lexer.h File Reference

Go to the source code of this file.

Data Structures
struct	AttVal
struct	IStack
struct	Lexer
struct	Node
struct	TagStyle
struct	StyleProp

Macros
#define	CM_BLOCK   (1 << 3)
#define	CM_DEFLIST   (1 << 6)
#define	CM_EMPTY   (1 << 0)
#define	CM_FIELD   (1 << 10)
#define	CM_FRAMES   (1 << 13)
#define	CM_HEAD   (1 << 2)
#define	CM_HEADING   (1 << 14)
#define	CM_HTML   (1 << 1)
#define	CM_IMG   (1 << 16)
#define	CM_INLINE   (1 << 4)
#define	CM_LIST   (1 << 5)
#define	CM_MIXED   (1 << 17)
#define	CM_NEW   (1 << 20)
#define	CM_NO_INDENT   (1 << 18)
#define	CM_OBJECT   (1 << 11)
#define	CM_OBSOLETE   (1 << 19)
#define	CM_OMITST   (1 << 21)
#define	CM_OPT   (1 << 15)
#define	CM_PARAM   (1 << 12)
#define	CM_ROW   (1 << 9)
#define	CM_ROWGRP   (1 << 8)
#define	CM_TABLE   (1 << 7)
#define	CM_UNKNOWN   0
#define	digit   1u
#define	digithex   128u
#define	H40F   16u
#define	H40S   4u
#define	H40T   8u
#define	H41F   128u
#define	H41S   32u
#define	H41T   64u
#define	HT20   1u
#define	HT32   2u
#define	HT50   131072u
#define	letter   2u
#define	lowercase   32u
#define	namechar   4u
#define	newline   16u
#define	uppercase   64u
#define	VERS_ALL   (VERS_HTML20|VERS_HTML32|VERS_FROM40|XH50|HT50)
#define	VERS_BASIC   (XB10)
#define	VERS_EVENTS   (VERS_HTML40|VERS_XHTML11)
#define	VERS_FRAMESET   (H40F|H41F|X10F)
#define	VERS_FROM32   (VERS_HTML32|VERS_HTML40|HT50)
#define	VERS_FROM40   (VERS_HTML40|VERS_XHTML11|VERS_BASIC|VERS_HTML5)
#define	VERS_HTML20   (HT20)
#define	VERS_HTML32   (HT32)
#define	VERS_HTML40   (VERS_HTML40_STRICT|VERS_HTML40_LOOSE|VERS_FRAMESET)
#define	VERS_HTML40_LOOSE   (H40T|H41T|X10T)
#define	VERS_HTML40_STRICT   (H40S|H41S|X10S)
#define	VERS_HTML5   (HT50|XH50)
#define	VERS_IFRAME   (VERS_HTML40_LOOSE|VERS_FRAMESET)
#define	VERS_LOOSE   (VERS_HTML20|VERS_HTML32|VERS_IFRAME)
#define	VERS_MICROSOFT   32768u
#define	VERS_NETSCAPE   16384u
#define	VERS_PROPRIETARY   (VERS_NETSCAPE|VERS_MICROSOFT|VERS_SUN)
#define	VERS_STRICT   (VERS_HTML5|VERS_HTML40_STRICT)
#define	VERS_SUN   8192u
#define	VERS_UNKNOWN   (xxxx)
#define	VERS_XHTML   (X10S|X10T|X10F|XH11|XB10|XH50)
#define	VERS_XHTML11   (XH11)
#define	VERS_XML   65536u
#define	white   8u
#define	X10F   1024u
#define	X10S   256u
#define	X10T   512u
#define	XB10   4096u
#define	XH11   2048u
#define	XH50   262144u
#define	xxxx   0u

Enumerations
enum	GetTokenMode {   IgnoreWhitespace,   MixedContent,   Preformatted,   IgnoreMarkup,   OtherNamespace,   CdataContent }
enum	LexerState {   LEX_CONTENT,   LEX_GT,   LEX_ENDTAG,   LEX_STARTTAG,   LEX_COMMENT,   LEX_DOCTYPE,   LEX_PROCINSTR,   LEX_CDATA,   LEX_SECTION,   LEX_ASP,   LEX_JSTE,   LEX_PHP,   LEX_XMLDECL }
enum	NodeType {   RootNode,   DocTypeTag,   CommentTag,   ProcInsTag,   TextNode,   StartTag,   EndTag,   StartEndTag,   CDATATag,   SectionTag,   AspTag,   JsteTag,   PhpTag,   XmlDecl }
enum	ParseDocTypeDeclState {   DT_INTERMEDIATE,   DT_DOCTYPENAME,   DT_PUBLICSYSTEM,   DT_QUOTEDSTRING,   DT_INTSUBSET }

Functions
void	TY_❪AddCharToLexer❫ (Lexer *lexer, uint c)
Bool	TY_❪AddGenerator❫ (TidyDocImpl *doc)
void	TY_❪AddStringLiteral❫ (Lexer *lexer, ctmbstr str)
uint	TY_❪ApparentVersion❫ (TidyDocImpl *doc)
Node *	TY_❪CloneNode❫ (TidyDocImpl *doc, Node *element)
void	TY_❪ConstrainVersion❫ (TidyDocImpl *doc, uint vers)
void	TY_❪DeferDup❫ (TidyDocImpl *doc)
void	TY_❪DetachAttribute❫ (Node *node, AttVal *attr)
AttVal *	TY_❪DupAttrs❫ (TidyDocImpl *doc, AttVal *attrs)
Node *	TY_❪FindBody❫ (TidyDocImpl *doc)
Node *	TY_❪FindContainer❫ (Node *node)
Node *	TY_❪FindDocType❫ (TidyDocImpl *doc)
Node *	TY_❪FindHEAD❫ (TidyDocImpl *doc)
Node *	TY_❪FindHTML❫ (TidyDocImpl *doc)
Node *	TY_❪FindTITLE❫ (TidyDocImpl *doc)
Node *	TY_❪FindXmlDecl❫ (TidyDocImpl *doc)
Bool	TY_❪FixDocType❫ (TidyDocImpl *doc)
Bool	TY_❪FixXmlDecl❫ (TidyDocImpl *doc)
void	TY_❪FreeAttribute❫ (TidyDocImpl *doc, AttVal *av)
void	TY_❪FreeAttrs❫ (TidyDocImpl *doc, Node *node)
void	TY_❪FreeLexer❫ (TidyDocImpl *doc)
void	TY_❪FreeNode❫ (TidyDocImpl *doc, Node *node)
Node *	TY_❪GetToken❫ (TidyDocImpl *doc, GetTokenMode mode)
ctmbstr	TY_❪HTMLVersionNameFromCode❫ (uint vers, Bool isXhtml)
uint	TY_❪HTMLVersionNumberFromCode❫ (uint vers)
int	TY_❪HTMLVersion❫ (TidyDocImpl *doc)
Node *	TY_❪InferredTag❫ (TidyDocImpl *doc, TidyTagId id)
void	TY_❪InitMap❫ (void)
Bool	TY_❪InlineDup1❫ (TidyDocImpl *doc, Node *node, Node *element)
int	TY_❪InlineDup❫ (TidyDocImpl *doc, Node *node)
void	TY_❪InsertAttributeAtEnd❫ (Node *node, AttVal *av)
void	TY_❪InsertAttributeAtStart❫ (Node *node, AttVal *av)
Node *	TY_❪InsertedToken❫ (TidyDocImpl *doc)
Bool	TY_❪IsDigit❫ (uint c)
Bool	TY_❪IsHTMLSpace❫ (uint c)
Bool	TY_❪IsLetter❫ (uint c)
Bool	TY_❪IsNamechar❫ (uint c)
Bool	TY_❪IsNewline❫ (uint c)
Bool	TY_❪IsPushedLast❫ (TidyDocImpl *doc, Node *element, Node *node)
Bool	TY_❪IsPushed❫ (TidyDocImpl *doc, Node *node)
Bool	TY_❪IsUpper❫ (uint c)
Bool	TY_❪IsWhite❫ (uint c)
Bool	TY_❪IsXMLLetter❫ (uint c)
Bool	TY_❪IsXMLNamechar❫ (uint c)
AttVal *	TY_❪NewAttributeEx❫ (TidyDocImpl *doc, ctmbstr name, ctmbstr value, int delim)
AttVal *	TY_❪NewAttribute❫ (TidyDocImpl *doc)
Lexer *	TY_❪NewLexer❫ (TidyDocImpl *doc)
Node *	TY_❪NewLineNode❫ (Lexer *lexer)
Node *	TY_❪NewLiteralTextNode❫ (Lexer *lexer, ctmbstr txt)
Node *	TY_❪NewNode❫ (TidyAllocator *allocator, Lexer *lexer)
void	TY_❪PopInline❫ (TidyDocImpl *doc, Node *node)
void	TY_❪PushInline❫ (TidyDocImpl *doc, Node *node)
void	TY_❪RemoveAttribute❫ (TidyDocImpl *doc, Node *node, AttVal *attr)
Bool	TY_❪SetXHTMLDocType❫ (TidyDocImpl *doc)
Bool	TY_❪SwitchInline❫ (TidyDocImpl *doc, Node *element, Node *node)
Node *	TY_❪TextToken❫ (Lexer *lexer)
uint	TY_❪ToLower❫ (uint c)
uint	TY_❪ToUpper❫ (uint c)
void	TY_❪UngetToken❫ (TidyDocImpl *doc)
Bool	TY_❪WarnMissingSIInEmittedDocType❫ (TidyDocImpl *doc)

---

Data Structure Documentation

struct _AttVal

Data Fields
Node *	asp
tmbstr	attribute
int	delim
const Attribute *	dict
AttVal *	next
Node *	php
tmbstr	value

struct _IStack

Data Fields
AttVal *	attributes
tmbstr	element
IStack *	next
const Dict *	tag

struct _Lexer

Data Fields
TidyAllocator *	allocator
Bool	bad_doctype
uint	columns
uint	doctype
Bool	excludeBlocks
Bool	exiled
Node *	inode
IStack *	insert
Bool	insertspace
IStack *	istack
uint	istackbase
uint	istacklength
uint	istacksize
Bool	isvoyager
Node *	itoken
tmbstr	lexbuf
uint	lexlength
uint	lexsize
uint	lines
Node *	parent
Bool	pushed
Node *	root
Bool	seenEndBody
Bool	seenEndHtml
LexerState	state
TagStyle *	styles
Node *	token
uint	txtend
uint	txtstart
uint	versionEmitted
uint	versions
Bool	waswhite

struct _Node

Data Fields
AttVal *	attributes
Bool	closed
uint	column
Node *	content
tmbstr	element
uint	end
Bool	implicit
Node *	last
uint	line
Bool	linebreak
Node *	next
Node *	parent
Node *	prev
uint	start
const Dict *	tag
NodeType	type
const Dict *	was

struct _Style

Data Fields
TagStyle *	next
tmbstr	properties
tmbstr	tag
tmbstr	tag_class

struct _StyleProp

Data Fields
tmbstr	name
StyleProp *	next
tmbstr	value

Macro Definition Documentation

#define CM_BLOCK   (1 << 3)

#define CM_DEFLIST   (1 << 6)

#define CM_EMPTY   (1 << 0)

#define CM_FIELD   (1 << 10)

#define CM_FRAMES   (1 << 13)

#define CM_HEAD   (1 << 2)

#define CM_HEADING   (1 << 14)

#define CM_HTML   (1 << 1)

#define CM_IMG   (1 << 16)

#define CM_INLINE   (1 << 4)

#define CM_LIST   (1 << 5)

#define CM_MIXED   (1 << 17)

#define CM_NEW   (1 << 20)

#define CM_NO_INDENT   (1 << 18)

#define CM_OBJECT   (1 << 11)

#define CM_OBSOLETE   (1 << 19)

#define CM_OMITST   (1 << 21)

#define CM_OPT   (1 << 15)

#define CM_PARAM   (1 << 12)

#define CM_ROW   (1 << 9)

#define CM_ROWGRP   (1 << 8)

#define CM_TABLE   (1 << 7)

#define CM_UNKNOWN   0

#define digit   1u

#define digithex   128u

#define H40F   16u

#define H40S   4u

#define H40T   8u

#define H41F   128u

#define H41S   32u

#define H41T   64u

#define HT20   1u

#define HT32   2u

#define HT50   131072u

#define letter   2u

#define lowercase   32u

#define namechar   4u

#define newline   16u

#define uppercase   64u

#define VERS_ALL   (VERS_HTML20|VERS_HTML32|VERS_FROM40|XH50|HT50)

#define VERS_BASIC   (XB10)

#define VERS_EVENTS   (VERS_HTML40|VERS_XHTML11)

#define VERS_FRAMESET   (H40F|H41F|X10F)

#define VERS_FROM32   (VERS_HTML32|VERS_HTML40|HT50)

#define VERS_FROM40   (VERS_HTML40|VERS_XHTML11|VERS_BASIC|VERS_HTML5)

#define VERS_HTML20   (HT20)

#define VERS_HTML32   (HT32)

#define VERS_HTML40   (VERS_HTML40_STRICT|VERS_HTML40_LOOSE|VERS_FRAMESET)

#define VERS_HTML40_LOOSE   (H40T|H41T|X10T)

#define VERS_HTML40_STRICT   (H40S|H41S|X10S)

#define VERS_HTML5   (HT50|XH50)

#define VERS_IFRAME   (VERS_HTML40_LOOSE|VERS_FRAMESET)

#define VERS_LOOSE   (VERS_HTML20|VERS_HTML32|VERS_IFRAME)

#define VERS_MICROSOFT   32768u

#define VERS_NETSCAPE   16384u

#define VERS_PROPRIETARY   (VERS_NETSCAPE|VERS_MICROSOFT|VERS_SUN)

#define VERS_STRICT   (VERS_HTML5|VERS_HTML40_STRICT)

#define VERS_SUN   8192u

#define VERS_UNKNOWN   (xxxx)

#define VERS_XHTML   (X10S|X10T|X10F|XH11|XB10|XH50)

#define VERS_XHTML11   (XH11)

#define VERS_XML   65536u

#define white   8u

#define X10F   1024u

#define X10S   256u

#define X10T   512u

#define XB10   4096u

#define XH11   2048u

#define XH50   262144u

#define xxxx   0u

Enumeration Type Documentation

enum GetTokenMode

Enumerator
IgnoreWhitespace
MixedContent
Preformatted
IgnoreMarkup
OtherNamespace
CdataContent

enum LexerState

Enumerator
LEX_CONTENT
LEX_GT
LEX_ENDTAG
LEX_STARTTAG
LEX_COMMENT
LEX_DOCTYPE
LEX_PROCINSTR
LEX_CDATA
LEX_SECTION
LEX_ASP
LEX_JSTE
LEX_PHP
LEX_XMLDECL

enum NodeType

Enumerator
RootNode
DocTypeTag
CommentTag
ProcInsTag
TextNode
StartTag
EndTag
StartEndTag
CDATATag
SectionTag
AspTag
JsteTag
PhpTag
XmlDecl

enum ParseDocTypeDeclState

Enumerator
DT_INTERMEDIATE
DT_DOCTYPENAME
DT_PUBLICSYSTEM
DT_QUOTEDSTRING
DT_INTSUBSET

Function Documentation

void TY_❪AddCharToLexer❫	(	Lexer *	lexer,
		uint	c
	)

Bool TY_❪AddGenerator❫

(

TidyDocImpl *

doc

)

void TY_❪AddStringLiteral❫	(	Lexer *	lexer,
		ctmbstr	str
	)

uint TY_❪ApparentVersion❫

(

TidyDocImpl *

doc

)

Node* TY_❪CloneNode❫	(	TidyDocImpl *	doc,
		Node *	element
	)

void TY_❪ConstrainVersion❫	(	TidyDocImpl *	doc,
		uint	vers
	)

void TY_❪DeferDup❫

(

TidyDocImpl *

doc

)

void TY_❪DetachAttribute❫	(	Node *	node,
		AttVal *	attr
	)

AttVal* TY_❪DupAttrs❫	(	TidyDocImpl *	doc,
		AttVal *	attrs
	)

Node* TY_❪FindBody❫

(

TidyDocImpl *

doc

)

Node* TY_❪FindContainer❫

(

Node *

node

)

Node* TY_❪FindDocType❫

(

TidyDocImpl *

doc

)

Node* TY_❪FindHEAD❫

(

TidyDocImpl *

doc

)

Node* TY_❪FindHTML❫

(

TidyDocImpl *

doc

)

Node* TY_❪FindTITLE❫

(

TidyDocImpl *

doc

)

Node* TY_❪FindXmlDecl❫

(

TidyDocImpl *

doc

)

Bool TY_❪FixDocType❫

(

TidyDocImpl *

doc

)

Bool TY_❪FixXmlDecl❫

(

TidyDocImpl *

doc

)

void TY_❪FreeAttribute❫	(	TidyDocImpl *	doc,
		AttVal *	av
	)

void TY_❪FreeAttrs❫	(	TidyDocImpl *	doc,
		Node *	node
	)

void TY_❪FreeLexer❫

(

TidyDocImpl *

doc

)

void TY_❪FreeNode❫	(	TidyDocImpl *	doc,
		Node *	node
	)

Node* TY_❪GetToken❫	(	TidyDocImpl *	doc,
		GetTokenMode	mode
	)

ctmbstr TY_❪HTMLVersionNameFromCode❫	(	uint	vers,
		Bool	isXhtml
	)

uint TY_❪HTMLVersionNumberFromCode❫

(

uint

vers

)

int TY_❪HTMLVersion❫

(

TidyDocImpl *

doc

)

Node* TY_❪InferredTag❫	(	TidyDocImpl *	doc,
		TidyTagId	id
	)

void TY_❪InitMap❫

(

void

)

Bool TY_❪InlineDup1❫	(	TidyDocImpl *	doc,
		Node *	node,
		Node *	element
	)

int TY_❪InlineDup❫	(	TidyDocImpl *	doc,
		Node *	node
	)

void TY_❪InsertAttributeAtEnd❫	(	Node *	node,
		AttVal *	av
	)

void TY_❪InsertAttributeAtStart❫	(	Node *	node,
		AttVal *	av
	)

Node* TY_❪InsertedToken❫

(

TidyDocImpl *

doc

)

Bool TY_❪IsDigit❫

(

uint

c

)

Bool TY_❪IsHTMLSpace❫

(

uint

c

)

Bool TY_❪IsLetter❫

(

uint

c

)

Bool TY_❪IsNamechar❫

(

uint

c

)

Bool TY_❪IsNewline❫

(

uint

c

)

Bool TY_❪IsPushedLast❫	(	TidyDocImpl *	doc,
		Node *	element,
		Node *	node
	)

Bool TY_❪IsPushed❫	(	TidyDocImpl *	doc,
		Node *	node
	)

Bool TY_❪IsUpper❫

(

uint

c

)

Bool TY_❪IsWhite❫

(

uint

c

)

Bool TY_❪IsXMLLetter❫

(

uint

c

)

Bool TY_❪IsXMLNamechar❫

(

uint

c

)

AttVal* TY_❪NewAttributeEx❫	(	TidyDocImpl *	doc,
		ctmbstr	name,
		ctmbstr	value,
		int	delim
	)

AttVal* TY_❪NewAttribute❫

(

TidyDocImpl *

doc

)

Lexer* TY_❪NewLexer❫

(

TidyDocImpl *

doc

)

Node* TY_❪NewLineNode❫

(

Lexer *

lexer

)

Node* TY_❪NewLiteralTextNode❫	(	Lexer *	lexer,
		ctmbstr	txt
	)

Node* TY_❪NewNode❫	(	TidyAllocator *	allocator,
		Lexer *	lexer
	)

void TY_❪PopInline❫	(	TidyDocImpl *	doc,
		Node *	node
	)

void TY_❪PushInline❫	(	TidyDocImpl *	doc,
		Node *	node
	)

void TY_❪RemoveAttribute❫	(	TidyDocImpl *	doc,
		Node *	node,
		AttVal *	attr
	)

Bool TY_❪SetXHTMLDocType❫

(

TidyDocImpl *

doc

)

Bool TY_❪SwitchInline❫	(	TidyDocImpl *	doc,
		Node *	element,
		Node *	node
	)

Node* TY_❪TextToken❫

(

Lexer *

lexer

)

uint TY_❪ToLower❫

(

uint

c

)

uint TY_❪ToUpper❫

(

uint

c

)

void TY_❪UngetToken❫

(

TidyDocImpl *

doc

)

Bool TY_❪WarnMissingSIInEmittedDocType❫

(

TidyDocImpl *

doc

)