org.w3c.tidy
Class Lexer
java.lang.Object
|
+--org.w3c.tidy.Lexer
-
public class Lexer
-
extends java.lang.Object
Clase de Tidy adaptada a las necesidades del paquete uaw.
Method Summary
|
void
|
addByte (int
c)
|
void
|
addCharToLexer (int
c)
|
boolean
|
addGenerator
(org.w3c.tidy.Node root)
|
void
|
addStringLiteral
(java.lang.String str)
|
void
|
addStringToLexer
(java.lang.String str)
|
short
|
apparentVersion ()
|
boolean
|
canPrune
(org.w3c.tidy.Node element)
|
void
|
changeChar
(byte c)
|
boolean
|
checkDocTypeKeyWords
(org.w3c.tidy.Node doctype)
|
org.w3c.tidy.AttVal
|
cloneAttributes
(org.w3c.tidy.AttVal attrs)
|
org.w3c.tidy.Node
|
cloneNode
(org.w3c.tidy.Node node)
|
void
|
deferDup ()
|
boolean
|
endOfInput
()
|
static boolean
|
expectsContent
(org.w3c.tidy.Node node)
|
private static boolean
|
findBadSubString (java.lang.String s,
java.lang.String p, int len)
|
short
|
findGivenVersion
(org.w3c.tidy.Node doctype)
|
boolean
|
fixDocType
(org.w3c.tidy.Node root)
|
void
|
fixHTMLNameSpace (org.w3c.tidy.Node root,
java.lang.String profile)
|
void
|
fixId
(org.w3c.tidy.Node node)
|
boolean
|
fixXMLPI
(org.w3c.tidy.Node root)
|
static char
|
foldCase
(char c, boolean tocaps, boolean xmlTags)
|
static byte[]
|
getBytes
(java.lang.String str)
|
org.w3c.tidy.Node
|
getCDATA
(org.w3c.tidy.Node container)
|
static java.lang.String
|
getString (byte[]
bytes, int offset, int length)
|
org.w3c.tidy.Node
|
getToken (short
mode)
|
short
|
HTMLVersion
()
|
java.lang.String
|
HTMLVersionName ()
|
org.w3c.tidy.Node
|
inferredTag
(java.lang.String name)
|
int
|
inlineDup
(org.w3c.tidy.Node node)
|
org.w3c.tidy.Node
|
insertedToken
()
|
private static boolean
|
isDigit (char
c)
|
private static boolean
|
isLetter (char
c)
|
boolean
|
isPushed
(org.w3c.tidy.Node node)
|
static boolean
|
isValidAttrName
(java.lang.String attr)
|
private static boolean
|
isWhite (char
c)
|
private static short
|
MAP (char c)
|
private static void
|
mapStr
(java.lang.String str, short code)
|
org.w3c.tidy.Node
|
newLineNode
()
|
org.w3c.tidy.Node
|
newNode ()
|
org.w3c.tidy.Node
|
newNode
(short type, byte[] textarray, int start, int end)
|
org.w3c.tidy.Node
|
newNode
(short type, byte[] textarray, int start, int end,
java.lang.String element)
|
org.w3c.tidy.Node
|
parseAsp ()
|
java.lang.String
|
parseAttribute (org.w3c.tidy.MutableBoolean
isempty, org.w3c.tidy.MutableObject asp,
org.w3c.tidy.MutableObject php)
|
org.w3c.tidy.AttVal
|
parseAttrs
(org.w3c.tidy.MutableBoolean isempty)
|
void
|
parseEntity
(short mode)
|
org.w3c.tidy.Node
|
parsePhp ()
|
int
|
parseServerInstruction
()
|
char
|
parseTagName
()
|
java.lang.String
|
parseValue (java.lang.String name, boolean
foldCase, org.w3c.tidy.MutableBoolean isempty,
org.w3c.tidy.MutableInteger pdelim)
|
void
|
popInline
(org.w3c.tidy.Node node)
|
void
|
pushInline
(org.w3c.tidy.Node node)
|
boolean
|
setXHTMLDocType
(org.w3c.tidy.Node root)
|
private static char
|
toLower (char
c)
|
private static char
|
toUpper (char
c)
|
void
|
ungetToken
()
|
protected void
|
updateNodeTextArrays
(byte[] oldtextarray, byte[] newtextarray)
|
static int
|
wstrcasecmp
(java.lang.String s1, java.lang.String s2)
|
static int
|
wstrcaselexcmp
(java.lang.String s1, java.lang.String s2)
|
static boolean
|
wsubstr
(java.lang.String s1, java.lang.String s2)
|
Methods inherited from class java.lang.Object
|
clone, equals, finalize, getClass, hashCode, notify,
notifyAll, toString, wait, wait, wait
|
in
public org.w3c.tidy.StreamIn in
errout
public java.io.PrintWriter errout
badAccess
public short badAccess
badLayout
public short badLayout
badChars
public short badChars
badForm
public short badForm
warnings
public short warnings
errors
public short errors
lines
public int lines
columns
public int columns
waswhite
public boolean waswhite
pushed
public boolean pushed
insertspace
public boolean insertspace
excludeBlocks
public boolean excludeBlocks
exiled
public boolean exiled
isvoyager
public boolean isvoyager
versions
public short versions
doctype
public int doctype
badDoctype
public boolean badDoctype
txtstart
public int txtstart
txtend
public int txtend
state
public short state
token
public org.w3c.tidy.Node token
lexbuf
public byte[] lexbuf
lexlength
public int lexlength
lexsize
public int lexsize
inode
public org.w3c.tidy.Node inode
insert
public int insert
istack
public java.util.Stack istack
istackbase
public int istackbase
styles
public org.w3c.tidy.Style styles
configuration
public Configuration configuration
seenBodyEndTag
protected int seenBodyEndTag
nodeList
private java.util.Vector nodeList
IgnoreWhitespace
public static final short IgnoreWhitespace
-
-
See Also:
-
Constant Field Values
MixedContent
public static final short MixedContent
-
-
See Also:
-
Constant Field Values
Preformatted
public static final short Preformatted
-
-
See Also:
-
Constant Field Values
IgnoreMarkup
public static final short IgnoreMarkup
-
-
See Also:
-
Constant Field Values
DIGIT
private static final short DIGIT
-
-
See Also:
-
Constant Field Values
LETTER
private static final short LETTER
-
-
See Also:
-
Constant Field Values
NAMECHAR
private static final short NAMECHAR
-
-
See Also:
-
Constant Field Values
WHITE
private static final short WHITE
-
-
See Also:
-
Constant Field Values
NEWLINE
private static final short NEWLINE
-
-
See Also:
-
Constant Field Values
LOWERCASE
private static final short LOWERCASE
-
-
See Also:
-
Constant Field Values
UPPERCASE
private static final short UPPERCASE
-
-
See Also:
-
Constant Field Values
LEX_CONTENT
private static final short LEX_CONTENT
-
-
See Also:
-
Constant Field Values
LEX_GT
private static final short LEX_GT
-
-
See Also:
-
Constant Field Values
LEX_ENDTAG
private static final short LEX_ENDTAG
-
-
See Also:
-
Constant Field Values
LEX_STARTTAG
private static final short LEX_STARTTAG
-
-
See Also:
-
Constant Field Values
LEX_COMMENT
private static final short LEX_COMMENT
-
-
See Also:
-
Constant Field Values
LEX_DOCTYPE
private static final short LEX_DOCTYPE
-
-
See Also:
-
Constant Field Values
LEX_PROCINSTR
private static final short LEX_PROCINSTR
-
-
See Also:
-
Constant Field Values
LEX_ENDCOMMENT
private static final short LEX_ENDCOMMENT
-
-
See Also:
-
Constant Field Values
LEX_CDATA
private static final short LEX_CDATA
-
-
See Also:
-
Constant Field Values
LEX_SECTION
private static final short LEX_SECTION
-
-
See Also:
-
Constant Field Values
LEX_ASP
private static final short LEX_ASP
-
-
See Also:
-
Constant Field Values
LEX_JSTE
private static final short LEX_JSTE
-
-
See Also:
-
Constant Field Values
LEX_PHP
private static final short LEX_PHP
-
-
See Also:
-
Constant Field Values
lexmap
private static short[] lexmap
voyager_loose
private static final java.lang.String voyager_loose
-
-
See Also:
-
Constant Field Values
voyager_strict
private static final java.lang.String voyager_strict
-
-
See Also:
-
Constant Field Values
voyager_frameset
private static final java.lang.String voyager_frameset
-
-
See Also:
-
Constant Field Values
XHTML_NAMESPACE
private static final java.lang.String XHTML_NAMESPACE
-
-
See Also:
-
Constant Field Values
W3CVersion
private static Lexer.W3CVersionInfo[] W3CVersion
Lexer
public Lexer(org.w3c.tidy.StreamIn in,
Configuration configuration)
newNode
public org.w3c.tidy.Node newNode()
newNode
public org.w3c.tidy.Node newNode(short type,
byte[] textarray,
int start,
int end)
newNode
public org.w3c.tidy.Node newNode(short type,
byte[] textarray,
int start,
int end,
java.lang.String element)
cloneNode
public org.w3c.tidy.Node cloneNode(org.w3c.tidy.Node node)
cloneAttributes
public org.w3c.tidy.AttVal cloneAttributes(org.w3c.tidy.AttVal attrs)
updateNodeTextArrays
protected void updateNodeTextArrays(byte[] oldtextarray,
byte[] newtextarray)
newLineNode
public org.w3c.tidy.Node newLineNode()
getBytes
public static byte[] getBytes(java.lang.String str)
getString
public static java.lang.String getString(byte[] bytes,
int offset,
int length)
endOfInput
public boolean endOfInput()
addByte
public void addByte(int c)
changeChar
public void changeChar(byte c)
addCharToLexer
public void addCharToLexer(int c)
addStringToLexer
public void addStringToLexer(java.lang.String str)
parseEntity
public void parseEntity(short mode)
parseTagName
public char parseTagName()
addStringLiteral
public void addStringLiteral(java.lang.String str)
HTMLVersion
public short HTMLVersion()
HTMLVersionName
public java.lang.String HTMLVersionName()
addGenerator
public boolean addGenerator(org.w3c.tidy.Node root)
findBadSubString
private static boolean findBadSubString(java.lang.String s,
java.lang.String p,
int len)
checkDocTypeKeyWords
public boolean checkDocTypeKeyWords(org.w3c.tidy.Node doctype)
findGivenVersion
public short findGivenVersion(org.w3c.tidy.Node doctype)
fixHTMLNameSpace
public void fixHTMLNameSpace(org.w3c.tidy.Node root,
java.lang.String profile)
setXHTMLDocType
public boolean setXHTMLDocType(org.w3c.tidy.Node root)
apparentVersion
public short apparentVersion()
fixDocType
public boolean fixDocType(org.w3c.tidy.Node root)
fixXMLPI
public boolean fixXMLPI(org.w3c.tidy.Node root)
inferredTag
public org.w3c.tidy.Node inferredTag(java.lang.String name)
expectsContent
public static boolean expectsContent(org.w3c.tidy.Node node)
getCDATA
public org.w3c.tidy.Node getCDATA(org.w3c.tidy.Node container)
ungetToken
public void ungetToken()
getToken
public org.w3c.tidy.Node getToken(short mode)
parseAsp
public org.w3c.tidy.Node parseAsp()
parsePhp
public org.w3c.tidy.Node parsePhp()
parseAttribute
public java.lang.String parseAttribute(org.w3c.tidy.MutableBoolean isempty,
org.w3c.tidy.MutableObject asp,
org.w3c.tidy.MutableObject php)
parseServerInstruction
public int parseServerInstruction()
parseValue
public java.lang.String parseValue(java.lang.String name,
boolean foldCase,
org.w3c.tidy.MutableBoolean isempty,
org.w3c.tidy.MutableInteger pdelim)
isValidAttrName
public static boolean isValidAttrName(java.lang.String attr)
parseAttrs
public org.w3c.tidy.AttVal parseAttrs(org.w3c.tidy.MutableBoolean isempty)
pushInline
public void pushInline(org.w3c.tidy.Node node)
popInline
public void popInline(org.w3c.tidy.Node node)
isPushed
public boolean isPushed(org.w3c.tidy.Node node)
inlineDup
public int inlineDup(org.w3c.tidy.Node node)
insertedToken
public org.w3c.tidy.Node insertedToken()
wstrcasecmp
public static int wstrcasecmp(java.lang.String s1,
java.lang.String s2)
wstrcaselexcmp
public static int wstrcaselexcmp(java.lang.String s1,
java.lang.String s2)
wsubstr
public static boolean wsubstr(java.lang.String s1,
java.lang.String s2)
canPrune
public boolean canPrune(org.w3c.tidy.Node element)
fixId
public void fixId(org.w3c.tidy.Node node)
deferDup
public void deferDup()
mapStr
private static void mapStr(java.lang.String str,
short code)
MAP
private static short MAP(char c)
isWhite
private static boolean isWhite(char c)
isDigit
private static boolean isDigit(char c)
isLetter
private static boolean isLetter(char c)
toLower
private static char toLower(char c)
toUpper
private static char toUpper(char c)
foldCase
public static char foldCase(char c,
boolean tocaps,
boolean xmlTags)