#include "xmldef.h"
#include "xmltok.h"
#include "nametab.h"
#include "xmltok_impl.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	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_INVALID3(p)

#define	UTF8_INVALID4(p) ((*p) == 0xF4 && ((p)[1] & 0x30) != 0)

#define	utf8_isName4 isNever

#define	utf8_isNmstrt4 isNever

#define	utf8_isInvalid2 isNever

#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) (((const struct normal_encoding *)(enc))->isName ## n(enc, p))

#define	IS_NMSTRT_CHAR(enc, p, n) (((const struct normal_encoding *)(enc))->isNmstrt ## n(enc, p))

#define	IS_INVALID_CHAR(enc, p, n) (((const struct 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	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	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	BT_COLON BT_NMSTRT

#define	BT_COLON BT_NMSTRT

#define	INIT_ENC_INDEX(enc) ((enc)->initEnc.isUtf16)

#define	NS(x) x

#define	ns(x) x

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	XmlUtf8Encode (int c, char *buf)

int	XmlUtf16Encode (int charNum, unsigned short *buf)

int	XmlSizeOfUnknownEncoding ()

ENCODING *	XmlInitUnknownEncoding (void mem, int table, int(convert)(void userData, const char p), void userData)

Macro Definition Documentation

◆ BIG2_BYTE_TO_ASCII

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

Definition at line 740 of file xmltok.c.

◆ 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]))

Definition at line 736 of file xmltok.c.

◆ BIG2_CHAR_MATCHES

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

Definition at line 741 of file xmltok.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])

Definition at line 742 of file xmltok.c.

◆ 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])

Definition at line 744 of file xmltok.c.

◆ 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)

Definition at line 789 of file xmltok.c.

◆ BYTE_TO_ASCII [2/3]

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

Definition at line 789 of file xmltok.c.

◆ BYTE_TO_ASCII [3/3]

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

Definition at line 789 of file xmltok.c.

◆ BYTE_TYPE [1/3]

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

Definition at line 788 of file xmltok.c.

◆ BYTE_TYPE [2/3]

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

Definition at line 788 of file xmltok.c.

◆ BYTE_TYPE [3/3]

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

Definition at line 788 of file xmltok.c.

◆ CHAR_MATCHES [1/3]

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

Definition at line 790 of file xmltok.c.

◆ CHAR_MATCHES [2/3]

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

Definition at line 790 of file xmltok.c.

◆ CHAR_MATCHES [3/3]

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

Definition at line 790 of file xmltok.c.

◆ DEFINE_UTF16_TO_UTF16

#define DEFINE_UTF16_TO_UTF16 ( E )

Value:

static \
void 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); \
}

Definition at line 559 of file xmltok.c.

◆ DEFINE_UTF16_TO_UTF8

#define DEFINE_UTF16_TO_UTF8 ( E )

Definition at line 496 of file xmltok.c.

◆ GET_HI [1/2]

#define GET_HI ( ptr ) ((unsigned char)(ptr)[1])

Definition at line 588 of file xmltok.c.

◆ GET_HI [2/2]

#define GET_HI ( ptr ) ((unsigned char)(ptr)[0])

Definition at line 588 of file xmltok.c.

◆ GET_LO [1/2]

#define GET_LO ( ptr ) ((unsigned char)(ptr)[0])

Definition at line 587 of file xmltok.c.

◆ GET_LO [2/2]

#define GET_LO ( ptr ) ((unsigned char)(ptr)[1])

Definition at line 587 of file xmltok.c.

◆ INIT_ENC_INDEX

#define INIT_ENC_INDEX ( enc ) ((enc)->initEnc.isUtf16)

Definition at line 1378 of file xmltok.c.

◆ IS_INVALID_CHAR

#define IS_INVALID_CHAR	(	enc,
		p,
		n
	)	(((const struct normal_encoding *)(enc))->isInvalid ## n(enc, p))

Definition at line 237 of file xmltok.c.

◆ IS_NAME_CHAR [1/3]

#define IS_NAME_CHAR	(	enc,
		p,
		n
	)	(((const struct normal_encoding *)(enc))->isName ## n(enc, p))

Definition at line 791 of file xmltok.c.

◆ IS_NAME_CHAR [2/3]

#define IS_NAME_CHAR	(	enc,
		p,
		n
	)	0

Definition at line 791 of file xmltok.c.

◆ IS_NAME_CHAR [3/3]

#define IS_NAME_CHAR	(	enc,
		p,
		n
	)	0

Definition at line 791 of file xmltok.c.

◆ IS_NAME_CHAR_MINBPC [1/3]

#define IS_NAME_CHAR_MINBPC	(	enc,
		p
	)	(0)

Definition at line 792 of file xmltok.c.

◆ IS_NAME_CHAR_MINBPC [2/3]

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

Definition at line 792 of file xmltok.c.

◆ IS_NAME_CHAR_MINBPC [3/3]

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

Definition at line 792 of file xmltok.c.

◆ IS_NMSTRT_CHAR [1/3]

#define IS_NMSTRT_CHAR	(	enc,
		p,
		n
	)	(((const struct normal_encoding *)(enc))->isNmstrt ## n(enc, p))

Definition at line 793 of file xmltok.c.

◆ IS_NMSTRT_CHAR [2/3]

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

Definition at line 793 of file xmltok.c.

◆ IS_NMSTRT_CHAR [3/3]

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

Definition at line 793 of file xmltok.c.

◆ IS_NMSTRT_CHAR_MINBPC [1/3]

#define IS_NMSTRT_CHAR_MINBPC	(	enc,
		p
	)	(0)

Definition at line 794 of file xmltok.c.

◆ IS_NMSTRT_CHAR_MINBPC [2/3]

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

Definition at line 794 of file xmltok.c.

◆ IS_NMSTRT_CHAR_MINBPC [3/3]

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

Definition at line 794 of file xmltok.c.

◆ LITTLE2_BYTE_TO_ASCII

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

Definition at line 601 of file xmltok.c.

◆ 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]))

Definition at line 597 of file xmltok.c.

◆ LITTLE2_CHAR_MATCHES

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

Definition at line 602 of file xmltok.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])

Definition at line 603 of file xmltok.c.

◆ 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])

Definition at line 605 of file xmltok.c.

◆ MINBPC [1/3]

#define MINBPC ( enc ) 1

Definition at line 786 of file xmltok.c.

◆ MINBPC [2/3]

#define MINBPC ( enc ) 2

Definition at line 786 of file xmltok.c.

◆ MINBPC [3/3]

#define MINBPC ( enc ) 2

Definition at line 786 of file xmltok.c.

◆ 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

Definition at line 179 of file xmltok.c.

◆ NS

#define NS ( x ) x

Definition at line 1503 of file xmltok.c.

◆ ns

#define ns ( x ) x

Definition at line 1504 of file xmltok.c.

◆ PREFIX [1/3]

#define PREFIX ( ident ) normal_ ## ident

Definition at line 785 of file xmltok.c.

◆ PREFIX [2/3]

#define PREFIX ( ident ) little2_ ## ident

Definition at line 785 of file xmltok.c.

◆ PREFIX [3/3]

#define PREFIX ( ident ) big2_ ## ident

Definition at line 785 of file xmltok.c.

◆ SB_BYTE_TYPE

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

Definition at line 206 of file xmltok.c.

◆ SET2 [1/2]

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

Definition at line 585 of file xmltok.c.

◆ SET2 [2/2]

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

Definition at line 585 of file xmltok.c.

◆ STANDARD_VTABLE

#define STANDARD_VTABLE ( E ) /* as nothing */

Definition at line 175 of file xmltok.c.

◆ UCS2_GET_NAMING

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

Definition at line 54 of file xmltok.c.

◆ 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))

Definition at line 79 of file xmltok.c.

◆ 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)))

Definition at line 61 of file xmltok.c.

◆ 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)))

Definition at line 71 of file xmltok.c.

◆ UTF8_INVALID3

#define UTF8_INVALID3 ( p )

Value:

  ((*p) == 0xED \
  ? (((p)[1] & 0x20) != 0) \
  : ((*p) == 0xEF \
     ? ((p)[1] == 0xBF && ((p)[2] == 0xBF || (p)[2] == 0xBE)) \
     : 0))

Definition at line 86 of file xmltok.c.

◆ UTF8_INVALID4

#define UTF8_INVALID4 ( p ) ((*p) == 0xF4 && ((p)[1] & 0x30) != 0)

Definition at line 93 of file xmltok.c.

◆ utf8_isInvalid2

#define utf8_isInvalid2 isNever

Definition at line 129 of file xmltok.c.

◆ utf8_isName4

#define utf8_isName4 isNever

Definition at line 113 of file xmltok.c.

◆ utf8_isNmstrt4

#define utf8_isNmstrt4 isNever

Definition at line 127 of file xmltok.c.

◆ VTABLE

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

Definition at line 52 of file xmltok.c.

◆ VTABLE1

#define VTABLE1

Value:

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

Definition at line 39 of file xmltok.c.

Enumeration Type Documentation

◆ anonymous enum

anonymous enum

Enumerator
UTF8_cval1
UTF8_cval2
UTF8_cval3
UTF8_cval4

Definition at line 276 of file xmltok.c.

     {  /* UTF8_cvalN is value of masked first byte of N byte sequence */
  UTF8_cval1 = 0x00,
  UTF8_cval2 = 0xc0,
  UTF8_cval3 = 0xe0,
  UTF8_cval4 = 0xf0
};

◆ 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

Definition at line 1343 of file xmltok.c.

     {
  UNKNOWN_ENC = -1,
  ISO_8859_1_ENC = 0,
  US_ASCII_ENC,
  UTF_8_ENC,
  UTF_16_ENC,
  UTF_16BE_ENC,
  UTF_16LE_ENC,
  /* must match encodingNames up to here */
  NO_ENC
};

Function Documentation

◆ XmlInitUnknownEncoding()

ENCODING * XmlInitUnknownEncoding	(	void *	mem,
		int *	table,
		int()(void userData, const char *p)	convert,
		void *	userData
	)

Definition at line 1263 of file xmltok.c.

{
  int i;
  struct unknown_encoding *e = mem;
  for (i = 0; i < sizeof(struct normal_encoding); i++)
    ((char *)mem)[i] = ((char *)&latin1_encoding)[i];
  for (i = 0; i < 128; i++)
    if (latin1_encoding.type[i] != BT_OTHER
        && latin1_encoding.type[i] != BT_NONXML
    && table[i] != i)
      return 0;
  for (i = 0; i < 256; i++) {
    int c = table[i];
    if (c == -1) {
      e->normal.type[i] = BT_MALFORM;
      /* This shouldn't really get used. */
      e->utf16[i] = 0xFFFF;
      e->utf8[i][0] = 1;
      e->utf8[i][1] = 0;
    }
    else if (c < 0) {
      if (c < -4)
    return 0;
      e->normal.type[i] = BT_LEAD2 - (c + 2);
      e->utf8[i][0] = 0;
      e->utf16[i] = 0;
    }
    else if (c < 0x80) {
      if (latin1_encoding.type[c] != BT_OTHER
      && latin1_encoding.type[c] != BT_NONXML
      && c != i)
    return 0;
      e->normal.type[i] = latin1_encoding.type[c];
      e->utf8[i][0] = 1;
      e->utf8[i][1] = (char)c;
      e->utf16[i] = c == 0 ? 0xFFFF : c;
    }
    else if (checkCharRefNumber(c) < 0) {
      e->normal.type[i] = BT_NONXML;
      /* This shouldn't really get used. */
      e->utf16[i] = 0xFFFF;
      e->utf8[i][0] = 1;
      e->utf8[i][1] = 0;
    }
    else {
      if (c > 0xFFFF)
    return 0;
      if (UCS2_GET_NAMING(nmstrtPages, c >> 8, c & 0xff))
    e->normal.type[i] = BT_NMSTRT;
      else if (UCS2_GET_NAMING(namePages, c >> 8, c & 0xff))
    e->normal.type[i] = BT_NAME;
      else
    e->normal.type[i] = BT_OTHER;
      e->utf8[i][0] = (char)XmlUtf8Encode(c, e->utf8[i] + 1);
      e->utf16[i] = c;
    }
  }
  e->userData = userData;
  e->convert = convert;
  if (convert) {
    e->normal.isName2 = unknown_isName;
    e->normal.isName3 = unknown_isName;
    e->normal.isName4 = unknown_isName;
    e->normal.isNmstrt2 = unknown_isNmstrt;
    e->normal.isNmstrt3 = unknown_isNmstrt;
    e->normal.isNmstrt4 = unknown_isNmstrt;
    e->normal.isInvalid2 = unknown_isInvalid;
    e->normal.isInvalid3 = unknown_isInvalid;
    e->normal.isInvalid4 = unknown_isInvalid;
  }
  e->normal.enc.utf8Convert = unknown_toUtf8;
  e->normal.enc.utf16Convert = unknown_toUtf16;
  return &(e->normal.enc);
}

◆ XmlSizeOfUnknownEncoding()

int XmlSizeOfUnknownEncoding ( )

Definition at line 1175 of file xmltok.c.

{
  return sizeof(struct unknown_encoding);
}

◆ XmlUtf16Encode()

int XmlUtf16Encode	(	int	charNum,
		unsigned short *	buf
	)

Definition at line 1150 of file xmltok.c.

{
  if (charNum < 0)
    return 0;
  if (charNum < 0x10000) {
    buf[0] = charNum;
    return 1;
  }
  if (charNum < 0x110000) {
    charNum -= 0x10000;
    buf[0] = (charNum >> 10) + 0xD800;
    buf[1] = (charNum & 0x3FF) + 0xDC00;
    return 2;
  }
  return 0;
}

◆ XmlUtf8Encode()

int XmlUtf8Encode	(	int	c,
		char *	buf
	)

Definition at line 1114 of file xmltok.c.

{
  enum {
    /* minN is minimum legal resulting value for N byte sequence */
    min2 = 0x80,
    min3 = 0x800,
    min4 = 0x10000
  };
 
  if (c < 0)
    return 0;
  if (c < min2) {
    buf[0] = (c | UTF8_cval1);
    return 1;
  }
  if (c < min3) {
    buf[0] = ((c >> 6) | UTF8_cval2);
    buf[1] = ((c & 0x3f) | 0x80);
    return 2;
  }
  if (c < min4) {
    buf[0] = ((c >> 12) | UTF8_cval3);
    buf[1] = (((c >> 6) & 0x3f) | 0x80);
    buf[2] = ((c & 0x3f) | 0x80);
    return 3;
  }
  if (c < 0x110000) {
    buf[0] = ((c >> 18) | UTF8_cval4);
    buf[1] = (((c >> 12) & 0x3f) | 0x80);
    buf[2] = (((c >> 6) & 0x3f) | 0x80);
    buf[3] = ((c & 0x3f) | 0x80);
    return 4;
  }
  return 0;
}

Referenced by XmlInitUnknownEncoding().

Classes

Macros

Enumerations

Functions

Macro Definition Documentation

◆ BIG2_BYTE_TO_ASCII

◆ BIG2_BYTE_TYPE

◆ BIG2_CHAR_MATCHES

◆ BIG2_IS_NAME_CHAR_MINBPC

◆ BIG2_IS_NMSTRT_CHAR_MINBPC

◆ BT_COLON [1/8]

◆ BT_COLON [2/8]

◆ BT_COLON [3/8]

◆ BT_COLON [4/8]

◆ BT_COLON [5/8]

◆ BT_COLON [6/8]

◆ BT_COLON [7/8]

◆ BT_COLON [8/8]

◆ BYTE_TO_ASCII [1/3]

◆ BYTE_TO_ASCII [2/3]

◆ BYTE_TO_ASCII [3/3]

◆ BYTE_TYPE [1/3]

◆ BYTE_TYPE [2/3]

◆ BYTE_TYPE [3/3]

◆ CHAR_MATCHES [1/3]

◆ CHAR_MATCHES [2/3]

◆ CHAR_MATCHES [3/3]

◆ DEFINE_UTF16_TO_UTF16

◆ DEFINE_UTF16_TO_UTF8

◆ GET_HI [1/2]

◆ GET_HI [2/2]

◆ GET_LO [1/2]

◆ GET_LO [2/2]

◆ INIT_ENC_INDEX

◆ IS_INVALID_CHAR

◆ IS_NAME_CHAR [1/3]

◆ IS_NAME_CHAR [2/3]

◆ IS_NAME_CHAR [3/3]

◆ IS_NAME_CHAR_MINBPC [1/3]

◆ IS_NAME_CHAR_MINBPC [2/3]

◆ IS_NAME_CHAR_MINBPC [3/3]

◆ IS_NMSTRT_CHAR [1/3]

◆ IS_NMSTRT_CHAR [2/3]

◆ IS_NMSTRT_CHAR [3/3]

◆ IS_NMSTRT_CHAR_MINBPC [1/3]

◆ IS_NMSTRT_CHAR_MINBPC [2/3]

◆ IS_NMSTRT_CHAR_MINBPC [3/3]

◆ LITTLE2_BYTE_TO_ASCII

◆ LITTLE2_BYTE_TYPE

◆ LITTLE2_CHAR_MATCHES

◆ LITTLE2_IS_NAME_CHAR_MINBPC

◆ LITTLE2_IS_NMSTRT_CHAR_MINBPC

◆ MINBPC [1/3]

◆ MINBPC [2/3]

◆ MINBPC [3/3]

◆ NORMAL_VTABLE

◆ NS

◆ ns

◆ PREFIX [1/3]

◆ PREFIX [2/3]

◆ PREFIX [3/3]

◆ SB_BYTE_TYPE

◆ SET2 [1/2]

◆ SET2 [2/2]

◆ STANDARD_VTABLE

◆ UCS2_GET_NAMING

◆ UTF8_GET_NAMING

◆ UTF8_GET_NAMING2

◆ UTF8_GET_NAMING3

◆ UTF8_INVALID3

◆ UTF8_INVALID4

◆ utf8_isInvalid2

◆ utf8_isName4

◆ utf8_isNmstrt4

◆ VTABLE

◆ VTABLE1

Enumeration Type Documentation

◆ anonymous enum

◆ anonymous enum

Function Documentation