| 6 |
and semantics are as close as possible to those of the Perl 5 language. |
and semantics are as close as possible to those of the Perl 5 language. |
| 7 |
|
|
| 8 |
Written by Philip Hazel |
Written by Philip Hazel |
| 9 |
Copyright (c) 1997-2005 University of Cambridge |
Copyright (c) 1997-2012 University of Cambridge |
| 10 |
|
|
| 11 |
----------------------------------------------------------------------------- |
----------------------------------------------------------------------------- |
| 12 |
Redistribution and use in source and binary forms, with or without |
Redistribution and use in source and binary forms, with or without |
| 43 |
auxiliary local function to flip the appropriate bytes. */ |
auxiliary local function to flip the appropriate bytes. */ |
| 44 |
|
|
| 45 |
|
|
| 46 |
|
#ifdef HAVE_CONFIG_H |
| 47 |
|
#include "config.h" |
| 48 |
|
#endif |
| 49 |
|
|
| 50 |
#include "pcre_internal.h" |
#include "pcre_internal.h" |
| 51 |
|
|
| 52 |
|
|
| 53 |
/************************************************* |
/************************************************* |
| 54 |
* Flip bytes in an integer * |
* Swap byte functions * |
| 55 |
*************************************************/ |
*************************************************/ |
| 56 |
|
|
| 57 |
/* This function is called when the magic number in a regex doesn't match, in |
/* The following functions swap the bytes of a pcre_uint16 |
| 58 |
order to flip its bytes to see if we are dealing with a pattern that was |
and pcre_uint32 value. |
|
compiled on a host of different endianness. If so, this function is used to |
|
|
flip other byte values. |
|
| 59 |
|
|
| 60 |
Arguments: |
Arguments: |
| 61 |
value the number to flip |
value any number |
|
n the number of bytes to flip (assumed to be 2 or 4) |
|
| 62 |
|
|
| 63 |
Returns: the flipped value |
Returns: the byte swapped value |
| 64 |
*/ |
*/ |
| 65 |
|
|
| 66 |
static long int |
static pcre_uint32 |
| 67 |
byteflip(long int value, int n) |
swap_uint32(pcre_uint32 value) |
| 68 |
{ |
{ |
|
if (n == 2) return ((value & 0x00ff) << 8) | ((value & 0xff00) >> 8); |
|
| 69 |
return ((value & 0x000000ff) << 24) | |
return ((value & 0x000000ff) << 24) | |
| 70 |
((value & 0x0000ff00) << 8) | |
((value & 0x0000ff00) << 8) | |
| 71 |
((value & 0x00ff0000) >> 8) | |
((value & 0x00ff0000) >> 8) | |
| 72 |
((value & 0xff000000) >> 24); |
(value >> 24); |
| 73 |
} |
} |
| 74 |
|
|
| 75 |
|
static pcre_uint16 |
| 76 |
|
swap_uint16(pcre_uint16 value) |
| 77 |
|
{ |
| 78 |
|
return (value >> 8) | (value << 8); |
| 79 |
|
} |
| 80 |
|
|
| 81 |
|
|
| 82 |
/************************************************* |
/************************************************* |
| 83 |
* Test for a byte-flipped compiled regex * |
* Test for a byte-flipped compiled regex * |
| 84 |
*************************************************/ |
*************************************************/ |
| 85 |
|
|
| 86 |
/* This function is called from pcre_exec(), pcre_dfa_exec(), and also from |
/* This function swaps the bytes of a compiled pattern usually |
| 87 |
pcre_fullinfo(). Its job is to test whether the regex is byte-flipped - that |
loaded form the disk. It also sets the tables pointer, which |
| 88 |
is, it was compiled on a system of opposite endianness. The function is called |
is likely an invalid pointer after reload. |
|
only when the native MAGIC_NUMBER test fails. If the regex is indeed flipped, |
|
|
we flip all the relevant values into a different data block, and return it. |
|
| 89 |
|
|
| 90 |
Arguments: |
Arguments: |
| 91 |
re points to the regex |
argument_re points to the compiled expression |
| 92 |
study points to study data, or NULL |
extra_data points to extra data or is NULL |
| 93 |
internal_re points to a new regex block |
tables points to the character tables or NULL |
|
internal_study points to a new study block |
|
| 94 |
|
|
| 95 |
Returns: the new block if is is indeed a byte-flipped regex |
Returns: 0 if the swap is successful, negative on error |
|
NULL if it is not |
|
| 96 |
*/ |
*/ |
| 97 |
|
|
| 98 |
PCRE_EXPORT real_pcre * |
#ifdef COMPILE_PCRE8 |
| 99 |
_pcre_try_flipped(const real_pcre *re, real_pcre *internal_re, |
PCRE_EXP_DECL int pcre_pattern_to_host_byte_order(pcre *argument_re, |
| 100 |
const pcre_study_data *study, pcre_study_data *internal_study) |
pcre_extra *extra_data, const unsigned char *tables) |
| 101 |
|
#else |
| 102 |
|
PCRE_EXP_DECL int pcre16_pattern_to_host_byte_order(pcre *argument_re, |
| 103 |
|
pcre_extra *extra_data, const unsigned char *tables) |
| 104 |
|
#endif |
| 105 |
{ |
{ |
| 106 |
if (byteflip(re->magic_number, sizeof(re->magic_number)) != MAGIC_NUMBER) |
real_pcre *re = (real_pcre *)argument_re; |
| 107 |
return NULL; |
pcre_study_data *study; |
| 108 |
|
#ifndef COMPILE_PCRE8 |
| 109 |
|
pcre_uchar *ptr; |
| 110 |
|
int length; |
| 111 |
|
#ifdef SUPPORT_UTF |
| 112 |
|
BOOL utf; |
| 113 |
|
BOOL utf16_char; |
| 114 |
|
#endif /* SUPPORT_UTF */ |
| 115 |
|
#endif /* !COMPILE_PCRE8 */ |
| 116 |
|
|
| 117 |
|
if (re == NULL) return PCRE_ERROR_NULL; |
| 118 |
|
if (re->magic_number == MAGIC_NUMBER) |
| 119 |
|
{ |
| 120 |
|
if ((re->flags & PCRE_MODE) == 0) return PCRE_ERROR_BADMODE; |
| 121 |
|
re->tables = tables; |
| 122 |
|
return 0; |
| 123 |
|
} |
| 124 |
|
|
| 125 |
|
if (re->magic_number != REVERSED_MAGIC_NUMBER) return PCRE_ERROR_BADMAGIC; |
| 126 |
|
if ((swap_uint16(re->flags) & PCRE_MODE) == 0) return PCRE_ERROR_BADMODE; |
| 127 |
|
|
| 128 |
|
re->magic_number = MAGIC_NUMBER; |
| 129 |
|
re->size = swap_uint32(re->size); |
| 130 |
|
re->options = swap_uint32(re->options); |
| 131 |
|
re->flags = swap_uint16(re->flags); |
| 132 |
|
re->top_bracket = swap_uint16(re->top_bracket); |
| 133 |
|
re->top_backref = swap_uint16(re->top_backref); |
| 134 |
|
re->first_char = swap_uint16(re->first_char); |
| 135 |
|
re->req_char = swap_uint16(re->req_char); |
| 136 |
|
re->name_table_offset = swap_uint16(re->name_table_offset); |
| 137 |
|
re->name_entry_size = swap_uint16(re->name_entry_size); |
| 138 |
|
re->name_count = swap_uint16(re->name_count); |
| 139 |
|
re->ref_count = swap_uint16(re->ref_count); |
| 140 |
|
re->tables = tables; |
| 141 |
|
|
| 142 |
|
if (extra_data != NULL && (re->flags & PCRE_EXTRA_STUDY_DATA) != 0) |
| 143 |
|
{ |
| 144 |
|
study = (pcre_study_data *)extra_data->study_data; |
| 145 |
|
study->size = swap_uint32(study->size); |
| 146 |
|
study->flags = swap_uint32(study->flags); |
| 147 |
|
study->minlength = swap_uint32(study->minlength); |
| 148 |
|
} |
| 149 |
|
|
| 150 |
*internal_re = *re; /* To copy other fields */ |
#ifndef COMPILE_PCRE8 |
| 151 |
internal_re->size = byteflip(re->size, sizeof(re->size)); |
ptr = (pcre_uchar *)re + re->name_table_offset; |
| 152 |
internal_re->options = byteflip(re->options, sizeof(re->options)); |
length = re->name_count * re->name_entry_size; |
| 153 |
internal_re->top_bracket = |
#ifdef SUPPORT_UTF |
| 154 |
(pcre_uint16)byteflip(re->top_bracket, sizeof(re->top_bracket)); |
utf = (re->options & PCRE_UTF16) != 0; |
| 155 |
internal_re->top_backref = |
utf16_char = FALSE; |
| 156 |
(pcre_uint16)byteflip(re->top_backref, sizeof(re->top_backref)); |
#endif |
|
internal_re->first_byte = |
|
|
(pcre_uint16)byteflip(re->first_byte, sizeof(re->first_byte)); |
|
|
internal_re->req_byte = |
|
|
(pcre_uint16)byteflip(re->req_byte, sizeof(re->req_byte)); |
|
|
internal_re->name_table_offset = |
|
|
(pcre_uint16)byteflip(re->name_table_offset, sizeof(re->name_table_offset)); |
|
|
internal_re->name_entry_size = |
|
|
(pcre_uint16)byteflip(re->name_entry_size, sizeof(re->name_entry_size)); |
|
|
internal_re->name_count = |
|
|
(pcre_uint16)byteflip(re->name_count, sizeof(re->name_count)); |
|
| 157 |
|
|
| 158 |
if (study != NULL) |
while(TRUE) |
| 159 |
{ |
{ |
| 160 |
*internal_study = *study; /* To copy other fields */ |
/* Swap previous characters. */ |
| 161 |
internal_study->size = byteflip(study->size, sizeof(study->size)); |
while (length-- > 0) |
| 162 |
internal_study->options = byteflip(study->options, sizeof(study->options)); |
{ |
| 163 |
|
*ptr = swap_uint16(*ptr); |
| 164 |
|
ptr++; |
| 165 |
|
} |
| 166 |
|
#ifdef SUPPORT_UTF |
| 167 |
|
if (utf16_char) |
| 168 |
|
{ |
| 169 |
|
if (HAS_EXTRALEN(ptr[-1])) |
| 170 |
|
{ |
| 171 |
|
/* We know that there is only one extra character in UTF-16. */ |
| 172 |
|
*ptr = swap_uint16(*ptr); |
| 173 |
|
ptr++; |
| 174 |
|
} |
| 175 |
|
} |
| 176 |
|
utf16_char = FALSE; |
| 177 |
|
#endif /* SUPPORT_UTF */ |
| 178 |
|
|
| 179 |
|
/* Get next opcode. */ |
| 180 |
|
length = 0; |
| 181 |
|
*ptr = swap_uint16(*ptr); |
| 182 |
|
switch (*ptr) |
| 183 |
|
{ |
| 184 |
|
case OP_END: |
| 185 |
|
return 0; |
| 186 |
|
|
| 187 |
|
#ifdef SUPPORT_UTF |
| 188 |
|
case OP_CHAR: |
| 189 |
|
case OP_CHARI: |
| 190 |
|
case OP_NOT: |
| 191 |
|
case OP_NOTI: |
| 192 |
|
case OP_STAR: |
| 193 |
|
case OP_MINSTAR: |
| 194 |
|
case OP_PLUS: |
| 195 |
|
case OP_MINPLUS: |
| 196 |
|
case OP_QUERY: |
| 197 |
|
case OP_MINQUERY: |
| 198 |
|
case OP_UPTO: |
| 199 |
|
case OP_MINUPTO: |
| 200 |
|
case OP_EXACT: |
| 201 |
|
case OP_POSSTAR: |
| 202 |
|
case OP_POSPLUS: |
| 203 |
|
case OP_POSQUERY: |
| 204 |
|
case OP_POSUPTO: |
| 205 |
|
case OP_STARI: |
| 206 |
|
case OP_MINSTARI: |
| 207 |
|
case OP_PLUSI: |
| 208 |
|
case OP_MINPLUSI: |
| 209 |
|
case OP_QUERYI: |
| 210 |
|
case OP_MINQUERYI: |
| 211 |
|
case OP_UPTOI: |
| 212 |
|
case OP_MINUPTOI: |
| 213 |
|
case OP_EXACTI: |
| 214 |
|
case OP_POSSTARI: |
| 215 |
|
case OP_POSPLUSI: |
| 216 |
|
case OP_POSQUERYI: |
| 217 |
|
case OP_POSUPTOI: |
| 218 |
|
case OP_NOTSTAR: |
| 219 |
|
case OP_NOTMINSTAR: |
| 220 |
|
case OP_NOTPLUS: |
| 221 |
|
case OP_NOTMINPLUS: |
| 222 |
|
case OP_NOTQUERY: |
| 223 |
|
case OP_NOTMINQUERY: |
| 224 |
|
case OP_NOTUPTO: |
| 225 |
|
case OP_NOTMINUPTO: |
| 226 |
|
case OP_NOTEXACT: |
| 227 |
|
case OP_NOTPOSSTAR: |
| 228 |
|
case OP_NOTPOSPLUS: |
| 229 |
|
case OP_NOTPOSQUERY: |
| 230 |
|
case OP_NOTPOSUPTO: |
| 231 |
|
case OP_NOTSTARI: |
| 232 |
|
case OP_NOTMINSTARI: |
| 233 |
|
case OP_NOTPLUSI: |
| 234 |
|
case OP_NOTMINPLUSI: |
| 235 |
|
case OP_NOTQUERYI: |
| 236 |
|
case OP_NOTMINQUERYI: |
| 237 |
|
case OP_NOTUPTOI: |
| 238 |
|
case OP_NOTMINUPTOI: |
| 239 |
|
case OP_NOTEXACTI: |
| 240 |
|
case OP_NOTPOSSTARI: |
| 241 |
|
case OP_NOTPOSPLUSI: |
| 242 |
|
case OP_NOTPOSQUERYI: |
| 243 |
|
case OP_NOTPOSUPTOI: |
| 244 |
|
if (utf) utf16_char = TRUE; |
| 245 |
|
#endif |
| 246 |
|
/* Fall through. */ |
| 247 |
|
|
| 248 |
|
default: |
| 249 |
|
length = PRIV(OP_lengths)[*ptr] - 1; |
| 250 |
|
break; |
| 251 |
|
|
| 252 |
|
case OP_CLASS: |
| 253 |
|
case OP_NCLASS: |
| 254 |
|
/* Skip the character bit map. */ |
| 255 |
|
ptr += 32/sizeof(pcre_uchar); |
| 256 |
|
length = 0; |
| 257 |
|
break; |
| 258 |
|
|
| 259 |
|
case OP_XCLASS: |
| 260 |
|
/* Reverse the size of the XCLASS instance. */ |
| 261 |
|
ptr++; |
| 262 |
|
*ptr = swap_uint16(*ptr); |
| 263 |
|
if (LINK_SIZE > 1) |
| 264 |
|
{ |
| 265 |
|
/* LINK_SIZE can be 1 or 2 in 16 bit mode. */ |
| 266 |
|
ptr++; |
| 267 |
|
*ptr = swap_uint16(*ptr); |
| 268 |
|
} |
| 269 |
|
ptr++; |
| 270 |
|
length = (GET(ptr, -LINK_SIZE)) - (1 + LINK_SIZE + 1); |
| 271 |
|
*ptr = swap_uint16(*ptr); |
| 272 |
|
if ((*ptr & XCL_MAP) != 0) |
| 273 |
|
{ |
| 274 |
|
/* Skip the character bit map. */ |
| 275 |
|
ptr += 32/sizeof(pcre_uchar); |
| 276 |
|
length -= 32/sizeof(pcre_uchar); |
| 277 |
|
} |
| 278 |
|
break; |
| 279 |
|
} |
| 280 |
|
ptr++; |
| 281 |
} |
} |
| 282 |
|
/* Control should never reach here in 16 bit mode. */ |
| 283 |
|
#endif /* !COMPILE_PCRE8 */ |
| 284 |
|
|
| 285 |
return internal_re; |
return 0; |
| 286 |
} |
} |
| 287 |
|
|
| 288 |
/* End of pcre_tryflipped.c */ |
/* End of pcre_byte_order.c */ |