|
version 1.1.1.2, 2012/10/09 09:19:17
|
version 1.1.1.3, 2013/07/22 08:25:56
|
|
Line 78 having a separate .h file just for this. */
|
Line 78 having a separate .h file just for this. */
|
| #ifdef PCRE_INCLUDED |
#ifdef PCRE_INCLUDED |
| static /* Keep the following function as private. */ |
static /* Keep the following function as private. */ |
| #endif |
#endif |
| #ifdef COMPILE_PCRE8 | |
| | #if defined COMPILE_PCRE8 |
| void pcre_printint(pcre *external_re, FILE *f, BOOL print_lengths); |
void pcre_printint(pcre *external_re, FILE *f, BOOL print_lengths); |
| #else | #elif defined COMPILE_PCRE16 |
| void pcre16_printint(pcre *external_re, FILE *f, BOOL print_lengths); |
void pcre16_printint(pcre *external_re, FILE *f, BOOL print_lengths); |
| |
#elif defined COMPILE_PCRE32 |
| |
void pcre32_printint(pcre *external_re, FILE *f, BOOL print_lengths); |
| #endif |
#endif |
| |
|
| /* Macro that decides whether a character should be output as a literal or in |
/* Macro that decides whether a character should be output as a literal or in |
|
Line 111 static const pcre_uint8 priv_OP_lengths[] = { OP_LENGT
|
Line 114 static const pcre_uint8 priv_OP_lengths[] = { OP_LENGT
|
| * Print single- or multi-byte character * |
* Print single- or multi-byte character * |
| *************************************************/ |
*************************************************/ |
| |
|
| static int | static unsigned int |
| print_char(FILE *f, pcre_uchar *ptr, BOOL utf) |
print_char(FILE *f, pcre_uchar *ptr, BOOL utf) |
| { |
{ |
| int c = *ptr; | pcre_uint32 c = *ptr; |
| |
|
| #ifndef SUPPORT_UTF |
#ifndef SUPPORT_UTF |
| |
|
| (void)utf; /* Avoid compiler warning */ |
(void)utf; /* Avoid compiler warning */ |
| if (PRINTABLE(c)) fprintf(f, "%c", c); | if (PRINTABLE(c)) fprintf(f, "%c", (char)c); |
| else if (c <= 0xff) fprintf(f, "\\x%02x", c); | else if (c <= 0x80) fprintf(f, "\\x%02x", c); |
| else fprintf(f, "\\x{%x}", c); |
else fprintf(f, "\\x{%x}", c); |
| return 0; |
return 0; |
| |
|
| #else |
#else |
| |
|
| #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
| |
|
| if (!utf || (c & 0xc0) != 0xc0) |
if (!utf || (c & 0xc0) != 0xc0) |
| { |
{ |
| if (PRINTABLE(c)) fprintf(f, "%c", c); else fprintf(f, "\\x%02x", c); | if (PRINTABLE(c)) fprintf(f, "%c", (char)c); |
| | else if (c < 0x80) fprintf(f, "\\x%02x", c); |
| | else fprintf(f, "\\x{%02x}", c); |
| return 0; |
return 0; |
| } |
} |
| else |
else |
|
Line 160 else
|
Line 165 else
|
| return a; |
return a; |
| } |
} |
| |
|
| #else | #elif defined COMPILE_PCRE16 |
| |
|
| #ifdef COMPILE_PCRE16 |
|
| |
|
| if (!utf || (c & 0xfc00) != 0xd800) |
if (!utf || (c & 0xfc00) != 0xd800) |
| { |
{ |
| if (PRINTABLE(c)) fprintf(f, "%c", c); | if (PRINTABLE(c)) fprintf(f, "%c", (char)c); |
| else if (c <= 0xff) fprintf(f, "\\x%02x", c); | else if (c <= 0x80) fprintf(f, "\\x%02x", c); |
| else fprintf(f, "\\x{%x}", c); | else fprintf(f, "\\x{%02x}", c); |
| return 0; |
return 0; |
| } |
} |
| else |
else |
|
Line 188 else
|
Line 191 else
|
| return 1; |
return 1; |
| } |
} |
| |
|
| #endif /* COMPILE_PCRE16 */ | #elif defined COMPILE_PCRE32 |
| |
|
| #endif /* COMPILE_PCRE8 */ | if (!utf || (c & 0xfffff800u) != 0xd800u) |
| | { |
| | if (PRINTABLE(c)) fprintf(f, "%c", (char)c); |
| | else if (c <= 0x80) fprintf(f, "\\x%02x", c); |
| | else fprintf(f, "\\x{%x}", c); |
| | return 0; |
| | } |
| | else |
| | { |
| | /* This is a check for malformed UTF-32; it should only occur if the sanity |
| | check has been turned off. Rather than swallow a surrogate, just stop if |
| | we hit one. Print it with \X instead of \x as an indication. */ |
| | fprintf(f, "\\X{%x}", c); |
| | return 0; |
| | } |
| |
|
| |
#endif /* COMPILE_PCRE[8|16|32] */ |
| |
|
| #endif /* SUPPORT_UTF */ |
#endif /* SUPPORT_UTF */ |
| } |
} |
| |
|
|
Line 204 print_puchar(FILE *f, PCRE_PUCHAR ptr)
|
Line 223 print_puchar(FILE *f, PCRE_PUCHAR ptr)
|
| { |
{ |
| while (*ptr != '\0') |
while (*ptr != '\0') |
| { |
{ |
| register int c = *ptr++; | register pcre_uint32 c = *ptr++; |
| if (PRINTABLE(c)) fprintf(f, "%c", c); else fprintf(f, "\\x{%x}", c); |
if (PRINTABLE(c)) fprintf(f, "%c", c); else fprintf(f, "\\x{%x}", c); |
| } |
} |
| } |
} |
|
Line 214 while (*ptr != '\0')
|
Line 233 while (*ptr != '\0')
|
| *************************************************/ |
*************************************************/ |
| |
|
| static const char * |
static const char * |
| get_ucpname(int ptype, int pvalue) | get_ucpname(unsigned int ptype, unsigned int pvalue) |
| { |
{ |
| #ifdef SUPPORT_UCP |
#ifdef SUPPORT_UCP |
| int i; |
int i; |
|
Line 231 return (ptype == pvalue)? "??" : "??";
|
Line 250 return (ptype == pvalue)? "??" : "??";
|
| } |
} |
| |
|
| |
|
| |
/************************************************* |
| |
* Print Unicode property value * |
| |
*************************************************/ |
| |
|
| |
/* "Normal" properties can be printed from tables. The PT_CLIST property is a |
| |
pseudo-property that contains a pointer to a list of case-equivalent |
| |
characters. This is used only when UCP support is available and UTF mode is |
| |
selected. It should never occur otherwise, but just in case it does, have |
| |
something ready to print. */ |
| |
|
| |
static void |
| |
print_prop(FILE *f, pcre_uchar *code, const char *before, const char *after) |
| |
{ |
| |
if (code[1] != PT_CLIST) |
| |
{ |
| |
fprintf(f, "%s%s %s%s", before, priv_OP_names[*code], get_ucpname(code[1], |
| |
code[2]), after); |
| |
} |
| |
else |
| |
{ |
| |
const char *not = (*code == OP_PROP)? "" : "not "; |
| |
#ifndef SUPPORT_UCP |
| |
fprintf(f, "%s%sclist %d%s", before, not, code[2], after); |
| |
#else |
| |
const pcre_uint32 *p = PRIV(ucd_caseless_sets) + code[2]; |
| |
fprintf (f, "%s%sclist", before, not); |
| |
while (*p < NOTACHAR) fprintf(f, " %04x", *p++); |
| |
fprintf(f, "%s", after); |
| |
#endif |
| |
} |
| |
} |
| |
|
| |
|
| |
|
| |
|
| /************************************************* |
/************************************************* |
| * Print compiled regex * |
* Print compiled regex * |
| *************************************************/ |
*************************************************/ |
|
Line 245 written that do not depend on the value of LINK_SIZE.
|
Line 298 written that do not depend on the value of LINK_SIZE.
|
| #ifdef PCRE_INCLUDED |
#ifdef PCRE_INCLUDED |
| static /* Keep the following function as private. */ |
static /* Keep the following function as private. */ |
| #endif |
#endif |
| #ifdef COMPILE_PCRE8 | #if defined COMPILE_PCRE8 |
| void |
void |
| pcre_printint(pcre *external_re, FILE *f, BOOL print_lengths) |
pcre_printint(pcre *external_re, FILE *f, BOOL print_lengths) |
| #else | #elif defined COMPILE_PCRE16 |
| void |
void |
| pcre16_printint(pcre *external_re, FILE *f, BOOL print_lengths) |
pcre16_printint(pcre *external_re, FILE *f, BOOL print_lengths) |
| |
#elif defined COMPILE_PCRE32 |
| |
void |
| |
pcre32_printint(pcre *external_re, FILE *f, BOOL print_lengths) |
| #endif |
#endif |
| { |
{ |
| REAL_PCRE *re = (REAL_PCRE *)external_re; |
REAL_PCRE *re = (REAL_PCRE *)external_re; |
|
Line 274 if (re->magic_number != MAGIC_NUMBER)
|
Line 330 if (re->magic_number != MAGIC_NUMBER)
|
| } |
} |
| |
|
| code = codestart = (pcre_uchar *)re + offset + count * size; |
code = codestart = (pcre_uchar *)re + offset + count * size; |
| /* PCRE_UTF16 has the same value as PCRE_UTF8. */ | /* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */ |
| utf = (options & PCRE_UTF8) != 0; |
utf = (options & PCRE_UTF8) != 0; |
| |
|
| for(;;) |
for(;;) |
| { |
{ |
| pcre_uchar *ccode; |
pcre_uchar *ccode; |
| const char *flag = " "; |
const char *flag = " "; |
| int c; | pcre_uint32 c; |
| int extra = 0; | unsigned int extra = 0; |
| |
|
| if (print_lengths) |
if (print_lengths) |
| fprintf(f, "%3d ", (int)(code - codestart)); |
fprintf(f, "%3d ", (int)(code - codestart)); |
|
Line 425 for(;;)
|
Line 481 for(;;)
|
| fprintf(f, " %s ", flag); |
fprintf(f, " %s ", flag); |
| if (*code >= OP_TYPESTAR) |
if (*code >= OP_TYPESTAR) |
| { |
{ |
| fprintf(f, "%s", priv_OP_names[code[1]]); |
|
| if (code[1] == OP_PROP || code[1] == OP_NOTPROP) |
if (code[1] == OP_PROP || code[1] == OP_NOTPROP) |
| { |
{ |
| fprintf(f, " %s ", get_ucpname(code[2], code[3])); | print_prop(f, code + 1, "", " "); |
| extra = 2; |
extra = 2; |
| } |
} |
| |
else fprintf(f, "%s", priv_OP_names[code[1]]); |
| } |
} |
| else extra = print_char(f, code+1, utf); |
else extra = print_char(f, code+1, utf); |
| fprintf(f, "%s", priv_OP_names[*code]); |
fprintf(f, "%s", priv_OP_names[*code]); |
|
Line 459 for(;;)
|
Line 515 for(;;)
|
| case OP_TYPEUPTO: |
case OP_TYPEUPTO: |
| case OP_TYPEMINUPTO: |
case OP_TYPEMINUPTO: |
| case OP_TYPEPOSUPTO: |
case OP_TYPEPOSUPTO: |
| fprintf(f, " %s", priv_OP_names[code[1 + IMM2_SIZE]]); |
|
| if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP) |
if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP) |
| { |
{ |
| fprintf(f, " %s ", get_ucpname(code[1 + IMM2_SIZE + 1], | print_prop(f, code + IMM2_SIZE + 1, " ", " "); |
| code[1 + IMM2_SIZE + 2])); | |
| extra = 2; |
extra = 2; |
| } |
} |
| |
else fprintf(f, " %s", priv_OP_names[code[1 + IMM2_SIZE]]); |
| fprintf(f, "{"); |
fprintf(f, "{"); |
| if (*code != OP_TYPEEXACT) fprintf(f, "0,"); |
if (*code != OP_TYPEEXACT) fprintf(f, "0,"); |
| fprintf(f, "%d}", GET2(code,1)); |
fprintf(f, "%d}", GET2(code,1)); |
|
Line 550 for(;;)
|
Line 605 for(;;)
|
| |
|
| case OP_PROP: |
case OP_PROP: |
| case OP_NOTPROP: |
case OP_NOTPROP: |
| fprintf(f, " %s %s", priv_OP_names[*code], get_ucpname(code[1], code[2])); | print_prop(f, code, " ", ""); |
| break; |
break; |
| |
|
| /* OP_XCLASS can only occur in UTF or PCRE16 modes. However, there's no |
/* OP_XCLASS can only occur in UTF or PCRE16 modes. However, there's no |
|
Line 561 for(;;)
|
Line 616 for(;;)
|
| case OP_NCLASS: |
case OP_NCLASS: |
| case OP_XCLASS: |
case OP_XCLASS: |
| { |
{ |
| int i, min, max; | int i; |
| | unsigned int min, max; |
| BOOL printmap; |
BOOL printmap; |
| pcre_uint8 *map; |
pcre_uint8 *map; |
| |
|
|
Line 612 for(;;)
|
Line 668 for(;;)
|
| |
|
| if (*code == OP_XCLASS) |
if (*code == OP_XCLASS) |
| { |
{ |
| int ch; | pcre_uchar ch; |
| while ((ch = *ccode++) != XCL_END) |
while ((ch = *ccode++) != XCL_END) |
| { |
{ |
| if (ch == XCL_PROP) |
if (ch == XCL_PROP) |
| { |
{ |
| int ptype = *ccode++; | unsigned int ptype = *ccode++; |
| int pvalue = *ccode++; | unsigned int pvalue = *ccode++; |
| fprintf(f, "\\p{%s}", get_ucpname(ptype, pvalue)); |
fprintf(f, "\\p{%s}", get_ucpname(ptype, pvalue)); |
| } |
} |
| else if (ch == XCL_NOTPROP) |
else if (ch == XCL_NOTPROP) |
| { |
{ |
| int ptype = *ccode++; | unsigned int ptype = *ccode++; |
| int pvalue = *ccode++; | unsigned int pvalue = *ccode++; |
| fprintf(f, "\\P{%s}", get_ucpname(ptype, pvalue)); |
fprintf(f, "\\P{%s}", get_ucpname(ptype, pvalue)); |
| } |
} |
| else |
else |
|
Line 662 for(;;)
|
Line 718 for(;;)
|
| case OP_CRMINRANGE: |
case OP_CRMINRANGE: |
| min = GET2(ccode,1); |
min = GET2(ccode,1); |
| max = GET2(ccode,1 + IMM2_SIZE); |
max = GET2(ccode,1 + IMM2_SIZE); |
| if (max == 0) fprintf(f, "{%d,}", min); | if (max == 0) fprintf(f, "{%u,}", min); |
| else fprintf(f, "{%d,%d}", min, max); | else fprintf(f, "{%u,%u}", min, max); |
| if (*ccode == OP_CRMINRANGE) fprintf(f, "?"); |
if (*ccode == OP_CRMINRANGE) fprintf(f, "?"); |
| extra += priv_OP_lengths[*ccode]; |
extra += priv_OP_lengths[*ccode]; |
| break; |
break; |