xmltok.c File Reference

#include <stddef.h>
#include "expat_external.h"
#include "internal.h"
#include "xmltok.h"
#include "nametab.h"
#include "xmltok_impl.h"
#include "ascii.h"
#include "xmltok_impl.c"
#include "asciitab.h"
#include "utf8tab.h"
#include "iasciitab.h"
#include "latin1tab.h"
#include "xmltok_ns.c"

Classes
struct	normal_encoding
struct	unknown_encoding

Macros
#define	IGNORE_SECTION_TOK_VTABLE   /* as nothing */
#define	VTABLE1
#define	VTABLE   VTABLE1, PREFIX(toUtf8), PREFIX(toUtf16)
#define	UCS2_GET_NAMING(pages, hi, lo)   (namingBitmap[(pages[hi] << 3) + ((lo) >> 5)] & (1 << ((lo) & 0x1F)))
#define	UTF8_GET_NAMING2(pages, byte)
#define	UTF8_GET_NAMING3(pages, byte)
#define	UTF8_GET_NAMING(pages, p, n)
#define	UTF8_INVALID2(p)   ((*p) < 0xC2 || ((p)[1] & 0x80) == 0 || ((p)[1] & 0xC0) == 0xC0)
#define	UTF8_INVALID3(p)
#define	UTF8_INVALID4(p)
#define	utf8_isName4   isNever
#define	utf8_isNmstrt4   isNever
#define	AS_NORMAL_ENCODING(enc)   ((const struct normal_encoding *) (enc))
#define	STANDARD_VTABLE(E)   /* as nothing */
#define	NORMAL_VTABLE(E)
#define	MINBPC(enc)   1
#define	SB_BYTE_TYPE(enc, p)   (((struct normal_encoding *)(enc))->type[(unsigned char)*(p)])
#define	BYTE_TYPE(enc, p)   SB_BYTE_TYPE(enc, p)
#define	BYTE_TO_ASCII(enc, p)   (*(p))
#define	IS_NAME_CHAR(enc, p, n)   (AS_NORMAL_ENCODING(enc)->isName ## n(enc, p))
#define	IS_NMSTRT_CHAR(enc, p, n)   (AS_NORMAL_ENCODING(enc)->isNmstrt ## n(enc, p))
#define	IS_INVALID_CHAR(enc, p, n)   (AS_NORMAL_ENCODING(enc)->isInvalid ## n(enc, p))
#define	IS_NAME_CHAR_MINBPC(enc, p)   (0)
#define	IS_NMSTRT_CHAR_MINBPC(enc, p)   (0)
#define	CHAR_MATCHES(enc, p, c)   (*(p) == c)
#define	PREFIX(ident)   normal_ ## ident
#define	XML_TOK_IMPL_C
#define	BT_COLON   BT_NMSTRT
#define	BT_COLON   BT_NMSTRT
#define	BT_COLON   BT_NMSTRT
#define	BT_COLON   BT_NMSTRT
#define	DEFINE_UTF16_TO_UTF8(E)
#define	DEFINE_UTF16_TO_UTF16(E)
#define	SET2(ptr, ch)   (((ptr)[0] = ((ch) & 0xff)), ((ptr)[1] = ((ch) >> 8)))
#define	GET_LO(ptr)   ((unsigned char)(ptr)[0])
#define	GET_HI(ptr)   ((unsigned char)(ptr)[1])
#define	SET2(ptr, ch)   (((ptr)[0] = ((ch) >> 8)), ((ptr)[1] = ((ch) & 0xFF)))
#define	GET_LO(ptr)   ((unsigned char)(ptr)[1])
#define	GET_HI(ptr)   ((unsigned char)(ptr)[0])
#define	LITTLE2_BYTE_TYPE(enc, p)
#define	LITTLE2_BYTE_TO_ASCII(enc, p)   ((p)[1] == 0 ? (p)[0] : -1)
#define	LITTLE2_CHAR_MATCHES(enc, p, c)   ((p)[1] == 0 && (p)[0] == c)
#define	LITTLE2_IS_NAME_CHAR_MINBPC(enc, p)   UCS2_GET_NAMING(namePages, (unsigned char)p[1], (unsigned char)p[0])
#define	LITTLE2_IS_NMSTRT_CHAR_MINBPC(enc, p)   UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[1], (unsigned char)p[0])
#define	PREFIX(ident)   little2_ ## ident
#define	MINBPC(enc)   2
#define	BYTE_TYPE(enc, p)   LITTLE2_BYTE_TYPE(enc, p)
#define	BYTE_TO_ASCII(enc, p)   LITTLE2_BYTE_TO_ASCII(enc, p)
#define	CHAR_MATCHES(enc, p, c)   LITTLE2_CHAR_MATCHES(enc, p, c)
#define	IS_NAME_CHAR(enc, p, n)   0
#define	IS_NAME_CHAR_MINBPC(enc, p)   LITTLE2_IS_NAME_CHAR_MINBPC(enc, p)
#define	IS_NMSTRT_CHAR(enc, p, n)   (0)
#define	IS_NMSTRT_CHAR_MINBPC(enc, p)   LITTLE2_IS_NMSTRT_CHAR_MINBPC(enc, p)
#define	XML_TOK_IMPL_C
#define	BT_COLON   BT_NMSTRT
#define	BT_COLON   BT_NMSTRT
#define	BIG2_BYTE_TYPE(enc, p)
#define	BIG2_BYTE_TO_ASCII(enc, p)   ((p)[0] == 0 ? (p)[1] : -1)
#define	BIG2_CHAR_MATCHES(enc, p, c)   ((p)[0] == 0 && (p)[1] == c)
#define	BIG2_IS_NAME_CHAR_MINBPC(enc, p)   UCS2_GET_NAMING(namePages, (unsigned char)p[0], (unsigned char)p[1])
#define	BIG2_IS_NMSTRT_CHAR_MINBPC(enc, p)   UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[0], (unsigned char)p[1])
#define	PREFIX(ident)   big2_ ## ident
#define	MINBPC(enc)   2
#define	BYTE_TYPE(enc, p)   BIG2_BYTE_TYPE(enc, p)
#define	BYTE_TO_ASCII(enc, p)   BIG2_BYTE_TO_ASCII(enc, p)
#define	CHAR_MATCHES(enc, p, c)   BIG2_CHAR_MATCHES(enc, p, c)
#define	IS_NAME_CHAR(enc, p, n)   0
#define	IS_NAME_CHAR_MINBPC(enc, p)   BIG2_IS_NAME_CHAR_MINBPC(enc, p)
#define	IS_NMSTRT_CHAR(enc, p, n)   (0)
#define	IS_NMSTRT_CHAR_MINBPC(enc, p)   BIG2_IS_NMSTRT_CHAR_MINBPC(enc, p)
#define	XML_TOK_IMPL_C
#define	BT_COLON   BT_NMSTRT
#define	BT_COLON   BT_NMSTRT
#define	AS_UNKNOWN_ENCODING(enc)   ((const struct unknown_encoding *) (enc))
#define	INIT_ENC_INDEX(enc)   ((int)(enc)->initEnc.isUtf16)
#define	SET_INIT_ENC_INDEX(enc, i)   ((enc)->initEnc.isUtf16 = (char)i)
#define	NS(x)   x
#define	ns(x)   x
#define	XML_TOK_NS_C

Enumerations
enum	{ UTF8_cval1 = 0x00, UTF8_cval2 = 0xc0, UTF8_cval3 = 0xe0, UTF8_cval4 = 0xf0 }
enum	{   UNKNOWN_ENC = -1, ISO_8859_1_ENC = 0, US_ASCII_ENC, UTF_8_ENC,   UTF_16_ENC, UTF_16BE_ENC, UTF_16LE_ENC, NO_ENC }

Functions
int FASTCALL	XmlUtf8Encode (int c, char *buf)
int FASTCALL	XmlUtf16Encode (int charNum, unsigned short *buf)
int	XmlSizeOfUnknownEncoding (void)
ENCODING *	XmlInitUnknownEncoding (void *mem, int *table, CONVERTER convert, void *userData)

Macro Definition Documentation

AS_NORMAL_ENCODING

#define AS_NORMAL_ENCODING

(

enc

)

((const struct normal_encoding *) (enc))

AS_UNKNOWN_ENCODING

#define AS_UNKNOWN_ENCODING

(

enc

)

((const struct unknown_encoding *) (enc))

BIG2_BYTE_TO_ASCII

#define BIG2_BYTE_TO_ASCII	(		enc,
			p
	)		((p)[0] == 0 ? (p)[1] : -1)

BIG2_BYTE_TYPE

#define BIG2_BYTE_TYPE	(		enc,
			p
	)

Value:

((p)[0] == 0 \
  ? ((struct normal_encoding *)(enc))->type[(unsigned char)(p)[1]] \
  : unicode_byte_type((p)[0], (p)[1]))

BIG2_CHAR_MATCHES

#define BIG2_CHAR_MATCHES	(		enc,
			p,
			c
	)		((p)[0] == 0 && (p)[1] == c)

BIG2_IS_NAME_CHAR_MINBPC

#define BIG2_IS_NAME_CHAR_MINBPC	(		enc,
			p
	)		UCS2_GET_NAMING(namePages, (unsigned char)p[0], (unsigned char)p[1])

BIG2_IS_NMSTRT_CHAR_MINBPC

#define BIG2_IS_NMSTRT_CHAR_MINBPC	(		enc,
			p
	)		UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[0], (unsigned char)p[1])

BT_COLON [1/8]

#define BT_COLON   BT_NMSTRT

BT_COLON [2/8]

#define BT_COLON   BT_NMSTRT

BT_COLON [3/8]

#define BT_COLON   BT_NMSTRT

BT_COLON [4/8]

#define BT_COLON   BT_NMSTRT

BT_COLON [5/8]

#define BT_COLON   BT_NMSTRT

BT_COLON [6/8]

#define BT_COLON   BT_NMSTRT

BT_COLON [7/8]

#define BT_COLON   BT_NMSTRT

BT_COLON [8/8]

#define BT_COLON   BT_NMSTRT

BYTE_TO_ASCII [1/3]

#define BYTE_TO_ASCII	(		enc,
			p
	)		(*(p))

BYTE_TO_ASCII [2/3]

#define BYTE_TO_ASCII	(		enc,
			p
	)		LITTLE2_BYTE_TO_ASCII(enc, p)

BYTE_TO_ASCII [3/3]

#define BYTE_TO_ASCII	(		enc,
			p
	)		BIG2_BYTE_TO_ASCII(enc, p)

BYTE_TYPE [1/3]

#define BYTE_TYPE	(		enc,
			p
	)		SB_BYTE_TYPE(enc, p)

BYTE_TYPE [2/3]

#define BYTE_TYPE	(		enc,
			p
	)		LITTLE2_BYTE_TYPE(enc, p)

BYTE_TYPE [3/3]

#define BYTE_TYPE	(		enc,
			p
	)		BIG2_BYTE_TYPE(enc, p)

CHAR_MATCHES [1/3]

#define CHAR_MATCHES	(		enc,
			p,
			c
	)		(*(p) == c)

CHAR_MATCHES [2/3]

#define CHAR_MATCHES	(		enc,
			p,
			c
	)		LITTLE2_CHAR_MATCHES(enc, p, c)

CHAR_MATCHES [3/3]

#define CHAR_MATCHES	(		enc,
			p,
			c
	)		BIG2_CHAR_MATCHES(enc, p, c)

DEFINE_UTF16_TO_UTF16

#define DEFINE_UTF16_TO_UTF16

(

E

)

Value:

static void  PTRCALL \
E ## toUtf16(const ENCODING *enc, \
             const char **fromP, const char *fromLim, \
             unsigned short **toP, const unsigned short *toLim) \
{ \
  /* Avoid copying first half only of surrogate */ \
  if (fromLim - *fromP > ((toLim - *toP) << 1) \
      && (GET_HI(fromLim - 2) & 0xF8) == 0xD8) \
    fromLim -= 2; \
  for (; *fromP != fromLim && *toP != toLim; *fromP += 2) \
    *(*toP)++ = (GET_HI(*fromP) << 8) | GET_LO(*fromP); \
}

DEFINE_UTF16_TO_UTF8

#define DEFINE_UTF16_TO_UTF8

(

E

)

GET_HI [1/2]

#define GET_HI

(

ptr

)

((unsigned char)(ptr)[1])

GET_HI [2/2]

#define GET_HI

(

ptr

)

((unsigned char)(ptr)[0])

GET_LO [1/2]

#define GET_LO

(

ptr

)

((unsigned char)(ptr)[0])

GET_LO [2/2]

#define GET_LO

(

ptr

)

((unsigned char)(ptr)[1])

IGNORE_SECTION_TOK_VTABLE

#define IGNORE_SECTION_TOK_VTABLE   /* as nothing */

INIT_ENC_INDEX

#define INIT_ENC_INDEX

(

enc

)

((int)(enc)->initEnc.isUtf16)

IS_INVALID_CHAR

#define IS_INVALID_CHAR	(		enc,
			p,
			n
	)		(AS_NORMAL_ENCODING(enc)->isInvalid ## n(enc, p))

IS_NAME_CHAR [1/3]

#define IS_NAME_CHAR	(		enc,
			p,
			n
	)		(AS_NORMAL_ENCODING(enc)->isName ## n(enc, p))

IS_NAME_CHAR [2/3]

#define IS_NAME_CHAR	(		enc,
			p,
			n
	)		0

IS_NAME_CHAR [3/3]

#define IS_NAME_CHAR	(		enc,
			p,
			n
	)		0

IS_NAME_CHAR_MINBPC [1/3]

#define IS_NAME_CHAR_MINBPC	(		enc,
			p
	)		(0)

IS_NAME_CHAR_MINBPC [2/3]

#define IS_NAME_CHAR_MINBPC	(		enc,
			p
	)		LITTLE2_IS_NAME_CHAR_MINBPC(enc, p)

IS_NAME_CHAR_MINBPC [3/3]

#define IS_NAME_CHAR_MINBPC	(		enc,
			p
	)		BIG2_IS_NAME_CHAR_MINBPC(enc, p)

IS_NMSTRT_CHAR [1/3]

#define IS_NMSTRT_CHAR	(		enc,
			p,
			n
	)		(AS_NORMAL_ENCODING(enc)->isNmstrt ## n(enc, p))

IS_NMSTRT_CHAR [2/3]

#define IS_NMSTRT_CHAR	(		enc,
			p,
			n
	)		(0)

IS_NMSTRT_CHAR [3/3]

#define IS_NMSTRT_CHAR	(		enc,
			p,
			n
	)		(0)

IS_NMSTRT_CHAR_MINBPC [1/3]

#define IS_NMSTRT_CHAR_MINBPC	(		enc,
			p
	)		(0)

IS_NMSTRT_CHAR_MINBPC [2/3]

#define IS_NMSTRT_CHAR_MINBPC	(		enc,
			p
	)		LITTLE2_IS_NMSTRT_CHAR_MINBPC(enc, p)

IS_NMSTRT_CHAR_MINBPC [3/3]

#define IS_NMSTRT_CHAR_MINBPC	(		enc,
			p
	)		BIG2_IS_NMSTRT_CHAR_MINBPC(enc, p)

LITTLE2_BYTE_TO_ASCII

#define LITTLE2_BYTE_TO_ASCII	(		enc,
			p
	)		((p)[1] == 0 ? (p)[0] : -1)

LITTLE2_BYTE_TYPE

#define LITTLE2_BYTE_TYPE	(		enc,
			p
	)

Value:

((p)[1] == 0 \
  ? ((struct normal_encoding *)(enc))->type[(unsigned char)*(p)] \
  : unicode_byte_type((p)[1], (p)[0]))

LITTLE2_CHAR_MATCHES

#define LITTLE2_CHAR_MATCHES	(		enc,
			p,
			c
	)		((p)[1] == 0 && (p)[0] == c)

LITTLE2_IS_NAME_CHAR_MINBPC

#define LITTLE2_IS_NAME_CHAR_MINBPC	(		enc,
			p
	)		UCS2_GET_NAMING(namePages, (unsigned char)p[1], (unsigned char)p[0])

LITTLE2_IS_NMSTRT_CHAR_MINBPC

#define LITTLE2_IS_NMSTRT_CHAR_MINBPC	(		enc,
			p
	)		UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[1], (unsigned char)p[0])

MINBPC [1/3]

#define MINBPC

(

enc

)

1

MINBPC [2/3]

#define MINBPC

(

enc

)

2

MINBPC [3/3]

#define MINBPC

(

enc

)

2

NORMAL_VTABLE

#define NORMAL_VTABLE

(

E

)

Value:

E ## isName2, \
 E ## isName3, \
 E ## isName4, \
 E ## isNmstrt2, \
 E ## isNmstrt3, \
 E ## isNmstrt4, \
 E ## isInvalid2, \
 E ## isInvalid3, \
 E ## isInvalid4

NS

#define NS

(

x

)

x

ns

#define ns

(

x

)

x

PREFIX [1/3]

#define PREFIX

(

ident

)

normal_ ## ident

PREFIX [2/3]

#define PREFIX

(

ident

)

little2_ ## ident

PREFIX [3/3]

#define PREFIX

(

ident

)

big2_ ## ident

SB_BYTE_TYPE

#define SB_BYTE_TYPE	(		enc,
			p
	)		(((struct normal_encoding *)(enc))->type[(unsigned char)*(p)])

SET2 [1/2]

#define SET2	(		ptr,
			ch
	)		(((ptr)[0] = ((ch) & 0xff)), ((ptr)[1] = ((ch) >> 8)))

SET2 [2/2]

#define SET2	(		ptr,
			ch
	)		(((ptr)[0] = ((ch) >> 8)), ((ptr)[1] = ((ch) & 0xFF)))

SET_INIT_ENC_INDEX

#define SET_INIT_ENC_INDEX	(		enc,
			i
	)		((enc)->initEnc.isUtf16 = (char)i)

STANDARD_VTABLE

#define STANDARD_VTABLE

(

E

)

/* as nothing */

UCS2_GET_NAMING

#define UCS2_GET_NAMING	(		pages,
			hi,
			lo
	)		(namingBitmap[(pages[hi] << 3) + ((lo) >> 5)] & (1 << ((lo) & 0x1F)))

UTF8_GET_NAMING

#define UTF8_GET_NAMING	(		pages,
			p,
			n
	)

Value:

((n) == 2 \
  ? UTF8_GET_NAMING2(pages, (const unsigned char *)(p)) \
  : ((n) == 3 \
     ? UTF8_GET_NAMING3(pages, (const unsigned char *)(p)) \
     : 0))

UTF8_GET_NAMING2

#define UTF8_GET_NAMING2	(		pages,
			byte
	)

Value:

(namingBitmap[((pages)[(((byte)[0]) >> 2) & 7] << 3) \
                      + ((((byte)[0]) & 3) << 1) \
                      + ((((byte)[1]) >> 5) & 1)] \
         & (1 << (((byte)[1]) & 0x1F)))

UTF8_GET_NAMING3

#define UTF8_GET_NAMING3	(		pages,
			byte
	)

Value:

(namingBitmap[((pages)[((((byte)[0]) & 0xF) << 4) \
                             + ((((byte)[1]) >> 2) & 0xF)] \
                       << 3) \
                      + ((((byte)[1]) & 3) << 1) \
                      + ((((byte)[2]) >> 5) & 1)] \
         & (1 << (((byte)[2]) & 0x1F)))

UTF8_INVALID2

#define UTF8_INVALID2

(

p

)

((*p) < 0xC2 || ((p)[1] & 0x80) == 0 || ((p)[1] & 0xC0) == 0xC0)

UTF8_INVALID3

#define UTF8_INVALID3

(

p

)

Value:

(((p)[2] & 0x80) == 0 \
  || \
  ((*p) == 0xEF && (p)[1] == 0xBF \
    ? \
    (p)[2] > 0xBD \
    : \
    ((p)[2] & 0xC0) == 0xC0) \
  || \
  ((*p) == 0xE0 \
    ? \
    (p)[1] < 0xA0 || ((p)[1] & 0xC0) == 0xC0 \
    : \
    ((p)[1] & 0x80) == 0 \
    || \
    ((*p) == 0xED ? (p)[1] > 0x9F : ((p)[1] & 0xC0) == 0xC0)))

UTF8_INVALID4

#define UTF8_INVALID4

(

p

)

Value:

(((p)[3] & 0x80) == 0 || ((p)[3] & 0xC0) == 0xC0 \
  || \
  ((p)[2] & 0x80) == 0 || ((p)[2] & 0xC0) == 0xC0 \
  || \
  ((*p) == 0xF0 \
    ? \
    (p)[1] < 0x90 || ((p)[1] & 0xC0) == 0xC0 \
    : \
    ((p)[1] & 0x80) == 0 \
    || \
    ((*p) == 0xF4 ? (p)[1] > 0x8F : ((p)[1] & 0xC0) == 0xC0)))

utf8_isName4

#define utf8_isName4   isNever

utf8_isNmstrt4

#define utf8_isNmstrt4   isNever

VTABLE

#define VTABLE   VTABLE1, PREFIX(toUtf8), PREFIX(toUtf16)

VTABLE1

#define VTABLE1

Value:

{ PREFIX(prologTok), PREFIX(contentTok), \
    PREFIX(cdataSectionTok) IGNORE_SECTION_TOK_VTABLE }, \
  { PREFIX(attributeValueTok), PREFIX(entityValueTok) }, \
  PREFIX(sameName), \
  PREFIX(nameMatchesAscii), \
  PREFIX(nameLength), \
  PREFIX(skipS), \
  PREFIX(getAtts), \
  PREFIX(charRefNumber), \
  PREFIX(predefinedEntityName), \
  PREFIX(updatePosition), \
  PREFIX(isPublicId)

XML_TOK_IMPL_C [1/3]

#define XML_TOK_IMPL_C

XML_TOK_IMPL_C [2/3]

#define XML_TOK_IMPL_C

XML_TOK_IMPL_C [3/3]

#define XML_TOK_IMPL_C

XML_TOK_NS_C

#define XML_TOK_NS_C

Enumeration Type Documentation

anonymous enum

anonymous enum

Enumerator
UTF8_cval1
UTF8_cval2
UTF8_cval3
UTF8_cval4

anonymous enum

anonymous enum

Enumerator
UNKNOWN_ENC
ISO_8859_1_ENC
US_ASCII_ENC
UTF_8_ENC
UTF_16_ENC
UTF_16BE_ENC
UTF_16LE_ENC
NO_ENC

Function Documentation

XmlInitUnknownEncoding()

ENCODING* XmlInitUnknownEncoding	(	void *	mem,
		int *	table,
		CONVERTER	convert,
		void *	userData
	)

XmlSizeOfUnknownEncoding()

int XmlSizeOfUnknownEncoding

(

void

)

XmlUtf16Encode()

int FASTCALL XmlUtf16Encode	(	int	charNum,
		unsigned short *	buf
	)

XmlUtf8Encode()

int FASTCALL XmlUtf8Encode	(	int	c,
		char *	buf
	)