|
Lines 330-335
Link Here
|
| 330 |
static int use_dfa; |
330 |
static int use_dfa; |
| 331 |
static int use_dfa_checked = 0; |
331 |
static int use_dfa_checked = 0; |
| 332 |
#ifdef MBS_SUPPORT |
332 |
#ifdef MBS_SUPPORT |
|
|
333 |
const char *last_char = NULL; |
| 333 |
int mb_cur_max = MB_CUR_MAX; |
334 |
int mb_cur_max = MB_CUR_MAX; |
| 334 |
mbstate_t mbs; |
335 |
mbstate_t mbs; |
| 335 |
memset (&mbs, '\0', sizeof (mbstate_t)); |
336 |
memset (&mbs, '\0', sizeof (mbstate_t)); |
|
Lines 385-390
Link Here
|
| 385 |
while (bytes_left) |
386 |
while (bytes_left) |
| 386 |
{ |
387 |
{ |
| 387 |
size_t mlen = mbrlen (beg, bytes_left, &mbs); |
388 |
size_t mlen = mbrlen (beg, bytes_left, &mbs); |
|
|
389 |
|
| 390 |
last_char = beg; |
| 388 |
if (mlen == (size_t) -1 || mlen == 0) |
391 |
if (mlen == (size_t) -1 || mlen == 0) |
| 389 |
{ |
392 |
{ |
| 390 |
/* Incomplete character: treat as single-byte. */ |
393 |
/* Incomplete character: treat as single-byte. */ |
|
Lines 445-450
Link Here
|
| 445 |
while (bytes_left) |
448 |
while (bytes_left) |
| 446 |
{ |
449 |
{ |
| 447 |
size_t mlen = mbrlen (beg, bytes_left, &mbs); |
450 |
size_t mlen = mbrlen (beg, bytes_left, &mbs); |
|
|
451 |
|
| 452 |
last_char = beg; |
| 448 |
if (mlen == (size_t) -1 || mlen == 0) |
453 |
if (mlen == (size_t) -1 || mlen == 0) |
| 449 |
{ |
454 |
{ |
| 450 |
/* Incomplete character: treat as single-byte. */ |
455 |
/* Incomplete character: treat as single-byte. */ |
|
Lines 507-516
Link Here
|
| 507 |
if (match_words) |
512 |
if (match_words) |
| 508 |
while (start >= 0) |
513 |
while (start >= 0) |
| 509 |
{ |
514 |
{ |
| 510 |
if ((start == 0 || !WCHAR ((unsigned char) beg[start - 1])) |
515 |
int lword_match = 0; |
| 511 |
&& (len == end - beg - 1 |
516 |
if (start == 0) |
| 512 |
|| !WCHAR ((unsigned char) beg[start + len]))) |
517 |
lword_match = 1; |
| 513 |
goto success_in_beg_and_end; |
518 |
else |
|
|
519 |
{ |
| 520 |
assert (start > 0); |
| 521 |
#ifdef MBS_SUPPORT |
| 522 |
if (mb_cur_max > 1) |
| 523 |
{ |
| 524 |
const char *s; |
| 525 |
int mr; |
| 526 |
wchar_t pwc; |
| 527 |
|
| 528 |
if (using_utf8) |
| 529 |
{ |
| 530 |
s = beg + start - 1; |
| 531 |
while (s > buf |
| 532 |
&& (unsigned char) *s >= 0x80 |
| 533 |
&& (unsigned char) *s <= 0xbf) |
| 534 |
--s; |
| 535 |
} |
| 536 |
else |
| 537 |
s = last_char; |
| 538 |
mr = mbtowc (&pwc, s, beg + start - s); |
| 539 |
if (mr <= 0) |
| 540 |
{ |
| 541 |
memset (&mbs, '\0', sizeof (mbstate_t)); |
| 542 |
lword_match = 1; |
| 543 |
} |
| 544 |
else if (!(iswalnum (pwc) || pwc == L'_') |
| 545 |
&& mr == (int) (beg + start - s)) |
| 546 |
lword_match = 1; |
| 547 |
} |
| 548 |
else |
| 549 |
#endif /* MBS_SUPPORT */ |
| 550 |
if (!WCHAR ((unsigned char) beg[start - 1])) |
| 551 |
lword_match = 1; |
| 552 |
} |
| 553 |
|
| 554 |
if (lword_match) |
| 555 |
{ |
| 556 |
int rword_match = 0; |
| 557 |
if (start + len == end - beg - 1) |
| 558 |
rword_match = 1; |
| 559 |
else |
| 560 |
{ |
| 561 |
#ifdef MBS_SUPPORT |
| 562 |
if (mb_cur_max > 1) |
| 563 |
{ |
| 564 |
wchar_t nwc; |
| 565 |
int mr; |
| 566 |
|
| 567 |
mr = mbtowc (&nwc, beg + start + len, |
| 568 |
end - beg - start - len - 1); |
| 569 |
if (mr <= 0) |
| 570 |
{ |
| 571 |
memset (&mbs, '\0', sizeof (mbstate_t)); |
| 572 |
rword_match = 1; |
| 573 |
} |
| 574 |
else if (!iswalnum (nwc) && nwc != L'_') |
| 575 |
rword_match = 1; |
| 576 |
} |
| 577 |
else |
| 578 |
#endif /* MBS_SUPPORT */ |
| 579 |
if (!WCHAR ((unsigned char) beg[start + len])) |
| 580 |
rword_match = 1; |
| 581 |
} |
| 582 |
|
| 583 |
if (rword_match) |
| 584 |
{ |
| 585 |
if (!exact) |
| 586 |
/* Returns the whole line. */ |
| 587 |
goto success_in_beg_and_end; |
| 588 |
else |
| 589 |
/* Returns just this word match. */ |
| 590 |
goto success_in_start_and_len; |
| 591 |
} |
| 592 |
} |
| 514 |
if (len > 0) |
593 |
if (len > 0) |
| 515 |
{ |
594 |
{ |
| 516 |
/* Try a shorter length anchored at the same place. */ |
595 |
/* Try a shorter length anchored at the same place. */ |
