more it must
                      be present in even numbers. Hence if strict is true, we
                      presume the comment has been terminated and return.
                      However if strict is false, then there is no even number
                      requirement and this character can appear anywhere in the
                      comment.  The parser reads on until it sees the following
                      pattern: "-->" or "--!>".
                   **/
                if (!strict && (strpos != 0) && (str[strpos - 1] == '-')) {
                    if ((ch = readCh()) == '>') {
                        return;
                    }
                    if (ch == '!') {
                        if ((ch = readCh()) == '>') {
                            return;
                        } else {
                            /* to account for extra read()'s that happened */
                            addString('-');
                            addString('!');
                            continue;
                        }
                    }
                    break;
                }

                if ((ch = readCh()) == '-') {
                    ch = readCh();
                    if (strict || ch == '>') {
                        return;
                    }
                    if (ch == '!') {
                        if ((ch = readCh()) == '>') {
                            return;
                        } else {
                            /* to account for extra read()'s that happened */
                            addString('-');
                            addString('!');
                            continue;
                        }
                    }
                    /* to account for the extra read() */
                    addString('-');
                }
                break;

              case -1:
                  handleEOFInComment();
                  return;

              case '\n':
                ln++;
                ch = readCh();
                lfCount++;
                break;

              case '>':
                ch = readCh();
                break;

              case '\r':
                ln++;
                if ((ch = readCh()) == '\n') {
                    ch = readCh();
                    crlfCount++;
                }
                else {
                    crCount++;
                }
                c = '\n';
                break;
              default:
                ch = readCh();
                break;
            }

            addString(c);
        }
    }

    /**
     * Parse literal content. [46] 343:1 and [47] 344:1
     */
    void parseLiteral(boolean replace) throws IOException {
        while (true) {
            int c = ch;
            switch (c) {
              case -1:
                error("eof.literal", stack.elem.getName());
                endTag(true);
                return;

              case '>':
                ch = readCh();
                int i = textpos - (stack.elem.name.length() + 2), j = 0;

                // match end tag
                if ((i >= 0) && (text[i++] == '<') && (text[i] == '/')) {
                    while ((++i < textpos) &&
                           (Character.toLowerCase(text[i]) == stack.elem.name.charAt(j++)));
                    if (i == textpos) {
                        textpos -= (stack.elem.name.length() + 2);
                        if ((textpos > 0) && (text[textpos-1] == '\n')) {
                            textpos--;
                        }
                        endTag(false);
                        return;
                    }
                }
                break;

              case '&':
                char[] data = parseEntityReference();
                if (textpos + data.length > text.length) {
                    char[] newtext = new char[Math.max(textpos + data.length + 128, text.length * 2)];
                    System.arraycopy(text, 0, newtext, 0, text.length);
                    text = newtext;
                }
                System.arraycopy(data, 0, text, textpos, data.length);
                textpos += data.length;
                continue;

              case '\n':
                ln++;
                ch = readCh();
                lfCount++;
                break;

              case '\r':
                ln++;
                if ((ch = readCh()) == '\n') {
                    ch = readCh();
                    crlfCount++;
                }
                else {
                    crCount++;
                }
                c = '\n';
                break;
              default:
                ch = readCh();
                break;
            }

            // output character
            if (textpos == text.length) {
                char[] newtext = new char[text.length + 128];
                System.arraycopy(text, 0, newtext, 0, text.length);
                text = newtext;
            }
            text[textpos++] = (char)c;
        }
    }

    /**
     * Parse attribute value. [33] 331:1
     */
    @SuppressWarnings("fallthrough")
    String parseAttributeValue(boolean lower) throws IOException {
        int delim = -1;

        // Check for a delimiter
        switch(ch) {
          case '\'':
          case '"':
            delim = ch;
            ch = readCh();
            break;
        }

        // Parse the rest of the value
        while (true) {
            int c = ch;

            switch (c) {
              case '\n':
                ln++;
                ch = readCh();
                lfCount++;
                if (delim < 0) {
                    return getString(0);
                }
                break;

              case '\r':
                ln++;

                if ((ch = readCh()) == '\n') {
                    ch = readCh();
                    crlfCount++;
                }
                else {
                    crCount++;
                }
                if (delim < 0) {
                    return getString(0);
                }
                break;

              case '\t':
                  if (delim < 0)
                      c = ' ';
                  // Fall through
              case ' ':
                ch = readCh();
                if (delim < 0) {
                    return getString(0);
                }
                break;

              case '>':
              case '<':
                if (delim < 0) {
                    return getString(0);
                }
                ch = readCh();
                break;

              case '\'':
              case '"':
                ch = readCh();
                if (c == delim) {
                    return getString(0);
                } else if (delim == -1) {
                    error("attvalerr");
                    if (strict || ch == ' ') {
                        return getString(0);
                    } else {
                        continue;
                    }
                }
                break;

            case '=':
                if (delim < 0) {
                    /* In SGML a construct like <img src=/cgi-bin/foo?x=1>
                       is considered invalid since an = sign can only be contained
                       in an attributes value if the string is quoted.
                       */
                    error("attvalerr");
                    /* If strict is true then we return with the string we have thus far.
                       Otherwise we accept the = sign as part of the attribute's value and
                       process the rest of the img tag. */
                    if (strict) {
                        return getString(0);
                    }
                }
                ch = readCh();
                break;

              case '&':
                if (strict && delim < 0) {
                    ch = readCh();
                    break;
                }

                char[] data = parseEntityReference();
                for (int i = 0 ; i < data.length ; i++) {
                    c = data[i];
                    addString((lower && (c >= 'A') && (c <= 'Z')) ? 'a' + c - 'A' : c);
                }
                continue;

              case -1:
                return getString(0);

              default:
                if (lower && (c >= 'A') && (c <= 'Z')) {
                    c = 'a' + c - 'A';
                }
                ch = readCh();
                break;
            }
            addString(c);
        }
    }


    /**
     * Parse attribute specification List. [31] 327:17
     */
    void parseAttributeSpecificationList(Element elem) throws IOException {

        while (true) {
            skipSpace();

            switch (ch) {
              case '/':
              case '>':
              case '<':
              case -1:
                return;

              case '-':
                if ((ch = readCh()) == '-') {
                    ch = readCh();
                    parseComment();
                    strpos = 0;
                } else {
                    error("invalid.tagchar", "-", elem.getName());
                    ch = readCh();
                }
                continue;
            }

            AttributeList att;
            String attname;
            String attvalue;

            if (parseIdentifier(true)) {
                attname = getString(0);
                skipSpace();
                if (ch == '=') {
                    ch = readCh();
                    skipSpace();
                    att = elem.getAttribute(attname);
//  Bug ID 4102750
//  Load the NAME of an Attribute Case Sensitive
//  The case of the NAME  must be intact
//  MG 021898
                    attvalue = parseAttributeValue((att != null) && (att.type != CDATA) && (att.type != NOTATION) && (att.type != NAME));
//                  attvalue = parseAttributeValue((att != null) && (att.type != CDATA) && (att.type != NOTATION));
                } else {
                    attvalue = attname;
                    att = elem.getAttributeByValue(attvalue);
                    if (att == null) {
                        att = elem.getAttribute(attname);
                        if (att != null) {
                            attvalue = att.getValue();
                        }
                        else {
                            // Make it null so that NULL_ATTRIBUTE_VALUE is
                            // used
                            attvalue = null;
                        }
                    }
                }
            } else if (!strict && ch == ',') { // allows for comma separated attribute-value pairs
                ch = readCh();
                continue;
            } else if (!strict && ch == '"') { // allows for quoted attributes
                ch = readCh();
                skipSpace();
                if (parseIdentifier(true)) {
                    attname = getString(0);
                    if (ch == '"') {
                        ch = readCh();
                    }
                    skipSpace();
                    if (ch == '=') {
                        ch = readCh();
                        skipSpace();
                        att = elem.getAttribute(attname);
                        attvalue = parseAttributeValue((att != null) &&
                                                (att.type != CDATA) &&
                                                (att.type != NOTATION));
                    } else {
                        attvalue = attname;
                        att = elem.getAttributeByValue(attvalue);
                        if (att == null) {
                            att = elem.getAttribute(attname);
                            if (att != null) {
                                attvalue = att.getValue();
                            }
                        }
                    }
                } else {
                    char[] str = {(char)ch};
                    error("invalid.tagchar", new String(str), elem.getName());
                    ch = readCh();
                    continue;
                }
            } else if (!strict && (attributes.isEmpty()) && (ch == '=')) {
                ch = readCh();
                skipSpace();
                attname = elem.getName();
                att = elem.getAttribute(attname);
                attvalue = parseAttributeValue((att != null) &&
                                               (att.type != CDATA) &&
                                               (att.type != NOTATION));
            } else if (!strict && (ch == '=')) {
                ch = readCh();
                skipSpace();
                attvalue = parseAttributeValue(true);
                error("attvalerr");
                return;
            } else {
                char[] str = {(char)ch};
                error("invalid.tagchar", new String(str), elem.getName());
                if (!strict) {
                    ch = readCh();
                    continue;
                } else {
                    return;
                }
            }

            if (att != null) {
                attname = att.getName();
            } else {
                error("invalid.tagatt", attname, elem.getName());
            }

            // Check out the value
            if (attributes.isDefined(attname)) {
                error("multi.tagatt", attname, elem.getName());
            }
            if (attvalue == null) {
                attvalue = ((att != null) && (att.value != null)) ? att.value :
                    HTML.NULL_ATTRIBUTE_VALUE;
            } else if ((att != null) && (att.values != null) && !att.values.contains(attvalue)) {
                error("invalid.tagattval", attname, elem.getName());
            }
            HTML.Attribute attkey = HTML.getAttributeKey(attname);
            if (attkey == null) {
                attributes.addAttribute(attname, attvalue);
            } else {
                attributes.addAttribute(attkey, attvalue);
            }
        }
    }

    /**
     * Parses the Document Type Declaration markup declaration.
     * Currently ignores it.
     *
     * @return the string representation of the markup declaration
     * @throws IOException if an I/O error occurs
     */
    public String parseDTDMarkup() throws IOException {

        StringBuilder strBuff = new StringBuilder();
        ch = readCh();
        while(true) {
            switch (ch) {
            case '>':
                ch = readCh();
                return strBuff.toString();
            case -1:
                error("invalid.markup");
                return strBuff.toString();
            case '\n':
                ln++;
                ch = readCh();
                lfCount++;
                break;
            case '"':
                ch = readCh();
                break;
            case '\r':
                ln++;
                if ((ch = readCh()) == '\n') {
                    ch = readCh();
                    crlfCount++;
                }
                else {
                    crCount++;
                }
                break;
            default:
                strBuff.append((char)(ch & 0xFF));
                ch = readCh();
                break;
            }
        }
    }

    /**
     * Parse markup declarations.
     * Currently only handles the Document Type Declaration markup.
     * Returns true if it is a markup declaration false otherwise.
     *
     * @param strBuff  the markup declaration
     * @return {@code true} if this is a valid markup declaration;
     *         otherwise {@code false}
     * @throws IOException if an I/O error occurs
     */
    protected boolean parseMarkupDeclarations(StringBuffer strBuff) throws IOException {

        /* Currently handles only the DOCTYPE */
        if ((strBuff.length() == "DOCTYPE".length()) &&
            (strBuff.toString().toUpperCase().equals("DOCTYPE"))) {
            parseDTDMarkup();
            return true;
        }
        return false;
    }

    /**
     * Parse an invalid tag.
     */
    void parseInvalidTag() throws IOException {
        // ignore all data up to the close bracket '>'
        while (true) {
            skipSpace();
            switch (ch) {
              case '>':
              case -1:
                  ch = readCh();
                return;
              case '<':
                  return;
              default:
                  ch = readCh();

            }
        }
    }

    /**
     * Parse a start or end tag.
     */
    @SuppressWarnings("fallthrough")
    void parseTag() throws IOException {
        Element elem;
        boolean net = false;
        boolean warned = false;
        boolean unknown = false;

        switch (ch = readCh()) {
          case '!':
            switch (ch = readCh()) {
              case '-':
                // Parse comment. [92] 391:7
                while (true) {
                    if (ch == '-') {
                        if (!strict || ((ch = readCh()) == '-')) {
                            ch = readCh();
                            if (!strict && ch == '-') {
                                ch = readCh();
                            }
                            // send over any text you might see
                            // before parsing and sending the
                            // comment
                            if (textpos != 0) {
                                char[] newtext = new char[textpos];
                                System.arraycopy(text, 0, newtext, 0, textpos);
                                handleText(newtext);
                                lastBlockStartPos = currentBlockStartPos;
                                textpos = 0;
                            }
                            parseComment();
                            last = makeTag(dtd.getElement("comment"), true);
                            handleComment(getChars(0));
                            continue;
                        } else if (!warned) {
                            warned = true;
                            error("invalid.commentchar", "-");
                        }
                    }
                    skipSpace();
                    switch (ch) {
                      case '-':
                        continue;
                      case '>':
                        ch = readCh();
                        return;
                      case -1:
                        return;
                      default:
                        ch = readCh();
                        if (!warned) {
                            warned = true;
                            error("invalid.commentchar",
                                  String.valueOf((char)ch));
                        }
                        break;
                    }
                }

              default:
                // deal with marked sections
                StringBuffer strBuff = new StringBuffer();
                while (true) {
                    strBuff.append((char)ch);
                    if (parseMarkupDeclarations(strBuff)) {
                        return;
                    }
                    switch(ch) {
                      case '>':
                        ch = readCh();
                        // Fall through
                      case -1:
                        error("invalid.markup");
                        return;
                      case '\n':
                        ln++;
                        ch = readCh();
                        lfCount++;
                        break;
                      case '\r':
                        ln++;
                        if ((ch = readCh()) == '\n') {
                            ch = readCh();
                            crlfCount++;
                        }
                        else {
                            crCount++;
                        }
                        break;

                      default:
                        ch = readCh();
                        break;
                    }
                }
            }

          case '/':
            // parse end tag [19] 317:4
            switch (ch = readCh()) {
              case '>':
                ch = readCh();
                // Fall through
              case '<':
                // empty end tag. either </> or </<
                if (recent == null) {
                    error("invalid.shortend");
                    return;
                }
                elem = recent;
                break;

              default:
                if (!parseIdentifier(true)) {
                    error("expected.endtagname");
                    return;
                }
                skipSpace();
                switch (ch) {
                  case '>':
                    ch = readCh();
                    break;
                  case '<':
                    break;

                  default:
                    error("expected", "'>'");
                    while ((ch != -1) && (ch != '\n') && (ch != '>')) {
                        ch = readCh();
                    }
                    if (ch == '>') {
                        ch = readCh();
                    }
                    break;
                }
                String elemStr = getString(0);
                if (!dtd.elementExists(elemStr)) {
                    error("end.unrecognized", elemStr);
                    // Ignore RE before end tag
                    if ((textpos > 0) && (text[textpos-1] == '\n')) {
                        textpos--;
                    }
                    elem = dtd.getElement("unknown");
                    elem.name = elemStr;
                    unknown = true;
                } else {
                    elem = dtd.getElement(elemStr);
                }
                break;
            }


            // If the stack is null, we're seeing end tags without any begin
            // tags.  Ignore them.

            if (stack == null) {
                error("end.extra.tag", elem.getName());
                return;
            }

            // Ignore RE before end tag
            if ((textpos > 0) && (text[textpos-1] == '\n')) {
                // In a pre tag, if there are blank lines
                // we do not want to remove the newline
                // before the end tag.  Hence this code.
                //
                if (stack.pre) {
                    if ((textpos > 1) && (text[textpos-2] != '\n')) {
                        textpos--;
                    }
                } else {
                    textpos--;
                }
            }

            // If the end tag is a form, since we did not put it
            // on the tag stack, there is no corresponding start
            // start tag to find. Hence do not touch the tag stack.
            //

            /*
            if (!strict && elem.getName().equals("form")) {
                if (lastFormSent != null) {
                    handleEndTag(lastFormSent);
                    return;
                } else {
                    // do nothing.
                    return;
                }
            }
            */

            if (unknown) {
                // we will not see a corresponding start tag
                // on the stack.  If we are seeing an
                // end tag, lets send this on as an empty
                // tag with the end tag attribute set to
                // true.
                TagElement t = makeTag(elem);
                handleText(t);
                attributes.addAttribute(HTML.Attribute.ENDTAG, "true");
                handleEmptyTag(makeTag(elem));
                unknown = false;
                return;
            }

            // find the corresponding start tag

            // A commonly occurring error appears to be the insertion
            // of extra end tags in a table.  The intent here is ignore
            // such extra end tags.
            //
            if (!strict) {
                String stackElem = stack.elem.getName();

                if (stackElem.equals("table")) {
                    // If it is not a valid end tag ignore it and return
                    //
                    if (!elem.getName().equals(stackElem)) {
                        error("tag.ignore", elem.getName());
                        return;
                    }
                }



                if (stackElem.equals("tr") ||
                    stackElem.equals("td")) {
                    if ((!elem.getName().equals("table")) &&
                        (!elem.getName().equals(stackElem))) {
                        error("tag.ignore", elem.getName());
                        return;
                    }
                }
            }
            TagStack sp = stack;

            while ((sp != null) && (elem != sp.elem)) {
                sp = sp.next;
            }
            if (sp == null) {
                error("unmatched.endtag", elem.getName());
                return;
            }

            // People put font ending tags in the darndest places.
            // Don't close other contexts based on them being between
            // a font tag and the corresponding end tag.  Instead,
            // ignore the end tag like it doesn't exist and allow the end
            // of the document to close us out.
            String elemName = elem.getName();
            if (stack != sp &&
                (elemName.equals("font") ||
                 elemName.equals("center"))) {

                // Since closing out a center tag can have real weird
                // effects on the formatting,  make sure that tags
                // for which omitting an end tag is legimitate
                // get closed out.
                //
                if (elemName.equals("center")) {
                    while(stack.elem.omitEnd() && stack != sp) {
                        endTag(true);
                    }
                    if (stack.elem == elem) {
                        endTag(false);
                    }
                }
                return;
            }
            // People do the same thing with center tags.  In this
            // case we would like to close off the center tag but
            // not necessarily all enclosing tags.



            // end tags
            while (stack != sp) {
                endTag(true);
            }

            endTag(false);
            return;

          case -1:
            error("eof");
            return;
        }

        // start tag [14] 314:1
        if (!parseIdentifier(true)) {
            elem = recent;
            if ((ch != '>') || (elem == null)) {
                error("expected.tagname");
                return;
            }
        } else {
            String elemStr = getString(0);

            if (elemStr.equals("image")) {
                elemStr = "img";
            }

            /* determine if this element is part of the dtd. */

            if (!dtd.elementExists(elemStr)) {
                //              parseInvalidTag();
                error("tag.unrecognized ", elemStr);
                elem = dtd.getElement("unknown");
                elem.name = elemStr;
                unknown = true;
            } else {
                elem = dtd.getElement(elemStr);
            }
        }

        // Parse attributes
        parseAttributeSpecificationList(elem);

        switch (ch) {
          case '/':
            net = true;
            // Fall through
          case '>':
            ch = readCh();
            if (ch == '>' && net) {
                ch = readCh();
            }
          case '<':
            break;

          default:
            error("expected", "'>'");
            break;
        }

        if (!strict) {
          if (elem.getName().equals("script")) {
            error("javascript.unsupported");
          }
        }

        // ignore RE after start tag
        //
        if (!elem.isEmpty())  {
            if (ch == '\n') {
                ln++;
                lfCount++;
                ch = readCh();
            } else if (ch == '\r') {
                ln++;
                if ((ch = readCh()) == '\n') {
                    ch = readCh();
                    crlfCount++;
                }
                else {
                    crCount++;
                }
            }
        }

        // ensure a legal context for the tag
        TagElement tag = makeTag(elem, false);


        /** In dealing with forms, we have decided to treat
            them as legal in any context.  Also, even though
            they do have a start and an end tag, we will
            not put this tag on the stack.  This is to deal
            several pages in the web oasis that choose to
            start and end forms in any possible location. **/

        /*
        if (!strict && elem.getName().equals("form")) {
            if (lastFormSent == null) {
                lastFormSent = tag;
            } else {
                handleEndTag(lastFormSent);
                lastFormSent = tag;
            }
        } else {
        */
            // Smlly, if a tag is unknown, we will apply
            // no legalTagContext logic to it.
            //
            if (!unknown) {
                legalTagContext(tag);

                // If skip tag is true,  this implies that
                // the tag was illegal and that the error
                // recovery strategy adopted is to ignore
                // the tag.
                if (!strict && skipTag) {
                    skipTag = false;
                    return;
                }
            }
            /*
        }
            */

        startTag(tag);

        if (!elem.isEmpty()) {
            switch (elem.getType()) {
              case CDATA:
                parseLiteral(false);
                break;
              case RCDATA:
                parseLiteral(true);
                break;
              default:
                if (stack != null) {
                    stack.net = net;
                }
                break;
            }
        }
    }

    private static final String START_COMMENT = "<!--";
    private static final String END_COMMENT = "-->";
    private static final char[] SCRIPT_END_TAG = "</script>".toCharArray();
    private static final char[] SCRIPT_END_TAG_UPPER_CASE =
                                        "</SCRIPT>".toCharArray();

    void parseScript() throws IOException {
        char[] charsToAdd = new char[SCRIPT_END_TAG.length];
        boolean insideComment = false;

        /* Here, ch should be the first character after <script> */
        while (true) {
            int i = 0;
            while (!insideComment && i < SCRIPT_END_TAG.length
                    && (SCRIPT_END_TAG[i] == ch
                    || SCRIPT_END_TAG_UPPER_CASE[i] == ch)) {
                charsToAdd[i] = (char) ch;
                ch = readCh();
                i++;
            }
            if (i == SCRIPT_END_TAG.length) {
                return;
            }

            if (!insideComment && i == 1 && charsToAdd[0] == START_COMMENT.charAt(0)) {
                // it isn't end script tag, but may be it's start comment tag?
                while (i < START_COMMENT.length()
                        && START_COMMENT.charAt(i) == ch) {
                    charsToAdd[i] = (char) ch;
                    ch = readCh();
                    i++;
                }
                if (i == START_COMMENT.length()) {
                    insideComment = true;
                }
            }
            if (insideComment) {
                while (i < END_COMMENT.length()
                        && END_COMMENT.charAt(i) == ch) {
                    charsToAdd[i] = (char) ch;
                    ch = readCh();
                    i++;
                }
                if (i == END_COMMENT.length()) {
                    insideComment = false;
                }
            }

            /* To account for extra read()'s that happened */
            if (i > 0) {
                for (int j = 0; j < i; j++) {
                    addString(charsToAdd[j]);
                }
                continue;
            }
            switch (ch) {
            case -1:
                error("eof.script");
                return;
            case '\n':
                ln++;
                ch = readCh();
                lfCount++;
                addString('\n');
                break;
            case '\r':
                ln++;
                if ((ch = readCh()) == '\n') {
                    ch = readCh();
                    crlfCount++;
                } else {
                    crCount++;
                }
                addString('\n');
                break;
            default:
                addString(ch);
                ch = readCh();
                break;
            } // switch
        } // while
    }

    /**
     * Parse Content. [24] 320:1
     */
    void parseContent() throws IOException {
        Thread curThread = Thread.currentThread();

        for (;;) {
            if (curThread.isInterrupted()) {
                curThread.interrupt(); // resignal the interrupt
                break;
            }

            int c = ch;
            currentBlockStartPos = currentPosition;

            if (recent == dtd.script) { // means: if after starting <script> tag

                /* Here, ch has to be the first character after <script> */
                parseScript();
                last = makeTag(dtd.getElement("comment"), true);

                /* Remove leading and trailing HTML comment declarations */
                String str = new String(getChars(0)).trim();
                int minLength = START_COMMENT.length() + END_COMMENT.length();
                if (str.startsWith(START_COMMENT) && str.endsWith(END_COMMENT)
                       && str.length() >= (minLength)) {
                    str = str.substring(START_COMMENT.length(),
                                      str.length() - END_COMMENT.length());
                }

                /* Handle resulting chars as comment */
                handleComment(str.toCharArray());
                endTag(false);
                lastBlockStartPos = currentPosition;

                continue;
            } else {
                switch (c) {
                  case '<':
                    parseTag();
                    lastBlockStartPos = currentPosition;
                    continue;

                  case '/':
                    ch = readCh();
                    if ((stack != null) && stack.net) {
                        // null end tag.
                        endTag(false);
                        continue;
                    } else if (textpos == 0) {
                        if (!legalElementContext(dtd.pcdata)) {
                            error("unexpected.pcdata");
                        }
                        if (last.breaksFlow()) {
                            space = false;
                        }
                    }
                    break;

                  case -1:
                    return;

                  case '&':
                    if (textpos == 0) {
                        if (!legalElementContext(dtd.pcdata)) {
                            error("unexpected.pcdata");
                        }
                        if (last.breaksFlow()) {
                            space = false;
                        }
                    }
                    char[] data = parseEntityReference();
                    if (textpos + data.length + 1 > text.length) {
                        char[] newtext = new char[Math.max(textpos + data.length + 128, text.length * 2)];
                        System.arraycopy(text, 0, newtext, 0, text.length);
                        text = newtext;
                    }
                    if (space) {
                        space = false;
                        text[textpos++] = ' ';
                    }
                    System.arraycopy(data, 0, text, textpos, data.length);
                    textpos += data.length;
                    ignoreSpace = false;
                    continue;

                  case '\n':
                    ln++;
                    lfCount++;
                    ch = readCh();
                    if ((stack != null) && stack.pre) {
                        break;
                    }
                    if (textpos == 0) {
                        lastBlockStartPos = currentPosition;
                    }
                    if (!ignoreSpace) {
                        space = true;
                    }
                    continue;

                  case '\r':
                    ln++;
                    c = '\n';
                    if ((ch = readCh()) == '\n') {
                        ch = readCh();
                        crlfCount++;
                    }
                    else {
                        crCount++;
                    }
                    if ((stack != null) && stack.pre) {
                        break;
                    }
                    if (textpos == 0) {
                        lastBlockStartPos = currentPosition;
                    }
                    if (!ignoreSpace) {
                        space = true;
                    }
                    continue;


                  case '\t':
                  case ' ':
                    ch = readCh();
                    if ((stack != null) && stack.pre) {
                        break;
                    }
                    if (textpos == 0) {
                        lastBlockStartPos = currentPosition;
                    }
                    if (!ignoreSpace) {
                        space = true;
                    }
                    continue;

                  default:
                    if (textpos == 0) {
                        if (!legalElementContext(dtd.pcdata)) {
                            error("unexpected.pcdata");
                        }
                        if (last.breaksFlow()) {
                            space = false;
                        }
                    }
                    ch = readCh();
                    break;
                }
            }

            // enlarge buffer if needed
            if (textpos + 2 > text.length) {
                char[] newtext = new char[text.length + 128];
                System.arraycopy(text, 0, newtext, 0, text.length);
                text = newtext;
            }

            // output pending space
            if (space) {
                if (textpos == 0) {
                    lastBlockStartPos--;
                }
                text[textpos++] = ' ';
                space = false;
            }
            text[textpos++] = (char)c;
            ignoreSpace = false;
        }
    }

    /**
     * Returns the end of line string. This will return the end of line
     * string that has been encountered the most, one of \r, \n or \r\n.
     */
    String getEndOfLineString() {
        if (crlfCount >= crCount) {
            if (lfCount >= crlfCount) {
                return "\n";
            }
            else {
                return "\r\n";
            }
        }
        else {
            if (crCount > lfCount) {
                return "\r";
            }
            else {
                return "\n";
            }
        }
    }

    /**
     * Parse an HTML stream, given a DTD.
     *
     * @param in  the reader to read the source from
     * @throws IOException if an I/O error occurs
     */
    public synchronized void parse(Reader in) throws IOException {
        this.in = in;

        this.ln = 1;

        seenHtml = false;
        seenHead = false;
        seenBody = false;

        crCount = lfCount = crlfCount = 0;

        try {
            ch = readCh();
            text = new char[1024];
            str = new char[128];

            parseContent();
            // NOTE: interruption may have occurred.  Control flows out
            // of here normally.
            while (stack != null) {
                endTag(true);
            }
            in.close();
        } catch (IOException e) {
            errorContext();
            error("ioexception");
            throw e;
        } catch (Exception e) {
            errorContext();
            error("exception", e.getClass().getName(), e.getMessage());
            e.printStackTrace();
        } finally {
            for (; stack != null ; stack = stack.next) {
                handleEndTag(stack.tag);
            }

            text = null;
            str = null;
        }

    }


    /*
     * Input cache.  This is much faster than calling down to a synchronized
     * method of BufferedReader for each byte.  Measurements done 5/30/97
     * show that there's no point in having a bigger buffer:  Increasing
     * the buffer to 8192 had no measurable impact for a program discarding
     * one character at a time (reading from an http URL to a local machine).
     * NOTE: If the current encoding is bogus, and we read too much
     * (past the content-type) we may suffer a MalformedInputException. For
     * this reason the initial size is 1 and when the body is encountered the
     * size is adjusted to 256.
     */
    private char[] buf = new char[1];
    private int pos;
    private int len;
    /*
        tracks position relative to the beginning of the
        document.
    */
    private int currentPosition;


    private int readCh() throws IOException {

        if (pos >= len) {

            // This loop allows us to ignore interrupts if the flag
            // says so
            for (;;) {
                try {
                    len = in.read(buf);
                    break;
                } catch (InterruptedIOException ex) {
                    throw ex;
                }
            }

            if (len <= 0) {
                return -1;      // eof
            }
            pos = 0;
        }
        ++currentPosition;

        return buf[pos++];
    }


    /**
     * Returns the current position.
     *
     * @return the current position
     */
    protected int getCurrentPos() {
        return currentPosition;
    }
}