Contents of /trunk/coreutils/patches-6.9/coreutils-6.9-i18n.patch
Parent Directory | Revision Log
Revision 223 -
(show annotations)
(download)
Sat Jun 23 13:51:30 2007 UTC (17 years, 3 months ago) by niro
File size: 103622 byte(s)
Sat Jun 23 13:51:30 2007 UTC (17 years, 3 months ago) by niro
File size: 103622 byte(s)
-rev bump to 6.9
1 | Submitted by: Matt Burgess (matthew at linuxfromscratch.org) |
2 | Date: 2007-04-07 |
3 | Initial Package Version: 6.9 |
4 | Upstream Status: Rejected |
5 | Origin: Based on coreutils-5.93-i18n-2.patch by Alexander Patrakov |
6 | Description: This patch fixes various problems with multibyte character support. |
7 | LSB >= 2.0 tests for features added by this patch, but only Coreutils-5.2.1 plus |
8 | http://www.linuxfromscratch.org/~alexander/patches/coreutils-5.2.1-i18n_fixes-1.patch |
9 | actually pass the Li18nux2000-level1 testsuite. |
10 | |
11 | diff -Naur coreutils-6.9.orig/lib/linebuffer.h coreutils-6.9/lib/linebuffer.h |
12 | --- coreutils-6.9.orig/lib/linebuffer.h 2005-05-14 06:03:58.000000000 +0000 |
13 | +++ coreutils-6.9/lib/linebuffer.h 2007-04-07 16:59:55.000000000 +0000 |
14 | @@ -22,6 +22,11 @@ |
15 | |
16 | # include <stdio.h> |
17 | |
18 | +/* Get mbstate_t. */ |
19 | +# if HAVE_WCHAR_H |
20 | +# include <wchar.h> |
21 | +# endif |
22 | + |
23 | /* A `struct linebuffer' holds a line of text. */ |
24 | |
25 | struct linebuffer |
26 | @@ -29,6 +34,9 @@ |
27 | size_t size; /* Allocated. */ |
28 | size_t length; /* Used. */ |
29 | char *buffer; |
30 | +# if HAVE_WCHAR_H |
31 | + mbstate_t state; |
32 | +# endif |
33 | }; |
34 | |
35 | /* Initialize linebuffer LINEBUFFER for use. */ |
36 | diff -Naur coreutils-6.9.orig/src/cut.c coreutils-6.9/src/cut.c |
37 | --- coreutils-6.9.orig/src/cut.c 2007-03-18 21:36:43.000000000 +0000 |
38 | +++ coreutils-6.9/src/cut.c 2007-04-07 16:59:55.000000000 +0000 |
39 | @@ -29,6 +29,11 @@ |
40 | #include <assert.h> |
41 | #include <getopt.h> |
42 | #include <sys/types.h> |
43 | + |
44 | +/* Get mbstate_t, mbrtowc(). */ |
45 | +#if HAVE_WCHAR_H |
46 | +# include <wchar.h> |
47 | +#endif |
48 | #include "system.h" |
49 | |
50 | #include "error.h" |
51 | @@ -37,6 +42,18 @@ |
52 | #include "quote.h" |
53 | #include "xstrndup.h" |
54 | |
55 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
56 | + installation; work around this configuration error. */ |
57 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
58 | +# undef MB_LEN_MAX |
59 | +# define MB_LEN_MAX 16 |
60 | +#endif |
61 | + |
62 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
63 | +#if HAVE_MBRTOWC && defined mbstate_t |
64 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
65 | +#endif |
66 | + |
67 | /* The official name of this program (e.g., no `g' prefix). */ |
68 | #define PROGRAM_NAME "cut" |
69 | |
70 | @@ -67,6 +84,52 @@ |
71 | } \ |
72 | while (0) |
73 | |
74 | +/* Refill the buffer BUF to get a multibyte character. */ |
75 | +#define REFILL_BUFFER(BUF, BUFPOS, BUFLEN, STREAM) \ |
76 | + do \ |
77 | + { \ |
78 | + if (BUFLEN < MB_LEN_MAX && !feof (STREAM) && !ferror (STREAM)) \ |
79 | + { \ |
80 | + memmove (BUF, BUFPOS, BUFLEN); \ |
81 | + BUFLEN += fread (BUF + BUFLEN, sizeof(char), BUFSIZ, STREAM); \ |
82 | + BUFPOS = BUF; \ |
83 | + } \ |
84 | + } \ |
85 | + while (0) |
86 | + |
87 | +/* Get wide character on BUFPOS. BUFPOS is not included after that. |
88 | + If byte sequence is not valid as a character, CONVFAIL is 1. Otherwise 0. */ |
89 | +#define GET_NEXT_WC_FROM_BUFFER(WC, BUFPOS, BUFLEN, MBLENGTH, STATE, CONVFAIL) \ |
90 | + do \ |
91 | + { \ |
92 | + mbstate_t state_bak; \ |
93 | + \ |
94 | + if (BUFLEN < 1) \ |
95 | + { \ |
96 | + WC = WEOF; \ |
97 | + break; \ |
98 | + } \ |
99 | + \ |
100 | + /* Get a wide character. */ \ |
101 | + CONVFAIL = 0; \ |
102 | + state_bak = STATE; \ |
103 | + MBLENGTH = mbrtowc ((wchar_t *)&WC, BUFPOS, BUFLEN, &STATE); \ |
104 | + \ |
105 | + switch (MBLENGTH) \ |
106 | + { \ |
107 | + case (size_t)-1: \ |
108 | + case (size_t)-2: \ |
109 | + CONVFAIL++; \ |
110 | + STATE = state_bak; \ |
111 | + /* Fall througn. */ \ |
112 | + \ |
113 | + case 0: \ |
114 | + MBLENGTH = 1; \ |
115 | + break; \ |
116 | + } \ |
117 | + } \ |
118 | + while (0) |
119 | + |
120 | struct range_pair |
121 | { |
122 | size_t lo; |
123 | @@ -85,7 +148,7 @@ |
124 | /* The number of bytes allocated for FIELD_1_BUFFER. */ |
125 | static size_t field_1_bufsize; |
126 | |
127 | -/* The largest field or byte index used as an endpoint of a closed |
128 | +/* The largest byte, character or field index used as an endpoint of a closed |
129 | or degenerate range specification; this doesn't include the starting |
130 | index of right-open-ended ranges. For example, with either range spec |
131 | `2-5,9-', `2-3,5,9-' this variable would be set to 5. */ |
132 | @@ -97,10 +160,11 @@ |
133 | |
134 | /* This is a bit vector. |
135 | In byte mode, which bytes to output. |
136 | + In character mode, which characters to output. |
137 | In field mode, which DELIM-separated fields to output. |
138 | - Both bytes and fields are numbered starting with 1, |
139 | + Bytes, characters and fields are numbered starting with 1, |
140 | so the zeroth bit of this array is unused. |
141 | - A field or byte K has been selected if |
142 | + A byte, character or field K has been selected if |
143 | (K <= MAX_RANGE_ENDPOINT and is_printable_field(K)) |
144 | || (EOL_RANGE_START > 0 && K >= EOL_RANGE_START). */ |
145 | static unsigned char *printable_field; |
146 | @@ -109,9 +173,12 @@ |
147 | { |
148 | undefined_mode, |
149 | |
150 | - /* Output characters that are in the given bytes. */ |
151 | + /* Output bytes that are at the given positions. */ |
152 | byte_mode, |
153 | |
154 | + /* Output characters that are at the given positions. */ |
155 | + character_mode, |
156 | + |
157 | /* Output the given delimeter-separated fields. */ |
158 | field_mode |
159 | }; |
160 | @@ -121,6 +188,13 @@ |
161 | |
162 | static enum operating_mode operating_mode; |
163 | |
164 | +/* If nonzero, when in byte mode, don't split multibyte characters. */ |
165 | +static int byte_mode_character_aware; |
166 | + |
167 | +/* If nonzero, the function for single byte locale is work |
168 | + if this program runs on multibyte locale. */ |
169 | +static int force_singlebyte_mode; |
170 | + |
171 | /* If true do not output lines containing no delimeter characters. |
172 | Otherwise, all such lines are printed. This option is valid only |
173 | with field mode. */ |
174 | @@ -132,6 +206,9 @@ |
175 | |
176 | /* The delimeter character for field mode. */ |
177 | static unsigned char delim; |
178 | +#if HAVE_WCHAR_H |
179 | +static wchar_t wcdelim; |
180 | +#endif |
181 | |
182 | /* True if the --output-delimiter=STRING option was specified. */ |
183 | static bool output_delimiter_specified; |
184 | @@ -205,7 +282,7 @@ |
185 | -f, --fields=LIST select only these fields; also print any line\n\ |
186 | that contains no delimiter character, unless\n\ |
187 | the -s option is specified\n\ |
188 | - -n (ignored)\n\ |
189 | + -n with -b: don't split multibyte characters\n\ |
190 | "), stdout); |
191 | fputs (_("\ |
192 | --complement complement the set of selected bytes, characters\n\ |
193 | @@ -362,7 +439,7 @@ |
194 | in_digits = false; |
195 | /* Starting a range. */ |
196 | if (dash_found) |
197 | - FATAL_ERROR (_("invalid byte or field list")); |
198 | + FATAL_ERROR (_("invalid byte, character or field list")); |
199 | dash_found = true; |
200 | fieldstr++; |
201 | |
202 | @@ -387,14 +464,16 @@ |
203 | if (value == 0) |
204 | { |
205 | /* `n-'. From `initial' to end of line. */ |
206 | - eol_range_start = initial; |
207 | + if (eol_range_start == 0 || |
208 | + (eol_range_start != 0 && eol_range_start > initial)) |
209 | + eol_range_start = initial; |
210 | field_found = true; |
211 | } |
212 | else |
213 | { |
214 | /* `m-n' or `-n' (1-n). */ |
215 | if (value < initial) |
216 | - FATAL_ERROR (_("invalid byte or field list")); |
217 | + FATAL_ERROR (_("invalid byte, character or field list")); |
218 | |
219 | /* Is there already a range going to end of line? */ |
220 | if (eol_range_start != 0) |
221 | @@ -467,6 +546,9 @@ |
222 | if (operating_mode == byte_mode) |
223 | error (0, 0, |
224 | _("byte offset %s is too large"), quote (bad_num)); |
225 | + else if (operating_mode == character_mode) |
226 | + error (0, 0, |
227 | + _("character offset %s is too large"), quote (bad_num)); |
228 | else |
229 | error (0, 0, |
230 | _("field number %s is too large"), quote (bad_num)); |
231 | @@ -477,7 +559,7 @@ |
232 | fieldstr++; |
233 | } |
234 | else |
235 | - FATAL_ERROR (_("invalid byte or field list")); |
236 | + FATAL_ERROR (_("invalid byte, character or field list")); |
237 | } |
238 | |
239 | max_range_endpoint = 0; |
240 | @@ -570,6 +652,63 @@ |
241 | } |
242 | } |
243 | |
244 | +#if HAVE_MBRTOWC |
245 | +/* This function is in use for the following case. |
246 | + |
247 | + 1. Read from the stream STREAM, printing to standard output any selected |
248 | + characters. |
249 | + |
250 | + 2. Read from stream STREAM, printing to standard output any selected bytes, |
251 | + without splitting multibyte characters. */ |
252 | + |
253 | +static void |
254 | +cut_characters_or_cut_bytes_no_split (FILE *stream) |
255 | +{ |
256 | + int idx; /* number of bytes or characters in the line so far. */ |
257 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
258 | + char *bufpos; /* Next read position of BUF. */ |
259 | + size_t buflen; /* The length of the byte sequence in buf. */ |
260 | + wint_t wc; /* A gotten wide character. */ |
261 | + size_t mblength; /* The byte size of a multibyte character which shows |
262 | + as same character as WC. */ |
263 | + mbstate_t state; /* State of the stream. */ |
264 | + int convfail; /* 1, when conversion is failed. Otherwise 0. */ |
265 | + |
266 | + idx = 0; |
267 | + buflen = 0; |
268 | + bufpos = buf; |
269 | + memset (&state, '\0', sizeof(mbstate_t)); |
270 | + |
271 | + while (1) |
272 | + { |
273 | + REFILL_BUFFER (buf, bufpos, buflen, stream); |
274 | + |
275 | + GET_NEXT_WC_FROM_BUFFER (wc, bufpos, buflen, mblength, state, convfail); |
276 | + |
277 | + if (wc == WEOF) |
278 | + { |
279 | + if (idx > 0) |
280 | + putchar ('\n'); |
281 | + break; |
282 | + } |
283 | + else if (wc == L'\n') |
284 | + { |
285 | + putchar ('\n'); |
286 | + idx = 0; |
287 | + } |
288 | + else |
289 | + { |
290 | + idx += (operating_mode == byte_mode) ? mblength : 1; |
291 | + if (print_kth (idx, NULL)) |
292 | + fwrite (bufpos, mblength, sizeof(char), stdout); |
293 | + } |
294 | + |
295 | + buflen -= mblength; |
296 | + bufpos += mblength; |
297 | + } |
298 | +} |
299 | +#endif |
300 | + |
301 | /* Read from stream STREAM, printing to standard output any selected fields. */ |
302 | |
303 | static void |
304 | @@ -692,13 +831,192 @@ |
305 | } |
306 | } |
307 | |
308 | +#if HAVE_MBRTOWC |
309 | +static void |
310 | +cut_fields_mb (FILE *stream) |
311 | +{ |
312 | + int c; |
313 | + unsigned int field_idx; |
314 | + int found_any_selected_field; |
315 | + int buffer_first_field; |
316 | + int empty_input; |
317 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
318 | + char *bufpos; /* Next read position of BUF. */ |
319 | + size_t buflen; /* The length of the byte sequence in buf. */ |
320 | + wint_t wc = 0; /* A gotten wide character. */ |
321 | + size_t mblength; /* The byte size of a multibyte character which shows |
322 | + as same character as WC. */ |
323 | + mbstate_t state; /* State of the stream. */ |
324 | + int convfail; /* 1, when conversion is failed. Otherwise 0. */ |
325 | + |
326 | + found_any_selected_field = 0; |
327 | + field_idx = 1; |
328 | + bufpos = buf; |
329 | + buflen = 0; |
330 | + memset (&state, '\0', sizeof(mbstate_t)); |
331 | + |
332 | + c = getc (stream); |
333 | + empty_input = (c == EOF); |
334 | + if (c != EOF) |
335 | + ungetc (c, stream); |
336 | + else |
337 | + wc = WEOF; |
338 | + |
339 | + /* To support the semantics of the -s flag, we may have to buffer |
340 | + all of the first field to determine whether it is `delimited.' |
341 | + But that is unnecessary if all non-delimited lines must be printed |
342 | + and the first field has been selected, or if non-delimited lines |
343 | + must be suppressed and the first field has *not* been selected. |
344 | + That is because a non-delimited line has exactly one field. */ |
345 | + buffer_first_field = (suppress_non_delimited ^ !print_kth (1, NULL)); |
346 | + |
347 | + while (1) |
348 | + { |
349 | + if (field_idx == 1 && buffer_first_field) |
350 | + { |
351 | + int len = 0; |
352 | + |
353 | + while (1) |
354 | + { |
355 | + REFILL_BUFFER (buf, bufpos, buflen, stream); |
356 | + |
357 | + GET_NEXT_WC_FROM_BUFFER |
358 | + (wc, bufpos, buflen, mblength, state, convfail); |
359 | + |
360 | + if (wc == WEOF) |
361 | + break; |
362 | + |
363 | + field_1_buffer = xrealloc (field_1_buffer, len + mblength); |
364 | + memcpy (field_1_buffer + len, bufpos, mblength); |
365 | + len += mblength; |
366 | + buflen -= mblength; |
367 | + bufpos += mblength; |
368 | + |
369 | + if (!convfail && (wc == L'\n' || wc == wcdelim)) |
370 | + break; |
371 | + } |
372 | + |
373 | + if (wc == WEOF) |
374 | + break; |
375 | + |
376 | + /* If the first field extends to the end of line (it is not |
377 | + delimited) and we are printing all non-delimited lines, |
378 | + print this one. */ |
379 | + if (convfail || (!convfail && wc != wcdelim)) |
380 | + { |
381 | + if (suppress_non_delimited) |
382 | + { |
383 | + /* Empty. */ |
384 | + } |
385 | + else |
386 | + { |
387 | + fwrite (field_1_buffer, sizeof (char), len, stdout); |
388 | + /* Make sure the output line is newline terminated. */ |
389 | + if (convfail || (!convfail && wc != L'\n')) |
390 | + putchar ('\n'); |
391 | + } |
392 | + continue; |
393 | + } |
394 | + |
395 | + if (print_kth (1, NULL)) |
396 | + { |
397 | + /* Print the field, but not the trailing delimiter. */ |
398 | + fwrite (field_1_buffer, sizeof (char), len - 1, stdout); |
399 | + found_any_selected_field = 1; |
400 | + } |
401 | + ++field_idx; |
402 | + } |
403 | + |
404 | + if (wc != WEOF) |
405 | + { |
406 | + if (print_kth (field_idx, NULL)) |
407 | + { |
408 | + if (found_any_selected_field) |
409 | + { |
410 | + fwrite (output_delimiter_string, sizeof (char), |
411 | + output_delimiter_length, stdout); |
412 | + } |
413 | + found_any_selected_field = 1; |
414 | + } |
415 | + |
416 | + while (1) |
417 | + { |
418 | + REFILL_BUFFER (buf, bufpos, buflen, stream); |
419 | + |
420 | + GET_NEXT_WC_FROM_BUFFER |
421 | + (wc, bufpos, buflen, mblength, state, convfail); |
422 | + |
423 | + if (wc == WEOF) |
424 | + break; |
425 | + else if (!convfail && (wc == wcdelim || wc == L'\n')) |
426 | + { |
427 | + buflen -= mblength; |
428 | + bufpos += mblength; |
429 | + break; |
430 | + } |
431 | + |
432 | + if (print_kth (field_idx, NULL)) |
433 | + fwrite (bufpos, mblength, sizeof(char), stdout); |
434 | + |
435 | + buflen -= mblength; |
436 | + bufpos += mblength; |
437 | + } |
438 | + } |
439 | + |
440 | + if ((!convfail || wc == L'\n') && buflen < 1) |
441 | + wc = WEOF; |
442 | + |
443 | + if (!convfail && wc == wcdelim) |
444 | + ++field_idx; |
445 | + else if (wc == WEOF || (!convfail && wc == L'\n')) |
446 | + { |
447 | + if (found_any_selected_field |
448 | + || (!empty_input && !(suppress_non_delimited && field_idx == 1))) |
449 | + putchar ('\n'); |
450 | + if (wc == WEOF) |
451 | + break; |
452 | + field_idx = 1; |
453 | + found_any_selected_field = 0; |
454 | + } |
455 | + } |
456 | +} |
457 | +#endif |
458 | + |
459 | static void |
460 | cut_stream (FILE *stream) |
461 | { |
462 | - if (operating_mode == byte_mode) |
463 | - cut_bytes (stream); |
464 | +#if HAVE_MBRTOWC |
465 | + if (MB_CUR_MAX > 1 && !force_singlebyte_mode) |
466 | + { |
467 | + switch (operating_mode) |
468 | + { |
469 | + case byte_mode: |
470 | + if (byte_mode_character_aware) |
471 | + cut_characters_or_cut_bytes_no_split (stream); |
472 | + else |
473 | + cut_bytes (stream); |
474 | + break; |
475 | + |
476 | + case character_mode: |
477 | + cut_characters_or_cut_bytes_no_split (stream); |
478 | + break; |
479 | + |
480 | + case field_mode: |
481 | + cut_fields_mb (stream); |
482 | + break; |
483 | + |
484 | + default: |
485 | + abort (); |
486 | + } |
487 | + } |
488 | else |
489 | - cut_fields (stream); |
490 | +#endif |
491 | + { |
492 | + if (operating_mode == field_mode) |
493 | + cut_fields (stream); |
494 | + else |
495 | + cut_bytes (stream); |
496 | + } |
497 | } |
498 | |
499 | /* Process file FILE to standard output. |
500 | @@ -748,6 +1066,8 @@ |
501 | bool ok; |
502 | bool delim_specified = false; |
503 | char *spec_list_string IF_LINT(= NULL); |
504 | + char mbdelim[MB_LEN_MAX + 1]; |
505 | + size_t delimlen = 0; |
506 | |
507 | initialize_main (&argc, &argv); |
508 | program_name = argv[0]; |
509 | @@ -770,7 +1090,6 @@ |
510 | switch (optc) |
511 | { |
512 | case 'b': |
513 | - case 'c': |
514 | /* Build the byte list. */ |
515 | if (operating_mode != undefined_mode) |
516 | FATAL_ERROR (_("only one type of list may be specified")); |
517 | @@ -778,6 +1097,14 @@ |
518 | spec_list_string = optarg; |
519 | break; |
520 | |
521 | + case 'c': |
522 | + /* Build the character list. */ |
523 | + if (operating_mode != undefined_mode) |
524 | + FATAL_ERROR (_("only one type of list may be specified")); |
525 | + operating_mode = character_mode; |
526 | + spec_list_string = optarg; |
527 | + break; |
528 | + |
529 | case 'f': |
530 | /* Build the field list. */ |
531 | if (operating_mode != undefined_mode) |
532 | @@ -789,10 +1116,35 @@ |
533 | case 'd': |
534 | /* New delimiter. */ |
535 | /* Interpret -d '' to mean `use the NUL byte as the delimiter.' */ |
536 | - if (optarg[0] != '\0' && optarg[1] != '\0') |
537 | - FATAL_ERROR (_("the delimiter must be a single character")); |
538 | - delim = optarg[0]; |
539 | - delim_specified = true; |
540 | +#if HAVE_MBRTOWC |
541 | + { |
542 | + if(MB_CUR_MAX > 1) |
543 | + { |
544 | + mbstate_t state; |
545 | + |
546 | + memset (&state, '\0', sizeof(mbstate_t)); |
547 | + delimlen = mbrtowc (&wcdelim, optarg, strnlen(optarg, MB_LEN_MAX), &state); |
548 | + |
549 | + if (delimlen == (size_t)-1 || delimlen == (size_t)-2) |
550 | + ++force_singlebyte_mode; |
551 | + else |
552 | + { |
553 | + delimlen = (delimlen < 1) ? 1 : delimlen; |
554 | + if (wcdelim != L'\0' && *(optarg + delimlen) != '\0') |
555 | + FATAL_ERROR (_("the delimiter must be a single character")); |
556 | + memcpy (mbdelim, optarg, delimlen); |
557 | + } |
558 | + } |
559 | + |
560 | + if (MB_CUR_MAX <= 1 || force_singlebyte_mode) |
561 | +#endif |
562 | + { |
563 | + if (optarg[0] != '\0' && optarg[1] != '\0') |
564 | + FATAL_ERROR (_("the delimiter must be a single character")); |
565 | + delim = (unsigned char) optarg[0]; |
566 | + } |
567 | + delim_specified = true; |
568 | + } |
569 | break; |
570 | |
571 | case OUTPUT_DELIMITER_OPTION: |
572 | @@ -805,6 +1157,7 @@ |
573 | break; |
574 | |
575 | case 'n': |
576 | + byte_mode_character_aware = 1; |
577 | break; |
578 | |
579 | case 's': |
580 | @@ -827,7 +1180,7 @@ |
581 | if (operating_mode == undefined_mode) |
582 | FATAL_ERROR (_("you must specify a list of bytes, characters, or fields")); |
583 | |
584 | - if (delim != '\0' && operating_mode != field_mode) |
585 | + if (delim_specified && operating_mode != field_mode) |
586 | FATAL_ERROR (_("an input delimiter may be specified only\ |
587 | when operating on fields")); |
588 | |
589 | @@ -854,15 +1207,34 @@ |
590 | } |
591 | |
592 | if (!delim_specified) |
593 | - delim = '\t'; |
594 | + { |
595 | + delim = '\t'; |
596 | +#ifdef HAVE_MBRTOWC |
597 | + wcdelim = L'\t'; |
598 | + mbdelim[0] = '\t'; |
599 | + mbdelim[1] = '\0'; |
600 | + delimlen = 1; |
601 | +#endif |
602 | + } |
603 | |
604 | if (output_delimiter_string == NULL) |
605 | { |
606 | - static char dummy[2]; |
607 | - dummy[0] = delim; |
608 | - dummy[1] = '\0'; |
609 | - output_delimiter_string = dummy; |
610 | - output_delimiter_length = 1; |
611 | +#ifdef HAVE_MBRTOWC |
612 | + if (MB_CUR_MAX > 1 && !force_singlebyte_mode) |
613 | + { |
614 | + output_delimiter_string = xstrdup(mbdelim); |
615 | + output_delimiter_length = delimlen; |
616 | + } |
617 | + |
618 | + if (MB_CUR_MAX <= 1 || force_singlebyte_mode) |
619 | +#endif |
620 | + { |
621 | + static char dummy[2]; |
622 | + dummy[0] = delim; |
623 | + dummy[1] = '\0'; |
624 | + output_delimiter_string = dummy; |
625 | + output_delimiter_length = 1; |
626 | + } |
627 | } |
628 | |
629 | if (optind == argc) |
630 | diff -Naur coreutils-6.9.orig/src/expand.c coreutils-6.9/src/expand.c |
631 | --- coreutils-6.9.orig/src/expand.c 2007-03-18 21:36:43.000000000 +0000 |
632 | +++ coreutils-6.9/src/expand.c 2007-04-07 16:59:55.000000000 +0000 |
633 | @@ -38,11 +38,28 @@ |
634 | #include <stdio.h> |
635 | #include <getopt.h> |
636 | #include <sys/types.h> |
637 | + |
638 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ |
639 | +#if HAVE_WCHAR_H |
640 | +# include <wchar.h> |
641 | +#endif |
642 | + |
643 | #include "system.h" |
644 | #include "error.h" |
645 | #include "quote.h" |
646 | #include "xstrndup.h" |
647 | |
648 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
649 | + installation; work around this configuration error. */ |
650 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
651 | +# define MB_LEN_MAX 16 |
652 | +#endif |
653 | + |
654 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
655 | +#if HAVE_MBRTOWC && defined mbstate_t |
656 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
657 | +#endif |
658 | + |
659 | /* The official name of this program (e.g., no `g' prefix). */ |
660 | #define PROGRAM_NAME "expand" |
661 | |
662 | @@ -183,6 +200,7 @@ |
663 | stops = num_start + len - 1; |
664 | } |
665 | } |
666 | + |
667 | else |
668 | { |
669 | error (0, 0, _("tab size contains invalid character(s): %s"), |
670 | @@ -365,6 +383,142 @@ |
671 | } |
672 | } |
673 | |
674 | +#if HAVE_MBRTOWC |
675 | +static void |
676 | +expand_multibyte (void) |
677 | +{ |
678 | + FILE *fp; /* Input strem. */ |
679 | + mbstate_t i_state; /* Current shift state of the input stream. */ |
680 | + mbstate_t i_state_bak; /* Back up the I_STATE. */ |
681 | + mbstate_t o_state; /* Current shift state of the output stream. */ |
682 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
683 | + char *bufpos; /* Next read position of BUF. */ |
684 | + size_t buflen = 0; /* The length of the byte sequence in buf. */ |
685 | + wchar_t wc; /* A gotten wide character. */ |
686 | + size_t mblength; /* The byte size of a multibyte character |
687 | + which shows as same character as WC. */ |
688 | + int tab_index = 0; /* Index in `tab_list' of next tabstop. */ |
689 | + int column = 0; /* Column on screen of the next char. */ |
690 | + int next_tab_column; /* Column the next tab stop is on. */ |
691 | + int convert = 1; /* If nonzero, perform translations. */ |
692 | + |
693 | + fp = next_file ((FILE *) NULL); |
694 | + if (fp == NULL) |
695 | + return; |
696 | + |
697 | + memset (&o_state, '\0', sizeof(mbstate_t)); |
698 | + memset (&i_state, '\0', sizeof(mbstate_t)); |
699 | + |
700 | + for (;;) |
701 | + { |
702 | + /* Refill the buffer BUF. */ |
703 | + if (buflen < MB_LEN_MAX && !feof(fp) && !ferror(fp)) |
704 | + { |
705 | + memmove (buf, bufpos, buflen); |
706 | + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, fp); |
707 | + bufpos = buf; |
708 | + } |
709 | + |
710 | + /* No character is left in BUF. */ |
711 | + if (buflen < 1) |
712 | + { |
713 | + fp = next_file (fp); |
714 | + |
715 | + if (fp == NULL) |
716 | + break; /* No more files. */ |
717 | + else |
718 | + { |
719 | + memset (&i_state, '\0', sizeof(mbstate_t)); |
720 | + continue; |
721 | + } |
722 | + } |
723 | + |
724 | + /* Get a wide character. */ |
725 | + i_state_bak = i_state; |
726 | + mblength = mbrtowc (&wc, bufpos, buflen, &i_state); |
727 | + |
728 | + switch (mblength) |
729 | + { |
730 | + case (size_t)-1: /* illegal byte sequence. */ |
731 | + case (size_t)-2: |
732 | + mblength = 1; |
733 | + i_state = i_state_bak; |
734 | + if (convert) |
735 | + { |
736 | + ++column; |
737 | + if (convert_entire_line == 0) |
738 | + convert = 0; |
739 | + } |
740 | + putchar (*bufpos); |
741 | + break; |
742 | + |
743 | + case 0: /* null. */ |
744 | + mblength = 1; |
745 | + if (convert && convert_entire_line == 0) |
746 | + convert = 0; |
747 | + putchar ('\0'); |
748 | + break; |
749 | + |
750 | + default: |
751 | + if (wc == L'\n') /* LF. */ |
752 | + { |
753 | + tab_index = 0; |
754 | + column = 0; |
755 | + convert = 1; |
756 | + putchar ('\n'); |
757 | + } |
758 | + else if (wc == L'\t' && convert) /* Tab. */ |
759 | + { |
760 | + if (tab_size == 0) |
761 | + { |
762 | + /* Do not let tab_index == first_free_tab; |
763 | + stop when it is 1 less. */ |
764 | + while (tab_index < first_free_tab - 1 |
765 | + && column >= tab_list[tab_index]) |
766 | + tab_index++; |
767 | + next_tab_column = tab_list[tab_index]; |
768 | + if (tab_index < first_free_tab - 1) |
769 | + tab_index++; |
770 | + if (column >= next_tab_column) |
771 | + next_tab_column = column + 1; |
772 | + } |
773 | + else |
774 | + next_tab_column = column + tab_size - column % tab_size; |
775 | + |
776 | + while (column < next_tab_column) |
777 | + { |
778 | + putchar (' '); |
779 | + ++column; |
780 | + } |
781 | + } |
782 | + else /* Others. */ |
783 | + { |
784 | + if (convert) |
785 | + { |
786 | + if (wc == L'\b') |
787 | + { |
788 | + if (column > 0) |
789 | + --column; |
790 | + } |
791 | + else |
792 | + { |
793 | + int width; /* The width of WC. */ |
794 | + |
795 | + width = wcwidth (wc); |
796 | + column += (width > 0) ? width : 0; |
797 | + if (convert_entire_line == 0) |
798 | + convert = 0; |
799 | + } |
800 | + } |
801 | + fwrite (bufpos, sizeof(char), mblength, stdout); |
802 | + } |
803 | + } |
804 | + buflen -= mblength; |
805 | + bufpos += mblength; |
806 | + } |
807 | +} |
808 | +#endif |
809 | + |
810 | int |
811 | main (int argc, char **argv) |
812 | { |
813 | @@ -429,7 +583,12 @@ |
814 | |
815 | file_list = (optind < argc ? &argv[optind] : stdin_argv); |
816 | |
817 | - expand (); |
818 | +#if HAVE_MBRTOWC |
819 | + if (MB_CUR_MAX > 1) |
820 | + expand_multibyte (); |
821 | + else |
822 | +#endif |
823 | + expand (); |
824 | |
825 | if (have_read_stdin && fclose (stdin) != 0) |
826 | error (EXIT_FAILURE, errno, "-"); |
827 | diff -Naur coreutils-6.9.orig/src/fold.c coreutils-6.9/src/fold.c |
828 | --- coreutils-6.9.orig/src/fold.c 2007-03-18 21:36:43.000000000 +0000 |
829 | +++ coreutils-6.9/src/fold.c 2007-04-07 16:59:55.000000000 +0000 |
830 | @@ -23,11 +23,33 @@ |
831 | #include <getopt.h> |
832 | #include <sys/types.h> |
833 | |
834 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ |
835 | +#if HAVE_WCHAR_H |
836 | +# include <wchar.h> |
837 | +#endif |
838 | + |
839 | +/* Get iswprint(), iswblank(), wcwidth(). */ |
840 | +#if HAVE_WCTYPE_H |
841 | +# include <wctype.h> |
842 | +#endif |
843 | + |
844 | #include "system.h" |
845 | #include "error.h" |
846 | #include "quote.h" |
847 | #include "xstrtol.h" |
848 | |
849 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
850 | + installation; work around this configuration error. */ |
851 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
852 | +# undef MB_LEN_MAX |
853 | +# define MB_LEN_MAX 16 |
854 | +#endif |
855 | + |
856 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
857 | +#if HAVE_MBRTOWC && defined mbstate_t |
858 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
859 | +#endif |
860 | + |
861 | #define TAB_WIDTH 8 |
862 | |
863 | /* The official name of this program (e.g., no `g' prefix). */ |
864 | @@ -35,23 +57,44 @@ |
865 | |
866 | #define AUTHORS "David MacKenzie" |
867 | |
868 | +#define FATAL_ERROR(Message) \ |
869 | + do \ |
870 | + { \ |
871 | + error (0, 0, (Message)); \ |
872 | + usage (2); \ |
873 | + } \ |
874 | + while (0) |
875 | + |
876 | +enum operating_mode |
877 | +{ |
878 | + /* Fold texts by columns that are at the given positions. */ |
879 | + column_mode, |
880 | + |
881 | + /* Fold texts by bytes that are at the given positions. */ |
882 | + byte_mode, |
883 | + |
884 | + /* Fold texts by characters that are at the given positions. */ |
885 | + character_mode, |
886 | +}; |
887 | + |
888 | /* The name this program was run with. */ |
889 | char *program_name; |
890 | |
891 | +/* The argument shows current mode. (Default: column_mode) */ |
892 | +static enum operating_mode operating_mode; |
893 | + |
894 | /* If nonzero, try to break on whitespace. */ |
895 | static bool break_spaces; |
896 | |
897 | -/* If nonzero, count bytes, not column positions. */ |
898 | -static bool count_bytes; |
899 | - |
900 | /* If nonzero, at least one of the files we read was standard input. */ |
901 | static bool have_read_stdin; |
902 | |
903 | -static char const shortopts[] = "bsw:0::1::2::3::4::5::6::7::8::9::"; |
904 | +static char const shortopts[] = "bcsw:0::1::2::3::4::5::6::7::8::9::"; |
905 | |
906 | static struct option const longopts[] = |
907 | { |
908 | {"bytes", no_argument, NULL, 'b'}, |
909 | + {"characters", no_argument, NULL, 'c'}, |
910 | {"spaces", no_argument, NULL, 's'}, |
911 | {"width", required_argument, NULL, 'w'}, |
912 | {GETOPT_HELP_OPTION_DECL}, |
913 | @@ -81,6 +124,7 @@ |
914 | "), stdout); |
915 | fputs (_("\ |
916 | -b, --bytes count bytes rather than columns\n\ |
917 | + -c, --characters count characters rather than columns\n\ |
918 | -s, --spaces break at spaces\n\ |
919 | -w, --width=WIDTH use WIDTH columns instead of 80\n\ |
920 | "), stdout); |
921 | @@ -98,7 +142,7 @@ |
922 | static size_t |
923 | adjust_column (size_t column, char c) |
924 | { |
925 | - if (!count_bytes) |
926 | + if (operating_mode != byte_mode) |
927 | { |
928 | if (c == '\b') |
929 | { |
930 | @@ -117,34 +161,14 @@ |
931 | return column; |
932 | } |
933 | |
934 | -/* Fold file FILENAME, or standard input if FILENAME is "-", |
935 | - to stdout, with maximum line length WIDTH. |
936 | - Return true if successful. */ |
937 | - |
938 | -static bool |
939 | -fold_file (char const *filename, size_t width) |
940 | +static void |
941 | +fold_text (FILE *istream, size_t width, int *saved_errno) |
942 | { |
943 | - FILE *istream; |
944 | int c; |
945 | size_t column = 0; /* Screen column where next char will go. */ |
946 | size_t offset_out = 0; /* Index in `line_out' for next char. */ |
947 | static char *line_out = NULL; |
948 | static size_t allocated_out = 0; |
949 | - int saved_errno; |
950 | - |
951 | - if (STREQ (filename, "-")) |
952 | - { |
953 | - istream = stdin; |
954 | - have_read_stdin = true; |
955 | - } |
956 | - else |
957 | - istream = fopen (filename, "r"); |
958 | - |
959 | - if (istream == NULL) |
960 | - { |
961 | - error (0, errno, "%s", filename); |
962 | - return false; |
963 | - } |
964 | |
965 | while ((c = getc (istream)) != EOF) |
966 | { |
967 | @@ -172,6 +196,15 @@ |
968 | bool found_blank = false; |
969 | size_t logical_end = offset_out; |
970 | |
971 | + /* If LINE_OUT has no wide character, |
972 | + put a new wide character in LINE_OUT |
973 | + if column is bigger than width. */ |
974 | + if (offset_out == 0) |
975 | + { |
976 | + line_out[offset_out++] = c; |
977 | + continue; |
978 | + } |
979 | + |
980 | /* Look for the last blank. */ |
981 | while (logical_end) |
982 | { |
983 | @@ -218,11 +251,225 @@ |
984 | line_out[offset_out++] = c; |
985 | } |
986 | |
987 | - saved_errno = errno; |
988 | + *saved_errno = errno; |
989 | |
990 | if (offset_out) |
991 | fwrite (line_out, sizeof (char), (size_t) offset_out, stdout); |
992 | |
993 | + free(line_out); |
994 | +} |
995 | + |
996 | +#if HAVE_MBRTOWC |
997 | +static void |
998 | +fold_multibyte_text (FILE *istream, int width, int *saved_errno) |
999 | +{ |
1000 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
1001 | + size_t buflen = 0; /* The length of the byte sequence in buf. */ |
1002 | + char *bufpos; /* Next read position of BUF. */ |
1003 | + wint_t wc; /* A gotten wide character. */ |
1004 | + size_t mblength; /* The byte size of a multibyte character which shows |
1005 | + as same character as WC. */ |
1006 | + mbstate_t state, state_bak; /* State of the stream. */ |
1007 | + int convfail; /* 1, when conversion is failed. Otherwise 0. */ |
1008 | + |
1009 | + char *line_out = NULL; |
1010 | + size_t offset_out = 0; /* Index in `line_out' for next char. */ |
1011 | + size_t allocated_out = 0; |
1012 | + |
1013 | + int increment; |
1014 | + size_t column = 0; |
1015 | + |
1016 | + size_t last_blank_pos; |
1017 | + size_t last_blank_column; |
1018 | + int is_blank_seen; |
1019 | + int last_blank_increment; |
1020 | + int is_bs_following_last_blank; |
1021 | + size_t bs_following_last_blank_num; |
1022 | + int is_cr_after_last_blank; |
1023 | + |
1024 | +#define CLEAR_FLAGS \ |
1025 | + do \ |
1026 | + { \ |
1027 | + last_blank_pos = 0; \ |
1028 | + last_blank_column = 0; \ |
1029 | + is_blank_seen = 0; \ |
1030 | + is_bs_following_last_blank = 0; \ |
1031 | + bs_following_last_blank_num = 0; \ |
1032 | + is_cr_after_last_blank = 0; \ |
1033 | + } \ |
1034 | + while (0) |
1035 | + |
1036 | +#define START_NEW_LINE \ |
1037 | + do \ |
1038 | + { \ |
1039 | + putchar ('\n'); \ |
1040 | + column = 0; \ |
1041 | + offset_out = 0; \ |
1042 | + CLEAR_FLAGS; \ |
1043 | + } \ |
1044 | + while (0) |
1045 | + |
1046 | + CLEAR_FLAGS; |
1047 | + memset (&state, '\0', sizeof(mbstate_t)); |
1048 | + |
1049 | + for (;; bufpos += mblength, buflen -= mblength) |
1050 | + { |
1051 | + if (buflen < MB_LEN_MAX && !feof (istream) && !ferror (istream)) |
1052 | + { |
1053 | + memmove (buf, bufpos, buflen); |
1054 | + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, istream); |
1055 | + bufpos = buf; |
1056 | + } |
1057 | + |
1058 | + if (buflen < 1) |
1059 | + break; |
1060 | + |
1061 | + /* Get a wide character. */ |
1062 | + convfail = 0; |
1063 | + state_bak = state; |
1064 | + mblength = mbrtowc ((wchar_t *)&wc, bufpos, buflen, &state); |
1065 | + |
1066 | + switch (mblength) |
1067 | + { |
1068 | + case (size_t)-1: |
1069 | + case (size_t)-2: |
1070 | + convfail++; |
1071 | + state = state_bak; |
1072 | + /* Fall through. */ |
1073 | + |
1074 | + case 0: |
1075 | + mblength = 1; |
1076 | + break; |
1077 | + } |
1078 | + |
1079 | +rescan: |
1080 | + if (operating_mode == byte_mode) /* byte mode */ |
1081 | + increment = mblength; |
1082 | + else if (operating_mode == character_mode) /* character mode */ |
1083 | + increment = 1; |
1084 | + else /* column mode */ |
1085 | + { |
1086 | + if (convfail) |
1087 | + increment = 1; |
1088 | + else |
1089 | + { |
1090 | + switch (wc) |
1091 | + { |
1092 | + case L'\n': |
1093 | + fwrite (line_out, sizeof(char), offset_out, stdout); |
1094 | + START_NEW_LINE; |
1095 | + continue; |
1096 | + |
1097 | + case L'\b': |
1098 | + increment = (column > 0) ? -1 : 0; |
1099 | + break; |
1100 | + |
1101 | + case L'\r': |
1102 | + increment = -1 * column; |
1103 | + break; |
1104 | + |
1105 | + case L'\t': |
1106 | + increment = 8 - column % 8; |
1107 | + break; |
1108 | + |
1109 | + default: |
1110 | + increment = wcwidth (wc); |
1111 | + increment = (increment < 0) ? 0 : increment; |
1112 | + } |
1113 | + } |
1114 | + } |
1115 | + |
1116 | + if (column + increment > width && break_spaces && last_blank_pos) |
1117 | + { |
1118 | + fwrite (line_out, sizeof(char), last_blank_pos, stdout); |
1119 | + putchar ('\n'); |
1120 | + |
1121 | + offset_out = offset_out - last_blank_pos; |
1122 | + column = column - last_blank_column + ((is_cr_after_last_blank) |
1123 | + ? last_blank_increment : bs_following_last_blank_num); |
1124 | + memmove (line_out, line_out + last_blank_pos, offset_out); |
1125 | + CLEAR_FLAGS; |
1126 | + goto rescan; |
1127 | + } |
1128 | + |
1129 | + if (column + increment > width && column != 0) |
1130 | + { |
1131 | + fwrite (line_out, sizeof(char), offset_out, stdout); |
1132 | + START_NEW_LINE; |
1133 | + goto rescan; |
1134 | + } |
1135 | + |
1136 | + if (allocated_out < offset_out + mblength) |
1137 | + { |
1138 | + allocated_out += 1024; |
1139 | + line_out = xrealloc (line_out, allocated_out); |
1140 | + } |
1141 | + |
1142 | + memcpy (line_out + offset_out, bufpos, mblength); |
1143 | + offset_out += mblength; |
1144 | + column += increment; |
1145 | + |
1146 | + if (is_blank_seen && !convfail && wc == L'\r') |
1147 | + is_cr_after_last_blank = 1; |
1148 | + |
1149 | + if (is_bs_following_last_blank && !convfail && wc == L'\b') |
1150 | + ++bs_following_last_blank_num; |
1151 | + else |
1152 | + is_bs_following_last_blank = 0; |
1153 | + |
1154 | + if (break_spaces && !convfail && iswblank (wc)) |
1155 | + { |
1156 | + last_blank_pos = offset_out; |
1157 | + last_blank_column = column; |
1158 | + is_blank_seen = 1; |
1159 | + last_blank_increment = increment; |
1160 | + is_bs_following_last_blank = 1; |
1161 | + bs_following_last_blank_num = 0; |
1162 | + is_cr_after_last_blank = 0; |
1163 | + } |
1164 | + } |
1165 | + |
1166 | + *saved_errno = errno; |
1167 | + |
1168 | + if (offset_out) |
1169 | + fwrite (line_out, sizeof (char), (size_t) offset_out, stdout); |
1170 | + |
1171 | + free(line_out); |
1172 | +} |
1173 | +#endif |
1174 | + |
1175 | +/* Fold file FILENAME, or standard input if FILENAME is "-", |
1176 | + to stdout, with maximum line length WIDTH. |
1177 | + Return 0 if successful, 1 if an error occurs. */ |
1178 | + |
1179 | +static int |
1180 | +fold_file (char *filename, int width) |
1181 | +{ |
1182 | + FILE *istream; |
1183 | + int saved_errno; |
1184 | + |
1185 | + if (STREQ (filename, "-")) |
1186 | + { |
1187 | + istream = stdin; |
1188 | + have_read_stdin = 1; |
1189 | + } |
1190 | + else |
1191 | + istream = fopen (filename, "r"); |
1192 | + |
1193 | + if (istream == NULL) |
1194 | + { |
1195 | + error (0, errno, "%s", filename); |
1196 | + return 1; |
1197 | + } |
1198 | + |
1199 | + /* Define how ISTREAM is being folded. */ |
1200 | +#if HAVE_MBRTOWC |
1201 | + if (MB_CUR_MAX > 1) |
1202 | + fold_multibyte_text (istream, width, &saved_errno); |
1203 | + else |
1204 | +#endif |
1205 | + fold_text (istream, width, &saved_errno); |
1206 | + |
1207 | if (ferror (istream)) |
1208 | { |
1209 | error (0, saved_errno, "%s", filename); |
1210 | @@ -255,7 +502,8 @@ |
1211 | |
1212 | atexit (close_stdout); |
1213 | |
1214 | - break_spaces = count_bytes = have_read_stdin = false; |
1215 | + operating_mode = column_mode; |
1216 | + break_spaces = have_read_stdin = false; |
1217 | |
1218 | while ((optc = getopt_long (argc, argv, shortopts, longopts, NULL)) != -1) |
1219 | { |
1220 | @@ -264,7 +512,15 @@ |
1221 | switch (optc) |
1222 | { |
1223 | case 'b': /* Count bytes rather than columns. */ |
1224 | - count_bytes = true; |
1225 | + if (operating_mode != column_mode) |
1226 | + FATAL_ERROR (_("only one way of folding may be specified")); |
1227 | + operating_mode = byte_mode; |
1228 | + break; |
1229 | + |
1230 | + case 'c': |
1231 | + if (operating_mode != column_mode) |
1232 | + FATAL_ERROR (_("only one way of folding may be specified")); |
1233 | + operating_mode = character_mode; |
1234 | break; |
1235 | |
1236 | case 's': /* Break at word boundaries. */ |
1237 | diff -Naur coreutils-6.9.orig/src/join.c coreutils-6.9/src/join.c |
1238 | --- coreutils-6.9.orig/src/join.c 2007-03-18 21:36:43.000000000 +0000 |
1239 | +++ coreutils-6.9/src/join.c 2007-04-07 16:59:55.000000000 +0000 |
1240 | @@ -23,16 +23,30 @@ |
1241 | #include <sys/types.h> |
1242 | #include <getopt.h> |
1243 | |
1244 | +/* Get mbstate_t, mbrtowc(), mbrtowc(), wcwidth(). */ |
1245 | +#if HAVE_WCHAR_H |
1246 | +# include <wchar.h> |
1247 | +#endif |
1248 | + |
1249 | +/* Get iswblank(), towupper. */ |
1250 | +#if HAVE_WCTYPE_H |
1251 | +# include <wctype.h> |
1252 | +#endif |
1253 | + |
1254 | #include "system.h" |
1255 | #include "error.h" |
1256 | #include "hard-locale.h" |
1257 | #include "linebuffer.h" |
1258 | -#include "memcasecmp.h" |
1259 | #include "quote.h" |
1260 | #include "stdio--.h" |
1261 | #include "xmemcoll.h" |
1262 | #include "xstrtol.h" |
1263 | |
1264 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
1265 | +#if HAVE_MBRTOWC && defined mbstate_t |
1266 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
1267 | +#endif |
1268 | + |
1269 | /* The official name of this program (e.g., no `g' prefix). */ |
1270 | #define PROGRAM_NAME "join" |
1271 | |
1272 | @@ -104,10 +118,12 @@ |
1273 | /* Last element in `outlist', where a new element can be added. */ |
1274 | static struct outlist *outlist_end = &outlist_head; |
1275 | |
1276 | -/* Tab character separating fields. If negative, fields are separated |
1277 | - by any nonempty string of blanks, otherwise by exactly one |
1278 | - tab character whose value (when cast to unsigned char) equals TAB. */ |
1279 | -static int tab = -1; |
1280 | +/* Tab character separating fields. If NULL, fields are separated |
1281 | + by any nonempty string of blanks. */ |
1282 | +static char *tab = NULL; |
1283 | + |
1284 | +/* The number of bytes used for tab. */ |
1285 | +static size_t tablen = 0; |
1286 | |
1287 | static struct option const longopts[] = |
1288 | { |
1289 | @@ -190,6 +206,8 @@ |
1290 | |
1291 | /* Fill in the `fields' structure in LINE. */ |
1292 | |
1293 | +/* Fill in the `fields' structure in LINE. */ |
1294 | + |
1295 | static void |
1296 | xfields (struct line *line) |
1297 | { |
1298 | @@ -199,10 +217,11 @@ |
1299 | if (ptr == lim) |
1300 | return; |
1301 | |
1302 | - if (0 <= tab) |
1303 | + if (tab != NULL) |
1304 | { |
1305 | + unsigned char t = tab[0]; |
1306 | char *sep; |
1307 | - for (; (sep = memchr (ptr, tab, lim - ptr)) != NULL; ptr = sep + 1) |
1308 | + for (; (sep = memchr (ptr, t, lim - ptr)) != NULL; ptr = sep + 1) |
1309 | extract_field (line, ptr, sep - ptr); |
1310 | } |
1311 | else |
1312 | @@ -229,6 +248,148 @@ |
1313 | extract_field (line, ptr, lim - ptr); |
1314 | } |
1315 | |
1316 | +#if HAVE_MBRTOWC |
1317 | +static void |
1318 | +xfields_multibyte (struct line *line) |
1319 | +{ |
1320 | + char *ptr = line->buf.buffer; |
1321 | + char const *lim = ptr + line->buf.length - 1; |
1322 | + wchar_t wc = 0; |
1323 | + size_t mblength = 1; |
1324 | + mbstate_t state, state_bak; |
1325 | + |
1326 | + memset (&state, 0, sizeof (mbstate_t)); |
1327 | + |
1328 | + if (ptr == lim) |
1329 | + return; |
1330 | + |
1331 | + if (tab != NULL) |
1332 | + { |
1333 | + unsigned char t = tab[0]; |
1334 | + char *sep = ptr; |
1335 | + for (; ptr < lim; ptr = sep + mblength) |
1336 | + { |
1337 | + sep = ptr; |
1338 | + while (sep < lim) |
1339 | + { |
1340 | + state_bak = state; |
1341 | + mblength = mbrtowc (&wc, sep, lim - sep + 1, &state); |
1342 | + |
1343 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
1344 | + { |
1345 | + mblength = 1; |
1346 | + state = state_bak; |
1347 | + } |
1348 | + mblength = (mblength < 1) ? 1 : mblength; |
1349 | + |
1350 | + if (mblength == tablen && !memcmp (sep, tab, mblength)) |
1351 | + break; |
1352 | + else |
1353 | + { |
1354 | + sep += mblength; |
1355 | + continue; |
1356 | + } |
1357 | + } |
1358 | + |
1359 | + if (sep == lim) |
1360 | + break; |
1361 | + |
1362 | + extract_field (line, ptr, sep - ptr); |
1363 | + } |
1364 | + } |
1365 | + else |
1366 | + { |
1367 | + /* Skip leading blanks before the first field. */ |
1368 | + while(ptr < lim) |
1369 | + { |
1370 | + state_bak = state; |
1371 | + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); |
1372 | + |
1373 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
1374 | + { |
1375 | + mblength = 1; |
1376 | + state = state_bak; |
1377 | + break; |
1378 | + } |
1379 | + mblength = (mblength < 1) ? 1 : mblength; |
1380 | + |
1381 | + if (!iswblank(wc)) |
1382 | + break; |
1383 | + ptr += mblength; |
1384 | + } |
1385 | + |
1386 | + do |
1387 | + { |
1388 | + char *sep; |
1389 | + state_bak = state; |
1390 | + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); |
1391 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
1392 | + { |
1393 | + mblength = 1; |
1394 | + state = state_bak; |
1395 | + break; |
1396 | + } |
1397 | + mblength = (mblength < 1) ? 1 : mblength; |
1398 | + |
1399 | + sep = ptr + mblength; |
1400 | + while (sep != lim) |
1401 | + { |
1402 | + state_bak = state; |
1403 | + mblength = mbrtowc (&wc, sep, lim - sep + 1, &state); |
1404 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
1405 | + { |
1406 | + mblength = 1; |
1407 | + state = state_bak; |
1408 | + break; |
1409 | + } |
1410 | + mblength = (mblength < 1) ? 1 : mblength; |
1411 | + |
1412 | + if (iswblank (wc)) |
1413 | + break; |
1414 | + |
1415 | + sep += mblength; |
1416 | + } |
1417 | + |
1418 | + extract_field (line, ptr, sep - ptr); |
1419 | + if (sep == lim) |
1420 | + return; |
1421 | + |
1422 | + state_bak = state; |
1423 | + mblength = mbrtowc (&wc, sep, lim - sep + 1, &state); |
1424 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
1425 | + { |
1426 | + mblength = 1; |
1427 | + state = state_bak; |
1428 | + break; |
1429 | + } |
1430 | + mblength = (mblength < 1) ? 1 : mblength; |
1431 | + |
1432 | + ptr = sep + mblength; |
1433 | + while (ptr != lim) |
1434 | + { |
1435 | + state_bak = state; |
1436 | + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); |
1437 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
1438 | + { |
1439 | + mblength = 1; |
1440 | + state = state_bak; |
1441 | + break; |
1442 | + } |
1443 | + mblength = (mblength < 1) ? 1 : mblength; |
1444 | + |
1445 | + if (!iswblank (wc)) |
1446 | + break; |
1447 | + |
1448 | + ptr += mblength; |
1449 | + } |
1450 | + } |
1451 | + while (ptr != lim); |
1452 | + } |
1453 | + |
1454 | + extract_field (line, ptr, lim - ptr); |
1455 | +} |
1456 | +#endif |
1457 | + |
1458 | /* Read a line from FP into LINE and split it into fields. |
1459 | Return true if successful. */ |
1460 | |
1461 | @@ -249,6 +410,11 @@ |
1462 | line->nfields_allocated = 0; |
1463 | line->nfields = 0; |
1464 | line->fields = NULL; |
1465 | +#if HAVE_MBRTOWC |
1466 | + if (MB_CUR_MAX > 1) |
1467 | + xfields_multibyte (line); |
1468 | + else |
1469 | +#endif |
1470 | xfields (line); |
1471 | return true; |
1472 | } |
1473 | @@ -303,56 +469,114 @@ |
1474 | keycmp (struct line const *line1, struct line const *line2) |
1475 | { |
1476 | /* Start of field to compare in each file. */ |
1477 | - char *beg1; |
1478 | - char *beg2; |
1479 | - |
1480 | - size_t len1; |
1481 | - size_t len2; /* Length of fields to compare. */ |
1482 | + char *beg[2]; |
1483 | + char *copy[2]; |
1484 | + size_t len[2]; /* Length of fields to compare. */ |
1485 | int diff; |
1486 | + int i, j; |
1487 | |
1488 | if (join_field_1 < line1->nfields) |
1489 | { |
1490 | - beg1 = line1->fields[join_field_1].beg; |
1491 | - len1 = line1->fields[join_field_1].len; |
1492 | + beg[0] = line1->fields[join_field_1].beg; |
1493 | + len[0] = line1->fields[join_field_1].len; |
1494 | } |
1495 | else |
1496 | { |
1497 | - beg1 = NULL; |
1498 | - len1 = 0; |
1499 | + beg[0] = NULL; |
1500 | + len[0] = 0; |
1501 | } |
1502 | |
1503 | if (join_field_2 < line2->nfields) |
1504 | { |
1505 | - beg2 = line2->fields[join_field_2].beg; |
1506 | - len2 = line2->fields[join_field_2].len; |
1507 | + beg[1] = line2->fields[join_field_2].beg; |
1508 | + len[1] = line2->fields[join_field_2].len; |
1509 | } |
1510 | else |
1511 | { |
1512 | - beg2 = NULL; |
1513 | - len2 = 0; |
1514 | + beg[1] = NULL; |
1515 | + len[1] = 0; |
1516 | } |
1517 | |
1518 | - if (len1 == 0) |
1519 | - return len2 == 0 ? 0 : -1; |
1520 | - if (len2 == 0) |
1521 | + if (len[0] == 0) |
1522 | + return len[1] == 0 ? 0 : -1; |
1523 | + if (len[1] == 0) |
1524 | return 1; |
1525 | |
1526 | if (ignore_case) |
1527 | { |
1528 | - /* FIXME: ignore_case does not work with NLS (in particular, |
1529 | - with multibyte chars). */ |
1530 | - diff = memcasecmp (beg1, beg2, MIN (len1, len2)); |
1531 | +#ifdef HAVE_MBRTOWC |
1532 | + if (MB_CUR_MAX > 1) |
1533 | + { |
1534 | + size_t mblength; |
1535 | + wchar_t wc, uwc; |
1536 | + mbstate_t state, state_bak; |
1537 | + |
1538 | + memset (&state, '\0', sizeof (mbstate_t)); |
1539 | + |
1540 | + for (i = 0; i < 2; i++) |
1541 | + { |
1542 | + copy[i] = alloca (len[i] + 1); |
1543 | + |
1544 | + for (j = 0; j < MIN (len[0], len[1]);) |
1545 | + { |
1546 | + state_bak = state; |
1547 | + mblength = mbrtowc (&wc, beg[i] + j, len[i] - j, &state); |
1548 | + |
1549 | + switch (mblength) |
1550 | + { |
1551 | + case (size_t) -1: |
1552 | + case (size_t) -2: |
1553 | + state = state_bak; |
1554 | + /* Fall through */ |
1555 | + case 0: |
1556 | + mblength = 1; |
1557 | + break; |
1558 | + |
1559 | + default: |
1560 | + uwc = towupper (wc); |
1561 | + |
1562 | + if (uwc != wc) |
1563 | + { |
1564 | + mbstate_t state_wc; |
1565 | + |
1566 | + memset (&state_wc, '\0', sizeof (mbstate_t)); |
1567 | + wcrtomb (copy[i] + j, uwc, &state_wc); |
1568 | + } |
1569 | + else |
1570 | + memcpy (copy[i] + j, beg[i] + j, mblength); |
1571 | + } |
1572 | + j += mblength; |
1573 | + } |
1574 | + copy[i][j] = '\0'; |
1575 | + } |
1576 | + } |
1577 | + else |
1578 | +#endif |
1579 | + { |
1580 | + for (i = 0; i < 2; i++) |
1581 | + { |
1582 | + copy[i] = alloca (len[i] + 1); |
1583 | + |
1584 | + for (j = 0; j < MIN (len[0], len[1]); j++) |
1585 | + copy[i][j] = toupper (beg[i][j]); |
1586 | + |
1587 | + copy[i][j] = '\0'; |
1588 | + } |
1589 | + } |
1590 | } |
1591 | else |
1592 | { |
1593 | - if (hard_LC_COLLATE) |
1594 | - return xmemcoll (beg1, len1, beg2, len2); |
1595 | - diff = memcmp (beg1, beg2, MIN (len1, len2)); |
1596 | + copy[0] = (unsigned char *) beg[0]; |
1597 | + copy[1] = (unsigned char *) beg[1]; |
1598 | } |
1599 | |
1600 | + if (hard_LC_COLLATE) |
1601 | + return xmemcoll ((char *) copy[0], len[0], (char *) copy[1], len[1]); |
1602 | + diff = memcmp (copy[0], copy[1], MIN (len[0], len[1])); |
1603 | + |
1604 | if (diff) |
1605 | return diff; |
1606 | - return len1 < len2 ? -1 : len1 != len2; |
1607 | + return len[0] - len[1]; |
1608 | } |
1609 | |
1610 | /* Print field N of LINE if it exists and is nonempty, otherwise |
1611 | @@ -377,11 +601,18 @@ |
1612 | |
1613 | /* Print the join of LINE1 and LINE2. */ |
1614 | |
1615 | +#define PUT_TAB_CHAR \ |
1616 | + do \ |
1617 | + { \ |
1618 | + (tab != NULL) ? \ |
1619 | + fwrite(tab, sizeof(char), tablen, stdout) : putchar (' '); \ |
1620 | + } \ |
1621 | + while (0) |
1622 | + |
1623 | static void |
1624 | prjoin (struct line const *line1, struct line const *line2) |
1625 | { |
1626 | const struct outlist *outlist; |
1627 | - char output_separator = tab < 0 ? ' ' : tab; |
1628 | |
1629 | outlist = outlist_head.next; |
1630 | if (outlist) |
1631 | @@ -397,12 +628,12 @@ |
1632 | if (o->file == 0) |
1633 | { |
1634 | if (line1 == &uni_blank) |
1635 | - { |
1636 | + { |
1637 | line = line2; |
1638 | field = join_field_2; |
1639 | } |
1640 | else |
1641 | - { |
1642 | + { |
1643 | line = line1; |
1644 | field = join_field_1; |
1645 | } |
1646 | @@ -416,7 +647,7 @@ |
1647 | o = o->next; |
1648 | if (o == NULL) |
1649 | break; |
1650 | - putchar (output_separator); |
1651 | + PUT_TAB_CHAR; |
1652 | } |
1653 | putchar ('\n'); |
1654 | } |
1655 | @@ -434,23 +665,23 @@ |
1656 | prfield (join_field_1, line1); |
1657 | for (i = 0; i < join_field_1 && i < line1->nfields; ++i) |
1658 | { |
1659 | - putchar (output_separator); |
1660 | + PUT_TAB_CHAR; |
1661 | prfield (i, line1); |
1662 | } |
1663 | for (i = join_field_1 + 1; i < line1->nfields; ++i) |
1664 | { |
1665 | - putchar (output_separator); |
1666 | + PUT_TAB_CHAR; |
1667 | prfield (i, line1); |
1668 | } |
1669 | |
1670 | for (i = 0; i < join_field_2 && i < line2->nfields; ++i) |
1671 | { |
1672 | - putchar (output_separator); |
1673 | + PUT_TAB_CHAR; |
1674 | prfield (i, line2); |
1675 | } |
1676 | for (i = join_field_2 + 1; i < line2->nfields; ++i) |
1677 | { |
1678 | - putchar (output_separator); |
1679 | + PUT_TAB_CHAR; |
1680 | prfield (i, line2); |
1681 | } |
1682 | putchar ('\n'); |
1683 | @@ -859,20 +1090,41 @@ |
1684 | |
1685 | case 't': |
1686 | { |
1687 | - unsigned char newtab = optarg[0]; |
1688 | - if (! newtab) |
1689 | + char *newtab; |
1690 | + size_t newtablen; |
1691 | + if (! optarg[0]) |
1692 | error (EXIT_FAILURE, 0, _("empty tab")); |
1693 | - if (optarg[1]) |
1694 | + newtab = xstrdup (optarg); |
1695 | +#if HAVE_MBRTOWC |
1696 | + if (MB_CUR_MAX > 1) |
1697 | + { |
1698 | + mbstate_t state; |
1699 | + |
1700 | + memset (&state, 0, sizeof (mbstate_t)); |
1701 | + newtablen = mbrtowc (NULL, newtab, |
1702 | + strnlen (newtab, MB_LEN_MAX), |
1703 | + &state); |
1704 | + if (newtablen == (size_t) 0 |
1705 | + || newtablen == (size_t) -1 |
1706 | + || newtablen == (size_t) -2) |
1707 | + newtablen = 1; |
1708 | + } |
1709 | + else |
1710 | +#endif |
1711 | + newtablen = 1; |
1712 | + |
1713 | + if (newtablen == 1 && newtab[1]) |
1714 | + { |
1715 | + if (STREQ (newtab, "\\0")) |
1716 | + newtab[0] = '\0'; |
1717 | + } |
1718 | + if (tab != NULL && strcmp (tab, newtab)) |
1719 | { |
1720 | - if (STREQ (optarg, "\\0")) |
1721 | - newtab = '\0'; |
1722 | - else |
1723 | - error (EXIT_FAILURE, 0, _("multi-character tab %s"), |
1724 | - quote (optarg)); |
1725 | + free (newtab); |
1726 | + error (EXIT_FAILURE, 0, _("incompatible tabs")); |
1727 | } |
1728 | - if (0 <= tab && tab != newtab) |
1729 | - error (EXIT_FAILURE, 0, _("incompatible tabs")); |
1730 | tab = newtab; |
1731 | + tablen = newtablen; |
1732 | } |
1733 | break; |
1734 | |
1735 | diff -Naur coreutils-6.9.orig/src/pr.c coreutils-6.9/src/pr.c |
1736 | --- coreutils-6.9.orig/src/pr.c 2007-03-18 21:36:43.000000000 +0000 |
1737 | +++ coreutils-6.9/src/pr.c 2007-04-07 16:59:55.000000000 +0000 |
1738 | @@ -313,6 +313,32 @@ |
1739 | |
1740 | #include <getopt.h> |
1741 | #include <sys/types.h> |
1742 | + |
1743 | +/* Get MB_LEN_MAX. */ |
1744 | +#include <limits.h> |
1745 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
1746 | + installation; work around this configuration error. */ |
1747 | +#if !defined MB_LEN_MAX || MB_LEN_MAX == 1 |
1748 | +# define MB_LEN_MAX 16 |
1749 | +#endif |
1750 | + |
1751 | +/* Get MB_CUR_MAX. */ |
1752 | +#include <stdlib.h> |
1753 | + |
1754 | +/* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */ |
1755 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ |
1756 | +#if HAVE_WCHAR_H |
1757 | +# include <wchar.h> |
1758 | +#endif |
1759 | + |
1760 | +/* Get iswprint(). -- for wcwidth(). */ |
1761 | +#if HAVE_WCTYPE_H |
1762 | +# include <wctype.h> |
1763 | +#endif |
1764 | +#if !defined iswprint && !HAVE_ISWPRINT |
1765 | +# define iswprint(wc) 1 |
1766 | +#endif |
1767 | + |
1768 | #include "system.h" |
1769 | #include "error.h" |
1770 | #include "hard-locale.h" |
1771 | @@ -324,6 +350,18 @@ |
1772 | #include "strftime.h" |
1773 | #include "xstrtol.h" |
1774 | |
1775 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
1776 | +#if HAVE_MBRTOWC && defined mbstate_t |
1777 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
1778 | +#endif |
1779 | + |
1780 | +#ifndef HAVE_DECL_WCWIDTH |
1781 | +"this configure-time declaration test was not run" |
1782 | +#endif |
1783 | +#if !HAVE_DECL_WCWIDTH |
1784 | +extern int wcwidth (); |
1785 | +#endif |
1786 | + |
1787 | /* The official name of this program (e.g., no `g' prefix). */ |
1788 | #define PROGRAM_NAME "pr" |
1789 | |
1790 | @@ -416,7 +454,20 @@ |
1791 | |
1792 | #define NULLCOL (COLUMN *)0 |
1793 | |
1794 | -static int char_to_clump (char c); |
1795 | +/* Funtion pointers to switch functions for single byte locale or for |
1796 | + multibyte locale. If multibyte functions do not exist in your sysytem, |
1797 | + these pointers always point the function for single byte locale. */ |
1798 | +static void (*print_char) (char c); |
1799 | +static int (*char_to_clump) (char c); |
1800 | + |
1801 | +/* Functions for single byte locale. */ |
1802 | +static void print_char_single (char c); |
1803 | +static int char_to_clump_single (char c); |
1804 | + |
1805 | +/* Functions for multibyte locale. */ |
1806 | +static void print_char_multi (char c); |
1807 | +static int char_to_clump_multi (char c); |
1808 | + |
1809 | static bool read_line (COLUMN *p); |
1810 | static bool print_page (void); |
1811 | static bool print_stored (COLUMN *p); |
1812 | @@ -426,6 +477,7 @@ |
1813 | static void pad_across_to (int position); |
1814 | static void add_line_number (COLUMN *p); |
1815 | static void getoptarg (char *arg, char switch_char, char *character, |
1816 | + int *character_length, int *character_width, |
1817 | int *number); |
1818 | void usage (int status); |
1819 | static void print_files (int number_of_files, char **av); |
1820 | @@ -440,7 +492,6 @@ |
1821 | static void pad_down (int lines); |
1822 | static void read_rest_of_line (COLUMN *p); |
1823 | static void skip_read (COLUMN *p, int column_number); |
1824 | -static void print_char (char c); |
1825 | static void cleanup (void); |
1826 | static void print_sep_string (void); |
1827 | static void separator_string (const char *optarg_S); |
1828 | @@ -455,7 +506,7 @@ |
1829 | we store the leftmost columns contiguously in buff. |
1830 | To print a line from buff, get the index of the first character |
1831 | from line_vector[i], and print up to line_vector[i + 1]. */ |
1832 | -static char *buff; |
1833 | +static unsigned char *buff; |
1834 | |
1835 | /* Index of the position in buff where the next character |
1836 | will be stored. */ |
1837 | @@ -559,7 +610,7 @@ |
1838 | static bool untabify_input = false; |
1839 | |
1840 | /* (-e) The input tab character. */ |
1841 | -static char input_tab_char = '\t'; |
1842 | +static char input_tab_char[MB_LEN_MAX] = "\t"; |
1843 | |
1844 | /* (-e) Tabstops are at chars_per_tab, 2*chars_per_tab, 3*chars_per_tab, ... |
1845 | where the leftmost column is 1. */ |
1846 | @@ -569,7 +620,10 @@ |
1847 | static bool tabify_output = false; |
1848 | |
1849 | /* (-i) The output tab character. */ |
1850 | -static char output_tab_char = '\t'; |
1851 | +static char output_tab_char[MB_LEN_MAX] = "\t"; |
1852 | + |
1853 | +/* (-i) The byte length of output tab character. */ |
1854 | +static int output_tab_char_length = 1; |
1855 | |
1856 | /* (-i) The width of the output tab. */ |
1857 | static int chars_per_output_tab = 8; |
1858 | @@ -643,7 +697,13 @@ |
1859 | static bool numbered_lines = false; |
1860 | |
1861 | /* (-n) Character which follows each line number. */ |
1862 | -static char number_separator = '\t'; |
1863 | +static char number_separator[MB_LEN_MAX] = "\t"; |
1864 | + |
1865 | +/* (-n) The byte length of the character which follows each line number. */ |
1866 | +static int number_separator_length = 1; |
1867 | + |
1868 | +/* (-n) The character width of the character which follows each line number. */ |
1869 | +static int number_separator_width = 0; |
1870 | |
1871 | /* (-n) line counting starts with 1st line of input file (not with 1st |
1872 | line of 1st page printed). */ |
1873 | @@ -696,6 +756,7 @@ |
1874 | -a|COLUMN|-m is a `space' and with the -J option a `tab'. */ |
1875 | static char *col_sep_string = ""; |
1876 | static int col_sep_length = 0; |
1877 | +static int col_sep_width = 0; |
1878 | static char *column_separator = " "; |
1879 | static char *line_separator = "\t"; |
1880 | |
1881 | @@ -852,6 +913,13 @@ |
1882 | col_sep_length = (int) strlen (optarg_S); |
1883 | col_sep_string = xmalloc (col_sep_length + 1); |
1884 | strcpy (col_sep_string, optarg_S); |
1885 | + |
1886 | +#if HAVE_MBRTOWC |
1887 | + if (MB_CUR_MAX > 1) |
1888 | + col_sep_width = mbswidth (col_sep_string, 0); |
1889 | + else |
1890 | +#endif |
1891 | + col_sep_width = col_sep_length; |
1892 | } |
1893 | |
1894 | int |
1895 | @@ -877,6 +945,21 @@ |
1896 | |
1897 | atexit (close_stdout); |
1898 | |
1899 | +/* Define which functions are used, the ones for single byte locale or the ones |
1900 | + for multibyte locale. */ |
1901 | +#if HAVE_MBRTOWC |
1902 | + if (MB_CUR_MAX > 1) |
1903 | + { |
1904 | + print_char = print_char_multi; |
1905 | + char_to_clump = char_to_clump_multi; |
1906 | + } |
1907 | + else |
1908 | +#endif |
1909 | + { |
1910 | + print_char = print_char_single; |
1911 | + char_to_clump = char_to_clump_single; |
1912 | + } |
1913 | + |
1914 | n_files = 0; |
1915 | file_names = (argc > 1 |
1916 | ? xmalloc ((argc - 1) * sizeof (char *)) |
1917 | @@ -949,8 +1032,12 @@ |
1918 | break; |
1919 | case 'e': |
1920 | if (optarg) |
1921 | - getoptarg (optarg, 'e', &input_tab_char, |
1922 | - &chars_per_input_tab); |
1923 | + { |
1924 | + int dummy_length, dummy_width; |
1925 | + |
1926 | + getoptarg (optarg, 'e', input_tab_char, &dummy_length, |
1927 | + &dummy_width, &chars_per_input_tab); |
1928 | + } |
1929 | /* Could check tab width > 0. */ |
1930 | untabify_input = true; |
1931 | break; |
1932 | @@ -963,8 +1050,12 @@ |
1933 | break; |
1934 | case 'i': |
1935 | if (optarg) |
1936 | - getoptarg (optarg, 'i', &output_tab_char, |
1937 | - &chars_per_output_tab); |
1938 | + { |
1939 | + int dummy_width; |
1940 | + |
1941 | + getoptarg (optarg, 'i', output_tab_char, &output_tab_char_length, |
1942 | + &dummy_width, &chars_per_output_tab); |
1943 | + } |
1944 | /* Could check tab width > 0. */ |
1945 | tabify_output = true; |
1946 | break; |
1947 | @@ -991,8 +1082,8 @@ |
1948 | case 'n': |
1949 | numbered_lines = true; |
1950 | if (optarg) |
1951 | - getoptarg (optarg, 'n', &number_separator, |
1952 | - &chars_per_number); |
1953 | + getoptarg (optarg, 'n', number_separator, &number_separator_length, |
1954 | + &number_separator_width, &chars_per_number); |
1955 | break; |
1956 | case 'N': |
1957 | skip_count = false; |
1958 | @@ -1031,7 +1122,7 @@ |
1959 | old_s = false; |
1960 | /* Reset an additional input of -s, -S dominates -s */ |
1961 | col_sep_string = ""; |
1962 | - col_sep_length = 0; |
1963 | + col_sep_length = col_sep_width = 0; |
1964 | use_col_separator = true; |
1965 | if (optarg) |
1966 | separator_string (optarg); |
1967 | @@ -1188,10 +1279,45 @@ |
1968 | a number. */ |
1969 | |
1970 | static void |
1971 | -getoptarg (char *arg, char switch_char, char *character, int *number) |
1972 | +getoptarg (char *arg, char switch_char, char *character, int *character_length, |
1973 | + int *character_width, int *number) |
1974 | { |
1975 | if (!ISDIGIT (*arg)) |
1976 | - *character = *arg++; |
1977 | + { |
1978 | +#ifdef HAVE_MBRTOWC |
1979 | + if (MB_CUR_MAX > 1) /* for multibyte locale. */ |
1980 | + { |
1981 | + wchar_t wc; |
1982 | + size_t mblength; |
1983 | + int width; |
1984 | + mbstate_t state = {'\0'}; |
1985 | + |
1986 | + mblength = mbrtowc (&wc, arg, strnlen(arg, MB_LEN_MAX), &state); |
1987 | + |
1988 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
1989 | + { |
1990 | + *character_length = 1; |
1991 | + *character_width = 1; |
1992 | + } |
1993 | + else |
1994 | + { |
1995 | + *character_length = (mblength < 1) ? 1 : mblength; |
1996 | + width = wcwidth (wc); |
1997 | + *character_width = (width < 0) ? 0 : width; |
1998 | + } |
1999 | + |
2000 | + strncpy (character, arg, *character_length); |
2001 | + arg += *character_length; |
2002 | + } |
2003 | + else /* for single byte locale. */ |
2004 | +#endif |
2005 | + { |
2006 | + *character = *arg++; |
2007 | + *character_length = 1; |
2008 | + *character_width = 1; |
2009 | + } |
2010 | + } |
2011 | + |
2012 | if (*arg) |
2013 | { |
2014 | long int tmp_long; |
2015 | @@ -1256,7 +1382,7 @@ |
2016 | else |
2017 | col_sep_string = column_separator; |
2018 | |
2019 | - col_sep_length = 1; |
2020 | + col_sep_length = col_sep_width = 1; |
2021 | use_col_separator = true; |
2022 | } |
2023 | /* It's rather pointless to define a TAB separator with column |
2024 | @@ -1287,11 +1413,11 @@ |
2025 | TAB_WIDTH (chars_per_input_tab, chars_per_number); */ |
2026 | |
2027 | /* Estimate chars_per_text without any margin and keep it constant. */ |
2028 | - if (number_separator == '\t') |
2029 | + if (number_separator[0] == '\t') |
2030 | number_width = chars_per_number + |
2031 | TAB_WIDTH (chars_per_default_tab, chars_per_number); |
2032 | else |
2033 | - number_width = chars_per_number + 1; |
2034 | + number_width = chars_per_number + number_separator_width; |
2035 | |
2036 | /* The number is part of the column width unless we are |
2037 | printing files in parallel. */ |
2038 | @@ -1306,7 +1432,7 @@ |
2039 | } |
2040 | |
2041 | chars_per_column = (chars_per_line - chars_used_by_number - |
2042 | - (columns - 1) * col_sep_length) / columns; |
2043 | + (columns - 1) * col_sep_width) / columns; |
2044 | |
2045 | if (chars_per_column < 1) |
2046 | error (EXIT_FAILURE, 0, _("page width too narrow")); |
2047 | @@ -1431,7 +1557,7 @@ |
2048 | |
2049 | /* Enlarge p->start_position of first column to use the same form of |
2050 | padding_not_printed with all columns. */ |
2051 | - h = h + col_sep_length; |
2052 | + h = h + col_sep_width; |
2053 | |
2054 | /* This loop takes care of all but the rightmost column. */ |
2055 | |
2056 | @@ -1465,7 +1591,7 @@ |
2057 | } |
2058 | else |
2059 | { |
2060 | - h = h_next + col_sep_length; |
2061 | + h = h_next + col_sep_width; |
2062 | h_next = h + chars_per_column; |
2063 | } |
2064 | } |
2065 | @@ -1755,9 +1881,9 @@ |
2066 | align_column (COLUMN *p) |
2067 | { |
2068 | padding_not_printed = p->start_position; |
2069 | - if (padding_not_printed - col_sep_length > 0) |
2070 | + if (padding_not_printed - col_sep_width > 0) |
2071 | { |
2072 | - pad_across_to (padding_not_printed - col_sep_length); |
2073 | + pad_across_to (padding_not_printed - col_sep_width); |
2074 | padding_not_printed = ANYWHERE; |
2075 | } |
2076 | |
2077 | @@ -2028,13 +2154,13 @@ |
2078 | /* May be too generous. */ |
2079 | buff = X2REALLOC (buff, &buff_allocated); |
2080 | } |
2081 | - buff[buff_current++] = c; |
2082 | + buff[buff_current++] = (unsigned char) c; |
2083 | } |
2084 | |
2085 | static void |
2086 | add_line_number (COLUMN *p) |
2087 | { |
2088 | - int i; |
2089 | + int i, j; |
2090 | char *s; |
2091 | int left_cut; |
2092 | |
2093 | @@ -2057,22 +2183,24 @@ |
2094 | /* Tabification is assumed for multiple columns, also for n-separators, |
2095 | but `default n-separator = TAB' hasn't been given priority over |
2096 | equal column_width also specified by POSIX. */ |
2097 | - if (number_separator == '\t') |
2098 | + if (number_separator[0] == '\t') |
2099 | { |
2100 | i = number_width - chars_per_number; |
2101 | while (i-- > 0) |
2102 | (p->char_func) (' '); |
2103 | } |
2104 | else |
2105 | - (p->char_func) (number_separator); |
2106 | + for (j = 0; j < number_separator_length; j++) |
2107 | + (p->char_func) (number_separator[j]); |
2108 | } |
2109 | else |
2110 | /* To comply with POSIX, we avoid any expansion of default TAB |
2111 | separator with a single column output. No column_width requirement |
2112 | has to be considered. */ |
2113 | { |
2114 | - (p->char_func) (number_separator); |
2115 | - if (number_separator == '\t') |
2116 | + for (j = 0; j < number_separator_length; j++) |
2117 | + (p->char_func) (number_separator[j]); |
2118 | + if (number_separator[0] == '\t') |
2119 | output_position = POS_AFTER_TAB (chars_per_output_tab, |
2120 | output_position); |
2121 | } |
2122 | @@ -2233,7 +2361,7 @@ |
2123 | while (goal - h_old > 1 |
2124 | && (h_new = POS_AFTER_TAB (chars_per_output_tab, h_old)) <= goal) |
2125 | { |
2126 | - putchar (output_tab_char); |
2127 | + fwrite (output_tab_char, sizeof(char), output_tab_char_length, stdout); |
2128 | h_old = h_new; |
2129 | } |
2130 | while (++h_old <= goal) |
2131 | @@ -2253,6 +2381,7 @@ |
2132 | { |
2133 | char *s; |
2134 | int l = col_sep_length; |
2135 | + int not_space_flag; |
2136 | |
2137 | s = col_sep_string; |
2138 | |
2139 | @@ -2266,6 +2395,7 @@ |
2140 | { |
2141 | for (; separators_not_printed > 0; --separators_not_printed) |
2142 | { |
2143 | + not_space_flag = 0; |
2144 | while (l-- > 0) |
2145 | { |
2146 | /* 3 types of sep_strings: spaces only, spaces and chars, |
2147 | @@ -2279,12 +2409,15 @@ |
2148 | } |
2149 | else |
2150 | { |
2151 | + not_space_flag = 1; |
2152 | if (spaces_not_printed > 0) |
2153 | print_white_space (); |
2154 | putchar (*s++); |
2155 | - ++output_position; |
2156 | } |
2157 | } |
2158 | + if (not_space_flag) |
2159 | + output_position += col_sep_width; |
2160 | + |
2161 | /* sep_string ends with some spaces */ |
2162 | if (spaces_not_printed > 0) |
2163 | print_white_space (); |
2164 | @@ -2312,7 +2445,7 @@ |
2165 | required number of tabs and spaces. */ |
2166 | |
2167 | static void |
2168 | -print_char (char c) |
2169 | +print_char_single (char c) |
2170 | { |
2171 | if (tabify_output) |
2172 | { |
2173 | @@ -2336,6 +2469,74 @@ |
2174 | putchar (c); |
2175 | } |
2176 | |
2177 | +#ifdef HAVE_MBRTOWC |
2178 | +static void |
2179 | +print_char_multi (char c) |
2180 | +{ |
2181 | + static size_t mbc_pos = 0; |
2182 | + static unsigned char mbc[MB_LEN_MAX] = {'\0'}; |
2183 | + static mbstate_t state = {'\0'}; |
2184 | + mbstate_t state_bak; |
2185 | + wchar_t wc; |
2186 | + size_t mblength; |
2187 | + int width; |
2188 | + |
2189 | + if (tabify_output) |
2190 | + { |
2191 | + state_bak = state; |
2192 | + mbc[mbc_pos++] = (unsigned char)c; |
2193 | + mblength = mbrtowc (&wc, mbc, mbc_pos, &state); |
2194 | + |
2195 | + while (mbc_pos > 0) |
2196 | + { |
2197 | + switch (mblength) |
2198 | + { |
2199 | + case (size_t)-2: |
2200 | + state = state_bak; |
2201 | + return; |
2202 | + |
2203 | + case (size_t)-1: |
2204 | + state = state_bak; |
2205 | + ++output_position; |
2206 | + putchar (mbc[0]); |
2207 | + memmove (mbc, mbc + 1, MB_CUR_MAX - 1); |
2208 | + --mbc_pos; |
2209 | + break; |
2210 | + |
2211 | + case 0: |
2212 | + mblength = 1; |
2213 | + |
2214 | + default: |
2215 | + if (wc == L' ') |
2216 | + { |
2217 | + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); |
2218 | + --mbc_pos; |
2219 | + ++spaces_not_printed; |
2220 | + return; |
2221 | + } |
2222 | + else if (spaces_not_printed > 0) |
2223 | + print_white_space (); |
2224 | + |
2225 | + /* Nonprintables are assumed to have width 0, except L'\b'. */ |
2226 | + if ((width = wcwidth (wc)) < 1) |
2227 | + { |
2228 | + if (wc == L'\b') |
2229 | + --output_position; |
2230 | + } |
2231 | + else |
2232 | + output_position += width; |
2233 | + |
2234 | + fwrite (mbc, sizeof(char), mblength, stdout); |
2235 | + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); |
2236 | + mbc_pos -= mblength; |
2237 | + } |
2238 | + } |
2239 | + return; |
2240 | + } |
2241 | + putchar (c); |
2242 | +} |
2243 | +#endif |
2244 | + |
2245 | /* Skip to page PAGE before printing. |
2246 | PAGE may be larger than total number of pages. */ |
2247 | |
2248 | @@ -2516,9 +2717,9 @@ |
2249 | align_empty_cols = false; |
2250 | } |
2251 | |
2252 | - if (padding_not_printed - col_sep_length > 0) |
2253 | + if (padding_not_printed - col_sep_width > 0) |
2254 | { |
2255 | - pad_across_to (padding_not_printed - col_sep_length); |
2256 | + pad_across_to (padding_not_printed - col_sep_width); |
2257 | padding_not_printed = ANYWHERE; |
2258 | } |
2259 | |
2260 | @@ -2619,9 +2820,9 @@ |
2261 | } |
2262 | } |
2263 | |
2264 | - if (padding_not_printed - col_sep_length > 0) |
2265 | + if (padding_not_printed - col_sep_width > 0) |
2266 | { |
2267 | - pad_across_to (padding_not_printed - col_sep_length); |
2268 | + pad_across_to (padding_not_printed - col_sep_width); |
2269 | padding_not_printed = ANYWHERE; |
2270 | } |
2271 | |
2272 | @@ -2634,8 +2835,8 @@ |
2273 | if (spaces_not_printed == 0) |
2274 | { |
2275 | output_position = p->start_position + end_vector[line]; |
2276 | - if (p->start_position - col_sep_length == chars_per_margin) |
2277 | - output_position -= col_sep_length; |
2278 | + if (p->start_position - col_sep_width == chars_per_margin) |
2279 | + output_position -= col_sep_width; |
2280 | } |
2281 | |
2282 | return true; |
2283 | @@ -2654,7 +2855,7 @@ |
2284 | number of characters is 1.) */ |
2285 | |
2286 | static int |
2287 | -char_to_clump (char c) |
2288 | +char_to_clump_single (char c) |
2289 | { |
2290 | unsigned char uc = c; |
2291 | char *s = clump_buff; |
2292 | @@ -2664,10 +2865,10 @@ |
2293 | int chars; |
2294 | int chars_per_c = 8; |
2295 | |
2296 | - if (c == input_tab_char) |
2297 | + if (c == input_tab_char[0]) |
2298 | chars_per_c = chars_per_input_tab; |
2299 | |
2300 | - if (c == input_tab_char || c == '\t') |
2301 | + if (c == input_tab_char[0] || c == '\t') |
2302 | { |
2303 | width = TAB_WIDTH (chars_per_c, input_position); |
2304 | |
2305 | @@ -2738,6 +2939,154 @@ |
2306 | return chars; |
2307 | } |
2308 | |
2309 | +#ifdef HAVE_MBRTOWC |
2310 | +static int |
2311 | +char_to_clump_multi (char c) |
2312 | +{ |
2313 | + static size_t mbc_pos = 0; |
2314 | + static char mbc[MB_LEN_MAX] = {'\0'}; |
2315 | + static mbstate_t state = {'\0'}; |
2316 | + mbstate_t state_bak; |
2317 | + wchar_t wc; |
2318 | + size_t mblength; |
2319 | + int wc_width; |
2320 | + register int *s = clump_buff; |
2321 | + register int i, j; |
2322 | + char esc_buff[4]; |
2323 | + int width; |
2324 | + int chars; |
2325 | + int chars_per_c = 8; |
2326 | + |
2327 | + state_bak = state; |
2328 | + mbc[mbc_pos++] = c; |
2329 | + mblength = mbrtowc (&wc, mbc, mbc_pos, &state); |
2330 | + |
2331 | + width = 0; |
2332 | + chars = 0; |
2333 | + while (mbc_pos > 0) |
2334 | + { |
2335 | + switch (mblength) |
2336 | + { |
2337 | + case (size_t)-2: |
2338 | + state = state_bak; |
2339 | + return 0; |
2340 | + |
2341 | + case (size_t)-1: |
2342 | + state = state_bak; |
2343 | + mblength = 1; |
2344 | + |
2345 | + if (use_esc_sequence || use_cntrl_prefix) |
2346 | + { |
2347 | + width = +4; |
2348 | + chars = +4; |
2349 | + *s++ = '\\'; |
2350 | + sprintf (esc_buff, "%03o", mbc[0]); |
2351 | + for (i = 0; i <= 2; ++i) |
2352 | + *s++ = (int) esc_buff[i]; |
2353 | + } |
2354 | + else |
2355 | + { |
2356 | + width += 1; |
2357 | + chars += 1; |
2358 | + *s++ = mbc[0]; |
2359 | + } |
2360 | + break; |
2361 | + |
2362 | + case 0: |
2363 | + mblength = 1; |
2364 | + /* Fall through */ |
2365 | + |
2366 | + default: |
2367 | + if (memcmp (mbc, input_tab_char, mblength) == 0) |
2368 | + chars_per_c = chars_per_input_tab; |
2369 | + |
2370 | + if (memcmp (mbc, input_tab_char, mblength) == 0 || c == '\t') |
2371 | + { |
2372 | + int width_inc; |
2373 | + |
2374 | + width_inc = TAB_WIDTH (chars_per_c, input_position); |
2375 | + width += width_inc; |
2376 | + |
2377 | + if (untabify_input) |
2378 | + { |
2379 | + for (i = width_inc; i; --i) |
2380 | + *s++ = ' '; |
2381 | + chars += width_inc; |
2382 | + } |
2383 | + else |
2384 | + { |
2385 | + for (i = 0; i < mblength; i++) |
2386 | + *s++ = mbc[i]; |
2387 | + chars += mblength; |
2388 | + } |
2389 | + } |
2390 | + else if ((wc_width = wcwidth (wc)) < 1) |
2391 | + { |
2392 | + if (use_esc_sequence) |
2393 | + { |
2394 | + for (i = 0; i < mblength; i++) |
2395 | + { |
2396 | + width += 4; |
2397 | + chars += 4; |
2398 | + *s++ = '\\'; |
2399 | + sprintf (esc_buff, "%03o", c); |
2400 | + for (j = 0; j <= 2; ++j) |
2401 | + *s++ = (int) esc_buff[j]; |
2402 | + } |
2403 | + } |
2404 | + else if (use_cntrl_prefix) |
2405 | + { |
2406 | + if (wc < 0200) |
2407 | + { |
2408 | + width += 2; |
2409 | + chars += 2; |
2410 | + *s++ = '^'; |
2411 | + *s++ = wc ^ 0100; |
2412 | + } |
2413 | + else |
2414 | + { |
2415 | + for (i = 0; i < mblength; i++) |
2416 | + { |
2417 | + width += 4; |
2418 | + chars += 4; |
2419 | + *s++ = '\\'; |
2420 | + sprintf (esc_buff, "%03o", c); |
2421 | + for (j = 0; j <= 2; ++j) |
2422 | + *s++ = (int) esc_buff[j]; |
2423 | + } |
2424 | + } |
2425 | + } |
2426 | + else if (wc == L'\b') |
2427 | + { |
2428 | + width += -1; |
2429 | + chars += 1; |
2430 | + *s++ = c; |
2431 | + } |
2432 | + else |
2433 | + { |
2434 | + width += 0; |
2435 | + chars += mblength; |
2436 | + for (i = 0; i < mblength; i++) |
2437 | + *s++ = mbc[i]; |
2438 | + } |
2439 | + } |
2440 | + else |
2441 | + { |
2442 | + width += wc_width; |
2443 | + chars += mblength; |
2444 | + for (i = 0; i < mblength; i++) |
2445 | + *s++ = mbc[i]; |
2446 | + } |
2447 | + } |
2448 | + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); |
2449 | + mbc_pos -= mblength; |
2450 | + } |
2451 | + |
2452 | + input_position += width; |
2453 | + return chars; |
2454 | +} |
2455 | +#endif |
2456 | + |
2457 | /* We've just printed some files and need to clean up things before |
2458 | looking for more options and printing the next batch of files. |
2459 | |
2460 | diff -Naur coreutils-6.9.orig/src/sort.c coreutils-6.9/src/sort.c |
2461 | --- coreutils-6.9.orig/src/sort.c 2007-03-18 21:36:43.000000000 +0000 |
2462 | +++ coreutils-6.9/src/sort.c 2007-04-07 17:11:06.000000000 +0000 |
2463 | @@ -23,10 +23,18 @@ |
2464 | |
2465 | #include <config.h> |
2466 | |
2467 | +#include <assert.h> |
2468 | #include <getopt.h> |
2469 | #include <sys/types.h> |
2470 | #include <sys/wait.h> |
2471 | #include <signal.h> |
2472 | +#if HAVE_WCHAR_H |
2473 | +# include <wchar.h> |
2474 | +#endif |
2475 | +/* Get isw* functions. */ |
2476 | +#if HAVE_WCTYPE_H |
2477 | +# include <wctype.h> |
2478 | +#endif |
2479 | #include "system.h" |
2480 | #include "argmatch.h" |
2481 | #include "error.h" |
2482 | @@ -116,14 +124,38 @@ |
2483 | /* Thousands separator; if -1, then there isn't one. */ |
2484 | static int thousands_sep; |
2485 | |
2486 | +static int force_general_numcompare = 0; |
2487 | + |
2488 | /* Nonzero if the corresponding locales are hard. */ |
2489 | static bool hard_LC_COLLATE; |
2490 | -#if HAVE_NL_LANGINFO |
2491 | +#if HAVE_LANGINFO_CODESET |
2492 | static bool hard_LC_TIME; |
2493 | #endif |
2494 | |
2495 | #define NONZERO(x) ((x) != 0) |
2496 | |
2497 | +/* get a multibyte character's byte length. */ |
2498 | +#define GET_BYTELEN_OF_CHAR(LIM, PTR, MBLENGTH, STATE) \ |
2499 | + do \ |
2500 | + { \ |
2501 | + wchar_t wc; \ |
2502 | + mbstate_t state_bak; \ |
2503 | + \ |
2504 | + state_bak = STATE; \ |
2505 | + mblength = mbrtowc (&wc, PTR, LIM - PTR, &STATE); \ |
2506 | + \ |
2507 | + switch (MBLENGTH) \ |
2508 | + { \ |
2509 | + case (size_t)-1: \ |
2510 | + case (size_t)-2: \ |
2511 | + STATE = state_bak; \ |
2512 | + /* Fall through. */ \ |
2513 | + case 0: \ |
2514 | + MBLENGTH = 1; \ |
2515 | + } \ |
2516 | + } \ |
2517 | + while (0) |
2518 | + |
2519 | /* The kind of blanks for '-b' to skip in various options. */ |
2520 | enum blanktype { bl_start, bl_end, bl_both }; |
2521 | |
2522 | @@ -261,13 +293,11 @@ |
2523 | they were read if all keys compare equal. */ |
2524 | static bool stable; |
2525 | |
2526 | -/* If TAB has this value, blanks separate fields. */ |
2527 | -enum { TAB_DEFAULT = CHAR_MAX + 1 }; |
2528 | - |
2529 | -/* Tab character separating fields. If TAB_DEFAULT, then fields are |
2530 | +/* Tab character separating fields. If tab_length is 0, then fields are |
2531 | separated by the empty string between a non-blank character and a blank |
2532 | character. */ |
2533 | -static int tab = TAB_DEFAULT; |
2534 | +static char tab[MB_LEN_MAX + 1]; |
2535 | +static size_t tab_length = 0; |
2536 | |
2537 | /* Flag to remove consecutive duplicate lines from the output. |
2538 | Only the last of a sequence of equal lines will be output. */ |
2539 | @@ -639,6 +669,44 @@ |
2540 | update_proc (pid); |
2541 | } |
2542 | |
2543 | +/* Function pointers. */ |
2544 | +static void |
2545 | +(*inittables) (void); |
2546 | +static char * |
2547 | +(*begfield) (const struct line*, const struct keyfield *); |
2548 | +static char * |
2549 | +(*limfield) (const struct line*, const struct keyfield *); |
2550 | +static int |
2551 | +(*getmonth) (char const *, size_t); |
2552 | +static int |
2553 | +(*keycompare) (const struct line *, const struct line *); |
2554 | +static int |
2555 | +(*numcompare) (const char *, const char *); |
2556 | + |
2557 | +/* Test for white space multibyte character. |
2558 | + Set LENGTH the byte length of investigated multibyte character. */ |
2559 | +#if HAVE_MBRTOWC |
2560 | +static int |
2561 | +ismbblank (const char *str, size_t len, size_t *length) |
2562 | +{ |
2563 | + size_t mblength; |
2564 | + wchar_t wc; |
2565 | + mbstate_t state; |
2566 | + |
2567 | + memset (&state, '\0', sizeof(mbstate_t)); |
2568 | + mblength = mbrtowc (&wc, str, len, &state); |
2569 | + |
2570 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
2571 | + { |
2572 | + *length = 1; |
2573 | + return 0; |
2574 | + } |
2575 | + |
2576 | + *length = (mblength < 1) ? 1 : mblength; |
2577 | + return iswblank (wc); |
2578 | +} |
2579 | +#endif |
2580 | + |
2581 | /* Clean up any remaining temporary files. */ |
2582 | |
2583 | static void |
2584 | @@ -978,7 +1046,7 @@ |
2585 | free (node); |
2586 | } |
2587 | |
2588 | -#if HAVE_NL_LANGINFO |
2589 | +#if HAVE_LANGINFO_CODESET |
2590 | |
2591 | static int |
2592 | struct_month_cmp (const void *m1, const void *m2) |
2593 | @@ -993,7 +1061,7 @@ |
2594 | /* Initialize the character class tables. */ |
2595 | |
2596 | static void |
2597 | -inittables (void) |
2598 | +inittables_uni (void) |
2599 | { |
2600 | size_t i; |
2601 | |
2602 | @@ -1005,7 +1073,7 @@ |
2603 | fold_toupper[i] = toupper (i); |
2604 | } |
2605 | |
2606 | -#if HAVE_NL_LANGINFO |
2607 | +#if HAVE_LANGINFO_CODESET |
2608 | /* If we're not in the "C" locale, read different names for months. */ |
2609 | if (hard_LC_TIME) |
2610 | { |
2611 | @@ -1031,6 +1099,64 @@ |
2612 | #endif |
2613 | } |
2614 | |
2615 | +#if HAVE_MBRTOWC |
2616 | +static void |
2617 | +inittables_mb (void) |
2618 | +{ |
2619 | + int i, j, k, l; |
2620 | + char *name, *s; |
2621 | + size_t s_len, mblength; |
2622 | + char mbc[MB_LEN_MAX]; |
2623 | + wchar_t wc, pwc; |
2624 | + mbstate_t state_mb, state_wc; |
2625 | + |
2626 | + for (i = 0; i < MONTHS_PER_YEAR; i++) |
2627 | + { |
2628 | + s = (char *) nl_langinfo (ABMON_1 + i); |
2629 | + s_len = strlen (s); |
2630 | + monthtab[i].name = name = (char *) xmalloc (s_len + 1); |
2631 | + monthtab[i].val = i + 1; |
2632 | + |
2633 | + memset (&state_mb, '\0', sizeof (mbstate_t)); |
2634 | + memset (&state_wc, '\0', sizeof (mbstate_t)); |
2635 | + |
2636 | + for (j = 0; j < s_len;) |
2637 | + { |
2638 | + if (!ismbblank (s + j, s_len - j, &mblength)) |
2639 | + break; |
2640 | + j += mblength; |
2641 | + } |
2642 | + |
2643 | + for (k = 0; j < s_len;) |
2644 | + { |
2645 | + mblength = mbrtowc (&wc, (s + j), (s_len - j), &state_mb); |
2646 | + assert (mblength != (size_t)-1 && mblength != (size_t)-2); |
2647 | + if (mblength == 0) |
2648 | + break; |
2649 | + |
2650 | + pwc = towupper (wc); |
2651 | + if (pwc == wc) |
2652 | + { |
2653 | + memcpy (mbc, s + j, mblength); |
2654 | + j += mblength; |
2655 | + } |
2656 | + else |
2657 | + { |
2658 | + j += mblength; |
2659 | + mblength = wcrtomb (mbc, pwc, &state_wc); |
2660 | + assert (mblength != (size_t)0 && mblength != (size_t)-1); |
2661 | + } |
2662 | + |
2663 | + for (l = 0; l < mblength; l++) |
2664 | + name[k++] = mbc[l]; |
2665 | + } |
2666 | + name[k] = '\0'; |
2667 | + } |
2668 | + qsort ((void *) monthtab, MONTHS_PER_YEAR, |
2669 | + sizeof (struct month), struct_month_cmp); |
2670 | +} |
2671 | +#endif |
2672 | + |
2673 | /* Specify the amount of main memory to use when sorting. */ |
2674 | static void |
2675 | specify_sort_size (char const *s) |
2676 | @@ -1241,7 +1367,7 @@ |
2677 | by KEY in LINE. */ |
2678 | |
2679 | static char * |
2680 | -begfield (const struct line *line, const struct keyfield *key) |
2681 | +begfield_uni (const struct line *line, const struct keyfield *key) |
2682 | { |
2683 | char *ptr = line->text, *lim = ptr + line->length - 1; |
2684 | size_t sword = key->sword; |
2685 | @@ -1251,10 +1377,10 @@ |
2686 | /* The leading field separator itself is included in a field when -t |
2687 | is absent. */ |
2688 | |
2689 | - if (tab != TAB_DEFAULT) |
2690 | + if (tab_length) |
2691 | while (ptr < lim && sword--) |
2692 | { |
2693 | - while (ptr < lim && *ptr != tab) |
2694 | + while (ptr < lim && *ptr != tab[0]) |
2695 | ++ptr; |
2696 | if (ptr < lim) |
2697 | ++ptr; |
2698 | @@ -1282,11 +1408,70 @@ |
2699 | return ptr; |
2700 | } |
2701 | |
2702 | +#if HAVE_MBRTOWC |
2703 | +static char * |
2704 | +begfield_mb (const struct line *line, const struct keyfield *key) |
2705 | +{ |
2706 | + int i; |
2707 | + char *ptr = line->text, *lim = ptr + line->length - 1; |
2708 | + size_t sword = key->sword; |
2709 | + size_t schar = key->schar; |
2710 | + size_t mblength; |
2711 | + mbstate_t state; |
2712 | + |
2713 | + memset (&state, '\0', sizeof(mbstate_t)); |
2714 | + |
2715 | + if (tab_length) |
2716 | + while (ptr < lim && sword--) |
2717 | + { |
2718 | + while (ptr < lim && memcmp (ptr, tab, tab_length) != 0) |
2719 | + { |
2720 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2721 | + ptr += mblength; |
2722 | + } |
2723 | + if (ptr < lim) |
2724 | + { |
2725 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2726 | + ptr += mblength; |
2727 | + } |
2728 | + } |
2729 | + else |
2730 | + while (ptr < lim && sword--) |
2731 | + { |
2732 | + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) |
2733 | + ptr += mblength; |
2734 | + if (ptr < lim) |
2735 | + { |
2736 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2737 | + ptr += mblength; |
2738 | + } |
2739 | + while (ptr < lim && !ismbblank (ptr, lim - ptr, &mblength)) |
2740 | + ptr += mblength; |
2741 | + } |
2742 | + |
2743 | + if (key->skipsblanks) |
2744 | + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) |
2745 | + ptr += mblength; |
2746 | + |
2747 | + for (i = 0; i < schar; i++) |
2748 | + { |
2749 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2750 | + |
2751 | + if (ptr + mblength > lim) |
2752 | + break; |
2753 | + else |
2754 | + ptr += mblength; |
2755 | + } |
2756 | + |
2757 | + return ptr; |
2758 | +} |
2759 | +#endif |
2760 | + |
2761 | /* Return the limit of (a pointer to the first character after) the field |
2762 | in LINE specified by KEY. */ |
2763 | |
2764 | static char * |
2765 | -limfield (const struct line *line, const struct keyfield *key) |
2766 | +limfield_uni (const struct line *line, const struct keyfield *key) |
2767 | { |
2768 | char *ptr = line->text, *lim = ptr + line->length - 1; |
2769 | size_t eword = key->eword, echar = key->echar; |
2770 | @@ -1299,10 +1484,10 @@ |
2771 | `beginning' is the first character following the delimiting TAB. |
2772 | Otherwise, leave PTR pointing at the first `blank' character after |
2773 | the preceding field. */ |
2774 | - if (tab != TAB_DEFAULT) |
2775 | + if (tab_length) |
2776 | while (ptr < lim && eword--) |
2777 | { |
2778 | - while (ptr < lim && *ptr != tab) |
2779 | + while (ptr < lim && *ptr != tab[0]) |
2780 | ++ptr; |
2781 | if (ptr < lim && (eword | echar)) |
2782 | ++ptr; |
2783 | @@ -1348,10 +1533,10 @@ |
2784 | */ |
2785 | |
2786 | /* Make LIM point to the end of (one byte past) the current field. */ |
2787 | - if (tab != TAB_DEFAULT) |
2788 | + if (tab_length) |
2789 | { |
2790 | char *newlim; |
2791 | - newlim = memchr (ptr, tab, lim - ptr); |
2792 | + newlim = memchr (ptr, tab[0], lim - ptr); |
2793 | if (newlim) |
2794 | lim = newlim; |
2795 | } |
2796 | @@ -1384,6 +1569,107 @@ |
2797 | return ptr; |
2798 | } |
2799 | |
2800 | +#if HAVE_MBRTOWC |
2801 | +static char * |
2802 | +limfield_mb (const struct line *line, const struct keyfield *key) |
2803 | +{ |
2804 | + char *ptr = line->text, *lim = ptr + line->length - 1; |
2805 | + size_t eword = key->eword, echar = key->echar; |
2806 | + int i; |
2807 | + size_t mblength; |
2808 | + mbstate_t state; |
2809 | + |
2810 | + memset (&state, '\0', sizeof(mbstate_t)); |
2811 | + |
2812 | + if (tab_length) |
2813 | + while (ptr < lim && eword--) |
2814 | + { |
2815 | + while (ptr < lim && memcmp (ptr, tab, tab_length) != 0) |
2816 | + { |
2817 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2818 | + ptr += mblength; |
2819 | + } |
2820 | + if (ptr < lim && (eword | echar)) |
2821 | + { |
2822 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2823 | + ptr += mblength; |
2824 | + } |
2825 | + } |
2826 | + else |
2827 | + while (ptr < lim && eword--) |
2828 | + { |
2829 | + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) |
2830 | + ptr += mblength; |
2831 | + if (ptr < lim) |
2832 | + { |
2833 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2834 | + ptr += mblength; |
2835 | + } |
2836 | + while (ptr < lim && !ismbblank (ptr, lim - ptr, &mblength)) |
2837 | + ptr += mblength; |
2838 | + } |
2839 | + |
2840 | + |
2841 | +# ifdef POSIX_UNSPECIFIED |
2842 | + /* Make LIM point to the end of (one byte past) the current field. */ |
2843 | + if (tab_length) |
2844 | + { |
2845 | + char *newlim, *p; |
2846 | + |
2847 | + newlim = NULL; |
2848 | + for (p = ptr; p < lim;) |
2849 | + { |
2850 | + if (memcmp (p, tab, tab_length) == 0) |
2851 | + { |
2852 | + newlim = p; |
2853 | + break; |
2854 | + } |
2855 | + |
2856 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2857 | + p += mblength; |
2858 | + } |
2859 | + } |
2860 | + else |
2861 | + { |
2862 | + char *newlim; |
2863 | + newlim = ptr; |
2864 | + |
2865 | + while (newlim < lim && ismbblank (newlim, lim - newlim, &mblength)) |
2866 | + newlim += mblength; |
2867 | + if (ptr < lim) |
2868 | + { |
2869 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2870 | + ptr += mblength; |
2871 | + } |
2872 | + while (newlim < lim && !ismbblank (newlim, lim - newlim, &mblength)) |
2873 | + newlim += mblength; |
2874 | + lim = newlim; |
2875 | + } |
2876 | +# endif |
2877 | + |
2878 | + /* If we're skipping leading blanks, don't start counting characters |
2879 | + * until after skipping past any leading blanks. */ |
2880 | + if (key->skipsblanks) |
2881 | + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) |
2882 | + ptr += mblength; |
2883 | + |
2884 | + memset (&state, '\0', sizeof(mbstate_t)); |
2885 | + |
2886 | + /* Advance PTR by ECHAR (if possible), but no further than LIM. */ |
2887 | + for (i = 0; i < echar; i++) |
2888 | + { |
2889 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2890 | + |
2891 | + if (ptr + mblength > lim) |
2892 | + break; |
2893 | + else |
2894 | + ptr += mblength; |
2895 | + } |
2896 | + |
2897 | + return ptr; |
2898 | +} |
2899 | +#endif |
2900 | + |
2901 | /* Fill BUF reading from FP, moving buf->left bytes from the end |
2902 | of buf->buf to the beginning first. If EOF is reached and the |
2903 | file wasn't terminated by a newline, supply one. Set up BUF's line |
2904 | @@ -1500,7 +1786,7 @@ |
2905 | hideously fast. */ |
2906 | |
2907 | static int |
2908 | -numcompare (const char *a, const char *b) |
2909 | +numcompare_uni (const char *a, const char *b) |
2910 | { |
2911 | while (blanks[to_uchar (*a)]) |
2912 | a++; |
2913 | @@ -1510,6 +1796,25 @@ |
2914 | return strnumcmp (a, b, decimal_point, thousands_sep); |
2915 | } |
2916 | |
2917 | +#if HAVE_MBRTOWC |
2918 | +static int |
2919 | +numcompare_mb (const char *a, const char *b) |
2920 | +{ |
2921 | + size_t mblength, len; |
2922 | + len = strlen (a); /* okay for UTF-8 */ |
2923 | + while (*a && ismbblank (a, len > MB_CUR_MAX ? MB_CUR_MAX : len, &mblength)) |
2924 | + { |
2925 | + a += mblength; |
2926 | + len -= mblength; |
2927 | + } |
2928 | + len = strlen (b); /* okay for UTF-8 */ |
2929 | + while (*b && ismbblank (b, len > MB_CUR_MAX ? MB_CUR_MAX : len, &mblength)) |
2930 | + b += mblength; |
2931 | + |
2932 | + return strnumcmp (a, b, decimal_point, thousands_sep); |
2933 | +} |
2934 | +#endif /* HAV_EMBRTOWC */ |
2935 | + |
2936 | static int |
2937 | general_numcompare (const char *sa, const char *sb) |
2938 | { |
2939 | @@ -1543,7 +1848,7 @@ |
2940 | Return 0 if the name in S is not recognized. */ |
2941 | |
2942 | static int |
2943 | -getmonth (char const *month, size_t len) |
2944 | +getmonth_uni (char const *month, size_t len) |
2945 | { |
2946 | size_t lo = 0; |
2947 | size_t hi = MONTHS_PER_YEAR; |
2948 | @@ -1698,11 +2003,79 @@ |
2949 | return diff; |
2950 | } |
2951 | |
2952 | +#if HAVE_MBRTOWC |
2953 | +static int |
2954 | +getmonth_mb (const char *s, size_t len) |
2955 | +{ |
2956 | + char *month; |
2957 | + register size_t i; |
2958 | + register int lo = 0, hi = MONTHS_PER_YEAR, result; |
2959 | + char *tmp; |
2960 | + size_t wclength, mblength; |
2961 | + const char **pp; |
2962 | + const wchar_t **wpp; |
2963 | + wchar_t *month_wcs; |
2964 | + mbstate_t state; |
2965 | + |
2966 | + while (len > 0 && ismbblank (s, len, &mblength)) |
2967 | + { |
2968 | + s += mblength; |
2969 | + len -= mblength; |
2970 | + } |
2971 | + |
2972 | + if (len == 0) |
2973 | + return 0; |
2974 | + |
2975 | + month = (char *) alloca (len + 1); |
2976 | + |
2977 | + tmp = (char *) alloca (len + 1); |
2978 | + memcpy (tmp, s, len); |
2979 | + tmp[len] = '\0'; |
2980 | + pp = (const char **)&tmp; |
2981 | + month_wcs = (wchar_t *) alloca ((len + 1) * sizeof (wchar_t)); |
2982 | + memset (&state, '\0', sizeof(mbstate_t)); |
2983 | + |
2984 | + wclength = mbsrtowcs (month_wcs, pp, len + 1, &state); |
2985 | + assert (wclength != (size_t)-1 && *pp == NULL); |
2986 | + |
2987 | + for (i = 0; i < wclength; i++) |
2988 | + { |
2989 | + month_wcs[i] = towupper(month_wcs[i]); |
2990 | + if (iswblank (month_wcs[i])) |
2991 | + { |
2992 | + month_wcs[i] = L'\0'; |
2993 | + break; |
2994 | + } |
2995 | + } |
2996 | + |
2997 | + wpp = (const wchar_t **)&month_wcs; |
2998 | + |
2999 | + mblength = wcsrtombs (month, wpp, len + 1, &state); |
3000 | + assert (mblength != (-1) && *wpp == NULL); |
3001 | + |
3002 | + do |
3003 | + { |
3004 | + int ix = (lo + hi) / 2; |
3005 | + |
3006 | + if (strncmp (month, monthtab[ix].name, strlen (monthtab[ix].name)) < 0) |
3007 | + hi = ix; |
3008 | + else |
3009 | + lo = ix; |
3010 | + } |
3011 | + while (hi - lo > 1); |
3012 | + |
3013 | + result = (!strncmp (month, monthtab[lo].name, strlen (monthtab[lo].name)) |
3014 | + ? monthtab[lo].val : 0); |
3015 | + |
3016 | + return result; |
3017 | +} |
3018 | +#endif |
3019 | + |
3020 | /* Compare two lines A and B trying every key in sequence until there |
3021 | are no more keys or a difference is found. */ |
3022 | |
3023 | static int |
3024 | -keycompare (const struct line *a, const struct line *b) |
3025 | +keycompare_uni (const struct line *a, const struct line *b) |
3026 | { |
3027 | struct keyfield const *key = keylist; |
3028 | |
3029 | @@ -1875,6 +2248,177 @@ |
3030 | return key->reverse ? -diff : diff; |
3031 | } |
3032 | |
3033 | +#if HAVE_MBRTOWC |
3034 | +static int |
3035 | +keycompare_mb (const struct line *a, const struct line *b) |
3036 | +{ |
3037 | + struct keyfield *key = keylist; |
3038 | + |
3039 | + /* For the first iteration only, the key positions have been |
3040 | + precomputed for us. */ |
3041 | + char *texta = a->keybeg; |
3042 | + char *textb = b->keybeg; |
3043 | + char *lima = a->keylim; |
3044 | + char *limb = b->keylim; |
3045 | + |
3046 | + size_t mblength_a, mblength_b; |
3047 | + wchar_t wc_a, wc_b; |
3048 | + mbstate_t state_a, state_b; |
3049 | + |
3050 | + int diff; |
3051 | + |
3052 | + memset (&state_a, '\0', sizeof(mbstate_t)); |
3053 | + memset (&state_b, '\0', sizeof(mbstate_t)); |
3054 | + |
3055 | + for (;;) |
3056 | + { |
3057 | + unsigned char *translate = (unsigned char *) key->translate; |
3058 | + bool const *ignore = key->ignore; |
3059 | + |
3060 | + /* Find the lengths. */ |
3061 | + size_t lena = lima <= texta ? 0 : lima - texta; |
3062 | + size_t lenb = limb <= textb ? 0 : limb - textb; |
3063 | + |
3064 | + /* Actually compare the fields. */ |
3065 | + if (key->numeric | key->general_numeric) |
3066 | + { |
3067 | + char savea = *lima, saveb = *limb; |
3068 | + |
3069 | + *lima = *limb = '\0'; |
3070 | + if (force_general_numcompare) |
3071 | + diff = general_numcompare (texta, textb); |
3072 | + else |
3073 | + diff = ((key->numeric ? numcompare : general_numcompare) |
3074 | + (texta, textb)); |
3075 | + *lima = savea, *limb = saveb; |
3076 | + } |
3077 | + else if (key->month) |
3078 | + diff = getmonth (texta, lena) - getmonth (textb, lenb); |
3079 | + else |
3080 | + { |
3081 | + if (ignore || translate) |
3082 | + { |
3083 | + char *copy_a = (char *) alloca (lena + 1 + lenb + 1); |
3084 | + char *copy_b = copy_a + lena + 1; |
3085 | + size_t new_len_a, new_len_b; |
3086 | + size_t i, j; |
3087 | + |
3088 | + /* Ignore and/or translate chars before comparing. */ |
3089 | +# define IGNORE_CHARS(NEW_LEN, LEN, TEXT, COPY, WC, MBLENGTH, STATE) \ |
3090 | + do \ |
3091 | + { \ |
3092 | + wchar_t uwc; \ |
3093 | + char mbc[MB_LEN_MAX]; \ |
3094 | + mbstate_t state_wc; \ |
3095 | + \ |
3096 | + for (NEW_LEN = i = 0; i < LEN;) \ |
3097 | + { \ |
3098 | + mbstate_t state_bak; \ |
3099 | + \ |
3100 | + state_bak = STATE; \ |
3101 | + MBLENGTH = mbrtowc (&WC, TEXT + i, LEN - i, &STATE); \ |
3102 | + \ |
3103 | + if (MBLENGTH == (size_t)-2 || MBLENGTH == (size_t)-1 \ |
3104 | + || MBLENGTH == 0) \ |
3105 | + { \ |
3106 | + if (MBLENGTH == (size_t)-2 || MBLENGTH == (size_t)-1) \ |
3107 | + STATE = state_bak; \ |
3108 | + if (!ignore) \ |
3109 | + COPY[NEW_LEN++] = TEXT[i++]; \ |
3110 | + continue; \ |
3111 | + } \ |
3112 | + \ |
3113 | + if (ignore) \ |
3114 | + { \ |
3115 | + if ((ignore == nonprinting && !iswprint (WC)) \ |
3116 | + || (ignore == nondictionary \ |
3117 | + && !iswalnum (WC) && !iswblank (WC))) \ |
3118 | + { \ |
3119 | + i += MBLENGTH; \ |
3120 | + continue; \ |
3121 | + } \ |
3122 | + } \ |
3123 | + \ |
3124 | + if (translate) \ |
3125 | + { \ |
3126 | + \ |
3127 | + uwc = towupper(WC); \ |
3128 | + if (WC == uwc) \ |
3129 | + { \ |
3130 | + memcpy (mbc, TEXT + i, MBLENGTH); \ |
3131 | + i += MBLENGTH; \ |
3132 | + } \ |
3133 | + else \ |
3134 | + { \ |
3135 | + i += MBLENGTH; \ |
3136 | + WC = uwc; \ |
3137 | + memset (&state_wc, '\0', sizeof (mbstate_t)); \ |
3138 | + \ |
3139 | + MBLENGTH = wcrtomb (mbc, WC, &state_wc); \ |
3140 | + assert (MBLENGTH != (size_t)-1 && MBLENGTH != 0); \ |
3141 | + } \ |
3142 | + \ |
3143 | + for (j = 0; j < MBLENGTH; j++) \ |
3144 | + COPY[NEW_LEN++] = mbc[j]; \ |
3145 | + } \ |
3146 | + else \ |
3147 | + for (j = 0; j < MBLENGTH; j++) \ |
3148 | + COPY[NEW_LEN++] = TEXT[i++]; \ |
3149 | + } \ |
3150 | + COPY[NEW_LEN] = '\0'; \ |
3151 | + } \ |
3152 | + while (0) |
3153 | + IGNORE_CHARS (new_len_a, lena, texta, copy_a, |
3154 | + wc_a, mblength_a, state_a); |
3155 | + IGNORE_CHARS (new_len_b, lenb, textb, copy_b, |
3156 | + wc_b, mblength_b, state_b); |
3157 | + diff = xmemcoll (copy_a, new_len_a, copy_b, new_len_b); |
3158 | + } |
3159 | + else if (lena == 0) |
3160 | + diff = - NONZERO (lenb); |
3161 | + else if (lenb == 0) |
3162 | + goto greater; |
3163 | + else |
3164 | + diff = xmemcoll (texta, lena, textb, lenb); |
3165 | + } |
3166 | + |
3167 | + if (diff) |
3168 | + goto not_equal; |
3169 | + |
3170 | + key = key->next; |
3171 | + if (! key) |
3172 | + break; |
3173 | + |
3174 | + /* Find the beginning and limit of the next field. */ |
3175 | + if (key->eword != -1) |
3176 | + lima = limfield (a, key), limb = limfield (b, key); |
3177 | + else |
3178 | + lima = a->text + a->length - 1, limb = b->text + b->length - 1; |
3179 | + |
3180 | + if (key->sword != -1) |
3181 | + texta = begfield (a, key), textb = begfield (b, key); |
3182 | + else |
3183 | + { |
3184 | + texta = a->text, textb = b->text; |
3185 | + if (key->skipsblanks) |
3186 | + { |
3187 | + while (texta < lima && ismbblank (texta, lima - texta, &mblength_a)) |
3188 | + texta += mblength_a; |
3189 | + while (textb < limb && ismbblank (textb, limb - textb, &mblength_b)) |
3190 | + textb += mblength_b; |
3191 | + } |
3192 | + } |
3193 | + } |
3194 | + |
3195 | + return 0; |
3196 | + |
3197 | +greater: |
3198 | + diff = 1; |
3199 | +not_equal: |
3200 | + return key->reverse ? -diff : diff; |
3201 | +} |
3202 | +#endif |
3203 | + |
3204 | /* Compare two lines A and B, returning negative, zero, or positive |
3205 | depending on whether A compares less than, equal to, or greater than B. */ |
3206 | |
3207 | @@ -2744,7 +3288,7 @@ |
3208 | initialize_exit_failure (SORT_FAILURE); |
3209 | |
3210 | hard_LC_COLLATE = hard_locale (LC_COLLATE); |
3211 | -#if HAVE_NL_LANGINFO |
3212 | +#if HAVE_LANGINFO_CODESET |
3213 | hard_LC_TIME = hard_locale (LC_TIME); |
3214 | #endif |
3215 | |
3216 | @@ -2765,6 +3309,27 @@ |
3217 | thousands_sep = -1; |
3218 | } |
3219 | |
3220 | +#if HAVE_MBRTOWC |
3221 | + if (MB_CUR_MAX > 1) |
3222 | + { |
3223 | + inittables = inittables_mb; |
3224 | + begfield = begfield_mb; |
3225 | + limfield = limfield_mb; |
3226 | + getmonth = getmonth_mb; |
3227 | + keycompare = keycompare_mb; |
3228 | + numcompare = numcompare_mb; |
3229 | + } |
3230 | + else |
3231 | +#endif |
3232 | + { |
3233 | + inittables = inittables_uni; |
3234 | + begfield = begfield_uni; |
3235 | + limfield = limfield_uni; |
3236 | + getmonth = getmonth_uni; |
3237 | + keycompare = keycompare_uni; |
3238 | + numcompare = numcompare_uni; |
3239 | + } |
3240 | + |
3241 | have_read_stdin = false; |
3242 | inittables (); |
3243 | |
3244 | @@ -3015,13 +3580,35 @@ |
3245 | |
3246 | case 't': |
3247 | { |
3248 | - char newtab = optarg[0]; |
3249 | - if (! newtab) |
3250 | + char newtab[MB_LEN_MAX + 1]; |
3251 | + size_t newtab_length = 1; |
3252 | + strncpy (newtab, optarg, MB_LEN_MAX); |
3253 | + if (! newtab[0]) |
3254 | error (SORT_FAILURE, 0, _("empty tab")); |
3255 | - if (optarg[1]) |
3256 | +#if HAVE_MBRTOWC |
3257 | + if (MB_CUR_MAX > 1) |
3258 | + { |
3259 | + wchar_t wc; |
3260 | + mbstate_t state; |
3261 | + size_t i; |
3262 | + |
3263 | + memset (&state, '\0', sizeof (mbstate_t)); |
3264 | + newtab_length = mbrtowc (&wc, newtab, strnlen (newtab, |
3265 | + MB_LEN_MAX), |
3266 | + &state); |
3267 | + switch (newtab_length) |
3268 | + { |
3269 | + case (size_t) -1: |
3270 | + case (size_t) -2: |
3271 | + case 0: |
3272 | + newtab_length = 1; |
3273 | + } |
3274 | + } |
3275 | +#endif |
3276 | + if (newtab_length == 1 && optarg[1]) |
3277 | { |
3278 | if (STREQ (optarg, "\\0")) |
3279 | - newtab = '\0'; |
3280 | + newtab[0] = '\0'; |
3281 | else |
3282 | { |
3283 | /* Provoke with `sort -txx'. Complain about |
3284 | @@ -3032,9 +3619,12 @@ |
3285 | quote (optarg)); |
3286 | } |
3287 | } |
3288 | - if (tab != TAB_DEFAULT && tab != newtab) |
3289 | + if (tab_length |
3290 | + && (tab_length != newtab_length |
3291 | + || memcmp (tab, newtab, tab_length) != 0)) |
3292 | error (SORT_FAILURE, 0, _("incompatible tabs")); |
3293 | - tab = newtab; |
3294 | + memcpy (tab, newtab, newtab_length); |
3295 | + tab_length = newtab_length; |
3296 | } |
3297 | break; |
3298 | |
3299 | diff -Naur coreutils-6.9.orig/src/unexpand.c coreutils-6.9/src/unexpand.c |
3300 | --- coreutils-6.9.orig/src/unexpand.c 2007-03-18 21:36:43.000000000 +0000 |
3301 | +++ coreutils-6.9/src/unexpand.c 2007-04-07 16:59:55.000000000 +0000 |
3302 | @@ -39,11 +39,28 @@ |
3303 | #include <stdio.h> |
3304 | #include <getopt.h> |
3305 | #include <sys/types.h> |
3306 | + |
3307 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ |
3308 | +#if HAVE_WCHAR_H |
3309 | +# include <wchar.h> |
3310 | +#endif |
3311 | + |
3312 | #include "system.h" |
3313 | #include "error.h" |
3314 | #include "quote.h" |
3315 | #include "xstrndup.h" |
3316 | |
3317 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
3318 | + installation; work around this configuration error. */ |
3319 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
3320 | +# define MB_LEN_MAX 16 |
3321 | +#endif |
3322 | + |
3323 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
3324 | +#if HAVE_MBRTOWC && defined mbstate_t |
3325 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
3326 | +#endif |
3327 | + |
3328 | /* The official name of this program (e.g., no `g' prefix). */ |
3329 | #define PROGRAM_NAME "unexpand" |
3330 | |
3331 | @@ -110,6 +127,208 @@ |
3332 | {NULL, 0, NULL, 0} |
3333 | }; |
3334 | |
3335 | +static FILE *next_file (FILE *fp); |
3336 | + |
3337 | +#if HAVE_MBRTOWC |
3338 | +static void |
3339 | +unexpand_multibyte (void) |
3340 | +{ |
3341 | + FILE *fp; /* Input stream. */ |
3342 | + mbstate_t i_state; /* Current shift state of the input stream. */ |
3343 | + mbstate_t i_state_bak; /* Back up the I_STATE. */ |
3344 | + mbstate_t o_state; /* Current shift state of the output stream. */ |
3345 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
3346 | + char *bufpos; /* Next read position of BUF. */ |
3347 | + size_t buflen = 0; /* The length of the byte sequence in buf. */ |
3348 | + wint_t wc; /* A gotten wide character. */ |
3349 | + size_t mblength; /* The byte size of a multibyte character |
3350 | + which shows as same character as WC. */ |
3351 | + |
3352 | + /* Index in `tab_list' of next tabstop: */ |
3353 | + int tab_index = 0; /* For calculating width of pending tabs. */ |
3354 | + int print_tab_index = 0; /* For printing as many tabs as possible. */ |
3355 | + unsigned int column = 0; /* Column on screen of next char. */ |
3356 | + int next_tab_column; /* Column the next tab stop is on. */ |
3357 | + int convert = 1; /* If nonzero, perform translations. */ |
3358 | + unsigned int pending = 0; /* Pending columns of blanks. */ |
3359 | + |
3360 | + fp = next_file ((FILE *) NULL); |
3361 | + if (fp == NULL) |
3362 | + return; |
3363 | + |
3364 | + memset (&o_state, '\0', sizeof(mbstate_t)); |
3365 | + memset (&i_state, '\0', sizeof(mbstate_t)); |
3366 | + |
3367 | + for (;;) |
3368 | + { |
3369 | + if (buflen < MB_LEN_MAX && !feof(fp) && !ferror(fp)) |
3370 | + { |
3371 | + memmove (buf, bufpos, buflen); |
3372 | + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, fp); |
3373 | + bufpos = buf; |
3374 | + } |
3375 | + |
3376 | + /* Get a wide character. */ |
3377 | + if (buflen < 1) |
3378 | + { |
3379 | + mblength = 1; |
3380 | + wc = WEOF; |
3381 | + } |
3382 | + else |
3383 | + { |
3384 | + i_state_bak = i_state; |
3385 | + mblength = mbrtowc ((wchar_t *)&wc, bufpos, buflen, &i_state); |
3386 | + } |
3387 | + |
3388 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
3389 | + { |
3390 | + i_state = i_state_bak; |
3391 | + wc = L'\0'; |
3392 | + } |
3393 | + |
3394 | + if (wc == L' ' && convert && column < INT_MAX) |
3395 | + { |
3396 | + ++pending; |
3397 | + ++column; |
3398 | + } |
3399 | + else if (wc == L'\t' && convert) |
3400 | + { |
3401 | + if (tab_size == 0) |
3402 | + { |
3403 | + /* Do not let tab_index == first_free_tab; |
3404 | + stop when it is 1 less. */ |
3405 | + while (tab_index < first_free_tab - 1 |
3406 | + && column >= tab_list[tab_index]) |
3407 | + tab_index++; |
3408 | + next_tab_column = tab_list[tab_index]; |
3409 | + if (tab_index < first_free_tab - 1) |
3410 | + tab_index++; |
3411 | + if (column >= next_tab_column) |
3412 | + { |
3413 | + convert = 0; /* Ran out of tab stops. */ |
3414 | + goto flush_pend_mb; |
3415 | + } |
3416 | + } |
3417 | + else |
3418 | + { |
3419 | + next_tab_column = column + tab_size - column % tab_size; |
3420 | + } |
3421 | + pending += next_tab_column - column; |
3422 | + column = next_tab_column; |
3423 | + } |
3424 | + else |
3425 | + { |
3426 | +flush_pend_mb: |
3427 | + /* Flush pending spaces. Print as many tabs as possible, |
3428 | + then print the rest as spaces. */ |
3429 | + if (pending == 1) |
3430 | + { |
3431 | + putchar (' '); |
3432 | + pending = 0; |
3433 | + } |
3434 | + column -= pending; |
3435 | + while (pending > 0) |
3436 | + { |
3437 | + if (tab_size == 0) |
3438 | + { |
3439 | + /* Do not let print_tab_index == first_free_tab; |
3440 | + stop when it is 1 less. */ |
3441 | + while (print_tab_index < first_free_tab - 1 |
3442 | + && column >= tab_list[print_tab_index]) |
3443 | + print_tab_index++; |
3444 | + next_tab_column = tab_list[print_tab_index]; |
3445 | + if (print_tab_index < first_free_tab - 1) |
3446 | + print_tab_index++; |
3447 | + } |
3448 | + else |
3449 | + { |
3450 | + next_tab_column = |
3451 | + column + tab_size - column % tab_size; |
3452 | + } |
3453 | + if (next_tab_column - column <= pending) |
3454 | + { |
3455 | + putchar ('\t'); |
3456 | + pending -= next_tab_column - column; |
3457 | + column = next_tab_column; |
3458 | + } |
3459 | + else |
3460 | + { |
3461 | + --print_tab_index; |
3462 | + column += pending; |
3463 | + while (pending != 0) |
3464 | + { |
3465 | + putchar (' '); |
3466 | + pending--; |
3467 | + } |
3468 | + } |
3469 | + } |
3470 | + |
3471 | + if (wc == WEOF) |
3472 | + { |
3473 | + fp = next_file (fp); |
3474 | + if (fp == NULL) |
3475 | + break; /* No more files. */ |
3476 | + else |
3477 | + { |
3478 | + memset (&i_state, '\0', sizeof(mbstate_t)); |
3479 | + continue; |
3480 | + } |
3481 | + } |
3482 | + |
3483 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
3484 | + { |
3485 | + if (convert) |
3486 | + { |
3487 | + ++column; |
3488 | + if (convert_entire_line == 0) |
3489 | + convert = 0; |
3490 | + } |
3491 | + mblength = 1; |
3492 | + putchar (buf[0]); |
3493 | + } |
3494 | + else if (mblength == 0) |
3495 | + { |
3496 | + if (convert && convert_entire_line == 0) |
3497 | + convert = 0; |
3498 | + mblength = 1; |
3499 | + putchar ('\0'); |
3500 | + } |
3501 | + else |
3502 | + { |
3503 | + if (convert) |
3504 | + { |
3505 | + if (wc == L'\b') |
3506 | + { |
3507 | + if (column > 0) |
3508 | + --column; |
3509 | + } |
3510 | + else |
3511 | + { |
3512 | + int width; /* The width of WC. */ |
3513 | + |
3514 | + width = wcwidth (wc); |
3515 | + column += (width > 0) ? width : 0; |
3516 | + if (convert_entire_line == 0) |
3517 | + convert = 0; |
3518 | + } |
3519 | + } |
3520 | + |
3521 | + if (wc == L'\n') |
3522 | + { |
3523 | + tab_index = print_tab_index = 0; |
3524 | + column = pending = 0; |
3525 | + convert = 1; |
3526 | + } |
3527 | + fwrite (bufpos, sizeof(char), mblength, stdout); |
3528 | + } |
3529 | + } |
3530 | + buflen -= mblength; |
3531 | + bufpos += mblength; |
3532 | + } |
3533 | +} |
3534 | +#endif |
3535 | + |
3536 | + |
3537 | void |
3538 | usage (int status) |
3539 | { |
3540 | @@ -531,7 +750,12 @@ |
3541 | |
3542 | file_list = (optind < argc ? &argv[optind] : stdin_argv); |
3543 | |
3544 | - unexpand (); |
3545 | +#if HAVE_MBRTOWC |
3546 | + if (MB_CUR_MAX > 1) |
3547 | + unexpand_multibyte (); |
3548 | + else |
3549 | +#endif |
3550 | + unexpand (); |
3551 | |
3552 | if (have_read_stdin && fclose (stdin) != 0) |
3553 | error (EXIT_FAILURE, errno, "-"); |
3554 | diff -Naur coreutils-6.9.orig/src/uniq.c coreutils-6.9/src/uniq.c |
3555 | --- coreutils-6.9.orig/src/uniq.c 2007-03-18 21:36:43.000000000 +0000 |
3556 | +++ coreutils-6.9/src/uniq.c 2007-04-07 16:59:55.000000000 +0000 |
3557 | @@ -23,6 +23,16 @@ |
3558 | #include <getopt.h> |
3559 | #include <sys/types.h> |
3560 | |
3561 | +/* Get mbstate_t, mbrtowc(). */ |
3562 | +#if HAVE_WCHAR_H |
3563 | +# include <wchar.h> |
3564 | +#endif |
3565 | + |
3566 | +/* Get isw* functions. */ |
3567 | +#if HAVE_WCTYPE_H |
3568 | +# include <wctype.h> |
3569 | +#endif |
3570 | + |
3571 | #include "system.h" |
3572 | #include "argmatch.h" |
3573 | #include "linebuffer.h" |
3574 | @@ -32,7 +42,19 @@ |
3575 | #include "quote.h" |
3576 | #include "xmemcoll.h" |
3577 | #include "xstrtol.h" |
3578 | -#include "memcasecmp.h" |
3579 | +#include "xmemcoll.h" |
3580 | + |
3581 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
3582 | + installation; work around this configuration error. */ |
3583 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
3584 | +# define MB_LEN_MAX 16 |
3585 | +#endif |
3586 | + |
3587 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
3588 | +#if HAVE_MBRTOWC && defined mbstate_t |
3589 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
3590 | +#endif |
3591 | + |
3592 | |
3593 | /* The official name of this program (e.g., no `g' prefix). */ |
3594 | #define PROGRAM_NAME "uniq" |
3595 | @@ -109,6 +131,10 @@ |
3596 | /* Select whether/how to delimit groups of duplicate lines. */ |
3597 | static enum delimit_method delimit_groups; |
3598 | |
3599 | +/* Function pointers. */ |
3600 | +static char * |
3601 | +(*find_field) (struct linebuffer *line); |
3602 | + |
3603 | static struct option const longopts[] = |
3604 | { |
3605 | {"count", no_argument, NULL, 'c'}, |
3606 | @@ -198,7 +224,7 @@ |
3607 | return a pointer to the beginning of the line's field to be compared. */ |
3608 | |
3609 | static char * |
3610 | -find_field (const struct linebuffer *line) |
3611 | +find_field_uni (struct linebuffer *line) |
3612 | { |
3613 | size_t count; |
3614 | char *lp = line->buffer; |
3615 | @@ -219,6 +245,83 @@ |
3616 | return lp + i; |
3617 | } |
3618 | |
3619 | +#if HAVE_MBRTOWC |
3620 | + |
3621 | +# define MBCHAR_TO_WCHAR(WC, MBLENGTH, LP, POS, SIZE, STATEP, CONVFAIL) \ |
3622 | + do \ |
3623 | + { \ |
3624 | + mbstate_t state_bak; \ |
3625 | + \ |
3626 | + CONVFAIL = 0; \ |
3627 | + state_bak = *STATEP; \ |
3628 | + \ |
3629 | + MBLENGTH = mbrtowc (&WC, LP + POS, SIZE - POS, STATEP); \ |
3630 | + \ |
3631 | + switch (MBLENGTH) \ |
3632 | + { \ |
3633 | + case (size_t)-2: \ |
3634 | + case (size_t)-1: \ |
3635 | + *STATEP = state_bak; \ |
3636 | + CONVFAIL++; \ |
3637 | + /* Fall through */ \ |
3638 | + case 0: \ |
3639 | + MBLENGTH = 1; \ |
3640 | + } \ |
3641 | + } \ |
3642 | + while (0) |
3643 | + |
3644 | +static char * |
3645 | +find_field_multi (struct linebuffer *line) |
3646 | +{ |
3647 | + size_t count; |
3648 | + char *lp = line->buffer; |
3649 | + size_t size = line->length - 1; |
3650 | + size_t pos; |
3651 | + size_t mblength; |
3652 | + wchar_t wc; |
3653 | + mbstate_t *statep; |
3654 | + int convfail; |
3655 | + |
3656 | + pos = 0; |
3657 | + statep = &(line->state); |
3658 | + |
3659 | + /* skip fields. */ |
3660 | + for (count = 0; count < skip_fields && pos < size; count++) |
3661 | + { |
3662 | + while (pos < size) |
3663 | + { |
3664 | + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); |
3665 | + |
3666 | + if (convfail || !iswblank (wc)) |
3667 | + { |
3668 | + pos += mblength; |
3669 | + break; |
3670 | + } |
3671 | + pos += mblength; |
3672 | + } |
3673 | + |
3674 | + while (pos < size) |
3675 | + { |
3676 | + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); |
3677 | + |
3678 | + if (!convfail && iswblank (wc)) |
3679 | + break; |
3680 | + |
3681 | + pos += mblength; |
3682 | + } |
3683 | + } |
3684 | + |
3685 | + /* skip fields. */ |
3686 | + for (count = 0; count < skip_chars && pos < size; count++) |
3687 | + { |
3688 | + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); |
3689 | + pos += mblength; |
3690 | + } |
3691 | + |
3692 | + return lp + pos; |
3693 | +} |
3694 | +#endif |
3695 | + |
3696 | /* Return false if two strings OLD and NEW match, true if not. |
3697 | OLD and NEW point not to the beginnings of the lines |
3698 | but rather to the beginnings of the fields to compare. |
3699 | @@ -227,6 +330,8 @@ |
3700 | static bool |
3701 | different (char *old, char *new, size_t oldlen, size_t newlen) |
3702 | { |
3703 | + char *copy_old, *copy_new; |
3704 | + |
3705 | if (check_chars < oldlen) |
3706 | oldlen = check_chars; |
3707 | if (check_chars < newlen) |
3708 | @@ -234,14 +339,92 @@ |
3709 | |
3710 | if (ignore_case) |
3711 | { |
3712 | - /* FIXME: This should invoke strcoll somehow. */ |
3713 | - return oldlen != newlen || memcasecmp (old, new, oldlen); |
3714 | + size_t i; |
3715 | + |
3716 | + copy_old = alloca (oldlen + 1); |
3717 | + copy_new = alloca (oldlen + 1); |
3718 | + |
3719 | + for (i = 0; i < oldlen; i++) |
3720 | + { |
3721 | + copy_old[i] = toupper (old[i]); |
3722 | + copy_new[i] = toupper (new[i]); |
3723 | + } |
3724 | } |
3725 | - else if (hard_LC_COLLATE) |
3726 | - return xmemcoll (old, oldlen, new, newlen) != 0; |
3727 | else |
3728 | - return oldlen != newlen || memcmp (old, new, oldlen); |
3729 | + { |
3730 | + copy_old = (char *)old; |
3731 | + copy_new = (char *)new; |
3732 | + } |
3733 | + |
3734 | + return xmemcoll (copy_old, oldlen, copy_new, newlen); |
3735 | +} |
3736 | + |
3737 | +#if HAVE_MBRTOWC |
3738 | +static int |
3739 | +different_multi (const char *old, const char *new, size_t oldlen, size_t newlen, mbstate_t oldstate, mbstate_t newstate) |
3740 | +{ |
3741 | + size_t i, j, chars; |
3742 | + const char *str[2]; |
3743 | + char *copy[2]; |
3744 | + size_t len[2]; |
3745 | + mbstate_t state[2]; |
3746 | + size_t mblength; |
3747 | + wchar_t wc, uwc; |
3748 | + mbstate_t state_bak; |
3749 | + |
3750 | + str[0] = old; |
3751 | + str[1] = new; |
3752 | + len[0] = oldlen; |
3753 | + len[1] = newlen; |
3754 | + state[0] = oldstate; |
3755 | + state[1] = newstate; |
3756 | + |
3757 | + for (i = 0; i < 2; i++) |
3758 | + { |
3759 | + copy[i] = alloca (len[i] + 1); |
3760 | + |
3761 | + for (j = 0, chars = 0; j < len[i] && chars < check_chars; chars++) |
3762 | + { |
3763 | + state_bak = state[i]; |
3764 | + mblength = mbrtowc (&wc, str[i] + j, len[i] - j, &(state[i])); |
3765 | + |
3766 | + switch (mblength) |
3767 | + { |
3768 | + case (size_t)-1: |
3769 | + case (size_t)-2: |
3770 | + state[i] = state_bak; |
3771 | + /* Fall through */ |
3772 | + case 0: |
3773 | + mblength = 1; |
3774 | + break; |
3775 | + |
3776 | + default: |
3777 | + if (ignore_case) |
3778 | + { |
3779 | + uwc = towupper (wc); |
3780 | + |
3781 | + if (uwc != wc) |
3782 | + { |
3783 | + mbstate_t state_wc; |
3784 | + |
3785 | + memset (&state_wc, '\0', sizeof(mbstate_t)); |
3786 | + wcrtomb (copy[i] + j, uwc, &state_wc); |
3787 | + } |
3788 | + else |
3789 | + memcpy (copy[i] + j, str[i] + j, mblength); |
3790 | + } |
3791 | + else |
3792 | + memcpy (copy[i] + j, str[i] + j, mblength); |
3793 | + } |
3794 | + j += mblength; |
3795 | + } |
3796 | + copy[i][j] = '\0'; |
3797 | + len[i] = j; |
3798 | + } |
3799 | + |
3800 | + return xmemcoll (copy[0], len[0], copy[1], len[1]); |
3801 | } |
3802 | +#endif |
3803 | |
3804 | /* Output the line in linebuffer LINE to standard output |
3805 | provided that the switches say it should be output. |
3806 | @@ -295,15 +478,43 @@ |
3807 | { |
3808 | char *prevfield IF_LINT (= NULL); |
3809 | size_t prevlen IF_LINT (= 0); |
3810 | +#if HAVE_MBRTOWC |
3811 | + mbstate_t prevstate; |
3812 | + |
3813 | + memset (&prevstate, '\0', sizeof (mbstate_t)); |
3814 | +#endif |
3815 | |
3816 | while (!feof (stdin)) |
3817 | { |
3818 | char *thisfield; |
3819 | size_t thislen; |
3820 | +#if HAVE_MBRTOWC |
3821 | + mbstate_t thisstate; |
3822 | +#endif |
3823 | + |
3824 | if (readlinebuffer (thisline, stdin) == 0) |
3825 | break; |
3826 | thisfield = find_field (thisline); |
3827 | thislen = thisline->length - 1 - (thisfield - thisline->buffer); |
3828 | +#if HAVE_MBRTOWC |
3829 | + if (MB_CUR_MAX > 1) |
3830 | + { |
3831 | + thisstate = thisline->state; |
3832 | + |
3833 | + if (prevline->length == 0 || different_multi |
3834 | + (thisfield, prevfield, thislen, prevlen, thisstate, prevstate)) |
3835 | + { |
3836 | + fwrite (thisline->buffer, sizeof (char), |
3837 | + thisline->length, stdout); |
3838 | + |
3839 | + SWAP_LINES (prevline, thisline); |
3840 | + prevfield = thisfield; |
3841 | + prevlen = thislen; |
3842 | + prevstate = thisstate; |
3843 | + } |
3844 | + } |
3845 | + else |
3846 | +#endif |
3847 | if (prevline->length == 0 |
3848 | || different (thisfield, prevfield, thislen, prevlen)) |
3849 | { |
3850 | @@ -322,17 +533,26 @@ |
3851 | size_t prevlen; |
3852 | uintmax_t match_count = 0; |
3853 | bool first_delimiter = true; |
3854 | +#if HAVE_MBRTOWC |
3855 | + mbstate_t prevstate; |
3856 | +#endif |
3857 | |
3858 | if (readlinebuffer (prevline, stdin) == 0) |
3859 | goto closefiles; |
3860 | prevfield = find_field (prevline); |
3861 | prevlen = prevline->length - 1 - (prevfield - prevline->buffer); |
3862 | +#if HAVE_MBRTOWC |
3863 | + prevstate = prevline->state; |
3864 | +#endif |
3865 | |
3866 | while (!feof (stdin)) |
3867 | { |
3868 | bool match; |
3869 | char *thisfield; |
3870 | size_t thislen; |
3871 | +#if HAVE_MBRTOWC |
3872 | + mbstate_t thisstate; |
3873 | +#endif |
3874 | if (readlinebuffer (thisline, stdin) == 0) |
3875 | { |
3876 | if (ferror (stdin)) |
3877 | @@ -341,6 +561,15 @@ |
3878 | } |
3879 | thisfield = find_field (thisline); |
3880 | thislen = thisline->length - 1 - (thisfield - thisline->buffer); |
3881 | +#if HAVE_MBRTOWC |
3882 | + if (MB_CUR_MAX > 1) |
3883 | + { |
3884 | + thisstate = thisline->state; |
3885 | + match = !different_multi (thisfield, prevfield, |
3886 | + thislen, prevlen, thisstate, prevstate); |
3887 | + } |
3888 | + else |
3889 | +#endif |
3890 | match = !different (thisfield, prevfield, thislen, prevlen); |
3891 | match_count += match; |
3892 | |
3893 | @@ -373,6 +602,9 @@ |
3894 | SWAP_LINES (prevline, thisline); |
3895 | prevfield = thisfield; |
3896 | prevlen = thislen; |
3897 | +#if HAVE_MBRTOWC |
3898 | + prevstate = thisstate; |
3899 | +#endif |
3900 | if (!match) |
3901 | match_count = 0; |
3902 | } |
3903 | @@ -417,6 +649,19 @@ |
3904 | |
3905 | atexit (close_stdout); |
3906 | |
3907 | +#if HAVE_MBRTOWC |
3908 | + if (MB_CUR_MAX > 1) |
3909 | + { |
3910 | + find_field = find_field_multi; |
3911 | + } |
3912 | + else |
3913 | +#endif |
3914 | + { |
3915 | + find_field = find_field_uni; |
3916 | + } |
3917 | + |
3918 | + |
3919 | + |
3920 | skip_chars = 0; |
3921 | skip_fields = 0; |
3922 | check_chars = SIZE_MAX; |
3923 | diff -Naur coreutils-6.9.orig/tests/sort/Makefile.am coreutils-6.9/tests/sort/Makefile.am |
3924 | --- coreutils-6.9.orig/tests/sort/Makefile.am 2007-03-18 21:36:44.000000000 +0000 |
3925 | +++ coreutils-6.9/tests/sort/Makefile.am 2007-04-07 17:03:36.000000000 +0000 |
3926 | @@ -66,15 +66,17 @@ |
3927 | bigfield.O bigfield.E |
3928 | ##test-files-end |
3929 | |
3930 | -EXTRA_DIST = Test.pm $x-tests $(explicit) $(maint_gen) |
3931 | -noinst_SCRIPTS = $x-tests |
3932 | +run_gen += mb1.0 mb2.0 |
3933 | + |
3934 | +EXTRA_DIST = Test.pm $x-tests $(explicit) $(maint_gen) mb1.I mb1.X mb2.I mb2.X |
3935 | +noinst_SCRIPTS = $x-tests # $x-mb-tests |
3936 | TESTS_ENVIRONMENT = \ |
3937 | CU_TEST_NAME=`basename $(abs_srcdir)`,$$tst \ |
3938 | PATH="$(VG_PATH_PREFIX)`pwd`/../../src$(PATH_SEPARATOR)$$PATH" |
3939 | |
3940 | editpl = sed -e 's,@''PERL''@,$(PERL),g' -e 's,@''srcdir''@,$(srcdir),g' |
3941 | |
3942 | -TESTS = $x-tests |
3943 | +TESTS = $x-tests $x-mb-tests |
3944 | |
3945 | mk_script = $(srcdir)/../mk-script |
3946 | $(srcdir)/$x-tests: $(mk_script) Test.pm Makefile.am |
3947 | diff -Naur coreutils-6.9.orig/tests/sort/Makefile.in coreutils-6.9/tests/sort/Makefile.in |
3948 | --- coreutils-6.9.orig/tests/sort/Makefile.in 2007-03-22 21:20:25.000000000 +0000 |
3949 | +++ coreutils-6.9/tests/sort/Makefile.in 2007-04-07 17:01:55.000000000 +0000 |
3950 | @@ -540,14 +540,16 @@ |
3951 | incompat5.O incompat5.E incompat6.O incompat6.E nul-tab.O nul-tab.E \ |
3952 | bigfield.O bigfield.E |
3953 | |
3954 | -EXTRA_DIST = Test.pm $x-tests $(explicit) $(maint_gen) |
3955 | -noinst_SCRIPTS = $x-tests |
3956 | +run_gen += mb1.0 mb2.0 |
3957 | + |
3958 | +EXTRA_DIST = Test.pm $x-tests $(explicit) $(maint_gen) mb1.I mb1.X mb2.I mb2.X |
3959 | +noinst_SCRIPTS = $x-tests # $x-mb-tests |
3960 | TESTS_ENVIRONMENT = \ |
3961 | CU_TEST_NAME=`basename $(abs_srcdir)`,$$tst \ |
3962 | PATH="$(VG_PATH_PREFIX)`pwd`/../../src$(PATH_SEPARATOR)$$PATH" |
3963 | |
3964 | editpl = sed -e 's,@''PERL''@,$(PERL),g' -e 's,@''srcdir''@,$(srcdir),g' |
3965 | -TESTS = $x-tests |
3966 | +TESTS = $x-tests $x-mb-tests |
3967 | mk_script = $(srcdir)/../mk-script |
3968 | MAINTAINERCLEANFILES = $x-tests $(maint_gen) |
3969 | CLEANFILES = $(run_gen) |
3970 | diff -Naur coreutils-6.9.orig/tests/sort/mb1.I coreutils-6.9/tests/sort/mb1.I |
3971 | --- coreutils-6.9.orig/tests/sort/mb1.I 1970-01-01 00:00:00.000000000 +0000 |
3972 | +++ coreutils-6.9/tests/sort/mb1.I 2007-04-07 16:59:55.000000000 +0000 |
3973 | @@ -0,0 +1,4 @@ |
3974 | +Apple@10 |
3975 | +Banana@5 |
3976 | +Citrus@20 |
3977 | +Cherry@30 |
3978 | diff -Naur coreutils-6.9.orig/tests/sort/mb1.X coreutils-6.9/tests/sort/mb1.X |
3979 | --- coreutils-6.9.orig/tests/sort/mb1.X 1970-01-01 00:00:00.000000000 +0000 |
3980 | +++ coreutils-6.9/tests/sort/mb1.X 2007-04-07 16:59:55.000000000 +0000 |
3981 | @@ -0,0 +1,4 @@ |
3982 | +Banana@5 |
3983 | +Apple@10 |
3984 | +Citrus@20 |
3985 | +Cherry@30 |
3986 | diff -Naur coreutils-6.9.orig/tests/sort/mb2.I coreutils-6.9/tests/sort/mb2.I |
3987 | --- coreutils-6.9.orig/tests/sort/mb2.I 1970-01-01 00:00:00.000000000 +0000 |
3988 | +++ coreutils-6.9/tests/sort/mb2.I 2007-04-07 16:59:55.000000000 +0000 |
3989 | @@ -0,0 +1,4 @@ |
3990 | +Apple@AA10@@20 |
3991 | +Banana@AA5@@30 |
3992 | +Citrus@AA20@@5 |
3993 | +Cherry@AA30@@10 |
3994 | diff -Naur coreutils-6.9.orig/tests/sort/mb2.X coreutils-6.9/tests/sort/mb2.X |
3995 | --- coreutils-6.9.orig/tests/sort/mb2.X 1970-01-01 00:00:00.000000000 +0000 |
3996 | +++ coreutils-6.9/tests/sort/mb2.X 2007-04-07 16:59:55.000000000 +0000 |
3997 | @@ -0,0 +1,4 @@ |
3998 | +Citrus@AA20@@5 |
3999 | +Cherry@AA30@@10 |
4000 | +Apple@AA10@@20 |
4001 | +Banana@AA5@@30 |
4002 | diff -Naur coreutils-6.9.orig/tests/sort/sort-mb-tests coreutils-6.9/tests/sort/sort-mb-tests |
4003 | --- coreutils-6.9.orig/tests/sort/sort-mb-tests 1970-01-01 00:00:00.000000000 +0000 |
4004 | +++ coreutils-6.9/tests/sort/sort-mb-tests 2007-04-07 16:59:55.000000000 +0000 |
4005 | @@ -0,0 +1,58 @@ |
4006 | +#! /bin/sh |
4007 | +case $# in |
4008 | + 0) xx='../../src/sort';; |
4009 | + *) xx="$1";; |
4010 | +esac |
4011 | +test "$VERBOSE" && echo=echo || echo=: |
4012 | +$echo testing program: $xx |
4013 | +errors=0 |
4014 | +test "$srcdir" || srcdir=. |
4015 | +test "$VERBOSE" && $xx --version 2> /dev/null |
4016 | + |
4017 | +export LC_ALL=en_US.UTF-8 |
4018 | +locale -k LC_CTYPE 2>&1 | grep -q charmap.*UTF-8 || exit 77 |
4019 | +errors=0 |
4020 | + |
4021 | +$xx -t @ -k2 -n mb1.I > mb1.O |
4022 | +code=$? |
4023 | +if test $code != 0; then |
4024 | + $echo "Test mb1 failed: $xx return code $code differs from expected value 0" 1>&2 |
4025 | + errors=`expr $errors + 1` |
4026 | +else |
4027 | + cmp mb1.O $srcdir/mb1.X > /dev/null 2>&1 |
4028 | + case $? in |
4029 | + 0) if test "$VERBOSE"; then $echo "passed mb1"; fi;; |
4030 | + 1) $echo "Test mb1 failed: files mb1.O and $srcdir/mb1.X differ" 1>&2 |
4031 | + (diff -c mb1.O $srcdir/mb1.X) 2> /dev/null |
4032 | + errors=`expr $errors + 1`;; |
4033 | + 2) $echo "Test mb1 may have failed." 1>&2 |
4034 | + $echo The command "cmp mb1.O $srcdir/mb1.X" failed. 1>&2 |
4035 | + errors=`expr $errors + 1`;; |
4036 | + esac |
4037 | +fi |
4038 | + |
4039 | +$xx -t @ -k4 -n mb2.I > mb2.O |
4040 | +code=$? |
4041 | +if test $code != 0; then |
4042 | + $echo "Test mb2 failed: $xx return code $code differs from expected value 0" 1>&2 |
4043 | + errors=`expr $errors + 1` |
4044 | +else |
4045 | + cmp mb2.O $srcdir/mb2.X > /dev/null 2>&1 |
4046 | + case $? in |
4047 | + 0) if test "$VERBOSE"; then $echo "passed mb2"; fi;; |
4048 | + 1) $echo "Test mb2 failed: files mb2.O and $srcdir/mb2.X differ" 1>&2 |
4049 | + (diff -c mb2.O $srcdir/mb2.X) 2> /dev/null |
4050 | + errors=`expr $errors + 1`;; |
4051 | + 2) $echo "Test mb2 may have failed." 1>&2 |
4052 | + $echo The command "cmp mb2.O $srcdir/mb2.X" failed. 1>&2 |
4053 | + errors=`expr $errors + 1`;; |
4054 | + esac |
4055 | +fi |
4056 | + |
4057 | +if test $errors = 0; then |
4058 | + $echo Passed all 113 tests. 1>&2 |
4059 | +else |
4060 | + $echo Failed $errors tests. 1>&2 |
4061 | +fi |
4062 | +test $errors = 0 || errors=1 |
4063 | +exit $errors |