#include <expat_config.h>
#include <stddef.h>
#include <string.h>
#include <stdbool.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 , PREFIX(ignoreSectionTok)

#define	VTABLE1

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

#define	UCS2_GET_NAMING(pages, hi, lo)

#define	UTF8_GET_NAMING2(pages, byte)

#define	UTF8_GET_NAMING3(pages, byte)

#define	UTF8_INVALID2(p)

#define	UTF8_INVALID3(p)

#define	UTF8_INVALID4(p)

#define	utf8_isName4 isNever

#define	utf8_isNmstrt4 isNever

#define	AS_NORMAL_ENCODING(enc)

#define	STANDARD_VTABLE(E)

#define	NORMAL_VTABLE(E)

#define	NULL_VTABLE

#define	MINBPC(enc)

#define	SB_BYTE_TYPE(enc, p)

#define	BYTE_TYPE(enc, p)

#define	BYTE_TO_ASCII(enc, p)

#define	IS_NAME_CHAR(enc, p, n)

#define	IS_NMSTRT_CHAR(enc, p, n)

#define	IS_INVALID_CHAR(enc, p, n)

#define	IS_NAME_CHAR_MINBPC(enc, p)

#define	IS_NMSTRT_CHAR_MINBPC(enc, p)

#define	CHAR_MATCHES(enc, p, c)

#define	PREFIX(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)

#define	GET_LO(ptr)

#define	GET_HI(ptr)

#define	SET2(ptr, ch)

#define	GET_LO(ptr)

#define	GET_HI(ptr)

#define	LITTLE2_BYTE_TYPE(enc, p)

#define	LITTLE2_BYTE_TO_ASCII(p)

#define	LITTLE2_CHAR_MATCHES(p, c)

#define	LITTLE2_IS_NAME_CHAR_MINBPC(p)

#define	LITTLE2_IS_NMSTRT_CHAR_MINBPC(p)

#define	PREFIX(ident)

#define	MINBPC(enc)

#define	BYTE_TYPE(enc, p)

#define	BYTE_TO_ASCII(enc, p)

#define	CHAR_MATCHES(enc, p, c)

#define	IS_NAME_CHAR(enc, p, n)

#define	IS_NAME_CHAR_MINBPC(enc, p)

#define	IS_NMSTRT_CHAR(enc, p, n)

#define	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(p)

#define	BIG2_CHAR_MATCHES(p, c)

#define	BIG2_IS_NAME_CHAR_MINBPC(p)

#define	BIG2_IS_NMSTRT_CHAR_MINBPC(p)

#define	PREFIX(ident)

#define	MINBPC(enc)

#define	BYTE_TYPE(enc, p)

#define	BYTE_TO_ASCII(enc, p)

#define	CHAR_MATCHES(enc, p, c)

#define	IS_NAME_CHAR(enc, p, n)

#define	IS_NAME_CHAR_MINBPC(enc, p)

#define	IS_NMSTRT_CHAR(enc, p, n)

#define	IS_NMSTRT_CHAR_MINBPC(enc, p)

#define	XML_TOK_IMPL_C

#define	BT_COLON BT_NMSTRT

#define	AS_UNKNOWN_ENCODING(enc)

#define	INIT_ENC_INDEX(enc)

#define	SET_INIT_ENC_INDEX(enc, i)

#define	NS(x)

#define	ns(x)

#define	XML_TOK_NS_C

#define	NS(x)

#define	ns(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
void	_INTERNAL_trim_to_complete_utf8_characters (const char from, const char *fromLimRef)

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)

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

Macro Definition Documentation

◆ AS_NORMAL_ENCODING

#define AS_NORMAL_ENCODING ( enc )

Value:

((const struct normal_encoding *)(enc))

normal_encoding

Definition xmltok.c:185

normal_encoding::enc

ENCODING enc

Definition xmltok.c:186

Definition at line 206 of file xmltok.c.

◆ AS_UNKNOWN_ENCODING

#define AS_UNKNOWN_ENCODING ( enc )

Value:

((const struct unknown_encoding *)(enc))

unknown_encoding

Definition xmltok.c:1314

Definition at line 1322 of file xmltok.c.

◆ BIG2_BYTE_TO_ASCII

#define BIG2_BYTE_TO_ASCII ( p )

Value:

((p)[0] == 0 ? (p)[1] : -1)

Definition at line 878 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 874 of file xmltok.c.

#define BIG2_BYTE_TYPE(enc, p)                                                 \
  ((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	(		p,
			c )

Value:

((p)[0] == 0 && (p)[1] == (c))

Definition at line 879 of file xmltok.c.

◆ BIG2_IS_NAME_CHAR_MINBPC

#define BIG2_IS_NAME_CHAR_MINBPC ( p )

Value:

UCS2_GET_NAMING(namePages, (unsigned char)p[0], (unsigned char)p[1])

UCS2_GET_NAMING

#define UCS2_GET_NAMING(pages, hi, lo)

Definition xmltok.c:78

Definition at line 880 of file xmltok.c.

880#define BIG2_IS_NAME_CHAR_MINBPC(p) \

881 UCS2_GET_NAMING(namePages, (unsigned char)p[0], (unsigned char)p[1])

◆ BIG2_IS_NMSTRT_CHAR_MINBPC

#define BIG2_IS_NMSTRT_CHAR_MINBPC ( p )

Value:

UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[0], (unsigned char)p[1])

Definition at line 882 of file xmltok.c.

882#define BIG2_IS_NMSTRT_CHAR_MINBPC(p) \

883 UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[0], (unsigned char)p[1])

◆ BT_COLON [1/7]

#define BT_COLON BT_NMSTRT

Referenced by XmlInitUnknownEncodingNS().

◆ BT_COLON [2/7]

#define BT_COLON BT_NMSTRT

◆ BT_COLON [3/7]

#define BT_COLON BT_NMSTRT

◆ BT_COLON [4/7]

#define BT_COLON BT_NMSTRT

◆ BT_COLON [5/7]

#define BT_COLON BT_NMSTRT

◆ BT_COLON [6/7]

#define BT_COLON BT_NMSTRT

◆ BT_COLON [7/7]

#define BT_COLON BT_NMSTRT

◆ BYTE_TO_ASCII [1/3]

#define BYTE_TO_ASCII	(		enc,
			p )

Value:

(*(p))

Definition at line 266 of file xmltok.c.

◆ BYTE_TO_ASCII [2/3]

#define BYTE_TO_ASCII	(		enc,
			p )

Value:

LITTLE2_BYTE_TO_ASCII(p)

LITTLE2_BYTE_TO_ASCII

#define LITTLE2_BYTE_TO_ASCII(p)

Definition xmltok.c:743

Definition at line 266 of file xmltok.c.

◆ BYTE_TO_ASCII [3/3]

#define BYTE_TO_ASCII	(		enc,
			p )

Value:

BIG2_BYTE_TO_ASCII(p)

BIG2_BYTE_TO_ASCII

#define BIG2_BYTE_TO_ASCII(p)

Definition xmltok.c:878

Definition at line 266 of file xmltok.c.

◆ BYTE_TYPE [1/3]

#define BYTE_TYPE	(		enc,
			p )

Value:

SB_BYTE_TYPE(enc, p)

SB_BYTE_TYPE

#define SB_BYTE_TYPE(enc, p)

Definition xmltok.c:245

Definition at line 255 of file xmltok.c.

◆ BYTE_TYPE [2/3]

#define BYTE_TYPE	(		enc,
			p )

Value:

LITTLE2_BYTE_TYPE(enc, p)

LITTLE2_BYTE_TYPE

#define LITTLE2_BYTE_TYPE(enc, p)

Definition xmltok.c:740

Definition at line 255 of file xmltok.c.

◆ BYTE_TYPE [3/3]

#define BYTE_TYPE	(		enc,
			p )

Value:

BIG2_BYTE_TYPE(enc, p)

BIG2_BYTE_TYPE

#define BIG2_BYTE_TYPE(enc, p)

Definition xmltok.c:874

Definition at line 255 of file xmltok.c.

◆ CHAR_MATCHES [1/3]

#define CHAR_MATCHES	(	enc,
		p,
		c )

Value:

(*(p) == (c))

Definition at line 300 of file xmltok.c.

◆ CHAR_MATCHES [2/3]

#define CHAR_MATCHES	(	enc,
		p,
		c )

Value:

LITTLE2_CHAR_MATCHES(p, c)

LITTLE2_CHAR_MATCHES

#define LITTLE2_CHAR_MATCHES(p, c)

Definition xmltok.c:744

Definition at line 300 of file xmltok.c.

◆ CHAR_MATCHES [3/3]

#define CHAR_MATCHES	(	enc,
		p,
		c )

Value:

BIG2_CHAR_MATCHES(p, c)

BIG2_CHAR_MATCHES

#define BIG2_CHAR_MATCHES(p, c)

Definition xmltok.c:879

Definition at line 300 of file xmltok.c.

◆ DEFINE_UTF16_TO_UTF16

#define DEFINE_UTF16_TO_UTF16 ( E )

Value:

  static enum XML_Convert_Result PTRCALL E##toUtf16(                           \
      const ENCODING *enc, const char **fromP, const char *fromLim,            \
      unsigned short **toP, const unsigned short *toLim) {                     \
    enum XML_Convert_Result res = XML_CONVERT_COMPLETED;                       \
    UNUSED_P(enc);                                                             \
    fromLim = *fromP + (((fromLim - *fromP) >> 1) << 1); /* shrink to even */  \
    /* Avoid copying first half only of surrogate */                           \
    if (fromLim - *fromP > ((toLim - *toP) << 1)                               \
        && (GET_HI(fromLim - 2) & 0xF8) == 0xD8) {                             \
      fromLim -= 2;                                                            \
      res = XML_CONVERT_INPUT_INCOMPLETE;                                      \
    }                                                                          \
    for (; *fromP < fromLim && *toP < toLim; *fromP += 2)                      \
      *(*toP)++ = (GET_HI(*fromP) << 8) | GET_LO(*fromP);                      \
    if ((*toP == toLim) && (*fromP < fromLim))                                 \
      return XML_CONVERT_OUTPUT_EXHAUSTED;                                     \
    else                                                                       \
      return res;                                                              \
  }

Definition at line 697 of file xmltok.c.

#define DEFINE_UTF16_TO_UTF16(E)                                               \
  static enum XML_Convert_Result PTRCALL E##toUtf16(                           \
      const ENCODING *enc, const char **fromP, const char *fromLim,            \
      unsigned short **toP, const unsigned short *toLim) {                     \
    enum XML_Convert_Result res = XML_CONVERT_COMPLETED;                       \
    UNUSED_P(enc);                                                             \
    fromLim = *fromP + (((fromLim - *fromP) >> 1) << 1); /* shrink to even */  \
    /* Avoid copying first half only of surrogate */                           \
    if (fromLim - *fromP > ((toLim - *toP) << 1)                               \
        && (GET_HI(fromLim - 2) & 0xF8) == 0xD8) {                             \
      fromLim -= 2;                                                            \
      res = XML_CONVERT_INPUT_INCOMPLETE;                                      \
    }                                                                          \
    for (; *fromP < fromLim && *toP < toLim; *fromP += 2)                      \
      *(*toP)++ = (GET_HI(*fromP) << 8) | GET_LO(*fromP);                      \
    if ((*toP == toLim) && (*fromP < fromLim))                                 \
      return XML_CONVERT_OUTPUT_EXHAUSTED;                                     \
    else                                                                       \
      return res;                                                              \
  }

◆ DEFINE_UTF16_TO_UTF8

#define DEFINE_UTF16_TO_UTF8 ( E )

Definition at line 620 of file xmltok.c.

#define DEFINE_UTF16_TO_UTF8(E)                                                \
  static enum XML_Convert_Result PTRCALL E##toUtf8(                            \
      const ENCODING *enc, const char **fromP, const char *fromLim,            \
      char **toP, const char *toLim) {                                         \
    const char *from = *fromP;                                                 \
    UNUSED_P(enc);                                                             \
    fromLim = from + (((fromLim - from) >> 1) << 1); /* shrink to even */      \
    for (; from < fromLim; from += 2) {                                        \
      int plane;                                                               \
      unsigned char lo2;                                                       \
      unsigned char lo = GET_LO(from);                                         \
      unsigned char hi = GET_HI(from);                                         \
      switch (hi) {                                                            \
      case 0:                                                                  \
        if (lo < 0x80) {                                                       \
          if (*toP == toLim) {                                                 \
            *fromP = from;                                                     \
            return XML_CONVERT_OUTPUT_EXHAUSTED;                               \
          }                                                                    \
          *(*toP)++ = lo;                                                      \
          break;                                                               \
        }                                                                      \
        /* fall through */                                                     \
      case 0x1:                                                                \
      case 0x2:                                                                \
      case 0x3:                                                                \
      case 0x4:                                                                \
      case 0x5:                                                                \
      case 0x6:                                                                \
      case 0x7:                                                                \
        if (toLim - *toP < 2) {                                                \
          *fromP = from;                                                       \
          return XML_CONVERT_OUTPUT_EXHAUSTED;                                 \
        }                                                                      \
        *(*toP)++ = ((lo >> 6) | (hi << 2) | UTF8_cval2);                      \
        *(*toP)++ = ((lo & 0x3f) | 0x80);                                      \
        break;                                                                 \
      default:                                                                 \
        if (toLim - *toP < 3) {                                                \
          *fromP = from;                                                       \
          return XML_CONVERT_OUTPUT_EXHAUSTED;                                 \
        }                                                                      \
        /* 16 bits divided 4, 6, 6 amongst 3 bytes */                          \
        *(*toP)++ = ((hi >> 4) | UTF8_cval3);                                  \
        *(*toP)++ = (((hi & 0xf) << 2) | (lo >> 6) | 0x80);                    \
        *(*toP)++ = ((lo & 0x3f) | 0x80);                                      \
        break;                                                                 \
      case 0xD8:                                                               \
      case 0xD9:                                                               \
      case 0xDA:                                                               \
      case 0xDB:                                                               \
        if (toLim - *toP < 4) {                                                \
          *fromP = from;                                                       \
          return XML_CONVERT_OUTPUT_EXHAUSTED;                                 \
        }                                                                      \
        if (fromLim - from < 4) {                                              \
          *fromP = from;                                                       \
          return XML_CONVERT_INPUT_INCOMPLETE;                                 \
        }                                                                      \
        plane = (((hi & 0x3) << 2) | ((lo >> 6) & 0x3)) + 1;                   \
        *(*toP)++ = (char)((plane >> 2) | UTF8_cval4);                         \
        *(*toP)++ = (((lo >> 2) & 0xF) | ((plane & 0x3) << 4) | 0x80);         \
        from += 2;                                                             \
        lo2 = GET_LO(from);                                                    \
        *(*toP)++ = (((lo & 0x3) << 4) | ((GET_HI(from) & 0x3) << 2)           \
                     | (lo2 >> 6) | 0x80);                                     \
        *(*toP)++ = ((lo2 & 0x3f) | 0x80);                                     \
        break;                                                                 \
      }                                                                        \
    }                                                                          \
    *fromP = from;                                                             \
    if (from < fromLim)                                                        \
      return XML_CONVERT_INPUT_INCOMPLETE;                                     \
    else                                                                       \
      return XML_CONVERT_COMPLETED;                                            \
  }

◆ GET_HI [1/2]

#define GET_HI ( ptr )

Value:

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

Definition at line 720 of file xmltok.c.

◆ GET_HI [2/2]

#define GET_HI ( ptr )

Value:

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

Definition at line 720 of file xmltok.c.

◆ GET_LO [1/2]

#define GET_LO ( ptr )

Value:

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

Definition at line 719 of file xmltok.c.

◆ GET_LO [2/2]

#define GET_LO ( ptr )

Value:

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

Definition at line 719 of file xmltok.c.

◆ IGNORE_SECTION_TOK_VTABLE

#define IGNORE_SECTION_TOK_VTABLE , PREFIX(ignoreSectionTok)

Definition at line 63 of file xmltok.c.

◆ INIT_ENC_INDEX

#define INIT_ENC_INDEX ( enc )

Value:

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

Definition at line 1526 of file xmltok.c.

◆ IS_INVALID_CHAR

#define IS_INVALID_CHAR	(	enc,
		p,
		n )

Value:

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

AS_NORMAL_ENCODING

#define AS_NORMAL_ENCODING(enc)

Definition xmltok.c:206

Definition at line 276 of file xmltok.c.

276# define IS_INVALID_CHAR(enc, p, n) \

277 (AS_NORMAL_ENCODING(enc)->isInvalid##n(enc, p))

◆ IS_NAME_CHAR [1/3]

#define IS_NAME_CHAR	(	enc,
		p,
		n )

Value:

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

Definition at line 269 of file xmltok.c.

◆ IS_NAME_CHAR [2/3]

#define IS_NAME_CHAR	(	enc,
		p,
		n )

Value:

0

Definition at line 269 of file xmltok.c.

◆ IS_NAME_CHAR [3/3]

#define IS_NAME_CHAR	(	enc,
		p,
		n )

Value:

0

Definition at line 269 of file xmltok.c.

◆ IS_NAME_CHAR_MINBPC [1/3]

#define IS_NAME_CHAR_MINBPC	(		enc,
			p )

Value:

(0)

Definition at line 286 of file xmltok.c.

◆ IS_NAME_CHAR_MINBPC [2/3]

#define IS_NAME_CHAR_MINBPC	(		enc,
			p )

Value:

LITTLE2_IS_NAME_CHAR_MINBPC(p)

LITTLE2_IS_NAME_CHAR_MINBPC

#define LITTLE2_IS_NAME_CHAR_MINBPC(p)

Definition xmltok.c:745

Definition at line 286 of file xmltok.c.

◆ IS_NAME_CHAR_MINBPC [3/3]

#define IS_NAME_CHAR_MINBPC	(		enc,
			p )

Value:

BIG2_IS_NAME_CHAR_MINBPC(p)

BIG2_IS_NAME_CHAR_MINBPC

#define BIG2_IS_NAME_CHAR_MINBPC(p)

Definition xmltok.c:880

Definition at line 286 of file xmltok.c.

◆ IS_NMSTRT_CHAR [1/3]

#define IS_NMSTRT_CHAR	(	enc,
		p,
		n )

Value:

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

Definition at line 270 of file xmltok.c.

◆ IS_NMSTRT_CHAR [2/3]

#define IS_NMSTRT_CHAR	(	enc,
		p,
		n )

Value:

(0)

Definition at line 270 of file xmltok.c.

◆ IS_NMSTRT_CHAR [3/3]

#define IS_NMSTRT_CHAR	(	enc,
		p,
		n )

Value:

(0)

Definition at line 270 of file xmltok.c.

◆ IS_NMSTRT_CHAR_MINBPC [1/3]

#define IS_NMSTRT_CHAR_MINBPC	(		enc,
			p )

Value:

(0)

Definition at line 287 of file xmltok.c.

◆ IS_NMSTRT_CHAR_MINBPC [2/3]

#define IS_NMSTRT_CHAR_MINBPC	(		enc,
			p )

Value:

LITTLE2_IS_NMSTRT_CHAR_MINBPC(p)

LITTLE2_IS_NMSTRT_CHAR_MINBPC

#define LITTLE2_IS_NMSTRT_CHAR_MINBPC(p)

Definition xmltok.c:747

Definition at line 287 of file xmltok.c.

◆ IS_NMSTRT_CHAR_MINBPC [3/3]

#define IS_NMSTRT_CHAR_MINBPC	(		enc,
			p )

Value:

BIG2_IS_NMSTRT_CHAR_MINBPC(p)

BIG2_IS_NMSTRT_CHAR_MINBPC

#define BIG2_IS_NMSTRT_CHAR_MINBPC(p)

Definition xmltok.c:882

Definition at line 287 of file xmltok.c.

◆ LITTLE2_BYTE_TO_ASCII

#define LITTLE2_BYTE_TO_ASCII ( p )

Value:

((p)[1] == 0 ? (p)[0] : -1)

Definition at line 743 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 740 of file xmltok.c.

#define LITTLE2_BYTE_TYPE(enc, p)                                              \
  ((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	(		p,
			c )

Value:

((p)[1] == 0 && (p)[0] == (c))

Definition at line 744 of file xmltok.c.

◆ LITTLE2_IS_NAME_CHAR_MINBPC

#define LITTLE2_IS_NAME_CHAR_MINBPC ( p )

Value:

UCS2_GET_NAMING(namePages, (unsigned char)p[1], (unsigned char)p[0])

Definition at line 745 of file xmltok.c.

745#define LITTLE2_IS_NAME_CHAR_MINBPC(p) \

746 UCS2_GET_NAMING(namePages, (unsigned char)p[1], (unsigned char)p[0])

◆ LITTLE2_IS_NMSTRT_CHAR_MINBPC

#define LITTLE2_IS_NMSTRT_CHAR_MINBPC ( p )

Value:

UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[1], (unsigned char)p[0])

Definition at line 747 of file xmltok.c.

747#define LITTLE2_IS_NMSTRT_CHAR_MINBPC(p) \

748 UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[1], (unsigned char)p[0])

◆ MINBPC [1/3]

#define MINBPC ( enc )

Value:

1

Definition at line 242 of file xmltok.c.

◆ MINBPC [2/3]

#define MINBPC ( enc )

Value:

2

Definition at line 242 of file xmltok.c.

◆ MINBPC [3/3]

#define MINBPC ( enc )

Value:

2

Definition at line 242 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 219 of file xmltok.c.

#define NORMAL_VTABLE(E)                                                       \
  E##isName2, E##isName3, E##isName4, E##isNmstrt2, E##isNmstrt3,              \
      E##isNmstrt4, E##isInvalid2, E##isInvalid3, E##isInvalid4

◆ NS [1/2]

#define NS ( x )

Value:

x

Definition at line 1648 of file xmltok.c.

◆ NS [2/2]

#define NS ( x )

Value:

x##NS

NS

#define NS(x)

Definition xmltok.c:1648

Definition at line 1648 of file xmltok.c.

◆ ns [1/2]

#define ns ( x )

Value:

x

Definition at line 1649 of file xmltok.c.

Referenced by G4ErrorSurfaceTrajState::BuildErrorMatrix(), G4FastStep::CheckIt(), G4NuclideTable::GenerateNuclide(), and G4IRTUtils::GetRCutOff().

◆ ns [2/2]

#define ns ( x )

Value:

x##_ns

Definition at line 1649 of file xmltok.c.

◆ NULL_VTABLE

#define NULL_VTABLE

Value:

  /* isName2 */ NULL, /* isName3 */ NULL, /* isName4 */ NULL,                  \
      /* isNmstrt2 */ NULL, /* isNmstrt3 */ NULL, /* isNmstrt4 */ NULL,        \
      /* isInvalid2 */ NULL, /* isInvalid3 */ NULL, /* isInvalid4 */ NULL

Definition at line 223 of file xmltok.c.

#define NULL_VTABLE                                                            \
  /* isName2 */ NULL, /* isName3 */ NULL, /* isName4 */ NULL,                  \
      /* isNmstrt2 */ NULL, /* isNmstrt3 */ NULL, /* isNmstrt4 */ NULL,        \
      /* isInvalid2 */ NULL, /* isInvalid3 */ NULL, /* isInvalid4 */ NULL

◆ PREFIX [1/3]

#define PREFIX ( ident )

Value:

normal_##ident

Definition at line 303 of file xmltok.c.

◆ PREFIX [2/3]

#define PREFIX ( ident )

Value:

little2_##ident

Definition at line 303 of file xmltok.c.

◆ PREFIX [3/3]

#define PREFIX ( ident )

Value:

big2_##ident

Definition at line 303 of file xmltok.c.

◆ SB_BYTE_TYPE

#define SB_BYTE_TYPE	(		enc,
			p )

Value:

(((struct normal_encoding *)(enc))->type[(unsigned char)*(p)])

Definition at line 245 of file xmltok.c.

245#define SB_BYTE_TYPE(enc, p) \

246 (((struct normal_encoding *)(enc))->type[(unsigned char)*(p)])

◆ SET2 [1/2]

#define SET2	(		ptr,
			ch )

Value:

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

Definition at line 718 of file xmltok.c.

◆ SET2 [2/2]

#define SET2	(		ptr,
			ch )

Value:

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

Definition at line 718 of file xmltok.c.

◆ SET_INIT_ENC_INDEX

#define SET_INIT_ENC_INDEX	(		enc,
			i )

Value:

((enc)->initEnc.isUtf16 = (char)i)

Definition at line 1527 of file xmltok.c.

◆ STANDARD_VTABLE

#define STANDARD_VTABLE ( E )

Value:

/* as nothing */

Definition at line 215 of file xmltok.c.

◆ UCS2_GET_NAMING

#define UCS2_GET_NAMING	(	pages,
		hi,
		lo )

Value:

(namingBitmap[(pages[hi] << 3) + ((lo) >> 5)] & (1u << ((lo)&0x1F)))

Definition at line 78 of file xmltok.c.

78#define UCS2_GET_NAMING(pages, hi, lo) \

79 (namingBitmap[(pages[hi] << 3) + ((lo) >> 5)] & (1u << ((lo)&0x1F)))

Referenced by XmlInitUnknownEncoding().

◆ 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)]         \
   & (1u << (((byte)[1]) & 0x1F)))

Definition at line 85 of file xmltok.c.

#define UTF8_GET_NAMING2(pages, byte)                                          \
  (namingBitmap[((pages)[(((byte)[0]) >> 2) & 7] << 3)                         \
                + ((((byte)[0]) & 3) << 1) + ((((byte)[1]) >> 5) & 1)]         \
   & (1u << (((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)]                 \
   & (1u << (((byte)[2]) & 0x1F)))

Definition at line 95 of file xmltok.c.

#define UTF8_GET_NAMING3(pages, byte)                                          \
  (namingBitmap                                                                \
       [((pages)[((((byte)[0]) & 0xF) << 4) + ((((byte)[1]) >> 2) & 0xF)]      \
         << 3)                                                                 \
        + ((((byte)[1]) & 3) << 1) + ((((byte)[2]) >> 5) & 1)]                 \
   & (1u << (((byte)[2]) & 0x1F)))

◆ UTF8_INVALID2

#define UTF8_INVALID2 ( p )

Value:

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

Definition at line 112 of file xmltok.c.

112#define UTF8_INVALID2(p) \

113 ((*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)))

Definition at line 115 of file xmltok.c.

#define UTF8_INVALID3(p)                                                       \
  (((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)))

Definition at line 124 of file xmltok.c.

#define UTF8_INVALID4(p)                                                       \
  (((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

Definition at line 151 of file xmltok.c.

◆ utf8_isNmstrt4

#define utf8_isNmstrt4 isNever

Definition at line 165 of file xmltok.c.

◆ VTABLE

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

Definition at line 76 of file xmltok.c.

◆ VTABLE1

#define VTABLE1

Value:

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

Definition at line 68 of file xmltok.c.

#define VTABLE1                                                                \
  {PREFIX(prologTok), PREFIX(contentTok),                                      \
   PREFIX(cdataSectionTok) IGNORE_SECTION_TOK_VTABLE},                         \
      {PREFIX(attributeValueTok), PREFIX(entityValueTok)},                     \
      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

Definition at line 304 of file xmltok.c.

◆ XML_TOK_IMPL_C [2/3]

#define XML_TOK_IMPL_C

Definition at line 304 of file xmltok.c.

◆ XML_TOK_IMPL_C [3/3]

#define XML_TOK_IMPL_C

Definition at line 304 of file xmltok.c.

◆ XML_TOK_NS_C [1/2]

#define XML_TOK_NS_C

Definition at line 1650 of file xmltok.c.

◆ XML_TOK_NS_C [2/2]

#define XML_TOK_NS_C

Definition at line 1650 of file xmltok.c.

Enumeration Type Documentation

◆ anonymous enum

anonymous enum

Enumerator
UTF8_cval1
UTF8_cval2
UTF8_cval3
UTF8_cval4

Definition at line 318 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 1479 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

◆ _INTERNAL_trim_to_complete_utf8_characters()

void _INTERNAL_trim_to_complete_utf8_characters	(	const char *	from,
		const char **	fromLimRef )

Definition at line 326 of file xmltok.c.

                                                                    {
  const char *fromLim = *fromLimRef;
  size_t walked = 0;
  for (; fromLim > from; fromLim--, walked++) {
    const unsigned char prev = (unsigned char)fromLim[-1];
    if ((prev & 0xf8u)
        == 0xf0u) { /* 4-byte character, lead by 0b11110xxx byte */
      if (walked + 1 >= 4) {
        fromLim += 4 - 1;
        break;
      } else {
        walked = 0;
      }
    } else if ((prev & 0xf0u)
               == 0xe0u) { /* 3-byte character, lead by 0b1110xxxx byte */
      if (walked + 1 >= 3) {
        fromLim += 3 - 1;
        break;
      } else {
        walked = 0;
      }
    } else if ((prev & 0xe0u)
               == 0xc0u) { /* 2-byte character, lead by 0b110xxxxx byte */
      if (walked + 1 >= 2) {
        fromLim += 2 - 1;
        break;
      } else {
        walked = 0;
      }
    } else if ((prev & 0x80u)
               == 0x00u) { /* 1-byte character, matching 0b0xxxxxxx */
      break;
    }
  }
  *fromLimRef = fromLim;
}

◆ XmlInitUnknownEncoding()

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

Definition at line 1406 of file xmltok.c.

                                       {
  int i;
  struct unknown_encoding *e = (struct unknown_encoding *)mem;
  memcpy(mem, &latin1_encoding, sizeof(struct normal_encoding));
  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;
      /* Multi-byte sequences need a converter function */
      if (! convert)
        return 0;
      e->normal.type[i] = (unsigned char)(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] = (unsigned short)(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] = (unsigned short)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);
}

Referenced by XmlInitUnknownEncodingNS().

◆ XmlInitUnknownEncodingNS()

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

Definition at line 1669 of file xmltok.c.

                                         {
  ENCODING *enc = XmlInitUnknownEncoding(mem, table, convert, userData);
  if (enc)
    ((struct normal_encoding *)enc)->type[ASCII_COLON] = BT_COLON;
  return enc;
}

◆ XmlSizeOfUnknownEncoding()

int XmlSizeOfUnknownEncoding ( void )

Definition at line 1325 of file xmltok.c.

                               {
  return sizeof(struct unknown_encoding);
}

◆ XmlUtf16Encode()

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

Definition at line 1298 of file xmltok.c.

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

◆ XmlUtf8Encode()

int FASTCALL XmlUtf8Encode	(	int	c,
		char *	buf )

Definition at line 1262 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; /* LCOV_EXCL_LINE: this case is always eliminated beforehand */
  if (c < min2) {
    buf[0] = (char)(c | UTF8_cval1);
    return 1;
  }
  if (c < min3) {
    buf[0] = (char)((c >> 6) | UTF8_cval2);
    buf[1] = (char)((c & 0x3f) | 0x80);
    return 2;
  }
  if (c < min4) {
    buf[0] = (char)((c >> 12) | UTF8_cval3);
    buf[1] = (char)(((c >> 6) & 0x3f) | 0x80);
    buf[2] = (char)((c & 0x3f) | 0x80);
    return 3;
  }
  if (c < 0x110000) {
    buf[0] = (char)((c >> 18) | UTF8_cval4);
    buf[1] = (char)(((c >> 12) & 0x3f) | 0x80);
    buf[2] = (char)(((c >> 6) & 0x3f) | 0x80);
    buf[3] = (char)((c & 0x3f) | 0x80);
    return 4;
  }
  return 0; /* LCOV_EXCL_LINE: this case too is eliminated before calling */
}

Referenced by XmlInitUnknownEncoding().

Classes

Macros

Enumerations

Functions

Macro Definition Documentation

◆ AS_NORMAL_ENCODING

◆ AS_UNKNOWN_ENCODING

◆ BIG2_BYTE_TO_ASCII

◆ BIG2_BYTE_TYPE

◆ BIG2_CHAR_MATCHES

◆ BIG2_IS_NAME_CHAR_MINBPC

◆ BIG2_IS_NMSTRT_CHAR_MINBPC

◆ BT_COLON [1/7]

◆ BT_COLON [2/7]

◆ BT_COLON [3/7]

◆ BT_COLON [4/7]

◆ BT_COLON [5/7]

◆ BT_COLON [6/7]

◆ BT_COLON [7/7]

◆ 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]

◆ IGNORE_SECTION_TOK_VTABLE

◆ 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 [1/2]

◆ NS [2/2]

◆ ns [1/2]

◆ ns [2/2]

◆ NULL_VTABLE

◆ PREFIX [1/3]

◆ PREFIX [2/3]

◆ PREFIX [3/3]

◆ SB_BYTE_TYPE

◆ SET2 [1/2]

◆ SET2 [2/2]

◆ SET_INIT_ENC_INDEX

◆ STANDARD_VTABLE

◆ UCS2_GET_NAMING

◆ UTF8_GET_NAMING2

◆ UTF8_GET_NAMING3

◆ UTF8_INVALID2

◆ UTF8_INVALID3

◆ UTF8_INVALID4

◆ utf8_isName4

◆ utf8_isNmstrt4

◆ VTABLE