| 490 |
|
|
| 491 |
|
|
| 492 |
/************************************************* |
/************************************************* |
| 493 |
|
* Set bits for a positive character type * |
| 494 |
|
*************************************************/ |
| 495 |
|
|
| 496 |
|
/* This function sets starting bits for a character type. In UTF-8 mode, we can |
| 497 |
|
only do a direct setting for bytes less than 128, as otherwise there can be |
| 498 |
|
confusion with bytes in the middle of UTF-8 characters. In a "traditional" |
| 499 |
|
environment, the tables will only recognize ASCII characters anyway, but in at |
| 500 |
|
least one Windows environment, some higher bytes bits were set in the tables. |
| 501 |
|
So we deal with that case by considering the UTF-8 encoding. |
| 502 |
|
|
| 503 |
|
Arguments: |
| 504 |
|
start_bits the starting bitmap |
| 505 |
|
cbit type the type of character wanted |
| 506 |
|
table_limit 32 for non-UTF-8; 16 for UTF-8 |
| 507 |
|
cd the block with char table pointers |
| 508 |
|
|
| 509 |
|
Returns: nothing |
| 510 |
|
*/ |
| 511 |
|
|
| 512 |
|
static void |
| 513 |
|
set_type_bits(uschar *start_bits, int cbit_type, int table_limit, |
| 514 |
|
compile_data *cd) |
| 515 |
|
{ |
| 516 |
|
register int c; |
| 517 |
|
for (c = 0; c < table_limit; c++) start_bits[c] |= cd->cbits[c+cbit_type]; |
| 518 |
|
if (table_limit == 32) return; |
| 519 |
|
for (c = 128; c < 256; c++) |
| 520 |
|
{ |
| 521 |
|
if ((cd->cbits[c/8] & (1 << (c&7))) != 0) |
| 522 |
|
{ |
| 523 |
|
uschar buff[8]; |
| 524 |
|
(void)_pcre_ord2utf8(c, buff); |
| 525 |
|
SET_BIT(buff[0]); |
| 526 |
|
} |
| 527 |
|
} |
| 528 |
|
} |
| 529 |
|
|
| 530 |
|
|
| 531 |
|
/************************************************* |
| 532 |
|
* Set bits for a negative character type * |
| 533 |
|
*************************************************/ |
| 534 |
|
|
| 535 |
|
/* This function sets starting bits for a negative character type such as \D. |
| 536 |
|
In UTF-8 mode, we can only do a direct setting for bytes less than 128, as |
| 537 |
|
otherwise there can be confusion with bytes in the middle of UTF-8 characters. |
| 538 |
|
Unlike in the positive case, where we can set appropriate starting bits for |
| 539 |
|
specific high-valued UTF-8 characters, in this case we have to set the bits for |
| 540 |
|
all high-valued characters. The lowest is 0xc2, but we overkill by starting at |
| 541 |
|
0xc0 (192) for simplicity. |
| 542 |
|
|
| 543 |
|
Arguments: |
| 544 |
|
start_bits the starting bitmap |
| 545 |
|
cbit type the type of character wanted |
| 546 |
|
table_limit 32 for non-UTF-8; 16 for UTF-8 |
| 547 |
|
cd the block with char table pointers |
| 548 |
|
|
| 549 |
|
Returns: nothing |
| 550 |
|
*/ |
| 551 |
|
|
| 552 |
|
static void |
| 553 |
|
set_nottype_bits(uschar *start_bits, int cbit_type, int table_limit, |
| 554 |
|
compile_data *cd) |
| 555 |
|
{ |
| 556 |
|
register int c; |
| 557 |
|
for (c = 0; c < table_limit; c++) start_bits[c] |= ~cd->cbits[c+cbit_type]; |
| 558 |
|
if (table_limit != 32) for (c = 24; c < 32; c++) start_bits[c] = 0xff; |
| 559 |
|
} |
| 560 |
|
|
| 561 |
|
|
| 562 |
|
|
| 563 |
|
/************************************************* |
| 564 |
* Create bitmap of starting bytes * |
* Create bitmap of starting bytes * |
| 565 |
*************************************************/ |
*************************************************/ |
| 566 |
|
|
| 776 |
|
|
| 777 |
/* Single character types set the bits and stop. Note that if PCRE_UCP |
/* Single character types set the bits and stop. Note that if PCRE_UCP |
| 778 |
is set, we do not see these op codes because \d etc are converted to |
is set, we do not see these op codes because \d etc are converted to |
| 779 |
properties. Therefore, these apply in the case when only ASCII characters |
properties. Therefore, these apply in the case when only characters less |
| 780 |
are recognized to match the types. In UTF-8 mode, we must restrict |
than 256 are recognized to match the types. */ |
|
ourselves to bytes less than 128, as otherwise there can be confusion |
|
|
with bytes in the middle of UTF-8 characters. (In a "traditional" |
|
|
environment, the tables will only recognize ASCII characters anyway, but |
|
|
in at least one Windows environment, some higher bytes bits were set in |
|
|
the tables.) */ |
|
| 781 |
|
|
| 782 |
case OP_NOT_DIGIT: |
case OP_NOT_DIGIT: |
| 783 |
for (c = 0; c < table_limit; c++) |
set_nottype_bits(start_bits, cbit_digit, table_limit, cd); |
|
start_bits[c] |= ~cd->cbits[c+cbit_digit]; |
|
| 784 |
try_next = FALSE; |
try_next = FALSE; |
| 785 |
break; |
break; |
| 786 |
|
|
| 787 |
case OP_DIGIT: |
case OP_DIGIT: |
| 788 |
for (c = 0; c < table_limit; c++) |
set_type_bits(start_bits, cbit_digit, table_limit, cd); |
|
start_bits[c] |= cd->cbits[c+cbit_digit]; |
|
| 789 |
try_next = FALSE; |
try_next = FALSE; |
| 790 |
break; |
break; |
| 791 |
|
|
| 792 |
/* The cbit_space table has vertical tab as whitespace; we have to |
/* The cbit_space table has vertical tab as whitespace; we have to |
| 793 |
discard it. */ |
ensure it is set as not whitespace. */ |
| 794 |
|
|
| 795 |
case OP_NOT_WHITESPACE: |
case OP_NOT_WHITESPACE: |
| 796 |
for (c = 0; c < table_limit; c++) |
set_nottype_bits(start_bits, cbit_space, table_limit, cd); |
| 797 |
{ |
start_bits[1] |= 0x08; |
|
int d = cd->cbits[c+cbit_space]; |
|
|
if (c == 1) d &= ~0x08; |
|
|
start_bits[c] |= ~d; |
|
|
} |
|
| 798 |
try_next = FALSE; |
try_next = FALSE; |
| 799 |
break; |
break; |
| 800 |
|
|
| 801 |
/* The cbit_space table has vertical tab as whitespace; we have to |
/* The cbit_space table has vertical tab as whitespace; we have to |
| 802 |
discard it. */ |
not set it from the table. */ |
| 803 |
|
|
| 804 |
case OP_WHITESPACE: |
case OP_WHITESPACE: |
| 805 |
for (c = 0; c < table_limit; c++) |
c = start_bits[1]; /* Save in case it was already set */ |
| 806 |
{ |
set_type_bits(start_bits, cbit_space, table_limit, cd); |
| 807 |
int d = cd->cbits[c+cbit_space]; |
start_bits[1] = (start_bits[1] & ~0x08) | c; |
|
if (c == 1) d &= ~0x08; |
|
|
start_bits[c] |= d; |
|
|
} |
|
| 808 |
try_next = FALSE; |
try_next = FALSE; |
| 809 |
break; |
break; |
| 810 |
|
|
| 811 |
case OP_NOT_WORDCHAR: |
case OP_NOT_WORDCHAR: |
| 812 |
for (c = 0; c < table_limit; c++) |
set_nottype_bits(start_bits, cbit_word, table_limit, cd); |
|
start_bits[c] |= ~cd->cbits[c+cbit_word]; |
|
| 813 |
try_next = FALSE; |
try_next = FALSE; |
| 814 |
break; |
break; |
| 815 |
|
|
| 816 |
case OP_WORDCHAR: |
case OP_WORDCHAR: |
| 817 |
for (c = 0; c < table_limit; c++) |
set_type_bits(start_bits, cbit_word, table_limit, cd); |
|
start_bits[c] |= cd->cbits[c+cbit_word]; |
|
| 818 |
try_next = FALSE; |
try_next = FALSE; |
| 819 |
break; |
break; |
| 820 |
|
|
| 821 |
/* One or more character type fudges the pointer and restarts, knowing |
/* One or more character type fudges the pointer and restarts, knowing |
| 822 |
it will hit a single character type and stop there. */ |
it will hit a single character type and stop there. */ |
| 823 |
|
|
| 880 |
break; |
break; |
| 881 |
|
|
| 882 |
case OP_NOT_DIGIT: |
case OP_NOT_DIGIT: |
| 883 |
for (c = 0; c < table_limit; c++) |
set_nottype_bits(start_bits, cbit_digit, table_limit, cd); |
|
start_bits[c] |= ~cd->cbits[c+cbit_digit]; |
|
| 884 |
break; |
break; |
| 885 |
|
|
| 886 |
case OP_DIGIT: |
case OP_DIGIT: |
| 887 |
for (c = 0; c < table_limit; c++) |
set_type_bits(start_bits, cbit_digit, table_limit, cd); |
|
start_bits[c] |= cd->cbits[c+cbit_digit]; |
|
| 888 |
break; |
break; |
| 889 |
|
|
| 890 |
/* The cbit_space table has vertical tab as whitespace; we have to |
/* The cbit_space table has vertical tab as whitespace; we have to |
| 891 |
discard it. */ |
ensure it gets set as not whitespace. */ |
| 892 |
|
|
| 893 |
case OP_NOT_WHITESPACE: |
case OP_NOT_WHITESPACE: |
| 894 |
for (c = 0; c < table_limit; c++) |
set_nottype_bits(start_bits, cbit_space, table_limit, cd); |
| 895 |
{ |
start_bits[1] |= 0x08; |
|
int d = cd->cbits[c+cbit_space]; |
|
|
if (c == 1) d &= ~0x08; |
|
|
start_bits[c] |= ~d; |
|
|
} |
|
| 896 |
break; |
break; |
| 897 |
|
|
| 898 |
/* The cbit_space table has vertical tab as whitespace; we have to |
/* The cbit_space table has vertical tab as whitespace; we have to |
| 899 |
discard it. */ |
avoid setting it. */ |
| 900 |
|
|
| 901 |
case OP_WHITESPACE: |
case OP_WHITESPACE: |
| 902 |
for (c = 0; c < table_limit; c++) |
c = start_bits[1]; /* Save in case it was already set */ |
| 903 |
{ |
set_type_bits(start_bits, cbit_space, table_limit, cd); |
| 904 |
int d = cd->cbits[c+cbit_space]; |
start_bits[1] = (start_bits[1] & ~0x08) | c; |
|
if (c == 1) d &= ~0x08; |
|
|
start_bits[c] |= d; |
|
|
} |
|
| 905 |
break; |
break; |
| 906 |
|
|
| 907 |
case OP_NOT_WORDCHAR: |
case OP_NOT_WORDCHAR: |
| 908 |
for (c = 0; c < table_limit; c++) |
set_nottype_bits(start_bits, cbit_word, table_limit, cd); |
|
start_bits[c] |= ~cd->cbits[c+cbit_word]; |
|
| 909 |
break; |
break; |
| 910 |
|
|
| 911 |
case OP_WORDCHAR: |
case OP_WORDCHAR: |
| 912 |
for (c = 0; c < table_limit; c++) |
set_type_bits(start_bits, cbit_word, table_limit, cd); |
|
start_bits[c] |= cd->cbits[c+cbit_word]; |
|
| 913 |
break; |
break; |
| 914 |
} |
} |
| 915 |
|
|