Contents of /trunk/coreutils/patches-5.3.0/coreutils-5.3.0-i18n.patch
Parent Directory | Revision Log
Revision 44 -
(show annotations)
(download)
Thu Oct 13 21:17:16 2005 UTC (18 years, 11 months ago) by niro
File size: 102675 byte(s)
Thu Oct 13 21:17:16 2005 UTC (18 years, 11 months ago) by niro
File size: 102675 byte(s)
patch set for coretutils-5.3.0
1 | http://www.openi18n.org/download/utildev/coreutils-5.3.0-i18n-0.1.patch |
2 | |
3 | --- coreutils/lib/linebuffer.h |
4 | +++ coreutils/lib/linebuffer.h |
5 | @@ -22,6 +22,11 @@ |
6 | |
7 | # include <stdio.h> |
8 | |
9 | +/* Get mbstate_t. */ |
10 | +# if HAVE_WCHAR_H |
11 | +# include <wchar.h> |
12 | +# endif |
13 | + |
14 | /* A `struct linebuffer' holds a line of text. */ |
15 | |
16 | struct linebuffer |
17 | @@ -29,6 +34,9 @@ struct linebuffer |
18 | size_t size; /* Allocated. */ |
19 | size_t length; /* Used. */ |
20 | char *buffer; |
21 | +# if HAVE_WCHAR_H |
22 | + mbstate_t state; |
23 | +# endif |
24 | }; |
25 | |
26 | /* Initialize linebuffer LINEBUFFER for use. */ |
27 | --- coreutils/src/cut.c |
28 | +++ coreutils/src/cut.c |
29 | @@ -29,6 +29,12 @@ |
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 | + |
39 | #include "system.h" |
40 | |
41 | #include "error.h" |
42 | @@ -37,6 +43,13 @@ |
43 | #include "quote.h" |
44 | #include "xstrndup.h" |
45 | |
46 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
47 | + installation; work around this configuration error. */ |
48 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
49 | +# undef MB_LEN_MAX |
50 | +# define MB_LEN_MAX 16 |
51 | +#endif |
52 | + |
53 | /* The official name of this program (e.g., no `g' prefix). */ |
54 | #define PROGRAM_NAME "cut" |
55 | |
56 | @@ -73,6 +86,54 @@ struct range_pair |
57 | size_t hi; |
58 | }; |
59 | |
60 | +/* Refill the buffer BUF. */ |
61 | +#define REFILL_BUFFER(BUF, BUFPOS, BUFLEN, STREAM) \ |
62 | + do \ |
63 | + { \ |
64 | + if (BUFLEN < MB_LEN_MAX && !feof (STREAM) && !ferror (STREAM)) \ |
65 | + { \ |
66 | + memmove (BUF, BUFPOS, BUFLEN); \ |
67 | + BUFLEN += fread (BUF + BUFLEN, sizeof(char), BUFSIZ, STREAM); \ |
68 | + BUFPOS = BUF; \ |
69 | + } \ |
70 | + } \ |
71 | + while (0) |
72 | + |
73 | +/* Get wide character which starts at BUFPOS. If the byte sequence is |
74 | + not valid as a character, CONVFAIL is 1. Otherwise 0. */ |
75 | +#define GET_NEXT_WC_FROM_BUFFER(WC, BUFPOS, BUFLEN, MBLENGTH, STATE, CONVFAIL) \ |
76 | + do \ |
77 | + { \ |
78 | + wchar_t tmp; \ |
79 | + mbstate_t state_bak; \ |
80 | + \ |
81 | + if (BUFLEN < 1) \ |
82 | + { \ |
83 | + WC = WEOF; \ |
84 | + break; \ |
85 | + } \ |
86 | + \ |
87 | + /* Get a wide character. */ \ |
88 | + CONVFAIL = 0; \ |
89 | + state_bak = STATE; \ |
90 | + MBLENGTH = mbrtowc (&tmp, BUFPOS, BUFLEN, &STATE); \ |
91 | + WC = tmp; \ |
92 | + \ |
93 | + switch (MBLENGTH) \ |
94 | + { \ |
95 | + case (size_t)-1: \ |
96 | + case (size_t)-2: \ |
97 | + ++CONVFAIL; \ |
98 | + STATE = state_bak; \ |
99 | + /* Fall througn. */ \ |
100 | + \ |
101 | + case 0: \ |
102 | + MBLENGTH = 1; \ |
103 | + break; \ |
104 | + } \ |
105 | + } \ |
106 | + while (0) |
107 | + |
108 | /* This buffer is used to support the semantics of the -s option |
109 | (or lack of same) when the specified field list includes (does |
110 | not include) the first field. In both of those cases, the entire |
111 | @@ -85,7 +146,7 @@ static char *field_1_buffer; |
112 | /* The number of bytes allocated for FIELD_1_BUFFER. */ |
113 | static size_t field_1_bufsize; |
114 | |
115 | -/* The largest field or byte index used as an endpoint of a closed |
116 | +/* The largest field, character or byte index used as an endpoint of a closed |
117 | or degenerate range specification; this doesn't include the starting |
118 | index of right-open-ended ranges. For example, with either range spec |
119 | `2-5,9-', `2-3,5,9-' this variable would be set to 5. */ |
120 | @@ -97,10 +158,11 @@ static size_t eol_range_start; |
121 | |
122 | /* This is a bit vector. |
123 | In byte mode, which bytes to output. |
124 | + In character mode, which characters to output. |
125 | In field mode, which DELIM-separated fields to output. |
126 | - Both bytes and fields are numbered starting with 1, |
127 | + Bytes, characters and fields are numbered starting with 1, |
128 | so the zeroth bit of this array is unused. |
129 | - A field or byte K has been selected if |
130 | + A byte, character or field K has been selected if |
131 | (K <= MAX_RANGE_ENDPOINT and is_printable_field(K)) |
132 | || (EOL_RANGE_START > 0 && K >= EOL_RANGE_START). */ |
133 | static unsigned char *printable_field; |
134 | @@ -109,9 +171,12 @@ enum operating_mode |
135 | { |
136 | undefined_mode, |
137 | |
138 | - /* Output characters that are in the given bytes. */ |
139 | + /* Output bytes that are in the given bytes. */ |
140 | byte_mode, |
141 | |
142 | + /* Output characters that are at the given positions. */ |
143 | + character_mode, |
144 | + |
145 | /* Output the given delimeter-separated fields. */ |
146 | field_mode |
147 | }; |
148 | @@ -121,6 +186,13 @@ char *program_name; |
149 | |
150 | static enum operating_mode operating_mode; |
151 | |
152 | +/* If true, when in byte mode, don't split multibyte characters. */ |
153 | +static bool byte_mode_character_aware; |
154 | + |
155 | +/* If true, the function for single byte locale is work |
156 | + if this program runs on multibyte locale. */ |
157 | +static bool force_singlebyte_mode; |
158 | + |
159 | /* If true do not output lines containing no delimeter characters. |
160 | Otherwise, all such lines are printed. This option is valid only |
161 | with field mode. */ |
162 | @@ -132,6 +204,9 @@ static bool complement; |
163 | |
164 | /* The delimeter character for field mode. */ |
165 | static unsigned char delim; |
166 | +#if HAVE_WCHAR_H |
167 | +static wchar_t wcdelim; |
168 | +#endif |
169 | |
170 | /* True if the --output-delimiter=STRING option was specified. */ |
171 | static bool output_delimiter_specified; |
172 | @@ -205,7 +280,7 @@ Mandatory arguments to long options are |
173 | -f, --fields=LIST select only these fields; also print any line\n\ |
174 | that contains no delimiter character, unless\n\ |
175 | the -s option is specified\n\ |
176 | - -n (ignored)\n\ |
177 | + -n with -b: don't split multibyte characters\n\ |
178 | "), stdout); |
179 | fputs (_("\ |
180 | --complement complement the set of selected bytes, characters\n\ |
181 | @@ -360,7 +435,7 @@ set_fields (const char *fieldstr) |
182 | in_digits = false; |
183 | /* Starting a range. */ |
184 | if (dash_found) |
185 | - FATAL_ERROR (_("invalid byte or field list")); |
186 | + FATAL_ERROR (_("invalid byte, character or field list")); |
187 | dash_found = true; |
188 | fieldstr++; |
189 | |
190 | @@ -385,14 +460,16 @@ set_fields (const char *fieldstr) |
191 | if (value == 0) |
192 | { |
193 | /* `n-'. From `initial' to end of line. */ |
194 | - eol_range_start = initial; |
195 | + if(eol_range_start == 0 || |
196 | + (eol_range_start != 0 && eol_range_start > initial)) |
197 | + eol_range_start = initial; |
198 | field_found = true; |
199 | } |
200 | else |
201 | { |
202 | /* `m-n' or `-n' (1-n). */ |
203 | if (value < initial) |
204 | - FATAL_ERROR (_("invalid byte or field list")); |
205 | + FATAL_ERROR (_("invalid byte, character or field list")); |
206 | |
207 | /* Is there already a range going to end of line? */ |
208 | if (eol_range_start != 0) |
209 | @@ -478,7 +555,7 @@ set_fields (const char *fieldstr) |
210 | fieldstr++; |
211 | } |
212 | else |
213 | - FATAL_ERROR (_("invalid byte or field list")); |
214 | + FATAL_ERROR (_("invalid byte, character or field list")); |
215 | } |
216 | |
217 | max_range_endpoint = 0; |
218 | @@ -571,6 +648,81 @@ cut_bytes (FILE *stream) |
219 | } |
220 | } |
221 | |
222 | +#if HAVE_MBRTOWC |
223 | +/* This function is in use for the following case. |
224 | + |
225 | + 1. Read from the stream STREAM, printing to standard output any selected |
226 | + characters. |
227 | + |
228 | + 2. Read from stream STREAM, printing to standard output any selected bytes, |
229 | + without splitting multibyte characters. */ |
230 | + |
231 | +static void |
232 | +cut_characters_or_cut_bytes_no_split (FILE *stream) |
233 | +{ |
234 | + size_t idx; /* Number of bytes or characters in the line so far. */ |
235 | + /* Whether to begin printing delimiters between ranges for the current line. |
236 | + Set after we've begun printing data corresponding to the first range. */ |
237 | + bool print_delimiter; |
238 | + |
239 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
240 | + char *bufpos; /* Next read position of BUF. */ |
241 | + size_t buflen; /* The length of the byte sequence in buf. */ |
242 | + wint_t wc; /* A gotten wide character. */ |
243 | + size_t mblength; /* The byte size of a multibyte character which shows |
244 | + as same character as WC. */ |
245 | + mbstate_t state; /* State of the stream. */ |
246 | + int convfail; /* 1, when conversion is failed. Otherwise 0. */ |
247 | + |
248 | + |
249 | + idx = 0; |
250 | + print_delimiter = false; |
251 | + buflen = 0; |
252 | + bufpos = buf; |
253 | + memset (&state, '\0', sizeof(mbstate_t)); |
254 | + |
255 | + while (1) |
256 | + { |
257 | + REFILL_BUFFER (buf, bufpos, buflen, stream); |
258 | + |
259 | + GET_NEXT_WC_FROM_BUFFER (wc, bufpos, buflen, mblength, state, convfail); |
260 | + |
261 | + if (wc == WEOF) |
262 | + { |
263 | + if (idx > 0) |
264 | + putchar ('\n'); |
265 | + break; |
266 | + } |
267 | + else if (wc == L'\n') |
268 | + { |
269 | + putchar ('\n'); |
270 | + idx = 0; |
271 | + print_delimiter = false; |
272 | + } |
273 | + else |
274 | + { |
275 | + bool range_start; |
276 | + bool *rs = output_delimiter_specified ? &range_start : NULL; |
277 | + |
278 | + idx += (operating_mode == byte_mode) ? mblength : 1; |
279 | + if (print_kth (idx, rs)) |
280 | + { |
281 | + if (rs && *rs && print_delimiter) |
282 | + { |
283 | + fwrite (output_delimiter_string, sizeof (char), |
284 | + output_delimiter_length, stdout); |
285 | + } |
286 | + print_delimiter = true; |
287 | + fwrite (bufpos, mblength, sizeof(char), stdout); |
288 | + } |
289 | + } |
290 | + |
291 | + buflen -= mblength; |
292 | + bufpos += mblength; |
293 | + } |
294 | +} |
295 | +#endif |
296 | + |
297 | /* Read from stream STREAM, printing to standard output any selected fields. */ |
298 | |
299 | static void |
300 | @@ -692,13 +844,190 @@ cut_fields (FILE *stream) |
301 | } |
302 | } |
303 | |
304 | +#if HAVE_MBRTOWC |
305 | +static void |
306 | +cut_fields_mb (FILE *stream) |
307 | +{ |
308 | + int c; |
309 | + size_t field_idx = 1; |
310 | + bool found_any_selected_field = false; |
311 | + bool buffer_first_field; |
312 | + int empty_input; |
313 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
314 | + char *bufpos; /* Next read position of BUF. */ |
315 | + size_t buflen; /* The length of the byte sequence in buf. */ |
316 | + wint_t wc = 0; /* A gotten wide character. */ |
317 | + size_t mblength; /* The byte size of a multibyte character which shows |
318 | + as same character as WC. */ |
319 | + mbstate_t state; /* State of the stream. */ |
320 | + int convfail; /* 1, when conversion is failed. Otherwise 0. */ |
321 | + |
322 | + bufpos = buf; |
323 | + buflen = 0; |
324 | + memset (&state, '\0', sizeof(mbstate_t)); |
325 | + |
326 | + c = getc (stream); |
327 | + empty_input = (c == EOF); |
328 | + if (c != EOF) |
329 | + ungetc (c, stream); |
330 | + else |
331 | + wc = WEOF; |
332 | + |
333 | + /* To support the semantics of the -s flag, we may have to buffer |
334 | + all of the first field to determine whether it is `delimited.' |
335 | + But that is unnecessary if all non-delimited lines must be printed |
336 | + and the first field has been selected, or if non-delimited lines |
337 | + must be suppressed and the first field has *not* been selected. |
338 | + That is because a non-delimited line has exactly one field. */ |
339 | + buffer_first_field = (suppress_non_delimited ^ !print_kth (1, NULL)); |
340 | + |
341 | + while (1) |
342 | + { |
343 | + if (field_idx == 1 && buffer_first_field) |
344 | + { |
345 | + size_t n_bytes = 0; |
346 | + |
347 | + while (1) |
348 | + { |
349 | + REFILL_BUFFER (buf, bufpos, buflen, stream); |
350 | + |
351 | + GET_NEXT_WC_FROM_BUFFER |
352 | + (wc, bufpos, buflen, mblength, state, convfail); |
353 | + |
354 | + if (wc == WEOF) |
355 | + break; |
356 | + |
357 | + field_1_buffer = xrealloc (field_1_buffer, n_bytes + mblength); |
358 | + memcpy (field_1_buffer + n_bytes, bufpos, mblength); |
359 | + n_bytes += mblength; |
360 | + buflen -= mblength; |
361 | + bufpos += mblength; |
362 | + |
363 | + if (!convfail && (wc == L'\n' || wc == wcdelim)) |
364 | + break; |
365 | + } |
366 | + |
367 | + if (wc == WEOF) |
368 | + break; |
369 | + |
370 | + /* If the first field extends to the end of line (it is not |
371 | + delimited) and we are printing all non-delimited lines, |
372 | + print this one. */ |
373 | + if (convfail || (!convfail && wc != wcdelim)) |
374 | + { |
375 | + if (suppress_non_delimited) |
376 | + { |
377 | + /* Empty. */ |
378 | + } |
379 | + else |
380 | + { |
381 | + fwrite (field_1_buffer, sizeof (char), n_bytes, stdout); |
382 | + /* Make sure the output line is newline terminated. */ |
383 | + if (convfail || (!convfail && wc != L'\n')) |
384 | + putchar ('\n'); |
385 | + } |
386 | + continue; |
387 | + } |
388 | + |
389 | + if (print_kth (1, NULL)) |
390 | + { |
391 | + /* Print the field, but not the trailing delimiter. */ |
392 | + fwrite (field_1_buffer, sizeof (char), n_bytes - 1, stdout); |
393 | + found_any_selected_field = true; |
394 | + } |
395 | + ++field_idx; |
396 | + } |
397 | + |
398 | + if (wc != WEOF) |
399 | + { |
400 | + if (print_kth (field_idx, NULL)) |
401 | + { |
402 | + if (found_any_selected_field) |
403 | + { |
404 | + fwrite (output_delimiter_string, sizeof (char), |
405 | + output_delimiter_length, stdout); |
406 | + } |
407 | + found_any_selected_field = true; |
408 | + } |
409 | + |
410 | + while (1) |
411 | + { |
412 | + REFILL_BUFFER (buf, bufpos, buflen, stream); |
413 | + |
414 | + GET_NEXT_WC_FROM_BUFFER |
415 | + (wc, bufpos, buflen, mblength, state, convfail); |
416 | + |
417 | + if (wc == WEOF) |
418 | + break; |
419 | + else if (!convfail && (wc == wcdelim || wc == L'\n')) |
420 | + { |
421 | + buflen -= mblength; |
422 | + bufpos += mblength; |
423 | + break; |
424 | + } |
425 | + |
426 | + if (print_kth (field_idx, NULL)) |
427 | + fwrite (bufpos, mblength, sizeof(char), stdout); |
428 | + |
429 | + buflen -= mblength; |
430 | + bufpos += mblength; |
431 | + } |
432 | + } |
433 | + |
434 | + if ((!convfail || wc == L'\n') && buflen < 1) |
435 | + wc = WEOF; |
436 | + |
437 | + if (!convfail && wc == wcdelim) |
438 | + ++field_idx; |
439 | + else if (wc == WEOF || (!convfail && wc == L'\n')) |
440 | + { |
441 | + if (found_any_selected_field |
442 | + || (!empty_input && !(suppress_non_delimited && field_idx == 1))) |
443 | + putchar ('\n'); |
444 | + if (wc == WEOF) |
445 | + break; |
446 | + field_idx = 1; |
447 | + found_any_selected_field = false; |
448 | + } |
449 | + } |
450 | +} |
451 | +#endif |
452 | + |
453 | static void |
454 | cut_stream (FILE *stream) |
455 | { |
456 | - if (operating_mode == byte_mode) |
457 | - cut_bytes (stream); |
458 | +#if HAVE_MBRTOWC |
459 | + if (MB_CUR_MAX > 1 && !force_singlebyte_mode) |
460 | + { |
461 | + switch (operating_mode) |
462 | + { |
463 | + case byte_mode: |
464 | + if (byte_mode_character_aware) |
465 | + cut_characters_or_cut_bytes_no_split (stream); |
466 | + else |
467 | + cut_bytes (stream); |
468 | + break; |
469 | + |
470 | + case character_mode: |
471 | + cut_characters_or_cut_bytes_no_split (stream); |
472 | + break; |
473 | + |
474 | + case field_mode: |
475 | + cut_fields_mb (stream); |
476 | + break; |
477 | + |
478 | + default: |
479 | + abort (); |
480 | + } |
481 | + } |
482 | else |
483 | - cut_fields (stream); |
484 | +#endif |
485 | + { |
486 | + if (operating_mode == field_mode) |
487 | + cut_fields (stream); |
488 | + else |
489 | + cut_bytes (stream); |
490 | + } |
491 | } |
492 | |
493 | /* Process file FILE to standard output. |
494 | @@ -748,6 +1077,8 @@ main (int argc, char **argv) |
495 | bool ok; |
496 | bool delim_specified = false; |
497 | char *spec_list_string IF_LINT(= NULL); |
498 | + char mbdelim[MB_LEN_MAX + 1]; |
499 | + size_t delimlen = 0; |
500 | |
501 | initialize_main (&argc, &argv); |
502 | program_name = argv[0]; |
503 | @@ -770,7 +1101,6 @@ main (int argc, char **argv) |
504 | switch (optc) |
505 | { |
506 | case 'b': |
507 | - case 'c': |
508 | /* Build the byte list. */ |
509 | if (operating_mode != undefined_mode) |
510 | FATAL_ERROR (_("only one type of list may be specified")); |
511 | @@ -778,6 +1108,14 @@ main (int argc, char **argv) |
512 | spec_list_string = optarg; |
513 | break; |
514 | |
515 | + case 'c': |
516 | + /* Build the character list. */ |
517 | + if (operating_mode != undefined_mode) |
518 | + FATAL_ERROR (_("only one type of list may be specified")); |
519 | + operating_mode = character_mode; |
520 | + spec_list_string = optarg; |
521 | + break; |
522 | + |
523 | case 'f': |
524 | /* Build the field list. */ |
525 | if (operating_mode != undefined_mode) |
526 | @@ -789,9 +1127,32 @@ main (int argc, char **argv) |
527 | case 'd': |
528 | /* New delimiter. */ |
529 | /* Interpret -d '' to mean `use the NUL byte as the delimiter.' */ |
530 | - if (optarg[0] != '\0' && optarg[1] != '\0') |
531 | - FATAL_ERROR (_("the delimiter must be a single character")); |
532 | - delim = optarg[0]; |
533 | +#if HAVE_MBRTOWC |
534 | + if(MB_CUR_MAX > 1) |
535 | + { |
536 | + mbstate_t state; |
537 | + |
538 | + memset (&state, '\0', sizeof(mbstate_t)); |
539 | + delimlen = mbrtowc (&wcdelim, optarg, MB_LEN_MAX, &state); |
540 | + |
541 | + if (delimlen == (size_t)-1 || delimlen == (size_t)-2) |
542 | + force_singlebyte_mode = true; |
543 | + else |
544 | + { |
545 | + delimlen = (delimlen < 1) ? 1 : delimlen; |
546 | + if (wcdelim != L'\0' && *(optarg + delimlen) != '\0') |
547 | + FATAL_ERROR (_("the delimiter must be a single character")); |
548 | + memcpy (mbdelim, optarg, delimlen); |
549 | + } |
550 | + } |
551 | + |
552 | + if (MB_CUR_MAX <= 1 || force_singlebyte_mode) |
553 | +#endif |
554 | + { |
555 | + if (optarg[0] != '\0' && optarg[1] != '\0') |
556 | + FATAL_ERROR (_("the delimiter must be a single character")); |
557 | + delim = (unsigned char) optarg[0]; |
558 | + } |
559 | delim_specified = true; |
560 | break; |
561 | |
562 | @@ -805,6 +1166,7 @@ main (int argc, char **argv) |
563 | break; |
564 | |
565 | case 'n': |
566 | + byte_mode_character_aware = true; |
567 | break; |
568 | |
569 | case 's': |
570 | @@ -827,7 +1189,7 @@ main (int argc, char **argv) |
571 | if (operating_mode == undefined_mode) |
572 | FATAL_ERROR (_("you must specify a list of bytes, characters, or fields")); |
573 | |
574 | - if (delim != '\0' && operating_mode != field_mode) |
575 | + if (delim_specified && operating_mode != field_mode) |
576 | FATAL_ERROR (_("an input delimiter may be specified only\ |
577 | when operating on fields")); |
578 | |
579 | @@ -854,15 +1216,34 @@ main (int argc, char **argv) |
580 | } |
581 | |
582 | if (!delim_specified) |
583 | - delim = '\t'; |
584 | + { |
585 | + delim = '\t'; |
586 | +#ifdef HAVE_MBRTOWC |
587 | + wcdelim = L'\t'; |
588 | + mbdelim[0] = '\t'; |
589 | + mbdelim[1] = '\0'; |
590 | + delimlen = 1; |
591 | + } |
592 | +#endif |
593 | |
594 | if (output_delimiter_string == NULL) |
595 | { |
596 | - static char dummy[2]; |
597 | - dummy[0] = delim; |
598 | - dummy[1] = '\0'; |
599 | - output_delimiter_string = dummy; |
600 | - output_delimiter_length = 1; |
601 | +#ifdef HAVE_MBRTOWC |
602 | + if (MB_CUR_MAX > 1 && !force_singlebyte_mode) |
603 | + { |
604 | + output_delimiter_string = xstrdup(mbdelim); |
605 | + output_delimiter_length = delimlen; |
606 | + } |
607 | + |
608 | + if (MB_CUR_MAX <= 1 || force_singlebyte_mode) |
609 | +#endif |
610 | + { |
611 | + static char dummy[2]; |
612 | + dummy[0] = delim; |
613 | + dummy[1] = '\0'; |
614 | + output_delimiter_string = dummy; |
615 | + output_delimiter_length = 1; |
616 | + } |
617 | } |
618 | |
619 | if (optind == argc) |
620 | --- coreutils/src/expand.c |
621 | +++ coreutils/src/expand.c |
622 | @@ -38,12 +38,32 @@ |
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 | +#if HAVE_WCTYPE_H |
632 | +# include <wctype.h> |
633 | +#endif |
634 | + |
635 | #include "system.h" |
636 | #include "error.h" |
637 | #include "posixver.h" |
638 | #include "quote.h" |
639 | #include "xstrndup.h" |
640 | |
641 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
642 | + installation; work around this configuration error. */ |
643 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
644 | +# define MB_LEN_MAX 16 |
645 | +#endif |
646 | + |
647 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
648 | +#if HAVE_MBRTOWC && defined mbstate_t |
649 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
650 | +#endif |
651 | + |
652 | /* The official name of this program (e.g., no `g' prefix). */ |
653 | #define PROGRAM_NAME "expand" |
654 | |
655 | @@ -352,8 +372,9 @@ expand (void) |
656 | } |
657 | else |
658 | { |
659 | - column++; |
660 | + if (!ISCNTRL (c)) |
661 | + column++; |
662 | - if (!column) |
663 | + if (column >= UINTMAX_MAX) |
664 | error (EXIT_FAILURE, 0, _("input line is too long")); |
665 | } |
666 | |
667 | @@ -370,6 +391,163 @@ expand (void) |
668 | } |
669 | } |
670 | |
671 | +#if HAVE_MBRTOWC && HAVE_WCTYPE_H |
672 | +static void |
673 | +expand_multibyte (void) |
674 | +{ |
675 | + /* Input stream. */ |
676 | + FILE *fp = next_file (NULL); |
677 | + |
678 | + mbstate_t i_state; /* Current shift state of the input stream. */ |
679 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
680 | + char *bufpos; /* Next read position of BUF. */ |
681 | + size_t buflen = 0; /* The length of the byte sequence in buf. */ |
682 | + |
683 | + if (!fp) |
684 | + return; |
685 | + |
686 | + /* Binary I/O will preserve the original EOL style (DOS/Unix) of files. */ |
687 | + SET_BINARY2 (fileno (fp), STDOUT_FILENO); |
688 | + |
689 | + for (;;) |
690 | + { |
691 | + /* Input character, or EOF. */ |
692 | + wint_t wc; |
693 | + |
694 | + /* If true, perform translations. */ |
695 | + bool convert = true; |
696 | + |
697 | + |
698 | + /* The following variables have valid values only when CONVERT |
699 | + is true: */ |
700 | + |
701 | + /* Column of next input character. */ |
702 | + uintmax_t column = 0; |
703 | + |
704 | + /* Index in TAB_LIST of next tab stop to examine. */ |
705 | + size_t tab_index = 0; |
706 | + |
707 | + |
708 | + /* Convert a line of text. */ |
709 | + |
710 | + do |
711 | + { |
712 | + wchar_t w; |
713 | + size_t mblength; /* The byte size of a multibyte character |
714 | + which shows as same character as WC. */ |
715 | + mbstate_t i_state_bak; /* Back up the I_STATE. */ |
716 | + |
717 | + /* Fill buffer */ |
718 | + if (buflen < MB_LEN_MAX) |
719 | + { |
720 | + if (!feof(fp) && !ferror(fp)) { |
721 | + if (buflen > 0) memmove(buf, bufpos, buflen); |
722 | + buflen += fread(buf + buflen, sizeof(char), BUFSIZ, fp); |
723 | + bufpos = buf; |
724 | + } |
725 | + } |
726 | + |
727 | + if (buflen < 1) { |
728 | + /* Move to the next file */ |
729 | + if (feof(fp) || ferror(fp)) { |
730 | + fp = next_file(fp); |
731 | + } |
732 | + if (!fp) |
733 | + return; |
734 | + memset (&i_state, '\0', sizeof (mbstate_t)); |
735 | + SET_BINARY2 (fileno (fp), STDOUT_FILENO); |
736 | + continue; |
737 | + } |
738 | + |
739 | + i_state_bak = i_state; |
740 | + mblength = mbrtowc (&w, bufpos, buflen, &i_state); |
741 | + wc = w; |
742 | + |
743 | + if (mblength == (size_t) -1 || mblength == (size_t) -2) { |
744 | + i_state = i_state_bak; |
745 | + wc = L'\0'; |
746 | + column += convert; |
747 | + mblength = 1; |
748 | + } |
749 | + |
750 | + if (convert) |
751 | + { |
752 | + if (wc == L'\t') |
753 | + { |
754 | + /* Column the next input tab stop is on. */ |
755 | + uintmax_t next_tab_column; |
756 | + |
757 | + if (tab_size) |
758 | + next_tab_column = column + (tab_size - column % tab_size); |
759 | + else |
760 | + for (;;) |
761 | + if (tab_index == first_free_tab) |
762 | + { |
763 | + next_tab_column = column + 1; |
764 | + break; |
765 | + } |
766 | + else |
767 | + { |
768 | + uintmax_t tab = tab_list[tab_index++]; |
769 | + if (column < tab) |
770 | + { |
771 | + next_tab_column = tab; |
772 | + break; |
773 | + } |
774 | + } |
775 | + |
776 | + if (next_tab_column < column) |
777 | + error (EXIT_FAILURE, 0, _("input line is too long")); |
778 | + |
779 | + while (++column < next_tab_column) |
780 | + if (putchar (' ') < 0) |
781 | + error (EXIT_FAILURE, errno, _("write error")); |
782 | + |
783 | + *bufpos = ' '; |
784 | + } |
785 | + else if (wc == L'\b') |
786 | + { |
787 | + /* Go back one column, and force recalculation of the |
788 | + next tab stop. */ |
789 | + column -= !!column; |
790 | + tab_index -= !!tab_index; |
791 | + } |
792 | + else |
793 | + { |
794 | + if (!iswcntrl (wc)) |
795 | + { |
796 | + int width = wcwidth (wc); |
797 | + if (width > 0) { |
798 | + if (column > (column + width)) |
799 | + error (EXIT_FAILURE, 0, _("input line is too long")); |
800 | + column += width; |
801 | + } |
802 | + } |
803 | + } |
804 | + |
805 | + convert &= convert_entire_line | iswblank (wc); |
806 | + } |
807 | + |
808 | + if (mblength) |
809 | + { |
810 | + if (fwrite (bufpos, sizeof(char), mblength, stdout) < mblength) |
811 | + error (EXIT_FAILURE, errno, _("write error")); |
812 | + } |
813 | + else |
814 | + { |
815 | + if (putchar('\0')) |
816 | + error (EXIT_FAILURE, errno, _("write error")); |
817 | + mblength = 1; |
818 | + } |
819 | + |
820 | + buflen -= mblength; |
821 | + bufpos += mblength; |
822 | + } |
823 | + while (wc != L'\n'); |
824 | + } |
825 | +} |
826 | +#endif |
827 | + |
828 | int |
829 | main (int argc, char **argv) |
830 | { |
831 | @@ -446,7 +624,12 @@ main (int argc, char **argv) |
832 | |
833 | file_list = (optind < argc ? &argv[optind] : stdin_argv); |
834 | |
835 | - expand (); |
836 | +#if HAVE_MBRTOWC |
837 | + if (MB_CUR_MAX > 1) |
838 | + expand_multibyte (); |
839 | + else |
840 | +#endif |
841 | + expand (); |
842 | |
843 | if (have_read_stdin && fclose (stdin) != 0) |
844 | error (EXIT_FAILURE, errno, "-"); |
845 | --- coreutils/src/fold.c |
846 | +++ coreutils/src/fold.c |
847 | @@ -23,6 +23,19 @@ |
848 | #include <getopt.h> |
849 | #include <sys/types.h> |
850 | |
851 | +/* Get MB_CUR_MAX. */ |
852 | +#include <stdlib.h> |
853 | + |
854 | +/* Get mbrtowc, mbstate_t, wcwidth(). */ |
855 | +#if HAVE_WCHAR_H |
856 | +# include <wchar.h> |
857 | +#endif |
858 | + |
859 | +/* Get iswprint(), iswctype(), wctype(). */ |
860 | +#if HAVE_WCTYPE_H |
861 | +# include <wctype.h> |
862 | +#endif |
863 | + |
864 | #include "system.h" |
865 | #include "error.h" |
866 | #include "posixver.h" |
867 | @@ -30,14 +43,57 @@ |
868 | |
869 | #define TAB_WIDTH 8 |
870 | |
871 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
872 | + installation; work around this configuration error. */ |
873 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
874 | +# undef MB_LEN_MAX |
875 | +# define MB_LEN_MAX 16 |
876 | +#endif |
877 | + |
878 | +#ifndef HAVE_DECL_WCWIDTH |
879 | +"this configure-time declaration test was not run" |
880 | +#endif |
881 | +#if !HAVE_DECL_WCWIDTH |
882 | +extern int wcwidth (); |
883 | +#endif |
884 | + |
885 | +/* If wcwidth() doesn't exist, assume all printable characters have |
886 | + width 1. */ |
887 | +#if !defined wcwidth && !HAVE_WCWIDTH |
888 | +# define wcwidth(wc) ((wc) == 0 ? 0 : iswprint (wc) ? 1 : -1) |
889 | +#endif |
890 | + |
891 | /* The official name of this program (e.g., no `g' prefix). */ |
892 | #define PROGRAM_NAME "fold" |
893 | |
894 | #define AUTHORS "David MacKenzie" |
895 | |
896 | +#define FATAL_ERROR(Message) \ |
897 | +do \ |
898 | +{ \ |
899 | + error (0, 0, (Message)); \ |
900 | + usage (2); \ |
901 | +} \ |
902 | +while (0) |
903 | + |
904 | +enum operating_mode |
905 | +{ |
906 | + /* Fold texts by columns that are at the given positions. */ |
907 | + column_mode, |
908 | + |
909 | + /* Fold texts by bytes that are at the given positions. */ |
910 | + byte_mode, |
911 | + |
912 | + /* Fold texts by characters that are at the given positions. */ |
913 | + character_mode, |
914 | +}; |
915 | + |
916 | /* The name this program was run with. */ |
917 | char *program_name; |
918 | |
919 | +/* The argument shows current mode. (Default: column_mode) */ |
920 | +static enum operating_mode operating_mode; |
921 | + |
922 | /* If nonzero, try to break on whitespace. */ |
923 | static bool break_spaces; |
924 | |
925 | @@ -47,9 +103,15 @@ static bool count_bytes; |
926 | /* If nonzero, at least one of the files we read was standard input. */ |
927 | static bool have_read_stdin; |
928 | |
929 | +/* wide character class `blank' */ |
930 | +#if HAVE_MBRTOWC |
931 | +wctype_t blank_type; |
932 | +#endif |
933 | + |
934 | static struct option const longopts[] = |
935 | { |
936 | {"bytes", no_argument, NULL, 'b'}, |
937 | + {"characters", no_argument, NULL, 'c'}, |
938 | {"spaces", no_argument, NULL, 's'}, |
939 | {"width", required_argument, NULL, 'w'}, |
940 | {GETOPT_HELP_OPTION_DECL}, |
941 | @@ -79,6 +141,7 @@ Mandatory arguments to long options are |
942 | "), stdout); |
943 | fputs (_("\ |
944 | -b, --bytes count bytes rather than columns\n\ |
945 | + -c, --characters count characters rather than columns\n\ |
946 | -s, --spaces break at spaces\n\ |
947 | -w, --width=WIDTH use WIDTH columns instead of 80\n\ |
948 | "), stdout); |
949 | @@ -96,7 +159,7 @@ Mandatory arguments to long options are |
950 | static size_t |
951 | adjust_column (size_t column, char c) |
952 | { |
953 | - if (!count_bytes) |
954 | + if (operating_mode != byte_mode) |
955 | { |
956 | if (c == '\b') |
957 | { |
958 | @@ -115,14 +178,9 @@ adjust_column (size_t column, char c) |
959 | return column; |
960 | } |
961 | |
962 | -/* Fold file FILENAME, or standard input if FILENAME is "-", |
963 | - to stdout, with maximum line length WIDTH. |
964 | - Return true if successful. */ |
965 | - |
966 | -static bool |
967 | -fold_file (char *filename, size_t width) |
968 | +static int |
969 | +fold_text (FILE *istream, int width) |
970 | { |
971 | - FILE *istream; |
972 | register int c; |
973 | size_t column = 0; /* Screen column where next char will go. */ |
974 | size_t offset_out = 0; /* Index in `line_out' for next char. */ |
975 | @@ -130,20 +188,6 @@ fold_file (char *filename, size_t width) |
976 | static size_t allocated_out = 0; |
977 | int saved_errno; |
978 | |
979 | - if (STREQ (filename, "-")) |
980 | - { |
981 | - istream = stdin; |
982 | - have_read_stdin = true; |
983 | - } |
984 | - else |
985 | - istream = fopen (filename, "r"); |
986 | - |
987 | - if (istream == NULL) |
988 | - { |
989 | - error (0, errno, "%s", filename); |
990 | - return false; |
991 | - } |
992 | - |
993 | while ((c = getc (istream)) != EOF) |
994 | { |
995 | if (offset_out + 1 >= allocated_out) |
996 | @@ -221,6 +265,233 @@ fold_file (char *filename, size_t width) |
997 | if (offset_out) |
998 | fwrite (line_out, sizeof (char), (size_t) offset_out, stdout); |
999 | |
1000 | + return saved_errno; |
1001 | +} |
1002 | + |
1003 | +#if HAVE_MBRTOWC |
1004 | +static void |
1005 | +fold_multibyte_text (FILE *istream, int width) |
1006 | +{ |
1007 | + int i; |
1008 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
1009 | + size_t buflen; /* The length of the byte sequence in buf. */ |
1010 | + char *bufpos; /* Next read position of BUF. */ |
1011 | + wint_t wc; /* A gotten wide character. */ |
1012 | + wchar_t tmp; |
1013 | + size_t mblength; /* The byte size of a multibyte character which shows |
1014 | + as same character as WC. */ |
1015 | + mbstate_t state, state_bak; /* State of the stream. */ |
1016 | + int convfail; /* 1, when conversion is failed. Otherwise 0. */ |
1017 | + |
1018 | + char *line_out = NULL; |
1019 | + size_t offset_out = 0; /* Index in `line_out' for next char. */ |
1020 | + size_t allocated_out = 1024; |
1021 | + |
1022 | + int increment; |
1023 | + size_t column = 0; |
1024 | + |
1025 | + size_t last_blank_pos; |
1026 | + size_t last_blank_column; |
1027 | + int is_blank_seen; |
1028 | + int last_blank_increment; |
1029 | + int is_bs_following_last_blank; |
1030 | + size_t bs_following_last_blank_num; |
1031 | + int is_cr_after_last_blank; |
1032 | + |
1033 | + |
1034 | +#define CLEAR_FLAGS \ |
1035 | + do \ |
1036 | + { \ |
1037 | + last_blank_pos = 0; \ |
1038 | + last_blank_column = 0; \ |
1039 | + is_blank_seen = 0; \ |
1040 | + is_bs_following_last_blank = 0; \ |
1041 | + bs_following_last_blank_num = 0; \ |
1042 | + is_cr_after_last_blank = 0; \ |
1043 | + } \ |
1044 | + while (0) |
1045 | + |
1046 | +#define START_NEW_LINE \ |
1047 | + do \ |
1048 | + { \ |
1049 | + putchar ('\n'); \ |
1050 | + column = 0; \ |
1051 | + offset_out = 0; \ |
1052 | + CLEAR_FLAGS; \ |
1053 | + } \ |
1054 | + while (0) |
1055 | + |
1056 | + CLEAR_FLAGS; |
1057 | + |
1058 | + memset (&state, '\0', sizeof(mbstate_t)); |
1059 | + line_out = xmalloc (allocated_out); |
1060 | + |
1061 | + buflen = fread (buf, sizeof(char), BUFSIZ, istream); |
1062 | + bufpos = buf; |
1063 | + |
1064 | + for (;; bufpos += mblength, buflen -= mblength) |
1065 | + { |
1066 | + if (buflen < MB_LEN_MAX && !feof (istream) && !ferror (istream)) |
1067 | + { |
1068 | + memmove (buf, bufpos, buflen); |
1069 | + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, istream); |
1070 | + bufpos = buf; |
1071 | + } |
1072 | + |
1073 | + if (buflen < 1) |
1074 | + break; |
1075 | + |
1076 | + /* Get a wide character. */ |
1077 | + convfail = 0; |
1078 | + state_bak = state; |
1079 | + mblength = mbrtowc (&tmp, bufpos, buflen, &state); |
1080 | + wc = tmp; |
1081 | + |
1082 | + switch (mblength) |
1083 | + { |
1084 | + case (size_t)-1: |
1085 | + case (size_t)-2: |
1086 | + convfail++; |
1087 | + state = state_bak; |
1088 | + /* Fall through. */ |
1089 | + |
1090 | + case 0: |
1091 | + mblength = 1; |
1092 | + break; |
1093 | + } |
1094 | + |
1095 | + if (!convfail && wc == L'\n') |
1096 | + { |
1097 | + if (offset_out > 0) |
1098 | + { |
1099 | + fwrite (line_out, sizeof(char), offset_out, stdout); |
1100 | + START_NEW_LINE; |
1101 | + } |
1102 | + continue; |
1103 | + } |
1104 | + |
1105 | +rescan: |
1106 | + if (operating_mode == byte_mode) /* byte mode */ |
1107 | + increment = mblength; |
1108 | + else if (operating_mode == character_mode) /* character mode */ |
1109 | + increment = 1; |
1110 | + else /* column mode */ |
1111 | + { |
1112 | + if (convfail) |
1113 | + increment = 1; |
1114 | + else |
1115 | + { |
1116 | + switch (wc) |
1117 | + { |
1118 | + case L'\b': |
1119 | + increment = (column > 0) ? -1 : 0; |
1120 | + break; |
1121 | + |
1122 | + case L'\r': |
1123 | + increment = -1 * column; |
1124 | + break; |
1125 | + |
1126 | + case L'\t': |
1127 | + increment = 8 - column % 8; |
1128 | + break; |
1129 | + |
1130 | + default: |
1131 | + increment = wcwidth (wc); |
1132 | + increment = (increment < 0) ? 0 : increment; |
1133 | + } |
1134 | + } |
1135 | + } |
1136 | + |
1137 | + if (column + increment > width && break_spaces && last_blank_pos) |
1138 | + { |
1139 | + fwrite (line_out, sizeof(char), last_blank_pos, stdout); |
1140 | + putchar ('\n'); |
1141 | + |
1142 | + offset_out = offset_out - last_blank_pos; |
1143 | + column = column - last_blank_column + ((is_cr_after_last_blank) |
1144 | + ? last_blank_increment : bs_following_last_blank_num); |
1145 | + memmove (line_out, line_out + last_blank_pos, offset_out); |
1146 | + CLEAR_FLAGS; |
1147 | + goto rescan; |
1148 | + } |
1149 | + |
1150 | + if (column + increment > width && column != 0) |
1151 | + { |
1152 | + fwrite (line_out, sizeof(char), offset_out, stdout); |
1153 | + START_NEW_LINE; |
1154 | + goto rescan; |
1155 | + } |
1156 | + |
1157 | + if (allocated_out < offset_out + mblength) |
1158 | + line_out = x2nrealloc (line_out, &allocated_out, sizeof *line_out); |
1159 | + |
1160 | + for (i = 0; i < mblength; i++) |
1161 | + { |
1162 | + *(line_out + offset_out) = *(bufpos + i); |
1163 | + ++offset_out; |
1164 | + } |
1165 | + |
1166 | + column += increment; |
1167 | + |
1168 | + if (is_blank_seen && !convfail && wc == L'\r') |
1169 | + is_cr_after_last_blank = 1; |
1170 | + |
1171 | + if (is_bs_following_last_blank && !convfail && wc == L'\b') |
1172 | + ++bs_following_last_blank_num; |
1173 | + else |
1174 | + is_bs_following_last_blank = 0; |
1175 | + |
1176 | + if (break_spaces && !convfail && iswctype (wc, blank_type)) |
1177 | + { |
1178 | + last_blank_pos = offset_out; |
1179 | + last_blank_column = column; |
1180 | + is_blank_seen = 1; |
1181 | + last_blank_increment = increment; |
1182 | + is_bs_following_last_blank = 1; |
1183 | + bs_following_last_blank_num = 0; |
1184 | + is_cr_after_last_blank = 0; |
1185 | + } |
1186 | + } |
1187 | + |
1188 | + if (offset_out) |
1189 | + fwrite (line_out, sizeof (char), (size_t) offset_out, stdout); |
1190 | + |
1191 | + free(line_out); |
1192 | +} |
1193 | +#endif |
1194 | + |
1195 | +/* Fold file FILENAME, or standard input if FILENAME is "-", |
1196 | + to stdout, with maximum line length WIDTH. |
1197 | + Return true if successful. */ |
1198 | + |
1199 | +static bool |
1200 | +fold_file (char *filename, int width) |
1201 | +{ |
1202 | + FILE *istream; |
1203 | + int saved_errno; |
1204 | + |
1205 | + if (STREQ (filename, "-")) |
1206 | + { |
1207 | + istream = stdin; |
1208 | + have_read_stdin = true; |
1209 | + } |
1210 | + else |
1211 | + istream = fopen (filename, "r"); |
1212 | + |
1213 | + if (istream == NULL) |
1214 | + { |
1215 | + error (0, errno, "%s", filename); |
1216 | + return false; |
1217 | + } |
1218 | + |
1219 | + /* Define how ISTREAM is being folded. */ |
1220 | +#if HAVE_MBRTOWC |
1221 | + if (MB_CUR_MAX > 1) |
1222 | + fold_multibyte_text (istream, width); |
1223 | + else |
1224 | +#endif |
1225 | + saved_errno = fold_text (istream, width); |
1226 | + |
1227 | if (ferror (istream)) |
1228 | { |
1229 | error (0, saved_errno, "%s", filename); |
1230 | @@ -253,6 +524,10 @@ main (int argc, char **argv) |
1231 | |
1232 | atexit (close_stdout); |
1233 | |
1234 | +#if HAVE_MBRTOWC |
1235 | + blank_type = wctype ("blank"); |
1236 | +#endif |
1237 | + operating_mode = column_mode; |
1238 | break_spaces = count_bytes = have_read_stdin = false; |
1239 | |
1240 | /* Turn any numeric options into -w options. */ |
1241 | @@ -280,12 +555,23 @@ main (int argc, char **argv) |
1242 | } |
1243 | } |
1244 | |
1245 | - while ((optc = getopt_long (argc, argv, "bsw:", longopts, NULL)) != -1) |
1246 | + while ((optc = getopt_long (argc, argv, "bcsw:", longopts, NULL)) != -1) |
1247 | { |
1248 | switch (optc) |
1249 | { |
1250 | + case 0: |
1251 | + break; |
1252 | + |
1253 | case 'b': /* Count bytes rather than columns. */ |
1254 | - count_bytes = true; |
1255 | + if (operating_mode != column_mode) |
1256 | + FATAL_ERROR (_("only one way of folding may be specified")); |
1257 | + operating_mode = byte_mode; |
1258 | + break; |
1259 | + |
1260 | + case 'c': /* Count characters rather than columns. */ |
1261 | + if (operating_mode != column_mode) |
1262 | + FATAL_ERROR (_("only one way of folding may be specified")); |
1263 | + operating_mode = character_mode; |
1264 | break; |
1265 | |
1266 | case 's': /* Break at word boundaries. */ |
1267 | --- coreutils/src/join.c |
1268 | +++ coreutils/src/join.c |
1269 | @@ -24,6 +24,16 @@ |
1270 | #include <sys/types.h> |
1271 | #include <getopt.h> |
1272 | |
1273 | +/* Get mbstate_t, mbrtowc, mbrtowc, wcwidth. */ |
1274 | +#if HAVE_WCHAR_H |
1275 | +# include <wchar.h> |
1276 | +#endif |
1277 | + |
1278 | +/* Get iswblank, towupper. */ |
1279 | +#if HAVE_WCTYPE_H |
1280 | +# include <wctype.h> |
1281 | +#endif |
1282 | + |
1283 | #include "system.h" |
1284 | #include "error.h" |
1285 | #include "hard-locale.h" |
1286 | @@ -34,6 +44,11 @@ |
1287 | #include "xmemcoll.h" |
1288 | #include "xstrtol.h" |
1289 | |
1290 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
1291 | +#if HAVE_MBRTOWC && defined mbstate_t |
1292 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
1293 | +#endif |
1294 | + |
1295 | /* The official name of this program (e.g., no `g' prefix). */ |
1296 | #define PROGRAM_NAME "join" |
1297 | |
1298 | @@ -110,7 +125,10 @@ static struct outlist *outlist_end = &ou |
1299 | /* Tab character separating fields; if this is NUL fields are separated |
1300 | by any nonempty string of white space, otherwise by exactly one |
1301 | tab character. */ |
1302 | -static char tab; |
1303 | +static char *tab = NULL; |
1304 | + |
1305 | +/* The number of bytes used for tab. */ |
1306 | +static size_t tablen = 0; |
1307 | |
1308 | /* When using getopt_long_only, no long option can start with |
1309 | a character that is a short option. */ |
1310 | @@ -233,7 +251,7 @@ xfields (struct line *line) |
1311 | |
1312 | if (tab) |
1313 | { |
1314 | - unsigned char t = tab; |
1315 | + unsigned char t = tab[0]; |
1316 | char *sep; |
1317 | for (; (sep = memchr (ptr, t, lim - ptr)) != NULL; ptr = sep + 1) |
1318 | extract_field (line, ptr, sep - ptr); |
1319 | @@ -262,6 +280,133 @@ xfields (struct line *line) |
1320 | extract_field (line, ptr, lim - ptr); |
1321 | } |
1322 | |
1323 | +#if HAVE_MBRTOWC |
1324 | +static void |
1325 | +xfields_multibyte (struct line *line) |
1326 | +{ |
1327 | + int i; |
1328 | + char *ptr0 = line->buf.buffer; |
1329 | + char *ptr; |
1330 | + char *lim; |
1331 | + wchar_t wc = 0; |
1332 | + size_t mblength; |
1333 | + mbstate_t state, state_bak; |
1334 | + |
1335 | + memset (&state, 0, sizeof (mbstate_t)); |
1336 | + |
1337 | + ptr = ptr0; |
1338 | + lim = ptr0 + line->buf.length - 1; |
1339 | + |
1340 | + if (tab == NULL) |
1341 | + { |
1342 | + /* Skip leading blanks before the first field. */ |
1343 | + while (ptr < lim) |
1344 | + { |
1345 | + state_bak = state; |
1346 | + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); |
1347 | + |
1348 | + if (mblength == (size_t) -1 || mblength == (size_t) -2) |
1349 | + { |
1350 | + mblength = 1; |
1351 | + state = state_bak; |
1352 | + break; |
1353 | + } |
1354 | + mblength = (mblength < 1) ? 1 : mblength; |
1355 | + |
1356 | + if (!iswblank (wc)) |
1357 | + break; |
1358 | + ptr += mblength; |
1359 | + } |
1360 | + } |
1361 | + |
1362 | + for (i = 0; ptr < lim; ++i) |
1363 | + { |
1364 | + if (tab != NULL) |
1365 | + { |
1366 | + char *beg = ptr; |
1367 | + while (ptr < lim) |
1368 | + { |
1369 | + state_bak = state; |
1370 | + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); |
1371 | + |
1372 | + if (mblength == (size_t) -1 || mblength == (size_t) -2) |
1373 | + { |
1374 | + mblength = 1; |
1375 | + state = state_bak; |
1376 | + } |
1377 | + mblength = (mblength < 1) ? 1 : mblength; |
1378 | + |
1379 | + if (mblength == tablen && !memcmp (ptr, tab, mblength)) |
1380 | + break; |
1381 | + else |
1382 | + { |
1383 | + ptr += mblength; |
1384 | + continue; |
1385 | + } |
1386 | + } |
1387 | + |
1388 | + extract_field (line, beg, ptr - beg); |
1389 | + if (ptr < lim) |
1390 | + ptr += mblength; |
1391 | + } |
1392 | + else |
1393 | + { |
1394 | + char *beg = ptr; |
1395 | + while (ptr < lim) |
1396 | + { |
1397 | + state_bak = state; |
1398 | + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); |
1399 | + |
1400 | + if (mblength == (size_t) -1 || mblength == (size_t) -2) |
1401 | + { |
1402 | + mblength = 1; |
1403 | + state = state_bak; |
1404 | + } |
1405 | + mblength = (mblength < 1) ? 1 : mblength; |
1406 | + |
1407 | + if (iswblank (wc)) |
1408 | + break; |
1409 | + else |
1410 | + { |
1411 | + ptr += mblength; |
1412 | + continue; |
1413 | + } |
1414 | + } |
1415 | + |
1416 | + extract_field (line, beg, ptr - beg); |
1417 | + if (ptr < lim) |
1418 | + ptr += mblength; |
1419 | + } |
1420 | + } |
1421 | + |
1422 | + if (ptr != ptr0) |
1423 | + { |
1424 | + mblength = mbrtowc (&wc, ptr - mblength, mblength, &state); |
1425 | + wc = (mbsinit (&state) && *(ptr - mblength) == '\0') ? L'\0' : wc; |
1426 | + if (tab != NULL) |
1427 | + { |
1428 | + if (mblength == (size_t) -1 || mblength == (size_t) -2) |
1429 | + mblength = 1; |
1430 | + |
1431 | + if (mblength == tablen && !memcmp (ptr - mblength, tab, mblength)) |
1432 | + /* Add one more (empty) field because the last character of |
1433 | + the line was a delimiter. */ |
1434 | + extract_field (line, NULL, 0); |
1435 | + } |
1436 | + else |
1437 | + { |
1438 | + if (mblength != (size_t) -1 && mblength != (size_t) -2) |
1439 | + { |
1440 | + if (iswblank (wc)) |
1441 | + /* Add one more (empty) field because the last character of |
1442 | + the line was a delimiter. */ |
1443 | + extract_field (line, NULL, 0); |
1444 | + } |
1445 | + } |
1446 | + } |
1447 | +} |
1448 | +#endif |
1449 | + |
1450 | /* Read a line from FP into LINE and split it into fields. |
1451 | Return true if successful. */ |
1452 | |
1453 | @@ -282,7 +427,13 @@ get_line (FILE *fp, struct line *line) |
1454 | line->nfields_allocated = 0; |
1455 | line->nfields = 0; |
1456 | line->fields = NULL; |
1457 | - xfields (line); |
1458 | + |
1459 | +#if HAVE_MBRTOWC |
1460 | + if (MB_CUR_MAX > 1) |
1461 | + xfields_multibyte (line); |
1462 | + else |
1463 | +#endif |
1464 | + xfields (line); |
1465 | return true; |
1466 | } |
1467 | |
1468 | @@ -336,56 +487,115 @@ static int |
1469 | keycmp (struct line const *line1, struct line const *line2) |
1470 | { |
1471 | /* Start of field to compare in each file. */ |
1472 | - char *beg1; |
1473 | - char *beg2; |
1474 | - |
1475 | - size_t len1; |
1476 | - size_t len2; /* Length of fields to compare. */ |
1477 | + char *beg[2]; |
1478 | + char *copy[2]; |
1479 | + size_t len[2]; /* Length of fields to compare. */ |
1480 | int diff; |
1481 | + int i, j; |
1482 | |
1483 | if (join_field_1 < line1->nfields) |
1484 | { |
1485 | - beg1 = line1->fields[join_field_1].beg; |
1486 | - len1 = line1->fields[join_field_1].len; |
1487 | + beg[0] = line1->fields[join_field_1].beg; |
1488 | + len[0] = line1->fields[join_field_1].len; |
1489 | } |
1490 | else |
1491 | { |
1492 | - beg1 = NULL; |
1493 | - len1 = 0; |
1494 | + beg[0] = NULL; |
1495 | + len[0] = 0; |
1496 | } |
1497 | |
1498 | if (join_field_2 < line2->nfields) |
1499 | { |
1500 | - beg2 = line2->fields[join_field_2].beg; |
1501 | - len2 = line2->fields[join_field_2].len; |
1502 | + beg[1] = line2->fields[join_field_2].beg; |
1503 | + len[1] = line2->fields[join_field_2].len; |
1504 | } |
1505 | else |
1506 | { |
1507 | - beg2 = NULL; |
1508 | - len2 = 0; |
1509 | + beg[1] = NULL; |
1510 | + len[1] = 0; |
1511 | } |
1512 | |
1513 | - if (len1 == 0) |
1514 | - return len2 == 0 ? 0 : -1; |
1515 | - if (len2 == 0) |
1516 | + if (len[0] == 0) |
1517 | + return len[1] == 0 ? 0 : -1; |
1518 | + if (len[1] == 0) |
1519 | return 1; |
1520 | |
1521 | if (ignore_case) |
1522 | { |
1523 | - /* FIXME: ignore_case does not work with NLS (in particular, |
1524 | - with multibyte chars). */ |
1525 | - diff = memcasecmp (beg1, beg2, MIN (len1, len2)); |
1526 | +#ifdef HAVE_MBRTOWC |
1527 | + if (MB_CUR_MAX > 1) |
1528 | + { |
1529 | + size_t mblength; |
1530 | + wchar_t wc, uwc; |
1531 | + mbstate_t state, state_bak; |
1532 | + |
1533 | + memset (&state, '\0', sizeof (mbstate_t)); |
1534 | + |
1535 | + for (i = 0; i < 2; i++) |
1536 | + { |
1537 | + copy[i] = alloca (len[i] + 1); |
1538 | + |
1539 | + for (j = 0; j < MIN (len[0], len[1]);) |
1540 | + { |
1541 | + state_bak = state; |
1542 | + mblength = mbrtowc (&wc, beg[i] + j, len[i] - j, &state); |
1543 | + |
1544 | + switch (mblength) |
1545 | + { |
1546 | + case (size_t) -1: |
1547 | + case (size_t) -2: |
1548 | + state = state_bak; |
1549 | + /* Fall through */ |
1550 | + case 0: |
1551 | + mblength = 1; |
1552 | + break; |
1553 | + |
1554 | + default: |
1555 | + uwc = towupper (wc); |
1556 | + |
1557 | + if (uwc != wc) |
1558 | + { |
1559 | + mbstate_t state_wc; |
1560 | + |
1561 | + memset (&state_wc, '\0', sizeof (mbstate_t)); |
1562 | + wcrtomb (copy[i] + j, uwc, &state_wc); |
1563 | + } |
1564 | + else |
1565 | + memcpy (copy[i] + j, beg[i] + j, mblength); |
1566 | + } |
1567 | + j += mblength; |
1568 | + } |
1569 | + copy[i][j] = '\0'; |
1570 | + } |
1571 | + return xmemcoll (copy[0], len[0], copy[1], len[1]); |
1572 | + } |
1573 | +#endif |
1574 | + if (hard_LC_COLLATE) |
1575 | + { |
1576 | + for (i = 0; i < 2; i++) |
1577 | + { |
1578 | + copy[i] = alloca (len[i] + 1); |
1579 | + |
1580 | + for (j = 0; j < MIN (len[0], len[1]); j++) |
1581 | + copy[i][j] = toupper (beg[i][j]); |
1582 | + |
1583 | + copy[i][j] = '\0'; |
1584 | + } |
1585 | + return xmemcoll (copy[0], len[0], copy[1], len[1]); |
1586 | + } |
1587 | + else |
1588 | + diff = memcasecmp (beg[0], beg[1], MIN (len[0], len[1])); |
1589 | } |
1590 | else |
1591 | { |
1592 | if (hard_LC_COLLATE) |
1593 | - return xmemcoll (beg1, len1, beg2, len2); |
1594 | - diff = memcmp (beg1, beg2, MIN (len1, len2)); |
1595 | + return xmemcoll (beg[0], len[0], beg[1], len[1]); |
1596 | + diff = memcmp (beg[0], beg[1], MIN (len[0], len[1])); |
1597 | } |
1598 | |
1599 | if (diff) |
1600 | return diff; |
1601 | - return len1 < len2 ? -1 : len1 != len2; |
1602 | + return len[0] < len[1] ? -1 : len[0] != len[1]; |
1603 | } |
1604 | |
1605 | /* Print field N of LINE if it exists and is nonempty, otherwise |
1606 | @@ -414,7 +624,8 @@ static void |
1607 | prjoin (struct line const *line1, struct line const *line2) |
1608 | { |
1609 | const struct outlist *outlist; |
1610 | - char output_separator = tab ? tab : ' '; |
1611 | + char *output_separator = tab ? tab : " "; |
1612 | + size_t output_separator_len = tab ? tablen : 1; |
1613 | |
1614 | outlist = outlist_head.next; |
1615 | if (outlist) |
1616 | @@ -449,7 +660,7 @@ prjoin (struct line const *line1, struct |
1617 | o = o->next; |
1618 | if (o == NULL) |
1619 | break; |
1620 | - putchar (output_separator); |
1621 | + fwrite (output_separator, 1, output_separator_len, stdout); |
1622 | } |
1623 | putchar ('\n'); |
1624 | } |
1625 | @@ -467,23 +678,23 @@ prjoin (struct line const *line1, struct |
1626 | prfield (join_field_1, line1); |
1627 | for (i = 0; i < join_field_1 && i < line1->nfields; ++i) |
1628 | { |
1629 | - putchar (output_separator); |
1630 | + fwrite (output_separator, 1, output_separator_len, stdout); |
1631 | prfield (i, line1); |
1632 | } |
1633 | for (i = join_field_1 + 1; i < line1->nfields; ++i) |
1634 | { |
1635 | - putchar (output_separator); |
1636 | + fwrite (output_separator, 1, output_separator_len, stdout); |
1637 | prfield (i, line1); |
1638 | } |
1639 | |
1640 | for (i = 0; i < join_field_2 && i < line2->nfields; ++i) |
1641 | { |
1642 | - putchar (output_separator); |
1643 | + fwrite (output_separator, 1, output_separator_len, stdout); |
1644 | prfield (i, line2); |
1645 | } |
1646 | for (i = join_field_2 + 1; i < line2->nfields; ++i) |
1647 | { |
1648 | - putchar (output_separator); |
1649 | + fwrite (output_separator, 1, output_separator_len, stdout); |
1650 | prfield (i, line2); |
1651 | } |
1652 | putchar ('\n'); |
1653 | @@ -814,7 +1025,21 @@ main (int argc, char **argv) |
1654 | break; |
1655 | |
1656 | case 't': |
1657 | - tab = *optarg; |
1658 | + tab = xstrdup (optarg); |
1659 | +#if HAVE_MBRTOWC |
1660 | + if (MB_CUR_MAX > 1) |
1661 | + { |
1662 | + mbstate_t state; |
1663 | + |
1664 | + memset (&state, 0, sizeof (mbstate_t)); |
1665 | + tablen = mbrtowc (NULL, optarg, strlen (optarg), &state); |
1666 | + if (tablen == (size_t) 0 |
1667 | + || tablen == (size_t) -1 || tablen == (size_t) -2) |
1668 | + tablen = 1; |
1669 | + } |
1670 | + else |
1671 | +#endif |
1672 | + tablen = 1; |
1673 | break; |
1674 | |
1675 | case 1: /* Non-option argument. */ |
1676 | --- coreutils/src/pr.c |
1677 | +++ coreutils/src/pr.c |
1678 | @@ -314,6 +314,32 @@ |
1679 | #include <stdio.h> |
1680 | #include <getopt.h> |
1681 | #include <sys/types.h> |
1682 | + |
1683 | +/* Get MB_LEN_MAX. */ |
1684 | +#include <limits.h> |
1685 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
1686 | + installation; work around this configuration error. */ |
1687 | +#if !defined MB_LEN_MAX || MB_LEN_MAX == 1 |
1688 | +# define MB_LEN_MAX 16 |
1689 | +#endif |
1690 | + |
1691 | +/* Get MB_CUR_MAX. */ |
1692 | +#include <stdlib.h> |
1693 | + |
1694 | +/* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */ |
1695 | +/* Get mbstate_t, mbrtowc(), wcwidth(). */ |
1696 | +#if HAVE_WCHAR_H |
1697 | +# include <wchar.h> |
1698 | +#endif |
1699 | + |
1700 | +/* Get iswprint(). -- for wcwidth(). */ |
1701 | +#if HAVE_WCTYPE_H |
1702 | +# include <wctype.h> |
1703 | +#endif |
1704 | +#if !defined iswprint && !HAVE_ISWPRINT |
1705 | +# define iswprint(wc) 1 |
1706 | +#endif |
1707 | + |
1708 | #include "system.h" |
1709 | #include "error.h" |
1710 | #include "hard-locale.h" |
1711 | @@ -321,6 +347,18 @@ |
1712 | #include "posixver.h" |
1713 | #include "xstrtol.h" |
1714 | |
1715 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
1716 | +#if HAVE_MBRTOWC && defined mbstate_t |
1717 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
1718 | +#endif |
1719 | + |
1720 | +#ifndef HAVE_DECL_WCWIDTH |
1721 | +"this configure-time declaration test was not run" |
1722 | +#endif |
1723 | +#if !HAVE_DECL_WCWIDTH |
1724 | +extern int wcwidth (); |
1725 | +#endif |
1726 | + |
1727 | #if ! (HAVE_DECL_STRTOUMAX || defined strtoumax) |
1728 | uintmax_t strtoumax (); |
1729 | #endif |
1730 | @@ -416,8 +454,21 @@ struct COLUMN |
1731 | typedef struct COLUMN COLUMN; |
1732 | |
1733 | #define NULLCOL (COLUMN *)0 |
1734 | + |
1735 | +/* Funtion pointers to switch functions for single byte locale or for |
1736 | + multibyte locale. If multibyte functions do not exist in your sysytem, |
1737 | + these pointers always point the function for single byte locale. */ |
1738 | +static void (*print_char) (char c); |
1739 | +static int (*char_to_clump) (char c); |
1740 | + |
1741 | +/* Functions for single byte locale. */ |
1742 | +static void print_char_single (char c); |
1743 | +static int char_to_clump_single (char c); |
1744 | + |
1745 | +/* Functions for multibyte locale. */ |
1746 | +static void print_char_multi (char c); |
1747 | +static int char_to_clump_multi (char c); |
1748 | |
1749 | -static int char_to_clump (char c); |
1750 | static bool read_line (COLUMN *p); |
1751 | static bool print_page (void); |
1752 | static bool print_stored (COLUMN *p); |
1753 | @@ -427,6 +478,7 @@ static void print_header (void); |
1754 | static void pad_across_to (int position); |
1755 | static void add_line_number (COLUMN *p); |
1756 | static void getoptarg (char *arg, char switch_char, char *character, |
1757 | + int *character_length, int *character_width, |
1758 | int *number); |
1759 | void usage (int status); |
1760 | static void print_files (int number_of_files, char **av); |
1761 | @@ -441,7 +493,6 @@ static void store_char (char c); |
1762 | static void pad_down (int lines); |
1763 | static void read_rest_of_line (COLUMN *p); |
1764 | static void skip_read (COLUMN *p, int column_number); |
1765 | -static void print_char (char c); |
1766 | static void cleanup (void); |
1767 | static void print_sep_string (void); |
1768 | static void separator_string (const char *optarg_S); |
1769 | @@ -456,7 +507,7 @@ static COLUMN *column_vector; |
1770 | we store the leftmost columns contiguously in buff. |
1771 | To print a line from buff, get the index of the first character |
1772 | from line_vector[i], and print up to line_vector[i + 1]. */ |
1773 | -static char *buff; |
1774 | +static unsigned char *buff; |
1775 | |
1776 | /* Index of the position in buff where the next character |
1777 | will be stored. */ |
1778 | @@ -560,7 +611,7 @@ static int chars_per_column; |
1779 | static bool untabify_input = false; |
1780 | |
1781 | /* (-e) The input tab character. */ |
1782 | -static char input_tab_char = '\t'; |
1783 | +static char input_tab_char[MB_LEN_MAX] = "\t"; |
1784 | |
1785 | /* (-e) Tabstops are at chars_per_tab, 2*chars_per_tab, 3*chars_per_tab, ... |
1786 | where the leftmost column is 1. */ |
1787 | @@ -570,7 +621,10 @@ static int chars_per_input_tab = 8; |
1788 | static bool tabify_output = false; |
1789 | |
1790 | /* (-i) The output tab character. */ |
1791 | -static char output_tab_char = '\t'; |
1792 | +static char output_tab_char[MB_LEN_MAX] = "\t"; |
1793 | + |
1794 | +/* (-i) The byte length of output tab character. */ |
1795 | +static int output_tab_char_length = 1; |
1796 | |
1797 | /* (-i) The width of the output tab. */ |
1798 | static int chars_per_output_tab = 8; |
1799 | @@ -644,7 +698,13 @@ static int power_10; |
1800 | static bool numbered_lines = false; |
1801 | |
1802 | /* (-n) Character which follows each line number. */ |
1803 | -static char number_separator = '\t'; |
1804 | +static char number_separator[MB_LEN_MAX] = "\t"; |
1805 | + |
1806 | +/* (-n) The byte length of the character which follows each line number. */ |
1807 | +static int number_separator_length = 1; |
1808 | + |
1809 | +/* (-n) The character width of the character which follows each line number. */ |
1810 | +static int number_separator_width = 0; |
1811 | |
1812 | /* (-n) line counting starts with 1st line of input file (not with 1st |
1813 | line of 1st page printed). */ |
1814 | @@ -697,6 +757,7 @@ static bool use_col_separator = false; |
1815 | -a|COLUMN|-m is a `space' and with the -J option a `tab'. */ |
1816 | static char *col_sep_string = ""; |
1817 | static int col_sep_length = 0; |
1818 | +static int col_sep_width = 0; |
1819 | static char *column_separator = " "; |
1820 | static char *line_separator = "\t"; |
1821 | |
1822 | @@ -840,6 +901,13 @@ separator_string (const char *optarg_S) |
1823 | col_sep_length = (int) strlen (optarg_S); |
1824 | col_sep_string = xmalloc (col_sep_length + 1); |
1825 | strcpy (col_sep_string, optarg_S); |
1826 | + |
1827 | +#if HAVE_MBRTOWC |
1828 | + if (MB_CUR_MAX > 1) |
1829 | + col_sep_width = mbswidth (col_sep_string, 0); |
1830 | + else |
1831 | +#endif |
1832 | + col_sep_width = col_sep_length; |
1833 | } |
1834 | |
1835 | int |
1836 | @@ -864,6 +932,21 @@ main (int argc, char **argv) |
1837 | |
1838 | atexit (close_stdout); |
1839 | |
1840 | +/* Define which functions are used, the ones for single byte locale or the ones |
1841 | + for multibyte locale. */ |
1842 | +#if HAVE_MBRTOWC |
1843 | + if (MB_CUR_MAX > 1) |
1844 | + { |
1845 | + print_char = print_char_multi; |
1846 | + char_to_clump = char_to_clump_multi; |
1847 | + } |
1848 | + else |
1849 | +#endif |
1850 | + { |
1851 | + print_char = print_char_single; |
1852 | + char_to_clump = char_to_clump_single; |
1853 | + } |
1854 | + |
1855 | n_files = 0; |
1856 | file_names = (argc > 1 |
1857 | ? xmalloc ((argc - 1) * sizeof (char *)) |
1858 | @@ -938,8 +1021,12 @@ main (int argc, char **argv) |
1859 | break; |
1860 | case 'e': |
1861 | if (optarg) |
1862 | - getoptarg (optarg, 'e', &input_tab_char, |
1863 | - &chars_per_input_tab); |
1864 | + { |
1865 | + int dummy_length, dummy_width; |
1866 | + |
1867 | + getoptarg (optarg, 'e', input_tab_char, &dummy_length, |
1868 | + &dummy_width, &chars_per_input_tab); |
1869 | + } |
1870 | /* Could check tab width > 0. */ |
1871 | untabify_input = true; |
1872 | break; |
1873 | @@ -952,8 +1039,12 @@ main (int argc, char **argv) |
1874 | break; |
1875 | case 'i': |
1876 | if (optarg) |
1877 | - getoptarg (optarg, 'i', &output_tab_char, |
1878 | - &chars_per_output_tab); |
1879 | + { |
1880 | + int dummy_width; |
1881 | + |
1882 | + getoptarg (optarg, 'i', output_tab_char, &output_tab_char_length, |
1883 | + &dummy_width, &chars_per_output_tab); |
1884 | + } |
1885 | /* Could check tab width > 0. */ |
1886 | tabify_output = true; |
1887 | break; |
1888 | @@ -980,8 +1071,8 @@ main (int argc, char **argv) |
1889 | case 'n': |
1890 | numbered_lines = true; |
1891 | if (optarg) |
1892 | - getoptarg (optarg, 'n', &number_separator, |
1893 | - &chars_per_number); |
1894 | + getoptarg (optarg, 'n', number_separator, &number_separator_length, |
1895 | + &number_separator_width, &chars_per_number); |
1896 | break; |
1897 | case 'N': |
1898 | skip_count = false; |
1899 | @@ -1020,7 +1111,7 @@ main (int argc, char **argv) |
1900 | old_s = false; |
1901 | /* Reset an additional input of -s, -S dominates -s */ |
1902 | col_sep_string = ""; |
1903 | - col_sep_length = 0; |
1904 | + col_sep_length = col_sep_width = 0; |
1905 | use_col_separator = true; |
1906 | if (optarg) |
1907 | separator_string (optarg); |
1908 | @@ -1169,10 +1260,45 @@ main (int argc, char **argv) |
1909 | a number. */ |
1910 | |
1911 | static void |
1912 | -getoptarg (char *arg, char switch_char, char *character, int *number) |
1913 | +getoptarg (char *arg, char switch_char, char *character, int *character_length, |
1914 | + int *character_width, int *number) |
1915 | { |
1916 | if (!ISDIGIT (*arg)) |
1917 | - *character = *arg++; |
1918 | + { |
1919 | +#ifdef HAVE_MBRTOWC |
1920 | + if (MB_CUR_MAX > 1) /* for multibyte locale. */ |
1921 | + { |
1922 | + wchar_t wc; |
1923 | + size_t mblength; |
1924 | + int width; |
1925 | + mbstate_t state = {'\0'}; |
1926 | + |
1927 | + mblength = mbrtowc (&wc, arg, strlen (arg), &state); |
1928 | + |
1929 | + if (mblength == (size_t) -1 || mblength == (size_t) -2) |
1930 | + { |
1931 | + *character_length = 1; |
1932 | + *character_width = 1; |
1933 | + } |
1934 | + else |
1935 | + { |
1936 | + *character_length = (mblength < 1) ? 1 : mblength; |
1937 | + width = wcwidth (wc); |
1938 | + *character_width = (width < 0) ? 0 : width; |
1939 | + } |
1940 | + |
1941 | + strncpy (character, arg, *character_length); |
1942 | + arg += *character_length; |
1943 | + } |
1944 | + else /* for single byte locale. */ |
1945 | +#endif |
1946 | + { |
1947 | + *character = *arg++; |
1948 | + *character_length = 1; |
1949 | + *character_width = 1; |
1950 | + } |
1951 | + } |
1952 | + |
1953 | if (*arg) |
1954 | { |
1955 | long int tmp_long; |
1956 | @@ -1237,7 +1363,7 @@ init_parameters (int number_of_files) |
1957 | else |
1958 | col_sep_string = column_separator; |
1959 | |
1960 | - col_sep_length = 1; |
1961 | + col_sep_length = col_sep_width = 1; |
1962 | use_col_separator = true; |
1963 | } |
1964 | /* It's rather pointless to define a TAB separator with column |
1965 | @@ -1269,11 +1395,11 @@ init_parameters (int number_of_files) |
1966 | TAB_WIDTH (chars_per_input_tab, chars_per_number); */ |
1967 | |
1968 | /* Estimate chars_per_text without any margin and keep it constant. */ |
1969 | - if (number_separator == '\t') |
1970 | + if (number_separator[0] == '\t') |
1971 | number_width = chars_per_number + |
1972 | TAB_WIDTH (chars_per_default_tab, chars_per_number); |
1973 | else |
1974 | - number_width = chars_per_number + 1; |
1975 | + number_width = chars_per_number + number_separator_width; |
1976 | |
1977 | /* The number is part of the column width unless we are |
1978 | printing files in parallel. */ |
1979 | @@ -1288,7 +1414,7 @@ init_parameters (int number_of_files) |
1980 | } |
1981 | |
1982 | chars_per_column = (chars_per_line - chars_used_by_number - |
1983 | - (columns - 1) * col_sep_length) / columns; |
1984 | + (columns - 1) * col_sep_width) / columns; |
1985 | |
1986 | if (chars_per_column < 1) |
1987 | error (EXIT_FAILURE, 0, _("page width too narrow")); |
1988 | @@ -1416,7 +1542,7 @@ init_funcs (void) |
1989 | |
1990 | /* Enlarge p->start_position of first column to use the same form of |
1991 | padding_not_printed with all columns. */ |
1992 | - h = h + col_sep_length; |
1993 | + h = h + col_sep_width; |
1994 | |
1995 | /* This loop takes care of all but the rightmost column. */ |
1996 | |
1997 | @@ -1450,7 +1576,7 @@ init_funcs (void) |
1998 | } |
1999 | else |
2000 | { |
2001 | - h = h_next + col_sep_length; |
2002 | + h = h_next + col_sep_width; |
2003 | h_next = h + chars_per_column; |
2004 | } |
2005 | } |
2006 | @@ -1734,9 +1860,9 @@ static void |
2007 | align_column (COLUMN *p) |
2008 | { |
2009 | padding_not_printed = p->start_position; |
2010 | - if (padding_not_printed - col_sep_length > 0) |
2011 | + if (padding_not_printed - col_sep_width > 0) |
2012 | { |
2013 | - pad_across_to (padding_not_printed - col_sep_length); |
2014 | + pad_across_to (padding_not_printed - col_sep_width); |
2015 | padding_not_printed = ANYWHERE; |
2016 | } |
2017 | |
2018 | @@ -2010,13 +2136,13 @@ store_char (char c) |
2019 | /* May be too generous. */ |
2020 | buff = x2nrealloc (buff, &buff_allocated, sizeof *buff); |
2021 | } |
2022 | - buff[buff_current++] = c; |
2023 | + buff[buff_current++] = (unsigned char) c; |
2024 | } |
2025 | |
2026 | static void |
2027 | add_line_number (COLUMN *p) |
2028 | { |
2029 | - int i; |
2030 | + int i, j; |
2031 | char *s; |
2032 | int left_cut; |
2033 | |
2034 | @@ -2039,22 +2165,24 @@ add_line_number (COLUMN *p) |
2035 | /* Tabification is assumed for multiple columns, also for n-separators, |
2036 | but `default n-separator = TAB' hasn't been given priority over |
2037 | equal column_width also specified by POSIX. */ |
2038 | - if (number_separator == '\t') |
2039 | + if (number_separator[0] == '\t') |
2040 | { |
2041 | i = number_width - chars_per_number; |
2042 | while (i-- > 0) |
2043 | (p->char_func) (' '); |
2044 | } |
2045 | else |
2046 | - (p->char_func) (number_separator); |
2047 | + for (j = 0; j < number_separator_length; j++) |
2048 | + (p->char_func) (number_separator[j]); |
2049 | } |
2050 | else |
2051 | /* To comply with POSIX, we avoid any expansion of default TAB |
2052 | separator with a single column output. No column_width requirement |
2053 | has to be considered. */ |
2054 | { |
2055 | - (p->char_func) (number_separator); |
2056 | - if (number_separator == '\t') |
2057 | + for (j = 0; j < number_separator_length; j++) |
2058 | + (p->char_func) (number_separator[j]); |
2059 | + if (number_separator[0] == '\t') |
2060 | output_position = POS_AFTER_TAB (chars_per_output_tab, |
2061 | output_position); |
2062 | } |
2063 | @@ -2215,7 +2343,7 @@ print_white_space (void) |
2064 | while (goal - h_old > 1 |
2065 | && (h_new = POS_AFTER_TAB (chars_per_output_tab, h_old)) <= goal) |
2066 | { |
2067 | - putchar (output_tab_char); |
2068 | + fwrite (output_tab_char, 1, output_tab_char_length, stdout); |
2069 | h_old = h_new; |
2070 | } |
2071 | while (++h_old <= goal) |
2072 | @@ -2235,6 +2363,7 @@ print_sep_string () |
2073 | { |
2074 | char *s; |
2075 | int l = col_sep_length; |
2076 | + int not_space_flag; |
2077 | |
2078 | s = col_sep_string; |
2079 | |
2080 | @@ -2248,6 +2377,7 @@ print_sep_string () |
2081 | { |
2082 | for (; separators_not_printed > 0; --separators_not_printed) |
2083 | { |
2084 | + not_space_flag = 0; |
2085 | while (l-- > 0) |
2086 | { |
2087 | /* 3 types of sep_strings: spaces only, spaces and chars, |
2088 | @@ -2261,12 +2391,15 @@ print_sep_string () |
2089 | } |
2090 | else |
2091 | { |
2092 | + not_space_flag = 1; |
2093 | if (spaces_not_printed > 0) |
2094 | print_white_space (); |
2095 | putchar (*s++); |
2096 | - ++output_position; |
2097 | } |
2098 | } |
2099 | + if (not_space_flag) |
2100 | + output_position += col_sep_width; |
2101 | + |
2102 | /* sep_string ends with some spaces */ |
2103 | if (spaces_not_printed > 0) |
2104 | print_white_space (); |
2105 | @@ -2293,8 +2426,9 @@ print_clump (COLUMN *p, int n, char *clu |
2106 | a nonspace is encountered, call print_white_space() to print the |
2107 | required number of tabs and spaces. */ |
2108 | |
2109 | + |
2110 | static void |
2111 | -print_char (char c) |
2112 | +print_char_single (char c) |
2113 | { |
2114 | if (tabify_output) |
2115 | { |
2116 | @@ -2318,6 +2452,75 @@ print_char (char c) |
2117 | putchar (c); |
2118 | } |
2119 | |
2120 | +#ifdef HAVE_MBRTOWC |
2121 | +static void |
2122 | +print_char_multi (char c) |
2123 | +{ |
2124 | + static size_t mbc_pos = 0; |
2125 | + static unsigned char mbc[MB_LEN_MAX] = {'\0'}; |
2126 | + static mbstate_t state = {'\0'}; |
2127 | + mbstate_t state_bak; |
2128 | + wchar_t wc; |
2129 | + unsigned char uc = (unsigned char) c; |
2130 | + size_t mblength; |
2131 | + int width; |
2132 | + |
2133 | + if (tabify_output) |
2134 | + { |
2135 | + state_bak = state; |
2136 | + mbc[mbc_pos++] = uc; |
2137 | + mblength = mbrtowc (&wc, mbc, mbc_pos, &state); |
2138 | + |
2139 | + while (mbc_pos > 0) |
2140 | + { |
2141 | + switch (mblength) |
2142 | + { |
2143 | + case (size_t) -2: |
2144 | + state = state_bak; |
2145 | + return; |
2146 | + |
2147 | + case (size_t) -1: |
2148 | + state = state_bak; |
2149 | + ++output_position; |
2150 | + putchar (mbc[0]); |
2151 | + memmove (mbc, mbc + 1, MB_CUR_MAX - 1); |
2152 | + --mbc_pos; |
2153 | + break; |
2154 | + |
2155 | + case 0: |
2156 | + mblength = 1; |
2157 | + |
2158 | + default: |
2159 | + if (wc == L' ') |
2160 | + { |
2161 | + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); |
2162 | + --mbc_pos; |
2163 | + ++spaces_not_printed; |
2164 | + return; |
2165 | + } |
2166 | + else if (spaces_not_printed > 0) |
2167 | + print_white_space (); |
2168 | + |
2169 | + /* Nonprintables are assumed to have width 0, except L'\b'. */ |
2170 | + if ((width = wcwidth (wc)) < 1) |
2171 | + { |
2172 | + if (wc == L'\b') |
2173 | + --output_position; |
2174 | + } |
2175 | + else |
2176 | + output_position += width; |
2177 | + |
2178 | + fwrite (mbc, 1, mblength, stdout); |
2179 | + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); |
2180 | + mbc_pos -= mblength; |
2181 | + } |
2182 | + } |
2183 | + return; |
2184 | + } |
2185 | + putchar (uc); |
2186 | +} |
2187 | +#endif |
2188 | + |
2189 | /* Skip to page PAGE before printing. |
2190 | PAGE may be larger than total number of pages. */ |
2191 | |
2192 | @@ -2498,9 +2701,9 @@ read_line (COLUMN *p) |
2193 | align_empty_cols = false; |
2194 | } |
2195 | |
2196 | - if (padding_not_printed - col_sep_length > 0) |
2197 | + if (padding_not_printed - col_sep_width > 0) |
2198 | { |
2199 | - pad_across_to (padding_not_printed - col_sep_length); |
2200 | + pad_across_to (padding_not_printed - col_sep_width); |
2201 | padding_not_printed = ANYWHERE; |
2202 | } |
2203 | |
2204 | @@ -2601,9 +2804,9 @@ print_stored (COLUMN *p) |
2205 | } |
2206 | } |
2207 | |
2208 | - if (padding_not_printed - col_sep_length > 0) |
2209 | + if (padding_not_printed - col_sep_width > 0) |
2210 | { |
2211 | - pad_across_to (padding_not_printed - col_sep_length); |
2212 | + pad_across_to (padding_not_printed - col_sep_width); |
2213 | padding_not_printed = ANYWHERE; |
2214 | } |
2215 | |
2216 | @@ -2616,8 +2819,8 @@ print_stored (COLUMN *p) |
2217 | if (spaces_not_printed == 0) |
2218 | { |
2219 | output_position = p->start_position + end_vector[line]; |
2220 | - if (p->start_position - col_sep_length == chars_per_margin) |
2221 | - output_position -= col_sep_length; |
2222 | + if (p->start_position - col_sep_width == chars_per_margin) |
2223 | + output_position -= col_sep_width; |
2224 | } |
2225 | |
2226 | return true; |
2227 | @@ -2635,8 +2838,9 @@ print_stored (COLUMN *p) |
2228 | characters in clump_buff. (e.g, the width of '\b' is -1, while the |
2229 | number of characters is 1.) */ |
2230 | |
2231 | + |
2232 | static int |
2233 | -char_to_clump (char c) |
2234 | +char_to_clump_single (char c) |
2235 | { |
2236 | unsigned char uc = c; |
2237 | register char *s = clump_buff; |
2238 | @@ -2646,10 +2850,10 @@ char_to_clump (char c) |
2239 | int chars; |
2240 | int chars_per_c = 8; |
2241 | |
2242 | - if (c == input_tab_char) |
2243 | + if (c == input_tab_char[0]) |
2244 | chars_per_c = chars_per_input_tab; |
2245 | |
2246 | - if (c == input_tab_char || c == '\t') |
2247 | + if (c == input_tab_char[0] || c == '\t') |
2248 | { |
2249 | width = TAB_WIDTH (chars_per_c, input_position); |
2250 | |
2251 | @@ -2720,6 +2924,155 @@ char_to_clump (char c) |
2252 | return chars; |
2253 | } |
2254 | |
2255 | +#ifdef HAVE_MBRTOWC |
2256 | +static int |
2257 | +char_to_clump_multi (char c) |
2258 | +{ |
2259 | + static size_t mbc_pos = 0; |
2260 | + static unsigned char mbc[MB_LEN_MAX] = {'\0'}; |
2261 | + static mbstate_t state = {'\0'}; |
2262 | + mbstate_t state_bak; |
2263 | + wchar_t wc; |
2264 | + unsigned char uc = (unsigned char) c; |
2265 | + size_t mblength; |
2266 | + int wc_width; |
2267 | + register char *s = clump_buff; |
2268 | + register int i, j; |
2269 | + char esc_buff[4]; |
2270 | + int width; |
2271 | + int chars; |
2272 | + int chars_per_c = 8; |
2273 | + |
2274 | + state_bak = state; |
2275 | + mbc[mbc_pos++] = uc; |
2276 | + mblength = mbrtowc (&wc, mbc, mbc_pos, &state); |
2277 | + |
2278 | + width = 0; |
2279 | + chars = 0; |
2280 | + while (mbc_pos > 0) |
2281 | + { |
2282 | + switch (mblength) |
2283 | + { |
2284 | + case (size_t) -2: |
2285 | + state = state_bak; |
2286 | + return 0; |
2287 | + |
2288 | + case (size_t) -1: |
2289 | + state = state_bak; |
2290 | + mblength = 1; |
2291 | + |
2292 | + if (use_esc_sequence || use_cntrl_prefix) |
2293 | + { |
2294 | + width = +4; |
2295 | + chars = +4; |
2296 | + *s++ = '\\'; |
2297 | + sprintf (esc_buff, "%03o", mbc[0]); |
2298 | + for (i = 0; i <= 2; ++i) |
2299 | + *s++ = (int) esc_buff[i]; |
2300 | + } |
2301 | + else |
2302 | + { |
2303 | + width += 1; |
2304 | + chars += 1; |
2305 | + *s++ = mbc[0]; |
2306 | + } |
2307 | + break; |
2308 | + |
2309 | + case 0: |
2310 | + mblength = 1; |
2311 | + /* Fall through */ |
2312 | + |
2313 | + default: |
2314 | + if (memcmp (mbc, input_tab_char, mblength) == 0) |
2315 | + chars_per_c = chars_per_input_tab; |
2316 | + |
2317 | + if (memcmp (mbc, input_tab_char, mblength) == 0 || c == '\t') |
2318 | + { |
2319 | + int width_inc; |
2320 | + |
2321 | + width_inc = TAB_WIDTH (chars_per_c, input_position); |
2322 | + width += width_inc; |
2323 | + |
2324 | + if (untabify_input) |
2325 | + { |
2326 | + for (i = width_inc; i; --i) |
2327 | + *s++ = ' '; |
2328 | + chars += width_inc; |
2329 | + } |
2330 | + else |
2331 | + { |
2332 | + for (i = 0; i < mblength; i++) |
2333 | + *s++ = mbc[i]; |
2334 | + chars += mblength; |
2335 | + } |
2336 | + } |
2337 | + else if ((wc_width = wcwidth (wc)) < 1) |
2338 | + { |
2339 | + if (use_esc_sequence) |
2340 | + { |
2341 | + for (i = 0; i < mblength; i++) |
2342 | + { |
2343 | + width += 4; |
2344 | + chars += 4; |
2345 | + *s++ = '\\'; |
2346 | + sprintf (esc_buff, "%03o", uc); |
2347 | + for (j = 0; j <= 2; ++j) |
2348 | + *s++ = (int) esc_buff[j]; |
2349 | + } |
2350 | + } |
2351 | + else if (use_cntrl_prefix) |
2352 | + { |
2353 | + if (wc < 0200) |
2354 | + { |
2355 | + width += 2; |
2356 | + chars += 2; |
2357 | + *s++ = '^'; |
2358 | + *s++ = wc ^ 0100; |
2359 | + } |
2360 | + else |
2361 | + { |
2362 | + for (i = 0; i < mblength; i++) |
2363 | + { |
2364 | + width += 4; |
2365 | + chars += 4; |
2366 | + *s++ = '\\'; |
2367 | + sprintf (esc_buff, "%03o", uc); |
2368 | + for (j = 0; j <= 2; ++j) |
2369 | + *s++ = (int) esc_buff[j]; |
2370 | + } |
2371 | + } |
2372 | + } |
2373 | + else if (wc == L'\b') |
2374 | + { |
2375 | + width += -1; |
2376 | + chars += 1; |
2377 | + *s++ = c; |
2378 | + } |
2379 | + else |
2380 | + { |
2381 | + width += 0; |
2382 | + chars += mblength; |
2383 | + for (i = 0; i < mblength; i++) |
2384 | + *s++ = mbc[i]; |
2385 | + } |
2386 | + } |
2387 | + else |
2388 | + { |
2389 | + width += wc_width; |
2390 | + chars += mblength; |
2391 | + for (i = 0; i < mblength; i++) |
2392 | + *s++ = mbc[i]; |
2393 | + } |
2394 | + } |
2395 | + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); |
2396 | + mbc_pos -= mblength; |
2397 | + } |
2398 | + |
2399 | + input_position += width; |
2400 | + return chars; |
2401 | +} |
2402 | +#endif |
2403 | + |
2404 | /* We've just printed some files and need to clean up things before |
2405 | looking for more options and printing the next batch of files. |
2406 | |
2407 | --- coreutils/src/sort.c |
2408 | +++ coreutils/src/sort.c |
2409 | @@ -27,6 +27,19 @@ |
2410 | #include <sys/types.h> |
2411 | #include <signal.h> |
2412 | #include <stdio.h> |
2413 | +#include <assert.h> |
2414 | + |
2415 | +/* Get mbstate_t, mbrtowc(), wcrtomb(). */ |
2416 | +#if HAVE_WCHAR_H |
2417 | +# include <wchar.h> |
2418 | +#endif |
2419 | + |
2420 | +/* Get iswprint(), iswctype() towupper(). */ |
2421 | +#if HAVE_WCTYPE_H |
2422 | +# include <wctype.h> |
2423 | +wctype_t blank_type; /* = wctype ("blank"); */ |
2424 | +#endif |
2425 | + |
2426 | #include "system.h" |
2427 | #include "error.h" |
2428 | #include "hard-locale.h" |
2429 | @@ -46,6 +59,17 @@ struct rlimit { size_t rlim_cur; }; |
2430 | # define getrlimit(Resource, Rlp) (-1) |
2431 | #endif |
2432 | |
2433 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
2434 | + installation; work around this configuration error. */ |
2435 | +#if !defined MB_LEN_MAX || MB_LEN_MAX == 1 |
2436 | +# define MB_LEN_MAX 16 |
2437 | +#endif |
2438 | + |
2439 | +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ |
2440 | +#if HAVE_MBRTOWC && defined mbstate_t |
2441 | +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) |
2442 | +#endif |
2443 | + |
2444 | /* The official name of this program (e.g., no `g' prefix). */ |
2445 | #define PROGRAM_NAME "sort" |
2446 | |
2447 | @@ -91,14 +115,38 @@ static char decimal_point; |
2448 | /* Thousands separator; if CHAR_MAX + 1, then there isn't one. */ |
2449 | static int thousands_sep; |
2450 | |
2451 | +static int force_general_numcompare = 0; |
2452 | + |
2453 | /* Nonzero if the corresponding locales are hard. */ |
2454 | static bool hard_LC_COLLATE; |
2455 | -#if HAVE_NL_LANGINFO |
2456 | +#if HAVE_LANGINFO_CODESET |
2457 | static bool hard_LC_TIME; |
2458 | #endif |
2459 | |
2460 | #define NONZERO(x) ((x) != 0) |
2461 | |
2462 | +/* get a multibyte character's byte length. */ |
2463 | +#define GET_BYTELEN_OF_CHAR(LIM, PTR, MBLENGTH, STATE) \ |
2464 | + do \ |
2465 | + { \ |
2466 | + wchar_t wc; \ |
2467 | + mbstate_t state_bak; \ |
2468 | + \ |
2469 | + state_bak = STATE; \ |
2470 | + mblength = mbrtowc (&wc, PTR, LIM - PTR, &STATE); \ |
2471 | + \ |
2472 | + switch (MBLENGTH) \ |
2473 | + { \ |
2474 | + case (size_t)-1: \ |
2475 | + case (size_t)-2: \ |
2476 | + STATE = state_bak; \ |
2477 | + /* Fall through. */ \ |
2478 | + case 0: \ |
2479 | + MBLENGTH = 1; \ |
2480 | + } \ |
2481 | + } \ |
2482 | + while (0) |
2483 | + |
2484 | /* The kind of blanks for '-b' to skip in various options. */ |
2485 | enum blanktype { bl_start, bl_end, bl_both }; |
2486 | |
2487 | @@ -235,13 +283,22 @@ static bool reverse; |
2488 | they were read if all keys compare equal. */ |
2489 | static bool stable; |
2490 | |
2491 | -/* If TAB has this value, blanks separate fields. */ |
2492 | +/* If TAB has this value, blanks separate fields. |
2493 | enum { TAB_DEFAULT = CHAR_MAX + 1 }; |
2494 | +*/ |
2495 | |
2496 | /* Tab character separating fields. If TAB_DEFAULT, then fields are |
2497 | separated by the empty string between a non-blank character and a blank |
2498 | - character. */ |
2499 | + character. |
2500 | static int tab = TAB_DEFAULT; |
2501 | +*/ |
2502 | + |
2503 | +/* Tab character separating fields. If NUL, then fields are separated |
2504 | + by the empty string between a non-whitespace character and a whitespace |
2505 | + character. */ |
2506 | +static bool tab_default = true; |
2507 | +static unsigned char tab[MB_LEN_MAX + 1]; |
2508 | +static size_t tab_length = 1; |
2509 | |
2510 | /* Flag to remove consecutive duplicate lines from the output. |
2511 | Only the last of a sequence of equal lines will be output. */ |
2512 | @@ -386,6 +443,43 @@ struct tempnode |
2513 | static struct tempnode *volatile temphead; |
2514 | static struct tempnode *volatile *temptail = &temphead; |
2515 | |
2516 | +/* Fucntion pointers. */ |
2517 | +static char * |
2518 | +(* begfield) (const struct line *line, const struct keyfield *key); |
2519 | + |
2520 | +static char * |
2521 | +(* limfield) (const struct line *line, const struct keyfield *key); |
2522 | + |
2523 | +static int |
2524 | +(*getmonth) (const char *s, size_t len); |
2525 | + |
2526 | +static int |
2527 | +(* keycompare) (const struct line *a, const struct line *b); |
2528 | + |
2529 | +/* Test for white space multibyte character. |
2530 | + Set LENGTH the byte length of investigated multibyte character. */ |
2531 | +#if HAVE_MBRTOWC |
2532 | +static int |
2533 | +ismbblank (const char *str, size_t *length) |
2534 | +{ |
2535 | + size_t mblength; |
2536 | + wchar_t wc; |
2537 | + mbstate_t state; |
2538 | + |
2539 | + memset (&state, '\0', sizeof(mbstate_t)); |
2540 | + mblength = mbrtowc (&wc, str, MB_LEN_MAX, &state); |
2541 | + |
2542 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
2543 | + { |
2544 | + *length = 1; |
2545 | + return 0; |
2546 | + } |
2547 | + |
2548 | + *length = (mblength < 1) ? 1 : mblength; |
2549 | + return (iswctype (wc, blank_type)); |
2550 | +} |
2551 | +#endif |
2552 | + |
2553 | /* Clean up any remaining temporary files. */ |
2554 | |
2555 | static void |
2556 | @@ -535,7 +629,7 @@ zaptemp (const char *name) |
2557 | free (node); |
2558 | } |
2559 | |
2560 | -#if HAVE_NL_LANGINFO |
2561 | +#if HAVE_LANGINFO_CODESET |
2562 | |
2563 | static int |
2564 | struct_month_cmp (const void *m1, const void *m2) |
2565 | @@ -562,7 +656,7 @@ inittables (void) |
2566 | fold_toupper[i] = (ISLOWER (i) ? toupper (i) : i); |
2567 | } |
2568 | |
2569 | -#if HAVE_NL_LANGINFO |
2570 | +#if HAVE_LANGINFO_CODESET |
2571 | /* If we're not in the "C" locale, read different names for months. */ |
2572 | if (hard_LC_TIME) |
2573 | { |
2574 | @@ -588,6 +682,71 @@ inittables (void) |
2575 | #endif |
2576 | } |
2577 | |
2578 | +#if HAVE_MBRTOWC |
2579 | +static void |
2580 | +inittables_mb (void) |
2581 | +{ |
2582 | + int i, j, k, l; |
2583 | + char *name, *s; |
2584 | + size_t s_len, mblength; |
2585 | + char mbc[MB_LEN_MAX]; |
2586 | + wchar_t wc, pwc; |
2587 | + mbstate_t state_mb, state_wc; |
2588 | + |
2589 | + for (i = 0; i < MONTHS_PER_YEAR; i++) |
2590 | + { |
2591 | + s = (char *) nl_langinfo (ABMON_1 + i); |
2592 | + s_len = strlen (s); |
2593 | + monthtab[i].name = name = (char *) xmalloc (s_len + 1); |
2594 | + monthtab[i].val = i + 1; |
2595 | + |
2596 | + memset (&state_mb, '\0', sizeof (mbstate_t)); |
2597 | + memset (&state_wc, '\0', sizeof (mbstate_t)); |
2598 | + |
2599 | + for (j = 0; j < s_len;) |
2600 | + { |
2601 | + if (!ismbblank (s + j, &mblength)) |
2602 | + break; |
2603 | + j += mblength; |
2604 | + } |
2605 | + |
2606 | + for (k = 0; j < s_len;) |
2607 | + { |
2608 | + mblength = mbrtowc (&wc, (s + j), (s_len - j), &state_mb); |
2609 | + /* If conversion is failed, fall back into single byte sorting. */ |
2610 | + if (mblength == (size_t)-1 || mblength == (size_t)-2) |
2611 | + { |
2612 | + for (l = 0; l <= i; l++) |
2613 | + free ((void *)monthtab[l].name); |
2614 | + inittables(); |
2615 | + return; |
2616 | + } |
2617 | + else if (mblength == 0) |
2618 | + break; |
2619 | + |
2620 | + pwc = towupper (wc); |
2621 | + if (pwc == wc) |
2622 | + { |
2623 | + memcpy (mbc, s + j, mblength); |
2624 | + j += mblength; |
2625 | + } |
2626 | + else |
2627 | + { |
2628 | + j += mblength; |
2629 | + mblength = wcrtomb (mbc, wc, &state_wc); |
2630 | + assert (mblength != (size_t)0 && mblength != (size_t)-1); |
2631 | + } |
2632 | + |
2633 | + for (l = 0; l < mblength; l++) |
2634 | + name[k++] = mbc[l]; |
2635 | + } |
2636 | + name[k] = '\0'; |
2637 | + } |
2638 | + qsort ((void *) monthtab, MONTHS_PER_YEAR, |
2639 | + sizeof *monthtab, struct_month_cmp); |
2640 | +} |
2641 | +#endif |
2642 | + |
2643 | /* Specify the amount of main memory to use when sorting. */ |
2644 | static void |
2645 | specify_sort_size (char const *s) |
2646 | @@ -798,7 +957,7 @@ buffer_linelim (struct buffer const *buf |
2647 | by KEY in LINE. */ |
2648 | |
2649 | static char * |
2650 | -begfield (const struct line *line, const struct keyfield *key) |
2651 | +begfield_uni (const struct line *line, const struct keyfield *key) |
2652 | { |
2653 | register char *ptr = line->text, *lim = ptr + line->length - 1; |
2654 | register size_t sword = key->sword; |
2655 | @@ -808,10 +967,10 @@ begfield (const struct line *line, const |
2656 | /* The leading field separator itself is included in a field when -t |
2657 | is absent. */ |
2658 | |
2659 | - if (tab != TAB_DEFAULT) |
2660 | + if (!tab_default) |
2661 | while (ptr < lim && sword--) |
2662 | { |
2663 | - while (ptr < lim && *ptr != tab) |
2664 | + while (ptr < lim && *ptr != tab[0]) |
2665 | ++ptr; |
2666 | if (ptr < lim) |
2667 | ++ptr; |
2668 | @@ -839,11 +998,70 @@ begfield (const struct line *line, const |
2669 | return ptr; |
2670 | } |
2671 | |
2672 | +#if HAVE_MBRTOWC |
2673 | +static char * |
2674 | +begfield_mb (const struct line *line, const struct keyfield *key) |
2675 | +{ |
2676 | + int i; |
2677 | + char *ptr = line->text, *lim = ptr + line->length - 1; |
2678 | + size_t sword = key->sword; |
2679 | + size_t schar = key->schar; |
2680 | + size_t mblength; |
2681 | + mbstate_t state; |
2682 | + |
2683 | + memset (&state, '\0', sizeof(mbstate_t)); |
2684 | + |
2685 | + if (!tab_default) |
2686 | + while (ptr < lim && sword--) |
2687 | + { |
2688 | + while (ptr < lim && memcmp (ptr, tab, tab_length) != 0) |
2689 | + { |
2690 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2691 | + ptr += mblength; |
2692 | + } |
2693 | + if (ptr < lim) |
2694 | + { |
2695 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2696 | + ptr += mblength; |
2697 | + } |
2698 | + } |
2699 | + else |
2700 | + while (ptr < lim && sword--) |
2701 | + { |
2702 | + while (ptr < lim && ismbblank (ptr, &mblength)) |
2703 | + ptr += mblength; |
2704 | + if (ptr < lim) |
2705 | + { |
2706 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2707 | + ptr += mblength; |
2708 | + } |
2709 | + while (ptr < lim && !ismbblank (ptr, &mblength)) |
2710 | + ptr += mblength; |
2711 | + } |
2712 | + |
2713 | + if (key->skipsblanks) |
2714 | + while (ptr < lim && ismbblank (ptr, &mblength)) |
2715 | + ptr += mblength; |
2716 | + |
2717 | + for (i = 0; i < schar; i++) |
2718 | + { |
2719 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2720 | + |
2721 | + if (ptr + mblength > lim) |
2722 | + break; |
2723 | + else |
2724 | + ptr += mblength; |
2725 | + } |
2726 | + |
2727 | + return ptr; |
2728 | +} |
2729 | +#endif |
2730 | + |
2731 | /* Return the limit of (a pointer to the first character after) the field |
2732 | in LINE specified by KEY. */ |
2733 | |
2734 | static char * |
2735 | -limfield (const struct line *line, const struct keyfield *key) |
2736 | +limfield_uni (const struct line *line, const struct keyfield *key) |
2737 | { |
2738 | register char *ptr = line->text, *lim = ptr + line->length - 1; |
2739 | register size_t eword = key->eword, echar = key->echar; |
2740 | @@ -856,10 +1074,10 @@ limfield (const struct line *line, const |
2741 | `beginning' is the first character following the delimiting TAB. |
2742 | Otherwise, leave PTR pointing at the first `blank' character after |
2743 | the preceding field. */ |
2744 | - if (tab != TAB_DEFAULT) |
2745 | + if (!tab_default) |
2746 | while (ptr < lim && eword--) |
2747 | { |
2748 | - while (ptr < lim && *ptr != tab) |
2749 | + while (ptr < lim && *ptr != tab[0]) |
2750 | ++ptr; |
2751 | if (ptr < lim && (eword | echar)) |
2752 | ++ptr; |
2753 | @@ -905,7 +1123,7 @@ limfield (const struct line *line, const |
2754 | */ |
2755 | |
2756 | /* Make LIM point to the end of (one byte past) the current field. */ |
2757 | - if (tab != TAB_DEFAULT) |
2758 | + if (!tab_default) |
2759 | { |
2760 | char *newlim; |
2761 | newlim = memchr (ptr, tab, lim - ptr); |
2762 | @@ -941,6 +1159,107 @@ limfield (const struct line *line, const |
2763 | return ptr; |
2764 | } |
2765 | |
2766 | +#if HAVE_MBRTOWC |
2767 | +static char * |
2768 | +limfield_mb (const struct line *line, const struct keyfield *key) |
2769 | +{ |
2770 | + char *ptr = line->text, *lim = ptr + line->length - 1; |
2771 | + size_t eword = key->eword, echar = key->echar; |
2772 | + int i; |
2773 | + size_t mblength; |
2774 | + mbstate_t state; |
2775 | + |
2776 | + memset (&state, '\0', sizeof(mbstate_t)); |
2777 | + |
2778 | + if (!tab_default) |
2779 | + while (ptr < lim && eword--) |
2780 | + { |
2781 | + while (ptr < lim && memcmp (ptr, tab, tab_length) != 0) |
2782 | + { |
2783 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2784 | + ptr += mblength; |
2785 | + } |
2786 | + if (ptr < lim) |
2787 | + { |
2788 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2789 | + ptr += mblength; |
2790 | + } |
2791 | + } |
2792 | + else |
2793 | + while (ptr < lim && eword--) |
2794 | + { |
2795 | + while (ptr < lim && ismbblank (ptr, &mblength)) |
2796 | + ptr += mblength; |
2797 | + if (ptr < lim) |
2798 | + { |
2799 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2800 | + ptr += mblength; |
2801 | + } |
2802 | + while (ptr < lim && !ismbblank (ptr, &mblength)) |
2803 | + ptr += mblength; |
2804 | + } |
2805 | + |
2806 | +# ifdef POSIX_UNSPECIFIED |
2807 | + |
2808 | + /* Make LIM point to the end of (one byte past) the current field. */ |
2809 | + if (!tab_default) |
2810 | + { |
2811 | + char *newlim, *p; |
2812 | + |
2813 | + newlim = NULL; |
2814 | + for (p = ptr; p < lim;) |
2815 | + { |
2816 | + if (memcmp (p, tab, tab_length) == 0) |
2817 | + { |
2818 | + newlim = p; |
2819 | + break; |
2820 | + } |
2821 | + |
2822 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2823 | + p += mblength; |
2824 | + } |
2825 | + } |
2826 | + else |
2827 | + { |
2828 | + char *newlim; |
2829 | + newlim = ptr; |
2830 | + |
2831 | + while (newlim < lim && ismbblank (newlim, &mblength)) |
2832 | + newlim += mblength; |
2833 | + if (ptr < lim) |
2834 | + { |
2835 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2836 | + ptr += mblength; |
2837 | + } |
2838 | + while (newlim < lim && !ismbblank (newlim, &mblength)) |
2839 | + newlim += mblength; |
2840 | + lim = newlim; |
2841 | + } |
2842 | +# endif |
2843 | + |
2844 | + /* If we're skipping leading blanks, don't start counting characters |
2845 | + until after skipping past any leading blanks. */ |
2846 | + if (key->skipeblanks) |
2847 | + while (ptr < lim && ismbblank (ptr, &mblength)) |
2848 | + ptr += mblength; |
2849 | + |
2850 | + memset (&state, '\0', sizeof(mbstate_t)); |
2851 | + |
2852 | + /* Advance PTR by ECHAR (if possible), but no further than LIM. */ |
2853 | + for (i = 0; i < echar; i++) |
2854 | + { |
2855 | + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); |
2856 | + |
2857 | + if (ptr + mblength > lim) |
2858 | + break; |
2859 | + else |
2860 | + ptr += mblength; |
2861 | + } |
2862 | + |
2863 | + return ptr; |
2864 | +} |
2865 | +#endif |
2866 | + |
2867 | /* Fill BUF reading from FP, moving buf->left bytes from the end |
2868 | of buf->buf to the beginning first. If EOF is reached and the |
2869 | file wasn't terminated by a newline, supply one. Set up BUF's line |
2870 | @@ -1023,8 +1342,22 @@ fillbuf (struct buffer *buf, register FI |
2871 | else |
2872 | { |
2873 | if (key->skipsblanks) |
2874 | - while (blanks[to_uchar (*line_start)]) |
2875 | - line_start++; |
2876 | + { |
2877 | +#if HAVE_MBRTOWC |
2878 | + if (MB_CUR_MAX > 1) |
2879 | + { |
2880 | + size_t mblength; |
2881 | + |
2882 | + while (ismbblank (line_start, &mblength)) |
2883 | + line_start += mblength; |
2884 | + } |
2885 | + else |
2886 | +#endif |
2887 | + { |
2888 | + while (blanks[to_uchar (*line_start)]) |
2889 | + line_start++; |
2890 | + } |
2891 | + } |
2892 | line->keybeg = line_start; |
2893 | } |
2894 | } |
2895 | @@ -1130,10 +1463,27 @@ numcompare (register const char *a, regi |
2896 | size_t log_a; |
2897 | size_t log_b; |
2898 | |
2899 | - while (blanks[to_uchar (tmpa = *a)]) |
2900 | - a++; |
2901 | - while (blanks[to_uchar (tmpb = *b)]) |
2902 | - b++; |
2903 | +#if HAVE_MBRTOWC |
2904 | + if (MB_CUR_MAX > 1) |
2905 | + { |
2906 | + size_t mblength; |
2907 | + |
2908 | + while (ismbblank (a, &mblength)) |
2909 | + a += mblength; |
2910 | + while (ismbblank (b, &mblength)) |
2911 | + b += mblength; |
2912 | + |
2913 | + tmpa = *a; |
2914 | + tmpb = *b; |
2915 | + } |
2916 | + else |
2917 | +#endif |
2918 | + { |
2919 | + while (blanks[to_uchar (tmpa = *a)]) |
2920 | + a++; |
2921 | + while (blanks[to_uchar (tmpb = *b)]) |
2922 | + b++; |
2923 | + } |
2924 | |
2925 | if (tmpa == NEGATION_SIGN) |
2926 | { |
2927 | @@ -1263,15 +1613,59 @@ general_numcompare (const char *sa, cons |
2928 | /* FIXME: maybe add option to try expensive FP conversion |
2929 | only if A and B can't be compared more cheaply/accurately. */ |
2930 | |
2931 | - char *ea; |
2932 | - char *eb; |
2933 | - double a = strtod (sa, &ea); |
2934 | - double b = strtod (sb, &eb); |
2935 | + char *bufa, *ea; |
2936 | + char *bufb, *eb; |
2937 | + double a; |
2938 | + double b; |
2939 | + |
2940 | + char *p; |
2941 | + struct lconv *lconvp = localeconv (); |
2942 | + size_t thousands_sep_len = strlen (lconvp->thousands_sep); |
2943 | + |
2944 | + bufa = (char *) xmalloc (strlen (sa) + 1); |
2945 | + bufb = (char *) xmalloc (strlen (sb) + 1); |
2946 | + strcpy (bufa, sa); |
2947 | + strcpy (bufb, sb); |
2948 | + |
2949 | + if (force_general_numcompare) |
2950 | + { |
2951 | + while (1) |
2952 | + { |
2953 | + a = strtod (bufa, &ea); |
2954 | + if (memcmp (ea, lconvp->thousands_sep, thousands_sep_len) == 0) |
2955 | + { |
2956 | + for (p = ea; *(p + thousands_sep_len) != '\0'; p++) |
2957 | + *p = *(p + thousands_sep_len); |
2958 | + *p = '\0'; |
2959 | + continue; |
2960 | + } |
2961 | + break; |
2962 | + } |
2963 | + |
2964 | + while (1) |
2965 | + { |
2966 | + b = strtod (bufb, &eb); |
2967 | + if (memcmp (eb, lconvp->thousands_sep, thousands_sep_len) == 0) |
2968 | + { |
2969 | + for (p = eb; *(p + thousands_sep_len) != '\0'; p++) |
2970 | + *p = *(p + thousands_sep_len); |
2971 | + *p = '\0'; |
2972 | + continue; |
2973 | + } |
2974 | + break; |
2975 | + } |
2976 | + } |
2977 | + else |
2978 | + { |
2979 | + a = strtod (bufa, &ea); |
2980 | + b = strtod (bufb, &eb); |
2981 | + } |
2982 | + |
2983 | |
2984 | /* Put conversion errors at the start of the collating sequence. */ |
2985 | - if (sa == ea) |
2986 | - return sb == eb ? 0 : -1; |
2987 | - if (sb == eb) |
2988 | + if (bufa == ea) |
2989 | + return bufb == eb ? 0 : -1; |
2990 | + if (bufb == eb) |
2991 | return 1; |
2992 | |
2993 | /* Sort numbers in the usual way, where -0 == +0. Put NaNs after |
2994 | @@ -1289,7 +1683,7 @@ general_numcompare (const char *sa, cons |
2995 | Return 0 if the name in S is not recognized. */ |
2996 | |
2997 | static int |
2998 | -getmonth (char const *month, size_t len) |
2999 | +getmonth_uni (char const *month, size_t len) |
3000 | { |
3001 | size_t lo = 0; |
3002 | size_t hi = MONTHS_PER_YEAR; |
3003 | @@ -1331,11 +1725,79 @@ getmonth (char const *month, size_t len) |
3004 | return 0; |
3005 | } |
3006 | |
3007 | +#if HAVE_MBRTOWC |
3008 | +static int |
3009 | +getmonth_mb (char const *s, size_t len) |
3010 | +{ |
3011 | + char *month; |
3012 | + register size_t i; |
3013 | + register int lo = 0, hi = MONTHS_PER_YEAR, result; |
3014 | + char *tmp; |
3015 | + size_t wclength, mblength; |
3016 | + const char **pp; |
3017 | + const wchar_t **wpp; |
3018 | + wchar_t *month_wcs; |
3019 | + mbstate_t state; |
3020 | + |
3021 | + while (len > 0 && ismbblank (s, &mblength)) |
3022 | + { |
3023 | + s += mblength; |
3024 | + len -= mblength; |
3025 | + } |
3026 | + |
3027 | + if (len == 0) |
3028 | + return 0; |
3029 | + |
3030 | + month = (char *) alloca (len + 1); |
3031 | + |
3032 | + tmp = (char *) alloca (len + 1); |
3033 | + memcpy (tmp, s, len); |
3034 | + tmp[len] = '\0'; |
3035 | + pp = (const char **)&tmp; |
3036 | + month_wcs = (wchar_t *) alloca ((len + 1) * sizeof (wchar_t)); |
3037 | + memset (&state, '\0', sizeof(mbstate_t)); |
3038 | + |
3039 | + wclength = mbsrtowcs (month_wcs, pp, len + 1, &state); |
3040 | + assert (wclength != (size_t)1 && *pp == NULL); |
3041 | + |
3042 | + for (i = 0; i < wclength; i++) |
3043 | + { |
3044 | + month_wcs[i] = towupper(month_wcs[i]); |
3045 | + if (iswctype (month_wcs[i], blank_type)) |
3046 | + { |
3047 | + month_wcs[i] = L'\0'; |
3048 | + break; |
3049 | + } |
3050 | + } |
3051 | + |
3052 | + wpp = (const wchar_t **)&month_wcs; |
3053 | + |
3054 | + mblength = wcsrtombs (month, wpp, len + 1, &state); |
3055 | + assert (mblength != (-1) && *wpp == NULL); |
3056 | + |
3057 | + do |
3058 | + { |
3059 | + int ix = (lo + hi) / 2; |
3060 | + |
3061 | + if (strncmp (month, monthtab[ix].name, strlen (monthtab[ix].name)) < 0) |
3062 | + hi = ix; |
3063 | + else |
3064 | + lo = ix; |
3065 | + } |
3066 | + while (hi - lo > 1); |
3067 | + |
3068 | + result = (!strncmp (month, monthtab[lo].name, strlen (monthtab[lo].name)) |
3069 | + ? monthtab[lo].val : 0); |
3070 | + |
3071 | + return result; |
3072 | +} |
3073 | +#endif |
3074 | + |
3075 | /* Compare two lines A and B trying every key in sequence until there |
3076 | are no more keys or a difference is found. */ |
3077 | |
3078 | static int |
3079 | -keycompare (const struct line *a, const struct line *b) |
3080 | +keycompare_uni (const struct line *a, const struct line *b) |
3081 | { |
3082 | struct keyfield const *key = keylist; |
3083 | |
3084 | @@ -1499,11 +1961,188 @@ keycompare (const struct line *a, const |
3085 | |
3086 | return 0; |
3087 | |
3088 | - greater: |
3089 | +greater: |
3090 | + diff = 1; |
3091 | +not_equal: |
3092 | + return key->reverse ? -diff : diff; |
3093 | +} |
3094 | + |
3095 | +#if HAVE_MBRTOWC |
3096 | +static int |
3097 | +keycompare_mb (const struct line *a, const struct line *b) |
3098 | +{ |
3099 | + struct keyfield *key = keylist; |
3100 | + |
3101 | + /* For the first iteration only, the key positions have been |
3102 | + precomputed for us. */ |
3103 | + char *texta = a->keybeg; |
3104 | + char *textb = b->keybeg; |
3105 | + char *lima = a->keylim; |
3106 | + char *limb = b->keylim; |
3107 | + |
3108 | + size_t mblength_a, mblength_b; |
3109 | + wchar_t wc_a, wc_b; |
3110 | + mbstate_t state_a, state_b; |
3111 | + |
3112 | + int diff; |
3113 | + |
3114 | + memset (&state_a, '\0', sizeof(mbstate_t)); |
3115 | + memset (&state_b, '\0', sizeof(mbstate_t)); |
3116 | + |
3117 | + for (;;) |
3118 | + { |
3119 | + register char const *translate = key->translate; |
3120 | + register bool const *ignore = key->ignore; |
3121 | + |
3122 | + /* Find the lengths. */ |
3123 | + size_t lena = lima <= texta ? 0 : lima - texta; |
3124 | + size_t lenb = limb <= textb ? 0 : limb - textb; |
3125 | + |
3126 | + /* Actually compare the fields. */ |
3127 | + if (key->numeric | key->general_numeric) |
3128 | + { |
3129 | + char savea = *lima, saveb = *limb; |
3130 | + |
3131 | + *lima = *limb = '\0'; |
3132 | + if (force_general_numcompare) |
3133 | + diff = general_numcompare (texta, textb); |
3134 | + else |
3135 | + diff = ((key->numeric ? numcompare : general_numcompare) |
3136 | + (texta, textb)); |
3137 | + *lima = savea, *limb = saveb; |
3138 | + } |
3139 | + else if (key->month) |
3140 | + diff = getmonth (texta, lena) - getmonth (textb, lenb); |
3141 | + else |
3142 | + { |
3143 | + if (ignore || translate) |
3144 | + { |
3145 | + char buf[4000]; |
3146 | + size_t size = lena + 1 + lenb + 1; |
3147 | + char *copy_a = (size <= sizeof buf ? buf : xmalloc (size)); |
3148 | + char *copy_b = copy_a + lena + 1; |
3149 | + size_t new_len_a, new_len_b; |
3150 | + size_t i, j; |
3151 | + |
3152 | + /* Ignore and/or translate chars before comparing. */ |
3153 | +# define IGNORE_CHARS(NEW_LEN, LEN, TEXT, COPY, WC, MBLENGTH, STATE) \ |
3154 | + do \ |
3155 | + { \ |
3156 | + wchar_t uwc; \ |
3157 | + char mbc[MB_LEN_MAX]; \ |
3158 | + mbstate_t state_wc; \ |
3159 | + \ |
3160 | + for (NEW_LEN = i = 0; i < LEN;) \ |
3161 | + { \ |
3162 | + mbstate_t state_bak; \ |
3163 | + \ |
3164 | + state_bak = STATE; \ |
3165 | + MBLENGTH = mbrtowc (&WC, TEXT + i, LEN - i, &STATE); \ |
3166 | + \ |
3167 | + if (MBLENGTH == (size_t)-2 || MBLENGTH == (size_t)-1 \ |
3168 | + || MBLENGTH == 0) \ |
3169 | + { \ |
3170 | + if (MBLENGTH == (size_t)-2 || MBLENGTH == (size_t)-1) \ |
3171 | + STATE = state_bak; \ |
3172 | + if (!ignore) \ |
3173 | + COPY[NEW_LEN++] = TEXT[i++]; \ |
3174 | + continue; \ |
3175 | + } \ |
3176 | + \ |
3177 | + if (ignore) \ |
3178 | + { \ |
3179 | + if ((ignore == nonprinting && !iswprint (WC)) \ |
3180 | + || (ignore == nondictionary \ |
3181 | + && !iswalnum (WC) && !iswctype (WC, blank_type))) \ |
3182 | + { \ |
3183 | + i += MBLENGTH; \ |
3184 | + continue; \ |
3185 | + } \ |
3186 | + } \ |
3187 | + \ |
3188 | + if (translate) \ |
3189 | + { \ |
3190 | + \ |
3191 | + uwc = toupper(WC); \ |
3192 | + if (WC == uwc) \ |
3193 | + { \ |
3194 | + memcpy (mbc, TEXT + i, MBLENGTH); \ |
3195 | + i += MBLENGTH; \ |
3196 | + } \ |
3197 | + else \ |
3198 | + { \ |
3199 | + i += MBLENGTH; \ |
3200 | + WC = uwc; \ |
3201 | + memset (&state_wc, '\0', sizeof (mbstate_t)); \ |
3202 | + \ |
3203 | + MBLENGTH = wcrtomb (mbc, WC, &state_wc); \ |
3204 | + assert (MBLENGTH != (size_t)-1 && MBLENGTH != 0); \ |
3205 | + } \ |
3206 | + \ |
3207 | + for (j = 0; j < MBLENGTH; j++) \ |
3208 | + COPY[NEW_LEN++] = mbc[j]; \ |
3209 | + } \ |
3210 | + else \ |
3211 | + for (j = 0; j < MBLENGTH; j++) \ |
3212 | + COPY[NEW_LEN++] = TEXT[i++]; \ |
3213 | + } \ |
3214 | + COPY[NEW_LEN] = '\0'; \ |
3215 | + } \ |
3216 | + while (0) |
3217 | + |
3218 | + IGNORE_CHARS (new_len_a, lena, texta, copy_a, |
3219 | + wc_a, mblength_a, state_a); |
3220 | + IGNORE_CHARS (new_len_b, lenb, textb, copy_b, |
3221 | + wc_b, mblength_b, state_b); |
3222 | + diff = xmemcoll (copy_a, new_len_a, copy_b, new_len_b); |
3223 | + |
3224 | + if (sizeof buf < size) |
3225 | + free (copy_a); |
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 != SIZE_MAX) |
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 != SIZE_MAX) |
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, &mblength_a)) |
3256 | + texta += mblength_a; |
3257 | + while (textb < limb && ismbblank (textb, &mblength_b)) |
3258 | + textb += mblength_b; |
3259 | + } |
3260 | + } |
3261 | + } |
3262 | + |
3263 | + return 0; |
3264 | + |
3265 | +greater: |
3266 | diff = 1; |
3267 | - not_equal: |
3268 | +not_equal: |
3269 | return key->reverse ? -diff : diff; |
3270 | } |
3271 | +#endif |
3272 | |
3273 | /* Compare two lines A and B, returning negative, zero, or positive |
3274 | depending on whether A compares less than, equal to, or greater than B. */ |
3275 | @@ -2243,6 +2882,11 @@ set_ordering (register const char *s, st |
3276 | break; |
3277 | case 'M': |
3278 | key->month = true; |
3279 | +#if HAVE_MBRTOWC |
3280 | + if (strcmp (setlocale (LC_CTYPE, NULL), setlocale (LC_TIME, NULL))) |
3281 | + error (0, 0, _("As LC_TIME differs from LC_CTYPE, the results may be strange.")); |
3282 | + inittables_mb (); |
3283 | +#endif |
3284 | break; |
3285 | case 'n': |
3286 | key->numeric = true; |
3287 | @@ -2296,7 +2940,7 @@ main (int argc, char **argv) |
3288 | atexit (close_stdout); |
3289 | |
3290 | hard_LC_COLLATE = hard_locale (LC_COLLATE); |
3291 | -#if HAVE_NL_LANGINFO |
3292 | +#if HAVE_LANGINFO_CODESET |
3293 | hard_LC_TIME = hard_locale (LC_TIME); |
3294 | #endif |
3295 | |
3296 | @@ -2309,14 +2953,40 @@ main (int argc, char **argv) |
3297 | add support for multibyte decimal points. */ |
3298 | decimal_point = locale->decimal_point[0]; |
3299 | if (! decimal_point || locale->decimal_point[1]) |
3300 | - decimal_point = '.'; |
3301 | + { |
3302 | + decimal_point = '.'; |
3303 | + if (locale->decimal_point[0] && locale->decimal_point[1]) |
3304 | + force_general_numcompare = 1; |
3305 | + } |
3306 | |
3307 | /* FIXME: add support for multibyte thousands separators. */ |
3308 | thousands_sep = *locale->thousands_sep; |
3309 | if (! thousands_sep || locale->thousands_sep[1]) |
3310 | - thousands_sep = CHAR_MAX + 1; |
3311 | + { |
3312 | + thousands_sep = CHAR_MAX + 1; |
3313 | + if (locale->thousands_sep[0] && locale->thousands_sep[1]) |
3314 | + force_general_numcompare = 1; |
3315 | + } |
3316 | } |
3317 | |
3318 | +#if HAVE_MBRTOWC |
3319 | + if (MB_CUR_MAX > 1) |
3320 | + { |
3321 | + blank_type = wctype ("blank"); |
3322 | + begfield = begfield_mb; |
3323 | + limfield = limfield_mb; |
3324 | + getmonth = getmonth_mb; |
3325 | + keycompare = keycompare_mb; |
3326 | + } |
3327 | + else |
3328 | +#endif |
3329 | + { |
3330 | + begfield = begfield_uni; |
3331 | + limfield = limfield_uni; |
3332 | + keycompare = keycompare_uni; |
3333 | + getmonth = getmonth_uni; |
3334 | + } |
3335 | + |
3336 | have_read_stdin = false; |
3337 | inittables (); |
3338 | |
3339 | @@ -2514,28 +3184,48 @@ main (int argc, char **argv) |
3340 | break; |
3341 | |
3342 | case 't': |
3343 | - { |
3344 | - char newtab = optarg[0]; |
3345 | - if (! newtab) |
3346 | + { |
3347 | + if (! optarg[0]) |
3348 | error (SORT_FAILURE, 0, _("empty tab")); |
3349 | + |
3350 | + strncpy (tab, optarg, MB_LEN_MAX); |
3351 | + |
3352 | if (optarg[1]) |
3353 | { |
3354 | - if (STREQ (optarg, "\\0")) |
3355 | - newtab = '\0'; |
3356 | + if (strcmp (optarg, "\\0") == 0) |
3357 | + tab[0] = '\0'; |
3358 | else |
3359 | { |
3360 | - /* Provoke with `sort -txx'. Complain about |
3361 | - "multi-character tab" instead of "multibyte tab", so |
3362 | - that the diagnostic's wording does not need to be |
3363 | - changed once multibyte characters are supported. */ |
3364 | - error (SORT_FAILURE, 0, _("multi-character tab `%s'"), |
3365 | - optarg); |
3366 | +#if HAVE_MBRTOWC |
3367 | + if (MB_CUR_MAX > 1) |
3368 | + { |
3369 | + wchar_t wc; |
3370 | + mbstate_t state; |
3371 | + |
3372 | + memset (&state, '\0', sizeof (mbstate_t)); |
3373 | + tab_length = mbrtowc (&wc, tab, MB_LEN_MAX, &state); |
3374 | + tab_length = (tab_length == (size_t)-1 |
3375 | + || tab_length == (size_t)-2 |
3376 | + || tab_length == 0) ? 1 : tab_length; |
3377 | + if (optarg[tab_length]) |
3378 | + { |
3379 | + error (SORT_FAILURE, 0, |
3380 | + _("multi-character tab `%s'"), optarg); |
3381 | + } |
3382 | + } |
3383 | + else |
3384 | +#endif |
3385 | + { |
3386 | + error (SORT_FAILURE, 0, _("multi-character tab `%s'"), |
3387 | + optarg); |
3388 | + } |
3389 | } |
3390 | } |
3391 | - if (tab != TAB_DEFAULT && tab != newtab) |
3392 | + if (!tab_default) |
3393 | error (SORT_FAILURE, 0, _("incompatible tabs")); |
3394 | - tab = newtab; |
3395 | + tab_default = false; |
3396 | } |
3397 | + |
3398 | break; |
3399 | |
3400 | case 'T': |
3401 | --- coreutils/src/unexpand.c |
3402 | +++ coreutils/src/unexpand.c |
3403 | @@ -39,12 +39,35 @@ |
3404 | #include <stdio.h> |
3405 | #include <getopt.h> |
3406 | #include <sys/types.h> |
3407 | + |
3408 | +/* Get mbstate_t, mbrtowc(), wcwidth() */ |
3409 | +#if HAVE_WCHAR_H |
3410 | +# include <wchar.h> |
3411 | +#endif |
3412 | +/* Get iswblank */ |
3413 | +#if HAVE_WCTYPE_H |
3414 | +# include <wctype.h> |
3415 | +#endif |
3416 | + |
3417 | + |
3418 | +/* A sentinel value that's placed at the end of the list of tab stops. |
3419 | + * This value must be a large number, but not so large that adding the |
3420 | + * length of a line to it would cause the column variable to overflow. */ |
3421 | +#define TAB_STOP_SENTINEL INT_MAX |
3422 | + |
3423 | #include "system.h" |
3424 | #include "error.h" |
3425 | #include "posixver.h" |
3426 | #include "quote.h" |
3427 | #include "xstrndup.h" |
3428 | |
3429 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
3430 | + installation; work around this configuration error. */ |
3431 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
3432 | +# undef MB_LEN_MAX |
3433 | +# define MB_LEN_MAX 16 |
3434 | +#endif |
3435 | + |
3436 | /* The official name of this program (e.g., no `g' prefix). */ |
3437 | #define PROGRAM_NAME "unexpand" |
3438 | |
3439 | @@ -460,6 +483,236 @@ unexpand (void) |
3440 | } |
3441 | } |
3442 | |
3443 | +#if HAVE_MBRTOWC && HAVE_WCTYPE_H |
3444 | +static void |
3445 | +unexpand_multibyte (void) |
3446 | +{ |
3447 | + /* Input stream. */ |
3448 | + FILE *fp = next_file (NULL); |
3449 | + |
3450 | + mbstate_t i_state; /* Current shift state of the input stream. */ |
3451 | + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ |
3452 | + char *bufpos; /* Next read position of BUF. */ |
3453 | + size_t buflen = 0; /* The length of the byte sequence in buf. */ |
3454 | + |
3455 | + /* The array of pending blanks. In non-POSIX locales, blanks can |
3456 | + include characters other than spaces, so the blanks must be |
3457 | + stored, not merely counted. */ |
3458 | + char *pending_blank; |
3459 | + |
3460 | + if (!fp) |
3461 | + return; |
3462 | + |
3463 | + /* Binary I/O will preserve the original EOL style (DOS/Unix) of files. */ |
3464 | + SET_BINARY2 (fileno (fp), STDOUT_FILENO); |
3465 | + |
3466 | + /* The worst case is a non-blank character, then one blank, then a |
3467 | + tab stop, then MAX_COLUMN_WIDTH - 1 blanks, then a non-blank; so |
3468 | + allocate MAX_COLUMN_WIDTH bytes to store the blanks. */ |
3469 | + pending_blank = xmalloc (max_column_width); |
3470 | + |
3471 | + memset (&i_state, '\0', sizeof(mbstate_t)); |
3472 | + |
3473 | + for (;;) |
3474 | + { |
3475 | + /* A gotten wide character. */ |
3476 | + wint_t wc; |
3477 | + |
3478 | + /* If true, perform translations. */ |
3479 | + bool convert = true; |
3480 | + |
3481 | + /* The following variables have valid values only when CONVERT |
3482 | + is true: */ |
3483 | + |
3484 | + /* Column of next input character. */ |
3485 | + uintmax_t column = 0; |
3486 | + |
3487 | + /* Column the next input tab stop is on. */ |
3488 | + uintmax_t next_tab_column = 0; |
3489 | + |
3490 | + /* Index in TAB_LIST of next tab stop to examine. */ |
3491 | + size_t tab_index = 0; |
3492 | + |
3493 | + /* If true, the first pending blank came just before a tab stop. */ |
3494 | + bool one_blank_before_tab_stop = false; |
3495 | + |
3496 | + /* If true, the previous input character was a blank. This is |
3497 | + initially true, since initial strings of blanks are treated |
3498 | + as if the line was preceded by a blank. */ |
3499 | + bool prev_blank = true; |
3500 | + |
3501 | + /* Number of pending columns of blanks. */ |
3502 | + size_t pending = 0; |
3503 | + |
3504 | + /* Convert a line of text. */ |
3505 | + do |
3506 | + { |
3507 | + wchar_t w; |
3508 | + size_t mblength; /* The byte size of a multibyte character |
3509 | + which shows as same character as WC. */ |
3510 | + mbstate_t i_state_bak; /* Back up the I_STATE. */ |
3511 | + |
3512 | + /* Fill buffer */ |
3513 | + if (buflen < MB_LEN_MAX) |
3514 | + { |
3515 | + if (!feof(fp) && !ferror(fp)) { |
3516 | + if (buflen > 0) memmove(buf, bufpos, buflen); |
3517 | + buflen += fread(buf + buflen, sizeof(char), BUFSIZ, fp); |
3518 | + bufpos = buf; |
3519 | + } |
3520 | + } |
3521 | + |
3522 | + if (buflen < 1) { |
3523 | + /* Move to the next file */ |
3524 | + if (feof(fp) || ferror(fp)) { |
3525 | + fp = next_file(fp); |
3526 | + } |
3527 | + if (!fp) { |
3528 | + if (pending) |
3529 | + { |
3530 | + if (fwrite (pending_blank, 1, pending, stdout) != pending) |
3531 | + error (EXIT_FAILURE, errno, _("write error")); |
3532 | + } |
3533 | + free (pending_blank); |
3534 | + return; |
3535 | + } |
3536 | + SET_BINARY2 (fileno (fp), STDOUT_FILENO); |
3537 | + continue; |
3538 | + } |
3539 | + |
3540 | + i_state_bak = i_state; |
3541 | + mblength = mbrtowc (&w, bufpos, buflen, &i_state); |
3542 | + wc = w; |
3543 | + |
3544 | + if (mblength == (size_t) -1 || mblength == (size_t) -2) { |
3545 | + i_state = i_state_bak; |
3546 | + wc = L'\0'; |
3547 | + column += convert; |
3548 | + mblength = 1; |
3549 | + } |
3550 | + |
3551 | + if (convert) |
3552 | + { |
3553 | + bool blank = iswblank (wc); |
3554 | + |
3555 | + if (blank) |
3556 | + { |
3557 | + if (next_tab_column <= column) |
3558 | + { |
3559 | + if (tab_size) |
3560 | + next_tab_column = |
3561 | + column + (tab_size - column % tab_size); |
3562 | + else |
3563 | + for (;;) |
3564 | + if (tab_index == first_free_tab) |
3565 | + { |
3566 | + convert = false; |
3567 | + break; |
3568 | + } |
3569 | + else |
3570 | + { |
3571 | + uintmax_t tab = tab_list[tab_index++]; |
3572 | + if (column < tab) |
3573 | + { |
3574 | + next_tab_column = tab; |
3575 | + break; |
3576 | + } |
3577 | + } |
3578 | + } |
3579 | + |
3580 | + if (convert) |
3581 | + { |
3582 | + if (next_tab_column < column) |
3583 | + error (EXIT_FAILURE, 0, _("input line is too long")); |
3584 | + |
3585 | + if (wc == L'\t') |
3586 | + { |
3587 | + column = next_tab_column; |
3588 | + |
3589 | + /* Discard pending blanks, unless it was a single |
3590 | + blank just before the previous tab stop. */ |
3591 | + if (! (pending == 1 && one_blank_before_tab_stop)) |
3592 | + { |
3593 | + pending = 0; |
3594 | + one_blank_before_tab_stop = false; |
3595 | + } |
3596 | + } |
3597 | + else |
3598 | + { |
3599 | + column++; |
3600 | + |
3601 | + if (! (prev_blank && column == next_tab_column)) |
3602 | + { |
3603 | + /* It is not yet known whether the pending blanks |
3604 | + will be replaced by tabs. */ |
3605 | + if (column == next_tab_column) |
3606 | + one_blank_before_tab_stop = true; |
3607 | + pending_blank[pending++] = ' '; |
3608 | + prev_blank = true; |
3609 | + buflen -= mblength; |
3610 | + bufpos += mblength; |
3611 | + continue; |
3612 | + } |
3613 | + |
3614 | + /* Replace the pending blanks by a tab or two. */ |
3615 | + pending_blank[0] = *bufpos = '\t'; |
3616 | + pending = one_blank_before_tab_stop; |
3617 | + } |
3618 | + } |
3619 | + } |
3620 | + else if (wc == L'\b') |
3621 | + { |
3622 | + /* Go back one column, and force recalculation of the |
3623 | + next tab stop. */ |
3624 | + column -= !!column; |
3625 | + next_tab_column = column; |
3626 | + tab_index -= !!tab_index; |
3627 | + } |
3628 | + else |
3629 | + { |
3630 | + if (!iswcntrl (wc)) |
3631 | + { |
3632 | + int width = wcwidth (wc); |
3633 | + if (width > 0) { |
3634 | + if (column > (column + width)) |
3635 | + error (EXIT_FAILURE, 0, _("input line is too long")); |
3636 | + column += width; |
3637 | + } |
3638 | + } |
3639 | + } |
3640 | + |
3641 | + if (pending) |
3642 | + { |
3643 | + if (fwrite (pending_blank, 1, pending, stdout) != pending) |
3644 | + error (EXIT_FAILURE, errno, _("write error")); |
3645 | + pending = 0; |
3646 | + one_blank_before_tab_stop = false; |
3647 | + } |
3648 | + |
3649 | + prev_blank = blank; |
3650 | + convert &= convert_entire_line | blank; |
3651 | + } |
3652 | + |
3653 | + if (mblength) |
3654 | + { |
3655 | + if (fwrite (bufpos, sizeof(char), mblength, stdout) < mblength) |
3656 | + error (EXIT_FAILURE, errno, _("write error")); |
3657 | + } |
3658 | + else |
3659 | + { |
3660 | + if (putchar('\0')) |
3661 | + error (EXIT_FAILURE, errno, _("write error")); |
3662 | + mblength = 1; |
3663 | + } |
3664 | + |
3665 | + buflen -= mblength; |
3666 | + bufpos += mblength; |
3667 | + } |
3668 | + while (wc != L'\n'); |
3669 | + } |
3670 | +} |
3671 | +#endif |
3672 | + |
3673 | int |
3674 | main (int argc, char **argv) |
3675 | { |
3676 | @@ -548,7 +801,12 @@ main (int argc, char **argv) |
3677 | |
3678 | file_list = (optind < argc ? &argv[optind] : stdin_argv); |
3679 | |
3680 | - unexpand (); |
3681 | +#if HAVE_MBRTOWC |
3682 | + if (MB_CUR_MAX > 1) |
3683 | + unexpand_multibyte (); |
3684 | + else |
3685 | +#endif |
3686 | + unexpand (); |
3687 | |
3688 | if (have_read_stdin && fclose (stdin) != 0) |
3689 | error (EXIT_FAILURE, errno, "-"); |
3690 | --- coreutils/src/uniq.c |
3691 | +++ coreutils/src/uniq.c |
3692 | @@ -23,6 +23,16 @@ |
3693 | #include <getopt.h> |
3694 | #include <sys/types.h> |
3695 | |
3696 | +/* Get mbstate_t, mbrtowc(), wcrtomb() */ |
3697 | +#if HAVE_WCHAR_H |
3698 | +# include <wchar.h> |
3699 | +#endif |
3700 | + |
3701 | +/* Get iswctype(), wctype(), towupper)(. */ |
3702 | +#if HAVE_WCTYPE_H |
3703 | +# include <wctype.h> |
3704 | +#endif |
3705 | + |
3706 | #include "system.h" |
3707 | #include "argmatch.h" |
3708 | #include "linebuffer.h" |
3709 | @@ -34,6 +44,13 @@ |
3710 | #include "xstrtol.h" |
3711 | #include "memcasecmp.h" |
3712 | |
3713 | +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC |
3714 | + installation; work around this configuration error. */ |
3715 | +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 |
3716 | +# undef MB_LEN_MAX |
3717 | +# define MB_LEN_MAX 16 |
3718 | +#endif |
3719 | + |
3720 | /* The official name of this program (e.g., no `g' prefix). */ |
3721 | #define PROGRAM_NAME "uniq" |
3722 | |
3723 | @@ -109,6 +126,12 @@ static enum delimit_method const delimit |
3724 | /* Select whether/how to delimit groups of duplicate lines. */ |
3725 | static enum delimit_method delimit_groups; |
3726 | |
3727 | +/* Function pointers. */ |
3728 | +static char * (*find_field) (struct linebuffer *line); |
3729 | + |
3730 | +/* Show the blank character class. */ |
3731 | +wctype_t blank_type; |
3732 | + |
3733 | static struct option const longopts[] = |
3734 | { |
3735 | {"count", no_argument, NULL, 'c'}, |
3736 | @@ -189,7 +212,7 @@ size_opt (char const *opt, char const *m |
3737 | return a pointer to the beginning of the line's field to be compared. */ |
3738 | |
3739 | static char * |
3740 | -find_field (const struct linebuffer *line) |
3741 | +find_field_uni (struct linebuffer *line) |
3742 | { |
3743 | register size_t count; |
3744 | register char *lp = line->buffer; |
3745 | @@ -210,6 +233,83 @@ find_field (const struct linebuffer *lin |
3746 | return lp + i; |
3747 | } |
3748 | |
3749 | +#if HAVE_MBRTOWC |
3750 | + |
3751 | +# define MBCHAR_TO_WCHAR(WC, MBLENGTH, LP, POS, SIZE, STATEP, CONVFAIL) \ |
3752 | + do \ |
3753 | + { \ |
3754 | + mbstate_t state_bak; \ |
3755 | + \ |
3756 | + CONVFAIL = 0; \ |
3757 | + state_bak = *STATEP; \ |
3758 | + \ |
3759 | + MBLENGTH = mbrtowc (&WC, LP + POS, SIZE - POS, STATEP); \ |
3760 | + \ |
3761 | + switch (MBLENGTH) \ |
3762 | + { \ |
3763 | + case (size_t)-2: \ |
3764 | + case (size_t)-1: \ |
3765 | + *STATEP = state_bak; \ |
3766 | + CONVFAIL++; \ |
3767 | + /* Fall through */ \ |
3768 | + case 0: \ |
3769 | + MBLENGTH = 1; \ |
3770 | + } \ |
3771 | + } \ |
3772 | + while (0) |
3773 | + |
3774 | +static char * |
3775 | +find_field_multi (struct linebuffer *line) |
3776 | +{ |
3777 | + size_t count; |
3778 | + char *lp = line->buffer; |
3779 | + size_t size = line->length - 1; |
3780 | + size_t pos; |
3781 | + size_t mblength; |
3782 | + wchar_t wc; |
3783 | + mbstate_t *statep; |
3784 | + int convfail; |
3785 | + |
3786 | + pos = 0; |
3787 | + statep = &(line->state); |
3788 | + |
3789 | + /* skip fields. */ |
3790 | + for (count = 0; count < skip_fields && pos < size; count++) |
3791 | + { |
3792 | + while (pos < size) |
3793 | + { |
3794 | + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); |
3795 | + |
3796 | + if (convfail || !iswctype (wc, blank_type)) |
3797 | + { |
3798 | + pos += mblength; |
3799 | + break; |
3800 | + } |
3801 | + pos += mblength; |
3802 | + } |
3803 | + |
3804 | + while (pos < size) |
3805 | + { |
3806 | + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); |
3807 | + |
3808 | + if (!convfail && iswctype (wc, blank_type)) |
3809 | + break; |
3810 | + |
3811 | + pos += mblength; |
3812 | + } |
3813 | + } |
3814 | + |
3815 | + /* skip fields. */ |
3816 | + for (count = 0; count < skip_chars && pos < size; count++) |
3817 | + { |
3818 | + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); |
3819 | + pos += mblength; |
3820 | + } |
3821 | + |
3822 | + return lp + pos; |
3823 | +} |
3824 | +#endif |
3825 | + |
3826 | /* Return false if two strings OLD and NEW match, true if not. |
3827 | OLD and NEW point not to the beginnings of the lines |
3828 | but rather to the beginnings of the fields to compare. |
3829 | @@ -234,6 +334,73 @@ different (char *old, char *new, size_t |
3830 | return oldlen != newlen || memcmp (old, new, oldlen); |
3831 | } |
3832 | |
3833 | +#if HAVE_MBRTOWC |
3834 | +static int |
3835 | +different_multi (const char *old, const char *new, size_t oldlen, size_t newlen, mbstate_t oldstate, mbstate_t newstate) |
3836 | +{ |
3837 | + size_t i, j, chars; |
3838 | + const char *str[2]; |
3839 | + char *copy[2]; |
3840 | + size_t len[2]; |
3841 | + mbstate_t state[2]; |
3842 | + size_t mblength; |
3843 | + wchar_t wc, uwc; |
3844 | + mbstate_t state_bak; |
3845 | + |
3846 | + str[0] = old; |
3847 | + str[1] = new; |
3848 | + len[0] = oldlen; |
3849 | + len[1] = newlen; |
3850 | + state[0] = oldstate; |
3851 | + state[1] = newstate; |
3852 | + |
3853 | + for (i = 0; i < 2; i++) |
3854 | + { |
3855 | + copy[i] = alloca (len[i] + 1); |
3856 | + |
3857 | + for (j = 0, chars = 0; j < len[i] && chars < check_chars; chars++) |
3858 | + { |
3859 | + state_bak = state[i]; |
3860 | + mblength = mbrtowc (&wc, str[i] + j, len[i] - j, &(state[i])); |
3861 | + |
3862 | + switch (mblength) |
3863 | + { |
3864 | + case (size_t)-1: |
3865 | + case (size_t)-2: |
3866 | + state[i] = state_bak; |
3867 | + /* Fall through */ |
3868 | + case 0: |
3869 | + mblength = 1; |
3870 | + break; |
3871 | + |
3872 | + default: |
3873 | + if (ignore_case) |
3874 | + { |
3875 | + uwc = towupper (wc); |
3876 | + |
3877 | + if (uwc != wc) |
3878 | + { |
3879 | + mbstate_t state_wc; |
3880 | + |
3881 | + memset (&state_wc, '\0', sizeof(mbstate_t)); |
3882 | + wcrtomb (copy[i] + j, uwc, &state_wc); |
3883 | + } |
3884 | + else |
3885 | + memcpy (copy[i] + j, str[i] + j, mblength); |
3886 | + } |
3887 | + else |
3888 | + memcpy (copy[i] + j, str[i] + j, mblength); |
3889 | + } |
3890 | + j += mblength; |
3891 | + } |
3892 | + copy[i][j] = '\0'; |
3893 | + len[i] = j; |
3894 | + } |
3895 | + |
3896 | + return xmemcoll (copy[0], len[0], copy[1], len[1]); |
3897 | +} |
3898 | +#endif |
3899 | + |
3900 | /* Output the line in linebuffer LINE to stream STREAM |
3901 | provided that the switches say it should be output. |
3902 | MATCH is true if the line matches the previous line. |
3903 | @@ -297,15 +464,42 @@ check_file (const char *infile, const ch |
3904 | { |
3905 | char *prevfield IF_LINT (= NULL); |
3906 | size_t prevlen IF_LINT (= 0); |
3907 | +#if HAVE_MBRTOWC |
3908 | + mbstate_t prevstate; |
3909 | + |
3910 | + memset (&prevstate, '\0', sizeof (mbstate_t)); |
3911 | +#endif |
3912 | |
3913 | while (!feof (istream)) |
3914 | { |
3915 | char *thisfield; |
3916 | size_t thislen; |
3917 | +#if HAVE_MBRTOWC |
3918 | + mbstate_t thisstate; |
3919 | +#endif |
3920 | if (readlinebuffer (thisline, istream) == 0) |
3921 | break; |
3922 | thisfield = find_field (thisline); |
3923 | thislen = thisline->length - 1 - (thisfield - thisline->buffer); |
3924 | +#if HAVE_MBRTOWC |
3925 | + if (MB_CUR_MAX > 1) |
3926 | + { |
3927 | + thisstate = thisline->state; |
3928 | + |
3929 | + if (prevline->length == 0 || different_multi |
3930 | + (thisfield, prevfield, thislen, prevlen, thisstate, prevstate)) |
3931 | + { |
3932 | + fwrite (thisline->buffer, sizeof (char), |
3933 | + thisline->length, ostream); |
3934 | + |
3935 | + SWAP_LINES (prevline, thisline); |
3936 | + prevfield = thisfield; |
3937 | + prevlen = thislen; |
3938 | + prevstate = thisstate; |
3939 | + } |
3940 | + } |
3941 | + else |
3942 | +#endif |
3943 | if (prevline->length == 0 |
3944 | || different (thisfield, prevfield, thislen, prevlen)) |
3945 | { |
3946 | @@ -324,17 +518,26 @@ check_file (const char *infile, const ch |
3947 | size_t prevlen; |
3948 | uintmax_t match_count = 0; |
3949 | bool first_delimiter = true; |
3950 | +#if HAVE_MBRTOWC |
3951 | + mbstate_t prevstate; |
3952 | +#endif |
3953 | |
3954 | if (readlinebuffer (prevline, istream) == 0) |
3955 | goto closefiles; |
3956 | prevfield = find_field (prevline); |
3957 | prevlen = prevline->length - 1 - (prevfield - prevline->buffer); |
3958 | +#if HAVE_MBRTOWC |
3959 | + prevstate = prevline->state; |
3960 | +#endif |
3961 | |
3962 | while (!feof (istream)) |
3963 | { |
3964 | bool match; |
3965 | char *thisfield; |
3966 | size_t thislen; |
3967 | +#if HAVE_MBRTOWC |
3968 | + mbstate_t thisstate; |
3969 | +#endif |
3970 | if (readlinebuffer (thisline, istream) == 0) |
3971 | { |
3972 | if (ferror (istream)) |
3973 | @@ -343,6 +546,15 @@ check_file (const char *infile, const ch |
3974 | } |
3975 | thisfield = find_field (thisline); |
3976 | thislen = thisline->length - 1 - (thisfield - thisline->buffer); |
3977 | +#if HAVE_MBRTOWC |
3978 | + if (MB_CUR_MAX > 1) |
3979 | + { |
3980 | + thisstate = thisline->state; |
3981 | + match = !different_multi (thisfield, prevfield, |
3982 | + thislen, prevlen, thisstate, prevstate); |
3983 | + } |
3984 | + else |
3985 | +#endif |
3986 | match = !different (thisfield, prevfield, thislen, prevlen); |
3987 | match_count += match; |
3988 | |
3989 | @@ -375,6 +587,9 @@ check_file (const char *infile, const ch |
3990 | SWAP_LINES (prevline, thisline); |
3991 | prevfield = thisfield; |
3992 | prevlen = thislen; |
3993 | +#if HAVE_MBRTOWC |
3994 | + prevstate = thisstate; |
3995 | +#endif |
3996 | if (!match) |
3997 | match_count = 0; |
3998 | } |
3999 | @@ -420,6 +635,18 @@ main (int argc, char **argv) |
4000 | |
4001 | atexit (close_stdout); |
4002 | |
4003 | +#if HAVE_MBRTOWC |
4004 | + if (MB_CUR_MAX > 1) |
4005 | + { |
4006 | + find_field = find_field_multi; |
4007 | + blank_type = wctype ("blank"); |
4008 | + } |
4009 | + else |
4010 | +#endif |
4011 | + { |
4012 | + find_field = find_field_uni; |
4013 | + } |
4014 | + |
4015 | skip_chars = 0; |
4016 | skip_fields = 0; |
4017 | check_chars = SIZE_MAX; |