]>
Commit | Line | Data |
---|---|---|
407c5be3 SS |
1 | diff -urNp coreutils-8.9-orig/lib/linebuffer.h coreutils-8.9/lib/linebuffer.h |
2 | --- coreutils-8.9-orig/lib/linebuffer.h 2011-01-01 22:19:27.000000000 +0100 | |
3 | +++ coreutils-8.9/lib/linebuffer.h 2011-01-04 17:41:55.358888521 +0100 | |
56ae3f82 SS |
4 | @@ -21,6 +21,11 @@ |
5 | ||
6 | # include <stdio.h> | |
7 | ||
8 | +/* Get mbstate_t. */ | |
9 | +# if HAVE_WCHAR_H | |
10 | +# include <wchar.h> | |
11 | +# endif | |
12 | + | |
13 | /* A `struct linebuffer' holds a line of text. */ | |
14 | ||
15 | struct linebuffer | |
16 | @@ -28,6 +33,9 @@ struct linebuffer | |
17 | size_t size; /* Allocated. */ | |
18 | size_t length; /* Used. */ | |
19 | char *buffer; | |
20 | +# if HAVE_WCHAR_H | |
21 | + mbstate_t state; | |
22 | +# endif | |
23 | }; | |
24 | ||
25 | /* Initialize linebuffer LINEBUFFER for use. */ | |
407c5be3 SS |
26 | diff -urNp coreutils-8.9-orig/src/cut.c coreutils-8.9/src/cut.c |
27 | --- coreutils-8.9-orig/src/cut.c 2011-01-01 22:19:23.000000000 +0100 | |
28 | +++ coreutils-8.9/src/cut.c 2011-01-04 17:41:55.361888730 +0100 | |
56ae3f82 SS |
29 | @@ -28,6 +28,11 @@ |
30 | #include <assert.h> | |
31 | #include <getopt.h> | |
32 | #include <sys/types.h> | |
33 | + | |
34 | +/* Get mbstate_t, mbrtowc(). */ | |
35 | +#if HAVE_WCHAR_H | |
36 | +# include <wchar.h> | |
37 | +#endif | |
38 | #include "system.h" | |
39 | ||
40 | #include "error.h" | |
1555d43c | 41 | @@ -37,6 +42,18 @@ |
56ae3f82 SS |
42 | #include "quote.h" |
43 | #include "xstrndup.h" | |
44 | ||
45 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC | |
46 | + installation; work around this configuration error. */ | |
47 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 | |
48 | +# undef MB_LEN_MAX | |
49 | +# define MB_LEN_MAX 16 | |
50 | +#endif | |
51 | + | |
52 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ | |
53 | +#if HAVE_MBRTOWC && defined mbstate_t | |
54 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) | |
55 | +#endif | |
56 | + | |
57 | /* The official name of this program (e.g., no `g' prefix). */ | |
58 | #define PROGRAM_NAME "cut" | |
59 | ||
1555d43c | 60 | @@ -72,6 +89,52 @@ |
56ae3f82 SS |
61 | } \ |
62 | while (0) | |
63 | ||
64 | +/* Refill the buffer BUF to get a multibyte character. */ | |
65 | +#define REFILL_BUFFER(BUF, BUFPOS, BUFLEN, STREAM) \ | |
66 | + do \ | |
67 | + { \ | |
68 | + if (BUFLEN < MB_LEN_MAX && !feof (STREAM) && !ferror (STREAM)) \ | |
69 | + { \ | |
70 | + memmove (BUF, BUFPOS, BUFLEN); \ | |
71 | + BUFLEN += fread (BUF + BUFLEN, sizeof(char), BUFSIZ, STREAM); \ | |
72 | + BUFPOS = BUF; \ | |
73 | + } \ | |
74 | + } \ | |
75 | + while (0) | |
76 | + | |
77 | +/* Get wide character on BUFPOS. BUFPOS is not included after that. | |
78 | + If byte sequence is not valid as a character, CONVFAIL is 1. Otherwise 0. */ | |
79 | +#define GET_NEXT_WC_FROM_BUFFER(WC, BUFPOS, BUFLEN, MBLENGTH, STATE, CONVFAIL) \ | |
80 | + do \ | |
81 | + { \ | |
82 | + mbstate_t state_bak; \ | |
83 | + \ | |
84 | + if (BUFLEN < 1) \ | |
85 | + { \ | |
86 | + WC = WEOF; \ | |
87 | + break; \ | |
88 | + } \ | |
89 | + \ | |
90 | + /* Get a wide character. */ \ | |
91 | + CONVFAIL = 0; \ | |
92 | + state_bak = STATE; \ | |
93 | + MBLENGTH = mbrtowc ((wchar_t *)&WC, BUFPOS, BUFLEN, &STATE); \ | |
94 | + \ | |
95 | + switch (MBLENGTH) \ | |
96 | + { \ | |
97 | + case (size_t)-1: \ | |
98 | + case (size_t)-2: \ | |
99 | + CONVFAIL++; \ | |
100 | + STATE = state_bak; \ | |
101 | + /* Fall througn. */ \ | |
102 | + \ | |
103 | + case 0: \ | |
104 | + MBLENGTH = 1; \ | |
105 | + break; \ | |
106 | + } \ | |
107 | + } \ | |
108 | + while (0) | |
109 | + | |
110 | struct range_pair | |
111 | { | |
112 | size_t lo; | |
1555d43c | 113 | @@ -90,7 +153,7 @@ static char *field_1_buffer; |
56ae3f82 SS |
114 | /* The number of bytes allocated for FIELD_1_BUFFER. */ |
115 | static size_t field_1_bufsize; | |
116 | ||
117 | -/* The largest field or byte index used as an endpoint of a closed | |
118 | +/* The largest byte, character or field index used as an endpoint of a closed | |
119 | or degenerate range specification; this doesn't include the starting | |
120 | index of right-open-ended ranges. For example, with either range spec | |
121 | `2-5,9-', `2-3,5,9-' this variable would be set to 5. */ | |
1555d43c | 122 | @@ -102,10 +165,11 @@ static size_t eol_range_start; |
56ae3f82 SS |
123 | |
124 | /* This is a bit vector. | |
125 | In byte mode, which bytes to output. | |
126 | + In character mode, which characters to output. | |
127 | In field mode, which DELIM-separated fields to output. | |
128 | - Both bytes and fields are numbered starting with 1, | |
129 | + Bytes, characters and fields are numbered starting with 1, | |
130 | so the zeroth bit of this array is unused. | |
131 | - A field or byte K has been selected if | |
132 | + A byte, character or field K has been selected if | |
133 | (K <= MAX_RANGE_ENDPOINT and is_printable_field(K)) | |
134 | || (EOL_RANGE_START > 0 && K >= EOL_RANGE_START). */ | |
135 | static unsigned char *printable_field; | |
1555d43c | 136 | @@ -114,15 +178,25 @@ enum operating_mode |
56ae3f82 SS |
137 | { |
138 | undefined_mode, | |
139 | ||
140 | - /* Output characters that are in the given bytes. */ | |
141 | + /* Output bytes that are at the given positions. */ | |
142 | byte_mode, | |
143 | ||
144 | + /* Output characters that are at the given positions. */ | |
145 | + character_mode, | |
146 | + | |
147 | /* Output the given delimeter-separated fields. */ | |
148 | field_mode | |
149 | }; | |
150 | ||
151 | static enum operating_mode operating_mode; | |
152 | ||
153 | +/* If nonzero, when in byte mode, don't split multibyte characters. */ | |
154 | +static int byte_mode_character_aware; | |
155 | + | |
156 | +/* If nonzero, the function for single byte locale is work | |
157 | + if this program runs on multibyte locale. */ | |
158 | +static int force_singlebyte_mode; | |
159 | + | |
160 | /* If true do not output lines containing no delimeter characters. | |
161 | Otherwise, all such lines are printed. This option is valid only | |
162 | with field mode. */ | |
1555d43c | 163 | @@ -134,6 +208,9 @@ static bool complement; |
56ae3f82 SS |
164 | |
165 | /* The delimeter character for field mode. */ | |
166 | static unsigned char delim; | |
167 | +#if HAVE_WCHAR_H | |
168 | +static wchar_t wcdelim; | |
169 | +#endif | |
170 | ||
171 | /* True if the --output-delimiter=STRING option was specified. */ | |
172 | static bool output_delimiter_specified; | |
1555d43c | 173 | @@ -207,7 +284,7 @@ Mandatory arguments to long options are |
56ae3f82 SS |
174 | -f, --fields=LIST select only these fields; also print any line\n\ |
175 | that contains no delimiter character, unless\n\ | |
176 | the -s option is specified\n\ | |
177 | - -n (ignored)\n\ | |
178 | + -n with -b: don't split multibyte characters\n\ | |
179 | "), stdout); | |
180 | fputs (_("\ | |
181 | --complement complement the set of selected bytes, characters\n\ | |
1555d43c | 182 | @@ -366,7 +443,7 @@ set_fields (const char *fieldstr) |
56ae3f82 SS |
183 | in_digits = false; |
184 | /* Starting a range. */ | |
185 | if (dash_found) | |
186 | - FATAL_ERROR (_("invalid byte or field list")); | |
187 | + FATAL_ERROR (_("invalid byte, character or field list")); | |
188 | dash_found = true; | |
189 | fieldstr++; | |
190 | ||
1555d43c | 191 | @@ -390,14 +467,16 @@ set_fields (const char *fieldstr) |
56ae3f82 SS |
192 | if (!rhs_specified) |
193 | { | |
194 | /* `n-'. From `initial' to end of line. */ | |
195 | - eol_range_start = initial; | |
196 | + if (eol_range_start == 0 || | |
197 | + (eol_range_start != 0 && eol_range_start > initial)) | |
198 | + eol_range_start = initial; | |
199 | field_found = true; | |
200 | } | |
201 | else | |
202 | { | |
203 | /* `m-n' or `-n' (1-n). */ | |
204 | if (value < initial) | |
205 | - FATAL_ERROR (_("invalid decreasing range")); | |
206 | + FATAL_ERROR (_("invalid byte, character or field list")); | |
207 | ||
208 | /* Is there already a range going to end of line? */ | |
209 | if (eol_range_start != 0) | |
1555d43c | 210 | @@ -477,6 +556,9 @@ set_fields (const char *fieldstr) |
56ae3f82 SS |
211 | if (operating_mode == byte_mode) |
212 | error (0, 0, | |
213 | _("byte offset %s is too large"), quote (bad_num)); | |
214 | + else if (operating_mode == character_mode) | |
215 | + error (0, 0, | |
216 | + _("character offset %s is too large"), quote (bad_num)); | |
217 | else | |
218 | error (0, 0, | |
219 | _("field number %s is too large"), quote (bad_num)); | |
1555d43c | 220 | @@ -487,7 +569,7 @@ set_fields (const char *fieldstr) |
56ae3f82 SS |
221 | fieldstr++; |
222 | } | |
223 | else | |
224 | - FATAL_ERROR (_("invalid byte or field list")); | |
225 | + FATAL_ERROR (_("invalid byte, character or field list")); | |
226 | } | |
227 | ||
228 | max_range_endpoint = 0; | |
1555d43c | 229 | @@ -580,6 +662,63 @@ cut_bytes (FILE *stream) |
56ae3f82 SS |
230 | } |
231 | } | |
232 | ||
233 | +#if HAVE_MBRTOWC | |
234 | +/* This function is in use for the following case. | |
235 | + | |
236 | + 1. Read from the stream STREAM, printing to standard output any selected | |
237 | + characters. | |
238 | + | |
239 | + 2. Read from stream STREAM, printing to standard output any selected bytes, | |
240 | + without splitting multibyte characters. */ | |
241 | + | |
242 | +static void | |
243 | +cut_characters_or_cut_bytes_no_split (FILE *stream) | |
244 | +{ | |
245 | + int idx; /* number of bytes or characters in the line so far. */ | |
246 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ | |
247 | + char *bufpos; /* Next read position of BUF. */ | |
248 | + size_t buflen; /* The length of the byte sequence in buf. */ | |
249 | + wint_t wc; /* A gotten wide character. */ | |
250 | + size_t mblength; /* The byte size of a multibyte character which shows | |
251 | + as same character as WC. */ | |
252 | + mbstate_t state; /* State of the stream. */ | |
253 | + int convfail; /* 1, when conversion is failed. Otherwise 0. */ | |
254 | + | |
255 | + idx = 0; | |
256 | + buflen = 0; | |
257 | + bufpos = buf; | |
258 | + memset (&state, '\0', sizeof(mbstate_t)); | |
259 | + | |
260 | + while (1) | |
261 | + { | |
262 | + REFILL_BUFFER (buf, bufpos, buflen, stream); | |
263 | + | |
264 | + GET_NEXT_WC_FROM_BUFFER (wc, bufpos, buflen, mblength, state, convfail); | |
265 | + | |
266 | + if (wc == WEOF) | |
267 | + { | |
268 | + if (idx > 0) | |
269 | + putchar ('\n'); | |
270 | + break; | |
271 | + } | |
272 | + else if (wc == L'\n') | |
273 | + { | |
274 | + putchar ('\n'); | |
275 | + idx = 0; | |
276 | + } | |
277 | + else | |
278 | + { | |
279 | + idx += (operating_mode == byte_mode) ? mblength : 1; | |
280 | + if (print_kth (idx, NULL)) | |
281 | + fwrite (bufpos, mblength, sizeof(char), stdout); | |
282 | + } | |
283 | + | |
284 | + buflen -= mblength; | |
285 | + bufpos += mblength; | |
286 | + } | |
287 | +} | |
288 | +#endif | |
289 | + | |
290 | /* Read from stream STREAM, printing to standard output any selected fields. */ | |
291 | ||
292 | static void | |
1555d43c | 293 | @@ -702,13 +841,192 @@ cut_fields (FILE *stream) |
56ae3f82 SS |
294 | } |
295 | } | |
296 | ||
297 | +#if HAVE_MBRTOWC | |
298 | +static void | |
299 | +cut_fields_mb (FILE *stream) | |
300 | +{ | |
301 | + int c; | |
302 | + unsigned int field_idx; | |
303 | + int found_any_selected_field; | |
304 | + int buffer_first_field; | |
305 | + int empty_input; | |
306 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ | |
307 | + char *bufpos; /* Next read position of BUF. */ | |
308 | + size_t buflen; /* The length of the byte sequence in buf. */ | |
309 | + wint_t wc = 0; /* A gotten wide character. */ | |
310 | + size_t mblength; /* The byte size of a multibyte character which shows | |
311 | + as same character as WC. */ | |
312 | + mbstate_t state; /* State of the stream. */ | |
313 | + int convfail; /* 1, when conversion is failed. Otherwise 0. */ | |
314 | + | |
315 | + found_any_selected_field = 0; | |
316 | + field_idx = 1; | |
317 | + bufpos = buf; | |
318 | + buflen = 0; | |
319 | + memset (&state, '\0', sizeof(mbstate_t)); | |
320 | + | |
321 | + c = getc (stream); | |
322 | + empty_input = (c == EOF); | |
323 | + if (c != EOF) | |
324 | + ungetc (c, stream); | |
325 | + else | |
326 | + wc = WEOF; | |
327 | + | |
328 | + /* To support the semantics of the -s flag, we may have to buffer | |
329 | + all of the first field to determine whether it is `delimited.' | |
330 | + But that is unnecessary if all non-delimited lines must be printed | |
331 | + and the first field has been selected, or if non-delimited lines | |
332 | + must be suppressed and the first field has *not* been selected. | |
333 | + That is because a non-delimited line has exactly one field. */ | |
334 | + buffer_first_field = (suppress_non_delimited ^ !print_kth (1, NULL)); | |
335 | + | |
336 | + while (1) | |
337 | + { | |
338 | + if (field_idx == 1 && buffer_first_field) | |
339 | + { | |
340 | + int len = 0; | |
341 | + | |
342 | + while (1) | |
343 | + { | |
344 | + REFILL_BUFFER (buf, bufpos, buflen, stream); | |
345 | + | |
346 | + GET_NEXT_WC_FROM_BUFFER | |
347 | + (wc, bufpos, buflen, mblength, state, convfail); | |
348 | + | |
349 | + if (wc == WEOF) | |
350 | + break; | |
351 | + | |
352 | + field_1_buffer = xrealloc (field_1_buffer, len + mblength); | |
353 | + memcpy (field_1_buffer + len, bufpos, mblength); | |
354 | + len += mblength; | |
355 | + buflen -= mblength; | |
356 | + bufpos += mblength; | |
357 | + | |
358 | + if (!convfail && (wc == L'\n' || wc == wcdelim)) | |
359 | + break; | |
360 | + } | |
361 | + | |
362 | + if (wc == WEOF) | |
363 | + break; | |
364 | + | |
365 | + /* If the first field extends to the end of line (it is not | |
366 | + delimited) and we are printing all non-delimited lines, | |
367 | + print this one. */ | |
368 | + if (convfail || (!convfail && wc != wcdelim)) | |
369 | + { | |
370 | + if (suppress_non_delimited) | |
371 | + { | |
372 | + /* Empty. */ | |
373 | + } | |
374 | + else | |
375 | + { | |
376 | + fwrite (field_1_buffer, sizeof (char), len, stdout); | |
377 | + /* Make sure the output line is newline terminated. */ | |
378 | + if (convfail || (!convfail && wc != L'\n')) | |
379 | + putchar ('\n'); | |
380 | + } | |
381 | + continue; | |
382 | + } | |
383 | + | |
384 | + if (print_kth (1, NULL)) | |
385 | + { | |
386 | + /* Print the field, but not the trailing delimiter. */ | |
387 | + fwrite (field_1_buffer, sizeof (char), len - 1, stdout); | |
388 | + found_any_selected_field = 1; | |
389 | + } | |
390 | + ++field_idx; | |
391 | + } | |
392 | + | |
393 | + if (wc != WEOF) | |
394 | + { | |
395 | + if (print_kth (field_idx, NULL)) | |
396 | + { | |
397 | + if (found_any_selected_field) | |
398 | + { | |
399 | + fwrite (output_delimiter_string, sizeof (char), | |
400 | + output_delimiter_length, stdout); | |
401 | + } | |
402 | + found_any_selected_field = 1; | |
403 | + } | |
404 | + | |
405 | + while (1) | |
406 | + { | |
407 | + REFILL_BUFFER (buf, bufpos, buflen, stream); | |
408 | + | |
409 | + GET_NEXT_WC_FROM_BUFFER | |
410 | + (wc, bufpos, buflen, mblength, state, convfail); | |
411 | + | |
412 | + if (wc == WEOF) | |
413 | + break; | |
414 | + else if (!convfail && (wc == wcdelim || wc == L'\n')) | |
415 | + { | |
416 | + buflen -= mblength; | |
417 | + bufpos += mblength; | |
418 | + break; | |
419 | + } | |
420 | + | |
421 | + if (print_kth (field_idx, NULL)) | |
422 | + fwrite (bufpos, mblength, sizeof(char), stdout); | |
423 | + | |
424 | + buflen -= mblength; | |
425 | + bufpos += mblength; | |
426 | + } | |
427 | + } | |
428 | + | |
429 | + if ((!convfail || wc == L'\n') && buflen < 1) | |
430 | + wc = WEOF; | |
431 | + | |
432 | + if (!convfail && wc == wcdelim) | |
433 | + ++field_idx; | |
434 | + else if (wc == WEOF || (!convfail && wc == L'\n')) | |
435 | + { | |
436 | + if (found_any_selected_field | |
437 | + || (!empty_input && !(suppress_non_delimited && field_idx == 1))) | |
438 | + putchar ('\n'); | |
439 | + if (wc == WEOF) | |
440 | + break; | |
441 | + field_idx = 1; | |
442 | + found_any_selected_field = 0; | |
443 | + } | |
444 | + } | |
445 | +} | |
446 | +#endif | |
447 | + | |
448 | static void | |
449 | cut_stream (FILE *stream) | |
450 | { | |
451 | - if (operating_mode == byte_mode) | |
452 | - cut_bytes (stream); | |
453 | +#if HAVE_MBRTOWC | |
454 | + if (MB_CUR_MAX > 1 && !force_singlebyte_mode) | |
455 | + { | |
456 | + switch (operating_mode) | |
457 | + { | |
458 | + case byte_mode: | |
459 | + if (byte_mode_character_aware) | |
460 | + cut_characters_or_cut_bytes_no_split (stream); | |
461 | + else | |
462 | + cut_bytes (stream); | |
463 | + break; | |
464 | + | |
465 | + case character_mode: | |
466 | + cut_characters_or_cut_bytes_no_split (stream); | |
467 | + break; | |
468 | + | |
469 | + case field_mode: | |
470 | + cut_fields_mb (stream); | |
471 | + break; | |
472 | + | |
473 | + default: | |
474 | + abort (); | |
475 | + } | |
476 | + } | |
477 | else | |
478 | - cut_fields (stream); | |
479 | +#endif | |
480 | + { | |
481 | + if (operating_mode == field_mode) | |
482 | + cut_fields (stream); | |
483 | + else | |
484 | + cut_bytes (stream); | |
485 | + } | |
486 | } | |
487 | ||
488 | /* Process file FILE to standard output. | |
1555d43c | 489 | @@ -760,6 +1078,8 @@ main (int argc, char **argv) |
56ae3f82 SS |
490 | bool ok; |
491 | bool delim_specified = false; | |
1555d43c | 492 | char *spec_list_string IF_LINT ( = NULL); |
56ae3f82 SS |
493 | + char mbdelim[MB_LEN_MAX + 1]; |
494 | + size_t delimlen = 0; | |
495 | ||
496 | initialize_main (&argc, &argv); | |
497 | set_program_name (argv[0]); | |
1555d43c | 498 | @@ -782,7 +1102,6 @@ main (int argc, char **argv) |
56ae3f82 SS |
499 | switch (optc) |
500 | { | |
501 | case 'b': | |
502 | - case 'c': | |
503 | /* Build the byte list. */ | |
504 | if (operating_mode != undefined_mode) | |
505 | FATAL_ERROR (_("only one type of list may be specified")); | |
1555d43c | 506 | @@ -790,6 +1109,14 @@ main (int argc, char **argv) |
56ae3f82 SS |
507 | spec_list_string = optarg; |
508 | break; | |
509 | ||
510 | + case 'c': | |
511 | + /* Build the character list. */ | |
512 | + if (operating_mode != undefined_mode) | |
513 | + FATAL_ERROR (_("only one type of list may be specified")); | |
514 | + operating_mode = character_mode; | |
515 | + spec_list_string = optarg; | |
516 | + break; | |
517 | + | |
518 | case 'f': | |
519 | /* Build the field list. */ | |
520 | if (operating_mode != undefined_mode) | |
1555d43c | 521 | @@ -801,10 +1128,35 @@ main (int argc, char **argv) |
56ae3f82 SS |
522 | case 'd': |
523 | /* New delimiter. */ | |
524 | /* Interpret -d '' to mean `use the NUL byte as the delimiter.' */ | |
525 | - if (optarg[0] != '\0' && optarg[1] != '\0') | |
526 | - FATAL_ERROR (_("the delimiter must be a single character")); | |
527 | - delim = optarg[0]; | |
528 | - delim_specified = true; | |
529 | + { | |
530 | +#if HAVE_MBRTOWC | |
531 | + if(MB_CUR_MAX > 1) | |
532 | + { | |
533 | + mbstate_t state; | |
534 | + | |
535 | + memset (&state, '\0', sizeof(mbstate_t)); | |
536 | + delimlen = mbrtowc (&wcdelim, optarg, strnlen(optarg, MB_LEN_MAX), &state); | |
537 | + | |
538 | + if (delimlen == (size_t)-1 || delimlen == (size_t)-2) | |
539 | + ++force_singlebyte_mode; | |
540 | + else | |
541 | + { | |
542 | + delimlen = (delimlen < 1) ? 1 : delimlen; | |
543 | + if (wcdelim != L'\0' && *(optarg + delimlen) != '\0') | |
544 | + FATAL_ERROR (_("the delimiter must be a single character")); | |
545 | + memcpy (mbdelim, optarg, delimlen); | |
546 | + } | |
547 | + } | |
548 | + | |
549 | + if (MB_CUR_MAX <= 1 || force_singlebyte_mode) | |
550 | +#endif | |
551 | + { | |
552 | + if (optarg[0] != '\0' && optarg[1] != '\0') | |
553 | + FATAL_ERROR (_("the delimiter must be a single character")); | |
554 | + delim = (unsigned char) optarg[0]; | |
555 | + } | |
556 | + delim_specified = true; | |
557 | + } | |
558 | break; | |
559 | ||
560 | case OUTPUT_DELIMITER_OPTION: | |
1555d43c | 561 | @@ -817,6 +1169,7 @@ main (int argc, char **argv) |
56ae3f82 SS |
562 | break; |
563 | ||
564 | case 'n': | |
565 | + byte_mode_character_aware = 1; | |
566 | break; | |
567 | ||
568 | case 's': | |
1555d43c | 569 | @@ -839,7 +1192,7 @@ main (int argc, char **argv) |
56ae3f82 SS |
570 | if (operating_mode == undefined_mode) |
571 | FATAL_ERROR (_("you must specify a list of bytes, characters, or fields")); | |
572 | ||
573 | - if (delim != '\0' && operating_mode != field_mode) | |
574 | + if (delim_specified && operating_mode != field_mode) | |
575 | FATAL_ERROR (_("an input delimiter may be specified only\ | |
576 | when operating on fields")); | |
577 | ||
1555d43c | 578 | @@ -866,15 +1219,34 @@ main (int argc, char **argv) |
56ae3f82 SS |
579 | } |
580 | ||
581 | if (!delim_specified) | |
582 | - delim = '\t'; | |
583 | + { | |
584 | + delim = '\t'; | |
585 | +#ifdef HAVE_MBRTOWC | |
586 | + wcdelim = L'\t'; | |
587 | + mbdelim[0] = '\t'; | |
588 | + mbdelim[1] = '\0'; | |
589 | + delimlen = 1; | |
590 | +#endif | |
591 | + } | |
592 | ||
593 | if (output_delimiter_string == NULL) | |
594 | { | |
595 | - static char dummy[2]; | |
596 | - dummy[0] = delim; | |
597 | - dummy[1] = '\0'; | |
598 | - output_delimiter_string = dummy; | |
599 | - output_delimiter_length = 1; | |
600 | +#ifdef HAVE_MBRTOWC | |
601 | + if (MB_CUR_MAX > 1 && !force_singlebyte_mode) | |
602 | + { | |
603 | + output_delimiter_string = xstrdup(mbdelim); | |
604 | + output_delimiter_length = delimlen; | |
605 | + } | |
606 | + | |
607 | + if (MB_CUR_MAX <= 1 || force_singlebyte_mode) | |
608 | +#endif | |
609 | + { | |
610 | + static char dummy[2]; | |
611 | + dummy[0] = delim; | |
612 | + dummy[1] = '\0'; | |
613 | + output_delimiter_string = dummy; | |
614 | + output_delimiter_length = 1; | |
615 | + } | |
616 | } | |
617 | ||
618 | if (optind == argc) | |
407c5be3 SS |
619 | diff -urNp coreutils-8.9-orig/src/expand.c coreutils-8.9/src/expand.c |
620 | --- coreutils-8.9-orig/src/expand.c 2011-01-01 22:19:23.000000000 +0100 | |
621 | +++ coreutils-8.9/src/expand.c 2011-01-04 17:41:55.363905562 +0100 | |
1555d43c | 622 | @@ -38,12 +38,29 @@ |
56ae3f82 SS |
623 | #include <stdio.h> |
624 | #include <getopt.h> | |
625 | #include <sys/types.h> | |
626 | + | |
627 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ | |
628 | +#if HAVE_WCHAR_H | |
629 | +# include <wchar.h> | |
630 | +#endif | |
631 | + | |
632 | #include "system.h" | |
633 | #include "error.h" | |
1555d43c | 634 | #include "fadvise.h" |
56ae3f82 SS |
635 | #include "quote.h" |
636 | #include "xstrndup.h" | |
637 | ||
638 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC | |
639 | + installation; work around this configuration error. */ | |
640 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 | |
641 | +# define MB_LEN_MAX 16 | |
642 | +#endif | |
643 | + | |
644 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ | |
645 | +#if HAVE_MBRTOWC && defined mbstate_t | |
646 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) | |
647 | +#endif | |
648 | + | |
649 | /* The official name of this program (e.g., no `g' prefix). */ | |
650 | #define PROGRAM_NAME "expand" | |
651 | ||
1555d43c | 652 | @@ -360,6 +377,142 @@ expand (void) |
56ae3f82 SS |
653 | } |
654 | } | |
655 | ||
656 | +#if HAVE_MBRTOWC | |
657 | +static void | |
658 | +expand_multibyte (void) | |
659 | +{ | |
660 | + FILE *fp; /* Input strem. */ | |
661 | + mbstate_t i_state; /* Current shift state of the input stream. */ | |
662 | + mbstate_t i_state_bak; /* Back up the I_STATE. */ | |
663 | + mbstate_t o_state; /* Current shift state of the output stream. */ | |
664 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ | |
665 | + char *bufpos; /* Next read position of BUF. */ | |
666 | + size_t buflen = 0; /* The length of the byte sequence in buf. */ | |
667 | + wchar_t wc; /* A gotten wide character. */ | |
668 | + size_t mblength; /* The byte size of a multibyte character | |
669 | + which shows as same character as WC. */ | |
670 | + int tab_index = 0; /* Index in `tab_list' of next tabstop. */ | |
671 | + int column = 0; /* Column on screen of the next char. */ | |
672 | + int next_tab_column; /* Column the next tab stop is on. */ | |
673 | + int convert = 1; /* If nonzero, perform translations. */ | |
674 | + | |
675 | + fp = next_file ((FILE *) NULL); | |
676 | + if (fp == NULL) | |
677 | + return; | |
678 | + | |
679 | + memset (&o_state, '\0', sizeof(mbstate_t)); | |
680 | + memset (&i_state, '\0', sizeof(mbstate_t)); | |
681 | + | |
682 | + for (;;) | |
683 | + { | |
684 | + /* Refill the buffer BUF. */ | |
685 | + if (buflen < MB_LEN_MAX && !feof(fp) && !ferror(fp)) | |
686 | + { | |
687 | + memmove (buf, bufpos, buflen); | |
688 | + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, fp); | |
689 | + bufpos = buf; | |
690 | + } | |
691 | + | |
692 | + /* No character is left in BUF. */ | |
693 | + if (buflen < 1) | |
694 | + { | |
695 | + fp = next_file (fp); | |
696 | + | |
697 | + if (fp == NULL) | |
698 | + break; /* No more files. */ | |
699 | + else | |
700 | + { | |
701 | + memset (&i_state, '\0', sizeof(mbstate_t)); | |
702 | + continue; | |
703 | + } | |
704 | + } | |
705 | + | |
706 | + /* Get a wide character. */ | |
707 | + i_state_bak = i_state; | |
708 | + mblength = mbrtowc (&wc, bufpos, buflen, &i_state); | |
709 | + | |
710 | + switch (mblength) | |
711 | + { | |
712 | + case (size_t)-1: /* illegal byte sequence. */ | |
713 | + case (size_t)-2: | |
714 | + mblength = 1; | |
715 | + i_state = i_state_bak; | |
716 | + if (convert) | |
717 | + { | |
718 | + ++column; | |
719 | + if (convert_entire_line == 0) | |
720 | + convert = 0; | |
721 | + } | |
722 | + putchar (*bufpos); | |
723 | + break; | |
724 | + | |
725 | + case 0: /* null. */ | |
726 | + mblength = 1; | |
727 | + if (convert && convert_entire_line == 0) | |
728 | + convert = 0; | |
729 | + putchar ('\0'); | |
730 | + break; | |
731 | + | |
732 | + default: | |
733 | + if (wc == L'\n') /* LF. */ | |
734 | + { | |
735 | + tab_index = 0; | |
736 | + column = 0; | |
737 | + convert = 1; | |
738 | + putchar ('\n'); | |
739 | + } | |
740 | + else if (wc == L'\t' && convert) /* Tab. */ | |
741 | + { | |
742 | + if (tab_size == 0) | |
743 | + { | |
744 | + /* Do not let tab_index == first_free_tab; | |
745 | + stop when it is 1 less. */ | |
746 | + while (tab_index < first_free_tab - 1 | |
747 | + && column >= tab_list[tab_index]) | |
748 | + tab_index++; | |
749 | + next_tab_column = tab_list[tab_index]; | |
750 | + if (tab_index < first_free_tab - 1) | |
751 | + tab_index++; | |
752 | + if (column >= next_tab_column) | |
753 | + next_tab_column = column + 1; | |
754 | + } | |
755 | + else | |
756 | + next_tab_column = column + tab_size - column % tab_size; | |
757 | + | |
758 | + while (column < next_tab_column) | |
759 | + { | |
760 | + putchar (' '); | |
761 | + ++column; | |
762 | + } | |
763 | + } | |
764 | + else /* Others. */ | |
765 | + { | |
766 | + if (convert) | |
767 | + { | |
768 | + if (wc == L'\b') | |
769 | + { | |
770 | + if (column > 0) | |
771 | + --column; | |
772 | + } | |
773 | + else | |
774 | + { | |
775 | + int width; /* The width of WC. */ | |
776 | + | |
777 | + width = wcwidth (wc); | |
778 | + column += (width > 0) ? width : 0; | |
779 | + if (convert_entire_line == 0) | |
780 | + convert = 0; | |
781 | + } | |
782 | + } | |
783 | + fwrite (bufpos, sizeof(char), mblength, stdout); | |
784 | + } | |
785 | + } | |
786 | + buflen -= mblength; | |
787 | + bufpos += mblength; | |
788 | + } | |
789 | +} | |
790 | +#endif | |
791 | + | |
792 | int | |
793 | main (int argc, char **argv) | |
794 | { | |
1555d43c | 795 | @@ -424,7 +577,12 @@ main (int argc, char **argv) |
56ae3f82 SS |
796 | |
797 | file_list = (optind < argc ? &argv[optind] : stdin_argv); | |
798 | ||
799 | - expand (); | |
800 | +#if HAVE_MBRTOWC | |
801 | + if (MB_CUR_MAX > 1) | |
802 | + expand_multibyte (); | |
803 | + else | |
804 | +#endif | |
805 | + expand (); | |
806 | ||
807 | if (have_read_stdin && fclose (stdin) != 0) | |
808 | error (EXIT_FAILURE, errno, "-"); | |
407c5be3 SS |
809 | diff -urNp coreutils-8.9-orig/src/fold.c coreutils-8.9/src/fold.c |
810 | --- coreutils-8.9-orig/src/fold.c 2011-01-01 22:19:23.000000000 +0100 | |
811 | +++ coreutils-8.9/src/fold.c 2011-01-04 17:41:55.366888520 +0100 | |
1555d43c | 812 | @@ -22,12 +22,34 @@ |
56ae3f82 SS |
813 | #include <getopt.h> |
814 | #include <sys/types.h> | |
815 | ||
816 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ | |
817 | +#if HAVE_WCHAR_H | |
818 | +# include <wchar.h> | |
819 | +#endif | |
820 | + | |
821 | +/* Get iswprint(), iswblank(), wcwidth(). */ | |
822 | +#if HAVE_WCTYPE_H | |
823 | +# include <wctype.h> | |
824 | +#endif | |
825 | + | |
826 | #include "system.h" | |
827 | #include "error.h" | |
1555d43c | 828 | #include "fadvise.h" |
56ae3f82 SS |
829 | #include "quote.h" |
830 | #include "xstrtol.h" | |
831 | ||
832 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC | |
833 | + installation; work around this configuration error. */ | |
834 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 | |
835 | +# undef MB_LEN_MAX | |
836 | +# define MB_LEN_MAX 16 | |
837 | +#endif | |
838 | + | |
839 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ | |
840 | +#if HAVE_MBRTOWC && defined mbstate_t | |
841 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) | |
842 | +#endif | |
843 | + | |
844 | #define TAB_WIDTH 8 | |
845 | ||
846 | /* The official name of this program (e.g., no `g' prefix). */ | |
1555d43c | 847 | @@ -35,20 +57,41 @@ |
56ae3f82 SS |
848 | |
849 | #define AUTHORS proper_name ("David MacKenzie") | |
850 | ||
851 | +#define FATAL_ERROR(Message) \ | |
852 | + do \ | |
853 | + { \ | |
854 | + error (0, 0, (Message)); \ | |
855 | + usage (2); \ | |
856 | + } \ | |
857 | + while (0) | |
858 | + | |
859 | +enum operating_mode | |
860 | +{ | |
861 | + /* Fold texts by columns that are at the given positions. */ | |
862 | + column_mode, | |
863 | + | |
864 | + /* Fold texts by bytes that are at the given positions. */ | |
865 | + byte_mode, | |
866 | + | |
867 | + /* Fold texts by characters that are at the given positions. */ | |
868 | + character_mode, | |
869 | +}; | |
870 | + | |
871 | +/* The argument shows current mode. (Default: column_mode) */ | |
872 | +static enum operating_mode operating_mode; | |
873 | + | |
874 | /* If nonzero, try to break on whitespace. */ | |
875 | static bool break_spaces; | |
876 | ||
877 | -/* If nonzero, count bytes, not column positions. */ | |
878 | -static bool count_bytes; | |
879 | - | |
880 | /* If nonzero, at least one of the files we read was standard input. */ | |
881 | static bool have_read_stdin; | |
882 | ||
883 | -static char const shortopts[] = "bsw:0::1::2::3::4::5::6::7::8::9::"; | |
884 | +static char const shortopts[] = "bcsw:0::1::2::3::4::5::6::7::8::9::"; | |
885 | ||
886 | static struct option const longopts[] = | |
887 | { | |
888 | {"bytes", no_argument, NULL, 'b'}, | |
889 | + {"characters", no_argument, NULL, 'c'}, | |
890 | {"spaces", no_argument, NULL, 's'}, | |
891 | {"width", required_argument, NULL, 'w'}, | |
892 | {GETOPT_HELP_OPTION_DECL}, | |
1555d43c | 893 | @@ -78,6 +121,7 @@ Mandatory arguments to long options are |
56ae3f82 SS |
894 | "), stdout); |
895 | fputs (_("\ | |
896 | -b, --bytes count bytes rather than columns\n\ | |
897 | + -c, --characters count characters rather than columns\n\ | |
898 | -s, --spaces break at spaces\n\ | |
899 | -w, --width=WIDTH use WIDTH columns instead of 80\n\ | |
900 | "), stdout); | |
1555d43c | 901 | @@ -95,7 +139,7 @@ Mandatory arguments to long options are |
56ae3f82 SS |
902 | static size_t |
903 | adjust_column (size_t column, char c) | |
904 | { | |
905 | - if (!count_bytes) | |
906 | + if (operating_mode != byte_mode) | |
907 | { | |
908 | if (c == '\b') | |
909 | { | |
1555d43c | 910 | @@ -118,30 +162,14 @@ adjust_column (size_t column, char c) |
56ae3f82 SS |
911 | to stdout, with maximum line length WIDTH. |
912 | Return true if successful. */ | |
913 | ||
914 | -static bool | |
915 | -fold_file (char const *filename, size_t width) | |
916 | +static void | |
917 | +fold_text (FILE *istream, size_t width, int *saved_errno) | |
918 | { | |
919 | - FILE *istream; | |
920 | int c; | |
921 | size_t column = 0; /* Screen column where next char will go. */ | |
922 | size_t offset_out = 0; /* Index in `line_out' for next char. */ | |
923 | static char *line_out = NULL; | |
924 | static size_t allocated_out = 0; | |
925 | - int saved_errno; | |
926 | - | |
927 | - if (STREQ (filename, "-")) | |
928 | - { | |
929 | - istream = stdin; | |
930 | - have_read_stdin = true; | |
931 | - } | |
932 | - else | |
933 | - istream = fopen (filename, "r"); | |
934 | - | |
935 | - if (istream == NULL) | |
936 | - { | |
937 | - error (0, errno, "%s", filename); | |
938 | - return false; | |
939 | - } | |
940 | ||
1555d43c SS |
941 | fadvise (istream, FADVISE_SEQUENTIAL); |
942 | ||
943 | @@ -171,6 +199,15 @@ fold_file (char const *filename, size_t | |
56ae3f82 SS |
944 | bool found_blank = false; |
945 | size_t logical_end = offset_out; | |
946 | ||
947 | + /* If LINE_OUT has no wide character, | |
948 | + put a new wide character in LINE_OUT | |
949 | + if column is bigger than width. */ | |
950 | + if (offset_out == 0) | |
951 | + { | |
952 | + line_out[offset_out++] = c; | |
953 | + continue; | |
954 | + } | |
955 | + | |
956 | /* Look for the last blank. */ | |
957 | while (logical_end) | |
958 | { | |
1555d43c | 959 | @@ -217,11 +254,222 @@ fold_file (char const *filename, size_t |
56ae3f82 SS |
960 | line_out[offset_out++] = c; |
961 | } | |
962 | ||
963 | - saved_errno = errno; | |
964 | + *saved_errno = errno; | |
965 | ||
966 | if (offset_out) | |
967 | fwrite (line_out, sizeof (char), (size_t) offset_out, stdout); | |
968 | ||
969 | +} | |
970 | + | |
971 | +#if HAVE_MBRTOWC | |
972 | +static void | |
973 | +fold_multibyte_text (FILE *istream, size_t width, int *saved_errno) | |
974 | +{ | |
975 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ | |
976 | + size_t buflen = 0; /* The length of the byte sequence in buf. */ | |
977 | + char *bufpos = NULL; /* Next read position of BUF. */ | |
978 | + wint_t wc; /* A gotten wide character. */ | |
979 | + size_t mblength; /* The byte size of a multibyte character which shows | |
980 | + as same character as WC. */ | |
981 | + mbstate_t state, state_bak; /* State of the stream. */ | |
982 | + int convfail; /* 1, when conversion is failed. Otherwise 0. */ | |
983 | + | |
984 | + static char *line_out = NULL; | |
985 | + size_t offset_out = 0; /* Index in `line_out' for next char. */ | |
986 | + static size_t allocated_out = 0; | |
987 | + | |
988 | + int increment; | |
989 | + size_t column = 0; | |
990 | + | |
991 | + size_t last_blank_pos; | |
992 | + size_t last_blank_column; | |
993 | + int is_blank_seen; | |
994 | + int last_blank_increment = 0; | |
995 | + int is_bs_following_last_blank; | |
996 | + size_t bs_following_last_blank_num; | |
997 | + int is_cr_after_last_blank; | |
998 | + | |
999 | +#define CLEAR_FLAGS \ | |
1000 | + do \ | |
1001 | + { \ | |
1002 | + last_blank_pos = 0; \ | |
1003 | + last_blank_column = 0; \ | |
1004 | + is_blank_seen = 0; \ | |
1005 | + is_bs_following_last_blank = 0; \ | |
1006 | + bs_following_last_blank_num = 0; \ | |
1007 | + is_cr_after_last_blank = 0; \ | |
1008 | + } \ | |
1009 | + while (0) | |
1010 | + | |
1011 | +#define START_NEW_LINE \ | |
1012 | + do \ | |
1013 | + { \ | |
1014 | + putchar ('\n'); \ | |
1015 | + column = 0; \ | |
1016 | + offset_out = 0; \ | |
1017 | + CLEAR_FLAGS; \ | |
1018 | + } \ | |
1019 | + while (0) | |
1020 | + | |
1021 | + CLEAR_FLAGS; | |
1022 | + memset (&state, '\0', sizeof(mbstate_t)); | |
1023 | + | |
1024 | + for (;; bufpos += mblength, buflen -= mblength) | |
1025 | + { | |
1026 | + if (buflen < MB_LEN_MAX && !feof (istream) && !ferror (istream)) | |
1027 | + { | |
1028 | + memmove (buf, bufpos, buflen); | |
1029 | + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, istream); | |
1030 | + bufpos = buf; | |
1031 | + } | |
1032 | + | |
1033 | + if (buflen < 1) | |
1034 | + break; | |
1035 | + | |
1036 | + /* Get a wide character. */ | |
1037 | + convfail = 0; | |
1038 | + state_bak = state; | |
1039 | + mblength = mbrtowc ((wchar_t *)&wc, bufpos, buflen, &state); | |
1040 | + | |
1041 | + switch (mblength) | |
1042 | + { | |
1043 | + case (size_t)-1: | |
1044 | + case (size_t)-2: | |
1045 | + convfail++; | |
1046 | + state = state_bak; | |
1047 | + /* Fall through. */ | |
1048 | + | |
1049 | + case 0: | |
1050 | + mblength = 1; | |
1051 | + break; | |
1052 | + } | |
1053 | + | |
1054 | +rescan: | |
1055 | + if (operating_mode == byte_mode) /* byte mode */ | |
1056 | + increment = mblength; | |
1057 | + else if (operating_mode == character_mode) /* character mode */ | |
1058 | + increment = 1; | |
1059 | + else /* column mode */ | |
1060 | + { | |
1061 | + if (convfail) | |
1062 | + increment = 1; | |
1063 | + else | |
1064 | + { | |
1065 | + switch (wc) | |
1066 | + { | |
1067 | + case L'\n': | |
1068 | + fwrite (line_out, sizeof(char), offset_out, stdout); | |
1069 | + START_NEW_LINE; | |
1070 | + continue; | |
1071 | + | |
1072 | + case L'\b': | |
1073 | + increment = (column > 0) ? -1 : 0; | |
1074 | + break; | |
1075 | + | |
1076 | + case L'\r': | |
1077 | + increment = -1 * column; | |
1078 | + break; | |
1079 | + | |
1080 | + case L'\t': | |
1081 | + increment = 8 - column % 8; | |
1082 | + break; | |
1083 | + | |
1084 | + default: | |
1085 | + increment = wcwidth (wc); | |
1086 | + increment = (increment < 0) ? 0 : increment; | |
1087 | + } | |
1088 | + } | |
1089 | + } | |
1090 | + | |
1091 | + if (column + increment > width && break_spaces && last_blank_pos) | |
1092 | + { | |
1093 | + fwrite (line_out, sizeof(char), last_blank_pos, stdout); | |
1094 | + putchar ('\n'); | |
1095 | + | |
1096 | + offset_out = offset_out - last_blank_pos; | |
1097 | + column = column - last_blank_column + ((is_cr_after_last_blank) | |
1098 | + ? last_blank_increment : bs_following_last_blank_num); | |
1099 | + memmove (line_out, line_out + last_blank_pos, offset_out); | |
1100 | + CLEAR_FLAGS; | |
1101 | + goto rescan; | |
1102 | + } | |
1103 | + | |
1104 | + if (column + increment > width && column != 0) | |
1105 | + { | |
1106 | + fwrite (line_out, sizeof(char), offset_out, stdout); | |
1107 | + START_NEW_LINE; | |
1108 | + goto rescan; | |
1109 | + } | |
1110 | + | |
1111 | + if (allocated_out < offset_out + mblength) | |
1112 | + { | |
1113 | + line_out = X2REALLOC (line_out, &allocated_out); | |
1114 | + } | |
1115 | + | |
1116 | + memcpy (line_out + offset_out, bufpos, mblength); | |
1117 | + offset_out += mblength; | |
1118 | + column += increment; | |
1119 | + | |
1120 | + if (is_blank_seen && !convfail && wc == L'\r') | |
1121 | + is_cr_after_last_blank = 1; | |
1122 | + | |
1123 | + if (is_bs_following_last_blank && !convfail && wc == L'\b') | |
1124 | + ++bs_following_last_blank_num; | |
1125 | + else | |
1126 | + is_bs_following_last_blank = 0; | |
1127 | + | |
1128 | + if (break_spaces && !convfail && iswblank (wc)) | |
1129 | + { | |
1130 | + last_blank_pos = offset_out; | |
1131 | + last_blank_column = column; | |
1132 | + is_blank_seen = 1; | |
1133 | + last_blank_increment = increment; | |
1134 | + is_bs_following_last_blank = 1; | |
1135 | + bs_following_last_blank_num = 0; | |
1136 | + is_cr_after_last_blank = 0; | |
1137 | + } | |
1138 | + } | |
1139 | + | |
1140 | + *saved_errno = errno; | |
1141 | + | |
1142 | + if (offset_out) | |
1143 | + fwrite (line_out, sizeof (char), (size_t) offset_out, stdout); | |
1144 | + | |
1145 | +} | |
1146 | +#endif | |
1147 | + | |
1148 | +/* Fold file FILENAME, or standard input if FILENAME is "-", | |
1149 | + to stdout, with maximum line length WIDTH. | |
1150 | + Return 0 if successful, 1 if an error occurs. */ | |
1151 | + | |
1152 | +static bool | |
1153 | +fold_file (char *filename, size_t width) | |
1154 | +{ | |
1155 | + FILE *istream; | |
1156 | + int saved_errno; | |
1157 | + | |
1158 | + if (STREQ (filename, "-")) | |
1159 | + { | |
1160 | + istream = stdin; | |
1161 | + have_read_stdin = 1; | |
1162 | + } | |
1163 | + else | |
1164 | + istream = fopen (filename, "r"); | |
1165 | + | |
1166 | + if (istream == NULL) | |
1167 | + { | |
1168 | + error (0, errno, "%s", filename); | |
1169 | + return 1; | |
1170 | + } | |
1171 | + | |
1172 | + /* Define how ISTREAM is being folded. */ | |
1173 | +#if HAVE_MBRTOWC | |
1174 | + if (MB_CUR_MAX > 1) | |
1175 | + fold_multibyte_text (istream, width, &saved_errno); | |
1176 | + else | |
1177 | +#endif | |
1178 | + fold_text (istream, width, &saved_errno); | |
1179 | + | |
1180 | if (ferror (istream)) | |
1181 | { | |
1182 | error (0, saved_errno, "%s", filename); | |
1555d43c | 1183 | @@ -254,7 +502,8 @@ main (int argc, char **argv) |
56ae3f82 SS |
1184 | |
1185 | atexit (close_stdout); | |
1186 | ||
1187 | - break_spaces = count_bytes = have_read_stdin = false; | |
1188 | + operating_mode = column_mode; | |
1189 | + break_spaces = have_read_stdin = false; | |
1190 | ||
1191 | while ((optc = getopt_long (argc, argv, shortopts, longopts, NULL)) != -1) | |
1192 | { | |
1555d43c | 1193 | @@ -263,7 +512,15 @@ main (int argc, char **argv) |
56ae3f82 SS |
1194 | switch (optc) |
1195 | { | |
1196 | case 'b': /* Count bytes rather than columns. */ | |
1197 | - count_bytes = true; | |
1198 | + if (operating_mode != column_mode) | |
1199 | + FATAL_ERROR (_("only one way of folding may be specified")); | |
1200 | + operating_mode = byte_mode; | |
1201 | + break; | |
1202 | + | |
1203 | + case 'c': | |
1204 | + if (operating_mode != column_mode) | |
1205 | + FATAL_ERROR (_("only one way of folding may be specified")); | |
1206 | + operating_mode = character_mode; | |
1207 | break; | |
1208 | ||
1209 | case 's': /* Break at word boundaries. */ | |
407c5be3 SS |
1210 | diff -urNp coreutils-8.9-orig/src/join.c coreutils-8.9/src/join.c |
1211 | --- coreutils-8.9-orig/src/join.c 2011-01-01 22:19:23.000000000 +0100 | |
1212 | +++ coreutils-8.9/src/join.c 2011-01-04 17:41:55.369888660 +0100 | |
1555d43c | 1213 | @@ -22,18 +22,32 @@ |
56ae3f82 SS |
1214 | #include <sys/types.h> |
1215 | #include <getopt.h> | |
1216 | ||
1217 | +/* Get mbstate_t, mbrtowc(), mbrtowc(), wcwidth(). */ | |
1218 | +#if HAVE_WCHAR_H | |
1219 | +# include <wchar.h> | |
1220 | +#endif | |
1221 | + | |
1222 | +/* Get iswblank(), towupper. */ | |
1223 | +#if HAVE_WCTYPE_H | |
1224 | +# include <wctype.h> | |
1225 | +#endif | |
1226 | + | |
1227 | #include "system.h" | |
1228 | #include "error.h" | |
1555d43c | 1229 | #include "fadvise.h" |
56ae3f82 SS |
1230 | #include "hard-locale.h" |
1231 | #include "linebuffer.h" | |
1232 | -#include "memcasecmp.h" | |
1233 | #include "quote.h" | |
1234 | #include "stdio--.h" | |
1235 | #include "xmemcoll.h" | |
1236 | #include "xstrtol.h" | |
1237 | #include "argmatch.h" | |
1238 | ||
1239 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ | |
1240 | +#if HAVE_MBRTOWC && defined mbstate_t | |
1241 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) | |
1242 | +#endif | |
1243 | + | |
1244 | /* The official name of this program (e.g., no `g' prefix). */ | |
1245 | #define PROGRAM_NAME "join" | |
1246 | ||
1555d43c | 1247 | @@ -122,10 +136,12 @@ static struct outlist outlist_head; |
56ae3f82 SS |
1248 | /* Last element in `outlist', where a new element can be added. */ |
1249 | static struct outlist *outlist_end = &outlist_head; | |
1250 | ||
1251 | -/* Tab character separating fields. If negative, fields are separated | |
1252 | - by any nonempty string of blanks, otherwise by exactly one | |
1253 | - tab character whose value (when cast to unsigned char) equals TAB. */ | |
1254 | -static int tab = -1; | |
1255 | +/* Tab character separating fields. If NULL, fields are separated | |
1256 | + by any nonempty string of blanks. */ | |
1257 | +static char *tab = NULL; | |
1258 | + | |
1259 | +/* The number of bytes used for tab. */ | |
1260 | +static size_t tablen = 0; | |
1261 | ||
1262 | /* If nonzero, check that the input is correctly ordered. */ | |
1263 | static enum | |
1555d43c | 1264 | @@ -249,13 +265,14 @@ xfields (struct line *line) |
56ae3f82 SS |
1265 | if (ptr == lim) |
1266 | return; | |
1267 | ||
1555d43c | 1268 | - if (0 <= tab && tab != '\n') |
56ae3f82 SS |
1269 | + if (tab != NULL) |
1270 | { | |
1271 | + unsigned char t = tab[0]; | |
1272 | char *sep; | |
1273 | - for (; (sep = memchr (ptr, tab, lim - ptr)) != NULL; ptr = sep + 1) | |
1274 | + for (; (sep = memchr (ptr, t, lim - ptr)) != NULL; ptr = sep + 1) | |
1275 | extract_field (line, ptr, sep - ptr); | |
1276 | } | |
1555d43c SS |
1277 | - else if (tab < 0) |
1278 | + else | |
1279 | { | |
1280 | /* Skip leading blanks before the first field. */ | |
1281 | while (isblank (to_uchar (*ptr))) | |
1282 | @@ -279,6 +296,148 @@ xfields (struct line *line) | |
56ae3f82 SS |
1283 | extract_field (line, ptr, lim - ptr); |
1284 | } | |
1285 | ||
1286 | +#if HAVE_MBRTOWC | |
1287 | +static void | |
1288 | +xfields_multibyte (struct line *line) | |
1289 | +{ | |
1290 | + char *ptr = line->buf.buffer; | |
1291 | + char const *lim = ptr + line->buf.length - 1; | |
1292 | + wchar_t wc = 0; | |
1293 | + size_t mblength = 1; | |
1294 | + mbstate_t state, state_bak; | |
1295 | + | |
1296 | + memset (&state, 0, sizeof (mbstate_t)); | |
1297 | + | |
1298 | + if (ptr >= lim) | |
1299 | + return; | |
1300 | + | |
1301 | + if (tab != NULL) | |
1302 | + { | |
1303 | + unsigned char t = tab[0]; | |
1304 | + char *sep = ptr; | |
1305 | + for (; ptr < lim; ptr = sep + mblength) | |
1306 | + { | |
1307 | + sep = ptr; | |
1308 | + while (sep < lim) | |
1309 | + { | |
1310 | + state_bak = state; | |
1311 | + mblength = mbrtowc (&wc, sep, lim - sep + 1, &state); | |
1312 | + | |
1313 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) | |
1314 | + { | |
1315 | + mblength = 1; | |
1316 | + state = state_bak; | |
1317 | + } | |
1318 | + mblength = (mblength < 1) ? 1 : mblength; | |
1319 | + | |
1320 | + if (mblength == tablen && !memcmp (sep, tab, mblength)) | |
1321 | + break; | |
1322 | + else | |
1323 | + { | |
1324 | + sep += mblength; | |
1325 | + continue; | |
1326 | + } | |
1327 | + } | |
1328 | + | |
1329 | + if (sep >= lim) | |
1330 | + break; | |
1331 | + | |
1332 | + extract_field (line, ptr, sep - ptr); | |
1333 | + } | |
1334 | + } | |
1335 | + else | |
1336 | + { | |
1337 | + /* Skip leading blanks before the first field. */ | |
1338 | + while(ptr < lim) | |
1339 | + { | |
1340 | + state_bak = state; | |
1341 | + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); | |
1342 | + | |
1343 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) | |
1344 | + { | |
1345 | + mblength = 1; | |
1346 | + state = state_bak; | |
1347 | + break; | |
1348 | + } | |
1349 | + mblength = (mblength < 1) ? 1 : mblength; | |
1350 | + | |
1351 | + if (!iswblank(wc)) | |
1352 | + break; | |
1353 | + ptr += mblength; | |
1354 | + } | |
1355 | + | |
1356 | + do | |
1357 | + { | |
1358 | + char *sep; | |
1359 | + state_bak = state; | |
1360 | + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); | |
1361 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) | |
1362 | + { | |
1363 | + mblength = 1; | |
1364 | + state = state_bak; | |
1365 | + break; | |
1366 | + } | |
1367 | + mblength = (mblength < 1) ? 1 : mblength; | |
1368 | + | |
1369 | + sep = ptr + mblength; | |
1370 | + while (sep < lim) | |
1371 | + { | |
1372 | + state_bak = state; | |
1373 | + mblength = mbrtowc (&wc, sep, lim - sep + 1, &state); | |
1374 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) | |
1375 | + { | |
1376 | + mblength = 1; | |
1377 | + state = state_bak; | |
1378 | + break; | |
1379 | + } | |
1380 | + mblength = (mblength < 1) ? 1 : mblength; | |
1381 | + | |
1382 | + if (iswblank (wc)) | |
1383 | + break; | |
1384 | + | |
1385 | + sep += mblength; | |
1386 | + } | |
1387 | + | |
1388 | + extract_field (line, ptr, sep - ptr); | |
1389 | + if (sep >= lim) | |
1390 | + return; | |
1391 | + | |
1392 | + state_bak = state; | |
1393 | + mblength = mbrtowc (&wc, sep, lim - sep + 1, &state); | |
1394 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) | |
1395 | + { | |
1396 | + mblength = 1; | |
1397 | + state = state_bak; | |
1398 | + break; | |
1399 | + } | |
1400 | + mblength = (mblength < 1) ? 1 : mblength; | |
1401 | + | |
1402 | + ptr = sep + mblength; | |
1403 | + while (ptr < lim) | |
1404 | + { | |
1405 | + state_bak = state; | |
1406 | + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); | |
1407 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) | |
1408 | + { | |
1409 | + mblength = 1; | |
1410 | + state = state_bak; | |
1411 | + break; | |
1412 | + } | |
1413 | + mblength = (mblength < 1) ? 1 : mblength; | |
1414 | + | |
1415 | + if (!iswblank (wc)) | |
1416 | + break; | |
1417 | + | |
1418 | + ptr += mblength; | |
1419 | + } | |
1420 | + } | |
1421 | + while (ptr < lim); | |
1422 | + } | |
1423 | + | |
1424 | + extract_field (line, ptr, lim - ptr); | |
1425 | +} | |
1426 | +#endif | |
1427 | + | |
1428 | static void | |
1429 | freeline (struct line *line) | |
1430 | { | |
1555d43c | 1431 | @@ -300,56 +459,115 @@ keycmp (struct line const *line1, struct |
56ae3f82 SS |
1432 | size_t jf_1, size_t jf_2) |
1433 | { | |
1434 | /* Start of field to compare in each file. */ | |
1435 | - char *beg1; | |
1436 | - char *beg2; | |
1437 | - | |
1438 | - size_t len1; | |
1439 | - size_t len2; /* Length of fields to compare. */ | |
1440 | + char *beg[2]; | |
1441 | + char *copy[2]; | |
1442 | + size_t len[2]; /* Length of fields to compare. */ | |
1443 | int diff; | |
1444 | + int i, j; | |
1445 | ||
1446 | if (jf_1 < line1->nfields) | |
1447 | { | |
1448 | - beg1 = line1->fields[jf_1].beg; | |
1449 | - len1 = line1->fields[jf_1].len; | |
1450 | + beg[0] = line1->fields[jf_1].beg; | |
1451 | + len[0] = line1->fields[jf_1].len; | |
1452 | } | |
1453 | else | |
1454 | { | |
1455 | - beg1 = NULL; | |
1456 | - len1 = 0; | |
1457 | + beg[0] = NULL; | |
1458 | + len[0] = 0; | |
1459 | } | |
1460 | ||
1461 | if (jf_2 < line2->nfields) | |
1462 | { | |
1463 | - beg2 = line2->fields[jf_2].beg; | |
1464 | - len2 = line2->fields[jf_2].len; | |
1465 | + beg[1] = line2->fields[jf_2].beg; | |
1466 | + len[1] = line2->fields[jf_2].len; | |
1467 | } | |
1468 | else | |
1469 | { | |
1470 | - beg2 = NULL; | |
1471 | - len2 = 0; | |
1472 | + beg[1] = NULL; | |
1473 | + len[1] = 0; | |
1474 | } | |
1475 | ||
1476 | - if (len1 == 0) | |
1477 | - return len2 == 0 ? 0 : -1; | |
1478 | - if (len2 == 0) | |
1479 | + if (len[0] == 0) | |
1480 | + return len[1] == 0 ? 0 : -1; | |
1481 | + if (len[1] == 0) | |
1482 | return 1; | |
1483 | ||
1484 | if (ignore_case) | |
1485 | { | |
1486 | - /* FIXME: ignore_case does not work with NLS (in particular, | |
1487 | - with multibyte chars). */ | |
1488 | - diff = memcasecmp (beg1, beg2, MIN (len1, len2)); | |
1489 | +#ifdef HAVE_MBRTOWC | |
1490 | + if (MB_CUR_MAX > 1) | |
1491 | + { | |
1492 | + size_t mblength; | |
1493 | + wchar_t wc, uwc; | |
1494 | + mbstate_t state, state_bak; | |
1495 | + | |
1496 | + memset (&state, '\0', sizeof (mbstate_t)); | |
1497 | + | |
1498 | + for (i = 0; i < 2; i++) | |
1499 | + { | |
1500 | + copy[i] = alloca (len[i] + 1); | |
1501 | + | |
1502 | + for (j = 0; j < MIN (len[0], len[1]);) | |
1503 | + { | |
1504 | + state_bak = state; | |
1505 | + mblength = mbrtowc (&wc, beg[i] + j, len[i] - j, &state); | |
1506 | + | |
1507 | + switch (mblength) | |
1508 | + { | |
1509 | + case (size_t) -1: | |
1510 | + case (size_t) -2: | |
1511 | + state = state_bak; | |
1512 | + /* Fall through */ | |
1513 | + case 0: | |
1514 | + mblength = 1; | |
1515 | + break; | |
1516 | + | |
1517 | + default: | |
1518 | + uwc = towupper (wc); | |
1519 | + | |
1520 | + if (uwc != wc) | |
1521 | + { | |
1522 | + mbstate_t state_wc; | |
1523 | + | |
1524 | + memset (&state_wc, '\0', sizeof (mbstate_t)); | |
1525 | + wcrtomb (copy[i] + j, uwc, &state_wc); | |
1526 | + } | |
1527 | + else | |
1528 | + memcpy (copy[i] + j, beg[i] + j, mblength); | |
1529 | + } | |
1530 | + j += mblength; | |
1531 | + } | |
1532 | + copy[i][j] = '\0'; | |
1533 | + } | |
1534 | + } | |
1535 | + else | |
1536 | +#endif | |
1537 | + { | |
1538 | + for (i = 0; i < 2; i++) | |
1539 | + { | |
1540 | + copy[i] = alloca (len[i] + 1); | |
1541 | + | |
1542 | + for (j = 0; j < MIN (len[0], len[1]); j++) | |
1543 | + copy[i][j] = toupper (beg[i][j]); | |
1544 | + | |
1545 | + copy[i][j] = '\0'; | |
1546 | + } | |
1547 | + } | |
1548 | } | |
1549 | else | |
1550 | { | |
1551 | - if (hard_LC_COLLATE) | |
1552 | - return xmemcoll (beg1, len1, beg2, len2); | |
1553 | - diff = memcmp (beg1, beg2, MIN (len1, len2)); | |
1554 | + copy[0] = (unsigned char *) beg[0]; | |
1555 | + copy[1] = (unsigned char *) beg[1]; | |
1556 | } | |
1557 | ||
1558 | + if (hard_LC_COLLATE) | |
1559 | + return xmemcoll ((char *) copy[0], len[0], (char *) copy[1], len[1]); | |
1560 | + diff = memcmp (copy[0], copy[1], MIN (len[0], len[1])); | |
1561 | + | |
1562 | + | |
1563 | if (diff) | |
1564 | return diff; | |
1565 | - return len1 < len2 ? -1 : len1 != len2; | |
1566 | + return len[0] - len[1]; | |
1567 | } | |
1568 | ||
1569 | /* Check that successive input lines PREV and CURRENT from input file | |
1555d43c | 1570 | @@ -430,6 +648,11 @@ get_line (FILE *fp, struct line **linep, |
56ae3f82 SS |
1571 | return false; |
1572 | } | |
1573 | ||
1574 | +#if HAVE_MBRTOWC | |
1575 | + if (MB_CUR_MAX > 1) | |
1576 | + xfields_multibyte (line); | |
1577 | + else | |
1578 | +#endif | |
1579 | xfields (line); | |
1580 | ||
1581 | if (prevline[which - 1]) | |
1555d43c | 1582 | @@ -529,11 +752,18 @@ prfield (size_t n, struct line const *li |
56ae3f82 SS |
1583 | |
1584 | /* Print the join of LINE1 and LINE2. */ | |
1585 | ||
1586 | +#define PUT_TAB_CHAR \ | |
1587 | + do \ | |
1588 | + { \ | |
1589 | + (tab != NULL) ? \ | |
1590 | + fwrite(tab, sizeof(char), tablen, stdout) : putchar (' '); \ | |
1591 | + } \ | |
1592 | + while (0) | |
1593 | + | |
1594 | static void | |
1595 | prjoin (struct line const *line1, struct line const *line2) | |
1596 | { | |
1597 | const struct outlist *outlist; | |
1598 | - char output_separator = tab < 0 ? ' ' : tab; | |
1599 | ||
1600 | outlist = outlist_head.next; | |
1601 | if (outlist) | |
1555d43c | 1602 | @@ -568,7 +798,7 @@ prjoin (struct line const *line1, struct |
56ae3f82 SS |
1603 | o = o->next; |
1604 | if (o == NULL) | |
1605 | break; | |
1606 | - putchar (output_separator); | |
1607 | + PUT_TAB_CHAR; | |
1608 | } | |
1609 | putchar ('\n'); | |
1610 | } | |
1555d43c | 1611 | @@ -586,23 +816,23 @@ prjoin (struct line const *line1, struct |
56ae3f82 SS |
1612 | prfield (join_field_1, line1); |
1613 | for (i = 0; i < join_field_1 && i < line1->nfields; ++i) | |
1614 | { | |
1615 | - putchar (output_separator); | |
1616 | + PUT_TAB_CHAR; | |
1617 | prfield (i, line1); | |
1618 | } | |
1619 | for (i = join_field_1 + 1; i < line1->nfields; ++i) | |
1620 | { | |
1621 | - putchar (output_separator); | |
1622 | + PUT_TAB_CHAR; | |
1623 | prfield (i, line1); | |
1624 | } | |
1625 | ||
1626 | for (i = 0; i < join_field_2 && i < line2->nfields; ++i) | |
1627 | { | |
1628 | - putchar (output_separator); | |
1629 | + PUT_TAB_CHAR; | |
1630 | prfield (i, line2); | |
1631 | } | |
1632 | for (i = join_field_2 + 1; i < line2->nfields; ++i) | |
1633 | { | |
1634 | - putchar (output_separator); | |
1635 | + PUT_TAB_CHAR; | |
1636 | prfield (i, line2); | |
1637 | } | |
1638 | putchar ('\n'); | |
1555d43c | 1639 | @@ -1043,21 +1273,46 @@ main (int argc, char **argv) |
56ae3f82 SS |
1640 | |
1641 | case 't': | |
1642 | { | |
1643 | - unsigned char newtab = optarg[0]; | |
1644 | + char *newtab; | |
1645 | + size_t newtablen; | |
1646 | + newtab = xstrdup (optarg); | |
1647 | +#if HAVE_MBRTOWC | |
1648 | + if (MB_CUR_MAX > 1) | |
1649 | + { | |
1650 | + mbstate_t state; | |
1651 | + | |
1652 | + memset (&state, 0, sizeof (mbstate_t)); | |
1653 | + newtablen = mbrtowc (NULL, newtab, | |
1654 | + strnlen (newtab, MB_LEN_MAX), | |
1655 | + &state); | |
1656 | + if (newtablen == (size_t) 0 | |
1657 | + || newtablen == (size_t) -1 | |
1658 | + || newtablen == (size_t) -2) | |
1659 | + newtablen = 1; | |
1660 | + } | |
1661 | + else | |
1662 | +#endif | |
1663 | + newtablen = 1; | |
1664 | if (! newtab) | |
1665 | + { | |
1666 | newtab = '\n'; /* '' => process the whole line. */ | |
1667 | + } | |
1668 | else if (optarg[1]) | |
1669 | { | |
1670 | - if (STREQ (optarg, "\\0")) | |
1671 | - newtab = '\0'; | |
1672 | - else | |
1673 | - error (EXIT_FAILURE, 0, _("multi-character tab %s"), | |
1674 | - quote (optarg)); | |
1675 | + if (newtablen == 1 && newtab[1]) | |
1676 | + { | |
1677 | + if (STREQ (newtab, "\\0")) | |
1678 | + newtab[0] = '\0'; | |
1679 | + } | |
1680 | + } | |
1681 | + if (tab != NULL && strcmp (tab, newtab)) | |
1682 | + { | |
1683 | + free (newtab); | |
1684 | + error (EXIT_FAILURE, 0, _("incompatible tabs")); | |
1685 | } | |
1686 | - if (0 <= tab && tab != newtab) | |
1687 | - error (EXIT_FAILURE, 0, _("incompatible tabs")); | |
1688 | tab = newtab; | |
1689 | - } | |
1690 | + tablen = newtablen; | |
1691 | + } | |
1692 | break; | |
1693 | ||
1694 | case NOCHECK_ORDER_OPTION: | |
407c5be3 SS |
1695 | diff -urNp coreutils-8.9-orig/src/pr.c coreutils-8.9/src/pr.c |
1696 | --- coreutils-8.9-orig/src/pr.c 2011-01-01 22:19:23.000000000 +0100 | |
1697 | +++ coreutils-8.9/src/pr.c 2011-01-04 17:41:55.377138275 +0100 | |
56ae3f82 SS |
1698 | @@ -312,6 +312,32 @@ |
1699 | ||
1700 | #include <getopt.h> | |
1701 | #include <sys/types.h> | |
1702 | + | |
1703 | +/* Get MB_LEN_MAX. */ | |
1704 | +#include <limits.h> | |
1705 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC | |
1706 | + installation; work around this configuration error. */ | |
1707 | +#if !defined MB_LEN_MAX || MB_LEN_MAX == 1 | |
1708 | +# define MB_LEN_MAX 16 | |
1709 | +#endif | |
1710 | + | |
1711 | +/* Get MB_CUR_MAX. */ | |
1712 | +#include <stdlib.h> | |
1713 | + | |
1714 | +/* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */ | |
1715 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ | |
1716 | +#if HAVE_WCHAR_H | |
1717 | +# include <wchar.h> | |
1718 | +#endif | |
1719 | + | |
1720 | +/* Get iswprint(). -- for wcwidth(). */ | |
1721 | +#if HAVE_WCTYPE_H | |
1722 | +# include <wctype.h> | |
1723 | +#endif | |
1724 | +#if !defined iswprint && !HAVE_ISWPRINT | |
1725 | +# define iswprint(wc) 1 | |
1726 | +#endif | |
1727 | + | |
1728 | #include "system.h" | |
1729 | #include "error.h" | |
1555d43c SS |
1730 | #include "fadvise.h" |
1731 | @@ -323,6 +349,18 @@ | |
56ae3f82 SS |
1732 | #include "strftime.h" |
1733 | #include "xstrtol.h" | |
1734 | ||
1735 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ | |
1736 | +#if HAVE_MBRTOWC && defined mbstate_t | |
1737 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) | |
1738 | +#endif | |
1739 | + | |
1740 | +#ifndef HAVE_DECL_WCWIDTH | |
1741 | +"this configure-time declaration test was not run" | |
1742 | +#endif | |
1743 | +#if !HAVE_DECL_WCWIDTH | |
1744 | +extern int wcwidth (); | |
1745 | +#endif | |
1746 | + | |
1747 | /* The official name of this program (e.g., no `g' prefix). */ | |
1748 | #define PROGRAM_NAME "pr" | |
1749 | ||
1555d43c | 1750 | @@ -415,7 +453,20 @@ struct COLUMN |
56ae3f82 SS |
1751 | |
1752 | typedef struct COLUMN COLUMN; | |
1753 | ||
1754 | -static int char_to_clump (char c); | |
1755 | +/* Funtion pointers to switch functions for single byte locale or for | |
1756 | + multibyte locale. If multibyte functions do not exist in your sysytem, | |
1757 | + these pointers always point the function for single byte locale. */ | |
1758 | +static void (*print_char) (char c); | |
1759 | +static int (*char_to_clump) (char c); | |
1760 | + | |
1761 | +/* Functions for single byte locale. */ | |
1762 | +static void print_char_single (char c); | |
1763 | +static int char_to_clump_single (char c); | |
1764 | + | |
1765 | +/* Functions for multibyte locale. */ | |
1766 | +static void print_char_multi (char c); | |
1767 | +static int char_to_clump_multi (char c); | |
1768 | + | |
1769 | static bool read_line (COLUMN *p); | |
1770 | static bool print_page (void); | |
1771 | static bool print_stored (COLUMN *p); | |
1555d43c | 1772 | @@ -425,6 +476,7 @@ static void print_header (void); |
56ae3f82 SS |
1773 | static void pad_across_to (int position); |
1774 | static void add_line_number (COLUMN *p); | |
1775 | static void getoptarg (char *arg, char switch_char, char *character, | |
1776 | + int *character_length, int *character_width, | |
1777 | int *number); | |
1778 | void usage (int status); | |
1779 | static void print_files (int number_of_files, char **av); | |
1555d43c | 1780 | @@ -439,7 +491,6 @@ static void store_char (char c); |
56ae3f82 SS |
1781 | static void pad_down (int lines); |
1782 | static void read_rest_of_line (COLUMN *p); | |
1783 | static void skip_read (COLUMN *p, int column_number); | |
1784 | -static void print_char (char c); | |
1785 | static void cleanup (void); | |
1786 | static void print_sep_string (void); | |
1787 | static void separator_string (const char *optarg_S); | |
1555d43c | 1788 | @@ -451,7 +502,7 @@ static COLUMN *column_vector; |
56ae3f82 SS |
1789 | we store the leftmost columns contiguously in buff. |
1790 | To print a line from buff, get the index of the first character | |
1791 | from line_vector[i], and print up to line_vector[i + 1]. */ | |
1792 | -static char *buff; | |
1793 | +static unsigned char *buff; | |
1794 | ||
1795 | /* Index of the position in buff where the next character | |
1796 | will be stored. */ | |
1555d43c | 1797 | @@ -555,7 +606,7 @@ static int chars_per_column; |
56ae3f82 SS |
1798 | static bool untabify_input = false; |
1799 | ||
1800 | /* (-e) The input tab character. */ | |
1801 | -static char input_tab_char = '\t'; | |
1802 | +static char input_tab_char[MB_LEN_MAX] = "\t"; | |
1803 | ||
1804 | /* (-e) Tabstops are at chars_per_tab, 2*chars_per_tab, 3*chars_per_tab, ... | |
1805 | where the leftmost column is 1. */ | |
1555d43c | 1806 | @@ -565,7 +616,10 @@ static int chars_per_input_tab = 8; |
56ae3f82 SS |
1807 | static bool tabify_output = false; |
1808 | ||
1809 | /* (-i) The output tab character. */ | |
1810 | -static char output_tab_char = '\t'; | |
1811 | +static char output_tab_char[MB_LEN_MAX] = "\t"; | |
1812 | + | |
1813 | +/* (-i) The byte length of output tab character. */ | |
1814 | +static int output_tab_char_length = 1; | |
1815 | ||
1816 | /* (-i) The width of the output tab. */ | |
1817 | static int chars_per_output_tab = 8; | |
1555d43c | 1818 | @@ -639,7 +693,13 @@ static int power_10; |
56ae3f82 SS |
1819 | static bool numbered_lines = false; |
1820 | ||
1821 | /* (-n) Character which follows each line number. */ | |
1822 | -static char number_separator = '\t'; | |
1823 | +static char number_separator[MB_LEN_MAX] = "\t"; | |
1824 | + | |
1825 | +/* (-n) The byte length of the character which follows each line number. */ | |
1826 | +static int number_separator_length = 1; | |
1827 | + | |
1828 | +/* (-n) The character width of the character which follows each line number. */ | |
1829 | +static int number_separator_width = 0; | |
1830 | ||
1831 | /* (-n) line counting starts with 1st line of input file (not with 1st | |
1832 | line of 1st page printed). */ | |
1555d43c | 1833 | @@ -692,6 +752,7 @@ static bool use_col_separator = false; |
56ae3f82 SS |
1834 | -a|COLUMN|-m is a `space' and with the -J option a `tab'. */ |
1835 | static char *col_sep_string = (char *) ""; | |
1836 | static int col_sep_length = 0; | |
1837 | +static int col_sep_width = 0; | |
1838 | static char *column_separator = (char *) " "; | |
1839 | static char *line_separator = (char *) "\t"; | |
1840 | ||
1555d43c | 1841 | @@ -848,6 +909,13 @@ separator_string (const char *optarg_S) |
56ae3f82 SS |
1842 | col_sep_length = (int) strlen (optarg_S); |
1843 | col_sep_string = xmalloc (col_sep_length + 1); | |
1844 | strcpy (col_sep_string, optarg_S); | |
1845 | + | |
1846 | +#if HAVE_MBRTOWC | |
1847 | + if (MB_CUR_MAX > 1) | |
1848 | + col_sep_width = mbswidth (col_sep_string, 0); | |
1849 | + else | |
1850 | +#endif | |
1851 | + col_sep_width = col_sep_length; | |
1852 | } | |
1853 | ||
1854 | int | |
1555d43c | 1855 | @@ -872,6 +940,21 @@ main (int argc, char **argv) |
56ae3f82 SS |
1856 | |
1857 | atexit (close_stdout); | |
1858 | ||
1859 | +/* Define which functions are used, the ones for single byte locale or the ones | |
1860 | + for multibyte locale. */ | |
1861 | +#if HAVE_MBRTOWC | |
1862 | + if (MB_CUR_MAX > 1) | |
1863 | + { | |
1864 | + print_char = print_char_multi; | |
1865 | + char_to_clump = char_to_clump_multi; | |
1866 | + } | |
1867 | + else | |
1868 | +#endif | |
1869 | + { | |
1870 | + print_char = print_char_single; | |
1871 | + char_to_clump = char_to_clump_single; | |
1872 | + } | |
1873 | + | |
1874 | n_files = 0; | |
1875 | file_names = (argc > 1 | |
1876 | ? xmalloc ((argc - 1) * sizeof (char *)) | |
1555d43c | 1877 | @@ -948,8 +1031,12 @@ main (int argc, char **argv) |
56ae3f82 SS |
1878 | break; |
1879 | case 'e': | |
1880 | if (optarg) | |
1881 | - getoptarg (optarg, 'e', &input_tab_char, | |
1882 | - &chars_per_input_tab); | |
1883 | + { | |
1884 | + int dummy_length, dummy_width; | |
1885 | + | |
1886 | + getoptarg (optarg, 'e', input_tab_char, &dummy_length, | |
1887 | + &dummy_width, &chars_per_input_tab); | |
1888 | + } | |
1889 | /* Could check tab width > 0. */ | |
1890 | untabify_input = true; | |
1891 | break; | |
1555d43c | 1892 | @@ -962,8 +1049,12 @@ main (int argc, char **argv) |
56ae3f82 SS |
1893 | break; |
1894 | case 'i': | |
1895 | if (optarg) | |
1896 | - getoptarg (optarg, 'i', &output_tab_char, | |
1897 | - &chars_per_output_tab); | |
1898 | + { | |
1899 | + int dummy_width; | |
1900 | + | |
1901 | + getoptarg (optarg, 'i', output_tab_char, &output_tab_char_length, | |
1902 | + &dummy_width, &chars_per_output_tab); | |
1903 | + } | |
1904 | /* Could check tab width > 0. */ | |
1905 | tabify_output = true; | |
1906 | break; | |
1555d43c | 1907 | @@ -990,8 +1081,8 @@ main (int argc, char **argv) |
56ae3f82 SS |
1908 | case 'n': |
1909 | numbered_lines = true; | |
1910 | if (optarg) | |
1911 | - getoptarg (optarg, 'n', &number_separator, | |
1912 | - &chars_per_number); | |
1913 | + getoptarg (optarg, 'n', number_separator, &number_separator_length, | |
1914 | + &number_separator_width, &chars_per_number); | |
1915 | break; | |
1916 | case 'N': | |
1917 | skip_count = false; | |
1555d43c | 1918 | @@ -1030,7 +1121,7 @@ main (int argc, char **argv) |
56ae3f82 SS |
1919 | old_s = false; |
1920 | /* Reset an additional input of -s, -S dominates -s */ | |
1921 | col_sep_string = bad_cast (""); | |
1922 | - col_sep_length = 0; | |
1923 | + col_sep_length = col_sep_width = 0; | |
1924 | use_col_separator = true; | |
1925 | if (optarg) | |
1926 | separator_string (optarg); | |
1555d43c | 1927 | @@ -1187,10 +1278,45 @@ main (int argc, char **argv) |
56ae3f82 SS |
1928 | a number. */ |
1929 | ||
1930 | static void | |
1931 | -getoptarg (char *arg, char switch_char, char *character, int *number) | |
1932 | +getoptarg (char *arg, char switch_char, char *character, int *character_length, | |
1933 | + int *character_width, int *number) | |
1934 | { | |
1935 | if (!ISDIGIT (*arg)) | |
1936 | - *character = *arg++; | |
1937 | + { | |
1938 | +#ifdef HAVE_MBRTOWC | |
1939 | + if (MB_CUR_MAX > 1) /* for multibyte locale. */ | |
1940 | + { | |
1941 | + wchar_t wc; | |
1942 | + size_t mblength; | |
1943 | + int width; | |
1944 | + mbstate_t state = {'\0'}; | |
1945 | + | |
1946 | + mblength = mbrtowc (&wc, arg, strnlen(arg, MB_LEN_MAX), &state); | |
1947 | + | |
1948 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) | |
1949 | + { | |
1950 | + *character_length = 1; | |
1951 | + *character_width = 1; | |
1952 | + } | |
1953 | + else | |
1954 | + { | |
1955 | + *character_length = (mblength < 1) ? 1 : mblength; | |
1956 | + width = wcwidth (wc); | |
1957 | + *character_width = (width < 0) ? 0 : width; | |
1958 | + } | |
1959 | + | |
1960 | + strncpy (character, arg, *character_length); | |
1961 | + arg += *character_length; | |
1962 | + } | |
1963 | + else /* for single byte locale. */ | |
1964 | +#endif | |
1965 | + { | |
1966 | + *character = *arg++; | |
1967 | + *character_length = 1; | |
1968 | + *character_width = 1; | |
1969 | + } | |
1970 | + } | |
1971 | + | |
1972 | if (*arg) | |
1973 | { | |
1974 | long int tmp_long; | |
1555d43c | 1975 | @@ -1249,7 +1375,7 @@ init_parameters (int number_of_files) |
56ae3f82 SS |
1976 | else |
1977 | col_sep_string = column_separator; | |
1978 | ||
1979 | - col_sep_length = 1; | |
1980 | + col_sep_length = col_sep_width = 1; | |
1981 | use_col_separator = true; | |
1982 | } | |
1983 | /* It's rather pointless to define a TAB separator with column | |
1555d43c | 1984 | @@ -1280,11 +1406,11 @@ init_parameters (int number_of_files) |
56ae3f82 SS |
1985 | TAB_WIDTH (chars_per_input_tab, chars_per_number); */ |
1986 | ||
1987 | /* Estimate chars_per_text without any margin and keep it constant. */ | |
1988 | - if (number_separator == '\t') | |
1989 | + if (number_separator[0] == '\t') | |
1990 | number_width = chars_per_number + | |
1991 | TAB_WIDTH (chars_per_default_tab, chars_per_number); | |
1992 | else | |
1993 | - number_width = chars_per_number + 1; | |
1994 | + number_width = chars_per_number + number_separator_width; | |
1995 | ||
1996 | /* The number is part of the column width unless we are | |
1997 | printing files in parallel. */ | |
1555d43c | 1998 | @@ -1299,7 +1425,7 @@ init_parameters (int number_of_files) |
56ae3f82 SS |
1999 | } |
2000 | ||
2001 | chars_per_column = (chars_per_line - chars_used_by_number - | |
2002 | - (columns - 1) * col_sep_length) / columns; | |
2003 | + (columns - 1) * col_sep_width) / columns; | |
2004 | ||
2005 | if (chars_per_column < 1) | |
2006 | error (EXIT_FAILURE, 0, _("page width too narrow")); | |
1555d43c | 2007 | @@ -1424,7 +1550,7 @@ init_funcs (void) |
56ae3f82 SS |
2008 | |
2009 | /* Enlarge p->start_position of first column to use the same form of | |
2010 | padding_not_printed with all columns. */ | |
2011 | - h = h + col_sep_length; | |
2012 | + h = h + col_sep_width; | |
2013 | ||
2014 | /* This loop takes care of all but the rightmost column. */ | |
2015 | ||
1555d43c | 2016 | @@ -1458,7 +1584,7 @@ init_funcs (void) |
56ae3f82 SS |
2017 | } |
2018 | else | |
2019 | { | |
2020 | - h = h_next + col_sep_length; | |
2021 | + h = h_next + col_sep_width; | |
2022 | h_next = h + chars_per_column; | |
2023 | } | |
2024 | } | |
1555d43c | 2025 | @@ -1749,9 +1875,9 @@ static void |
56ae3f82 SS |
2026 | align_column (COLUMN *p) |
2027 | { | |
2028 | padding_not_printed = p->start_position; | |
2029 | - if (padding_not_printed - col_sep_length > 0) | |
2030 | + if (padding_not_printed - col_sep_width > 0) | |
2031 | { | |
2032 | - pad_across_to (padding_not_printed - col_sep_length); | |
2033 | + pad_across_to (padding_not_printed - col_sep_width); | |
2034 | padding_not_printed = ANYWHERE; | |
2035 | } | |
2036 | ||
1555d43c | 2037 | @@ -2022,13 +2148,13 @@ store_char (char c) |
56ae3f82 SS |
2038 | /* May be too generous. */ |
2039 | buff = X2REALLOC (buff, &buff_allocated); | |
2040 | } | |
2041 | - buff[buff_current++] = c; | |
2042 | + buff[buff_current++] = (unsigned char) c; | |
2043 | } | |
2044 | ||
2045 | static void | |
2046 | add_line_number (COLUMN *p) | |
2047 | { | |
2048 | - int i; | |
2049 | + int i, j; | |
2050 | char *s; | |
2051 | int left_cut; | |
2052 | ||
1555d43c | 2053 | @@ -2051,22 +2177,24 @@ add_line_number (COLUMN *p) |
56ae3f82 SS |
2054 | /* Tabification is assumed for multiple columns, also for n-separators, |
2055 | but `default n-separator = TAB' hasn't been given priority over | |
2056 | equal column_width also specified by POSIX. */ | |
2057 | - if (number_separator == '\t') | |
2058 | + if (number_separator[0] == '\t') | |
2059 | { | |
2060 | i = number_width - chars_per_number; | |
2061 | while (i-- > 0) | |
2062 | (p->char_func) (' '); | |
2063 | } | |
2064 | else | |
2065 | - (p->char_func) (number_separator); | |
2066 | + for (j = 0; j < number_separator_length; j++) | |
2067 | + (p->char_func) (number_separator[j]); | |
2068 | } | |
2069 | else | |
2070 | /* To comply with POSIX, we avoid any expansion of default TAB | |
2071 | separator with a single column output. No column_width requirement | |
2072 | has to be considered. */ | |
2073 | { | |
2074 | - (p->char_func) (number_separator); | |
2075 | - if (number_separator == '\t') | |
2076 | + for (j = 0; j < number_separator_length; j++) | |
2077 | + (p->char_func) (number_separator[j]); | |
2078 | + if (number_separator[0] == '\t') | |
2079 | output_position = POS_AFTER_TAB (chars_per_output_tab, | |
2080 | output_position); | |
2081 | } | |
1555d43c | 2082 | @@ -2227,7 +2355,7 @@ print_white_space (void) |
56ae3f82 SS |
2083 | while (goal - h_old > 1 |
2084 | && (h_new = POS_AFTER_TAB (chars_per_output_tab, h_old)) <= goal) | |
2085 | { | |
2086 | - putchar (output_tab_char); | |
2087 | + fwrite (output_tab_char, sizeof(char), output_tab_char_length, stdout); | |
2088 | h_old = h_new; | |
2089 | } | |
2090 | while (++h_old <= goal) | |
1555d43c | 2091 | @@ -2247,6 +2375,7 @@ print_sep_string (void) |
56ae3f82 SS |
2092 | { |
2093 | char *s; | |
2094 | int l = col_sep_length; | |
2095 | + int not_space_flag; | |
2096 | ||
2097 | s = col_sep_string; | |
2098 | ||
1555d43c | 2099 | @@ -2260,6 +2389,7 @@ print_sep_string (void) |
56ae3f82 SS |
2100 | { |
2101 | for (; separators_not_printed > 0; --separators_not_printed) | |
2102 | { | |
2103 | + not_space_flag = 0; | |
2104 | while (l-- > 0) | |
2105 | { | |
2106 | /* 3 types of sep_strings: spaces only, spaces and chars, | |
1555d43c | 2107 | @@ -2273,12 +2403,15 @@ print_sep_string (void) |
56ae3f82 SS |
2108 | } |
2109 | else | |
2110 | { | |
2111 | + not_space_flag = 1; | |
2112 | if (spaces_not_printed > 0) | |
2113 | print_white_space (); | |
2114 | putchar (*s++); | |
2115 | - ++output_position; | |
2116 | } | |
2117 | } | |
2118 | + if (not_space_flag) | |
2119 | + output_position += col_sep_width; | |
2120 | + | |
2121 | /* sep_string ends with some spaces */ | |
2122 | if (spaces_not_printed > 0) | |
2123 | print_white_space (); | |
1555d43c | 2124 | @@ -2306,7 +2439,7 @@ print_clump (COLUMN *p, int n, char *clu |
56ae3f82 SS |
2125 | required number of tabs and spaces. */ |
2126 | ||
2127 | static void | |
2128 | -print_char (char c) | |
2129 | +print_char_single (char c) | |
2130 | { | |
2131 | if (tabify_output) | |
2132 | { | |
1555d43c | 2133 | @@ -2330,6 +2463,74 @@ print_char (char c) |
56ae3f82 SS |
2134 | putchar (c); |
2135 | } | |
2136 | ||
2137 | +#ifdef HAVE_MBRTOWC | |
2138 | +static void | |
2139 | +print_char_multi (char c) | |
2140 | +{ | |
2141 | + static size_t mbc_pos = 0; | |
2142 | + static char mbc[MB_LEN_MAX] = {'\0'}; | |
2143 | + static mbstate_t state = {'\0'}; | |
2144 | + mbstate_t state_bak; | |
2145 | + wchar_t wc; | |
2146 | + size_t mblength; | |
2147 | + int width; | |
2148 | + | |
2149 | + if (tabify_output) | |
2150 | + { | |
2151 | + state_bak = state; | |
2152 | + mbc[mbc_pos++] = c; | |
2153 | + mblength = mbrtowc (&wc, mbc, mbc_pos, &state); | |
2154 | + | |
2155 | + while (mbc_pos > 0) | |
2156 | + { | |
2157 | + switch (mblength) | |
2158 | + { | |
2159 | + case (size_t)-2: | |
2160 | + state = state_bak; | |
2161 | + return; | |
2162 | + | |
2163 | + case (size_t)-1: | |
2164 | + state = state_bak; | |
2165 | + ++output_position; | |
2166 | + putchar (mbc[0]); | |
2167 | + memmove (mbc, mbc + 1, MB_CUR_MAX - 1); | |
2168 | + --mbc_pos; | |
2169 | + break; | |
2170 | + | |
2171 | + case 0: | |
2172 | + mblength = 1; | |
2173 | + | |
2174 | + default: | |
2175 | + if (wc == L' ') | |
2176 | + { | |
2177 | + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); | |
2178 | + --mbc_pos; | |
2179 | + ++spaces_not_printed; | |
2180 | + return; | |
2181 | + } | |
2182 | + else if (spaces_not_printed > 0) | |
2183 | + print_white_space (); | |
2184 | + | |
2185 | + /* Nonprintables are assumed to have width 0, except L'\b'. */ | |
2186 | + if ((width = wcwidth (wc)) < 1) | |
2187 | + { | |
2188 | + if (wc == L'\b') | |
2189 | + --output_position; | |
2190 | + } | |
2191 | + else | |
2192 | + output_position += width; | |
2193 | + | |
2194 | + fwrite (mbc, sizeof(char), mblength, stdout); | |
2195 | + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); | |
2196 | + mbc_pos -= mblength; | |
2197 | + } | |
2198 | + } | |
2199 | + return; | |
2200 | + } | |
2201 | + putchar (c); | |
2202 | +} | |
2203 | +#endif | |
2204 | + | |
2205 | /* Skip to page PAGE before printing. | |
2206 | PAGE may be larger than total number of pages. */ | |
2207 | ||
1555d43c | 2208 | @@ -2509,9 +2710,9 @@ read_line (COLUMN *p) |
56ae3f82 SS |
2209 | align_empty_cols = false; |
2210 | } | |
2211 | ||
2212 | - if (padding_not_printed - col_sep_length > 0) | |
2213 | + if (padding_not_printed - col_sep_width > 0) | |
2214 | { | |
2215 | - pad_across_to (padding_not_printed - col_sep_length); | |
2216 | + pad_across_to (padding_not_printed - col_sep_width); | |
2217 | padding_not_printed = ANYWHERE; | |
2218 | } | |
2219 | ||
1555d43c | 2220 | @@ -2612,9 +2813,9 @@ print_stored (COLUMN *p) |
56ae3f82 SS |
2221 | } |
2222 | } | |
2223 | ||
2224 | - if (padding_not_printed - col_sep_length > 0) | |
2225 | + if (padding_not_printed - col_sep_width > 0) | |
2226 | { | |
2227 | - pad_across_to (padding_not_printed - col_sep_length); | |
2228 | + pad_across_to (padding_not_printed - col_sep_width); | |
2229 | padding_not_printed = ANYWHERE; | |
2230 | } | |
2231 | ||
1555d43c | 2232 | @@ -2627,8 +2828,8 @@ print_stored (COLUMN *p) |
56ae3f82 SS |
2233 | if (spaces_not_printed == 0) |
2234 | { | |
2235 | output_position = p->start_position + end_vector[line]; | |
2236 | - if (p->start_position - col_sep_length == chars_per_margin) | |
2237 | - output_position -= col_sep_length; | |
2238 | + if (p->start_position - col_sep_width == chars_per_margin) | |
2239 | + output_position -= col_sep_width; | |
2240 | } | |
2241 | ||
2242 | return true; | |
1555d43c | 2243 | @@ -2647,7 +2848,7 @@ print_stored (COLUMN *p) |
56ae3f82 SS |
2244 | number of characters is 1.) */ |
2245 | ||
2246 | static int | |
2247 | -char_to_clump (char c) | |
2248 | +char_to_clump_single (char c) | |
2249 | { | |
2250 | unsigned char uc = c; | |
2251 | char *s = clump_buff; | |
1555d43c | 2252 | @@ -2657,10 +2858,10 @@ char_to_clump (char c) |
56ae3f82 SS |
2253 | int chars; |
2254 | int chars_per_c = 8; | |
2255 | ||
2256 | - if (c == input_tab_char) | |
2257 | + if (c == input_tab_char[0]) | |
2258 | chars_per_c = chars_per_input_tab; | |
2259 | ||
2260 | - if (c == input_tab_char || c == '\t') | |
2261 | + if (c == input_tab_char[0] || c == '\t') | |
2262 | { | |
2263 | width = TAB_WIDTH (chars_per_c, input_position); | |
2264 | ||
1555d43c | 2265 | @@ -2741,6 +2942,154 @@ char_to_clump (char c) |
56ae3f82 SS |
2266 | return chars; |
2267 | } | |
2268 | ||
2269 | +#ifdef HAVE_MBRTOWC | |
2270 | +static int | |
2271 | +char_to_clump_multi (char c) | |
2272 | +{ | |
2273 | + static size_t mbc_pos = 0; | |
2274 | + static char mbc[MB_LEN_MAX] = {'\0'}; | |
2275 | + static mbstate_t state = {'\0'}; | |
2276 | + mbstate_t state_bak; | |
2277 | + wchar_t wc; | |
2278 | + size_t mblength; | |
2279 | + int wc_width; | |
2280 | + register char *s = clump_buff; | |
2281 | + register int i, j; | |
2282 | + char esc_buff[4]; | |
2283 | + int width; | |
2284 | + int chars; | |
2285 | + int chars_per_c = 8; | |
2286 | + | |
2287 | + state_bak = state; | |
2288 | + mbc[mbc_pos++] = c; | |
2289 | + mblength = mbrtowc (&wc, mbc, mbc_pos, &state); | |
2290 | + | |
2291 | + width = 0; | |
2292 | + chars = 0; | |
2293 | + while (mbc_pos > 0) | |
2294 | + { | |
2295 | + switch (mblength) | |
2296 | + { | |
2297 | + case (size_t)-2: | |
2298 | + state = state_bak; | |
2299 | + return 0; | |
2300 | + | |
2301 | + case (size_t)-1: | |
2302 | + state = state_bak; | |
2303 | + mblength = 1; | |
2304 | + | |
2305 | + if (use_esc_sequence || use_cntrl_prefix) | |
2306 | + { | |
2307 | + width = +4; | |
2308 | + chars = +4; | |
2309 | + *s++ = '\\'; | |
2310 | + sprintf (esc_buff, "%03o", mbc[0]); | |
2311 | + for (i = 0; i <= 2; ++i) | |
2312 | + *s++ = (int) esc_buff[i]; | |
2313 | + } | |
2314 | + else | |
2315 | + { | |
2316 | + width += 1; | |
2317 | + chars += 1; | |
2318 | + *s++ = mbc[0]; | |
2319 | + } | |
2320 | + break; | |
2321 | + | |
2322 | + case 0: | |
2323 | + mblength = 1; | |
2324 | + /* Fall through */ | |
2325 | + | |
2326 | + default: | |
2327 | + if (memcmp (mbc, input_tab_char, mblength) == 0) | |
2328 | + chars_per_c = chars_per_input_tab; | |
2329 | + | |
2330 | + if (memcmp (mbc, input_tab_char, mblength) == 0 || c == '\t') | |
2331 | + { | |
2332 | + int width_inc; | |
2333 | + | |
2334 | + width_inc = TAB_WIDTH (chars_per_c, input_position); | |
2335 | + width += width_inc; | |
2336 | + | |
2337 | + if (untabify_input) | |
2338 | + { | |
2339 | + for (i = width_inc; i; --i) | |
2340 | + *s++ = ' '; | |
2341 | + chars += width_inc; | |
2342 | + } | |
2343 | + else | |
2344 | + { | |
2345 | + for (i = 0; i < mblength; i++) | |
2346 | + *s++ = mbc[i]; | |
2347 | + chars += mblength; | |
2348 | + } | |
2349 | + } | |
2350 | + else if ((wc_width = wcwidth (wc)) < 1) | |
2351 | + { | |
2352 | + if (use_esc_sequence) | |
2353 | + { | |
2354 | + for (i = 0; i < mblength; i++) | |
2355 | + { | |
2356 | + width += 4; | |
2357 | + chars += 4; | |
2358 | + *s++ = '\\'; | |
2359 | + sprintf (esc_buff, "%03o", c); | |
2360 | + for (j = 0; j <= 2; ++j) | |
2361 | + *s++ = (int) esc_buff[j]; | |
2362 | + } | |
2363 | + } | |
2364 | + else if (use_cntrl_prefix) | |
2365 | + { | |
2366 | + if (wc < 0200) | |
2367 | + { | |
2368 | + width += 2; | |
2369 | + chars += 2; | |
2370 | + *s++ = '^'; | |
2371 | + *s++ = wc ^ 0100; | |
2372 | + } | |
2373 | + else | |
2374 | + { | |
2375 | + for (i = 0; i < mblength; i++) | |
2376 | + { | |
2377 | + width += 4; | |
2378 | + chars += 4; | |
2379 | + *s++ = '\\'; | |
2380 | + sprintf (esc_buff, "%03o", c); | |
2381 | + for (j = 0; j <= 2; ++j) | |
2382 | + *s++ = (int) esc_buff[j]; | |
2383 | + } | |
2384 | + } | |
2385 | + } | |
2386 | + else if (wc == L'\b') | |
2387 | + { | |
2388 | + width += -1; | |
2389 | + chars += 1; | |
2390 | + *s++ = c; | |
2391 | + } | |
2392 | + else | |
2393 | + { | |
2394 | + width += 0; | |
2395 | + chars += mblength; | |
2396 | + for (i = 0; i < mblength; i++) | |
2397 | + *s++ = mbc[i]; | |
2398 | + } | |
2399 | + } | |
2400 | + else | |
2401 | + { | |
2402 | + width += wc_width; | |
2403 | + chars += mblength; | |
2404 | + for (i = 0; i < mblength; i++) | |
2405 | + *s++ = mbc[i]; | |
2406 | + } | |
2407 | + } | |
2408 | + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); | |
2409 | + mbc_pos -= mblength; | |
2410 | + } | |
2411 | + | |
2412 | + input_position += width; | |
2413 | + return chars; | |
2414 | +} | |
2415 | +#endif | |
2416 | + | |
2417 | /* We've just printed some files and need to clean up things before | |
2418 | looking for more options and printing the next batch of files. | |
2419 | ||
407c5be3 SS |
2420 | diff -urNp coreutils-8.9-orig/src/sort.c coreutils-8.9/src/sort.c |
2421 | --- coreutils-8.9-orig/src/sort.c 2011-01-01 22:19:23.000000000 +0100 | |
2422 | +++ coreutils-8.9/src/sort.c 2011-01-04 17:41:55.384888730 +0100 | |
1555d43c | 2423 | @@ -22,11 +22,20 @@ |
56ae3f82 SS |
2424 | |
2425 | #include <config.h> | |
2426 | ||
2427 | +#include <assert.h> | |
2428 | #include <getopt.h> | |
1555d43c | 2429 | #include <pthread.h> |
56ae3f82 SS |
2430 | #include <sys/types.h> |
2431 | #include <sys/wait.h> | |
2432 | #include <signal.h> | |
2433 | +#if HAVE_WCHAR_H | |
2434 | +# include <wchar.h> | |
2435 | +#endif | |
2436 | +/* Get isw* functions. */ | |
2437 | +#if HAVE_WCTYPE_H | |
2438 | +# include <wctype.h> | |
2439 | +#endif | |
2440 | + | |
2441 | #include "system.h" | |
2442 | #include "argmatch.h" | |
2443 | #include "error.h" | |
407c5be3 | 2444 | @@ -163,12 +172,34 @@ static int thousands_sep; |
56ae3f82 | 2445 | |
56ae3f82 SS |
2446 | /* Nonzero if the corresponding locales are hard. */ |
2447 | static bool hard_LC_COLLATE; | |
2448 | -#if HAVE_NL_LANGINFO | |
2449 | +#if HAVE_LANGINFO_CODESET | |
2450 | static bool hard_LC_TIME; | |
2451 | #endif | |
2452 | ||
2453 | #define NONZERO(x) ((x) != 0) | |
2454 | ||
2455 | +/* get a multibyte character's byte length. */ | |
2456 | +#define GET_BYTELEN_OF_CHAR(LIM, PTR, MBLENGTH, STATE) \ | |
2457 | + do \ | |
2458 | + { \ | |
2459 | + wchar_t wc; \ | |
2460 | + mbstate_t state_bak; \ | |
2461 | + \ | |
2462 | + state_bak = STATE; \ | |
2463 | + mblength = mbrtowc (&wc, PTR, LIM - PTR, &STATE); \ | |
2464 | + \ | |
2465 | + switch (MBLENGTH) \ | |
2466 | + { \ | |
2467 | + case (size_t)-1: \ | |
2468 | + case (size_t)-2: \ | |
2469 | + STATE = state_bak; \ | |
2470 | + /* Fall through. */ \ | |
2471 | + case 0: \ | |
2472 | + MBLENGTH = 1; \ | |
2473 | + } \ | |
2474 | + } \ | |
2475 | + while (0) | |
2476 | + | |
2477 | /* The kind of blanks for '-b' to skip in various options. */ | |
2478 | enum blanktype { bl_start, bl_end, bl_both }; | |
2479 | ||
407c5be3 | 2480 | @@ -335,13 +366,11 @@ static bool reverse; |
56ae3f82 SS |
2481 | they were read if all keys compare equal. */ |
2482 | static bool stable; | |
2483 | ||
2484 | -/* If TAB has this value, blanks separate fields. */ | |
2485 | -enum { TAB_DEFAULT = CHAR_MAX + 1 }; | |
2486 | - | |
2487 | -/* Tab character separating fields. If TAB_DEFAULT, then fields are | |
2488 | +/* Tab character separating fields. If tab_length is 0, then fields are | |
2489 | separated by the empty string between a non-blank character and a blank | |
2490 | character. */ | |
2491 | -static int tab = TAB_DEFAULT; | |
2492 | +static char tab[MB_LEN_MAX + 1]; | |
2493 | +static size_t tab_length = 0; | |
2494 | ||
2495 | /* Flag to remove consecutive duplicate lines from the output. | |
2496 | Only the last of a sequence of equal lines will be output. */ | |
407c5be3 SS |
2497 | @@ -775,6 +804,46 @@ reap_all (void) |
2498 | reap (-1); | |
56ae3f82 SS |
2499 | } |
2500 | ||
2501 | +/* Function pointers. */ | |
2502 | +static void | |
2503 | +(*inittables) (void); | |
2504 | +static char * | |
2505 | +(*begfield) (const struct line*, const struct keyfield *); | |
2506 | +static char * | |
2507 | +(*limfield) (const struct line*, const struct keyfield *); | |
1555d43c SS |
2508 | +static void |
2509 | +(*skipblanks) (char **ptr, char *lim); | |
56ae3f82 | 2510 | +static int |
1555d43c | 2511 | +(*getmonth) (char const *, size_t, char **); |
56ae3f82 SS |
2512 | +static int |
2513 | +(*keycompare) (const struct line *, const struct line *); | |
2514 | +static int | |
2515 | +(*numcompare) (const char *, const char *); | |
2516 | + | |
2517 | +/* Test for white space multibyte character. | |
2518 | + Set LENGTH the byte length of investigated multibyte character. */ | |
2519 | +#if HAVE_MBRTOWC | |
2520 | +static int | |
2521 | +ismbblank (const char *str, size_t len, size_t *length) | |
2522 | +{ | |
2523 | + size_t mblength; | |
2524 | + wchar_t wc; | |
2525 | + mbstate_t state; | |
2526 | + | |
2527 | + memset (&state, '\0', sizeof(mbstate_t)); | |
2528 | + mblength = mbrtowc (&wc, str, len, &state); | |
2529 | + | |
2530 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) | |
2531 | + { | |
2532 | + *length = 1; | |
2533 | + return 0; | |
2534 | + } | |
2535 | + | |
2536 | + *length = (mblength < 1) ? 1 : mblength; | |
2537 | + return iswblank (wc); | |
2538 | +} | |
2539 | +#endif | |
2540 | + | |
2541 | /* Clean up any remaining temporary files. */ | |
2542 | ||
2543 | static void | |
407c5be3 | 2544 | @@ -1207,7 +1276,7 @@ zaptemp (char const *name) |
56ae3f82 SS |
2545 | free (node); |
2546 | } | |
2547 | ||
2548 | -#if HAVE_NL_LANGINFO | |
2549 | +#if HAVE_LANGINFO_CODESET | |
2550 | ||
2551 | static int | |
1555d43c | 2552 | struct_month_cmp (void const *m1, void const *m2) |
407c5be3 | 2553 | @@ -1222,7 +1291,7 @@ struct_month_cmp (void const *m1, void c |
56ae3f82 SS |
2554 | /* Initialize the character class tables. */ |
2555 | ||
2556 | static void | |
2557 | -inittables (void) | |
2558 | +inittables_uni (void) | |
2559 | { | |
2560 | size_t i; | |
2561 | ||
407c5be3 | 2562 | @@ -1234,7 +1303,7 @@ inittables (void) |
56ae3f82 SS |
2563 | fold_toupper[i] = toupper (i); |
2564 | } | |
2565 | ||
2566 | -#if HAVE_NL_LANGINFO | |
2567 | +#if HAVE_LANGINFO_CODESET | |
2568 | /* If we're not in the "C" locale, read different names for months. */ | |
2569 | if (hard_LC_TIME) | |
2570 | { | |
407c5be3 | 2571 | @@ -1316,6 +1385,84 @@ specify_nmerge (int oi, char c, char con |
56ae3f82 SS |
2572 | xstrtol_fatal (e, oi, c, long_options, s); |
2573 | } | |
2574 | ||
2575 | +#if HAVE_MBRTOWC | |
2576 | +static void | |
2577 | +inittables_mb (void) | |
2578 | +{ | |
2579 | + int i, j, k, l; | |
1555d43c | 2580 | + char *name, *s, *lc_time, *lc_ctype; |
56ae3f82 SS |
2581 | + size_t s_len, mblength; |
2582 | + char mbc[MB_LEN_MAX]; | |
2583 | + wchar_t wc, pwc; | |
2584 | + mbstate_t state_mb, state_wc; | |
2585 | + | |
1555d43c SS |
2586 | + lc_time = setlocale (LC_TIME, ""); |
2587 | + if (lc_time) | |
2588 | + lc_time = xstrdup (lc_time); | |
2589 | + | |
2590 | + lc_ctype = setlocale (LC_CTYPE, ""); | |
2591 | + if (lc_ctype) | |
2592 | + lc_ctype = xstrdup (lc_ctype); | |
2593 | + | |
2594 | + if (lc_time && lc_ctype) | |
2595 | + /* temporarily set LC_CTYPE to match LC_TIME, so that we can convert | |
2596 | + * the names of months to upper case */ | |
2597 | + setlocale (LC_CTYPE, lc_time); | |
2598 | + | |
56ae3f82 SS |
2599 | + for (i = 0; i < MONTHS_PER_YEAR; i++) |
2600 | + { | |
2601 | + s = (char *) nl_langinfo (ABMON_1 + i); | |
2602 | + s_len = strlen (s); | |
2603 | + monthtab[i].name = name = (char *) xmalloc (s_len + 1); | |
2604 | + monthtab[i].val = i + 1; | |
2605 | + | |
2606 | + memset (&state_mb, '\0', sizeof (mbstate_t)); | |
2607 | + memset (&state_wc, '\0', sizeof (mbstate_t)); | |
2608 | + | |
2609 | + for (j = 0; j < s_len;) | |
2610 | + { | |
2611 | + if (!ismbblank (s + j, s_len - j, &mblength)) | |
2612 | + break; | |
2613 | + j += mblength; | |
2614 | + } | |
2615 | + | |
2616 | + for (k = 0; j < s_len;) | |
2617 | + { | |
2618 | + mblength = mbrtowc (&wc, (s + j), (s_len - j), &state_mb); | |
2619 | + assert (mblength != (size_t)-1 && mblength != (size_t)-2); | |
2620 | + if (mblength == 0) | |
2621 | + break; | |
2622 | + | |
2623 | + pwc = towupper (wc); | |
2624 | + if (pwc == wc) | |
2625 | + { | |
2626 | + memcpy (mbc, s + j, mblength); | |
2627 | + j += mblength; | |
2628 | + } | |
2629 | + else | |
2630 | + { | |
2631 | + j += mblength; | |
2632 | + mblength = wcrtomb (mbc, pwc, &state_wc); | |
2633 | + assert (mblength != (size_t)0 && mblength != (size_t)-1); | |
2634 | + } | |
2635 | + | |
2636 | + for (l = 0; l < mblength; l++) | |
2637 | + name[k++] = mbc[l]; | |
2638 | + } | |
2639 | + name[k] = '\0'; | |
2640 | + } | |
2641 | + qsort ((void *) monthtab, MONTHS_PER_YEAR, | |
2642 | + sizeof (struct month), struct_month_cmp); | |
1555d43c SS |
2643 | + |
2644 | + if (lc_time && lc_ctype) | |
2645 | + /* restore the original locales */ | |
2646 | + setlocale (LC_CTYPE, lc_ctype); | |
2647 | + | |
2648 | + free (lc_ctype); | |
2649 | + free (lc_time); | |
56ae3f82 SS |
2650 | +} |
2651 | +#endif | |
2652 | + | |
2653 | /* Specify the amount of main memory to use when sorting. */ | |
2654 | static void | |
2655 | specify_sort_size (int oi, char c, char const *s) | |
407c5be3 | 2656 | @@ -1544,7 +1691,7 @@ buffer_linelim (struct buffer const *buf |
56ae3f82 SS |
2657 | by KEY in LINE. */ |
2658 | ||
2659 | static char * | |
1555d43c | 2660 | -begfield (struct line const *line, struct keyfield const *key) |
56ae3f82 SS |
2661 | +begfield_uni (const struct line *line, const struct keyfield *key) |
2662 | { | |
2663 | char *ptr = line->text, *lim = ptr + line->length - 1; | |
2664 | size_t sword = key->sword; | |
407c5be3 | 2665 | @@ -1553,10 +1700,10 @@ begfield (struct line const *line, struc |
56ae3f82 SS |
2666 | /* The leading field separator itself is included in a field when -t |
2667 | is absent. */ | |
2668 | ||
2669 | - if (tab != TAB_DEFAULT) | |
2670 | + if (tab_length) | |
2671 | while (ptr < lim && sword--) | |
2672 | { | |
2673 | - while (ptr < lim && *ptr != tab) | |
2674 | + while (ptr < lim && *ptr != tab[0]) | |
2675 | ++ptr; | |
2676 | if (ptr < lim) | |
2677 | ++ptr; | |
407c5be3 | 2678 | @@ -1582,11 +1729,70 @@ begfield (struct line const *line, struc |
56ae3f82 SS |
2679 | return ptr; |
2680 | } | |
2681 | ||
2682 | +#if HAVE_MBRTOWC | |
2683 | +static char * | |
2684 | +begfield_mb (const struct line *line, const struct keyfield *key) | |
2685 | +{ | |
2686 | + int i; | |
2687 | + char *ptr = line->text, *lim = ptr + line->length - 1; | |
2688 | + size_t sword = key->sword; | |
2689 | + size_t schar = key->schar; | |
2690 | + size_t mblength; | |
2691 | + mbstate_t state; | |
2692 | + | |
2693 | + memset (&state, '\0', sizeof(mbstate_t)); | |
2694 | + | |
2695 | + if (tab_length) | |
2696 | + while (ptr < lim && sword--) | |
2697 | + { | |
2698 | + while (ptr < lim && memcmp (ptr, tab, tab_length) != 0) | |
2699 | + { | |
2700 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); | |
2701 | + ptr += mblength; | |
2702 | + } | |
2703 | + if (ptr < lim) | |
2704 | + { | |
2705 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); | |
2706 | + ptr += mblength; | |
2707 | + } | |
2708 | + } | |
2709 | + else | |
2710 | + while (ptr < lim && sword--) | |
2711 | + { | |
2712 | + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) | |
2713 | + ptr += mblength; | |
2714 | + if (ptr < lim) | |
2715 | + { | |
2716 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); | |
2717 | + ptr += mblength; | |
2718 | + } | |
2719 | + while (ptr < lim && !ismbblank (ptr, lim - ptr, &mblength)) | |
2720 | + ptr += mblength; | |
2721 | + } | |
2722 | + | |
2723 | + if (key->skipsblanks) | |
2724 | + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) | |
2725 | + ptr += mblength; | |
2726 | + | |
2727 | + for (i = 0; i < schar; i++) | |
2728 | + { | |
2729 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); | |
2730 | + | |
2731 | + if (ptr + mblength > lim) | |
2732 | + break; | |
2733 | + else | |
2734 | + ptr += mblength; | |
2735 | + } | |
2736 | + | |
2737 | + return ptr; | |
2738 | +} | |
2739 | +#endif | |
2740 | + | |
2741 | /* Return the limit of (a pointer to the first character after) the field | |
2742 | in LINE specified by KEY. */ | |
2743 | ||
2744 | static char * | |
1555d43c | 2745 | -limfield (struct line const *line, struct keyfield const *key) |
56ae3f82 SS |
2746 | +limfield_uni (const struct line *line, const struct keyfield *key) |
2747 | { | |
2748 | char *ptr = line->text, *lim = ptr + line->length - 1; | |
2749 | size_t eword = key->eword, echar = key->echar; | |
407c5be3 | 2750 | @@ -1601,10 +1807,10 @@ limfield (struct line const *line, struc |
56ae3f82 SS |
2751 | `beginning' is the first character following the delimiting TAB. |
2752 | Otherwise, leave PTR pointing at the first `blank' character after | |
2753 | the preceding field. */ | |
2754 | - if (tab != TAB_DEFAULT) | |
2755 | + if (tab_length) | |
2756 | while (ptr < lim && eword--) | |
2757 | { | |
2758 | - while (ptr < lim && *ptr != tab) | |
2759 | + while (ptr < lim && *ptr != tab[0]) | |
2760 | ++ptr; | |
2761 | if (ptr < lim && (eword || echar)) | |
2762 | ++ptr; | |
407c5be3 | 2763 | @@ -1650,10 +1856,10 @@ limfield (struct line const *line, struc |
56ae3f82 SS |
2764 | */ |
2765 | ||
2766 | /* Make LIM point to the end of (one byte past) the current field. */ | |
2767 | - if (tab != TAB_DEFAULT) | |
2768 | + if (tab_length) | |
2769 | { | |
2770 | char *newlim; | |
2771 | - newlim = memchr (ptr, tab, lim - ptr); | |
2772 | + newlim = memchr (ptr, tab[0], lim - ptr); | |
2773 | if (newlim) | |
2774 | lim = newlim; | |
2775 | } | |
407c5be3 | 2776 | @@ -1684,6 +1890,130 @@ limfield (struct line const *line, struc |
56ae3f82 SS |
2777 | return ptr; |
2778 | } | |
2779 | ||
2780 | +#if HAVE_MBRTOWC | |
2781 | +static char * | |
2782 | +limfield_mb (const struct line *line, const struct keyfield *key) | |
2783 | +{ | |
2784 | + char *ptr = line->text, *lim = ptr + line->length - 1; | |
2785 | + size_t eword = key->eword, echar = key->echar; | |
2786 | + int i; | |
2787 | + size_t mblength; | |
2788 | + mbstate_t state; | |
2789 | + | |
2790 | + if (echar == 0) | |
2791 | + eword++; /* skip all of end field. */ | |
2792 | + | |
2793 | + memset (&state, '\0', sizeof(mbstate_t)); | |
2794 | + | |
2795 | + if (tab_length) | |
2796 | + while (ptr < lim && eword--) | |
2797 | + { | |
2798 | + while (ptr < lim && memcmp (ptr, tab, tab_length) != 0) | |
2799 | + { | |
2800 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); | |
2801 | + ptr += mblength; | |
2802 | + } | |
2803 | + if (ptr < lim && (eword | echar)) | |
2804 | + { | |
2805 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); | |
2806 | + ptr += mblength; | |
2807 | + } | |
2808 | + } | |
2809 | + else | |
2810 | + while (ptr < lim && eword--) | |
2811 | + { | |
2812 | + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) | |
2813 | + ptr += mblength; | |
2814 | + if (ptr < lim) | |
2815 | + { | |
2816 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); | |
2817 | + ptr += mblength; | |
2818 | + } | |
2819 | + while (ptr < lim && !ismbblank (ptr, lim - ptr, &mblength)) | |
2820 | + ptr += mblength; | |
2821 | + } | |
2822 | + | |
2823 | + | |
2824 | +# ifdef POSIX_UNSPECIFIED | |
2825 | + /* Make LIM point to the end of (one byte past) the current field. */ | |
2826 | + if (tab_length) | |
2827 | + { | |
2828 | + char *newlim, *p; | |
2829 | + | |
2830 | + newlim = NULL; | |
2831 | + for (p = ptr; p < lim;) | |
2832 | + { | |
2833 | + if (memcmp (p, tab, tab_length) == 0) | |
2834 | + { | |
2835 | + newlim = p; | |
2836 | + break; | |
2837 | + } | |
2838 | + | |
2839 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); | |
2840 | + p += mblength; | |
2841 | + } | |
2842 | + } | |
2843 | + else | |
2844 | + { | |
2845 | + char *newlim; | |
2846 | + newlim = ptr; | |
2847 | + | |
2848 | + while (newlim < lim && ismbblank (newlim, lim - newlim, &mblength)) | |
2849 | + newlim += mblength; | |
2850 | + if (ptr < lim) | |
2851 | + { | |
2852 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); | |
2853 | + ptr += mblength; | |
2854 | + } | |
2855 | + while (newlim < lim && !ismbblank (newlim, lim - newlim, &mblength)) | |
2856 | + newlim += mblength; | |
2857 | + lim = newlim; | |
2858 | + } | |
2859 | +# endif | |
2860 | + | |
2861 | + if (echar != 0) | |
2862 | + { | |
2863 | + /* If we're skipping leading blanks, don't start counting characters | |
2864 | + * until after skipping past any leading blanks. */ | |
2865 | + if (key->skipsblanks) | |
2866 | + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) | |
2867 | + ptr += mblength; | |
2868 | + | |
2869 | + memset (&state, '\0', sizeof(mbstate_t)); | |
2870 | + | |
2871 | + /* Advance PTR by ECHAR (if possible), but no further than LIM. */ | |
2872 | + for (i = 0; i < echar; i++) | |
2873 | + { | |
2874 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); | |
2875 | + | |
2876 | + if (ptr + mblength > lim) | |
2877 | + break; | |
2878 | + else | |
2879 | + ptr += mblength; | |
2880 | + } | |
2881 | + } | |
2882 | + | |
2883 | + return ptr; | |
2884 | +} | |
2885 | +#endif | |
1555d43c SS |
2886 | + |
2887 | +static void | |
2888 | +skipblanks_uni (char **ptr, char *lim) | |
2889 | +{ | |
2890 | + while (*ptr < lim && blanks[to_uchar (**ptr)]) | |
2891 | + ++(*ptr); | |
2892 | +} | |
2893 | + | |
2894 | +#if HAVE_MBRTOWC | |
2895 | +static void | |
2896 | +skipblanks_mb (char **ptr, char *lim) | |
2897 | +{ | |
2898 | + size_t mblength; | |
2899 | + while (*ptr < lim && ismbblank (*ptr, lim - *ptr, &mblength)) | |
2900 | + (*ptr) += mblength; | |
2901 | +} | |
2902 | +#endif | |
56ae3f82 SS |
2903 | + |
2904 | /* Fill BUF reading from FP, moving buf->left bytes from the end | |
2905 | of buf->buf to the beginning first. If EOF is reached and the | |
2906 | file wasn't terminated by a newline, supply one. Set up BUF's line | |
407c5be3 | 2907 | @@ -1770,8 +2100,22 @@ fillbuf (struct buffer *buf, FILE *fp, c |
56ae3f82 SS |
2908 | else |
2909 | { | |
2910 | if (key->skipsblanks) | |
2911 | - while (blanks[to_uchar (*line_start)]) | |
2912 | - line_start++; | |
2913 | + { | |
2914 | +#if HAVE_MBRTOWC | |
2915 | + if (MB_CUR_MAX > 1) | |
2916 | + { | |
2917 | + size_t mblength; | |
56ae3f82 SS |
2918 | + while (line_start < line->keylim && |
2919 | + ismbblank (line_start, | |
2920 | + line->keylim - line_start, | |
2921 | + &mblength)) | |
2922 | + line_start += mblength; | |
2923 | + } | |
2924 | + else | |
2925 | +#endif | |
2926 | + while (blanks[to_uchar (*line_start)]) | |
2927 | + line_start++; | |
2928 | + } | |
2929 | line->keybeg = line_start; | |
2930 | } | |
2931 | } | |
407c5be3 | 2932 | @@ -1892,7 +2236,7 @@ human_numcompare (char const *a, char co |
56ae3f82 SS |
2933 | hideously fast. */ |
2934 | ||
2935 | static int | |
1555d43c | 2936 | -numcompare (char const *a, char const *b) |
56ae3f82 SS |
2937 | +numcompare_uni (const char *a, const char *b) |
2938 | { | |
2939 | while (blanks[to_uchar (*a)]) | |
2940 | a++; | |
407c5be3 | 2941 | @@ -1902,6 +2246,25 @@ numcompare (char const *a, char const *b |
1555d43c | 2942 | return strnumcmp (a, b, decimal_point, thousands_sep); |
56ae3f82 SS |
2943 | } |
2944 | ||
2945 | +#if HAVE_MBRTOWC | |
2946 | +static int | |
2947 | +numcompare_mb (const char *a, const char *b) | |
2948 | +{ | |
2949 | + size_t mblength, len; | |
2950 | + len = strlen (a); /* okay for UTF-8 */ | |
2951 | + while (*a && ismbblank (a, len > MB_CUR_MAX ? MB_CUR_MAX : len, &mblength)) | |
2952 | + { | |
2953 | + a += mblength; | |
2954 | + len -= mblength; | |
2955 | + } | |
2956 | + len = strlen (b); /* okay for UTF-8 */ | |
2957 | + while (*b && ismbblank (b, len > MB_CUR_MAX ? MB_CUR_MAX : len, &mblength)) | |
2958 | + b += mblength; | |
2959 | + | |
2960 | + return strnumcmp (a, b, decimal_point, thousands_sep); | |
2961 | +} | |
2962 | +#endif /* HAV_EMBRTOWC */ | |
2963 | + | |
2964 | static int | |
1555d43c | 2965 | general_numcompare (char const *sa, char const *sb) |
56ae3f82 | 2966 | { |
407c5be3 | 2967 | @@ -1934,7 +2297,7 @@ general_numcompare (char const *sa, char |
56ae3f82 SS |
2968 | Return 0 if the name in S is not recognized. */ |
2969 | ||
2970 | static int | |
1555d43c SS |
2971 | -getmonth (char const *month, char **ea) |
2972 | +getmonth_uni (char const *month, size_t len, char **ea) | |
56ae3f82 SS |
2973 | { |
2974 | size_t lo = 0; | |
2975 | size_t hi = MONTHS_PER_YEAR; | |
407c5be3 SS |
2976 | @@ -2209,13 +2572,12 @@ debug_key (struct line const *line, stru |
2977 | char saved = *lim; | |
2978 | *lim = '\0'; | |
1555d43c SS |
2979 | |
2980 | - while (blanks[to_uchar (*beg)]) | |
2981 | - beg++; | |
2982 | + skipblanks (&beg, lim); | |
2983 | ||
2984 | char *tighter_lim = beg; | |
2985 | ||
2986 | if (key->month) | |
2987 | - getmonth (beg, &tighter_lim); | |
2988 | + getmonth (beg, lim-beg, &tighter_lim); | |
2989 | else if (key->general_numeric) | |
2990 | ignore_value (strtold (beg, &tighter_lim)); | |
2991 | else if (key->numeric || key->human_numeric) | |
407c5be3 | 2992 | @@ -2359,7 +2721,7 @@ key_warnings (struct keyfield const *gke |
1555d43c SS |
2993 | bool maybe_space_aligned = !hard_LC_COLLATE && default_key_compare (key) |
2994 | && !(key->schar || key->echar); | |
2995 | bool line_offset = key->eword == 0 && key->echar != 0; /* -k1.x,1.y */ | |
2996 | - if (!gkey_only && tab == TAB_DEFAULT && !line_offset | |
2997 | + if (!gkey_only && !tab_length && !line_offset | |
2998 | && ((!key->skipsblanks && !(implicit_skip || maybe_space_aligned)) | |
2999 | || (!key->skipsblanks && key->schar) | |
3000 | || (!key->skipeblanks && key->echar))) | |
407c5be3 | 3001 | @@ -2417,11 +2779,83 @@ key_warnings (struct keyfield const *gke |
1555d43c | 3002 | error (0, 0, _("option `-r' only applies to last-resort comparison")); |
56ae3f82 SS |
3003 | } |
3004 | ||
3005 | +#if HAVE_MBRTOWC | |
3006 | +static int | |
1555d43c | 3007 | +getmonth_mb (const char *s, size_t len, char **ea) |
56ae3f82 SS |
3008 | +{ |
3009 | + char *month; | |
3010 | + register size_t i; | |
3011 | + register int lo = 0, hi = MONTHS_PER_YEAR, result; | |
3012 | + char *tmp; | |
3013 | + size_t wclength, mblength; | |
3014 | + const char **pp; | |
3015 | + const wchar_t **wpp; | |
3016 | + wchar_t *month_wcs; | |
3017 | + mbstate_t state; | |
3018 | + | |
3019 | + while (len > 0 && ismbblank (s, len, &mblength)) | |
3020 | + { | |
3021 | + s += mblength; | |
3022 | + len -= mblength; | |
3023 | + } | |
3024 | + | |
3025 | + if (len == 0) | |
3026 | + return 0; | |
3027 | + | |
3028 | + month = (char *) alloca (len + 1); | |
3029 | + | |
3030 | + tmp = (char *) alloca (len + 1); | |
3031 | + memcpy (tmp, s, len); | |
3032 | + tmp[len] = '\0'; | |
3033 | + pp = (const char **)&tmp; | |
3034 | + month_wcs = (wchar_t *) alloca ((len + 1) * sizeof (wchar_t)); | |
3035 | + memset (&state, '\0', sizeof(mbstate_t)); | |
3036 | + | |
3037 | + wclength = mbsrtowcs (month_wcs, pp, len + 1, &state); | |
1555d43c SS |
3038 | + if (wclength == (size_t)-1 || *pp != NULL) |
3039 | + error (SORT_FAILURE, 0, _("Invalid multibyte input %s."), quote(s)); | |
56ae3f82 SS |
3040 | + |
3041 | + for (i = 0; i < wclength; i++) | |
3042 | + { | |
3043 | + month_wcs[i] = towupper(month_wcs[i]); | |
3044 | + if (iswblank (month_wcs[i])) | |
3045 | + { | |
3046 | + month_wcs[i] = L'\0'; | |
3047 | + break; | |
3048 | + } | |
3049 | + } | |
3050 | + | |
3051 | + wpp = (const wchar_t **)&month_wcs; | |
3052 | + | |
3053 | + mblength = wcsrtombs (month, wpp, len + 1, &state); | |
3054 | + assert (mblength != (-1) && *wpp == NULL); | |
3055 | + | |
3056 | + do | |
3057 | + { | |
3058 | + int ix = (lo + hi) / 2; | |
3059 | + | |
3060 | + if (strncmp (month, monthtab[ix].name, strlen (monthtab[ix].name)) < 0) | |
3061 | + hi = ix; | |
3062 | + else | |
3063 | + lo = ix; | |
3064 | + } | |
3065 | + while (hi - lo > 1); | |
3066 | + | |
1555d43c SS |
3067 | + if (ea) |
3068 | + *ea = (char *) month; | |
3069 | + | |
56ae3f82 SS |
3070 | + result = (!strncmp (month, monthtab[lo].name, strlen (monthtab[lo].name)) |
3071 | + ? monthtab[lo].val : 0); | |
3072 | + | |
3073 | + return result; | |
3074 | +} | |
3075 | +#endif | |
3076 | + | |
3077 | /* Compare two lines A and B trying every key in sequence until there | |
3078 | are no more keys or a difference is found. */ | |
3079 | ||
3080 | static int | |
1555d43c | 3081 | -keycompare (struct line const *a, struct line const *b) |
56ae3f82 SS |
3082 | +keycompare_uni (const struct line *a, const struct line *b) |
3083 | { | |
3084 | struct keyfield *key = keylist; | |
3085 | ||
407c5be3 | 3086 | @@ -2506,7 +2940,7 @@ keycompare (struct line const *a, struct |
1555d43c SS |
3087 | else if (key->human_numeric) |
3088 | diff = human_numcompare (ta, tb); | |
3089 | else if (key->month) | |
3090 | - diff = getmonth (ta, NULL) - getmonth (tb, NULL); | |
3091 | + diff = getmonth (ta, tlena, NULL) - getmonth (tb, tlenb, NULL); | |
3092 | else if (key->random) | |
3093 | diff = compare_random (ta, tlena, tb, tlenb); | |
3094 | else if (key->version) | |
407c5be3 | 3095 | @@ -2622,6 +3056,179 @@ keycompare (struct line const *a, struct |
56ae3f82 SS |
3096 | return key->reverse ? -diff : diff; |
3097 | } | |
3098 | ||
3099 | +#if HAVE_MBRTOWC | |
3100 | +static int | |
3101 | +keycompare_mb (const struct line *a, const struct line *b) | |
3102 | +{ | |
3103 | + struct keyfield *key = keylist; | |
3104 | + | |
3105 | + /* For the first iteration only, the key positions have been | |
3106 | + precomputed for us. */ | |
3107 | + char *texta = a->keybeg; | |
3108 | + char *textb = b->keybeg; | |
3109 | + char *lima = a->keylim; | |
3110 | + char *limb = b->keylim; | |
3111 | + | |
3112 | + size_t mblength_a, mblength_b; | |
3113 | + wchar_t wc_a, wc_b; | |
3114 | + mbstate_t state_a, state_b; | |
3115 | + | |
3116 | + int diff; | |
3117 | + | |
3118 | + memset (&state_a, '\0', sizeof(mbstate_t)); | |
3119 | + memset (&state_b, '\0', sizeof(mbstate_t)); | |
3120 | + | |
3121 | + for (;;) | |
3122 | + { | |
3123 | + char const *translate = key->translate; | |
3124 | + bool const *ignore = key->ignore; | |
3125 | + | |
3126 | + /* Find the lengths. */ | |
3127 | + size_t lena = lima <= texta ? 0 : lima - texta; | |
3128 | + size_t lenb = limb <= textb ? 0 : limb - textb; | |
3129 | + | |
3130 | + /* Actually compare the fields. */ | |
3131 | + if (key->random) | |
3132 | + diff = compare_random (texta, lena, textb, lenb); | |
3133 | + else if (key->numeric | key->general_numeric | key->human_numeric) | |
3134 | + { | |
3135 | + char savea = *lima, saveb = *limb; | |
3136 | + | |
3137 | + *lima = *limb = '\0'; | |
3138 | + diff = (key->numeric ? numcompare (texta, textb) | |
3139 | + : key->general_numeric ? general_numcompare (texta, textb) | |
1555d43c | 3140 | + : human_numcompare (texta, textb)); |
56ae3f82 SS |
3141 | + *lima = savea, *limb = saveb; |
3142 | + } | |
3143 | + else if (key->version) | |
1555d43c | 3144 | + diff = filevercmp (texta, textb); |
56ae3f82 | 3145 | + else if (key->month) |
1555d43c | 3146 | + diff = getmonth (texta, lena, NULL) - getmonth (textb, lenb, NULL); |
56ae3f82 SS |
3147 | + else |
3148 | + { | |
3149 | + if (ignore || translate) | |
3150 | + { | |
3151 | + char *copy_a = (char *) alloca (lena + 1 + lenb + 1); | |
3152 | + char *copy_b = copy_a + lena + 1; | |
3153 | + size_t new_len_a, new_len_b; | |
3154 | + size_t i, j; | |
3155 | + | |
3156 | + /* Ignore and/or translate chars before comparing. */ | |
3157 | +# define IGNORE_CHARS(NEW_LEN, LEN, TEXT, COPY, WC, MBLENGTH, STATE) \ | |
3158 | + do \ | |
3159 | + { \ | |
3160 | + wchar_t uwc; \ | |
3161 | + char mbc[MB_LEN_MAX]; \ | |
3162 | + mbstate_t state_wc; \ | |
3163 | + \ | |
3164 | + for (NEW_LEN = i = 0; i < LEN;) \ | |
3165 | + { \ | |
3166 | + mbstate_t state_bak; \ | |
3167 | + \ | |
3168 | + state_bak = STATE; \ | |
3169 | + MBLENGTH = mbrtowc (&WC, TEXT + i, LEN - i, &STATE); \ | |
3170 | + \ | |
3171 | + if (MBLENGTH == (size_t)-2 || MBLENGTH == (size_t)-1 \ | |
3172 | + || MBLENGTH == 0) \ | |
3173 | + { \ | |
3174 | + if (MBLENGTH == (size_t)-2 || MBLENGTH == (size_t)-1) \ | |
3175 | + STATE = state_bak; \ | |
3176 | + if (!ignore) \ | |
3177 | + COPY[NEW_LEN++] = TEXT[i++]; \ | |
3178 | + continue; \ | |
3179 | + } \ | |
3180 | + \ | |
3181 | + if (ignore) \ | |
3182 | + { \ | |
3183 | + if ((ignore == nonprinting && !iswprint (WC)) \ | |
3184 | + || (ignore == nondictionary \ | |
3185 | + && !iswalnum (WC) && !iswblank (WC))) \ | |
3186 | + { \ | |
3187 | + i += MBLENGTH; \ | |
3188 | + continue; \ | |
3189 | + } \ | |
3190 | + } \ | |
3191 | + \ | |
3192 | + if (translate) \ | |
3193 | + { \ | |
3194 | + \ | |
3195 | + uwc = towupper(WC); \ | |
3196 | + if (WC == uwc) \ | |
3197 | + { \ | |
3198 | + memcpy (mbc, TEXT + i, MBLENGTH); \ | |
3199 | + i += MBLENGTH; \ | |
3200 | + } \ | |
3201 | + else \ | |
3202 | + { \ | |
3203 | + i += MBLENGTH; \ | |
3204 | + WC = uwc; \ | |
3205 | + memset (&state_wc, '\0', sizeof (mbstate_t)); \ | |
3206 | + \ | |
3207 | + MBLENGTH = wcrtomb (mbc, WC, &state_wc); \ | |
3208 | + assert (MBLENGTH != (size_t)-1 && MBLENGTH != 0); \ | |
3209 | + } \ | |
3210 | + \ | |
3211 | + for (j = 0; j < MBLENGTH; j++) \ | |
3212 | + COPY[NEW_LEN++] = mbc[j]; \ | |
3213 | + } \ | |
3214 | + else \ | |
3215 | + for (j = 0; j < MBLENGTH; j++) \ | |
3216 | + COPY[NEW_LEN++] = TEXT[i++]; \ | |
3217 | + } \ | |
3218 | + COPY[NEW_LEN] = '\0'; \ | |
3219 | + } \ | |
3220 | + while (0) | |
3221 | + IGNORE_CHARS (new_len_a, lena, texta, copy_a, | |
3222 | + wc_a, mblength_a, state_a); | |
3223 | + IGNORE_CHARS (new_len_b, lenb, textb, copy_b, | |
3224 | + wc_b, mblength_b, state_b); | |
3225 | + diff = xmemcoll (copy_a, new_len_a, copy_b, new_len_b); | |
3226 | + } | |
3227 | + else if (lena == 0) | |
3228 | + diff = - NONZERO (lenb); | |
3229 | + else if (lenb == 0) | |
3230 | + goto greater; | |
3231 | + else | |
3232 | + diff = xmemcoll (texta, lena, textb, lenb); | |
3233 | + } | |
3234 | + | |
3235 | + if (diff) | |
3236 | + goto not_equal; | |
3237 | + | |
3238 | + key = key->next; | |
3239 | + if (! key) | |
3240 | + break; | |
3241 | + | |
3242 | + /* Find the beginning and limit of the next field. */ | |
3243 | + if (key->eword != -1) | |
3244 | + lima = limfield (a, key), limb = limfield (b, key); | |
3245 | + else | |
3246 | + lima = a->text + a->length - 1, limb = b->text + b->length - 1; | |
3247 | + | |
3248 | + if (key->sword != -1) | |
3249 | + texta = begfield (a, key), textb = begfield (b, key); | |
3250 | + else | |
3251 | + { | |
3252 | + texta = a->text, textb = b->text; | |
3253 | + if (key->skipsblanks) | |
3254 | + { | |
3255 | + while (texta < lima && ismbblank (texta, lima - texta, &mblength_a)) | |
3256 | + texta += mblength_a; | |
3257 | + while (textb < limb && ismbblank (textb, limb - textb, &mblength_b)) | |
3258 | + textb += mblength_b; | |
3259 | + } | |
3260 | + } | |
3261 | + } | |
3262 | + | |
3263 | + return 0; | |
3264 | + | |
3265 | +greater: | |
3266 | + diff = 1; | |
3267 | +not_equal: | |
3268 | + return key->reverse ? -diff : diff; | |
3269 | +} | |
3270 | +#endif | |
3271 | + | |
3272 | /* Compare two lines A and B, returning negative, zero, or positive | |
3273 | depending on whether A compares less than, equal to, or greater than B. */ | |
3274 | ||
407c5be3 | 3275 | @@ -4084,7 +4691,7 @@ main (int argc, char **argv) |
56ae3f82 SS |
3276 | initialize_exit_failure (SORT_FAILURE); |
3277 | ||
3278 | hard_LC_COLLATE = hard_locale (LC_COLLATE); | |
3279 | -#if HAVE_NL_LANGINFO | |
3280 | +#if HAVE_LANGINFO_CODESET | |
3281 | hard_LC_TIME = hard_locale (LC_TIME); | |
3282 | #endif | |
3283 | ||
407c5be3 | 3284 | @@ -4105,6 +4712,29 @@ main (int argc, char **argv) |
56ae3f82 SS |
3285 | thousands_sep = -1; |
3286 | } | |
3287 | ||
3288 | +#if HAVE_MBRTOWC | |
3289 | + if (MB_CUR_MAX > 1) | |
3290 | + { | |
3291 | + inittables = inittables_mb; | |
3292 | + begfield = begfield_mb; | |
3293 | + limfield = limfield_mb; | |
1555d43c | 3294 | + skipblanks = skipblanks_mb; |
56ae3f82 SS |
3295 | + getmonth = getmonth_mb; |
3296 | + keycompare = keycompare_mb; | |
3297 | + numcompare = numcompare_mb; | |
3298 | + } | |
3299 | + else | |
3300 | +#endif | |
3301 | + { | |
3302 | + inittables = inittables_uni; | |
3303 | + begfield = begfield_uni; | |
3304 | + limfield = limfield_uni; | |
1555d43c | 3305 | + skipblanks = skipblanks_uni; |
56ae3f82 SS |
3306 | + getmonth = getmonth_uni; |
3307 | + keycompare = keycompare_uni; | |
3308 | + numcompare = numcompare_uni; | |
3309 | + } | |
3310 | + | |
3311 | have_read_stdin = false; | |
3312 | inittables (); | |
3313 | ||
407c5be3 | 3314 | @@ -4375,13 +5005,34 @@ main (int argc, char **argv) |
56ae3f82 SS |
3315 | |
3316 | case 't': | |
3317 | { | |
3318 | - char newtab = optarg[0]; | |
3319 | - if (! newtab) | |
3320 | + char newtab[MB_LEN_MAX + 1]; | |
3321 | + size_t newtab_length = 1; | |
3322 | + strncpy (newtab, optarg, MB_LEN_MAX); | |
3323 | + if (! newtab[0]) | |
3324 | error (SORT_FAILURE, 0, _("empty tab")); | |
3325 | - if (optarg[1]) | |
3326 | +#if HAVE_MBRTOWC | |
3327 | + if (MB_CUR_MAX > 1) | |
3328 | + { | |
3329 | + wchar_t wc; | |
3330 | + mbstate_t state; | |
56ae3f82 SS |
3331 | + |
3332 | + memset (&state, '\0', sizeof (mbstate_t)); | |
3333 | + newtab_length = mbrtowc (&wc, newtab, strnlen (newtab, | |
3334 | + MB_LEN_MAX), | |
3335 | + &state); | |
3336 | + switch (newtab_length) | |
3337 | + { | |
3338 | + case (size_t) -1: | |
3339 | + case (size_t) -2: | |
3340 | + case 0: | |
3341 | + newtab_length = 1; | |
3342 | + } | |
3343 | + } | |
3344 | +#endif | |
3345 | + if (newtab_length == 1 && optarg[1]) | |
3346 | { | |
3347 | if (STREQ (optarg, "\\0")) | |
3348 | - newtab = '\0'; | |
3349 | + newtab[0] = '\0'; | |
3350 | else | |
3351 | { | |
3352 | /* Provoke with `sort -txx'. Complain about | |
407c5be3 | 3353 | @@ -4392,9 +5043,12 @@ main (int argc, char **argv) |
56ae3f82 SS |
3354 | quote (optarg)); |
3355 | } | |
3356 | } | |
3357 | - if (tab != TAB_DEFAULT && tab != newtab) | |
3358 | + if (tab_length | |
3359 | + && (tab_length != newtab_length | |
3360 | + || memcmp (tab, newtab, tab_length) != 0)) | |
3361 | error (SORT_FAILURE, 0, _("incompatible tabs")); | |
3362 | - tab = newtab; | |
3363 | + memcpy (tab, newtab, newtab_length); | |
3364 | + tab_length = newtab_length; | |
3365 | } | |
3366 | break; | |
3367 | ||
407c5be3 SS |
3368 | diff -urNp coreutils-8.9-orig/src/unexpand.c coreutils-8.9/src/unexpand.c |
3369 | --- coreutils-8.9-orig/src/unexpand.c 2011-01-01 22:19:23.000000000 +0100 | |
3370 | +++ coreutils-8.9/src/unexpand.c 2011-01-04 17:41:55.387888171 +0100 | |
1555d43c | 3371 | @@ -39,12 +39,29 @@ |
56ae3f82 SS |
3372 | #include <stdio.h> |
3373 | #include <getopt.h> | |
3374 | #include <sys/types.h> | |
3375 | + | |
3376 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ | |
3377 | +#if HAVE_WCHAR_H | |
3378 | +# include <wchar.h> | |
3379 | +#endif | |
3380 | + | |
3381 | #include "system.h" | |
3382 | #include "error.h" | |
1555d43c | 3383 | #include "fadvise.h" |
56ae3f82 SS |
3384 | #include "quote.h" |
3385 | #include "xstrndup.h" | |
3386 | ||
3387 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC | |
3388 | + installation; work around this configuration error. */ | |
3389 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 | |
3390 | +# define MB_LEN_MAX 16 | |
3391 | +#endif | |
3392 | + | |
3393 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ | |
3394 | +#if HAVE_MBRTOWC && defined mbstate_t | |
3395 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) | |
3396 | +#endif | |
3397 | + | |
3398 | /* The official name of this program (e.g., no `g' prefix). */ | |
3399 | #define PROGRAM_NAME "unexpand" | |
3400 | ||
1555d43c | 3401 | @@ -104,6 +121,208 @@ static struct option const longopts[] = |
56ae3f82 SS |
3402 | {NULL, 0, NULL, 0} |
3403 | }; | |
3404 | ||
3405 | +static FILE *next_file (FILE *fp); | |
3406 | + | |
3407 | +#if HAVE_MBRTOWC | |
3408 | +static void | |
3409 | +unexpand_multibyte (void) | |
3410 | +{ | |
3411 | + FILE *fp; /* Input stream. */ | |
3412 | + mbstate_t i_state; /* Current shift state of the input stream. */ | |
3413 | + mbstate_t i_state_bak; /* Back up the I_STATE. */ | |
3414 | + mbstate_t o_state; /* Current shift state of the output stream. */ | |
3415 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ | |
3416 | + char *bufpos; /* Next read position of BUF. */ | |
3417 | + size_t buflen = 0; /* The length of the byte sequence in buf. */ | |
3418 | + wint_t wc; /* A gotten wide character. */ | |
3419 | + size_t mblength; /* The byte size of a multibyte character | |
3420 | + which shows as same character as WC. */ | |
3421 | + | |
3422 | + /* Index in `tab_list' of next tabstop: */ | |
3423 | + int tab_index = 0; /* For calculating width of pending tabs. */ | |
3424 | + int print_tab_index = 0; /* For printing as many tabs as possible. */ | |
3425 | + unsigned int column = 0; /* Column on screen of next char. */ | |
3426 | + int next_tab_column; /* Column the next tab stop is on. */ | |
3427 | + int convert = 1; /* If nonzero, perform translations. */ | |
3428 | + unsigned int pending = 0; /* Pending columns of blanks. */ | |
3429 | + | |
3430 | + fp = next_file ((FILE *) NULL); | |
3431 | + if (fp == NULL) | |
3432 | + return; | |
3433 | + | |
3434 | + memset (&o_state, '\0', sizeof(mbstate_t)); | |
3435 | + memset (&i_state, '\0', sizeof(mbstate_t)); | |
3436 | + | |
3437 | + for (;;) | |
3438 | + { | |
3439 | + if (buflen < MB_LEN_MAX && !feof(fp) && !ferror(fp)) | |
3440 | + { | |
3441 | + memmove (buf, bufpos, buflen); | |
3442 | + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, fp); | |
3443 | + bufpos = buf; | |
3444 | + } | |
3445 | + | |
3446 | + /* Get a wide character. */ | |
3447 | + if (buflen < 1) | |
3448 | + { | |
3449 | + mblength = 1; | |
3450 | + wc = WEOF; | |
3451 | + } | |
3452 | + else | |
3453 | + { | |
3454 | + i_state_bak = i_state; | |
3455 | + mblength = mbrtowc ((wchar_t *)&wc, bufpos, buflen, &i_state); | |
3456 | + } | |
3457 | + | |
3458 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) | |
3459 | + { | |
3460 | + i_state = i_state_bak; | |
3461 | + wc = L'\0'; | |
3462 | + } | |
3463 | + | |
3464 | + if (wc == L' ' && convert && column < INT_MAX) | |
3465 | + { | |
3466 | + ++pending; | |
3467 | + ++column; | |
3468 | + } | |
3469 | + else if (wc == L'\t' && convert) | |
3470 | + { | |
3471 | + if (tab_size == 0) | |
3472 | + { | |
3473 | + /* Do not let tab_index == first_free_tab; | |
3474 | + stop when it is 1 less. */ | |
3475 | + while (tab_index < first_free_tab - 1 | |
3476 | + && column >= tab_list[tab_index]) | |
3477 | + tab_index++; | |
3478 | + next_tab_column = tab_list[tab_index]; | |
3479 | + if (tab_index < first_free_tab - 1) | |
3480 | + tab_index++; | |
3481 | + if (column >= next_tab_column) | |
3482 | + { | |
3483 | + convert = 0; /* Ran out of tab stops. */ | |
3484 | + goto flush_pend_mb; | |
3485 | + } | |
3486 | + } | |
3487 | + else | |
3488 | + { | |
3489 | + next_tab_column = column + tab_size - column % tab_size; | |
3490 | + } | |
3491 | + pending += next_tab_column - column; | |
3492 | + column = next_tab_column; | |
3493 | + } | |
3494 | + else | |
3495 | + { | |
3496 | +flush_pend_mb: | |
3497 | + /* Flush pending spaces. Print as many tabs as possible, | |
3498 | + then print the rest as spaces. */ | |
3499 | + if (pending == 1) | |
3500 | + { | |
3501 | + putchar (' '); | |
3502 | + pending = 0; | |
3503 | + } | |
3504 | + column -= pending; | |
3505 | + while (pending > 0) | |
3506 | + { | |
3507 | + if (tab_size == 0) | |
3508 | + { | |
3509 | + /* Do not let print_tab_index == first_free_tab; | |
3510 | + stop when it is 1 less. */ | |
3511 | + while (print_tab_index < first_free_tab - 1 | |
3512 | + && column >= tab_list[print_tab_index]) | |
3513 | + print_tab_index++; | |
3514 | + next_tab_column = tab_list[print_tab_index]; | |
3515 | + if (print_tab_index < first_free_tab - 1) | |
3516 | + print_tab_index++; | |
3517 | + } | |
3518 | + else | |
3519 | + { | |
3520 | + next_tab_column = | |
3521 | + column + tab_size - column % tab_size; | |
3522 | + } | |
3523 | + if (next_tab_column - column <= pending) | |
3524 | + { | |
3525 | + putchar ('\t'); | |
3526 | + pending -= next_tab_column - column; | |
3527 | + column = next_tab_column; | |
3528 | + } | |
3529 | + else | |
3530 | + { | |
3531 | + --print_tab_index; | |
3532 | + column += pending; | |
3533 | + while (pending != 0) | |
3534 | + { | |
3535 | + putchar (' '); | |
3536 | + pending--; | |
3537 | + } | |
3538 | + } | |
3539 | + } | |
3540 | + | |
3541 | + if (wc == WEOF) | |
3542 | + { | |
3543 | + fp = next_file (fp); | |
3544 | + if (fp == NULL) | |
3545 | + break; /* No more files. */ | |
3546 | + else | |
3547 | + { | |
3548 | + memset (&i_state, '\0', sizeof(mbstate_t)); | |
3549 | + continue; | |
3550 | + } | |
3551 | + } | |
3552 | + | |
3553 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) | |
3554 | + { | |
3555 | + if (convert) | |
3556 | + { | |
3557 | + ++column; | |
3558 | + if (convert_entire_line == 0) | |
3559 | + convert = 0; | |
3560 | + } | |
3561 | + mblength = 1; | |
3562 | + putchar (buf[0]); | |
3563 | + } | |
3564 | + else if (mblength == 0) | |
3565 | + { | |
3566 | + if (convert && convert_entire_line == 0) | |
3567 | + convert = 0; | |
3568 | + mblength = 1; | |
3569 | + putchar ('\0'); | |
3570 | + } | |
3571 | + else | |
3572 | + { | |
3573 | + if (convert) | |
3574 | + { | |
3575 | + if (wc == L'\b') | |
3576 | + { | |
3577 | + if (column > 0) | |
3578 | + --column; | |
3579 | + } | |
3580 | + else | |
3581 | + { | |
3582 | + int width; /* The width of WC. */ | |
3583 | + | |
3584 | + width = wcwidth (wc); | |
3585 | + column += (width > 0) ? width : 0; | |
3586 | + if (convert_entire_line == 0) | |
3587 | + convert = 0; | |
3588 | + } | |
3589 | + } | |
3590 | + | |
3591 | + if (wc == L'\n') | |
3592 | + { | |
3593 | + tab_index = print_tab_index = 0; | |
3594 | + column = pending = 0; | |
3595 | + convert = 1; | |
3596 | + } | |
3597 | + fwrite (bufpos, sizeof(char), mblength, stdout); | |
3598 | + } | |
3599 | + } | |
3600 | + buflen -= mblength; | |
3601 | + bufpos += mblength; | |
3602 | + } | |
3603 | +} | |
3604 | +#endif | |
3605 | + | |
3606 | + | |
3607 | void | |
3608 | usage (int status) | |
3609 | { | |
1555d43c | 3610 | @@ -526,7 +745,12 @@ main (int argc, char **argv) |
56ae3f82 SS |
3611 | |
3612 | file_list = (optind < argc ? &argv[optind] : stdin_argv); | |
3613 | ||
3614 | - unexpand (); | |
3615 | +#if HAVE_MBRTOWC | |
3616 | + if (MB_CUR_MAX > 1) | |
3617 | + unexpand_multibyte (); | |
3618 | + else | |
3619 | +#endif | |
3620 | + unexpand (); | |
3621 | ||
3622 | if (have_read_stdin && fclose (stdin) != 0) | |
3623 | error (EXIT_FAILURE, errno, "-"); | |
407c5be3 SS |
3624 | diff -urNp coreutils-8.9-orig/src/uniq.c coreutils-8.9/src/uniq.c |
3625 | --- coreutils-8.9-orig/src/uniq.c 2011-01-01 22:19:23.000000000 +0100 | |
3626 | +++ coreutils-8.9/src/uniq.c 2011-01-04 17:41:55.391888381 +0100 | |
56ae3f82 SS |
3627 | @@ -21,6 +21,16 @@ |
3628 | #include <getopt.h> | |
3629 | #include <sys/types.h> | |
3630 | ||
3631 | +/* Get mbstate_t, mbrtowc(). */ | |
3632 | +#if HAVE_WCHAR_H | |
3633 | +# include <wchar.h> | |
3634 | +#endif | |
3635 | + | |
3636 | +/* Get isw* functions. */ | |
3637 | +#if HAVE_WCTYPE_H | |
3638 | +# include <wctype.h> | |
3639 | +#endif | |
3640 | + | |
3641 | #include "system.h" | |
3642 | #include "argmatch.h" | |
3643 | #include "linebuffer.h" | |
1555d43c | 3644 | @@ -32,7 +42,19 @@ |
56ae3f82 SS |
3645 | #include "stdio--.h" |
3646 | #include "xmemcoll.h" | |
3647 | #include "xstrtol.h" | |
3648 | -#include "memcasecmp.h" | |
3649 | +#include "xmemcoll.h" | |
3650 | + | |
3651 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC | |
3652 | + installation; work around this configuration error. */ | |
3653 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 | |
3654 | +# define MB_LEN_MAX 16 | |
3655 | +#endif | |
3656 | + | |
3657 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ | |
3658 | +#if HAVE_MBRTOWC && defined mbstate_t | |
3659 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) | |
3660 | +#endif | |
3661 | + | |
3662 | ||
3663 | /* The official name of this program (e.g., no `g' prefix). */ | |
3664 | #define PROGRAM_NAME "uniq" | |
1555d43c | 3665 | @@ -108,6 +130,10 @@ static enum delimit_method const delimit |
56ae3f82 SS |
3666 | /* Select whether/how to delimit groups of duplicate lines. */ |
3667 | static enum delimit_method delimit_groups; | |
3668 | ||
3669 | +/* Function pointers. */ | |
3670 | +static char * | |
3671 | +(*find_field) (struct linebuffer *line); | |
3672 | + | |
3673 | static struct option const longopts[] = | |
3674 | { | |
3675 | {"count", no_argument, NULL, 'c'}, | |
1555d43c | 3676 | @@ -207,7 +233,7 @@ size_opt (char const *opt, char const *m |
56ae3f82 SS |
3677 | return a pointer to the beginning of the line's field to be compared. */ |
3678 | ||
3679 | static char * | |
3680 | -find_field (struct linebuffer const *line) | |
3681 | +find_field_uni (struct linebuffer *line) | |
3682 | { | |
3683 | size_t count; | |
3684 | char const *lp = line->buffer; | |
1555d43c | 3685 | @@ -228,6 +254,83 @@ find_field (struct linebuffer const *lin |
56ae3f82 SS |
3686 | return line->buffer + i; |
3687 | } | |
3688 | ||
3689 | +#if HAVE_MBRTOWC | |
3690 | + | |
3691 | +# define MBCHAR_TO_WCHAR(WC, MBLENGTH, LP, POS, SIZE, STATEP, CONVFAIL) \ | |
3692 | + do \ | |
3693 | + { \ | |
3694 | + mbstate_t state_bak; \ | |
3695 | + \ | |
3696 | + CONVFAIL = 0; \ | |
3697 | + state_bak = *STATEP; \ | |
3698 | + \ | |
3699 | + MBLENGTH = mbrtowc (&WC, LP + POS, SIZE - POS, STATEP); \ | |
3700 | + \ | |
3701 | + switch (MBLENGTH) \ | |
3702 | + { \ | |
3703 | + case (size_t)-2: \ | |
3704 | + case (size_t)-1: \ | |
3705 | + *STATEP = state_bak; \ | |
3706 | + CONVFAIL++; \ | |
3707 | + /* Fall through */ \ | |
3708 | + case 0: \ | |
3709 | + MBLENGTH = 1; \ | |
3710 | + } \ | |
3711 | + } \ | |
3712 | + while (0) | |
3713 | + | |
3714 | +static char * | |
3715 | +find_field_multi (struct linebuffer *line) | |
3716 | +{ | |
3717 | + size_t count; | |
3718 | + char *lp = line->buffer; | |
3719 | + size_t size = line->length - 1; | |
3720 | + size_t pos; | |
3721 | + size_t mblength; | |
3722 | + wchar_t wc; | |
3723 | + mbstate_t *statep; | |
3724 | + int convfail; | |
3725 | + | |
3726 | + pos = 0; | |
3727 | + statep = &(line->state); | |
3728 | + | |
3729 | + /* skip fields. */ | |
3730 | + for (count = 0; count < skip_fields && pos < size; count++) | |
3731 | + { | |
3732 | + while (pos < size) | |
3733 | + { | |
3734 | + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); | |
3735 | + | |
3736 | + if (convfail || !iswblank (wc)) | |
3737 | + { | |
3738 | + pos += mblength; | |
3739 | + break; | |
3740 | + } | |
3741 | + pos += mblength; | |
3742 | + } | |
3743 | + | |
3744 | + while (pos < size) | |
3745 | + { | |
3746 | + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); | |
3747 | + | |
3748 | + if (!convfail && iswblank (wc)) | |
3749 | + break; | |
3750 | + | |
3751 | + pos += mblength; | |
3752 | + } | |
3753 | + } | |
3754 | + | |
3755 | + /* skip fields. */ | |
3756 | + for (count = 0; count < skip_chars && pos < size; count++) | |
3757 | + { | |
3758 | + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); | |
3759 | + pos += mblength; | |
3760 | + } | |
3761 | + | |
3762 | + return lp + pos; | |
3763 | +} | |
3764 | +#endif | |
3765 | + | |
3766 | /* Return false if two strings OLD and NEW match, true if not. | |
3767 | OLD and NEW point not to the beginnings of the lines | |
3768 | but rather to the beginnings of the fields to compare. | |
1555d43c | 3769 | @@ -236,6 +339,8 @@ find_field (struct linebuffer const *lin |
56ae3f82 SS |
3770 | static bool |
3771 | different (char *old, char *new, size_t oldlen, size_t newlen) | |
3772 | { | |
3773 | + char *copy_old, *copy_new; | |
3774 | + | |
3775 | if (check_chars < oldlen) | |
3776 | oldlen = check_chars; | |
3777 | if (check_chars < newlen) | |
1555d43c | 3778 | @@ -243,14 +348,92 @@ different (char *old, char *new, size_t |
56ae3f82 SS |
3779 | |
3780 | if (ignore_case) | |
3781 | { | |
3782 | - /* FIXME: This should invoke strcoll somehow. */ | |
3783 | - return oldlen != newlen || memcasecmp (old, new, oldlen); | |
3784 | + size_t i; | |
3785 | + | |
3786 | + copy_old = alloca (oldlen + 1); | |
3787 | + copy_new = alloca (oldlen + 1); | |
3788 | + | |
3789 | + for (i = 0; i < oldlen; i++) | |
3790 | + { | |
3791 | + copy_old[i] = toupper (old[i]); | |
3792 | + copy_new[i] = toupper (new[i]); | |
3793 | + } | |
3794 | } | |
3795 | - else if (hard_LC_COLLATE) | |
3796 | - return xmemcoll (old, oldlen, new, newlen) != 0; | |
3797 | else | |
3798 | - return oldlen != newlen || memcmp (old, new, oldlen); | |
3799 | + { | |
3800 | + copy_old = (char *)old; | |
3801 | + copy_new = (char *)new; | |
3802 | + } | |
3803 | + | |
3804 | + return xmemcoll (copy_old, oldlen, copy_new, newlen); | |
3805 | +} | |
3806 | + | |
3807 | +#if HAVE_MBRTOWC | |
3808 | +static int | |
3809 | +different_multi (const char *old, const char *new, size_t oldlen, size_t newlen, mbstate_t oldstate, mbstate_t newstate) | |
3810 | +{ | |
3811 | + size_t i, j, chars; | |
3812 | + const char *str[2]; | |
3813 | + char *copy[2]; | |
3814 | + size_t len[2]; | |
3815 | + mbstate_t state[2]; | |
3816 | + size_t mblength; | |
3817 | + wchar_t wc, uwc; | |
3818 | + mbstate_t state_bak; | |
3819 | + | |
3820 | + str[0] = old; | |
3821 | + str[1] = new; | |
3822 | + len[0] = oldlen; | |
3823 | + len[1] = newlen; | |
3824 | + state[0] = oldstate; | |
3825 | + state[1] = newstate; | |
3826 | + | |
3827 | + for (i = 0; i < 2; i++) | |
3828 | + { | |
3829 | + copy[i] = alloca (len[i] + 1); | |
3830 | + | |
3831 | + for (j = 0, chars = 0; j < len[i] && chars < check_chars; chars++) | |
3832 | + { | |
3833 | + state_bak = state[i]; | |
3834 | + mblength = mbrtowc (&wc, str[i] + j, len[i] - j, &(state[i])); | |
3835 | + | |
3836 | + switch (mblength) | |
3837 | + { | |
3838 | + case (size_t)-1: | |
3839 | + case (size_t)-2: | |
3840 | + state[i] = state_bak; | |
3841 | + /* Fall through */ | |
3842 | + case 0: | |
3843 | + mblength = 1; | |
3844 | + break; | |
3845 | + | |
3846 | + default: | |
3847 | + if (ignore_case) | |
3848 | + { | |
3849 | + uwc = towupper (wc); | |
3850 | + | |
3851 | + if (uwc != wc) | |
3852 | + { | |
3853 | + mbstate_t state_wc; | |
3854 | + | |
3855 | + memset (&state_wc, '\0', sizeof(mbstate_t)); | |
3856 | + wcrtomb (copy[i] + j, uwc, &state_wc); | |
3857 | + } | |
3858 | + else | |
3859 | + memcpy (copy[i] + j, str[i] + j, mblength); | |
3860 | + } | |
3861 | + else | |
3862 | + memcpy (copy[i] + j, str[i] + j, mblength); | |
3863 | + } | |
3864 | + j += mblength; | |
3865 | + } | |
3866 | + copy[i][j] = '\0'; | |
3867 | + len[i] = j; | |
3868 | + } | |
3869 | + | |
3870 | + return xmemcoll (copy[0], len[0], copy[1], len[1]); | |
3871 | } | |
3872 | +#endif | |
3873 | ||
3874 | /* Output the line in linebuffer LINE to standard output | |
3875 | provided that the switches say it should be output. | |
1555d43c | 3876 | @@ -306,15 +489,43 @@ check_file (const char *infile, const ch |
56ae3f82 | 3877 | { |
1555d43c SS |
3878 | char *prevfield IF_LINT ( = NULL); |
3879 | size_t prevlen IF_LINT ( = 0); | |
56ae3f82 SS |
3880 | +#if HAVE_MBRTOWC |
3881 | + mbstate_t prevstate; | |
3882 | + | |
3883 | + memset (&prevstate, '\0', sizeof (mbstate_t)); | |
3884 | +#endif | |
3885 | ||
3886 | while (!feof (stdin)) | |
3887 | { | |
3888 | char *thisfield; | |
3889 | size_t thislen; | |
3890 | +#if HAVE_MBRTOWC | |
3891 | + mbstate_t thisstate; | |
3892 | +#endif | |
3893 | + | |
3894 | if (readlinebuffer_delim (thisline, stdin, delimiter) == 0) | |
3895 | break; | |
3896 | thisfield = find_field (thisline); | |
3897 | thislen = thisline->length - 1 - (thisfield - thisline->buffer); | |
3898 | +#if HAVE_MBRTOWC | |
3899 | + if (MB_CUR_MAX > 1) | |
3900 | + { | |
3901 | + thisstate = thisline->state; | |
3902 | + | |
3903 | + if (prevline->length == 0 || different_multi | |
3904 | + (thisfield, prevfield, thislen, prevlen, thisstate, prevstate)) | |
3905 | + { | |
3906 | + fwrite (thisline->buffer, sizeof (char), | |
3907 | + thisline->length, stdout); | |
3908 | + | |
3909 | + SWAP_LINES (prevline, thisline); | |
3910 | + prevfield = thisfield; | |
3911 | + prevlen = thislen; | |
3912 | + prevstate = thisstate; | |
3913 | + } | |
3914 | + } | |
3915 | + else | |
3916 | +#endif | |
3917 | if (prevline->length == 0 | |
3918 | || different (thisfield, prevfield, thislen, prevlen)) | |
3919 | { | |
1555d43c | 3920 | @@ -333,17 +544,26 @@ check_file (const char *infile, const ch |
56ae3f82 SS |
3921 | size_t prevlen; |
3922 | uintmax_t match_count = 0; | |
3923 | bool first_delimiter = true; | |
3924 | +#if HAVE_MBRTOWC | |
3925 | + mbstate_t prevstate; | |
3926 | +#endif | |
3927 | ||
3928 | if (readlinebuffer_delim (prevline, stdin, delimiter) == 0) | |
3929 | goto closefiles; | |
3930 | prevfield = find_field (prevline); | |
3931 | prevlen = prevline->length - 1 - (prevfield - prevline->buffer); | |
3932 | +#if HAVE_MBRTOWC | |
3933 | + prevstate = prevline->state; | |
3934 | +#endif | |
3935 | ||
3936 | while (!feof (stdin)) | |
3937 | { | |
3938 | bool match; | |
3939 | char *thisfield; | |
3940 | size_t thislen; | |
3941 | +#if HAVE_MBRTOWC | |
3942 | + mbstate_t thisstate; | |
3943 | +#endif | |
3944 | if (readlinebuffer_delim (thisline, stdin, delimiter) == 0) | |
3945 | { | |
3946 | if (ferror (stdin)) | |
1555d43c | 3947 | @@ -352,6 +572,15 @@ check_file (const char *infile, const ch |
56ae3f82 SS |
3948 | } |
3949 | thisfield = find_field (thisline); | |
3950 | thislen = thisline->length - 1 - (thisfield - thisline->buffer); | |
3951 | +#if HAVE_MBRTOWC | |
3952 | + if (MB_CUR_MAX > 1) | |
3953 | + { | |
3954 | + thisstate = thisline->state; | |
3955 | + match = !different_multi (thisfield, prevfield, | |
3956 | + thislen, prevlen, thisstate, prevstate); | |
3957 | + } | |
3958 | + else | |
3959 | +#endif | |
3960 | match = !different (thisfield, prevfield, thislen, prevlen); | |
3961 | match_count += match; | |
3962 | ||
1555d43c | 3963 | @@ -384,6 +613,9 @@ check_file (const char *infile, const ch |
56ae3f82 SS |
3964 | SWAP_LINES (prevline, thisline); |
3965 | prevfield = thisfield; | |
3966 | prevlen = thislen; | |
3967 | +#if HAVE_MBRTOWC | |
3968 | + prevstate = thisstate; | |
3969 | +#endif | |
3970 | if (!match) | |
3971 | match_count = 0; | |
3972 | } | |
1555d43c | 3973 | @@ -429,6 +661,19 @@ main (int argc, char **argv) |
56ae3f82 SS |
3974 | |
3975 | atexit (close_stdout); | |
3976 | ||
3977 | +#if HAVE_MBRTOWC | |
3978 | + if (MB_CUR_MAX > 1) | |
3979 | + { | |
3980 | + find_field = find_field_multi; | |
3981 | + } | |
3982 | + else | |
3983 | +#endif | |
3984 | + { | |
3985 | + find_field = find_field_uni; | |
3986 | + } | |
3987 | + | |
3988 | + | |
3989 | + | |
3990 | skip_chars = 0; | |
3991 | skip_fields = 0; | |
3992 | check_chars = SIZE_MAX; | |
407c5be3 SS |
3993 | diff -urNp coreutils-8.9-orig/tests/Makefile.am coreutils-8.9/tests/Makefile.am |
3994 | --- coreutils-8.9-orig/tests/Makefile.am 2011-01-04 17:41:12.682173268 +0100 | |
3995 | +++ coreutils-8.9/tests/Makefile.am 2011-01-04 17:41:55.392900534 +0100 | |
3996 | @@ -233,6 +233,7 @@ TESTS = \ | |
1555d43c SS |
3997 | misc/sort-debug-keys \ |
3998 | misc/sort-debug-warn \ | |
56ae3f82 SS |
3999 | misc/sort-files0-from \ |
4000 | + misc/sort-mb-tests \ | |
1555d43c | 4001 | misc/sort-float \ |
56ae3f82 SS |
4002 | misc/sort-merge \ |
4003 | misc/sort-merge-fdlimit \ | |
407c5be3 | 4004 | @@ -498,6 +499,10 @@ TESTS = \ |
56ae3f82 SS |
4005 | $(root_tests) |
4006 | ||
4007 | pr_data = \ | |
4008 | + misc/mb1.X \ | |
4009 | + misc/mb1.I \ | |
4010 | + misc/mb2.X \ | |
4011 | + misc/mb2.I \ | |
4012 | pr/0F \ | |
4013 | pr/0FF \ | |
4014 | pr/0FFnt \ | |
407c5be3 SS |
4015 | diff -urNp coreutils-8.9-orig/tests/misc/cut coreutils-8.9/tests/misc/cut |
4016 | --- coreutils-8.9-orig/tests/misc/cut 2011-01-01 22:19:23.000000000 +0100 | |
4017 | +++ coreutils-8.9/tests/misc/cut 2011-01-04 17:41:55.393887822 +0100 | |
56ae3f82 SS |
4018 | @@ -26,7 +26,7 @@ use strict; |
4019 | my $prog = 'cut'; | |
4020 | my $try = "Try \`$prog --help' for more information.\n"; | |
4021 | my $from_1 = "$prog: fields and positions are numbered from 1\n$try"; | |
4022 | -my $inval = "$prog: invalid byte or field list\n$try"; | |
4023 | +my $inval = "$prog: invalid byte, character or field list\n$try"; | |
4024 | my $no_endpoint = "$prog: invalid range with no endpoint: -\n$try"; | |
4025 | ||
4026 | my @Tests = | |
407c5be3 | 4027 | @@ -143,7 +143,7 @@ my @Tests = |
56ae3f82 SS |
4028 | |
4029 | # None of the following invalid ranges provoked an error up to coreutils-6.9. | |
4030 | ['inval1', qw(-f 2-0), {IN=>''}, {OUT=>''}, {EXIT=>1}, | |
4031 | - {ERR=>"$prog: invalid decreasing range\n$try"}], | |
4032 | + {ERR=>"$prog: invalid byte, character or field list\n$try"}], | |
4033 | ['inval2', qw(-f -), {IN=>''}, {OUT=>''}, {EXIT=>1}, {ERR=>$no_endpoint}], | |
4034 | ['inval3', '-f', '4,-', {IN=>''}, {OUT=>''}, {EXIT=>1}, {ERR=>$no_endpoint}], | |
407c5be3 SS |
4035 | ['inval4', '-f', '1-2,-', {IN=>''}, {OUT=>''}, {EXIT=>1}, |
4036 | diff -urNp coreutils-8.9-orig/tests/misc/mb1.I coreutils-8.9/tests/misc/mb1.I | |
4037 | --- coreutils-8.9-orig/tests/misc/mb1.I 1970-01-01 01:00:00.000000000 +0100 | |
4038 | +++ coreutils-8.9/tests/misc/mb1.I 2011-01-04 17:41:55.394899835 +0100 | |
56ae3f82 SS |
4039 | @@ -0,0 +1,4 @@ |
4040 | +Apple@10 | |
4041 | +Banana@5 | |
4042 | +Citrus@20 | |
4043 | +Cherry@30 | |
407c5be3 SS |
4044 | diff -urNp coreutils-8.9-orig/tests/misc/mb1.X coreutils-8.9/tests/misc/mb1.X |
4045 | --- coreutils-8.9-orig/tests/misc/mb1.X 1970-01-01 01:00:00.000000000 +0100 | |
4046 | +++ coreutils-8.9/tests/misc/mb1.X 2011-01-04 17:41:55.395888102 +0100 | |
56ae3f82 SS |
4047 | @@ -0,0 +1,4 @@ |
4048 | +Banana@5 | |
4049 | +Apple@10 | |
4050 | +Citrus@20 | |
4051 | +Cherry@30 | |
407c5be3 SS |
4052 | diff -urNp coreutils-8.9-orig/tests/misc/mb2.I coreutils-8.9/tests/misc/mb2.I |
4053 | --- coreutils-8.9-orig/tests/misc/mb2.I 1970-01-01 01:00:00.000000000 +0100 | |
4054 | +++ coreutils-8.9/tests/misc/mb2.I 2011-01-04 17:41:55.395888102 +0100 | |
56ae3f82 SS |
4055 | @@ -0,0 +1,4 @@ |
4056 | +Apple@AA10@@20 | |
4057 | +Banana@AA5@@30 | |
4058 | +Citrus@AA20@@5 | |
4059 | +Cherry@AA30@@10 | |
407c5be3 SS |
4060 | diff -urNp coreutils-8.9-orig/tests/misc/mb2.X coreutils-8.9/tests/misc/mb2.X |
4061 | --- coreutils-8.9-orig/tests/misc/mb2.X 1970-01-01 01:00:00.000000000 +0100 | |
4062 | +++ coreutils-8.9/tests/misc/mb2.X 2011-01-04 17:41:55.396892432 +0100 | |
56ae3f82 SS |
4063 | @@ -0,0 +1,4 @@ |
4064 | +Citrus@AA20@@5 | |
4065 | +Cherry@AA30@@10 | |
4066 | +Apple@AA10@@20 | |
4067 | +Banana@AA5@@30 | |
407c5be3 SS |
4068 | diff -urNp coreutils-8.9-orig/tests/misc/sort-mb-tests coreutils-8.9/tests/misc/sort-mb-tests |
4069 | --- coreutils-8.9-orig/tests/misc/sort-mb-tests 1970-01-01 01:00:00.000000000 +0100 | |
4070 | +++ coreutils-8.9/tests/misc/sort-mb-tests 2011-01-04 17:41:55.396892432 +0100 | |
56ae3f82 SS |
4071 | @@ -0,0 +1,58 @@ |
4072 | +#! /bin/sh | |
4073 | +case $# in | |
4074 | + 0) xx='../src/sort';; | |
4075 | + *) xx="$1";; | |
4076 | +esac | |
4077 | +test "$VERBOSE" && echo=echo || echo=: | |
4078 | +$echo testing program: $xx | |
4079 | +errors=0 | |
4080 | +test "$srcdir" || srcdir=. | |
4081 | +test "$VERBOSE" && $xx --version 2> /dev/null | |
4082 | + | |
4083 | +export LC_ALL=en_US.UTF-8 | |
4084 | +locale -k LC_CTYPE 2>&1 | grep -q charmap.*UTF-8 || exit 77 | |
4085 | +errors=0 | |
4086 | + | |
4087 | +$xx -t @ -k2 -n misc/mb1.I > misc/mb1.O | |
4088 | +code=$? | |
4089 | +if test $code != 0; then | |
4090 | + $echo "Test mb1 failed: $xx return code $code differs from expected value 0" 1>&2 | |
4091 | + errors=`expr $errors + 1` | |
4092 | +else | |
4093 | + cmp misc/mb1.O $srcdir/misc/mb1.X > /dev/null 2>&1 | |
4094 | + case $? in | |
4095 | + 0) if test "$VERBOSE"; then $echo "passed mb1"; fi;; | |
4096 | + 1) $echo "Test mb1 failed: files misc/mb1.O and $srcdir/misc/mb1.X differ" 1>&2 | |
4097 | + (diff -c misc/mb1.O $srcdir/misc/mb1.X) 2> /dev/null | |
4098 | + errors=`expr $errors + 1`;; | |
4099 | + 2) $echo "Test mb1 may have failed." 1>&2 | |
4100 | + $echo The command "cmp misc/mb1.O $srcdir/misc/mb1.X" failed. 1>&2 | |
4101 | + errors=`expr $errors + 1`;; | |
4102 | + esac | |
4103 | +fi | |
4104 | + | |
4105 | +$xx -t @ -k4 -n misc/mb2.I > misc/mb2.O | |
4106 | +code=$? | |
4107 | +if test $code != 0; then | |
4108 | + $echo "Test mb2 failed: $xx return code $code differs from expected value 0" 1>&2 | |
4109 | + errors=`expr $errors + 1` | |
4110 | +else | |
4111 | + cmp misc/mb2.O $srcdir/misc/mb2.X > /dev/null 2>&1 | |
4112 | + case $? in | |
4113 | + 0) if test "$VERBOSE"; then $echo "passed mb2"; fi;; | |
4114 | + 1) $echo "Test mb2 failed: files misc/mb2.O and $srcdir/misc/mb2.X differ" 1>&2 | |
4115 | + (diff -c misc/mb2.O $srcdir/misc/mb2.X) 2> /dev/null | |
4116 | + errors=`expr $errors + 1`;; | |
4117 | + 2) $echo "Test mb2 may have failed." 1>&2 | |
4118 | + $echo The command "cmp misc/mb2.O $srcdir/misc/mb2.X" failed. 1>&2 | |
4119 | + errors=`expr $errors + 1`;; | |
4120 | + esac | |
4121 | +fi | |
4122 | + | |
4123 | +if test $errors = 0; then | |
4124 | + $echo Passed all 113 tests. 1>&2 | |
4125 | +else | |
4126 | + $echo Failed $errors tests. 1>&2 | |
4127 | +fi | |
4128 | +test $errors = 0 || errors=1 | |
4129 | +exit $errors |