Package nu.validator.htmlparser.impl

Class Tokenizer

java.lang.Object
nu.validator.htmlparser.impl.Tokenizer
All Implemented Interfaces:
Locator
Direct Known Subclasses:
ErrorReportingTokenizer
public class Tokenizer extends Object implements Locator
An implementation of http://www.whatwg.org/specs/web-apps/current-work/multipage/tokenization.html This class implements the Locator interface. This is not an incidental implementation detail: Users of this class are encouraged to make use of the Locator nature. By default, the tokenizer may report data that XML 1.0 bans. The tokenizer can be configured to treat these conditions as fatal or to coerce the infoset to something that XML 1.0 allows.
Version:
$Id$

  • Field Details

    • DATA_AND_RCDATA_MASK

      private static final int DATA_AND_RCDATA_MASK
      See Also:

    • DATA

      public static final int DATA
      See Also:

    • RCDATA

      public static final int RCDATA
      See Also:

    • SCRIPT_DATA

      public static final int SCRIPT_DATA
      See Also:

    • RAWTEXT

      public static final int RAWTEXT
      See Also:

    • SCRIPT_DATA_ESCAPED

      public static final int SCRIPT_DATA_ESCAPED
      See Also:

    • ATTRIBUTE_VALUE_DOUBLE_QUOTED

      public static final int ATTRIBUTE_VALUE_DOUBLE_QUOTED
      See Also:

    • ATTRIBUTE_VALUE_SINGLE_QUOTED

      public static final int ATTRIBUTE_VALUE_SINGLE_QUOTED
      See Also:

    • ATTRIBUTE_VALUE_UNQUOTED

      public static final int ATTRIBUTE_VALUE_UNQUOTED
      See Also:

    • PLAINTEXT

      public static final int PLAINTEXT
      See Also:

    • TAG_OPEN

      public static final int TAG_OPEN
      See Also:

    • CLOSE_TAG_OPEN

      public static final int CLOSE_TAG_OPEN
      See Also:

    • TAG_NAME

      public static final int TAG_NAME
      See Also:

    • BEFORE_ATTRIBUTE_NAME

      public static final int BEFORE_ATTRIBUTE_NAME
      See Also:

    • ATTRIBUTE_NAME

      public static final int ATTRIBUTE_NAME
      See Also:

    • AFTER_ATTRIBUTE_NAME

      public static final int AFTER_ATTRIBUTE_NAME
      See Also:

    • BEFORE_ATTRIBUTE_VALUE

      public static final int BEFORE_ATTRIBUTE_VALUE
      See Also:

    • AFTER_ATTRIBUTE_VALUE_QUOTED

      public static final int AFTER_ATTRIBUTE_VALUE_QUOTED
      See Also:

    • BOGUS_COMMENT

      public static final int BOGUS_COMMENT
      See Also:

    • MARKUP_DECLARATION_OPEN

      public static final int MARKUP_DECLARATION_OPEN
      See Also:

    • DOCTYPE

      public static final int DOCTYPE
      See Also:

    • BEFORE_DOCTYPE_NAME

      public static final int BEFORE_DOCTYPE_NAME
      See Also:

    • DOCTYPE_NAME

      public static final int DOCTYPE_NAME
      See Also:

    • AFTER_DOCTYPE_NAME

      public static final int AFTER_DOCTYPE_NAME
      See Also:

    • BEFORE_DOCTYPE_PUBLIC_IDENTIFIER

      public static final int BEFORE_DOCTYPE_PUBLIC_IDENTIFIER
      See Also:

    • DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED

      public static final int DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED
      See Also:

    • DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED

      public static final int DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED
      See Also:

    • AFTER_DOCTYPE_PUBLIC_IDENTIFIER

      public static final int AFTER_DOCTYPE_PUBLIC_IDENTIFIER
      See Also:

    • BEFORE_DOCTYPE_SYSTEM_IDENTIFIER

      public static final int BEFORE_DOCTYPE_SYSTEM_IDENTIFIER
      See Also:

    • DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED

      public static final int DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED
      See Also:

    • DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED

      public static final int DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED
      See Also:

    • AFTER_DOCTYPE_SYSTEM_IDENTIFIER

      public static final int AFTER_DOCTYPE_SYSTEM_IDENTIFIER
      See Also:

    • BOGUS_DOCTYPE

      public static final int BOGUS_DOCTYPE
      See Also:

    • COMMENT_START

      public static final int COMMENT_START
      See Also:

    • COMMENT_START_DASH

      public static final int COMMENT_START_DASH
      See Also:

    • COMMENT

      public static final int COMMENT
      See Also:

    • COMMENT_END_DASH

      public static final int COMMENT_END_DASH
      See Also:

    • COMMENT_END

      public static final int COMMENT_END
      See Also:

    • COMMENT_END_BANG

      public static final int COMMENT_END_BANG
      See Also:

    • NON_DATA_END_TAG_NAME

      public static final int NON_DATA_END_TAG_NAME
      See Also:

    • MARKUP_DECLARATION_HYPHEN

      public static final int MARKUP_DECLARATION_HYPHEN
      See Also:

    • MARKUP_DECLARATION_OCTYPE

      public static final int MARKUP_DECLARATION_OCTYPE
      See Also:

    • DOCTYPE_UBLIC

      public static final int DOCTYPE_UBLIC
      See Also:

    • DOCTYPE_YSTEM

      public static final int DOCTYPE_YSTEM
      See Also:

    • AFTER_DOCTYPE_PUBLIC_KEYWORD

      public static final int AFTER_DOCTYPE_PUBLIC_KEYWORD
      See Also:

    • BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS

      public static final int BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS
      See Also:

    • AFTER_DOCTYPE_SYSTEM_KEYWORD

      public static final int AFTER_DOCTYPE_SYSTEM_KEYWORD
      See Also:

    • CONSUME_CHARACTER_REFERENCE

      public static final int CONSUME_CHARACTER_REFERENCE
      See Also:

    • CONSUME_NCR

      public static final int CONSUME_NCR
      See Also:

    • CHARACTER_REFERENCE_TAIL

      public static final int CHARACTER_REFERENCE_TAIL
      See Also:

    • HEX_NCR_LOOP

      public static final int HEX_NCR_LOOP
      See Also:

    • DECIMAL_NRC_LOOP

      public static final int DECIMAL_NRC_LOOP
      See Also:

    • HANDLE_NCR_VALUE

      public static final int HANDLE_NCR_VALUE
      See Also:

    • HANDLE_NCR_VALUE_RECONSUME

      public static final int HANDLE_NCR_VALUE_RECONSUME
      See Also:

    • CHARACTER_REFERENCE_HILO_LOOKUP

      public static final int CHARACTER_REFERENCE_HILO_LOOKUP
      See Also:

    • SELF_CLOSING_START_TAG

      public static final int SELF_CLOSING_START_TAG
      See Also:

    • CDATA_START

      public static final int CDATA_START
      See Also:

    • CDATA_SECTION

      public static final int CDATA_SECTION
      See Also:

    • CDATA_RSQB

      public static final int CDATA_RSQB
      See Also:

    • CDATA_RSQB_RSQB

      public static final int CDATA_RSQB_RSQB
      See Also:

    • SCRIPT_DATA_LESS_THAN_SIGN

      public static final int SCRIPT_DATA_LESS_THAN_SIGN
      See Also:

    • SCRIPT_DATA_ESCAPE_START

      public static final int SCRIPT_DATA_ESCAPE_START
      See Also:

    • SCRIPT_DATA_ESCAPE_START_DASH

      public static final int SCRIPT_DATA_ESCAPE_START_DASH
      See Also:

    • SCRIPT_DATA_ESCAPED_DASH

      public static final int SCRIPT_DATA_ESCAPED_DASH
      See Also:

    • SCRIPT_DATA_ESCAPED_DASH_DASH

      public static final int SCRIPT_DATA_ESCAPED_DASH_DASH
      See Also:

    • BOGUS_COMMENT_HYPHEN

      public static final int BOGUS_COMMENT_HYPHEN
      See Also:

    • RAWTEXT_RCDATA_LESS_THAN_SIGN

      public static final int RAWTEXT_RCDATA_LESS_THAN_SIGN
      See Also:

    • SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN

      public static final int SCRIPT_DATA_ESCAPED_LESS_THAN_SIGN
      See Also:

    • SCRIPT_DATA_DOUBLE_ESCAPE_START

      public static final int SCRIPT_DATA_DOUBLE_ESCAPE_START
      See Also:

    • SCRIPT_DATA_DOUBLE_ESCAPED

      public static final int SCRIPT_DATA_DOUBLE_ESCAPED
      See Also:

    • SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN

      public static final int SCRIPT_DATA_DOUBLE_ESCAPED_LESS_THAN_SIGN
      See Also:

    • SCRIPT_DATA_DOUBLE_ESCAPED_DASH

      public static final int SCRIPT_DATA_DOUBLE_ESCAPED_DASH
      See Also:

    • SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH

      public static final int SCRIPT_DATA_DOUBLE_ESCAPED_DASH_DASH
      See Also:

    • SCRIPT_DATA_DOUBLE_ESCAPE_END

      public static final int SCRIPT_DATA_DOUBLE_ESCAPE_END
      See Also:

    • PROCESSING_INSTRUCTION

      public static final int PROCESSING_INSTRUCTION
      See Also:

    • PROCESSING_INSTRUCTION_QUESTION_MARK

      public static final int PROCESSING_INSTRUCTION_QUESTION_MARK
      See Also:

    • LEAD_OFFSET

      private static final int LEAD_OFFSET
      Magic value for UTF-16 operations.
      See Also:

    • LT_GT

      private static final char[] LT_GT
      UTF-16 code unit array containing less than and greater than for emitting those characters on certain parse errors.

    • LT_SOLIDUS

      private static final char[] LT_SOLIDUS
      UTF-16 code unit array containing less than and solidus for emitting those characters on certain parse errors.

    • RSQB_RSQB

      private static final char[] RSQB_RSQB
      UTF-16 code unit array containing ]] for emitting those characters on state transitions.

    • REPLACEMENT_CHARACTER

      private static final char[] REPLACEMENT_CHARACTER
      Array version of U+FFFD.

    • SPACE

      private static final char[] SPACE
      Array version of space.

    • LF

      private static final char[] LF
      Array version of line feed.

    • BUFFER_GROW_BY

      private static final int BUFFER_GROW_BY
      Buffer growth parameter.
      See Also:

    • CDATA_LSQB

      private static final char[] CDATA_LSQB
      "CDATA[" as char[]

    • OCTYPE

      private static final char[] OCTYPE
      "octype" as char[]

    • UBLIC

      private static final char[] UBLIC
      "ublic" as char[]

    • YSTEM

      private static final char[] YSTEM
      "ystem" as char[]

    • TITLE_ARR

      private static final char[] TITLE_ARR

    • SCRIPT_ARR

      private static final char[] SCRIPT_ARR

    • STYLE_ARR

      private static final char[] STYLE_ARR

    • PLAINTEXT_ARR

      private static final char[] PLAINTEXT_ARR

    • XMP_ARR

      private static final char[] XMP_ARR

    • TEXTAREA_ARR

      private static final char[] TEXTAREA_ARR

    • IFRAME_ARR

      private static final char[] IFRAME_ARR

    • NOEMBED_ARR

      private static final char[] NOEMBED_ARR

    • NOSCRIPT_ARR

      private static final char[] NOSCRIPT_ARR

    • NOFRAMES_ARR

      private static final char[] NOFRAMES_ARR

    • tokenHandler

      protected final TokenHandler tokenHandler
      The token handler.

    • encodingDeclarationHandler

      protected EncodingDeclarationHandler encodingDeclarationHandler

    • errorHandler

      protected ErrorHandler errorHandler
      The error handler.

    • lastCR

      protected boolean lastCR
      Whether the previous char read was CR.

    • stateSave

      protected int stateSave

    • returnStateSave

      private int returnStateSave

    • index

      protected int index

    • forceQuirks

      private boolean forceQuirks

    • additional

      private char additional

    • entCol

      private int entCol

    • firstCharKey

      private int firstCharKey

    • lo

      private int lo

    • hi

      private int hi

    • candidate

      private int candidate

    • strBufMark

      private int strBufMark

    • prevValue

      private int prevValue

    • value

      protected int value

    • seenDigits

      private boolean seenDigits

    • cstart

      protected int cstart

    • publicId

      private String publicId
      The SAX public id for the resource being tokenized. (Only passed to back as part of locator data.)

    • systemId

      private String systemId
      The SAX system id for the resource being tokenized. (Only passed to back as part of locator data.)

    • strBuf

      private char[] strBuf
      Buffer for short identifiers.

    • strBufLen

      private int strBufLen
      Number of significant chars in strBuf.

    • longStrBuf

      private char[] longStrBuf
      Buffer for long strings.

    • longStrBufLen

      private int longStrBufLen
      Number of significant chars in longStrBuf.

    • bmpChar

      private final char[] bmpChar
      Buffer for expanding NCRs falling into the Basic Multilingual Plane.

    • astralChar

      private final char[] astralChar
      Buffer for expanding astral NCRs.

    • endTagExpectation

      protected ElementName endTagExpectation
      The element whose end tag closes the current CDATA or RCDATA element.

    • endTagExpectationAsArray

      private char[] endTagExpectationAsArray

    • endTag

      protected boolean endTag
      true if tokenizing an end tag

    • tagName

      private ElementName tagName
      The current tag token name.

    • attributeName

      protected AttributeName attributeName
      The current attribute name.

    • wantsComments

      private boolean wantsComments
      Whether comment tokens are emitted.

    • html4

      protected boolean html4
      true when HTML4-specific additional errors are requested.

    • metaBoundaryPassed

      private boolean metaBoundaryPassed
      Whether the stream is past the first 512 bytes.

    • doctypeName

      private String doctypeName
      The name of the current doctype token.

    • publicIdentifier

      private String publicIdentifier
      The public id of the current doctype token.

    • systemIdentifier

      private String systemIdentifier
      The system id of the current doctype token.

    • attributes

      private HtmlAttributes attributes
      The attribute holder.

    • contentSpacePolicy

      private XmlViolationPolicy contentSpacePolicy
      The policy for vertical tab and form feed.

    • commentPolicy

      private XmlViolationPolicy commentPolicy
      The policy for comments.

    • xmlnsPolicy

      private XmlViolationPolicy xmlnsPolicy

    • namePolicy

      private XmlViolationPolicy namePolicy

    • html4ModeCompatibleWithXhtml1Schemata

      private boolean html4ModeCompatibleWithXhtml1Schemata

    • newAttributesEachTime

      private final boolean newAttributesEachTime

    • mappingLangToXmlLang

      private int mappingLangToXmlLang

    • shouldSuspend

      private boolean shouldSuspend

    • confident

      protected boolean confident

    • line

      private int line

    • interner

      private Interner interner

    • ampersandLocation

      protected LocatorImpl ampersandLocation

  • Constructor Details

    • Tokenizer

      public Tokenizer(TokenHandler tokenHandler, boolean newAttributesEachTime)

    • Tokenizer

      public Tokenizer(TokenHandler tokenHandler)
      The constructor.
      Parameters:
      tokenHandler - the handler for receiving tokens

  • Method Details

    • setInterner

      public void setInterner(Interner interner)

    • initLocation

      public void initLocation(String newPublicId, String newSystemId)

    • isMappingLangToXmlLang

      public boolean isMappingLangToXmlLang()
      Returns the mappingLangToXmlLang.
      Returns:
      the mappingLangToXmlLang

    • setMappingLangToXmlLang

      public void setMappingLangToXmlLang(boolean mappingLangToXmlLang)
      Sets the mappingLangToXmlLang.
      Parameters:
      mappingLangToXmlLang - the mappingLangToXmlLang to set

    • setErrorHandler

      public void setErrorHandler(ErrorHandler eh)
      Sets the error handler.
      See Also:

    • getErrorHandler

      public ErrorHandler getErrorHandler()

    • setCommentPolicy

      public void setCommentPolicy(XmlViolationPolicy commentPolicy)
      Sets the commentPolicy.
      Parameters:
      commentPolicy - the commentPolicy to set

    • setContentNonXmlCharPolicy

      public void setContentNonXmlCharPolicy(XmlViolationPolicy contentNonXmlCharPolicy)
      Sets the contentNonXmlCharPolicy.
      Parameters:
      contentNonXmlCharPolicy - the contentNonXmlCharPolicy to set

    • setContentSpacePolicy

      public void setContentSpacePolicy(XmlViolationPolicy contentSpacePolicy)
      Sets the contentSpacePolicy.
      Parameters:
      contentSpacePolicy - the contentSpacePolicy to set

    • setXmlnsPolicy

      public void setXmlnsPolicy(XmlViolationPolicy xmlnsPolicy)
      Sets the xmlnsPolicy.
      Parameters:
      xmlnsPolicy - the xmlnsPolicy to set

    • setNamePolicy

      public void setNamePolicy(XmlViolationPolicy namePolicy)

    • setHtml4ModeCompatibleWithXhtml1Schemata

      public void setHtml4ModeCompatibleWithXhtml1Schemata(boolean html4ModeCompatibleWithXhtml1Schemata)
      Sets the html4ModeCompatibleWithXhtml1Schemata.
      Parameters:
      html4ModeCompatibleWithXhtml1Schemata - the html4ModeCompatibleWithXhtml1Schemata to set

    • setStateAndEndTagExpectation

      public void setStateAndEndTagExpectation(int specialTokenizerState, String endTagExpectation)
      Sets the tokenizer state and the associated element name. This should only ever used to put the tokenizer into one of the states that have a special end tag expectation.
      Parameters:
      specialTokenizerState - the tokenizer state to set
      endTagExpectation - the expected end tag for transitioning back to normal

    • setStateAndEndTagExpectation

      public void setStateAndEndTagExpectation(int specialTokenizerState, ElementName endTagExpectation)
      Sets the tokenizer state and the associated element name. This should only ever used to put the tokenizer into one of the states that have a special end tag expectation.
      Parameters:
      specialTokenizerState - the tokenizer state to set
      endTagExpectation - the expected end tag for transitioning back to normal

    • endTagExpectationToArray

      private void endTagExpectationToArray()

    • setLineNumber

      public void setLineNumber(int line)
      For C++ use only.

    • getLineNumber

      public int getLineNumber()
      Specified by:
      getLineNumber in interface Locator
      See Also:

    • getColumnNumber

      public int getColumnNumber()
      Specified by:
      getColumnNumber in interface Locator
      See Also:

    • getPublicId

      public String getPublicId()
      Specified by:
      getPublicId in interface Locator
      See Also:

    • getSystemId

      public String getSystemId()
      Specified by:
      getSystemId in interface Locator
      See Also:

    • notifyAboutMetaBoundary

      public void notifyAboutMetaBoundary()

    • turnOnAdditionalHtml4Errors

      void turnOnAdditionalHtml4Errors()

    • emptyAttributes

      HtmlAttributes emptyAttributes()

    • clearStrBufAndAppend

      private void clearStrBufAndAppend(char c)

    • clearStrBuf

      private void clearStrBuf()

    • appendStrBuf

      private void appendStrBuf(char c)
      Appends to the smaller buffer.
      Parameters:
      c - the UTF-16 code unit to append

    • strBufToString

      protected String strBufToString()
      The smaller buffer as a String. Currently only used for error reporting.

      C++ memory note: The return value must be released.

      Returns:
      the smaller buffer as a string

    • strBufToDoctypeName

      private void strBufToDoctypeName()
      Returns the short buffer as a local name. The return value is released in emitDoctypeToken().

    • emitStrBuf

      private void emitStrBuf() throws SAXException
      Emits the smaller buffer as character tokens.
      Throws:
      SAXException - if the token handler threw

    • clearLongStrBuf

      private void clearLongStrBuf()

    • clearLongStrBufAndAppend

      private void clearLongStrBufAndAppend(char c)

    • appendLongStrBuf

      private void appendLongStrBuf(char c)
      Appends to the larger buffer.
      Parameters:
      c - the UTF-16 code unit to append

    • appendSecondHyphenToBogusComment

      private void appendSecondHyphenToBogusComment() throws SAXException
      Throws:
      SAXException

    • maybeAppendSpaceToBogusComment

      private void maybeAppendSpaceToBogusComment() throws SAXException
      Throws:
      SAXException

    • adjustDoubleHyphenAndAppendToLongStrBufAndErr

      private void adjustDoubleHyphenAndAppendToLongStrBufAndErr(char c) throws SAXException
      Throws:
      SAXException

    • appendLongStrBuf

      private void appendLongStrBuf(char[] buffer, int offset, int length)

    • appendStrBufToLongStrBuf

      private void appendStrBufToLongStrBuf()
      Append the contents of the smaller buffer to the larger one.

    • longStrBufToString

      private String longStrBufToString()
      The larger buffer as a string.

      C++ memory note: The return value must be released.

      Returns:
      the larger buffer as a string

    • emitComment

      private void emitComment(int provisionalHyphens, int pos) throws SAXException
      Emits the current comment token.
      Parameters:
      pos - TODO
      Throws:
      SAXException

    • flushChars

      protected void flushChars(char[] buf, int pos) throws SAXException
      Flushes coalesced character tokens.
      Parameters:
      buf - TODO
      pos - TODO
      Throws:
      SAXException

    • fatal

      public void fatal(String message) throws SAXException
      Reports an condition that would make the infoset incompatible with XML 1.0 as fatal.
      Parameters:
      message - the message
      Throws:
      SAXException
      SAXParseException

    • err

      public void err(String message) throws SAXException
      Reports a Parse Error.
      Parameters:
      message - the message
      Throws:
      SAXException

    • errTreeBuilder

      public void errTreeBuilder(String message) throws SAXException
      Throws:
      SAXException

    • warn

      public void warn(String message) throws SAXException
      Reports a warning
      Parameters:
      message - the message
      Throws:
      SAXException

    • resetAttributes

      private void resetAttributes()

    • strBufToElementNameString

      private void strBufToElementNameString()

    • emitCurrentTagToken

      private int emitCurrentTagToken(boolean selfClosing, int pos) throws SAXException
      Throws:
      SAXException

    • attributeNameComplete

      private void attributeNameComplete() throws SAXException
      Throws:
      SAXException

    • addAttributeWithoutValue

      private void addAttributeWithoutValue() throws SAXException
      Throws:
      SAXException

    • addAttributeWithValue

      private void addAttributeWithValue() throws SAXException
      Throws:
      SAXException

    • newAsciiLowerCaseStringFromString

      private static String newAsciiLowerCaseStringFromString(String str)

    • startErrorReporting

      protected void startErrorReporting() throws SAXException
      Throws:
      SAXException

    • start

      public void start() throws SAXException
      Throws:
      SAXException

    • tokenizeBuffer

      public boolean tokenizeBuffer(UTF16Buffer buffer) throws SAXException
      Throws:
      SAXException

    • stateLoop

      private int stateLoop(int state, char c, int pos, char[] buf, boolean reconsume, int returnState, int endPos) throws SAXException
      Throws:
      SAXException

    • workAroundHotSpotHugeMethodLimit

      private long workAroundHotSpotHugeMethodLimit(int state, char c, int pos, char[] buf, boolean reconsume, int returnState, int endPos) throws SAXException
      compressed returnValue: int returnState = returnValue >> 33 boolean breakOuterState = ((returnValue >> 32) invalid input: '&' 0x1) != 0) int pos = returnValue invalid input: '&' 0xFFFFFFFF // same as (int)returnValue
      Throws:
      SAXException

    • transition

      protected int transition(int from, int to, boolean reconsume, int pos) throws SAXException
      Throws:
      SAXException

    • initDoctypeFields

      private void initDoctypeFields()

    • adjustDoubleHyphenAndAppendToLongStrBufCarriageReturn

      private void adjustDoubleHyphenAndAppendToLongStrBufCarriageReturn() throws SAXException
      Throws:
      SAXException

    • adjustDoubleHyphenAndAppendToLongStrBufLineFeed

      private void adjustDoubleHyphenAndAppendToLongStrBufLineFeed() throws SAXException
      Throws:
      SAXException

    • appendLongStrBufLineFeed

      private void appendLongStrBufLineFeed()

    • appendLongStrBufCarriageReturn

      private void appendLongStrBufCarriageReturn()

    • silentCarriageReturn

      protected void silentCarriageReturn()

    • silentLineFeed

      protected void silentLineFeed()

    • emitCarriageReturn

      private void emitCarriageReturn(char[] buf, int pos) throws SAXException
      Throws:
      SAXException

    • emitReplacementCharacter

      private void emitReplacementCharacter(char[] buf, int pos) throws SAXException
      Throws:
      SAXException

    • emitPlaintextReplacementCharacter

      private void emitPlaintextReplacementCharacter(char[] buf, int pos) throws SAXException
      Throws:
      SAXException

    • setAdditionalAndRememberAmpersandLocation

      private void setAdditionalAndRememberAmpersandLocation(char add)

    • bogusDoctype

      private void bogusDoctype() throws SAXException
      Throws:
      SAXException

    • bogusDoctypeWithoutQuirks

      private void bogusDoctypeWithoutQuirks() throws SAXException
      Throws:
      SAXException

    • emitOrAppendStrBuf

      private void emitOrAppendStrBuf(int returnState) throws SAXException
      Throws:
      SAXException

    • handleNcrValue

      private void handleNcrValue(int returnState) throws SAXException
      Throws:
      SAXException

    • eof

      public void eof() throws SAXException
      Throws:
      SAXException

    • emitDoctypeToken

      private void emitDoctypeToken(int pos) throws SAXException
      Throws:
      SAXException

    • checkChar

      protected char checkChar(char[] buf, int pos) throws SAXException
      Throws:
      SAXException

    • isAlreadyComplainedAboutNonAscii

      public boolean isAlreadyComplainedAboutNonAscii()
      Returns the alreadyComplainedAboutNonAscii.
      Returns:
      the alreadyComplainedAboutNonAscii

    • internalEncodingDeclaration

      public boolean internalEncodingDeclaration(String internalCharset) throws SAXException
      Throws:
      SAXException

    • emitOrAppendTwo

      private void emitOrAppendTwo(char[] val, int returnState) throws SAXException
      Parameters:
      val -
      Throws:
      SAXException

    • emitOrAppendOne

      private void emitOrAppendOne(char[] val, int returnState) throws SAXException
      Throws:
      SAXException

    • end

      public void end() throws SAXException
      Throws:
      SAXException

    • requestSuspension

      public void requestSuspension()

    • becomeConfident

      public void becomeConfident()

    • isNextCharOnNewLine

      public boolean isNextCharOnNewLine()
      Returns the nextCharOnNewLine.
      Returns:
      the nextCharOnNewLine

    • isPrevCR

      public boolean isPrevCR()

    • getLine

      public int getLine()
      Returns the line.
      Returns:
      the line

    • getCol

      public int getCol()
      Returns the col.
      Returns:
      the col

    • isInDataState

      public boolean isInDataState()

    • resetToDataState

      public void resetToDataState()

    • loadState

      public void loadState(Tokenizer other) throws SAXException
      Throws:
      SAXException

    • initializeWithoutStarting

      public void initializeWithoutStarting() throws SAXException
      Throws:
      SAXException

    • errGarbageAfterLtSlash

      protected void errGarbageAfterLtSlash() throws SAXException
      Throws:
      SAXException

    • errLtSlashGt

      protected void errLtSlashGt() throws SAXException
      Throws:
      SAXException

    • errWarnLtSlashInRcdata

      protected void errWarnLtSlashInRcdata() throws SAXException
      Throws:
      SAXException

    • errHtml4LtSlashInRcdata

      protected void errHtml4LtSlashInRcdata(char folded) throws SAXException
      Throws:
      SAXException

    • errCharRefLacksSemicolon

      protected void errCharRefLacksSemicolon() throws SAXException
      Throws:
      SAXException

    • errNoDigitsInNCR

      protected void errNoDigitsInNCR() throws SAXException
      Throws:
      SAXException

    • errGtInSystemId

      protected void errGtInSystemId() throws SAXException
      Throws:
      SAXException

    • errGtInPublicId

      protected void errGtInPublicId() throws SAXException
      Throws:
      SAXException

    • errNamelessDoctype

      protected void errNamelessDoctype() throws SAXException
      Throws:
      SAXException

    • errConsecutiveHyphens

      protected void errConsecutiveHyphens() throws SAXException
      Throws:
      SAXException

    • errPrematureEndOfComment

      protected void errPrematureEndOfComment() throws SAXException
      Throws:
      SAXException

    • errBogusComment

      protected void errBogusComment() throws SAXException
      Throws:
      SAXException

    • errUnquotedAttributeValOrNull

      protected void errUnquotedAttributeValOrNull(char c) throws SAXException
      Throws:
      SAXException

    • errSlashNotFollowedByGt

      protected void errSlashNotFollowedByGt() throws SAXException
      Throws:
      SAXException

    • errHtml4XmlVoidSyntax

      protected void errHtml4XmlVoidSyntax() throws SAXException
      Throws:
      SAXException

    • errNoSpaceBetweenAttributes

      protected void errNoSpaceBetweenAttributes() throws SAXException
      Throws:
      SAXException

    • errHtml4NonNameInUnquotedAttribute

      protected void errHtml4NonNameInUnquotedAttribute(char c) throws SAXException
      Throws:
      SAXException

    • errLtOrEqualsOrGraveInUnquotedAttributeOrNull

      protected void errLtOrEqualsOrGraveInUnquotedAttributeOrNull(char c) throws SAXException
      Throws:
      SAXException

    • errAttributeValueMissing

      protected void errAttributeValueMissing() throws SAXException
      Throws:
      SAXException

    • errBadCharBeforeAttributeNameOrNull

      protected void errBadCharBeforeAttributeNameOrNull(char c) throws SAXException
      Throws:
      SAXException

    • errEqualsSignBeforeAttributeName

      protected void errEqualsSignBeforeAttributeName() throws SAXException
      Throws:
      SAXException

    • errBadCharAfterLt

      protected void errBadCharAfterLt(char c) throws SAXException
      Throws:
      SAXException

    • errLtGt

      protected void errLtGt() throws SAXException
      Throws:
      SAXException

    • errProcessingInstruction

      protected void errProcessingInstruction() throws SAXException
      Throws:
      SAXException

    • errUnescapedAmpersandInterpretedAsCharacterReference

      protected void errUnescapedAmpersandInterpretedAsCharacterReference() throws SAXException
      Throws:
      SAXException

    • errNotSemicolonTerminated

      protected void errNotSemicolonTerminated() throws SAXException
      Throws:
      SAXException

    • errNoNamedCharacterMatch

      protected void errNoNamedCharacterMatch() throws SAXException
      Throws:
      SAXException

    • errQuoteBeforeAttributeName

      protected void errQuoteBeforeAttributeName(char c) throws SAXException
      Throws:
      SAXException

    • errQuoteOrLtInAttributeNameOrNull

      protected void errQuoteOrLtInAttributeNameOrNull(char c) throws SAXException
      Throws:
      SAXException

    • errExpectedPublicId

      protected void errExpectedPublicId() throws SAXException
      Throws:
      SAXException

    • errBogusDoctype

      protected void errBogusDoctype() throws SAXException
      Throws:
      SAXException

    • maybeWarnPrivateUseAstral

      protected void maybeWarnPrivateUseAstral() throws SAXException
      Throws:
      SAXException

    • maybeWarnPrivateUse

      protected void maybeWarnPrivateUse(char ch) throws SAXException
      Throws:
      SAXException

    • maybeErrAttributesOnEndTag

      protected void maybeErrAttributesOnEndTag(HtmlAttributes attrs) throws SAXException
      Throws:
      SAXException

    • maybeErrSlashInEndTag

      protected void maybeErrSlashInEndTag(boolean selfClosing) throws SAXException
      Throws:
      SAXException

    • errNcrNonCharacter

      protected char errNcrNonCharacter(char ch) throws SAXException
      Throws:
      SAXException

    • errAstralNonCharacter

      protected void errAstralNonCharacter(int ch) throws SAXException
      Throws:
      SAXException

    • errNcrSurrogate

      protected void errNcrSurrogate() throws SAXException
      Throws:
      SAXException

    • errNcrControlChar

      protected char errNcrControlChar(char ch) throws SAXException
      Throws:
      SAXException

    • errNcrCr

      protected void errNcrCr() throws SAXException
      Throws:
      SAXException

    • errNcrInC1Range

      protected void errNcrInC1Range() throws SAXException
      Throws:
      SAXException

    • errEofInPublicId

      protected void errEofInPublicId() throws SAXException
      Throws:
      SAXException

    • errEofInComment

      protected void errEofInComment() throws SAXException
      Throws:
      SAXException

    • errEofInDoctype

      protected void errEofInDoctype() throws SAXException
      Throws:
      SAXException

    • errEofInAttributeValue

      protected void errEofInAttributeValue() throws SAXException
      Throws:
      SAXException

    • errEofInAttributeName

      protected void errEofInAttributeName() throws SAXException
      Throws:
      SAXException

    • errEofWithoutGt

      protected void errEofWithoutGt() throws SAXException
      Throws:
      SAXException

    • errEofInTagName

      protected void errEofInTagName() throws SAXException
      Throws:
      SAXException

    • errEofInEndTag

      protected void errEofInEndTag() throws SAXException
      Throws:
      SAXException

    • errEofAfterLt

      protected void errEofAfterLt() throws SAXException
      Throws:
      SAXException

    • errNcrOutOfRange

      protected void errNcrOutOfRange() throws SAXException
      Throws:
      SAXException

    • errNcrUnassigned

      protected void errNcrUnassigned() throws SAXException
      Throws:
      SAXException

    • errDuplicateAttribute

      protected void errDuplicateAttribute() throws SAXException
      Throws:
      SAXException

    • errEofInSystemId

      protected void errEofInSystemId() throws SAXException
      Throws:
      SAXException

    • errExpectedSystemId

      protected void errExpectedSystemId() throws SAXException
      Throws:
      SAXException

    • errMissingSpaceBeforeDoctypeName

      protected void errMissingSpaceBeforeDoctypeName() throws SAXException
      Throws:
      SAXException

    • errHyphenHyphenBang

      protected void errHyphenHyphenBang() throws SAXException
      Throws:
      SAXException

    • errNcrControlChar

      protected void errNcrControlChar() throws SAXException
      Throws:
      SAXException

    • errNcrZero

      protected void errNcrZero() throws SAXException
      Throws:
      SAXException

    • errNoSpaceBetweenDoctypeSystemKeywordAndQuote

      protected void errNoSpaceBetweenDoctypeSystemKeywordAndQuote() throws SAXException
      Throws:
      SAXException

    • errNoSpaceBetweenPublicAndSystemIds

      protected void errNoSpaceBetweenPublicAndSystemIds() throws SAXException
      Throws:
      SAXException

    • errNoSpaceBetweenDoctypePublicKeywordAndQuote

      protected void errNoSpaceBetweenDoctypePublicKeywordAndQuote() throws SAXException
      Throws:
      SAXException

    • noteAttributeWithoutValue

      protected void noteAttributeWithoutValue() throws SAXException
      Throws:
      SAXException

    • noteUnquotedAttributeValue

      protected void noteUnquotedAttributeValue() throws SAXException
      Throws:
      SAXException

    • setEncodingDeclarationHandler

      public void setEncodingDeclarationHandler(EncodingDeclarationHandler encodingDeclarationHandler)
      Sets the encodingDeclarationHandler.
      Parameters:
      encodingDeclarationHandler - the encodingDeclarationHandler to set

    • destructor

      void destructor()

    • setTransitionBaseOffset

      public void setTransitionBaseOffset(int offset)
      Sets an offset to be added to the position reported to TransitionHandler.
      Parameters:
      offset - the offset