Contents of /trunk/coreutils/patches-5.94/coreutils-5.94-i18n.patch
Parent Directory | Revision Log
Revision 115 -
(show annotations)
(download)
Sun Mar 18 15:57:37 2007 UTC (17 years, 6 months ago) by niro
File size: 108739 byte(s)
Sun Mar 18 15:57:37 2007 UTC (17 years, 6 months ago) by niro
File size: 108739 byte(s)
patches for 5.94
1 | --- coreutils-5.93/lib/linebuffer.h.i18n 2005-05-14 08:58:06.000000000 +0100 |
2 | +++ coreutils-5.93/lib/linebuffer.h 2005-12-23 08:53:01.000000000 +0000 |
3 | @@ -22,6 +22,11 @@ |
4 | |
5 | # include <stdio.h> |
6 | |
7 | +/* Get mbstate_t. */ |
8 | +# if HAVE_WCHAR_H |
9 | +# include <wchar.h> |
10 | +# endif |
11 | + |
12 | /* A `struct linebuffer' holds a line of text. */ |
13 | |
14 | struct linebuffer |
15 | @@ -29,6 +34,9 @@ |
16 | size_t size; /* Allocated. */ |
17 | size_t length; /* Used. */ |
18 | char *buffer; |
19 | +# if HAVE_WCHAR_H |
20 | + mbstate_t state; |
21 | +# endif |
22 | }; |
23 | |
24 | /* Initialize linebuffer LINEBUFFER for use. */ |
25 | --- coreutils-5.93/src/cut.c.i18n 2005-08-12 08:16:25.000000000 +0100 |
26 | +++ coreutils-5.93/src/cut.c 2005-12-23 08:53:01.000000000 +0000 |
27 | @@ -29,6 +29,11 @@ |
28 | #include <assert.h> |
29 | #include <getopt.h> |
30 | #include <sys/types.h> |
31 | + |
32 | +/* Get mbstate_t, mbrtowc(). */ |
33 | +#if HAVE_WCHAR_H |
34 | +# include <wchar.h> |
35 | +#endif |
36 | #include "system.h" |
37 | |
38 | #include "error.h" |
39 | @@ -37,6 +42,18 @@ |
40 | #include "quote.h" |
41 | #include "xstrndup.h" |
42 | |
43 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
44 | + installation; work around this configuration error. */ |
45 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
46 | +# undef MB_LEN_MAX |
47 | +# define MB_LEN_MAX 16 |
48 | +#endif |
49 | + |
50 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
51 | +#if HAVE_MBRTOWC && defined mbstate_t |
52 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
53 | +#endif |
54 | + |
55 | /* The official name of this program (e.g., no `g' prefix). */ |
56 | #define PROGRAM_NAME "cut" |
57 | |
58 | @@ -67,6 +84,52 @@ |
59 | } \ |
60 | while (0) |
61 | |
62 | +/* Refill the buffer BUF to get a multibyte character. */ |
63 | +#define REFILL_BUFFER(BUF, BUFPOS, BUFLEN, STREAM) \ |
64 | + do \ |
65 | + { \ |
66 | + if (BUFLEN < MB_LEN_MAX && !feof (STREAM) && !ferror (STREAM)) \ |
67 | + { \ |
68 | + memmove (BUF, BUFPOS, BUFLEN); \ |
69 | + BUFLEN += fread (BUF + BUFLEN, sizeof(char), BUFSIZ, STREAM); \ |
70 | + BUFPOS = BUF; \ |
71 | + } \ |
72 | + } \ |
73 | + while (0) |
74 | + |
75 | +/* Get wide character on BUFPOS. BUFPOS is not included after that. |
76 | + If byte sequence is not valid as a character, CONVFAIL is 1. Otherwise 0. */ |
77 | +#define GET_NEXT_WC_FROM_BUFFER(WC, BUFPOS, BUFLEN, MBLENGTH, STATE, CONVFAIL) \ |
78 | + do \ |
79 | + { \ |
80 | + mbstate_t state_bak; \ |
81 | + \ |
82 | + if (BUFLEN < 1) \ |
83 | + { \ |
84 | + WC = WEOF; \ |
85 | + break; \ |
86 | + } \ |
87 | + \ |
88 | + /* Get a wide character. */ \ |
89 | + CONVFAIL = 0; \ |
90 | + state_bak = STATE; \ |
91 | + MBLENGTH = mbrtowc ((wchar_t *)&WC, BUFPOS, BUFLEN, &STATE); \ |
92 | + \ |
93 | + switch (MBLENGTH) \ |
94 | + { \ |
95 | + case (size_t)-1: \ |
96 | + case (size_t)-2: \ |
97 | + CONVFAIL++; \ |
98 | + STATE = state_bak; \ |
99 | + /* Fall througn. */ \ |
100 | + \ |
101 | + case 0: \ |
102 | + MBLENGTH = 1; \ |
103 | + break; \ |
104 | + } \ |
105 | + } \ |
106 | + while (0) |
107 | + |
108 | struct range_pair |
109 | { |
110 | size_t lo; |
111 | @@ -85,7 +148,7 @@ |
112 | /* The number of bytes allocated for FIELD_1_BUFFER. */ |
113 | static size_t field_1_bufsize; |
114 | |
115 | -/* The largest field or byte index used as an endpoint of a closed |
116 | +/* The largest byte, character or field index used as an endpoint of a closed |
117 | or degenerate range specification; this doesn't include the starting |
118 | index of right-open-ended ranges. For example, with either range spec |
119 | `2-5,9-', `2-3,5,9-' this variable would be set to 5. */ |
120 | @@ -97,10 +160,11 @@ |
121 | |
122 | /* This is a bit vector. |
123 | In byte mode, which bytes to output. |
124 | + In character mode, which characters to output. |
125 | In field mode, which DELIM-separated fields to output. |
126 | - Both bytes and fields are numbered starting with 1, |
127 | + Bytes, characters and fields are numbered starting with 1, |
128 | so the zeroth bit of this array is unused. |
129 | - A field or byte K has been selected if |
130 | + A byte, character or field K has been selected if |
131 | (K <= MAX_RANGE_ENDPOINT and is_printable_field(K)) |
132 | || (EOL_RANGE_START > 0 && K >= EOL_RANGE_START). */ |
133 | static unsigned char *printable_field; |
134 | @@ -109,9 +173,12 @@ |
135 | { |
136 | undefined_mode, |
137 | |
138 | - /* Output characters that are in the given bytes. */ |
139 | + /* Output bytes that are at the given positions. */ |
140 | byte_mode, |
141 | |
142 | + /* Output characters that are at the given positions. */ |
143 | + character_mode, |
144 | + |
145 | /* Output the given delimeter-separated fields. */ |
146 | field_mode |
147 | }; |
148 | @@ -121,6 +188,13 @@ |
149 | |
150 | static enum operating_mode operating_mode; |
151 | |
152 | +/* If nonzero, when in byte mode, don't split multibyte characters. */ |
153 | +static int byte_mode_character_aware; |
154 | + |
155 | +/* If nonzero, the function for single byte locale is work |
156 | + if this program runs on multibyte locale. */ |
157 | +static int force_singlebyte_mode; |
158 | + |
159 | /* If true do not output lines containing no delimeter characters. |
160 | Otherwise, all such lines are printed. This option is valid only |
161 | with field mode. */ |
162 | @@ -132,6 +206,9 @@ |
163 | |
164 | /* The delimeter character for field mode. */ |
165 | static unsigned char delim; |
166 | +#if HAVE_WCHAR_H |
167 | +static wchar_t wcdelim; |
168 | +#endif |
169 | |
170 | /* True if the --output-delimiter=STRING option was specified. */ |
171 | static bool output_delimiter_specified; |
172 | @@ -205,7 +282,7 @@ |
173 | -f, --fields=LIST select only these fields; also print any line\n\ |
174 | that contains no delimiter character, unless\n\ |
175 | the -s option is specified\n\ |
176 | - -n (ignored)\n\ |
177 | + -n with -b: don't split multibyte characters\n\ |
178 | "), stdout); |
179 | fputs (_("\ |
180 | --complement complement the set of selected bytes, characters\n\ |
181 | @@ -360,7 +437,7 @@ |
182 | in_digits = false; |
183 | /* Starting a range. */ |
184 | if (dash_found) |
185 | - FATAL_ERROR (_("invalid byte or field list")); |
186 | + FATAL_ERROR (_("invalid byte, character or field list")); |
187 | dash_found = true; |
188 | fieldstr++; |
189 | |
190 | @@ -385,14 +462,16 @@ |
191 | if (value == 0) |
192 | { |
193 | /* `n-'. From `initial' to end of line. */ |
194 | - eol_range_start = initial; |
195 | + if (eol_range_start == 0 || |
196 | + (eol_range_start != 0 && eol_range_start > initial)) |
197 | + eol_range_start = initial; |
198 | field_found = true; |
199 | } |
200 | else |
201 | { |
202 | /* `m-n' or `-n' (1-n). */ |
203 | if (value < initial) |
204 | - FATAL_ERROR (_("invalid byte or field list")); |
205 | + FATAL_ERROR (_("invalid byte, character or field list")); |
206 | |
207 | /* Is there already a range going to end of line? */ |
208 | if (eol_range_start != 0) |
209 | @@ -465,6 +544,9 @@ |
210 | if (operating_mode == byte_mode) |
211 | error (0, 0, |
212 | _("byte offset %s is too large"), quote (bad_num)); |
213 | + else if (operating_mode == character_mode) |
214 | + error (0, 0, |
215 | + _("character offset %s is too large"), quote (bad_num)); |
216 | else |
217 | error (0, 0, |
218 | _("field number %s is too large"), quote (bad_num)); |
219 | @@ -475,7 +557,7 @@ |
220 | fieldstr++; |
221 | } |
222 | else |
223 | - FATAL_ERROR (_("invalid byte or field list")); |
224 | + FATAL_ERROR (_("invalid byte, character or field list")); |
225 | } |
226 | |
227 | max_range_endpoint = 0; |
228 | @@ -568,6 +650,63 @@ |
229 | } |
230 | } |
231 | |
232 | +#if HAVE_MBRTOWC |
233 | +/* This function is in use for the following case. |
234 | + |
235 | + 1. Read from the stream STREAM, printing to standard output any selected |
236 | + characters. |
237 | + |
238 | + 2. Read from stream STREAM, printing to standard output any selected bytes, |
239 | + without splitting multibyte characters. */ |
240 | + |
241 | +static void |
242 | +cut_characters_or_cut_bytes_no_split (FILE *stream) |
243 | +{ |
244 | + int idx; /* number of bytes or characters in the line so far. */ |
245 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
246 | + char *bufpos; /* Next read position of BUF. */ |
247 | + size_t buflen; /* The length of the byte sequence in buf. */ |
248 | + wint_t wc = 0; /* A gotten wide character. */ |
249 | + size_t mblength; /* The byte size of a multibyte character which shows |
250 | + as same character as WC. */ |
251 | + mbstate_t state; /* State of the stream. */ |
252 | + int convfail; /* 1, when conversion is failed. Otherwise 0. */ |
253 | + |
254 | + idx = 0; |
255 | + buflen = 0; |
256 | + bufpos = buf; |
257 | + memset (&state, '\0', sizeof(mbstate_t)); |
258 | + |
259 | + while (1) |
260 | + { |
261 | + REFILL_BUFFER (buf, bufpos, buflen, stream); |
262 | + |
263 | + GET_NEXT_WC_FROM_BUFFER (wc, bufpos, buflen, mblength, state, convfail); |
264 | + |
265 | + if (wc == WEOF) |
266 | + { |
267 | + if (idx > 0) |
268 | + putchar ('\n'); |
269 | + break; |
270 | + } |
271 | + else if (wc == L'\n') |
272 | + { |
273 | + putchar ('\n'); |
274 | + idx = 0; |
275 | + } |
276 | + else |
277 | + { |
278 | + idx += (operating_mode == byte_mode) ? mblength : 1; |
279 | + if (print_kth (idx, NULL)) |
280 | + fwrite (bufpos, mblength, sizeof(char), stdout); |
281 | + } |
282 | + |
283 | + buflen -= mblength; |
284 | + bufpos += mblength; |
285 | + } |
286 | +} |
287 | +#endif |
288 | + |
289 | /* Read from stream STREAM, printing to standard output any selected fields. */ |
290 | |
291 | static void |
292 | @@ -689,13 +828,192 @@ |
293 | } |
294 | } |
295 | |
296 | +#if HAVE_MBRTOWC |
297 | +static void |
298 | +cut_fields_mb (FILE *stream) |
299 | +{ |
300 | + int c; |
301 | + unsigned int field_idx; |
302 | + int found_any_selected_field; |
303 | + int buffer_first_field; |
304 | + int empty_input; |
305 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
306 | + char *bufpos; /* Next read position of BUF. */ |
307 | + size_t buflen; /* The length of the byte sequence in buf. */ |
308 | + wint_t wc = 0; /* A gotten wide character. */ |
309 | + size_t mblength; /* The byte size of a multibyte character which shows |
310 | + as same character as WC. */ |
311 | + mbstate_t state; /* State of the stream. */ |
312 | + int convfail; /* 1, when conversion is failed. Otherwise 0. */ |
313 | + |
314 | + found_any_selected_field = 0; |
315 | + field_idx = 1; |
316 | + bufpos = buf; |
317 | + buflen = 0; |
318 | + memset (&state, '\0', sizeof(mbstate_t)); |
319 | + |
320 | + c = getc (stream); |
321 | + empty_input = (c == EOF); |
322 | + if (c != EOF) |
323 | + ungetc (c, stream); |
324 | + else |
325 | + wc = WEOF; |
326 | + |
327 | + /* To support the semantics of the -s flag, we may have to buffer |
328 | + all of the first field to determine whether it is `delimited.' |
329 | + But that is unnecessary if all non-delimited lines must be printed |
330 | + and the first field has been selected, or if non-delimited lines |
331 | + must be suppressed and the first field has *not* been selected. |
332 | + That is because a non-delimited line has exactly one field. */ |
333 | + buffer_first_field = (suppress_non_delimited ^ !print_kth (1, NULL)); |
334 | + |
335 | + while (1) |
336 | + { |
337 | + if (field_idx == 1 && buffer_first_field) |
338 | + { |
339 | + int len = 0; |
340 | + |
341 | + while (1) |
342 | + { |
343 | + REFILL_BUFFER (buf, bufpos, buflen, stream); |
344 | + |
345 | + GET_NEXT_WC_FROM_BUFFER |
346 | + (wc, bufpos, buflen, mblength, state, convfail); |
347 | + |
348 | + if (wc == WEOF) |
349 | + break; |
350 | + |
351 | + field_1_buffer = xrealloc (field_1_buffer, len + mblength); |
352 | + memcpy (field_1_buffer + len, bufpos, mblength); |
353 | + len += mblength; |
354 | + buflen -= mblength; |
355 | + bufpos += mblength; |
356 | + |
357 | + if (!convfail && (wc == L'\n' || wc == wcdelim)) |
358 | + break; |
359 | + } |
360 | + |
361 | + if (wc == WEOF) |
362 | + break; |
363 | + |
364 | + /* If the first field extends to the end of line (it is not |
365 | + delimited) and we are printing all non-delimited lines, |
366 | + print this one. */ |
367 | + if (convfail || (!convfail && wc != wcdelim)) |
368 | + { |
369 | + if (suppress_non_delimited) |
370 | + { |
371 | + /* Empty. */ |
372 | + } |
373 | + else |
374 | + { |
375 | + fwrite (field_1_buffer, sizeof (char), len, stdout); |
376 | + /* Make sure the output line is newline terminated. */ |
377 | + if (convfail || (!convfail && wc != L'\n')) |
378 | + putchar ('\n'); |
379 | + } |
380 | + continue; |
381 | + } |
382 | + |
383 | + if (print_kth (1, NULL)) |
384 | + { |
385 | + /* Print the field, but not the trailing delimiter. */ |
386 | + fwrite (field_1_buffer, sizeof (char), len - 1, stdout); |
387 | + found_any_selected_field = 1; |
388 | + } |
389 | + ++field_idx; |
390 | + } |
391 | + |
392 | + if (wc != WEOF) |
393 | + { |
394 | + if (print_kth (field_idx, NULL)) |
395 | + { |
396 | + if (found_any_selected_field) |
397 | + { |
398 | + fwrite (output_delimiter_string, sizeof (char), |
399 | + output_delimiter_length, stdout); |
400 | + } |
401 | + found_any_selected_field = 1; |
402 | + } |
403 | + |
404 | + while (1) |
405 | + { |
406 | + REFILL_BUFFER (buf, bufpos, buflen, stream); |
407 | + |
408 | + GET_NEXT_WC_FROM_BUFFER |
409 | + (wc, bufpos, buflen, mblength, state, convfail); |
410 | + |
411 | + if (wc == WEOF) |
412 | + break; |
413 | + else if (!convfail && (wc == wcdelim || wc == L'\n')) |
414 | + { |
415 | + buflen -= mblength; |
416 | + bufpos += mblength; |
417 | + break; |
418 | + } |
419 | + |
420 | + if (print_kth (field_idx, NULL)) |
421 | + fwrite (bufpos, mblength, sizeof(char), stdout); |
422 | + |
423 | + buflen -= mblength; |
424 | + bufpos += mblength; |
425 | + } |
426 | + } |
427 | + |
428 | + if ((!convfail || wc == L'\n') && buflen < 1) |
429 | + wc = WEOF; |
430 | + |
431 | + if (!convfail && wc == wcdelim) |
432 | + ++field_idx; |
433 | + else if (wc == WEOF || (!convfail && wc == L'\n')) |
434 | + { |
435 | + if (found_any_selected_field |
436 | + || (!empty_input && !(suppress_non_delimited && field_idx == 1))) |
437 | + putchar ('\n'); |
438 | + if (wc == WEOF) |
439 | + break; |
440 | + field_idx = 1; |
441 | + found_any_selected_field = 0; |
442 | + } |
443 | + } |
444 | +} |
445 | +#endif |
446 | + |
447 | static void |
448 | cut_stream (FILE *stream) |
449 | { |
450 | - if (operating_mode == byte_mode) |
451 | - cut_bytes (stream); |
452 | +#if HAVE_MBRTOWC |
453 | + if (MB_CUR_MAX > 1 && !force_singlebyte_mode) |
454 | + { |
455 | + switch (operating_mode) |
456 | + { |
457 | + case byte_mode: |
458 | + if (byte_mode_character_aware) |
459 | + cut_characters_or_cut_bytes_no_split (stream); |
460 | + else |
461 | + cut_bytes (stream); |
462 | + break; |
463 | + |
464 | + case character_mode: |
465 | + cut_characters_or_cut_bytes_no_split (stream); |
466 | + break; |
467 | + |
468 | + case field_mode: |
469 | + cut_fields_mb (stream); |
470 | + break; |
471 | + |
472 | + default: |
473 | + abort (); |
474 | + } |
475 | + } |
476 | else |
477 | - cut_fields (stream); |
478 | +#endif |
479 | + { |
480 | + if (operating_mode == field_mode) |
481 | + cut_fields (stream); |
482 | + else |
483 | + cut_bytes (stream); |
484 | + } |
485 | } |
486 | |
487 | /* Process file FILE to standard output. |
488 | @@ -745,6 +1063,8 @@ |
489 | bool ok; |
490 | bool delim_specified = false; |
491 | char *spec_list_string IF_LINT(= NULL); |
492 | + char mbdelim[MB_LEN_MAX + 1]; |
493 | + size_t delimlen = 0; |
494 | |
495 | initialize_main (&argc, &argv); |
496 | program_name = argv[0]; |
497 | @@ -767,7 +1087,6 @@ |
498 | switch (optc) |
499 | { |
500 | case 'b': |
501 | - case 'c': |
502 | /* Build the byte list. */ |
503 | if (operating_mode != undefined_mode) |
504 | FATAL_ERROR (_("only one type of list may be specified")); |
505 | @@ -775,6 +1094,14 @@ |
506 | spec_list_string = optarg; |
507 | break; |
508 | |
509 | + case 'c': |
510 | + /* Build the character list. */ |
511 | + if (operating_mode != undefined_mode) |
512 | + FATAL_ERROR (_("only one type of list may be specified")); |
513 | + operating_mode = character_mode; |
514 | + spec_list_string = optarg; |
515 | + break; |
516 | + |
517 | case 'f': |
518 | /* Build the field list. */ |
519 | if (operating_mode != undefined_mode) |
520 | @@ -786,10 +1113,35 @@ |
521 | case 'd': |
522 | /* New delimiter. */ |
523 | /* Interpret -d '' to mean `use the NUL byte as the delimiter.' */ |
524 | - if (optarg[0] != '\0' && optarg[1] != '\0') |
525 | - FATAL_ERROR (_("the delimiter must be a single character")); |
526 | - delim = optarg[0]; |
527 | - delim_specified = true; |
528 | +#if HAVE_MBRTOWC |
529 | + { |
530 | + if(MB_CUR_MAX > 1) |
531 | + { |
532 | + mbstate_t state; |
533 | + |
534 | + memset (&state, '\0', sizeof(mbstate_t)); |
535 | + delimlen = mbrtowc (&wcdelim, optarg, strnlen(optarg, MB_LEN_MAX), &state); |
536 | + |
537 | + if (delimlen == (size_t)-1 || delimlen == (size_t)-2) |
538 | + ++force_singlebyte_mode; |
539 | + else |
540 | + { |
541 | + delimlen = (delimlen < 1) ? 1 : delimlen; |
542 | + if (wcdelim != L'\0' && *(optarg + delimlen) != '\0') |
543 | + FATAL_ERROR (_("the delimiter must be a single character")); |
544 | + memcpy (mbdelim, optarg, delimlen); |
545 | + } |
546 | + } |
547 | + |
548 | + if (MB_CUR_MAX <= 1 || force_singlebyte_mode) |
549 | +#endif |
550 | + { |
551 | + if (optarg[0] != '\0' && optarg[1] != '\0') |
552 | + FATAL_ERROR (_("the delimiter must be a single character")); |
553 | + delim = (unsigned char) optarg[0]; |
554 | + } |
555 | + delim_specified = true; |
556 | + } |
557 | break; |
558 | |
559 | case OUTPUT_DELIMITER_OPTION: |
560 | @@ -802,6 +1154,7 @@ |
561 | break; |
562 | |
563 | case 'n': |
564 | + byte_mode_character_aware = 1; |
565 | break; |
566 | |
567 | case 's': |
568 | @@ -824,7 +1177,7 @@ |
569 | if (operating_mode == undefined_mode) |
570 | FATAL_ERROR (_("you must specify a list of bytes, characters, or fields")); |
571 | |
572 | - if (delim != '\0' && operating_mode != field_mode) |
573 | + if (delim_specified && operating_mode != field_mode) |
574 | FATAL_ERROR (_("an input delimiter may be specified only\ |
575 | when operating on fields")); |
576 | |
577 | @@ -851,15 +1204,34 @@ |
578 | } |
579 | |
580 | if (!delim_specified) |
581 | - delim = '\t'; |
582 | + { |
583 | + delim = '\t'; |
584 | +#ifdef HAVE_MBRTOWC |
585 | + wcdelim = L'\t'; |
586 | + mbdelim[0] = '\t'; |
587 | + mbdelim[1] = '\0'; |
588 | + delimlen = 1; |
589 | +#endif |
590 | + } |
591 | |
592 | if (output_delimiter_string == NULL) |
593 | { |
594 | - static char dummy[2]; |
595 | - dummy[0] = delim; |
596 | - dummy[1] = '\0'; |
597 | - output_delimiter_string = dummy; |
598 | - output_delimiter_length = 1; |
599 | +#ifdef HAVE_MBRTOWC |
600 | + if (MB_CUR_MAX > 1 && !force_singlebyte_mode) |
601 | + { |
602 | + output_delimiter_string = xstrdup(mbdelim); |
603 | + output_delimiter_length = delimlen; |
604 | + } |
605 | + |
606 | + if (MB_CUR_MAX <= 1 || force_singlebyte_mode) |
607 | +#endif |
608 | + { |
609 | + static char dummy[2]; |
610 | + dummy[0] = delim; |
611 | + dummy[1] = '\0'; |
612 | + output_delimiter_string = dummy; |
613 | + output_delimiter_length = 1; |
614 | + } |
615 | } |
616 | |
617 | if (optind == argc) |
618 | --- coreutils-5.93/src/pr.c.i18n 2005-09-16 08:50:33.000000000 +0100 |
619 | +++ coreutils-5.93/src/pr.c 2005-12-23 08:53:01.000000000 +0000 |
620 | @@ -313,6 +313,32 @@ |
621 | |
622 | #include <getopt.h> |
623 | #include <sys/types.h> |
624 | + |
625 | +/* Get MB_LEN_MAX. */ |
626 | +#include <limits.h> |
627 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
628 | + installation; work around this configuration error. */ |
629 | +#if !defined MB_LEN_MAX || MB_LEN_MAX == 1 |
630 | +# define MB_LEN_MAX 16 |
631 | +#endif |
632 | + |
633 | +/* Get MB_CUR_MAX. */ |
634 | +#include <stdlib.h> |
635 | + |
636 | +/* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */ |
637 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ |
638 | +#if HAVE_WCHAR_H |
639 | +# include <wchar.h> |
640 | +#endif |
641 | + |
642 | +/* Get iswprint(). -- for wcwidth(). */ |
643 | +#if HAVE_WCTYPE_H |
644 | +# include <wctype.h> |
645 | +#endif |
646 | +#if !defined iswprint && !HAVE_ISWPRINT |
647 | +# define iswprint(wc) 1 |
648 | +#endif |
649 | + |
650 | #include "system.h" |
651 | #include "error.h" |
652 | #include "hard-locale.h" |
653 | @@ -324,6 +350,18 @@ |
654 | #include "strftime.h" |
655 | #include "xstrtol.h" |
656 | |
657 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
658 | +#if HAVE_MBRTOWC && defined mbstate_t |
659 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
660 | +#endif |
661 | + |
662 | +#ifndef HAVE_DECL_WCWIDTH |
663 | +"this configure-time declaration test was not run" |
664 | +#endif |
665 | +#if !HAVE_DECL_WCWIDTH |
666 | +extern int wcwidth (); |
667 | +#endif |
668 | + |
669 | /* The official name of this program (e.g., no `g' prefix). */ |
670 | #define PROGRAM_NAME "pr" |
671 | |
672 | @@ -416,7 +454,20 @@ |
673 | |
674 | #define NULLCOL (COLUMN *)0 |
675 | |
676 | -static int char_to_clump (char c); |
677 | +/* Funtion pointers to switch functions for single byte locale or for |
678 | + multibyte locale. If multibyte functions do not exist in your sysytem, |
679 | + these pointers always point the function for single byte locale. */ |
680 | +static void (*print_char) (char c); |
681 | +static int (*char_to_clump) (char c); |
682 | + |
683 | +/* Functions for single byte locale. */ |
684 | +static void print_char_single (char c); |
685 | +static int char_to_clump_single (char c); |
686 | + |
687 | +/* Functions for multibyte locale. */ |
688 | +static void print_char_multi (char c); |
689 | +static int char_to_clump_multi (char c); |
690 | + |
691 | static bool read_line (COLUMN *p); |
692 | static bool print_page (void); |
693 | static bool print_stored (COLUMN *p); |
694 | @@ -426,6 +477,7 @@ |
695 | static void pad_across_to (int position); |
696 | static void add_line_number (COLUMN *p); |
697 | static void getoptarg (char *arg, char switch_char, char *character, |
698 | + int *character_length, int *character_width, |
699 | int *number); |
700 | void usage (int status); |
701 | static void print_files (int number_of_files, char **av); |
702 | @@ -440,7 +492,6 @@ |
703 | static void pad_down (int lines); |
704 | static void read_rest_of_line (COLUMN *p); |
705 | static void skip_read (COLUMN *p, int column_number); |
706 | -static void print_char (char c); |
707 | static void cleanup (void); |
708 | static void print_sep_string (void); |
709 | static void separator_string (const char *optarg_S); |
710 | @@ -455,7 +506,7 @@ |
711 | we store the leftmost columns contiguously in buff. |
712 | To print a line from buff, get the index of the first character |
713 | from line_vector[i], and print up to line_vector[i + 1]. */ |
714 | -static char *buff; |
715 | +static unsigned char *buff; |
716 | |
717 | /* Index of the position in buff where the next character |
718 | will be stored. */ |
719 | @@ -559,7 +610,7 @@ |
720 | static bool untabify_input = false; |
721 | |
722 | /* (-e) The input tab character. */ |
723 | -static char input_tab_char = '\t'; |
724 | +static char input_tab_char[MB_LEN_MAX] = "\t"; |
725 | |
726 | /* (-e) Tabstops are at chars_per_tab, 2*chars_per_tab, 3*chars_per_tab, ... |
727 | where the leftmost column is 1. */ |
728 | @@ -569,7 +620,10 @@ |
729 | static bool tabify_output = false; |
730 | |
731 | /* (-i) The output tab character. */ |
732 | -static char output_tab_char = '\t'; |
733 | +static char output_tab_char[MB_LEN_MAX] = "\t"; |
734 | + |
735 | +/* (-i) The byte length of output tab character. */ |
736 | +static int output_tab_char_length = 1; |
737 | |
738 | /* (-i) The width of the output tab. */ |
739 | static int chars_per_output_tab = 8; |
740 | @@ -643,7 +697,13 @@ |
741 | static bool numbered_lines = false; |
742 | |
743 | /* (-n) Character which follows each line number. */ |
744 | -static char number_separator = '\t'; |
745 | +static char number_separator[MB_LEN_MAX] = "\t"; |
746 | + |
747 | +/* (-n) The byte length of the character which follows each line number. */ |
748 | +static int number_separator_length = 1; |
749 | + |
750 | +/* (-n) The character width of the character which follows each line number. */ |
751 | +static int number_separator_width = 0; |
752 | |
753 | /* (-n) line counting starts with 1st line of input file (not with 1st |
754 | line of 1st page printed). */ |
755 | @@ -696,6 +756,7 @@ |
756 | -a|COLUMN|-m is a `space' and with the -J option a `tab'. */ |
757 | static char *col_sep_string = ""; |
758 | static int col_sep_length = 0; |
759 | +static int col_sep_width = 0; |
760 | static char *column_separator = " "; |
761 | static char *line_separator = "\t"; |
762 | |
763 | @@ -852,6 +913,13 @@ |
764 | col_sep_length = (int) strlen (optarg_S); |
765 | col_sep_string = xmalloc (col_sep_length + 1); |
766 | strcpy (col_sep_string, optarg_S); |
767 | + |
768 | +#if HAVE_MBRTOWC |
769 | + if (MB_CUR_MAX > 1) |
770 | + col_sep_width = mbswidth (col_sep_string, 0); |
771 | + else |
772 | +#endif |
773 | + col_sep_width = col_sep_length; |
774 | } |
775 | |
776 | int |
777 | @@ -877,6 +945,21 @@ |
778 | |
779 | atexit (close_stdout); |
780 | |
781 | +/* Define which functions are used, the ones for single byte locale or the ones |
782 | + for multibyte locale. */ |
783 | +#if HAVE_MBRTOWC |
784 | + if (MB_CUR_MAX > 1) |
785 | + { |
786 | + print_char = print_char_multi; |
787 | + char_to_clump = char_to_clump_multi; |
788 | + } |
789 | + else |
790 | +#endif |
791 | + { |
792 | + print_char = print_char_single; |
793 | + char_to_clump = char_to_clump_single; |
794 | + } |
795 | + |
796 | n_files = 0; |
797 | file_names = (argc > 1 |
798 | ? xmalloc ((argc - 1) * sizeof (char *)) |
799 | @@ -949,8 +1032,12 @@ |
800 | break; |
801 | case 'e': |
802 | if (optarg) |
803 | - getoptarg (optarg, 'e', &input_tab_char, |
804 | - &chars_per_input_tab); |
805 | + { |
806 | + int dummy_length, dummy_width; |
807 | + |
808 | + getoptarg (optarg, 'e', input_tab_char, &dummy_length, |
809 | + &dummy_width, &chars_per_input_tab); |
810 | + } |
811 | /* Could check tab width > 0. */ |
812 | untabify_input = true; |
813 | break; |
814 | @@ -963,8 +1050,12 @@ |
815 | break; |
816 | case 'i': |
817 | if (optarg) |
818 | - getoptarg (optarg, 'i', &output_tab_char, |
819 | - &chars_per_output_tab); |
820 | + { |
821 | + int dummy_width; |
822 | + |
823 | + getoptarg (optarg, 'i', output_tab_char, &output_tab_char_length, |
824 | + &dummy_width, &chars_per_output_tab); |
825 | + } |
826 | /* Could check tab width > 0. */ |
827 | tabify_output = true; |
828 | break; |
829 | @@ -991,8 +1082,8 @@ |
830 | case 'n': |
831 | numbered_lines = true; |
832 | if (optarg) |
833 | - getoptarg (optarg, 'n', &number_separator, |
834 | - &chars_per_number); |
835 | + getoptarg (optarg, 'n', number_separator, &number_separator_length, |
836 | + &number_separator_width, &chars_per_number); |
837 | break; |
838 | case 'N': |
839 | skip_count = false; |
840 | @@ -1031,7 +1122,7 @@ |
841 | old_s = false; |
842 | /* Reset an additional input of -s, -S dominates -s */ |
843 | col_sep_string = ""; |
844 | - col_sep_length = 0; |
845 | + col_sep_length = col_sep_width = 0; |
846 | use_col_separator = true; |
847 | if (optarg) |
848 | separator_string (optarg); |
849 | @@ -1188,10 +1279,45 @@ |
850 | a number. */ |
851 | |
852 | static void |
853 | -getoptarg (char *arg, char switch_char, char *character, int *number) |
854 | +getoptarg (char *arg, char switch_char, char *character, int *character_length, |
855 | + int *character_width, int *number) |
856 | { |
857 | if (!ISDIGIT (*arg)) |
858 | - *character = *arg++; |
859 | + { |
860 | +#ifdef HAVE_MBRTOWC |
861 | + if (MB_CUR_MAX > 1) /* for multibyte locale. */ |
862 | + { |
863 | + wchar_t wc; |
864 | + size_t mblength; |
865 | + int width; |
866 | + mbstate_t state = {'\0'}; |
867 | + |
868 | + mblength = mbrtowc (&wc, arg, strnlen(arg, MB_LEN_MAX), &state); |
869 | + |
870 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
871 | + { |
872 | + *character_length = 1; |
873 | + *character_width = 1; |
874 | + } |
875 | + else |
876 | + { |
877 | + *character_length = (mblength < 1) ? 1 : mblength; |
878 | + width = wcwidth (wc); |
879 | + *character_width = (width < 0) ? 0 : width; |
880 | + } |
881 | + |
882 | + strncpy (character, arg, *character_length); |
883 | + arg += *character_length; |
884 | + } |
885 | + else /* for single byte locale. */ |
886 | +#endif |
887 | + { |
888 | + *character = *arg++; |
889 | + *character_length = 1; |
890 | + *character_width = 1; |
891 | + } |
892 | + } |
893 | + |
894 | if (*arg) |
895 | { |
896 | long int tmp_long; |
897 | @@ -1256,7 +1382,7 @@ |
898 | else |
899 | col_sep_string = column_separator; |
900 | |
901 | - col_sep_length = 1; |
902 | + col_sep_length = col_sep_width = 1; |
903 | use_col_separator = true; |
904 | } |
905 | /* It's rather pointless to define a TAB separator with column |
906 | @@ -1288,11 +1414,11 @@ |
907 | TAB_WIDTH (chars_per_input_tab, chars_per_number); */ |
908 | |
909 | /* Estimate chars_per_text without any margin and keep it constant. */ |
910 | - if (number_separator == '\t') |
911 | + if (number_separator[0] == '\t') |
912 | number_width = chars_per_number + |
913 | TAB_WIDTH (chars_per_default_tab, chars_per_number); |
914 | else |
915 | - number_width = chars_per_number + 1; |
916 | + number_width = chars_per_number + number_separator_width; |
917 | |
918 | /* The number is part of the column width unless we are |
919 | printing files in parallel. */ |
920 | @@ -1307,7 +1433,7 @@ |
921 | } |
922 | |
923 | chars_per_column = (chars_per_line - chars_used_by_number - |
924 | - (columns - 1) * col_sep_length) / columns; |
925 | + (columns - 1) * col_sep_width) / columns; |
926 | |
927 | if (chars_per_column < 1) |
928 | error (EXIT_FAILURE, 0, _("page width too narrow")); |
929 | @@ -1432,7 +1558,7 @@ |
930 | |
931 | /* Enlarge p->start_position of first column to use the same form of |
932 | padding_not_printed with all columns. */ |
933 | - h = h + col_sep_length; |
934 | + h = h + col_sep_width; |
935 | |
936 | /* This loop takes care of all but the rightmost column. */ |
937 | |
938 | @@ -1466,7 +1592,7 @@ |
939 | } |
940 | else |
941 | { |
942 | - h = h_next + col_sep_length; |
943 | + h = h_next + col_sep_width; |
944 | h_next = h + chars_per_column; |
945 | } |
946 | } |
947 | @@ -1756,9 +1882,9 @@ |
948 | align_column (COLUMN *p) |
949 | { |
950 | padding_not_printed = p->start_position; |
951 | - if (padding_not_printed - col_sep_length > 0) |
952 | + if (padding_not_printed - col_sep_width > 0) |
953 | { |
954 | - pad_across_to (padding_not_printed - col_sep_length); |
955 | + pad_across_to (padding_not_printed - col_sep_width); |
956 | padding_not_printed = ANYWHERE; |
957 | } |
958 | |
959 | @@ -2029,13 +2155,13 @@ |
960 | /* May be too generous. */ |
961 | buff = X2REALLOC (buff, &buff_allocated); |
962 | } |
963 | - buff[buff_current++] = c; |
964 | + buff[buff_current++] = (unsigned char) c; |
965 | } |
966 | |
967 | static void |
968 | add_line_number (COLUMN *p) |
969 | { |
970 | - int i; |
971 | + int i, j; |
972 | char *s; |
973 | int left_cut; |
974 | |
975 | @@ -2058,22 +2184,24 @@ |
976 | /* Tabification is assumed for multiple columns, also for n-separators, |
977 | but `default n-separator = TAB' hasn't been given priority over |
978 | equal column_width also specified by POSIX. */ |
979 | - if (number_separator == '\t') |
980 | + if (number_separator[0] == '\t') |
981 | { |
982 | i = number_width - chars_per_number; |
983 | while (i-- > 0) |
984 | (p->char_func) (' '); |
985 | } |
986 | else |
987 | - (p->char_func) (number_separator); |
988 | + for (j = 0; j < number_separator_length; j++) |
989 | + (p->char_func) (number_separator[j]); |
990 | } |
991 | else |
992 | /* To comply with POSIX, we avoid any expansion of default TAB |
993 | separator with a single column output. No column_width requirement |
994 | has to be considered. */ |
995 | { |
996 | - (p->char_func) (number_separator); |
997 | - if (number_separator == '\t') |
998 | + for (j = 0; j < number_separator_length; j++) |
999 | + (p->char_func) (number_separator[j]); |
1000 | + if (number_separator[0] == '\t') |
1001 | output_position = POS_AFTER_TAB (chars_per_output_tab, |
1002 | output_position); |
1003 | } |
1004 | @@ -2234,7 +2362,7 @@ |
1005 | while (goal - h_old > 1 |
1006 | && (h_new = POS_AFTER_TAB (chars_per_output_tab, h_old)) <= goal) |
1007 | { |
1008 | - putchar (output_tab_char); |
1009 | + fwrite (output_tab_char, sizeof(char), output_tab_char_length, stdout); |
1010 | h_old = h_new; |
1011 | } |
1012 | while (++h_old <= goal) |
1013 | @@ -2254,6 +2382,7 @@ |
1014 | { |
1015 | char *s; |
1016 | int l = col_sep_length; |
1017 | + int not_space_flag; |
1018 | |
1019 | s = col_sep_string; |
1020 | |
1021 | @@ -2267,6 +2396,7 @@ |
1022 | { |
1023 | for (; separators_not_printed > 0; --separators_not_printed) |
1024 | { |
1025 | + not_space_flag = 0; |
1026 | while (l-- > 0) |
1027 | { |
1028 | /* 3 types of sep_strings: spaces only, spaces and chars, |
1029 | @@ -2280,12 +2410,15 @@ |
1030 | } |
1031 | else |
1032 | { |
1033 | + not_space_flag = 1; |
1034 | if (spaces_not_printed > 0) |
1035 | print_white_space (); |
1036 | putchar (*s++); |
1037 | - ++output_position; |
1038 | } |
1039 | } |
1040 | + if (not_space_flag) |
1041 | + output_position += col_sep_width; |
1042 | + |
1043 | /* sep_string ends with some spaces */ |
1044 | if (spaces_not_printed > 0) |
1045 | print_white_space (); |
1046 | @@ -2313,7 +2446,7 @@ |
1047 | required number of tabs and spaces. */ |
1048 | |
1049 | static void |
1050 | -print_char (char c) |
1051 | +print_char_single (char c) |
1052 | { |
1053 | if (tabify_output) |
1054 | { |
1055 | @@ -2337,6 +2470,74 @@ |
1056 | putchar (c); |
1057 | } |
1058 | |
1059 | +#ifdef HAVE_MBRTOWC |
1060 | +static void |
1061 | +print_char_multi (char c) |
1062 | +{ |
1063 | + static size_t mbc_pos = 0; |
1064 | + static unsigned char mbc[MB_LEN_MAX] = {'\0'}; |
1065 | + static mbstate_t state = {'\0'}; |
1066 | + mbstate_t state_bak; |
1067 | + wchar_t wc; |
1068 | + size_t mblength; |
1069 | + int width; |
1070 | + |
1071 | + if (tabify_output) |
1072 | + { |
1073 | + state_bak = state; |
1074 | + mbc[mbc_pos++] = (unsigned char)c; |
1075 | + mblength = mbrtowc (&wc, mbc, mbc_pos, &state); |
1076 | + |
1077 | + while (mbc_pos > 0) |
1078 | + { |
1079 | + switch (mblength) |
1080 | + { |
1081 | + case (size_t)-2: |
1082 | + state = state_bak; |
1083 | + return; |
1084 | + |
1085 | + case (size_t)-1: |
1086 | + state = state_bak; |
1087 | + ++output_position; |
1088 | + putchar (mbc[0]); |
1089 | + memmove (mbc, mbc + 1, MB_CUR_MAX - 1); |
1090 | + --mbc_pos; |
1091 | + break; |
1092 | + |
1093 | + case 0: |
1094 | + mblength = 1; |
1095 | + |
1096 | + default: |
1097 | + if (wc == L' ') |
1098 | + { |
1099 | + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); |
1100 | + --mbc_pos; |
1101 | + ++spaces_not_printed; |
1102 | + return; |
1103 | + } |
1104 | + else if (spaces_not_printed > 0) |
1105 | + print_white_space (); |
1106 | + |
1107 | + /* Nonprintables are assumed to have width 0, except L'\b'. */ |
1108 | + if ((width = wcwidth (wc)) < 1) |
1109 | + { |
1110 | + if (wc == L'\b') |
1111 | + --output_position; |
1112 | + } |
1113 | + else |
1114 | + output_position += width; |
1115 | + |
1116 | + fwrite (mbc, sizeof(char), mblength, stdout); |
1117 | + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); |
1118 | + mbc_pos -= mblength; |
1119 | + } |
1120 | + } |
1121 | + return; |
1122 | + } |
1123 | + putchar (c); |
1124 | +} |
1125 | +#endif |
1126 | + |
1127 | /* Skip to page PAGE before printing. |
1128 | PAGE may be larger than total number of pages. */ |
1129 | |
1130 | @@ -2517,9 +2718,9 @@ |
1131 | align_empty_cols = false; |
1132 | } |
1133 | |
1134 | - if (padding_not_printed - col_sep_length > 0) |
1135 | + if (padding_not_printed - col_sep_width > 0) |
1136 | { |
1137 | - pad_across_to (padding_not_printed - col_sep_length); |
1138 | + pad_across_to (padding_not_printed - col_sep_width); |
1139 | padding_not_printed = ANYWHERE; |
1140 | } |
1141 | |
1142 | @@ -2620,9 +2821,9 @@ |
1143 | } |
1144 | } |
1145 | |
1146 | - if (padding_not_printed - col_sep_length > 0) |
1147 | + if (padding_not_printed - col_sep_width > 0) |
1148 | { |
1149 | - pad_across_to (padding_not_printed - col_sep_length); |
1150 | + pad_across_to (padding_not_printed - col_sep_width); |
1151 | padding_not_printed = ANYWHERE; |
1152 | } |
1153 | |
1154 | @@ -2635,8 +2836,8 @@ |
1155 | if (spaces_not_printed == 0) |
1156 | { |
1157 | output_position = p->start_position + end_vector[line]; |
1158 | - if (p->start_position - col_sep_length == chars_per_margin) |
1159 | - output_position -= col_sep_length; |
1160 | + if (p->start_position - col_sep_width == chars_per_margin) |
1161 | + output_position -= col_sep_width; |
1162 | } |
1163 | |
1164 | return true; |
1165 | @@ -2655,7 +2856,7 @@ |
1166 | number of characters is 1.) */ |
1167 | |
1168 | static int |
1169 | -char_to_clump (char c) |
1170 | +char_to_clump_single (char c) |
1171 | { |
1172 | unsigned char uc = c; |
1173 | char *s = clump_buff; |
1174 | @@ -2665,10 +2866,10 @@ |
1175 | int chars; |
1176 | int chars_per_c = 8; |
1177 | |
1178 | - if (c == input_tab_char) |
1179 | + if (c == input_tab_char[0]) |
1180 | chars_per_c = chars_per_input_tab; |
1181 | |
1182 | - if (c == input_tab_char || c == '\t') |
1183 | + if (c == input_tab_char[0] || c == '\t') |
1184 | { |
1185 | width = TAB_WIDTH (chars_per_c, input_position); |
1186 | |
1187 | @@ -2739,6 +2940,154 @@ |
1188 | return chars; |
1189 | } |
1190 | |
1191 | +#ifdef HAVE_MBRTOWC |
1192 | +static int |
1193 | +char_to_clump_multi (char c) |
1194 | +{ |
1195 | + static size_t mbc_pos = 0; |
1196 | + static char mbc[MB_LEN_MAX] = {'\0'}; |
1197 | + static mbstate_t state = {'\0'}; |
1198 | + mbstate_t state_bak; |
1199 | + wchar_t wc; |
1200 | + size_t mblength; |
1201 | + int wc_width; |
1202 | + register int *s = clump_buff; |
1203 | + register int i, j; |
1204 | + char esc_buff[4]; |
1205 | + int width; |
1206 | + int chars; |
1207 | + int chars_per_c = 8; |
1208 | + |
1209 | + state_bak = state; |
1210 | + mbc[mbc_pos++] = c; |
1211 | + mblength = mbrtowc (&wc, mbc, mbc_pos, &state); |
1212 | + |
1213 | + width = 0; |
1214 | + chars = 0; |
1215 | + while (mbc_pos > 0) |
1216 | + { |
1217 | + switch (mblength) |
1218 | + { |
1219 | + case (size_t)-2: |
1220 | + state = state_bak; |
1221 | + return 0; |
1222 | + |
1223 | + case (size_t)-1: |
1224 | + state = state_bak; |
1225 | + mblength = 1; |
1226 | + |
1227 | + if (use_esc_sequence || use_cntrl_prefix) |
1228 | + { |
1229 | + width = +4; |
1230 | + chars = +4; |
1231 | + *s++ = '\\'; |
1232 | + sprintf (esc_buff, "%03o", mbc[0]); |
1233 | + for (i = 0; i <= 2; ++i) |
1234 | + *s++ = (int) esc_buff[i]; |
1235 | + } |
1236 | + else |
1237 | + { |
1238 | + width += 1; |
1239 | + chars += 1; |
1240 | + *s++ = mbc[0]; |
1241 | + } |
1242 | + break; |
1243 | + |
1244 | + case 0: |
1245 | + mblength = 1; |
1246 | + /* Fall through */ |
1247 | + |
1248 | + default: |
1249 | + if (memcmp (mbc, input_tab_char, mblength) == 0) |
1250 | + chars_per_c = chars_per_input_tab; |
1251 | + |
1252 | + if (memcmp (mbc, input_tab_char, mblength) == 0 || c == '\t') |
1253 | + { |
1254 | + int width_inc; |
1255 | + |
1256 | + width_inc = TAB_WIDTH (chars_per_c, input_position); |
1257 | + width += width_inc; |
1258 | + |
1259 | + if (untabify_input) |
1260 | + { |
1261 | + for (i = width_inc; i; --i) |
1262 | + *s++ = ' '; |
1263 | + chars += width_inc; |
1264 | + } |
1265 | + else |
1266 | + { |
1267 | + for (i = 0; i < mblength; i++) |
1268 | + *s++ = mbc[i]; |
1269 | + chars += mblength; |
1270 | + } |
1271 | + } |
1272 | + else if ((wc_width = wcwidth (wc)) < 1) |
1273 | + { |
1274 | + if (use_esc_sequence) |
1275 | + { |
1276 | + for (i = 0; i < mblength; i++) |
1277 | + { |
1278 | + width += 4; |
1279 | + chars += 4; |
1280 | + *s++ = '\\'; |
1281 | + sprintf (esc_buff, "%03o", c); |
1282 | + for (j = 0; j <= 2; ++j) |
1283 | + *s++ = (int) esc_buff[j]; |
1284 | + } |
1285 | + } |
1286 | + else if (use_cntrl_prefix) |
1287 | + { |
1288 | + if (wc < 0200) |
1289 | + { |
1290 | + width += 2; |
1291 | + chars += 2; |
1292 | + *s++ = '^'; |
1293 | + *s++ = wc ^ 0100; |
1294 | + } |
1295 | + else |
1296 | + { |
1297 | + for (i = 0; i < mblength; i++) |
1298 | + { |
1299 | + width += 4; |
1300 | + chars += 4; |
1301 | + *s++ = '\\'; |
1302 | + sprintf (esc_buff, "%03o", c); |
1303 | + for (j = 0; j <= 2; ++j) |
1304 | + *s++ = (int) esc_buff[j]; |
1305 | + } |
1306 | + } |
1307 | + } |
1308 | + else if (wc == L'\b') |
1309 | + { |
1310 | + width += -1; |
1311 | + chars += 1; |
1312 | + *s++ = c; |
1313 | + } |
1314 | + else |
1315 | + { |
1316 | + width += 0; |
1317 | + chars += mblength; |
1318 | + for (i = 0; i < mblength; i++) |
1319 | + *s++ = mbc[i]; |
1320 | + } |
1321 | + } |
1322 | + else |
1323 | + { |
1324 | + width += wc_width; |
1325 | + chars += mblength; |
1326 | + for (i = 0; i < mblength; i++) |
1327 | + *s++ = mbc[i]; |
1328 | + } |
1329 | + } |
1330 | + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); |
1331 | + mbc_pos -= mblength; |
1332 | + } |
1333 | + |
1334 | + input_position += width; |
1335 | + return chars; |
1336 | +} |
1337 | +#endif |
1338 | + |
1339 | /* We've just printed some files and need to clean up things before |
1340 | looking for more options and printing the next batch of files. |
1341 | |
1342 | --- coreutils-5.93/src/uniq.c.i18n 2005-07-05 07:32:54.000000000 +0100 |
1343 | +++ coreutils-5.93/src/uniq.c 2005-12-23 08:53:01.000000000 +0000 |
1344 | @@ -23,6 +23,16 @@ |
1345 | #include <getopt.h> |
1346 | #include <sys/types.h> |
1347 | |
1348 | +/* Get mbstate_t, mbrtowc(). */ |
1349 | +#if HAVE_WCHAR_H |
1350 | +# include <wchar.h> |
1351 | +#endif |
1352 | + |
1353 | +/* Get isw* functions. */ |
1354 | +#if HAVE_WCTYPE_H |
1355 | +# include <wctype.h> |
1356 | +#endif |
1357 | + |
1358 | #include "system.h" |
1359 | #include "argmatch.h" |
1360 | #include "linebuffer.h" |
1361 | @@ -32,7 +42,19 @@ |
1362 | #include "quote.h" |
1363 | #include "xmemcoll.h" |
1364 | #include "xstrtol.h" |
1365 | -#include "memcasecmp.h" |
1366 | +#include "xmemcoll.h" |
1367 | + |
1368 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
1369 | + installation; work around this configuration error. */ |
1370 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
1371 | +# define MB_LEN_MAX 16 |
1372 | +#endif |
1373 | + |
1374 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
1375 | +#if HAVE_MBRTOWC && defined mbstate_t |
1376 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
1377 | +#endif |
1378 | + |
1379 | |
1380 | /* The official name of this program (e.g., no `g' prefix). */ |
1381 | #define PROGRAM_NAME "uniq" |
1382 | @@ -109,6 +131,10 @@ |
1383 | /* Select whether/how to delimit groups of duplicate lines. */ |
1384 | static enum delimit_method delimit_groups; |
1385 | |
1386 | +/* Function pointers. */ |
1387 | +static char * |
1388 | +(*find_field) (struct linebuffer *line); |
1389 | + |
1390 | static struct option const longopts[] = |
1391 | { |
1392 | {"count", no_argument, NULL, 'c'}, |
1393 | @@ -189,7 +215,7 @@ |
1394 | return a pointer to the beginning of the line's field to be compared. */ |
1395 | |
1396 | static char * |
1397 | -find_field (const struct linebuffer *line) |
1398 | +find_field_uni (struct linebuffer *line) |
1399 | { |
1400 | size_t count; |
1401 | char *lp = line->buffer; |
1402 | @@ -210,6 +236,83 @@ |
1403 | return lp + i; |
1404 | } |
1405 | |
1406 | +#if HAVE_MBRTOWC |
1407 | + |
1408 | +# define MBCHAR_TO_WCHAR(WC, MBLENGTH, LP, POS, SIZE, STATEP, CONVFAIL) \ |
1409 | + do \ |
1410 | + { \ |
1411 | + mbstate_t state_bak; \ |
1412 | + \ |
1413 | + CONVFAIL = 0; \ |
1414 | + state_bak = *STATEP; \ |
1415 | + \ |
1416 | + MBLENGTH = mbrtowc (&WC, LP + POS, SIZE - POS, STATEP); \ |
1417 | + \ |
1418 | + switch (MBLENGTH) \ |
1419 | + { \ |
1420 | + case (size_t)-2: \ |
1421 | + case (size_t)-1: \ |
1422 | + *STATEP = state_bak; \ |
1423 | + CONVFAIL++; \ |
1424 | + /* Fall through */ \ |
1425 | + case 0: \ |
1426 | + MBLENGTH = 1; \ |
1427 | + } \ |
1428 | + } \ |
1429 | + while (0) |
1430 | + |
1431 | +static char * |
1432 | +find_field_multi (struct linebuffer *line) |
1433 | +{ |
1434 | + size_t count; |
1435 | + char *lp = line->buffer; |
1436 | + size_t size = line->length - 1; |
1437 | + size_t pos; |
1438 | + size_t mblength; |
1439 | + wchar_t wc; |
1440 | + mbstate_t *statep; |
1441 | + int convfail; |
1442 | + |
1443 | + pos = 0; |
1444 | + statep = &(line->state); |
1445 | + |
1446 | + /* skip fields. */ |
1447 | + for (count = 0; count < skip_fields && pos < size; count++) |
1448 | + { |
1449 | + while (pos < size) |
1450 | + { |
1451 | + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); |
1452 | + |
1453 | + if (convfail || !iswblank (wc)) |
1454 | + { |
1455 | + pos += mblength; |
1456 | + break; |
1457 | + } |
1458 | + pos += mblength; |
1459 | + } |
1460 | + |
1461 | + while (pos < size) |
1462 | + { |
1463 | + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); |
1464 | + |
1465 | + if (!convfail && iswblank (wc)) |
1466 | + break; |
1467 | + |
1468 | + pos += mblength; |
1469 | + } |
1470 | + } |
1471 | + |
1472 | + /* skip fields. */ |
1473 | + for (count = 0; count < skip_chars && pos < size; count++) |
1474 | + { |
1475 | + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); |
1476 | + pos += mblength; |
1477 | + } |
1478 | + |
1479 | + return lp + pos; |
1480 | +} |
1481 | +#endif |
1482 | + |
1483 | /* Return false if two strings OLD and NEW match, true if not. |
1484 | OLD and NEW point not to the beginnings of the lines |
1485 | but rather to the beginnings of the fields to compare. |
1486 | @@ -218,6 +321,8 @@ |
1487 | static bool |
1488 | different (char *old, char *new, size_t oldlen, size_t newlen) |
1489 | { |
1490 | + char *copy_old, *copy_new; |
1491 | + |
1492 | if (check_chars < oldlen) |
1493 | oldlen = check_chars; |
1494 | if (check_chars < newlen) |
1495 | @@ -225,14 +330,92 @@ |
1496 | |
1497 | if (ignore_case) |
1498 | { |
1499 | - /* FIXME: This should invoke strcoll somehow. */ |
1500 | - return oldlen != newlen || memcasecmp (old, new, oldlen); |
1501 | + size_t i; |
1502 | + |
1503 | + copy_old = alloca (oldlen + 1); |
1504 | + copy_new = alloca (oldlen + 1); |
1505 | + |
1506 | + for (i = 0; i < oldlen; i++) |
1507 | + { |
1508 | + copy_old[i] = toupper (old[i]); |
1509 | + copy_new[i] = toupper (new[i]); |
1510 | + } |
1511 | } |
1512 | - else if (hard_LC_COLLATE) |
1513 | - return xmemcoll (old, oldlen, new, newlen) != 0; |
1514 | else |
1515 | - return oldlen != newlen || memcmp (old, new, oldlen); |
1516 | + { |
1517 | + copy_old = (char *)old; |
1518 | + copy_new = (char *)new; |
1519 | + } |
1520 | + |
1521 | + return xmemcoll (copy_old, oldlen, copy_new, newlen); |
1522 | +} |
1523 | + |
1524 | +#if HAVE_MBRTOWC |
1525 | +static int |
1526 | +different_multi (const char *old, const char *new, size_t oldlen, size_t newlen, mbstate_t oldstate, mbstate_t newstate) |
1527 | +{ |
1528 | + size_t i, j, chars; |
1529 | + const char *str[2]; |
1530 | + char *copy[2]; |
1531 | + size_t len[2]; |
1532 | + mbstate_t state[2]; |
1533 | + size_t mblength; |
1534 | + wchar_t wc, uwc; |
1535 | + mbstate_t state_bak; |
1536 | + |
1537 | + str[0] = old; |
1538 | + str[1] = new; |
1539 | + len[0] = oldlen; |
1540 | + len[1] = newlen; |
1541 | + state[0] = oldstate; |
1542 | + state[1] = newstate; |
1543 | + |
1544 | + for (i = 0; i < 2; i++) |
1545 | + { |
1546 | + copy[i] = alloca (len[i] + 1); |
1547 | + |
1548 | + for (j = 0, chars = 0; j < len[i] && chars < check_chars; chars++) |
1549 | + { |
1550 | + state_bak = state[i]; |
1551 | + mblength = mbrtowc (&wc, str[i] + j, len[i] - j, &(state[i])); |
1552 | + |
1553 | + switch (mblength) |
1554 | + { |
1555 | + case (size_t)-1: |
1556 | + case (size_t)-2: |
1557 | + state[i] = state_bak; |
1558 | + /* Fall through */ |
1559 | + case 0: |
1560 | + mblength = 1; |
1561 | + break; |
1562 | + |
1563 | + default: |
1564 | + if (ignore_case) |
1565 | + { |
1566 | + uwc = towupper (wc); |
1567 | + |
1568 | + if (uwc != wc) |
1569 | + { |
1570 | + mbstate_t state_wc; |
1571 | + |
1572 | + memset (&state_wc, '\0', sizeof(mbstate_t)); |
1573 | + wcrtomb (copy[i] + j, uwc, &state_wc); |
1574 | + } |
1575 | + else |
1576 | + memcpy (copy[i] + j, str[i] + j, mblength); |
1577 | + } |
1578 | + else |
1579 | + memcpy (copy[i] + j, str[i] + j, mblength); |
1580 | + } |
1581 | + j += mblength; |
1582 | + } |
1583 | + copy[i][j] = '\0'; |
1584 | + len[i] = j; |
1585 | + } |
1586 | + |
1587 | + return xmemcoll (copy[0], len[0], copy[1], len[1]); |
1588 | } |
1589 | +#endif |
1590 | |
1591 | /* Output the line in linebuffer LINE to standard output |
1592 | provided that the switches say it should be output. |
1593 | @@ -286,15 +469,43 @@ |
1594 | { |
1595 | char *prevfield IF_LINT (= NULL); |
1596 | size_t prevlen IF_LINT (= 0); |
1597 | +#if HAVE_MBRTOWC |
1598 | + mbstate_t prevstate; |
1599 | + |
1600 | + memset (&prevstate, '\0', sizeof (mbstate_t)); |
1601 | +#endif |
1602 | |
1603 | while (!feof (stdin)) |
1604 | { |
1605 | char *thisfield; |
1606 | size_t thislen; |
1607 | +#if HAVE_MBRTOWC |
1608 | + mbstate_t thisstate; |
1609 | +#endif |
1610 | + |
1611 | if (readlinebuffer (thisline, stdin) == 0) |
1612 | break; |
1613 | thisfield = find_field (thisline); |
1614 | thislen = thisline->length - 1 - (thisfield - thisline->buffer); |
1615 | +#if HAVE_MBRTOWC |
1616 | + if (MB_CUR_MAX > 1) |
1617 | + { |
1618 | + thisstate = thisline->state; |
1619 | + |
1620 | + if (prevline->length == 0 || different_multi |
1621 | + (thisfield, prevfield, thislen, prevlen, thisstate, prevstate)) |
1622 | + { |
1623 | + fwrite (thisline->buffer, sizeof (char), |
1624 | + thisline->length, stdout); |
1625 | + |
1626 | + SWAP_LINES (prevline, thisline); |
1627 | + prevfield = thisfield; |
1628 | + prevlen = thislen; |
1629 | + prevstate = thisstate; |
1630 | + } |
1631 | + } |
1632 | + else |
1633 | +#endif |
1634 | if (prevline->length == 0 |
1635 | || different (thisfield, prevfield, thislen, prevlen)) |
1636 | { |
1637 | @@ -313,17 +524,26 @@ |
1638 | size_t prevlen; |
1639 | uintmax_t match_count = 0; |
1640 | bool first_delimiter = true; |
1641 | +#if HAVE_MBRTOWC |
1642 | + mbstate_t prevstate; |
1643 | +#endif |
1644 | |
1645 | if (readlinebuffer (prevline, stdin) == 0) |
1646 | goto closefiles; |
1647 | prevfield = find_field (prevline); |
1648 | prevlen = prevline->length - 1 - (prevfield - prevline->buffer); |
1649 | +#if HAVE_MBRTOWC |
1650 | + prevstate = prevline->state; |
1651 | +#endif |
1652 | |
1653 | while (!feof (stdin)) |
1654 | { |
1655 | bool match; |
1656 | char *thisfield; |
1657 | size_t thislen; |
1658 | +#if HAVE_MBRTOWC |
1659 | + mbstate_t thisstate; |
1660 | +#endif |
1661 | if (readlinebuffer (thisline, stdin) == 0) |
1662 | { |
1663 | if (ferror (stdin)) |
1664 | @@ -332,6 +552,15 @@ |
1665 | } |
1666 | thisfield = find_field (thisline); |
1667 | thislen = thisline->length - 1 - (thisfield - thisline->buffer); |
1668 | +#if HAVE_MBRTOWC |
1669 | + if (MB_CUR_MAX > 1) |
1670 | + { |
1671 | + thisstate = thisline->state; |
1672 | + match = !different_multi (thisfield, prevfield, |
1673 | + thislen, prevlen, thisstate, prevstate); |
1674 | + } |
1675 | + else |
1676 | +#endif |
1677 | match = !different (thisfield, prevfield, thislen, prevlen); |
1678 | match_count += match; |
1679 | |
1680 | @@ -364,6 +593,9 @@ |
1681 | SWAP_LINES (prevline, thisline); |
1682 | prevfield = thisfield; |
1683 | prevlen = thislen; |
1684 | +#if HAVE_MBRTOWC |
1685 | + prevstate = thisstate; |
1686 | +#endif |
1687 | if (!match) |
1688 | match_count = 0; |
1689 | } |
1690 | @@ -408,6 +640,19 @@ |
1691 | |
1692 | atexit (close_stdout); |
1693 | |
1694 | +#if HAVE_MBRTOWC |
1695 | + if (MB_CUR_MAX > 1) |
1696 | + { |
1697 | + find_field = find_field_multi; |
1698 | + } |
1699 | + else |
1700 | +#endif |
1701 | + { |
1702 | + find_field = find_field_uni; |
1703 | + } |
1704 | + |
1705 | + |
1706 | + |
1707 | skip_chars = 0; |
1708 | skip_fields = 0; |
1709 | check_chars = SIZE_MAX; |
1710 | --- coreutils-5.93/src/expand.c.i18n 2005-08-12 08:16:25.000000000 +0100 |
1711 | +++ coreutils-5.93/src/expand.c 2005-12-23 08:53:01.000000000 +0000 |
1712 | @@ -38,11 +38,28 @@ |
1713 | #include <stdio.h> |
1714 | #include <getopt.h> |
1715 | #include <sys/types.h> |
1716 | + |
1717 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ |
1718 | +#if HAVE_WCHAR_H |
1719 | +# include <wchar.h> |
1720 | +#endif |
1721 | + |
1722 | #include "system.h" |
1723 | #include "error.h" |
1724 | #include "quote.h" |
1725 | #include "xstrndup.h" |
1726 | |
1727 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
1728 | + installation; work around this configuration error. */ |
1729 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
1730 | +# define MB_LEN_MAX 16 |
1731 | +#endif |
1732 | + |
1733 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
1734 | +#if HAVE_MBRTOWC && defined mbstate_t |
1735 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
1736 | +#endif |
1737 | + |
1738 | /* The official name of this program (e.g., no `g' prefix). */ |
1739 | #define PROGRAM_NAME "expand" |
1740 | |
1741 | @@ -364,6 +382,142 @@ |
1742 | } |
1743 | } |
1744 | |
1745 | +#if HAVE_MBRTOWC |
1746 | +static void |
1747 | +expand_multibyte (void) |
1748 | +{ |
1749 | + FILE *fp; /* Input strem. */ |
1750 | + mbstate_t i_state; /* Current shift state of the input stream. */ |
1751 | + mbstate_t i_state_bak; /* Back up the I_STATE. */ |
1752 | + mbstate_t o_state; /* Current shift state of the output stream. */ |
1753 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
1754 | + char *bufpos; /* Next read position of BUF. */ |
1755 | + size_t buflen = 0; /* The length of the byte sequence in buf. */ |
1756 | + wchar_t wc; /* A gotten wide character. */ |
1757 | + size_t mblength; /* The byte size of a multibyte character |
1758 | + which shows as same character as WC. */ |
1759 | + int tab_index = 0; /* Index in `tab_list' of next tabstop. */ |
1760 | + int column = 0; /* Column on screen of the next char. */ |
1761 | + int next_tab_column; /* Column the next tab stop is on. */ |
1762 | + int convert = 1; /* If nonzero, perform translations. */ |
1763 | + |
1764 | + fp = next_file ((FILE *) NULL); |
1765 | + if (fp == NULL) |
1766 | + return; |
1767 | + |
1768 | + memset (&o_state, '\0', sizeof(mbstate_t)); |
1769 | + memset (&i_state, '\0', sizeof(mbstate_t)); |
1770 | + |
1771 | + for (;;) |
1772 | + { |
1773 | + /* Refill the buffer BUF. */ |
1774 | + if (buflen < MB_LEN_MAX && !feof(fp) && !ferror(fp)) |
1775 | + { |
1776 | + memmove (buf, bufpos, buflen); |
1777 | + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, fp); |
1778 | + bufpos = buf; |
1779 | + } |
1780 | + |
1781 | + /* No character is left in BUF. */ |
1782 | + if (buflen < 1) |
1783 | + { |
1784 | + fp = next_file (fp); |
1785 | + |
1786 | + if (fp == NULL) |
1787 | + break; /* No more files. */ |
1788 | + else |
1789 | + { |
1790 | + memset (&i_state, '\0', sizeof(mbstate_t)); |
1791 | + continue; |
1792 | + } |
1793 | + } |
1794 | + |
1795 | + /* Get a wide character. */ |
1796 | + i_state_bak = i_state; |
1797 | + mblength = mbrtowc (&wc, bufpos, buflen, &i_state); |
1798 | + |
1799 | + switch (mblength) |
1800 | + { |
1801 | + case (size_t)-1: /* illegal byte sequence. */ |
1802 | + case (size_t)-2: |
1803 | + mblength = 1; |
1804 | + i_state = i_state_bak; |
1805 | + if (convert) |
1806 | + { |
1807 | + ++column; |
1808 | + if (convert_entire_line == 0) |
1809 | + convert = 0; |
1810 | + } |
1811 | + putchar (*bufpos); |
1812 | + break; |
1813 | + |
1814 | + case 0: /* null. */ |
1815 | + mblength = 1; |
1816 | + if (convert && convert_entire_line == 0) |
1817 | + convert = 0; |
1818 | + putchar ('\0'); |
1819 | + break; |
1820 | + |
1821 | + default: |
1822 | + if (wc == L'\n') /* LF. */ |
1823 | + { |
1824 | + tab_index = 0; |
1825 | + column = 0; |
1826 | + convert = 1; |
1827 | + putchar ('\n'); |
1828 | + } |
1829 | + else if (wc == L'\t' && convert) /* Tab. */ |
1830 | + { |
1831 | + if (tab_size == 0) |
1832 | + { |
1833 | + /* Do not let tab_index == first_free_tab; |
1834 | + stop when it is 1 less. */ |
1835 | + while (tab_index < first_free_tab - 1 |
1836 | + && column >= tab_list[tab_index]) |
1837 | + tab_index++; |
1838 | + next_tab_column = tab_list[tab_index]; |
1839 | + if (tab_index < first_free_tab - 1) |
1840 | + tab_index++; |
1841 | + if (column >= next_tab_column) |
1842 | + next_tab_column = column + 1; |
1843 | + } |
1844 | + else |
1845 | + next_tab_column = column + tab_size - column % tab_size; |
1846 | + |
1847 | + while (column < next_tab_column) |
1848 | + { |
1849 | + putchar (' '); |
1850 | + ++column; |
1851 | + } |
1852 | + } |
1853 | + else /* Others. */ |
1854 | + { |
1855 | + if (convert) |
1856 | + { |
1857 | + if (wc == L'\b') |
1858 | + { |
1859 | + if (column > 0) |
1860 | + --column; |
1861 | + } |
1862 | + else |
1863 | + { |
1864 | + int width; /* The width of WC. */ |
1865 | + |
1866 | + width = wcwidth (wc); |
1867 | + column += (width > 0) ? width : 0; |
1868 | + if (convert_entire_line == 0) |
1869 | + convert = 0; |
1870 | + } |
1871 | + } |
1872 | + fwrite (bufpos, sizeof(char), mblength, stdout); |
1873 | + } |
1874 | + } |
1875 | + buflen -= mblength; |
1876 | + bufpos += mblength; |
1877 | + } |
1878 | +} |
1879 | +#endif |
1880 | + |
1881 | int |
1882 | main (int argc, char **argv) |
1883 | { |
1884 | @@ -428,7 +582,12 @@ |
1885 | |
1886 | file_list = (optind < argc ? &argv[optind] : stdin_argv); |
1887 | |
1888 | - expand (); |
1889 | +#if HAVE_MBRTOWC |
1890 | + if (MB_CUR_MAX > 1) |
1891 | + expand_multibyte (); |
1892 | + else |
1893 | +#endif |
1894 | + expand (); |
1895 | |
1896 | if (have_read_stdin && fclose (stdin) != 0) |
1897 | error (EXIT_FAILURE, errno, "-"); |
1898 | --- coreutils-5.93/src/fold.c.i18n 2005-08-12 08:29:38.000000000 +0100 |
1899 | +++ coreutils-5.93/src/fold.c 2005-12-23 08:53:01.000000000 +0000 |
1900 | @@ -23,11 +23,33 @@ |
1901 | #include <getopt.h> |
1902 | #include <sys/types.h> |
1903 | |
1904 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ |
1905 | +#if HAVE_WCHAR_H |
1906 | +# include <wchar.h> |
1907 | +#endif |
1908 | + |
1909 | +/* Get iswprint(), iswblank(), wcwidth(). */ |
1910 | +#if HAVE_WCTYPE_H |
1911 | +# include <wctype.h> |
1912 | +#endif |
1913 | + |
1914 | #include "system.h" |
1915 | #include "error.h" |
1916 | #include "quote.h" |
1917 | #include "xstrtol.h" |
1918 | |
1919 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
1920 | + installation; work around this configuration error. */ |
1921 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
1922 | +# undef MB_LEN_MAX |
1923 | +# define MB_LEN_MAX 16 |
1924 | +#endif |
1925 | + |
1926 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
1927 | +#if HAVE_MBRTOWC && defined mbstate_t |
1928 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
1929 | +#endif |
1930 | + |
1931 | #define TAB_WIDTH 8 |
1932 | |
1933 | /* The official name of this program (e.g., no `g' prefix). */ |
1934 | @@ -35,23 +57,44 @@ |
1935 | |
1936 | #define AUTHORS "David MacKenzie" |
1937 | |
1938 | +#define FATAL_ERROR(Message) \ |
1939 | + do \ |
1940 | + { \ |
1941 | + error (0, 0, (Message)); \ |
1942 | + usage (2); \ |
1943 | + } \ |
1944 | + while (0) |
1945 | + |
1946 | +enum operating_mode |
1947 | +{ |
1948 | + /* Fold texts by columns that are at the given positions. */ |
1949 | + column_mode, |
1950 | + |
1951 | + /* Fold texts by bytes that are at the given positions. */ |
1952 | + byte_mode, |
1953 | + |
1954 | + /* Fold texts by characters that are at the given positions. */ |
1955 | + character_mode, |
1956 | +}; |
1957 | + |
1958 | /* The name this program was run with. */ |
1959 | char *program_name; |
1960 | |
1961 | +/* The argument shows current mode. (Default: column_mode) */ |
1962 | +static enum operating_mode operating_mode; |
1963 | + |
1964 | /* If nonzero, try to break on whitespace. */ |
1965 | static bool break_spaces; |
1966 | |
1967 | -/* If nonzero, count bytes, not column positions. */ |
1968 | -static bool count_bytes; |
1969 | - |
1970 | /* If nonzero, at least one of the files we read was standard input. */ |
1971 | static bool have_read_stdin; |
1972 | |
1973 | -static char const shortopts[] = "bsw:0::1::2::3::4::5::6::7::8::9::"; |
1974 | +static char const shortopts[] = "bcsw:0::1::2::3::4::5::6::7::8::9::"; |
1975 | |
1976 | static struct option const longopts[] = |
1977 | { |
1978 | {"bytes", no_argument, NULL, 'b'}, |
1979 | + {"characters", no_argument, NULL, 'c'}, |
1980 | {"spaces", no_argument, NULL, 's'}, |
1981 | {"width", required_argument, NULL, 'w'}, |
1982 | {GETOPT_HELP_OPTION_DECL}, |
1983 | @@ -81,6 +124,7 @@ |
1984 | "), stdout); |
1985 | fputs (_("\ |
1986 | -b, --bytes count bytes rather than columns\n\ |
1987 | + -c, --characters count characters rather than columns\n\ |
1988 | -s, --spaces break at spaces\n\ |
1989 | -w, --width=WIDTH use WIDTH columns instead of 80\n\ |
1990 | "), stdout); |
1991 | @@ -98,7 +142,7 @@ |
1992 | static size_t |
1993 | adjust_column (size_t column, char c) |
1994 | { |
1995 | - if (!count_bytes) |
1996 | + if (operating_mode != byte_mode) |
1997 | { |
1998 | if (c == '\b') |
1999 | { |
2000 | @@ -117,35 +161,14 @@ |
2001 | return column; |
2002 | } |
2003 | |
2004 | -/* Fold file FILENAME, or standard input if FILENAME is "-", |
2005 | - to stdout, with maximum line length WIDTH. |
2006 | - Return true if successful. */ |
2007 | - |
2008 | -static bool |
2009 | -fold_file (char *filename, size_t width) |
2010 | +static void |
2011 | +fold_text (FILE *istream, size_t width, int *saved_errno) |
2012 | { |
2013 | - FILE *istream; |
2014 | int c; |
2015 | size_t column = 0; /* Screen column where next char will go. */ |
2016 | size_t offset_out = 0; /* Index in `line_out' for next char. */ |
2017 | static char *line_out = NULL; |
2018 | static size_t allocated_out = 0; |
2019 | - int saved_errno; |
2020 | - |
2021 | - if (STREQ (filename, "-")) |
2022 | - { |
2023 | - istream = stdin; |
2024 | - have_read_stdin = true; |
2025 | - } |
2026 | - else |
2027 | - istream = fopen (filename, "r"); |
2028 | - |
2029 | - if (istream == NULL) |
2030 | - { |
2031 | - error (0, errno, "%s", filename); |
2032 | - return false; |
2033 | - } |
2034 | - |
2035 | while ((c = getc (istream)) != EOF) |
2036 | { |
2037 | if (offset_out + 1 >= allocated_out) |
2038 | @@ -172,6 +195,15 @@ |
2039 | bool found_blank = false; |
2040 | size_t logical_end = offset_out; |
2041 | |
2042 | + /* If LINE_OUT has no wide character, |
2043 | + put a new wide character in LINE_OUT |
2044 | + if column is bigger than width. */ |
2045 | + if (offset_out == 0) |
2046 | + { |
2047 | + line_out[offset_out++] = c; |
2048 | + continue; |
2049 | + } |
2050 | + |
2051 | /* Look for the last blank. */ |
2052 | while (logical_end) |
2053 | { |
2054 | @@ -218,11 +250,225 @@ |
2055 | line_out[offset_out++] = c; |
2056 | } |
2057 | |
2058 | - saved_errno = errno; |
2059 | + *saved_errno = errno; |
2060 | + |
2061 | + if (offset_out) |
2062 | + fwrite (line_out, sizeof (char), (size_t) offset_out, stdout); |
2063 | + |
2064 | + free(line_out); |
2065 | +} |
2066 | + |
2067 | +#if HAVE_MBRTOWC |
2068 | +static void |
2069 | +fold_multibyte_text (FILE *istream, int width, int *saved_errno) |
2070 | +{ |
2071 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
2072 | + size_t buflen = 0; /* The length of the byte sequence in buf. */ |
2073 | + char *bufpos; /* Next read position of BUF. */ |
2074 | + wint_t wc; /* A gotten wide character. */ |
2075 | + size_t mblength; /* The byte size of a multibyte character which shows |
2076 | + as same character as WC. */ |
2077 | + mbstate_t state, state_bak; /* State of the stream. */ |
2078 | + int convfail; /* 1, when conversion is failed. Otherwise 0. */ |
2079 | + |
2080 | + char *line_out = NULL; |
2081 | + size_t offset_out = 0; /* Index in `line_out' for next char. */ |
2082 | + size_t allocated_out = 0; |
2083 | + |
2084 | + int increment; |
2085 | + size_t column = 0; |
2086 | + |
2087 | + size_t last_blank_pos; |
2088 | + size_t last_blank_column; |
2089 | + int is_blank_seen; |
2090 | + int last_blank_increment; |
2091 | + int is_bs_following_last_blank; |
2092 | + size_t bs_following_last_blank_num; |
2093 | + int is_cr_after_last_blank; |
2094 | + |
2095 | +#define CLEAR_FLAGS \ |
2096 | + do \ |
2097 | + { \ |
2098 | + last_blank_pos = 0; \ |
2099 | + last_blank_column = 0; \ |
2100 | + is_blank_seen = 0; \ |
2101 | + is_bs_following_last_blank = 0; \ |
2102 | + bs_following_last_blank_num = 0; \ |
2103 | + is_cr_after_last_blank = 0; \ |
2104 | + } \ |
2105 | + while (0) |
2106 | + |
2107 | +#define START_NEW_LINE \ |
2108 | + do \ |
2109 | + { \ |
2110 | + putchar ('\n'); \ |
2111 | + column = 0; \ |
2112 | + offset_out = 0; \ |
2113 | + CLEAR_FLAGS; \ |
2114 | + } \ |
2115 | + while (0) |
2116 | + |
2117 | + CLEAR_FLAGS; |
2118 | + memset (&state, '\0', sizeof(mbstate_t)); |
2119 | + |
2120 | + for (;; bufpos += mblength, buflen -= mblength) |
2121 | + { |
2122 | + if (buflen < MB_LEN_MAX && !feof (istream) && !ferror (istream)) |
2123 | + { |
2124 | + memmove (buf, bufpos, buflen); |
2125 | + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, istream); |
2126 | + bufpos = buf; |
2127 | + } |
2128 | + |
2129 | + if (buflen < 1) |
2130 | + break; |
2131 | + |
2132 | + /* Get a wide character. */ |
2133 | + convfail = 0; |
2134 | + state_bak = state; |
2135 | + mblength = mbrtowc ((wchar_t *)&wc, bufpos, buflen, &state); |
2136 | + |
2137 | + switch (mblength) |
2138 | + { |
2139 | + case (size_t)-1: |
2140 | + case (size_t)-2: |
2141 | + convfail++; |
2142 | + state = state_bak; |
2143 | + /* Fall through. */ |
2144 | + |
2145 | + case 0: |
2146 | + mblength = 1; |
2147 | + break; |
2148 | + } |
2149 | + |
2150 | +rescan: |
2151 | + if (operating_mode == byte_mode) /* byte mode */ |
2152 | + increment = mblength; |
2153 | + else if (operating_mode == character_mode) /* character mode */ |
2154 | + increment = 1; |
2155 | + else /* column mode */ |
2156 | + { |
2157 | + if (convfail) |
2158 | + increment = 1; |
2159 | + else |
2160 | + { |
2161 | + switch (wc) |
2162 | + { |
2163 | + case L'\n': |
2164 | + fwrite (line_out, sizeof(char), offset_out, stdout); |
2165 | + START_NEW_LINE; |
2166 | + continue; |
2167 | + |
2168 | + case L'\b': |
2169 | + increment = (column > 0) ? -1 : 0; |
2170 | + break; |
2171 | + |
2172 | + case L'\r': |
2173 | + increment = -1 * column; |
2174 | + break; |
2175 | + |
2176 | + case L'\t': |
2177 | + increment = 8 - column % 8; |
2178 | + break; |
2179 | + |
2180 | + default: |
2181 | + increment = wcwidth (wc); |
2182 | + increment = (increment < 0) ? 0 : increment; |
2183 | + } |
2184 | + } |
2185 | + } |
2186 | + |
2187 | + if (column + increment > width && break_spaces && last_blank_pos) |
2188 | + { |
2189 | + fwrite (line_out, sizeof(char), last_blank_pos, stdout); |
2190 | + putchar ('\n'); |
2191 | + |
2192 | + offset_out = offset_out - last_blank_pos; |
2193 | + column = column - last_blank_column + ((is_cr_after_last_blank) |
2194 | + ? last_blank_increment : bs_following_last_blank_num); |
2195 | + memmove (line_out, line_out + last_blank_pos, offset_out); |
2196 | + CLEAR_FLAGS; |
2197 | + goto rescan; |
2198 | + } |
2199 | + |
2200 | + if (column + increment > width && column != 0) |
2201 | + { |
2202 | + fwrite (line_out, sizeof(char), offset_out, stdout); |
2203 | + START_NEW_LINE; |
2204 | + goto rescan; |
2205 | + } |
2206 | + |
2207 | + if (allocated_out < offset_out + mblength) |
2208 | + { |
2209 | + allocated_out += 1024; |
2210 | + line_out = xrealloc (line_out, allocated_out); |
2211 | + } |
2212 | + |
2213 | + memcpy (line_out + offset_out, bufpos, mblength); |
2214 | + offset_out += mblength; |
2215 | + column += increment; |
2216 | + |
2217 | + if (is_blank_seen && !convfail && wc == L'\r') |
2218 | + is_cr_after_last_blank = 1; |
2219 | + |
2220 | + if (is_bs_following_last_blank && !convfail && wc == L'\b') |
2221 | + ++bs_following_last_blank_num; |
2222 | + else |
2223 | + is_bs_following_last_blank = 0; |
2224 | + |
2225 | + if (break_spaces && !convfail && iswblank (wc)) |
2226 | + { |
2227 | + last_blank_pos = offset_out; |
2228 | + last_blank_column = column; |
2229 | + is_blank_seen = 1; |
2230 | + last_blank_increment = increment; |
2231 | + is_bs_following_last_blank = 1; |
2232 | + bs_following_last_blank_num = 0; |
2233 | + is_cr_after_last_blank = 0; |
2234 | + } |
2235 | + } |
2236 | + |
2237 | + *saved_errno = errno; |
2238 | |
2239 | if (offset_out) |
2240 | fwrite (line_out, sizeof (char), (size_t) offset_out, stdout); |
2241 | |
2242 | + free(line_out); |
2243 | +} |
2244 | +#endif |
2245 | + |
2246 | +/* Fold file FILENAME, or standard input if FILENAME is "-", |
2247 | + to stdout, with maximum line length WIDTH. |
2248 | + Return 0 if successful, 1 if an error occurs. */ |
2249 | + |
2250 | +static int |
2251 | +fold_file (char *filename, int width) |
2252 | +{ |
2253 | + FILE *istream; |
2254 | + int saved_errno; |
2255 | + |
2256 | + if (STREQ (filename, "-")) |
2257 | + { |
2258 | + istream = stdin; |
2259 | + have_read_stdin = 1; |
2260 | + } |
2261 | + else |
2262 | + istream = fopen (filename, "r"); |
2263 | + |
2264 | + if (istream == NULL) |
2265 | + { |
2266 | + error (0, errno, "%s", filename); |
2267 | + return 1; |
2268 | + } |
2269 | + |
2270 | + /* Define how ISTREAM is being folded. */ |
2271 | +#if HAVE_MBRTOWC |
2272 | + if (MB_CUR_MAX > 1) |
2273 | + fold_multibyte_text (istream, width, &saved_errno); |
2274 | + else |
2275 | +#endif |
2276 | + fold_text (istream, width, &saved_errno); |
2277 | + |
2278 | if (ferror (istream)) |
2279 | { |
2280 | error (0, saved_errno, "%s", filename); |
2281 | @@ -255,7 +501,8 @@ |
2282 | |
2283 | atexit (close_stdout); |
2284 | |
2285 | - break_spaces = count_bytes = have_read_stdin = false; |
2286 | + operating_mode = column_mode; |
2287 | + break_spaces = have_read_stdin = false; |
2288 | |
2289 | while ((optc = getopt_long (argc, argv, shortopts, longopts, NULL)) != -1) |
2290 | { |
2291 | @@ -264,7 +511,15 @@ |
2292 | switch (optc) |
2293 | { |
2294 | case 'b': /* Count bytes rather than columns. */ |
2295 | - count_bytes = true; |
2296 | + if (operating_mode != column_mode) |
2297 | + FATAL_ERROR (_("only one way of folding may be specified")); |
2298 | + operating_mode = byte_mode; |
2299 | + break; |
2300 | + |
2301 | + case 'c': |
2302 | + if (operating_mode != column_mode) |
2303 | + FATAL_ERROR (_("only one way of folding may be specified")); |
2304 | + operating_mode = character_mode; |
2305 | break; |
2306 | |
2307 | case 's': /* Break at word boundaries. */ |
2308 | --- coreutils-5.93/src/join.c.i18n 2005-08-12 08:16:25.000000000 +0100 |
2309 | +++ coreutils-5.93/src/join.c 2005-12-23 08:53:01.000000000 +0000 |
2310 | @@ -23,16 +23,30 @@ |
2311 | #include <sys/types.h> |
2312 | #include <getopt.h> |
2313 | |
2314 | +/* Get mbstate_t, mbrtowc(), mbrtowc(), wcwidth(). */ |
2315 | +#if HAVE_WCHAR_H |
2316 | +# include <wchar.h> |
2317 | +#endif |
2318 | + |
2319 | +/* Get iswblank(), towupper. */ |
2320 | +#if HAVE_WCTYPE_H |
2321 | +# include <wctype.h> |
2322 | +#endif |
2323 | + |
2324 | #include "system.h" |
2325 | #include "error.h" |
2326 | #include "hard-locale.h" |
2327 | #include "linebuffer.h" |
2328 | -#include "memcasecmp.h" |
2329 | #include "quote.h" |
2330 | #include "stdio--.h" |
2331 | #include "xmemcoll.h" |
2332 | #include "xstrtol.h" |
2333 | |
2334 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
2335 | +#if HAVE_MBRTOWC && defined mbstate_t |
2336 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
2337 | +#endif |
2338 | + |
2339 | /* The official name of this program (e.g., no `g' prefix). */ |
2340 | #define PROGRAM_NAME "join" |
2341 | |
2342 | @@ -104,10 +118,12 @@ |
2343 | /* Last element in `outlist', where a new element can be added. */ |
2344 | static struct outlist *outlist_end = &outlist_head; |
2345 | |
2346 | -/* Tab character separating fields. If negative, fields are separated |
2347 | - by any nonempty string of blanks, otherwise by exactly one |
2348 | - tab character whose value (when cast to unsigned char) equals TAB. */ |
2349 | -static int tab = -1; |
2350 | +/* Tab character separating fields. If NULL, fields are separated |
2351 | + by any nonempty string of blanks. */ |
2352 | +static char *tab = NULL; |
2353 | + |
2354 | +/* The number of bytes used for tab. */ |
2355 | +static size_t tablen = 0; |
2356 | |
2357 | static struct option const longopts[] = |
2358 | { |
2359 | @@ -197,6 +213,8 @@ |
2360 | |
2361 | /* Fill in the `fields' structure in LINE. */ |
2362 | |
2363 | +/* Fill in the `fields' structure in LINE. */ |
2364 | + |
2365 | static void |
2366 | xfields (struct line *line) |
2367 | { |
2368 | @@ -206,10 +224,11 @@ |
2369 | if (ptr == lim) |
2370 | return; |
2371 | |
2372 | - if (0 <= tab) |
2373 | + if (tab != NULL) |
2374 | { |
2375 | + unsigned char t = tab[0]; |
2376 | char *sep; |
2377 | - for (; (sep = memchr (ptr, tab, lim - ptr)) != NULL; ptr = sep + 1) |
2378 | + for (; (sep = memchr (ptr, t, lim - ptr)) != NULL; ptr = sep + 1) |
2379 | extract_field (line, ptr, sep - ptr); |
2380 | } |
2381 | else |
2382 | @@ -236,6 +255,148 @@ |
2383 | extract_field (line, ptr, lim - ptr); |
2384 | } |
2385 | |
2386 | +#if HAVE_MBRTOWC |
2387 | +static void |
2388 | +xfields_multibyte (struct line *line) |
2389 | +{ |
2390 | + char *ptr = line->buf.buffer; |
2391 | + char const *lim = ptr + line->buf.length - 1; |
2392 | + wchar_t wc = 0; |
2393 | + size_t mblength = 1; |
2394 | + mbstate_t state, state_bak; |
2395 | + |
2396 | + memset (&state, 0, sizeof (mbstate_t)); |
2397 | + |
2398 | + if (ptr == lim) |
2399 | + return; |
2400 | + |
2401 | + if (tab != NULL) |
2402 | + { |
2403 | + unsigned char t = tab[0]; |
2404 | + char *sep = ptr; |
2405 | + for (; ptr < lim; ptr = sep + mblength) |
2406 | + { |
2407 | + sep = ptr; |
2408 | + while (sep < lim) |
2409 | + { |
2410 | + state_bak = state; |
2411 | + mblength = mbrtowc (&wc, sep, lim - sep + 1, &state); |
2412 | + |
2413 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
2414 | + { |
2415 | + mblength = 1; |
2416 | + state = state_bak; |
2417 | + } |
2418 | + mblength = (mblength < 1) ? 1 : mblength; |
2419 | + |
2420 | + if (mblength == tablen && !memcmp (sep, tab, mblength)) |
2421 | + break; |
2422 | + else |
2423 | + { |
2424 | + sep += mblength; |
2425 | + continue; |
2426 | + } |
2427 | + } |
2428 | + |
2429 | + if (sep == lim) |
2430 | + break; |
2431 | + |
2432 | + extract_field (line, ptr, sep - ptr); |
2433 | + } |
2434 | + } |
2435 | + else |
2436 | + { |
2437 | + /* Skip leading blanks before the first field. */ |
2438 | + while(ptr < lim) |
2439 | + { |
2440 | + state_bak = state; |
2441 | + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); |
2442 | + |
2443 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
2444 | + { |
2445 | + mblength = 1; |
2446 | + state = state_bak; |
2447 | + break; |
2448 | + } |
2449 | + mblength = (mblength < 1) ? 1 : mblength; |
2450 | + |
2451 | + if (!iswblank(wc)) |
2452 | + break; |
2453 | + ptr += mblength; |
2454 | + } |
2455 | + |
2456 | + do |
2457 | + { |
2458 | + char *sep; |
2459 | + state_bak = state; |
2460 | + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); |
2461 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
2462 | + { |
2463 | + mblength = 1; |
2464 | + state = state_bak; |
2465 | + break; |
2466 | + } |
2467 | + mblength = (mblength < 1) ? 1 : mblength; |
2468 | + |
2469 | + sep = ptr + mblength; |
2470 | + while (sep != lim) |
2471 | + { |
2472 | + state_bak = state; |
2473 | + mblength = mbrtowc (&wc, sep, lim - sep + 1, &state); |
2474 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
2475 | + { |
2476 | + mblength = 1; |
2477 | + state = state_bak; |
2478 | + break; |
2479 | + } |
2480 | + mblength = (mblength < 1) ? 1 : mblength; |
2481 | + |
2482 | + if (iswblank (wc)) |
2483 | + break; |
2484 | + |
2485 | + sep += mblength; |
2486 | + } |
2487 | + |
2488 | + extract_field (line, ptr, sep - ptr); |
2489 | + if (sep == lim) |
2490 | + return; |
2491 | + |
2492 | + state_bak = state; |
2493 | + mblength = mbrtowc (&wc, sep, lim - sep + 1, &state); |
2494 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
2495 | + { |
2496 | + mblength = 1; |
2497 | + state = state_bak; |
2498 | + break; |
2499 | + } |
2500 | + mblength = (mblength < 1) ? 1 : mblength; |
2501 | + |
2502 | + ptr = sep + mblength; |
2503 | + while (ptr != lim) |
2504 | + { |
2505 | + state_bak = state; |
2506 | + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); |
2507 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
2508 | + { |
2509 | + mblength = 1; |
2510 | + state = state_bak; |
2511 | + break; |
2512 | + } |
2513 | + mblength = (mblength < 1) ? 1 : mblength; |
2514 | + |
2515 | + if (!iswblank (wc)) |
2516 | + break; |
2517 | + |
2518 | + ptr += mblength; |
2519 | + } |
2520 | + } |
2521 | + while (ptr != lim); |
2522 | + } |
2523 | + |
2524 | + extract_field (line, ptr, lim - ptr); |
2525 | +} |
2526 | +#endif |
2527 | + |
2528 | /* Read a line from FP into LINE and split it into fields. |
2529 | Return true if successful. */ |
2530 | |
2531 | @@ -256,6 +417,11 @@ |
2532 | line->nfields_allocated = 0; |
2533 | line->nfields = 0; |
2534 | line->fields = NULL; |
2535 | +#if HAVE_MBRTOWC |
2536 | + if (MB_CUR_MAX > 1) |
2537 | + xfields_multibyte (line); |
2538 | + else |
2539 | +#endif |
2540 | xfields (line); |
2541 | return true; |
2542 | } |
2543 | @@ -310,56 +476,114 @@ |
2544 | keycmp (struct line const *line1, struct line const *line2) |
2545 | { |
2546 | /* Start of field to compare in each file. */ |
2547 | - char *beg1; |
2548 | - char *beg2; |
2549 | - |
2550 | - size_t len1; |
2551 | - size_t len2; /* Length of fields to compare. */ |
2552 | + char *beg[2]; |
2553 | + char *copy[2]; |
2554 | + size_t len[2]; /* Length of fields to compare. */ |
2555 | int diff; |
2556 | + int i, j; |
2557 | |
2558 | if (join_field_1 < line1->nfields) |
2559 | { |
2560 | - beg1 = line1->fields[join_field_1].beg; |
2561 | - len1 = line1->fields[join_field_1].len; |
2562 | + beg[0] = line1->fields[join_field_1].beg; |
2563 | + len[0] = line1->fields[join_field_1].len; |
2564 | } |
2565 | else |
2566 | { |
2567 | - beg1 = NULL; |
2568 | - len1 = 0; |
2569 | + beg[0] = NULL; |
2570 | + len[0] = 0; |
2571 | } |
2572 | |
2573 | if (join_field_2 < line2->nfields) |
2574 | { |
2575 | - beg2 = line2->fields[join_field_2].beg; |
2576 | - len2 = line2->fields[join_field_2].len; |
2577 | + beg[1] = line2->fields[join_field_2].beg; |
2578 | + len[1] = line2->fields[join_field_2].len; |
2579 | } |
2580 | else |
2581 | { |
2582 | - beg2 = NULL; |
2583 | - len2 = 0; |
2584 | + beg[1] = NULL; |
2585 | + len[1] = 0; |
2586 | } |
2587 | |
2588 | - if (len1 == 0) |
2589 | - return len2 == 0 ? 0 : -1; |
2590 | - if (len2 == 0) |
2591 | + if (len[0] == 0) |
2592 | + return len[1] == 0 ? 0 : -1; |
2593 | + if (len[1] == 0) |
2594 | return 1; |
2595 | |
2596 | if (ignore_case) |
2597 | { |
2598 | - /* FIXME: ignore_case does not work with NLS (in particular, |
2599 | - with multibyte chars). */ |
2600 | - diff = memcasecmp (beg1, beg2, MIN (len1, len2)); |
2601 | +#ifdef HAVE_MBRTOWC |
2602 | + if (MB_CUR_MAX > 1) |
2603 | + { |
2604 | + size_t mblength; |
2605 | + wchar_t wc, uwc; |
2606 | + mbstate_t state, state_bak; |
2607 | + |
2608 | + memset (&state, '\0', sizeof (mbstate_t)); |
2609 | + |
2610 | + for (i = 0; i < 2; i++) |
2611 | + { |
2612 | + copy[i] = alloca (len[i] + 1); |
2613 | + |
2614 | + for (j = 0; j < MIN (len[0], len[1]);) |
2615 | + { |
2616 | + state_bak = state; |
2617 | + mblength = mbrtowc (&wc, beg[i] + j, len[i] - j, &state); |
2618 | + |
2619 | + switch (mblength) |
2620 | + { |
2621 | + case (size_t) -1: |
2622 | + case (size_t) -2: |
2623 | + state = state_bak; |
2624 | + /* Fall through */ |
2625 | + case 0: |
2626 | + mblength = 1; |
2627 | + break; |
2628 | + |
2629 | + default: |
2630 | + uwc = towupper (wc); |
2631 | + |
2632 | + if (uwc != wc) |
2633 | + { |
2634 | + mbstate_t state_wc; |
2635 | + |
2636 | + memset (&state_wc, '\0', sizeof (mbstate_t)); |
2637 | + wcrtomb (copy[i] + j, uwc, &state_wc); |
2638 | + } |
2639 | + else |
2640 | + memcpy (copy[i] + j, beg[i] + j, mblength); |
2641 | + } |
2642 | + j += mblength; |
2643 | + } |
2644 | + copy[i][j] = '\0'; |
2645 | + } |
2646 | + } |
2647 | + else |
2648 | +#endif |
2649 | + { |
2650 | + for (i = 0; i < 2; i++) |
2651 | + { |
2652 | + copy[i] = alloca (len[i] + 1); |
2653 | + |
2654 | + for (j = 0; j < MIN (len[0], len[1]); j++) |
2655 | + copy[i][j] = toupper (beg[i][j]); |
2656 | + |
2657 | + copy[i][j] = '\0'; |
2658 | + } |
2659 | + } |
2660 | } |
2661 | else |
2662 | { |
2663 | - if (hard_LC_COLLATE) |
2664 | - return xmemcoll (beg1, len1, beg2, len2); |
2665 | - diff = memcmp (beg1, beg2, MIN (len1, len2)); |
2666 | + copy[0] = (unsigned char *) beg[0]; |
2667 | + copy[1] = (unsigned char *) beg[1]; |
2668 | } |
2669 | |
2670 | + if (HAVE_SETLOCALE && hard_LC_COLLATE) |
2671 | + return xmemcoll ((char *) copy[0], len[0], (char *) copy[1], len[1]); |
2672 | + diff = memcmp (copy[0], copy[1], MIN (len[0], len[1])); |
2673 | + |
2674 | if (diff) |
2675 | return diff; |
2676 | - return len1 < len2 ? -1 : len1 != len2; |
2677 | + return len[0] - len[1]; |
2678 | } |
2679 | |
2680 | /* Print field N of LINE if it exists and is nonempty, otherwise |
2681 | @@ -384,11 +608,18 @@ |
2682 | |
2683 | /* Print the join of LINE1 and LINE2. */ |
2684 | |
2685 | +#define PUT_TAB_CHAR \ |
2686 | + do \ |
2687 | + { \ |
2688 | + (tab != NULL) ? \ |
2689 | + fwrite(tab, sizeof(char), tablen, stdout) : putchar (' '); \ |
2690 | + } \ |
2691 | + while (0) |
2692 | + |
2693 | static void |
2694 | prjoin (struct line const *line1, struct line const *line2) |
2695 | { |
2696 | const struct outlist *outlist; |
2697 | - char output_separator = tab < 0 ? ' ' : tab; |
2698 | |
2699 | outlist = outlist_head.next; |
2700 | if (outlist) |
2701 | @@ -404,12 +635,12 @@ |
2702 | if (o->file == 0) |
2703 | { |
2704 | if (line1 == &uni_blank) |
2705 | - { |
2706 | + { |
2707 | line = line2; |
2708 | field = join_field_2; |
2709 | } |
2710 | else |
2711 | - { |
2712 | + { |
2713 | line = line1; |
2714 | field = join_field_1; |
2715 | } |
2716 | @@ -423,7 +654,7 @@ |
2717 | o = o->next; |
2718 | if (o == NULL) |
2719 | break; |
2720 | - putchar (output_separator); |
2721 | + PUT_TAB_CHAR; |
2722 | } |
2723 | putchar ('\n'); |
2724 | } |
2725 | @@ -441,23 +672,23 @@ |
2726 | prfield (join_field_1, line1); |
2727 | for (i = 0; i < join_field_1 && i < line1->nfields; ++i) |
2728 | { |
2729 | - putchar (output_separator); |
2730 | + PUT_TAB_CHAR; |
2731 | prfield (i, line1); |
2732 | } |
2733 | for (i = join_field_1 + 1; i < line1->nfields; ++i) |
2734 | { |
2735 | - putchar (output_separator); |
2736 | + PUT_TAB_CHAR; |
2737 | prfield (i, line1); |
2738 | } |
2739 | |
2740 | for (i = 0; i < join_field_2 && i < line2->nfields; ++i) |
2741 | { |
2742 | - putchar (output_separator); |
2743 | + PUT_TAB_CHAR; |
2744 | prfield (i, line2); |
2745 | } |
2746 | for (i = join_field_2 + 1; i < line2->nfields; ++i) |
2747 | { |
2748 | - putchar (output_separator); |
2749 | + PUT_TAB_CHAR; |
2750 | prfield (i, line2); |
2751 | } |
2752 | putchar ('\n'); |
2753 | @@ -869,20 +1100,41 @@ |
2754 | |
2755 | case 't': |
2756 | { |
2757 | - unsigned char newtab = optarg[0]; |
2758 | - if (! newtab) |
2759 | + char *newtab; |
2760 | + size_t newtablen; |
2761 | + if (! optarg[0]) |
2762 | error (EXIT_FAILURE, 0, _("empty tab")); |
2763 | - if (optarg[1]) |
2764 | + newtab = xstrdup (optarg); |
2765 | +#if HAVE_MBRTOWC |
2766 | + if (MB_CUR_MAX > 1) |
2767 | + { |
2768 | + mbstate_t state; |
2769 | + |
2770 | + memset (&state, 0, sizeof (mbstate_t)); |
2771 | + newtablen = mbrtowc (NULL, newtab, |
2772 | + strnlen (newtab, MB_LEN_MAX), |
2773 | + &state); |
2774 | + if (newtablen == (size_t) 0 |
2775 | + || newtablen == (size_t) -1 |
2776 | + || newtablen == (size_t) -2) |
2777 | + newtablen = 1; |
2778 | + } |
2779 | + else |
2780 | +#endif |
2781 | + newtablen = 1; |
2782 | + |
2783 | + if (newtablen == 1 && newtab[1]) |
2784 | + { |
2785 | + if (STREQ (newtab, "\\0")) |
2786 | + newtab[0] = '\0'; |
2787 | + } |
2788 | + if (tab != NULL && strcmp (tab, newtab)) |
2789 | { |
2790 | - if (STREQ (optarg, "\\0")) |
2791 | - newtab = '\0'; |
2792 | - else |
2793 | - error (EXIT_FAILURE, 0, _("multi-character tab %s"), |
2794 | - quote (optarg)); |
2795 | + free (newtab); |
2796 | + error (EXIT_FAILURE, 0, _("incompatible tabs")); |
2797 | } |
2798 | - if (0 <= tab && tab != newtab) |
2799 | - error (EXIT_FAILURE, 0, _("incompatible tabs")); |
2800 | tab = newtab; |
2801 | + tablen = newtablen; |
2802 | } |
2803 | break; |
2804 | |
2805 | --- coreutils-5.93/src/unexpand.c.i18n 2005-08-12 08:16:25.000000000 +0100 |
2806 | +++ coreutils-5.93/src/unexpand.c 2005-12-23 08:53:01.000000000 +0000 |
2807 | @@ -39,11 +39,28 @@ |
2808 | #include <stdio.h> |
2809 | #include <getopt.h> |
2810 | #include <sys/types.h> |
2811 | + |
2812 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ |
2813 | +#if HAVE_WCHAR_H |
2814 | +# include <wchar.h> |
2815 | +#endif |
2816 | + |
2817 | #include "system.h" |
2818 | #include "error.h" |
2819 | #include "quote.h" |
2820 | #include "xstrndup.h" |
2821 | |
2822 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
2823 | + installation; work around this configuration error. */ |
2824 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
2825 | +# define MB_LEN_MAX 16 |
2826 | +#endif |
2827 | + |
2828 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
2829 | +#if HAVE_MBRTOWC && defined mbstate_t |
2830 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
2831 | +#endif |
2832 | + |
2833 | /* The official name of this program (e.g., no `g' prefix). */ |
2834 | #define PROGRAM_NAME "unexpand" |
2835 | |
2836 | @@ -110,6 +127,208 @@ |
2837 | {NULL, 0, NULL, 0} |
2838 | }; |
2839 | |
2840 | +static FILE *next_file (FILE *fp); |
2841 | + |
2842 | +#if HAVE_MBRTOWC |
2843 | +static void |
2844 | +unexpand_multibyte (void) |
2845 | +{ |
2846 | + FILE *fp; /* Input stream. */ |
2847 | + mbstate_t i_state; /* Current shift state of the input stream. */ |
2848 | + mbstate_t i_state_bak; /* Back up the I_STATE. */ |
2849 | + mbstate_t o_state; /* Current shift state of the output stream. */ |
2850 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
2851 | + char *bufpos; /* Next read position of BUF. */ |
2852 | + size_t buflen = 0; /* The length of the byte sequence in buf. */ |
2853 | + wint_t wc; /* A gotten wide character. */ |
2854 | + size_t mblength; /* The byte size of a multibyte character |
2855 | + which shows as same character as WC. */ |
2856 | + |
2857 | + /* Index in `tab_list' of next tabstop: */ |
2858 | + int tab_index = 0; /* For calculating width of pending tabs. */ |
2859 | + int print_tab_index = 0; /* For printing as many tabs as possible. */ |
2860 | + unsigned int column = 0; /* Column on screen of next char. */ |
2861 | + int next_tab_column; /* Column the next tab stop is on. */ |
2862 | + int convert = 1; /* If nonzero, perform translations. */ |
2863 | + unsigned int pending = 0; /* Pending columns of blanks. */ |
2864 | + |
2865 | + fp = next_file ((FILE *) NULL); |
2866 | + if (fp == NULL) |
2867 | + return; |
2868 | + |
2869 | + memset (&o_state, '\0', sizeof(mbstate_t)); |
2870 | + memset (&i_state, '\0', sizeof(mbstate_t)); |
2871 | + |
2872 | + for (;;) |
2873 | + { |
2874 | + if (buflen < MB_LEN_MAX && !feof(fp) && !ferror(fp)) |
2875 | + { |
2876 | + memmove (buf, bufpos, buflen); |
2877 | + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, fp); |
2878 | + bufpos = buf; |
2879 | + } |
2880 | + |
2881 | + /* Get a wide character. */ |
2882 | + if (buflen < 1) |
2883 | + { |
2884 | + mblength = 1; |
2885 | + wc = WEOF; |
2886 | + } |
2887 | + else |
2888 | + { |
2889 | + i_state_bak = i_state; |
2890 | + mblength = mbrtowc ((wchar_t *)&wc, bufpos, buflen, &i_state); |
2891 | + } |
2892 | + |
2893 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
2894 | + { |
2895 | + i_state = i_state_bak; |
2896 | + wc = L'\0'; |
2897 | + } |
2898 | + |
2899 | + if (wc == L' ' && convert && column < INT_MAX) |
2900 | + { |
2901 | + ++pending; |
2902 | + ++column; |
2903 | + } |
2904 | + else if (wc == L'\t' && convert) |
2905 | + { |
2906 | + if (tab_size == 0) |
2907 | + { |
2908 | + /* Do not let tab_index == first_free_tab; |
2909 | + stop when it is 1 less. */ |
2910 | + while (tab_index < first_free_tab - 1 |
2911 | + && column >= tab_list[tab_index]) |
2912 | + tab_index++; |
2913 | + next_tab_column = tab_list[tab_index]; |
2914 | + if (tab_index < first_free_tab - 1) |
2915 | + tab_index++; |
2916 | + if (column >= next_tab_column) |
2917 | + { |
2918 | + convert = 0; /* Ran out of tab stops. */ |
2919 | + goto flush_pend_mb; |
2920 | + } |
2921 | + } |
2922 | + else |
2923 | + { |
2924 | + next_tab_column = column + tab_size - column % tab_size; |
2925 | + } |
2926 | + pending += next_tab_column - column; |
2927 | + column = next_tab_column; |
2928 | + } |
2929 | + else |
2930 | + { |
2931 | +flush_pend_mb: |
2932 | + /* Flush pending spaces. Print as many tabs as possible, |
2933 | + then print the rest as spaces. */ |
2934 | + if (pending == 1) |
2935 | + { |
2936 | + putchar (' '); |
2937 | + pending = 0; |
2938 | + } |
2939 | + column -= pending; |
2940 | + while (pending > 0) |
2941 | + { |
2942 | + if (tab_size == 0) |
2943 | + { |
2944 | + /* Do not let print_tab_index == first_free_tab; |
2945 | + stop when it is 1 less. */ |
2946 | + while (print_tab_index < first_free_tab - 1 |
2947 | + && column >= tab_list[print_tab_index]) |
2948 | + print_tab_index++; |
2949 | + next_tab_column = tab_list[print_tab_index]; |
2950 | + if (print_tab_index < first_free_tab - 1) |
2951 | + print_tab_index++; |
2952 | + } |
2953 | + else |
2954 | + { |
2955 | + next_tab_column = |
2956 | + column + tab_size - column % tab_size; |
2957 | + } |
2958 | + if (next_tab_column - column <= pending) |
2959 | + { |
2960 | + putchar ('\t'); |
2961 | + pending -= next_tab_column - column; |
2962 | + column = next_tab_column; |
2963 | + } |
2964 | + else |
2965 | + { |
2966 | + --print_tab_index; |
2967 | + column += pending; |
2968 | + while (pending != 0) |
2969 | + { |
2970 | + putchar (' '); |
2971 | + pending--; |
2972 | + } |
2973 | + } |
2974 | + } |
2975 | + |
2976 | + if (wc == WEOF) |
2977 | + { |
2978 | + fp = next_file (fp); |
2979 | + if (fp == NULL) |
2980 | + break; /* No more files. */ |
2981 | + else |
2982 | + { |
2983 | + memset (&i_state, '\0', sizeof(mbstate_t)); |
2984 | + continue; |
2985 | + } |
2986 | + } |
2987 | + |
2988 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
2989 | + { |
2990 | + if (convert) |
2991 | + { |
2992 | + ++column; |
2993 | + if (convert_entire_line == 0) |
2994 | + convert = 0; |
2995 | + } |
2996 | + mblength = 1; |
2997 | + putchar (buf[0]); |
2998 | + } |
2999 | + else if (mblength == 0) |
3000 | + { |
3001 | + if (convert && convert_entire_line == 0) |
3002 | + convert = 0; |
3003 | + mblength = 1; |
3004 | + putchar ('\0'); |
3005 | + } |
3006 | + else |
3007 | + { |
3008 | + if (convert) |
3009 | + { |
3010 | + if (wc == L'\b') |
3011 | + { |
3012 | + if (column > 0) |
3013 | + --column; |
3014 | + } |
3015 | + else |
3016 | + { |
3017 | + int width; /* The width of WC. */ |
3018 | + |
3019 | + width = wcwidth (wc); |
3020 | + column += (width > 0) ? width : 0; |
3021 | + if (convert_entire_line == 0) |
3022 | + convert = 0; |
3023 | + } |
3024 | + } |
3025 | + |
3026 | + if (wc == L'\n') |
3027 | + { |
3028 | + tab_index = print_tab_index = 0; |
3029 | + column = pending = 0; |
3030 | + convert = 1; |
3031 | + } |
3032 | + fwrite (bufpos, sizeof(char), mblength, stdout); |
3033 | + } |
3034 | + } |
3035 | + buflen -= mblength; |
3036 | + bufpos += mblength; |
3037 | + } |
3038 | +} |
3039 | +#endif |
3040 | + |
3041 | + |
3042 | void |
3043 | usage (int status) |
3044 | { |
3045 | @@ -532,7 +751,12 @@ |
3046 | |
3047 | file_list = (optind < argc ? &argv[optind] : stdin_argv); |
3048 | |
3049 | - unexpand (); |
3050 | +#if HAVE_MBRTOWC |
3051 | + if (MB_CUR_MAX > 1) |
3052 | + unexpand_multibyte (); |
3053 | + else |
3054 | +#endif |
3055 | + unexpand (); |
3056 | |
3057 | if (have_read_stdin && fclose (stdin) != 0) |
3058 | error (EXIT_FAILURE, errno, "-"); |
3059 | --- coreutils-5.93/src/sort.c.i18n 2005-10-07 19:48:28.000000000 +0100 |
3060 | +++ coreutils-5.93/src/sort.c 2005-12-23 10:38:44.000000000 +0000 |
3061 | @@ -23,9 +23,18 @@ |
3062 | |
3063 | #include <config.h> |
3064 | |
3065 | +#include <assert.h> |
3066 | #include <getopt.h> |
3067 | #include <sys/types.h> |
3068 | #include <signal.h> |
3069 | +#if HAVE_WCHAR_H |
3070 | +# include <wchar.h> |
3071 | +#endif |
3072 | +/* Get isw* functions. */ |
3073 | +#if HAVE_WCTYPE_H |
3074 | +# include <wctype.h> |
3075 | +#endif |
3076 | + |
3077 | #include "system.h" |
3078 | #include "error.h" |
3079 | #include "hard-locale.h" |
3080 | @@ -95,14 +104,38 @@ |
3081 | /* Thousands separator; if -1, then there isn't one. */ |
3082 | static int thousands_sep; |
3083 | |
3084 | +static int force_general_numcompare = 0; |
3085 | + |
3086 | /* Nonzero if the corresponding locales are hard. */ |
3087 | static bool hard_LC_COLLATE; |
3088 | -#if HAVE_NL_LANGINFO |
3089 | +#if HAVE_LANGINFO_CODESET |
3090 | static bool hard_LC_TIME; |
3091 | #endif |
3092 | |
3093 | #define NONZERO(x) ((x) != 0) |
3094 | |
3095 | +/* get a multibyte character's byte length. */ |
3096 | +#define GET_BYTELEN_OF_CHAR(LIM, PTR, MBLENGTH, STATE) \ |
3097 | + do \ |
3098 | + { \ |
3099 | + wchar_t wc; \ |
3100 | + mbstate_t state_bak; \ |
3101 | + \ |
3102 | + state_bak = STATE; \ |
3103 | + mblength = mbrtowc (&wc, PTR, LIM - PTR, &STATE); \ |
3104 | + \ |
3105 | + switch (MBLENGTH) \ |
3106 | + { \ |
3107 | + case (size_t)-1: \ |
3108 | + case (size_t)-2: \ |
3109 | + STATE = state_bak; \ |
3110 | + /* Fall through. */ \ |
3111 | + case 0: \ |
3112 | + MBLENGTH = 1; \ |
3113 | + } \ |
3114 | + } \ |
3115 | + while (0) |
3116 | + |
3117 | /* The kind of blanks for '-b' to skip in various options. */ |
3118 | enum blanktype { bl_start, bl_end, bl_both }; |
3119 | |
3120 | @@ -239,13 +272,11 @@ |
3121 | they were read if all keys compare equal. */ |
3122 | static bool stable; |
3123 | |
3124 | -/* If TAB has this value, blanks separate fields. */ |
3125 | -enum { TAB_DEFAULT = CHAR_MAX + 1 }; |
3126 | - |
3127 | -/* Tab character separating fields. If TAB_DEFAULT, then fields are |
3128 | +/* Tab character separating fields. If tab_length is 0, then fields are |
3129 | separated by the empty string between a non-blank character and a blank |
3130 | character. */ |
3131 | -static int tab = TAB_DEFAULT; |
3132 | +static char tab[MB_LEN_MAX + 1]; |
3133 | +static size_t tab_length = 0; |
3134 | |
3135 | /* Flag to remove consecutive duplicate lines from the output. |
3136 | Only the last of a sequence of equal lines will be output. */ |
3137 | @@ -392,6 +423,44 @@ |
3138 | static struct tempnode *volatile temphead; |
3139 | static struct tempnode *volatile *temptail = &temphead; |
3140 | |
3141 | +/* Function pointers. */ |
3142 | +static void |
3143 | +(*inittables) (void); |
3144 | +static char * |
3145 | +(*begfield) (const struct line*, const struct keyfield *); |
3146 | +static char * |
3147 | +(*limfield) (const struct line*, const struct keyfield *); |
3148 | +static int |
3149 | +(*getmonth) (char const *, size_t); |
3150 | +static int |
3151 | +(*keycompare) (const struct line *, const struct line *); |
3152 | +static int |
3153 | +(*numcompare) (const char *, const char *); |
3154 | + |
3155 | +/* Test for white space multibyte character. |
3156 | + Set LENGTH the byte length of investigated multibyte character. */ |
3157 | +#if HAVE_MBRTOWC |
3158 | +static int |
3159 | +ismbblank (const char *str, size_t len, size_t *length) |
3160 | +{ |
3161 | + size_t mblength; |
3162 | + wchar_t wc; |
3163 | + mbstate_t state; |
3164 | + |
3165 | + memset (&state, '\0', sizeof(mbstate_t)); |
3166 | + mblength = mbrtowc (&wc, str, len, &state); |
3167 | + |
3168 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
3169 | + { |
3170 | + *length = 1; |
3171 | + return 0; |
3172 | + } |
3173 | + |
3174 | + *length = (mblength < 1) ? 1 : mblength; |
3175 | + return iswblank (wc); |
3176 | +} |
3177 | +#endif |
3178 | + |
3179 | /* Clean up any remaining temporary files. */ |
3180 | |
3181 | static void |
3182 | @@ -545,7 +614,7 @@ |
3183 | free (node); |
3184 | } |
3185 | |
3186 | -#if HAVE_NL_LANGINFO |
3187 | +#if HAVE_LANGINFO_CODESET |
3188 | |
3189 | static int |
3190 | struct_month_cmp (const void *m1, const void *m2) |
3191 | @@ -560,7 +629,7 @@ |
3192 | /* Initialize the character class tables. */ |
3193 | |
3194 | static void |
3195 | -inittables (void) |
3196 | +inittables_uni (void) |
3197 | { |
3198 | size_t i; |
3199 | |
3200 | @@ -572,7 +641,7 @@ |
3201 | fold_toupper[i] = (ISLOWER (i) ? toupper (i) : i); |
3202 | } |
3203 | |
3204 | -#if HAVE_NL_LANGINFO |
3205 | +#if HAVE_LANGINFO_CODESET |
3206 | /* If we're not in the "C" locale, read different names for months. */ |
3207 | if (hard_LC_TIME) |
3208 | { |
3209 | @@ -598,6 +667,64 @@ |
3210 | #endif |
3211 | } |
3212 | |
3213 | +#if HAVE_MBRTOWC |
3214 | +static void |
3215 | +inittables_mb (void) |
3216 | +{ |
3217 | + int i, j, k, l; |
3218 | + char *name, *s; |
3219 | + size_t s_len, mblength; |
3220 | + char mbc[MB_LEN_MAX]; |
3221 | + wchar_t wc, pwc; |
3222 | + mbstate_t state_mb, state_wc; |
3223 | + |
3224 | + for (i = 0; i < MONTHS_PER_YEAR; i++) |
3225 | + { |
3226 | + s = (char *) nl_langinfo (ABMON_1 + i); |
3227 | + s_len = strlen (s); |
3228 | + monthtab[i].name = name = (char *) xmalloc (s_len + 1); |
3229 | + monthtab[i].val = i + 1; |
3230 | + |
3231 | + memset (&state_mb, '\0', sizeof (mbstate_t)); |
3232 | + memset (&state_wc, '\0', sizeof (mbstate_t)); |
3233 | + |
3234 | + for (j = 0; j < s_len;) |
3235 | + { |
3236 | + if (!ismbblank (s + j, s_len - j, &mblength)) |
3237 | + break; |
3238 | + j += mblength; |
3239 | + } |
3240 | + |
3241 | + for (k = 0; j < s_len;) |
3242 | + { |
3243 | + mblength = mbrtowc (&wc, (s + j), (s_len - j), &state_mb); |
3244 | + assert (mblength != (size_t)-1 && mblength != (size_t)-2); |
3245 | + if (mblength == 0) |
3246 | + break; |
3247 | + |
3248 | + pwc = towupper (wc); |
3249 | + if (pwc == wc) |
3250 | + { |
3251 | + memcpy (mbc, s + j, mblength); |
3252 | + j += mblength; |
3253 | + } |
3254 | + else |
3255 | + { |
3256 | + j += mblength; |
3257 | + mblength = wcrtomb (mbc, pwc, &state_wc); |
3258 | + assert (mblength != (size_t)0 && mblength != (size_t)-1); |
3259 | + } |
3260 | + |
3261 | + for (l = 0; l < mblength; l++) |
3262 | + name[k++] = mbc[l]; |
3263 | + } |
3264 | + name[k] = '\0'; |
3265 | + } |
3266 | + qsort ((void *) monthtab, MONTHS_PER_YEAR, |
3267 | + sizeof (struct month), struct_month_cmp); |
3268 | +} |
3269 | +#endif |
3270 | + |
3271 | /* Specify the amount of main memory to use when sorting. */ |
3272 | static void |
3273 | specify_sort_size (char const *s) |
3274 | @@ -808,7 +935,7 @@ |
3275 | by KEY in LINE. */ |
3276 | |
3277 | static char * |
3278 | -begfield (const struct line *line, const struct keyfield *key) |
3279 | +begfield_uni (const struct line *line, const struct keyfield *key) |
3280 | { |
3281 | char *ptr = line->text, *lim = ptr + line->length - 1; |
3282 | size_t sword = key->sword; |
3283 | @@ -818,10 +945,10 @@ |
3284 | /* The leading field separator itself is included in a field when -t |
3285 | is absent. */ |
3286 | |
3287 | - if (tab != TAB_DEFAULT) |
3288 | + if (tab_length) |
3289 | while (ptr < lim && sword--) |
3290 | { |
3291 | - while (ptr < lim && *ptr != tab) |
3292 | + while (ptr < lim && *ptr != tab[0]) |
3293 | ++ptr; |
3294 | if (ptr < lim) |
3295 | ++ptr; |
3296 | @@ -849,11 +976,70 @@ |
3297 | return ptr; |
3298 | } |
3299 | |
3300 | +#if HAVE_MBRTOWC |
3301 | +static char * |
3302 | +begfield_mb (const struct line *line, const struct keyfield *key) |
3303 | +{ |
3304 | + int i; |
3305 | + char *ptr = line->text, *lim = ptr + line->length - 1; |
3306 | + size_t sword = key->sword; |
3307 | + size_t schar = key->schar; |
3308 | + size_t mblength; |
3309 | + mbstate_t state; |
3310 | + |
3311 | + memset (&state, '\0', sizeof(mbstate_t)); |
3312 | + |
3313 | + if (tab_length) |
3314 | + while (ptr < lim && sword--) |
3315 | + { |
3316 | + while (ptr < lim && memcmp (ptr, tab, tab_length) != 0) |
3317 | + { |
3318 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
3319 | + ptr += mblength; |
3320 | + } |
3321 | + if (ptr < lim) |
3322 | + { |
3323 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
3324 | + ptr += mblength; |
3325 | + } |
3326 | + } |
3327 | + else |
3328 | + while (ptr < lim && sword--) |
3329 | + { |
3330 | + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) |
3331 | + ptr += mblength; |
3332 | + if (ptr < lim) |
3333 | + { |
3334 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
3335 | + ptr += mblength; |
3336 | + } |
3337 | + while (ptr < lim && !ismbblank (ptr, lim - ptr, &mblength)) |
3338 | + ptr += mblength; |
3339 | + } |
3340 | + |
3341 | + if (key->skipsblanks) |
3342 | + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) |
3343 | + ptr += mblength; |
3344 | + |
3345 | + for (i = 0; i < schar; i++) |
3346 | + { |
3347 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
3348 | + |
3349 | + if (ptr + mblength > lim) |
3350 | + break; |
3351 | + else |
3352 | + ptr += mblength; |
3353 | + } |
3354 | + |
3355 | + return ptr; |
3356 | +} |
3357 | +#endif |
3358 | + |
3359 | /* Return the limit of (a pointer to the first character after) the field |
3360 | in LINE specified by KEY. */ |
3361 | |
3362 | static char * |
3363 | -limfield (const struct line *line, const struct keyfield *key) |
3364 | +limfield_uni (const struct line *line, const struct keyfield *key) |
3365 | { |
3366 | char *ptr = line->text, *lim = ptr + line->length - 1; |
3367 | size_t eword = key->eword, echar = key->echar; |
3368 | @@ -866,10 +1052,10 @@ |
3369 | `beginning' is the first character following the delimiting TAB. |
3370 | Otherwise, leave PTR pointing at the first `blank' character after |
3371 | the preceding field. */ |
3372 | - if (tab != TAB_DEFAULT) |
3373 | + if (tab_length) |
3374 | while (ptr < lim && eword--) |
3375 | { |
3376 | - while (ptr < lim && *ptr != tab) |
3377 | + while (ptr < lim && *ptr != tab[0]) |
3378 | ++ptr; |
3379 | if (ptr < lim && (eword | echar)) |
3380 | ++ptr; |
3381 | @@ -915,10 +1101,10 @@ |
3382 | */ |
3383 | |
3384 | /* Make LIM point to the end of (one byte past) the current field. */ |
3385 | - if (tab != TAB_DEFAULT) |
3386 | + if (tab_length) |
3387 | { |
3388 | char *newlim; |
3389 | - newlim = memchr (ptr, tab, lim - ptr); |
3390 | + newlim = memchr (ptr, tab[0], lim - ptr); |
3391 | if (newlim) |
3392 | lim = newlim; |
3393 | } |
3394 | @@ -951,6 +1137,107 @@ |
3395 | return ptr; |
3396 | } |
3397 | |
3398 | +#if HAVE_MBRTOWC |
3399 | +static char * |
3400 | +limfield_mb (const struct line *line, const struct keyfield *key) |
3401 | +{ |
3402 | + char *ptr = line->text, *lim = ptr + line->length - 1; |
3403 | + size_t eword = key->eword, echar = key->echar; |
3404 | + int i; |
3405 | + size_t mblength; |
3406 | + mbstate_t state; |
3407 | + |
3408 | + memset (&state, '\0', sizeof(mbstate_t)); |
3409 | + |
3410 | + if (tab_length) |
3411 | + while (ptr < lim && eword--) |
3412 | + { |
3413 | + while (ptr < lim && memcmp (ptr, tab, tab_length) != 0) |
3414 | + { |
3415 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
3416 | + ptr += mblength; |
3417 | + } |
3418 | + if (ptr < lim && (eword | echar)) |
3419 | + { |
3420 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
3421 | + ptr += mblength; |
3422 | + } |
3423 | + } |
3424 | + else |
3425 | + while (ptr < lim && eword--) |
3426 | + { |
3427 | + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) |
3428 | + ptr += mblength; |
3429 | + if (ptr < lim) |
3430 | + { |
3431 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
3432 | + ptr += mblength; |
3433 | + } |
3434 | + while (ptr < lim && !ismbblank (ptr, lim - ptr, &mblength)) |
3435 | + ptr += mblength; |
3436 | + } |
3437 | + |
3438 | + |
3439 | +# ifdef POSIX_UNSPECIFIED |
3440 | + /* Make LIM point to the end of (one byte past) the current field. */ |
3441 | + if (tab_length) |
3442 | + { |
3443 | + char *newlim, *p; |
3444 | + |
3445 | + newlim = NULL; |
3446 | + for (p = ptr; p < lim;) |
3447 | + { |
3448 | + if (memcmp (p, tab, tab_length) == 0) |
3449 | + { |
3450 | + newlim = p; |
3451 | + break; |
3452 | + } |
3453 | + |
3454 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
3455 | + p += mblength; |
3456 | + } |
3457 | + } |
3458 | + else |
3459 | + { |
3460 | + char *newlim; |
3461 | + newlim = ptr; |
3462 | + |
3463 | + while (newlim < lim && ismbblank (newlim, lim - newlim, &mblength)) |
3464 | + newlim += mblength; |
3465 | + if (ptr < lim) |
3466 | + { |
3467 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
3468 | + ptr += mblength; |
3469 | + } |
3470 | + while (newlim < lim && !ismbblank (newlim, lim - newlim, &mblength)) |
3471 | + newlim += mblength; |
3472 | + lim = newlim; |
3473 | + } |
3474 | +# endif |
3475 | + |
3476 | + /* If we're skipping leading blanks, don't start counting characters |
3477 | + * until after skipping past any leading blanks. */ |
3478 | + if (key->skipsblanks) |
3479 | + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) |
3480 | + ptr += mblength; |
3481 | + |
3482 | + memset (&state, '\0', sizeof(mbstate_t)); |
3483 | + |
3484 | + /* Advance PTR by ECHAR (if possible), but no further than LIM. */ |
3485 | + for (i = 0; i < echar; i++) |
3486 | + { |
3487 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
3488 | + |
3489 | + if (ptr + mblength > lim) |
3490 | + break; |
3491 | + else |
3492 | + ptr += mblength; |
3493 | + } |
3494 | + |
3495 | + return ptr; |
3496 | +} |
3497 | +#endif |
3498 | + |
3499 | /* Fill BUF reading from FP, moving buf->left bytes from the end |
3500 | of buf->buf to the beginning first. If EOF is reached and the |
3501 | file wasn't terminated by a newline, supply one. Set up BUF's line |
3502 | @@ -1067,7 +1354,7 @@ |
3503 | hideously fast. */ |
3504 | |
3505 | static int |
3506 | -numcompare (const char *a, const char *b) |
3507 | +numcompare_uni (const char *a, const char *b) |
3508 | { |
3509 | while (blanks[to_uchar (*a)]) |
3510 | a++; |
3511 | @@ -1077,6 +1364,25 @@ |
3512 | return strnumcmp (a, b, decimal_point, thousands_sep); |
3513 | } |
3514 | |
3515 | +#if HAVE_MBRTOWC |
3516 | +static int |
3517 | +numcompare_mb (const char *a, const char *b) |
3518 | +{ |
3519 | + size_t mblength, len; |
3520 | + len = strlen (a); /* okay for UTF-8 */ |
3521 | + while (*a && ismbblank (a, len > MB_CUR_MAX ? MB_CUR_MAX : len, &mblength)) |
3522 | + { |
3523 | + a += mblength; |
3524 | + len -= mblength; |
3525 | + } |
3526 | + len = strlen (b); /* okay for UTF-8 */ |
3527 | + while (*b && ismbblank (b, len > MB_CUR_MAX ? MB_CUR_MAX : len, &mblength)) |
3528 | + b += mblength; |
3529 | + |
3530 | + return strnumcmp (a, b, decimal_point, thousands_sep); |
3531 | +} |
3532 | +#endif /* HAV_EMBRTOWC */ |
3533 | + |
3534 | static int |
3535 | general_numcompare (const char *sa, const char *sb) |
3536 | { |
3537 | @@ -1110,7 +1416,7 @@ |
3538 | Return 0 if the name in S is not recognized. */ |
3539 | |
3540 | static int |
3541 | -getmonth (char const *month, size_t len) |
3542 | +getmonth_uni (char const *month, size_t len) |
3543 | { |
3544 | size_t lo = 0; |
3545 | size_t hi = MONTHS_PER_YEAR; |
3546 | @@ -1152,11 +1458,79 @@ |
3547 | return 0; |
3548 | } |
3549 | |
3550 | +#if HAVE_MBRTOWC |
3551 | +static int |
3552 | +getmonth_mb (const char *s, size_t len) |
3553 | +{ |
3554 | + char *month; |
3555 | + register size_t i; |
3556 | + register int lo = 0, hi = MONTHS_PER_YEAR, result; |
3557 | + char *tmp; |
3558 | + size_t wclength, mblength; |
3559 | + const char **pp; |
3560 | + const wchar_t **wpp; |
3561 | + wchar_t *month_wcs; |
3562 | + mbstate_t state; |
3563 | + |
3564 | + while (len > 0 && ismbblank (s, len, &mblength)) |
3565 | + { |
3566 | + s += mblength; |
3567 | + len -= mblength; |
3568 | + } |
3569 | + |
3570 | + if (len == 0) |
3571 | + return 0; |
3572 | + |
3573 | + month = (char *) alloca (len + 1); |
3574 | + |
3575 | + tmp = (char *) alloca (len + 1); |
3576 | + memcpy (tmp, s, len); |
3577 | + tmp[len] = '\0'; |
3578 | + pp = (const char **)&tmp; |
3579 | + month_wcs = (wchar_t *) alloca ((len + 1) * sizeof (wchar_t)); |
3580 | + memset (&state, '\0', sizeof(mbstate_t)); |
3581 | + |
3582 | + wclength = mbsrtowcs (month_wcs, pp, len + 1, &state); |
3583 | + assert (wclength != (size_t)-1 && *pp == NULL); |
3584 | + |
3585 | + for (i = 0; i < wclength; i++) |
3586 | + { |
3587 | + month_wcs[i] = towupper(month_wcs[i]); |
3588 | + if (iswblank (month_wcs[i])) |
3589 | + { |
3590 | + month_wcs[i] = L'\0'; |
3591 | + break; |
3592 | + } |
3593 | + } |
3594 | + |
3595 | + wpp = (const wchar_t **)&month_wcs; |
3596 | + |
3597 | + mblength = wcsrtombs (month, wpp, len + 1, &state); |
3598 | + assert (mblength != (-1) && *wpp == NULL); |
3599 | + |
3600 | + do |
3601 | + { |
3602 | + int ix = (lo + hi) / 2; |
3603 | + |
3604 | + if (strncmp (month, monthtab[ix].name, strlen (monthtab[ix].name)) < 0) |
3605 | + hi = ix; |
3606 | + else |
3607 | + lo = ix; |
3608 | + } |
3609 | + while (hi - lo > 1); |
3610 | + |
3611 | + result = (!strncmp (month, monthtab[lo].name, strlen (monthtab[lo].name)) |
3612 | + ? monthtab[lo].val : 0); |
3613 | + |
3614 | + return result; |
3615 | +} |
3616 | +#endif |
3617 | + |
3618 | /* Compare two lines A and B trying every key in sequence until there |
3619 | are no more keys or a difference is found. */ |
3620 | |
3621 | static int |
3622 | -keycompare (const struct line *a, const struct line *b) |
3623 | +keycompare_uni (const struct line *a, const struct line *b) |
3624 | { |
3625 | struct keyfield const *key = keylist; |
3626 | |
3627 | @@ -1326,6 +1700,177 @@ |
3628 | return key->reverse ? -diff : diff; |
3629 | } |
3630 | |
3631 | +#if HAVE_MBRTOWC |
3632 | +static int |
3633 | +keycompare_mb (const struct line *a, const struct line *b) |
3634 | +{ |
3635 | + struct keyfield *key = keylist; |
3636 | + |
3637 | + /* For the first iteration only, the key positions have been |
3638 | + precomputed for us. */ |
3639 | + char *texta = a->keybeg; |
3640 | + char *textb = b->keybeg; |
3641 | + char *lima = a->keylim; |
3642 | + char *limb = b->keylim; |
3643 | + |
3644 | + size_t mblength_a, mblength_b; |
3645 | + wchar_t wc_a, wc_b; |
3646 | + mbstate_t state_a, state_b; |
3647 | + |
3648 | + int diff; |
3649 | + |
3650 | + memset (&state_a, '\0', sizeof(mbstate_t)); |
3651 | + memset (&state_b, '\0', sizeof(mbstate_t)); |
3652 | + |
3653 | + for (;;) |
3654 | + { |
3655 | + unsigned char *translate = (unsigned char *) key->translate; |
3656 | + bool const *ignore = key->ignore; |
3657 | + |
3658 | + /* Find the lengths. */ |
3659 | + size_t lena = lima <= texta ? 0 : lima - texta; |
3660 | + size_t lenb = limb <= textb ? 0 : limb - textb; |
3661 | + |
3662 | + /* Actually compare the fields. */ |
3663 | + if (key->numeric | key->general_numeric) |
3664 | + { |
3665 | + char savea = *lima, saveb = *limb; |
3666 | + |
3667 | + *lima = *limb = '\0'; |
3668 | + if (force_general_numcompare) |
3669 | + diff = general_numcompare (texta, textb); |
3670 | + else |
3671 | + diff = ((key->numeric ? numcompare : general_numcompare) |
3672 | + (texta, textb)); |
3673 | + *lima = savea, *limb = saveb; |
3674 | + } |
3675 | + else if (key->month) |
3676 | + diff = getmonth (texta, lena) - getmonth (textb, lenb); |
3677 | + else |
3678 | + { |
3679 | + if (ignore || translate) |
3680 | + { |
3681 | + char *copy_a = (char *) alloca (lena + 1 + lenb + 1); |
3682 | + char *copy_b = copy_a + lena + 1; |
3683 | + size_t new_len_a, new_len_b; |
3684 | + size_t i, j; |
3685 | + |
3686 | + /* Ignore and/or translate chars before comparing. */ |
3687 | +# define IGNORE_CHARS(NEW_LEN, LEN, TEXT, COPY, WC, MBLENGTH, STATE) \ |
3688 | + do \ |
3689 | + { \ |
3690 | + wchar_t uwc; \ |
3691 | + char mbc[MB_LEN_MAX]; \ |
3692 | + mbstate_t state_wc; \ |
3693 | + \ |
3694 | + for (NEW_LEN = i = 0; i < LEN;) \ |
3695 | + { \ |
3696 | + mbstate_t state_bak; \ |
3697 | + \ |
3698 | + state_bak = STATE; \ |
3699 | + MBLENGTH = mbrtowc (&WC, TEXT + i, LEN - i, &STATE); \ |
3700 | + \ |
3701 | + if (MBLENGTH == (size_t)-2 || MBLENGTH == (size_t)-1 \ |
3702 | + || MBLENGTH == 0) \ |
3703 | + { \ |
3704 | + if (MBLENGTH == (size_t)-2 || MBLENGTH == (size_t)-1) \ |
3705 | + STATE = state_bak; \ |
3706 | + if (!ignore) \ |
3707 | + COPY[NEW_LEN++] = TEXT[i++]; \ |
3708 | + continue; \ |
3709 | + } \ |
3710 | + \ |
3711 | + if (ignore) \ |
3712 | + { \ |
3713 | + if ((ignore == nonprinting && !iswprint (WC)) \ |
3714 | + || (ignore == nondictionary \ |
3715 | + && !iswalnum (WC) && !iswblank (WC))) \ |
3716 | + { \ |
3717 | + i += MBLENGTH; \ |
3718 | + continue; \ |
3719 | + } \ |
3720 | + } \ |
3721 | + \ |
3722 | + if (translate) \ |
3723 | + { \ |
3724 | + \ |
3725 | + uwc = towupper(WC); \ |
3726 | + if (WC == uwc) \ |
3727 | + { \ |
3728 | + memcpy (mbc, TEXT + i, MBLENGTH); \ |
3729 | + i += MBLENGTH; \ |
3730 | + } \ |
3731 | + else \ |
3732 | + { \ |
3733 | + i += MBLENGTH; \ |
3734 | + WC = uwc; \ |
3735 | + memset (&state_wc, '\0', sizeof (mbstate_t)); \ |
3736 | + \ |
3737 | + MBLENGTH = wcrtomb (mbc, WC, &state_wc); \ |
3738 | + assert (MBLENGTH != (size_t)-1 && MBLENGTH != 0); \ |
3739 | + } \ |
3740 | + \ |
3741 | + for (j = 0; j < MBLENGTH; j++) \ |
3742 | + COPY[NEW_LEN++] = mbc[j]; \ |
3743 | + } \ |
3744 | + else \ |
3745 | + for (j = 0; j < MBLENGTH; j++) \ |
3746 | + COPY[NEW_LEN++] = TEXT[i++]; \ |
3747 | + } \ |
3748 | + COPY[NEW_LEN] = '\0'; \ |
3749 | + } \ |
3750 | + while (0) |
3751 | + IGNORE_CHARS (new_len_a, lena, texta, copy_a, |
3752 | + wc_a, mblength_a, state_a); |
3753 | + IGNORE_CHARS (new_len_b, lenb, textb, copy_b, |
3754 | + wc_b, mblength_b, state_b); |
3755 | + diff = xmemcoll (copy_a, new_len_a, copy_b, new_len_b); |
3756 | + } |
3757 | + else if (lena == 0) |
3758 | + diff = - NONZERO (lenb); |
3759 | + else if (lenb == 0) |
3760 | + goto greater; |
3761 | + else |
3762 | + diff = xmemcoll (texta, lena, textb, lenb); |
3763 | + } |
3764 | + |
3765 | + if (diff) |
3766 | + goto not_equal; |
3767 | + |
3768 | + key = key->next; |
3769 | + if (! key) |
3770 | + break; |
3771 | + |
3772 | + /* Find the beginning and limit of the next field. */ |
3773 | + if (key->eword != -1) |
3774 | + lima = limfield (a, key), limb = limfield (b, key); |
3775 | + else |
3776 | + lima = a->text + a->length - 1, limb = b->text + b->length - 1; |
3777 | + |
3778 | + if (key->sword != -1) |
3779 | + texta = begfield (a, key), textb = begfield (b, key); |
3780 | + else |
3781 | + { |
3782 | + texta = a->text, textb = b->text; |
3783 | + if (key->skipsblanks) |
3784 | + { |
3785 | + while (texta < lima && ismbblank (texta, lima - texta, &mblength_a)) |
3786 | + texta += mblength_a; |
3787 | + while (textb < limb && ismbblank (textb, limb - textb, &mblength_b)) |
3788 | + textb += mblength_b; |
3789 | + } |
3790 | + } |
3791 | + } |
3792 | + |
3793 | + return 0; |
3794 | + |
3795 | +greater: |
3796 | + diff = 1; |
3797 | +not_equal: |
3798 | + return key->reverse ? -diff : diff; |
3799 | +} |
3800 | +#endif |
3801 | + |
3802 | /* Compare two lines A and B, returning negative, zero, or positive |
3803 | depending on whether A compares less than, equal to, or greater than B. */ |
3804 | |
3805 | @@ -2127,7 +2672,7 @@ |
3806 | atexit (close_stdout); |
3807 | |
3808 | hard_LC_COLLATE = hard_locale (LC_COLLATE); |
3809 | -#if HAVE_NL_LANGINFO |
3810 | +#if HAVE_LANGINFO_CODESET |
3811 | hard_LC_TIME = hard_locale (LC_TIME); |
3812 | #endif |
3813 | |
3814 | @@ -2148,6 +2693,27 @@ |
3815 | thousands_sep = -1; |
3816 | } |
3817 | |
3818 | +#if HAVE_MBRTOWC |
3819 | + if (MB_CUR_MAX > 1) |
3820 | + { |
3821 | + inittables = inittables_mb; |
3822 | + begfield = begfield_mb; |
3823 | + limfield = limfield_mb; |
3824 | + getmonth = getmonth_mb; |
3825 | + keycompare = keycompare_mb; |
3826 | + numcompare = numcompare_mb; |
3827 | + } |
3828 | + else |
3829 | +#endif |
3830 | + { |
3831 | + inittables = inittables_uni; |
3832 | + begfield = begfield_uni; |
3833 | + limfield = limfield_uni; |
3834 | + getmonth = getmonth_uni; |
3835 | + keycompare = keycompare_uni; |
3836 | + numcompare = numcompare_uni; |
3837 | + } |
3838 | + |
3839 | have_read_stdin = false; |
3840 | inittables (); |
3841 | |
3842 | @@ -2349,13 +2915,35 @@ |
3843 | |
3844 | case 't': |
3845 | { |
3846 | - char newtab = optarg[0]; |
3847 | - if (! newtab) |
3848 | + char newtab[MB_LEN_MAX + 1]; |
3849 | + size_t newtab_length = 1; |
3850 | + strncpy (newtab, optarg, MB_LEN_MAX); |
3851 | + if (! newtab[0]) |
3852 | error (SORT_FAILURE, 0, _("empty tab")); |
3853 | - if (optarg[1]) |
3854 | +#if HAVE_MBRTOWC |
3855 | + if (MB_CUR_MAX > 1) |
3856 | + { |
3857 | + wchar_t wc; |
3858 | + mbstate_t state; |
3859 | + size_t i; |
3860 | + |
3861 | + memset (&state, '\0', sizeof (mbstate_t)); |
3862 | + newtab_length = mbrtowc (&wc, newtab, strnlen (newtab, |
3863 | + MB_LEN_MAX), |
3864 | + &state); |
3865 | + switch (newtab_length) |
3866 | + { |
3867 | + case (size_t) -1: |
3868 | + case (size_t) -2: |
3869 | + case 0: |
3870 | + newtab_length = 1; |
3871 | + } |
3872 | + } |
3873 | +#endif |
3874 | + if (newtab_length == 1 && optarg[1]) |
3875 | { |
3876 | if (STREQ (optarg, "\\0")) |
3877 | - newtab = '\0'; |
3878 | + newtab[0] = '\0'; |
3879 | else |
3880 | { |
3881 | /* Provoke with `sort -txx'. Complain about |
3882 | @@ -2366,9 +2954,12 @@ |
3883 | quote (optarg)); |
3884 | } |
3885 | } |
3886 | - if (tab != TAB_DEFAULT && tab != newtab) |
3887 | + if (tab_length |
3888 | + && (tab_length != newtab_length |
3889 | + || memcmp (tab, newtab, tab_length) != 0)) |
3890 | error (SORT_FAILURE, 0, _("incompatible tabs")); |
3891 | - tab = newtab; |
3892 | + memcpy (tab, newtab, newtab_length); |
3893 | + tab_length = newtab_length; |
3894 | } |
3895 | break; |
3896 | |
3897 | --- /dev/null 2005-10-10 09:36:06.437701000 +0100 |
3898 | +++ coreutils-5.93/tests/sort/sort-mb-tests 2005-12-23 08:53:01.000000000 +0000 |
3899 | @@ -0,0 +1,58 @@ |
3900 | +#! /bin/sh |
3901 | +case $# in |
3902 | + 0) xx='../../src/sort';; |
3903 | + *) xx="$1";; |
3904 | +esac |
3905 | +test "$VERBOSE" && echo=echo || echo=: |
3906 | +$echo testing program: $xx |
3907 | +errors=0 |
3908 | +test "$srcdir" || srcdir=. |
3909 | +test "$VERBOSE" && $xx --version 2> /dev/null |
3910 | + |
3911 | +export LC_ALL=en_US.UTF-8 |
3912 | +locale -k LC_CTYPE 2>&1 | grep -q charmap.*UTF-8 || exit 77 |
3913 | +errors=0 |
3914 | + |
3915 | +$xx -t @ -k2 -n mb1.I > mb1.O |
3916 | +code=$? |
3917 | +if test $code != 0; then |
3918 | + $echo "Test mb1 failed: $xx return code $code differs from expected value 0" 1>&2 |
3919 | + errors=`expr $errors + 1` |
3920 | +else |
3921 | + cmp mb1.O $srcdir/mb1.X > /dev/null 2>&1 |
3922 | + case $? in |
3923 | + 0) if test "$VERBOSE"; then $echo "passed mb1"; fi;; |
3924 | + 1) $echo "Test mb1 failed: files mb1.O and $srcdir/mb1.X differ" 1>&2 |
3925 | + (diff -c mb1.O $srcdir/mb1.X) 2> /dev/null |
3926 | + errors=`expr $errors + 1`;; |
3927 | + 2) $echo "Test mb1 may have failed." 1>&2 |
3928 | + $echo The command "cmp mb1.O $srcdir/mb1.X" failed. 1>&2 |
3929 | + errors=`expr $errors + 1`;; |
3930 | + esac |
3931 | +fi |
3932 | + |
3933 | +$xx -t @ -k4 -n mb2.I > mb2.O |
3934 | +code=$? |
3935 | +if test $code != 0; then |
3936 | + $echo "Test mb2 failed: $xx return code $code differs from expected value 0" 1>&2 |
3937 | + errors=`expr $errors + 1` |
3938 | +else |
3939 | + cmp mb2.O $srcdir/mb2.X > /dev/null 2>&1 |
3940 | + case $? in |
3941 | + 0) if test "$VERBOSE"; then $echo "passed mb2"; fi;; |
3942 | + 1) $echo "Test mb2 failed: files mb2.O and $srcdir/mb2.X differ" 1>&2 |
3943 | + (diff -c mb2.O $srcdir/mb2.X) 2> /dev/null |
3944 | + errors=`expr $errors + 1`;; |
3945 | + 2) $echo "Test mb2 may have failed." 1>&2 |
3946 | + $echo The command "cmp mb2.O $srcdir/mb2.X" failed. 1>&2 |
3947 | + errors=`expr $errors + 1`;; |
3948 | + esac |
3949 | +fi |
3950 | + |
3951 | +if test $errors = 0; then |
3952 | + $echo Passed all 113 tests. 1>&2 |
3953 | +else |
3954 | + $echo Failed $errors tests. 1>&2 |
3955 | +fi |
3956 | +test $errors = 0 || errors=1 |
3957 | +exit $errors |
3958 | --- /dev/null 2005-10-10 09:36:06.437701000 +0100 |
3959 | +++ coreutils-5.93/tests/sort/mb1.I 2005-12-23 08:53:01.000000000 +0000 |
3960 | @@ -0,0 +1,4 @@ |
3961 | +Apple@10 |
3962 | +Banana@5 |
3963 | +Citrus@20 |
3964 | +Cherry@30 |
3965 | --- /dev/null 2005-10-10 09:36:06.437701000 +0100 |
3966 | +++ coreutils-5.93/tests/sort/mb2.I 2005-12-23 08:53:01.000000000 +0000 |
3967 | @@ -0,0 +1,4 @@ |
3968 | +Apple@AA10@@20 |
3969 | +Banana@AA5@@30 |
3970 | +Citrus@AA20@@5 |
3971 | +Cherry@AA30@@10 |
3972 | --- /dev/null 2005-10-10 09:36:06.437701000 +0100 |
3973 | +++ coreutils-5.93/tests/sort/mb1.X 2005-12-23 08:53:01.000000000 +0000 |
3974 | @@ -0,0 +1,4 @@ |
3975 | +Banana@5 |
3976 | +Apple@10 |
3977 | +Citrus@20 |
3978 | +Cherry@30 |
3979 | --- /dev/null 2005-10-10 09:36:06.437701000 +0100 |
3980 | +++ coreutils-5.93/tests/sort/mb2.X 2005-12-23 08:53:01.000000000 +0000 |
3981 | @@ -0,0 +1,4 @@ |
3982 | +Citrus@AA20@@5 |
3983 | +Cherry@AA30@@10 |
3984 | +Apple@AA10@@20 |
3985 | +Banana@AA5@@30 |
3986 | --- coreutils-5.93/tests/sort/Makefile.am.i18n 2005-10-24 22:02:25.000000000 +0100 |
3987 | +++ coreutils-5.93/tests/sort/Makefile.am 2005-12-23 08:53:01.000000000 +0000 |
3988 | @@ -43,14 +43,16 @@ |
3989 | nul-nls.E use-nl.O use-nl.E o2.O o2.E nul-tab.O nul-tab.E |
3990 | ##test-files-end |
3991 | |
3992 | -EXTRA_DIST = Test.pm $x-tests $(explicit) $(maint_gen) |
3993 | -noinst_SCRIPTS = $x-tests |
3994 | +run_gen += mb1.O mb2.O |
3995 | + |
3996 | +EXTRA_DIST = Test.pm $x-tests $(explicit) $(maint_gen) mb1.I mb1.X mb2.I mb2.X |
3997 | +noinst_SCRIPTS = $x-tests # $x-mb-tests |
3998 | TESTS_ENVIRONMENT = \ |
3999 | PATH="`pwd`/../../src$(PATH_SEPARATOR)$$PATH" |
4000 | |
4001 | editpl = sed -e 's,@''PERL''@,$(PERL),g' -e 's,@''srcdir''@,$(srcdir),g' |
4002 | |
4003 | -TESTS = $x-tests |
4004 | +TESTS = $x-tests $x-mb-tests |
4005 | |
4006 | mk_script = $(srcdir)/../mk-script |
4007 | $(srcdir)/$x-tests: $(mk_script) Test.pm Makefile.am |
4008 | --- coreutils-5.93/tests/sort/Makefile.in.i18n 2005-11-05 22:12:25.000000000 +0000 |
4009 | +++ coreutils-5.93/tests/sort/Makefile.in 2005-12-23 09:00:37.000000000 +0000 |
4010 | @@ -107,25 +107,25 @@ |
4011 | $(top_srcdir)/m4/onceonly_2_57.m4 $(top_srcdir)/m4/openat.m4 \ |
4012 | $(top_srcdir)/m4/pathmax.m4 $(top_srcdir)/m4/perl.m4 \ |
4013 | $(top_srcdir)/m4/physmem.m4 $(top_srcdir)/m4/po.m4 \ |
4014 | - $(top_srcdir)/m4/posixtm.m4 $(top_srcdir)/m4/posixver.m4 \ |
4015 | - $(top_srcdir)/m4/prereq.m4 $(top_srcdir)/m4/progtest.m4 \ |
4016 | - $(top_srcdir)/m4/putenv.m4 $(top_srcdir)/m4/quote.m4 \ |
4017 | - $(top_srcdir)/m4/quotearg.m4 $(top_srcdir)/m4/readlink.m4 \ |
4018 | - $(top_srcdir)/m4/readtokens.m4 $(top_srcdir)/m4/readutmp.m4 \ |
4019 | - $(top_srcdir)/m4/regex.m4 $(top_srcdir)/m4/rename.m4 \ |
4020 | - $(top_srcdir)/m4/restrict.m4 $(top_srcdir)/m4/rmdir-errno.m4 \ |
4021 | - $(top_srcdir)/m4/rmdir.m4 $(top_srcdir)/m4/root-dev-ino.m4 \ |
4022 | - $(top_srcdir)/m4/rpmatch.m4 $(top_srcdir)/m4/safe-read.m4 \ |
4023 | - $(top_srcdir)/m4/safe-write.m4 $(top_srcdir)/m4/same.m4 \ |
4024 | - $(top_srcdir)/m4/save-cwd.m4 $(top_srcdir)/m4/savedir.m4 \ |
4025 | - $(top_srcdir)/m4/setenv.m4 $(top_srcdir)/m4/settime.m4 \ |
4026 | - $(top_srcdir)/m4/sha1.m4 $(top_srcdir)/m4/sig2str.m4 \ |
4027 | - $(top_srcdir)/m4/signed.m4 $(top_srcdir)/m4/socklen.m4 \ |
4028 | - $(top_srcdir)/m4/sockpfaf.m4 $(top_srcdir)/m4/ssize_t.m4 \ |
4029 | - $(top_srcdir)/m4/st_dm_mode.m4 $(top_srcdir)/m4/stat-macros.m4 \ |
4030 | - $(top_srcdir)/m4/stat-prog.m4 $(top_srcdir)/m4/stat-time.m4 \ |
4031 | - $(top_srcdir)/m4/stdbool.m4 $(top_srcdir)/m4/stdint_h.m4 \ |
4032 | - $(top_srcdir)/m4/stdio-safer.m4 \ |
4033 | + $(top_srcdir)/m4/posix_acl.m4 $(top_srcdir)/m4/posixtm.m4 \ |
4034 | + $(top_srcdir)/m4/posixver.m4 $(top_srcdir)/m4/prereq.m4 \ |
4035 | + $(top_srcdir)/m4/progtest.m4 $(top_srcdir)/m4/putenv.m4 \ |
4036 | + $(top_srcdir)/m4/quote.m4 $(top_srcdir)/m4/quotearg.m4 \ |
4037 | + $(top_srcdir)/m4/readlink.m4 $(top_srcdir)/m4/readtokens.m4 \ |
4038 | + $(top_srcdir)/m4/readutmp.m4 $(top_srcdir)/m4/regex.m4 \ |
4039 | + $(top_srcdir)/m4/rename.m4 $(top_srcdir)/m4/restrict.m4 \ |
4040 | + $(top_srcdir)/m4/rmdir-errno.m4 $(top_srcdir)/m4/rmdir.m4 \ |
4041 | + $(top_srcdir)/m4/root-dev-ino.m4 $(top_srcdir)/m4/rpmatch.m4 \ |
4042 | + $(top_srcdir)/m4/safe-read.m4 $(top_srcdir)/m4/safe-write.m4 \ |
4043 | + $(top_srcdir)/m4/same.m4 $(top_srcdir)/m4/save-cwd.m4 \ |
4044 | + $(top_srcdir)/m4/savedir.m4 $(top_srcdir)/m4/setenv.m4 \ |
4045 | + $(top_srcdir)/m4/settime.m4 $(top_srcdir)/m4/sha1.m4 \ |
4046 | + $(top_srcdir)/m4/sig2str.m4 $(top_srcdir)/m4/signed.m4 \ |
4047 | + $(top_srcdir)/m4/socklen.m4 $(top_srcdir)/m4/sockpfaf.m4 \ |
4048 | + $(top_srcdir)/m4/ssize_t.m4 $(top_srcdir)/m4/st_dm_mode.m4 \ |
4049 | + $(top_srcdir)/m4/stat-macros.m4 $(top_srcdir)/m4/stat-prog.m4 \ |
4050 | + $(top_srcdir)/m4/stat-time.m4 $(top_srcdir)/m4/stdbool.m4 \ |
4051 | + $(top_srcdir)/m4/stdint_h.m4 $(top_srcdir)/m4/stdio-safer.m4 \ |
4052 | $(top_srcdir)/m4/stdlib-safer.m4 $(top_srcdir)/m4/stpcpy.m4 \ |
4053 | $(top_srcdir)/m4/strcase.m4 $(top_srcdir)/m4/strcspn.m4 \ |
4054 | $(top_srcdir)/m4/strdup.m4 $(top_srcdir)/m4/strftime.m4 \ |
4055 | @@ -196,7 +196,6 @@ |
4056 | GLIBC21 = @GLIBC21@ |
4057 | GMSGFMT = @GMSGFMT@ |
4058 | GNU_PACKAGE = @GNU_PACKAGE@ |
4059 | -GREP = @GREP@ |
4060 | HAVE__BOOL = @HAVE__BOOL@ |
4061 | HELP2MAN = @HELP2MAN@ |
4062 | INSTALL_DATA = @INSTALL_DATA@ |
4063 | @@ -207,6 +206,7 @@ |
4064 | INTL_MACOSX_LIBS = @INTL_MACOSX_LIBS@ |
4065 | KMEM_GROUP = @KMEM_GROUP@ |
4066 | LDFLAGS = @LDFLAGS@ |
4067 | +LIBACL = @LIBACL@ |
4068 | LIBICONV = @LIBICONV@ |
4069 | LIBINTL = @LIBINTL@ |
4070 | LIBOBJS = @LIBOBJS@ |
4071 | @@ -217,6 +217,8 @@ |
4072 | LIB_FDATASYNC = @LIB_FDATASYNC@ |
4073 | LIB_GETHRXTIME = @LIB_GETHRXTIME@ |
4074 | LIB_NANOSLEEP = @LIB_NANOSLEEP@ |
4075 | +LIB_PAM = @LIB_PAM@ |
4076 | +LIB_SELINUX = @LIB_SELINUX@ |
4077 | LN_S = @LN_S@ |
4078 | LTLIBICONV = @LTLIBICONV@ |
4079 | LTLIBINTL = @LTLIBINTL@ |
4080 | @@ -268,30 +270,23 @@ |
4081 | build_os = @build_os@ |
4082 | build_vendor = @build_vendor@ |
4083 | datadir = @datadir@ |
4084 | -datarootdir = @datarootdir@ |
4085 | -docdir = @docdir@ |
4086 | -dvidir = @dvidir@ |
4087 | exec_prefix = @exec_prefix@ |
4088 | host = @host@ |
4089 | host_alias = @host_alias@ |
4090 | host_cpu = @host_cpu@ |
4091 | host_os = @host_os@ |
4092 | host_vendor = @host_vendor@ |
4093 | -htmldir = @htmldir@ |
4094 | includedir = @includedir@ |
4095 | infodir = @infodir@ |
4096 | install_sh = @install_sh@ |
4097 | libdir = @libdir@ |
4098 | libexecdir = @libexecdir@ |
4099 | -localedir = @localedir@ |
4100 | localstatedir = @localstatedir@ |
4101 | mandir = @mandir@ |
4102 | mkdir_p = @mkdir_p@ |
4103 | oldincludedir = @oldincludedir@ |
4104 | -pdfdir = @pdfdir@ |
4105 | prefix = @prefix@ |
4106 | program_transform_name = @program_transform_name@ |
4107 | -psdir = @psdir@ |
4108 | sbindir = @sbindir@ |
4109 | sharedstatedir = @sharedstatedir@ |
4110 | sysconfdir = @sysconfdir@ |
4111 | @@ -318,33 +313,37 @@ |
4112 | neg-nls.I neg-nls.X nul-nls.I nul-nls.X use-nl.I use-nl.X o2.I o2.X nul-tab.I \ |
4113 | nul-tab.X |
4114 | |
4115 | -run_gen = n1.O n1.E n2.O n2.E n3.O n3.E n4.O n4.E n5.O n5.E n6.O n6.E n7.O \ |
4116 | -n7.E n8a.O n8a.E n8b.O n8b.E n9a.O n9a.E n9b.O n9b.E n10a.O n10a.E n10b.O \ |
4117 | -n10b.E n11a.O n11a.E n11b.O n11b.E 01a.O 01a.E 02a.O 02a.E 02b.O 02b.E 02c.O \ |
4118 | -02c.E 02m.O 02m.E 02n.O 02n.E 02o.O 02o.E 02p.O 02p.E 03a.O 03a.E 03b.O 03b.E \ |
4119 | -03c.O 03c.E 03d.O 03d.E 03e.O 03e.E 03f.O 03f.E 03g.O 03g.E 03h.O 03h.E 03i.O \ |
4120 | -03i.E 04a.O 04a.E 04b.O 04b.E 04c.O 04c.E 04d.O 04d.E 04e.O 04e.E 05a.O 05a.E \ |
4121 | -05b.O 05b.E 05c.O 05c.E 05d.O 05d.E 05e.O 05e.E 05f.O 05f.E 06a.O 06a.E 06b.O \ |
4122 | -06b.E 06c.O 06c.E 06d.O 06d.E 06e.O 06e.E 06f.O 06f.E 07a.O 07a.E 07b.O 07b.E \ |
4123 | -07c.O 07c.E 07d.O 07d.E 08a.O 08a.E 08b.O 08b.E 09a.O 09a.E 09b.O 09b.E 09c.O \ |
4124 | -09c.E 09d.O 09d.E 10a.O 10a.E 10b.O 10b.E 10c.O 10c.E 10d.O 10d.E 10a0.O \ |
4125 | -10a0.E 10a1.O 10a1.E 10a2.O 10a2.E 10e.O 10e.E 10f.O 10f.E 10g.O 10g.E 11a.O \ |
4126 | -11a.E 11b.O 11b.E 11c.O 11c.E 11d.O 11d.E 12a.O 12a.E 12b.O 12b.E 12c.O 12c.E \ |
4127 | -12d.O 12d.E 13a.O 13a.E 13b.O 13b.E 14a.O 14a.E 14b.O 14b.E 15a.O 15a.E 15b.O \ |
4128 | -15b.E 15c.O 15c.E 15d.O 15d.E 15e.O 15e.E 16a.O 16a.E 17.O 17.E 18a.O 18a.E \ |
4129 | -18b.O 18b.E 18c.O 18c.E 18d.O 18d.E 18e.O 18e.E 19a.O 19a.E 19b.O 19b.E 20a.O \ |
4130 | -20a.E 21a.O 21a.E 21b.O 21b.E 21c.O 21c.E 21d.O 21d.E 21e.O 21e.E 21f.O 21f.E \ |
4131 | -21g.O 21g.E 22a.O 22a.E 22b.O 22b.E no-file1.O no-file1.E o-no-file1.O \ |
4132 | -o-no-file1.E create-empty.O create-empty.E neg-nls.O neg-nls.E nul-nls.O \ |
4133 | -nul-nls.E use-nl.O use-nl.E o2.O o2.E nul-tab.O nul-tab.E |
4134 | - |
4135 | -EXTRA_DIST = Test.pm $x-tests $(explicit) $(maint_gen) |
4136 | -noinst_SCRIPTS = $x-tests |
4137 | +run_gen = n1.O n1.E n2.O n2.E n3.O n3.E n4.O n4.E n5.O n5.E n6.O n6.E \ |
4138 | + n7.O n7.E n8a.O n8a.E n8b.O n8b.E n9a.O n9a.E n9b.O n9b.E \ |
4139 | + n10a.O n10a.E n10b.O n10b.E n11a.O n11a.E n11b.O n11b.E 01a.O \ |
4140 | + 01a.E 02a.O 02a.E 02b.O 02b.E 02c.O 02c.E 02m.O 02m.E 02n.O \ |
4141 | + 02n.E 02o.O 02o.E 02p.O 02p.E 03a.O 03a.E 03b.O 03b.E 03c.O \ |
4142 | + 03c.E 03d.O 03d.E 03e.O 03e.E 03f.O 03f.E 03g.O 03g.E 03h.O \ |
4143 | + 03h.E 03i.O 03i.E 04a.O 04a.E 04b.O 04b.E 04c.O 04c.E 04d.O \ |
4144 | + 04d.E 04e.O 04e.E 05a.O 05a.E 05b.O 05b.E 05c.O 05c.E 05d.O \ |
4145 | + 05d.E 05e.O 05e.E 05f.O 05f.E 06a.O 06a.E 06b.O 06b.E 06c.O \ |
4146 | + 06c.E 06d.O 06d.E 06e.O 06e.E 06f.O 06f.E 07a.O 07a.E 07b.O \ |
4147 | + 07b.E 07c.O 07c.E 07d.O 07d.E 08a.O 08a.E 08b.O 08b.E 09a.O \ |
4148 | + 09a.E 09b.O 09b.E 09c.O 09c.E 09d.O 09d.E 10a.O 10a.E 10b.O \ |
4149 | + 10b.E 10c.O 10c.E 10d.O 10d.E 10a0.O 10a0.E 10a1.O 10a1.E \ |
4150 | + 10a2.O 10a2.E 10e.O 10e.E 10f.O 10f.E 10g.O 10g.E 11a.O 11a.E \ |
4151 | + 11b.O 11b.E 11c.O 11c.E 11d.O 11d.E 12a.O 12a.E 12b.O 12b.E \ |
4152 | + 12c.O 12c.E 12d.O 12d.E 13a.O 13a.E 13b.O 13b.E 14a.O 14a.E \ |
4153 | + 14b.O 14b.E 15a.O 15a.E 15b.O 15b.E 15c.O 15c.E 15d.O 15d.E \ |
4154 | + 15e.O 15e.E 16a.O 16a.E 17.O 17.E 18a.O 18a.E 18b.O 18b.E \ |
4155 | + 18c.O 18c.E 18d.O 18d.E 18e.O 18e.E 19a.O 19a.E 19b.O 19b.E \ |
4156 | + 20a.O 20a.E 21a.O 21a.E 21b.O 21b.E 21c.O 21c.E 21d.O 21d.E \ |
4157 | + 21e.O 21e.E 21f.O 21f.E 21g.O 21g.E 22a.O 22a.E 22b.O 22b.E \ |
4158 | + no-file1.O no-file1.E o-no-file1.O o-no-file1.E create-empty.O \ |
4159 | + create-empty.E neg-nls.O neg-nls.E nul-nls.O nul-nls.E \ |
4160 | + use-nl.O use-nl.E o2.O o2.E nul-tab.O nul-tab.E mb1.O mb2.O |
4161 | +EXTRA_DIST = Test.pm $x-tests $(explicit) $(maint_gen) mb1.I mb1.X mb2.I mb2.X |
4162 | +noinst_SCRIPTS = $x-tests # $x-mb-tests |
4163 | TESTS_ENVIRONMENT = \ |
4164 | PATH="`pwd`/../../src$(PATH_SEPARATOR)$$PATH" |
4165 | |
4166 | editpl = sed -e 's,@''PERL''@,$(PERL),g' -e 's,@''srcdir''@,$(srcdir),g' |
4167 | -TESTS = $x-tests |
4168 | +TESTS = $x-tests $x-mb-tests |
4169 | mk_script = $(srcdir)/../mk-script |
4170 | MAINTAINERCLEANFILES = $x-tests $(maint_gen) |
4171 | CLEANFILES = $(run_gen) |