fi.iki.hsivonen.htmlparser
Class HtmlParser

java.lang.Object
  fi.iki.hsivonen.htmlparser.HtmlParser

All Implemented Interfaces:

Locator, XMLReader

public final class HtmlParser
extends Object
implements XMLReader, Locator

WARNING: This parser is incomplete. It does not perform tag inference, yet. It does not yet perform case folding for attribute value like method="POST".

Version:

$Id: HtmlParser.java,v 1.20 2006/11/21 10:13:24 hsivonen Exp $

Author:

hsivonen

Field Summary

private boolean	alreadyWarnedAboutPrivateUseCharacters
private static char[]	APOS
private char[]	astralChar
private char[]	attrBuf
private int	attrBufLen
private AttributesImpl	attrs
private char[]	bmpChar
private char[]	buf
private int	bufLen
private static int	CASE_MASK
private int	cdataState
private CharacterEncodingDeclarationFilter	cedf
private ContentHandler	ch
private int	col
private int	cstart
private DoctypeHandler	doctypeHandler
private int	doctypeMode
private boolean	doctypeSeen
private DTDHandler	dtdHandler
private EmptyElementFilter	eef
private ErrorHandler	eh
private XhtmlSaxEmitter	emitter
private String	encoding
private EntityResolver	entityResolver
private boolean	foldedAttributeValue
private boolean	html5
private static int	LEAD_OFFSET
private int	line
private static char[]	LT
private boolean	nonWhiteSpaceAllowed
private NormalizationChecker	normalizationChecker
private static char[]	OCTYPE
private static int	PCDATA
private ContentHandlerFilter	pipelineLast
private int	pos
private char	prev
private String	publicId
private Reader	reader
private static int	SCRIPT
private char[]	strBuf
private int	strBufLen
private InputStream	stream
private static int	STYLE
private static int	SURROGATE_OFFSET
private String	systemId
private TagInferenceFilter	tif
private static char[]	TML
private static char[]	UBLIC
private boolean	wasLt

Constructor Summary

HtmlParser()

Method Summary

private void	appendAttrBuf(char c)
private void	appendAttrBuf(char[] cs)
private void	appendAttrBufAsciiLowerCase(char c)
private void	appendAttrBufAsciiLowerCase(char[] cs)
private void	appendStrBuf(char c)
private void	appendStrBufAsciiLowerCase(char c)
private String	attrBufToString()
private void	cannotDetermineEncoding()
private void	cdataStateEnd(String gi)
private void	checkPublicAndSystemIds(String publicId, String systemId)
private void	clearAttrBuf()
private void	clearStrBuf()
private char	consumeAttribute(char c)
private boolean	consumeCaseInsensitiveAsciiLetterString(char[] str)
private char[]	consumeCharRef()
private char	consumeComment()
private void	consumeDoctype()
private void	consumeEndTag()
private char[]	consumeEntityRef(char c)
private void	consumeMarkup()
private void	consumeMarkupDecl()
private char[]	consumeNCR()
private void	consumePI()
private void	consumeQuotedAttributeValue(char delim)
private void	consumeStartTag(char c)
private void	doctypeNotOk()
private Reader	draconianInputStreamReader(String encoding, InputStream stream, boolean requireAsciiSuperset)
private void	err(String message)
private void	fatal(String message)
private void	fatalIfAttributeExists(String name)
private void	flushChars()
int	getColumnNumber()
ContentHandler	getContentHandler()
DoctypeHandler	getDoctypeHandler() Returns the doctypeHandler.
int	getDoctypeMode() Returns the doctypeMode.
DTDHandler	getDTDHandler()
EntityResolver	getEntityResolver()
ErrorHandler	getErrorHandler()
boolean	getFeature(String key)
int	getLineNumber()
Object	getProperty(String key)
String	getPublicId()
String	getSystemId()
private boolean	isAstralPrivateUse(int c)
private boolean	isForbidden(char c)
private boolean	isNameChar(char c)
private boolean	isNameStart(char c)
private boolean	isNonCharacter(int c)
private boolean	isPrivateUse(char c)
private boolean	isUnquotedAttributeChar(char c)
private boolean	isWhiteSpace(char c)
private void	maybeBeginCdata(String gi)
private char	next()
private char	nextAfterZeroOrMoreWhiteSpace()
private char	nextMayEnd()
private void	parse()
void	parse(InputSource is)
void	parse(String url)
void	refireStart()
private void	sawHtml5Doctype()
void	setContentHandler(ContentHandler ch)
void	setDoctypeHandler(DoctypeHandler doctypeHandler) Sets the doctypeHandler.
void	setDoctypeMode(int doctypeMode) Sets the doctypeMode.
void	setDTDHandler(DTDHandler handler)
(package private) void	setEncoding(String enc)
void	setEntityResolver(EntityResolver entityResolver)
void	setErrorHandler(ErrorHandler eh)
void	setFeature(String key, boolean value)
(package private) void	setNonWhiteSpaceAllowed(boolean allow)
void	setProperty(String key, Object value)
private String	strBufToString()
private void	streamSetup(InputSource is)
private String	unescapedStringUntil(char delim)
private void	warn(String message)
private void	warnAboutPrivateUseChar()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

CASE_MASK

private static final int CASE_MASK

See Also:

Constant Field Values

LEAD_OFFSET

private static final int LEAD_OFFSET

See Also:

Constant Field Values

SURROGATE_OFFSET

private static final int SURROGATE_OFFSET

See Also:

Constant Field Values

LT

private static final char[] LT

APOS

private static final char[] APOS

OCTYPE

private static final char[] OCTYPE

TML

private static final char[] TML

UBLIC

private static final char[] UBLIC

PCDATA

private static final int PCDATA

See Also:

Constant Field Values

SCRIPT

private static final int SCRIPT

See Also:

Constant Field Values

STYLE

private static final int STYLE

See Also:

Constant Field Values

publicId

private String publicId

systemId

private String systemId

nonWhiteSpaceAllowed

private boolean nonWhiteSpaceAllowed

cdataState

private int cdataState

eh

private ErrorHandler eh

ch

private ContentHandler ch

doctypeHandler

private DoctypeHandler doctypeHandler

emitter

private XhtmlSaxEmitter emitter

reader

private Reader reader

pos

private int pos

cstart

private int cstart

buf

private char[] buf

bufLen

private int bufLen

line

private int line

col

private int col

doctypeSeen

private boolean doctypeSeen

doctypeMode

private int doctypeMode

html5

private boolean html5

prev

private char prev

wasLt

private boolean wasLt

strBuf

private char[] strBuf

strBufLen

private int strBufLen

attrBuf

private char[] attrBuf

attrBufLen

private int attrBufLen

attrs

private AttributesImpl attrs

bmpChar

private char[] bmpChar

astralChar

private char[] astralChar

dtdHandler

private DTDHandler dtdHandler

eef

private EmptyElementFilter eef

tif

private TagInferenceFilter tif

cedf

private CharacterEncodingDeclarationFilter cedf

pipelineLast

private ContentHandlerFilter pipelineLast

entityResolver

private EntityResolver entityResolver

encoding

private String encoding

stream

private InputStream stream

foldedAttributeValue

private boolean foldedAttributeValue

alreadyWarnedAboutPrivateUseCharacters

private boolean alreadyWarnedAboutPrivateUseCharacters

normalizationChecker

private NormalizationChecker normalizationChecker

Constructor Detail

HtmlParser

public HtmlParser()

Method Detail

clearStrBuf

private void clearStrBuf()

appendStrBufAsciiLowerCase

private void appendStrBufAsciiLowerCase(char c)
                                 throws SAXException,
                                        IOException

Throws:

SAXException

IOException

appendStrBuf

private void appendStrBuf(char c)
                   throws SAXException,
                          IOException

Throws:

SAXException

IOException

strBufToString

private String strBufToString()

clearAttrBuf

private void clearAttrBuf()

appendAttrBuf

private void appendAttrBuf(char c)
                    throws SAXException,
                           IOException

Throws:

SAXException

IOException

appendAttrBufAsciiLowerCase

private void appendAttrBufAsciiLowerCase(char c)
                                  throws SAXException,
                                         IOException

Throws:

SAXException

IOException

appendAttrBuf

private void appendAttrBuf(char[] cs)
                    throws SAXException,
                           IOException

Parameters:

cs -

Throws:

SAXException

IOException

appendAttrBufAsciiLowerCase

private void appendAttrBufAsciiLowerCase(char[] cs)
                                  throws SAXException,
                                         IOException

Parameters:

cs -

Throws:

SAXException

IOException

attrBufToString

private String attrBufToString()

parse

private void parse()
            throws SAXException,
                   IOException

Throws:

SAXException

IOException

doctypeNotOk

private void doctypeNotOk()
                   throws SAXException,
                          IOException

Throws:

SAXException

IOException

isWhiteSpace

private boolean isWhiteSpace(char c)

Parameters:

c -

Returns:

consumeCharRef

private char[] consumeCharRef()
                       throws SAXException,
                              IOException

Throws:

SAXException

IOException

consumeEntityRef

private char[] consumeEntityRef(char c)
                         throws SAXException,
                                IOException

Parameters:

c -

Throws:

SAXException

IOException

consumeNCR

private char[] consumeNCR()
                   throws SAXException,
                          IOException

Throws:

SAXException

IOException

consumeMarkup

private void consumeMarkup()
                    throws SAXException,
                           IOException

Throws:

SAXException

IOException

next

private char next()
           throws SAXException,
                  IOException

Returns:

Throws:

SAXException

IOException

consumeStartTag

private void consumeStartTag(char c)
                      throws SAXException,
                             IOException

Parameters:

c -

Throws:

SAXException

IOException

maybeBeginCdata

private void maybeBeginCdata(String gi)

Parameters:

gi -

consumeAttribute

private char consumeAttribute(char c)
                       throws SAXException,
                              IOException

Parameters:

c -

Returns:

Throws:

SAXException

IOException

isUnquotedAttributeChar

private boolean isUnquotedAttributeChar(char c)

Parameters:

c -

Returns:

consumeQuotedAttributeValue

private void consumeQuotedAttributeValue(char delim)
                                  throws SAXException,
                                         IOException

Parameters:

c -

Throws:

SAXException

IOException

fatalIfAttributeExists

private void fatalIfAttributeExists(String name)
                             throws SAXException,
                                    IOException

Parameters:

name -

Throws:

SAXException

IOException

isNameStart

private boolean isNameStart(char c)

Parameters:

c -

Returns:

isNameChar

private boolean isNameChar(char c)

Parameters:

c -

Returns:

consumeEndTag

private void consumeEndTag()
                    throws SAXException,
                           IOException

Throws:

SAXException

IOException

cdataStateEnd

private void cdataStateEnd(String gi)
                    throws SAXException,
                           IOException

Parameters:

gi -

Throws:

SAXException

IOException

nextAfterZeroOrMoreWhiteSpace

private char nextAfterZeroOrMoreWhiteSpace()
                                    throws SAXException,
                                           IOException

Returns:

Throws:

SAXException

IOException

consumePI

private void consumePI()
                throws SAXException,
                       IOException

Throws:

SAXException

IOException

consumeMarkupDecl

private void consumeMarkupDecl()
                        throws SAXException,
                               IOException

Throws:

SAXException

IOException

consumeDoctype

private void consumeDoctype()
                     throws SAXException,
                            IOException

Throws:

SAXException

IOException

sawHtml5Doctype

private void sawHtml5Doctype()
                      throws SAXException

Throws:

SAXException

checkPublicAndSystemIds

private void checkPublicAndSystemIds(String publicId,
                                     String systemId)
                              throws SAXException,
                                     IOException

Parameters:

publicId -

systemId -

Throws:

SAXException

IOException

unescapedStringUntil

private String unescapedStringUntil(char delim)
                             throws SAXException,
                                    IOException

Parameters:

c -

Returns:

Throws:

SAXException

IOException

consumeCaseInsensitiveAsciiLetterString

private boolean consumeCaseInsensitiveAsciiLetterString(char[] str)
                                                 throws SAXException,
                                                        IOException

Throws:

SAXException

IOException

consumeComment

private char consumeComment()
                     throws SAXException,
                            IOException

Throws:

SAXException

IOException

nextMayEnd

private char nextMayEnd()
                 throws SAXException,
                        IOException

Throws:

SAXException

IOException

warnAboutPrivateUseChar

private void warnAboutPrivateUseChar()
                              throws SAXException

Throws:

SAXException

isPrivateUse

private boolean isPrivateUse(char c)

isAstralPrivateUse

private boolean isAstralPrivateUse(int c)

isNonCharacter

private boolean isNonCharacter(int c)

Parameters:

intVal -

Returns:

isForbidden

private boolean isForbidden(char c)

Parameters:

c -

Returns:

flushChars

private void flushChars()
                 throws SAXException,
                        IOException

Throws:

SAXException

IOException

fatal

private void fatal(String message)
            throws SAXException

Throws:

SAXException

SAXParseException

err

private void err(String message)
          throws SAXException

Parameters:

string -

Throws:

SAXException

warn

private void warn(String message)
           throws SAXException

Parameters:

string -

Throws:

SAXException

getPublicId

public String getPublicId()

Specified by:

getPublicId in interface Locator

See Also:

Locator.getPublicId()

getSystemId

public String getSystemId()

Specified by:

getSystemId in interface Locator

See Also:

Locator.getSystemId()

getLineNumber

public int getLineNumber()

Specified by:

getLineNumber in interface Locator

See Also:

Locator.getLineNumber()

getColumnNumber

public int getColumnNumber()

Specified by:

getColumnNumber in interface Locator

See Also:

Locator.getColumnNumber()

getFeature

public boolean getFeature(String key)
                   throws SAXNotRecognizedException,
                          SAXNotSupportedException

Specified by:

getFeature in interface XMLReader

Throws:

SAXNotRecognizedException

SAXNotSupportedException

See Also:

XMLReader.getFeature(java.lang.String)

setFeature

public void setFeature(String key,
                       boolean value)
                throws SAXNotRecognizedException,
                       SAXNotSupportedException

Specified by:

setFeature in interface XMLReader

Throws:

SAXNotRecognizedException

SAXNotSupportedException

See Also:

XMLReader.setFeature(java.lang.String, boolean)

getProperty

public Object getProperty(String key)
                   throws SAXNotRecognizedException,
                          SAXNotSupportedException

Specified by:

getProperty in interface XMLReader

Throws:

SAXNotRecognizedException

SAXNotSupportedException

See Also:

XMLReader.getProperty(java.lang.String)

setProperty

public void setProperty(String key,
                        Object value)
                 throws SAXNotRecognizedException,
                        SAXNotSupportedException

Specified by:

setProperty in interface XMLReader

Throws:

SAXNotRecognizedException

SAXNotSupportedException

See Also:

XMLReader.setProperty(java.lang.String, java.lang.Object)

setEntityResolver

public void setEntityResolver(EntityResolver entityResolver)

Specified by:

setEntityResolver in interface XMLReader

See Also:

XMLReader.setEntityResolver(org.xml.sax.EntityResolver)

getEntityResolver

public EntityResolver getEntityResolver()

Specified by:

getEntityResolver in interface XMLReader

See Also:

XMLReader.getEntityResolver()

setDTDHandler

public void setDTDHandler(DTDHandler handler)

Specified by:

setDTDHandler in interface XMLReader

See Also:

XMLReader.setDTDHandler(org.xml.sax.DTDHandler)

getDTDHandler

public DTDHandler getDTDHandler()

Specified by:

getDTDHandler in interface XMLReader

See Also:

XMLReader.getDTDHandler()

setContentHandler

public void setContentHandler(ContentHandler ch)

Specified by:

setContentHandler in interface XMLReader

See Also:

XMLReader.setContentHandler(org.xml.sax.ContentHandler)

getContentHandler

public ContentHandler getContentHandler()

Specified by:

getContentHandler in interface XMLReader

See Also:

XMLReader.getContentHandler()

setErrorHandler

public void setErrorHandler(ErrorHandler eh)

Specified by:

setErrorHandler in interface XMLReader

See Also:

XMLReader.setErrorHandler(org.xml.sax.ErrorHandler)

getErrorHandler

public ErrorHandler getErrorHandler()

Specified by:

getErrorHandler in interface XMLReader

See Also:

XMLReader.getErrorHandler()

parse

public void parse(InputSource is)
           throws IOException,
                  SAXException

Specified by:

parse in interface XMLReader

Throws:

IOException

SAXException

See Also:

XMLReader.parse(org.xml.sax.InputSource)

streamSetup

private void streamSetup(InputSource is)
                  throws SAXException,
                         IOException

Parameters:

is -

swallowBom -

Throws:

IOException

SAXException

cannotDetermineEncoding

private void cannotDetermineEncoding()
                              throws IOException

Throws:

IOException

draconianInputStreamReader

private Reader draconianInputStreamReader(String encoding,
                                          InputStream stream,
                                          boolean requireAsciiSuperset)
                                   throws SAXException

Throws:

SAXException

parse

public void parse(String url)
           throws IOException,
                  SAXException

Specified by:

parse in interface XMLReader

Throws:

IOException

SAXException

See Also:

XMLReader.parse(java.lang.String)

setEncoding

void setEncoding(String enc)
           throws SAXException

Parameters:

string -

Throws:

SAXException

setNonWhiteSpaceAllowed

void setNonWhiteSpaceAllowed(boolean allow)

getDoctypeMode

public int getDoctypeMode()

Returns the doctypeMode.

Returns:

the doctypeMode

setDoctypeMode

public void setDoctypeMode(int doctypeMode)

Sets the doctypeMode.

Parameters:

doctypeMode - the doctypeMode to set

getDoctypeHandler

public DoctypeHandler getDoctypeHandler()

Returns the doctypeHandler.

Returns:

the doctypeHandler

setDoctypeHandler

public void setDoctypeHandler(DoctypeHandler doctypeHandler)

Sets the doctypeHandler.

Parameters:

doctypeHandler - the doctypeHandler to set

refireStart

public void refireStart()
                 throws SAXException

Throws:

SAXException