]>
Commit | Line | Data |
---|---|---|
6b98979f | 1 | /* Generate fastloading iconv module configuration files. |
b0c9067d | 2 | Copyright (C) 2000, 2001, 2002, 2003 Free Software Foundation, Inc. |
6b98979f UD |
3 | This file is part of the GNU C Library. |
4 | Contributed by Ulrich Drepper <drepper@redhat.com>, 2000. | |
5 | ||
6 | The GNU C Library is free software; you can redistribute it and/or | |
0588dfe7 AJ |
7 | modify it under the terms of the GNU Lesser General Public |
8 | License as published by the Free Software Foundation; either | |
9 | version 2.1 of the License, or (at your option) any later version. | |
6b98979f UD |
10 | |
11 | The GNU C Library is distributed in the hope that it will be useful, | |
12 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
0588dfe7 | 14 | Lesser General Public License for more details. |
6b98979f | 15 | |
0588dfe7 AJ |
16 | You should have received a copy of the GNU Lesser General Public |
17 | License along with the GNU C Library; if not, write to the Free | |
18 | Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA | |
19 | 02111-1307 USA. */ | |
6b98979f UD |
20 | |
21 | #include <argp.h> | |
22 | #include <assert.h> | |
23 | #include <error.h> | |
24 | #include <errno.h> | |
25 | #include <fcntl.h> | |
26 | #include <libintl.h> | |
27 | #include <locale.h> | |
28 | #include <mcheck.h> | |
29 | #include <search.h> | |
30 | #include <stdint.h> | |
31 | #include <stdio.h> | |
32 | #include <stdio_ext.h> | |
33 | #include <stdlib.h> | |
34 | #include <string.h> | |
35 | #include <unistd.h> | |
0588dfe7 | 36 | #include <sys/cdefs.h> |
6b98979f UD |
37 | #include <sys/uio.h> |
38 | ||
39 | #include "iconvconfig.h" | |
40 | ||
41 | /* Get libc version number. */ | |
42 | #include "../version.h" | |
43 | ||
44 | #define PACKAGE _libc_intl_domainname | |
45 | ||
46 | ||
47 | /* The hashing function we use. */ | |
48 | #include "../intl/hash-string.h" | |
49 | ||
50 | ||
51 | /* Types used. */ | |
52 | struct module | |
53 | { | |
54 | char *fromname; | |
55 | struct Strent *fromname_strent; | |
56 | char *filename; | |
57 | struct Strent *filename_strent; | |
58 | const char *directory; | |
59 | struct Strent *directory_strent; | |
60 | struct module *next; | |
61 | int cost; | |
62 | struct Strent *toname_strent; | |
63 | char toname[0]; | |
64 | }; | |
65 | ||
66 | struct alias | |
67 | { | |
68 | char *fromname; | |
69 | struct Strent *froment; | |
70 | struct module *module; | |
71 | struct Strent *toent; | |
72 | char toname[0]; | |
73 | }; | |
74 | ||
75 | struct name | |
76 | { | |
77 | const char *name; | |
78 | struct Strent *strent; | |
79 | int module_idx; | |
80 | uint32_t hashval; | |
81 | }; | |
82 | ||
83 | struct name_info | |
84 | { | |
85 | const char *canonical_name; | |
86 | struct Strent *canonical_strent; | |
87 | ||
88 | struct module *from_internal; | |
89 | struct module *to_internal; | |
90 | ||
91 | struct other_conv_list | |
92 | { | |
93 | int dest_idx; | |
94 | struct other_conv | |
95 | { | |
96 | gidx_t module_idx; | |
97 | struct module *module; | |
98 | struct other_conv *next; | |
99 | } other_conv; | |
100 | struct other_conv_list *next; | |
101 | } *other_conv_list; | |
102 | }; | |
103 | ||
104 | ||
105 | /* Name and version of program. */ | |
106 | static void print_version (FILE *stream, struct argp_state *state); | |
107 | void (*argp_program_version_hook) (FILE *, struct argp_state *) = print_version; | |
108 | ||
6b98979f UD |
109 | /* Short description of program. */ |
110 | static const char doc[] = N_("\ | |
111 | Create fastloading iconv module configuration file."); | |
112 | ||
113 | /* Strings for arguments in help texts. */ | |
114 | static const char args_doc[] = N_("[DIR...]"); | |
115 | ||
d8d5a647 UD |
116 | /* Prototype for option handler. */ |
117 | static error_t parse_opt (int key, char *arg, struct argp_state *state); | |
118 | ||
6b98979f UD |
119 | /* Function to print some extra text in the help message. */ |
120 | static char *more_help (int key, const char *text, void *input); | |
121 | ||
d8d5a647 UD |
122 | /* Definitions of arguments for argp functions. */ |
123 | #define OPT_PREFIX 300 | |
124 | static const struct argp_option options[] = | |
125 | { | |
126 | { "prefix", OPT_PREFIX, "PATH", 0, N_("Prefix used for all file accesses") }, | |
127 | { NULL, 0, NULL, 0, NULL } | |
128 | }; | |
129 | ||
6b98979f UD |
130 | /* Data structure to communicate with argp functions. */ |
131 | static struct argp argp = | |
132 | { | |
d8d5a647 UD |
133 | |
134 | options, parse_opt, args_doc, doc, NULL, more_help | |
6b98979f UD |
135 | }; |
136 | ||
137 | ||
138 | /* The function doing the actual work. */ | |
139 | static int handle_dir (const char *dir); | |
140 | ||
141 | /* Add all known builtin conversions and aliases. */ | |
142 | static void add_builtins (void); | |
143 | ||
144 | /* Create list of all aliases without circular aliases. */ | |
145 | static void get_aliases (void); | |
146 | ||
147 | /* Create list of all modules. */ | |
148 | static void get_modules (void); | |
149 | ||
150 | /* Get list of all the names and thereby indexing them. */ | |
151 | static void generate_name_list (void); | |
152 | ||
153 | /* Collect information about all the names. */ | |
154 | static void generate_name_info (void); | |
155 | ||
156 | /* Write the output file. */ | |
157 | static int write_output (void); | |
158 | ||
159 | ||
d8d5a647 UD |
160 | /* Prefix to be used for all file accesses. */ |
161 | static const char *prefix = ""; | |
162 | /* Its length. */ | |
163 | static size_t prefix_len; | |
164 | ||
6b98979f UD |
165 | /* Search tree of the modules we know. */ |
166 | static void *modules; | |
167 | ||
168 | /* Search tree of the aliases we know. */ | |
169 | static void *aliases; | |
170 | ||
171 | /* Search tree for name to index mapping. */ | |
172 | static void *names; | |
173 | ||
174 | /* Number of names we know about. */ | |
175 | static int nnames; | |
176 | ||
177 | /* List of all aliases. */ | |
178 | static struct alias **alias_list; | |
179 | static size_t nalias_list; | |
180 | static size_t nalias_list_max; | |
181 | ||
182 | /* List of all modules. */ | |
183 | static struct module **module_list; | |
184 | static size_t nmodule_list; | |
185 | static size_t nmodule_list_max; | |
186 | ||
187 | /* Names and information about them. */ | |
188 | static struct name_info *name_info; | |
189 | static size_t nname_info; | |
190 | ||
191 | /* Number of translations not from or to INTERNAL. */ | |
192 | static size_t nextra_modules; | |
193 | ||
194 | ||
195 | /* Names and aliases for the builtin transformations. */ | |
196 | static struct | |
197 | { | |
198 | const char *from; | |
199 | const char *to; | |
200 | } builtin_alias[] = | |
201 | { | |
202 | #define BUILTIN_ALIAS(alias, real) \ | |
203 | { .from = alias, .to = real }, | |
f9ad060c UD |
204 | #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, BtowcFct, \ |
205 | MinF, MaxF, MinT, MaxT) | |
6b98979f UD |
206 | #include <gconv_builtin.h> |
207 | }; | |
874d8461 UD |
208 | #undef BUILTIN_ALIAS |
209 | #undef BUILTIN_TRANSFORMATION | |
6b98979f UD |
210 | #define nbuiltin_alias (sizeof (builtin_alias) / sizeof (builtin_alias[0])) |
211 | ||
212 | static struct | |
213 | { | |
214 | const char *from; | |
215 | const char *to; | |
216 | const char *module; | |
217 | int cost; | |
218 | } builtin_trans[] = | |
219 | { | |
220 | #define BUILTIN_ALIAS(alias, real) | |
f9ad060c UD |
221 | #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, BtowcFct, \ |
222 | MinF, MaxF, MinT, MaxT) \ | |
6b98979f UD |
223 | { .from = From, .to = To, .module = Name, .cost = Cost }, |
224 | #include <gconv_builtin.h> | |
225 | }; | |
f9ad060c UD |
226 | #undef BUILTIN_ALIAS |
227 | #undef BUILTIN_TRANSFORMATION | |
6b98979f UD |
228 | #define nbuiltin_trans (sizeof (builtin_trans) / sizeof (builtin_trans[0])) |
229 | ||
230 | ||
231 | /* Filename extension for the modules. */ | |
232 | #ifndef MODULE_EXT | |
233 | # define MODULE_EXT ".so" | |
234 | #endif | |
235 | static const char gconv_module_ext[] = MODULE_EXT; | |
236 | ||
237 | ||
0588dfe7 AJ |
238 | extern void *xmalloc (size_t n) __attribute_malloc__; |
239 | extern void *xcalloc (size_t n, size_t m) __attribute_malloc__; | |
6b98979f UD |
240 | extern void *xrealloc (void *p, size_t n); |
241 | ||
242 | ||
243 | /* C string table handling. */ | |
244 | struct Strtab; | |
245 | struct Strent; | |
246 | ||
247 | /* Create new C string table object in memory. */ | |
248 | extern struct Strtab *strtabinit (void); | |
249 | ||
250 | /* Free resources allocated for C string table ST. */ | |
251 | extern void strtabfree (struct Strtab *st); | |
252 | ||
253 | /* Add string STR (length LEN is != 0) to C string table ST. */ | |
254 | extern struct Strent *strtabadd (struct Strtab *st, const char *str, | |
255 | size_t len); | |
256 | ||
257 | /* Finalize string table ST and store size in *SIZE and return a pointer. */ | |
258 | extern void *strtabfinalize (struct Strtab *st, size_t *size); | |
259 | ||
260 | /* Get offset in string table for string associated with SE. */ | |
261 | extern size_t strtaboffset (struct Strent *se); | |
262 | ||
263 | /* String table we construct. */ | |
264 | static struct Strtab *strtab; | |
265 | ||
266 | ||
267 | ||
268 | int | |
269 | main (int argc, char *argv[]) | |
270 | { | |
271 | int remaining; | |
272 | int status = 0; | |
273 | char *path; | |
274 | char *tp; | |
275 | ||
276 | /* Enable memory use testing. */ | |
47bd306e UD |
277 | /* mcheck_pedantic (NULL); */ |
278 | mtrace (); | |
6b98979f UD |
279 | |
280 | /* Set locale via LC_ALL. */ | |
281 | setlocale (LC_ALL, ""); | |
282 | ||
283 | /* Set the text message domain. */ | |
284 | textdomain (_libc_intl_domainname); | |
285 | ||
286 | /* Parse and process arguments. */ | |
287 | argp_parse (&argp, argc, argv, 0, &remaining, NULL); | |
288 | ||
289 | /* Initialize the string table. */ | |
290 | strtab = strtabinit (); | |
291 | ||
292 | /* Handle all directories mentioned. */ | |
293 | while (remaining < argc) | |
294 | status |= handle_dir (argv[remaining++]); | |
295 | ||
296 | /* In any case also handle the standard directory. */ | |
297 | path = strdupa (GCONV_PATH); | |
298 | tp = strtok (path, ":"); | |
299 | while (tp != NULL) | |
300 | { | |
301 | status |= handle_dir (tp); | |
302 | ||
303 | tp = strtok (NULL, ":"); | |
304 | } | |
305 | ||
306 | /* Add the builtin transformations and aliases without overwriting | |
307 | anything. */ | |
308 | add_builtins (); | |
309 | ||
310 | /* Store aliases in an array. */ | |
311 | get_aliases (); | |
312 | ||
313 | /* Get list of all modules. */ | |
314 | get_modules (); | |
315 | ||
316 | /* Generate list of all the names we know to handle in some way. */ | |
317 | generate_name_list (); | |
318 | ||
319 | /* Now we know all the names we will handle, collect information | |
320 | about them. */ | |
321 | generate_name_info (); | |
322 | ||
ba915a38 UD |
323 | /* Write the output file, but only if we haven't seen any error. */ |
324 | if (status == 0) | |
325 | status = write_output (); | |
326 | else | |
9fcddc2b | 327 | error (1, 0, _("no output file produced because warning were issued")); |
6b98979f UD |
328 | |
329 | return status; | |
330 | } | |
331 | ||
332 | ||
d8d5a647 UD |
333 | /* Handle program arguments. */ |
334 | static error_t | |
335 | parse_opt (int key, char *arg, struct argp_state *state) | |
336 | { | |
337 | switch (key) | |
338 | { | |
339 | case OPT_PREFIX: | |
340 | prefix = arg; | |
341 | prefix_len = strlen (prefix); | |
342 | break; | |
343 | default: | |
344 | return ARGP_ERR_UNKNOWN; | |
345 | } | |
346 | return 0; | |
347 | } | |
348 | ||
349 | ||
6b98979f UD |
350 | static char * |
351 | more_help (int key, const char *text, void *input) | |
352 | { | |
353 | switch (key) | |
354 | { | |
355 | case ARGP_KEY_HELP_EXTRA: | |
356 | /* We print some extra information. */ | |
357 | return strdup (gettext ("\ | |
358 | Report bugs using the `glibcbug' script to <bugs@gnu.org>.\n")); | |
359 | default: | |
360 | break; | |
361 | } | |
362 | return (char *) text; | |
363 | } | |
364 | ||
365 | ||
366 | /* Print the version information. */ | |
367 | static void | |
368 | print_version (FILE *stream, struct argp_state *state) | |
369 | { | |
370 | fprintf (stream, "iconvconfig (GNU %s) %s\n", PACKAGE, VERSION); | |
371 | fprintf (stream, gettext ("\ | |
372 | Copyright (C) %s Free Software Foundation, Inc.\n\ | |
373 | This is free software; see the source for copying conditions. There is NO\n\ | |
374 | warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n\ | |
b0c9067d | 375 | "), "2003"); |
6b98979f UD |
376 | fprintf (stream, gettext ("Written by %s.\n"), "Ulrich Drepper"); |
377 | } | |
378 | ||
379 | ||
380 | static int | |
381 | alias_compare (const void *p1, const void *p2) | |
382 | { | |
383 | const struct alias *a1 = (const struct alias *) p1; | |
384 | const struct alias *a2 = (const struct alias *) p2; | |
385 | ||
386 | return strcmp (a1->fromname, a2->fromname); | |
387 | } | |
388 | ||
389 | ||
390 | static void | |
391 | new_alias (const char *fromname, size_t fromlen, const char *toname, | |
392 | size_t tolen) | |
393 | { | |
394 | struct alias *newp; | |
395 | void **inserted; | |
396 | ||
397 | newp = (struct alias *) xmalloc (sizeof (struct alias) + fromlen + tolen); | |
398 | ||
399 | newp->fromname = mempcpy (newp->toname, toname, tolen); | |
400 | memcpy (newp->fromname, fromname, fromlen); | |
401 | newp->module = NULL; | |
402 | ||
403 | inserted = (void **) tsearch (newp, &aliases, alias_compare); | |
404 | if (inserted == NULL) | |
405 | error (EXIT_FAILURE, errno, gettext ("while inserting in search tree")); | |
406 | if (*inserted != newp) | |
407 | /* Something went wrong, free this entry. */ | |
408 | free (newp); | |
409 | else | |
410 | { | |
411 | newp->froment = strtabadd (strtab, newp->fromname, fromlen); | |
412 | newp->toent = strtabadd (strtab, newp->toname, tolen); | |
413 | } | |
414 | } | |
415 | ||
416 | ||
417 | /* Add new alias. */ | |
418 | static void | |
419 | add_alias (char *rp) | |
420 | { | |
421 | /* We now expect two more string. The strings are normalized | |
422 | (converted to UPPER case) and strored in the alias database. */ | |
423 | char *from; | |
424 | char *to; | |
425 | char *wp; | |
426 | ||
427 | while (isspace (*rp)) | |
428 | ++rp; | |
429 | from = wp = rp; | |
430 | while (*rp != '\0' && !isspace (*rp)) | |
431 | *wp++ = toupper (*rp++); | |
432 | if (*rp == '\0') | |
433 | /* There is no `to' string on the line. Ignore it. */ | |
434 | return; | |
435 | *wp++ = '\0'; | |
436 | to = ++rp; | |
437 | while (isspace (*rp)) | |
438 | ++rp; | |
439 | while (*rp != '\0' && !isspace (*rp)) | |
440 | *wp++ = toupper (*rp++); | |
441 | if (to == wp) | |
442 | /* No `to' string, ignore the line. */ | |
443 | return; | |
444 | *wp++ = '\0'; | |
445 | ||
6dd67bd5 UD |
446 | assert (strlen (from) + 1 == (size_t) (to - from)); |
447 | assert (strlen (to) + 1 == (size_t) (wp - to)); | |
47bd306e | 448 | |
6b98979f UD |
449 | new_alias (from, to - from, to, wp - to); |
450 | } | |
451 | ||
452 | ||
453 | static void | |
454 | append_alias (const void *nodep, VISIT value, int level) | |
455 | { | |
456 | if (value != leaf && value != postorder) | |
457 | return; | |
458 | ||
459 | if (nalias_list_max == nalias_list) | |
460 | { | |
461 | nalias_list_max += 50; | |
462 | alias_list = (struct alias **) xrealloc (alias_list, | |
463 | (nalias_list_max | |
464 | * sizeof (struct alias *))); | |
465 | } | |
466 | ||
467 | alias_list[nalias_list++] = *(struct alias **) nodep; | |
468 | } | |
469 | ||
470 | ||
471 | static void | |
472 | get_aliases (void) | |
473 | { | |
474 | twalk (aliases, append_alias); | |
475 | } | |
476 | ||
477 | ||
478 | static int | |
479 | module_compare (const void *p1, const void *p2) | |
480 | { | |
481 | const struct module *m1 = (const struct module *) p1; | |
482 | const struct module *m2 = (const struct module *) p2; | |
483 | int result; | |
484 | ||
485 | result = strcmp (m1->fromname, m2->fromname); | |
486 | if (result == 0) | |
487 | result = strcmp (m1->toname, m2->toname); | |
488 | ||
489 | return result; | |
490 | } | |
491 | ||
492 | ||
493 | /* Create new module record. */ | |
494 | static void | |
495 | new_module (const char *fromname, size_t fromlen, const char *toname, | |
496 | size_t tolen, const char *directory, | |
497 | const char *filename, size_t filelen, int cost, size_t need_ext) | |
498 | { | |
499 | struct module *new_module; | |
500 | size_t dirlen = strlen (directory) + 1; | |
501 | char *tmp; | |
502 | void **inserted; | |
503 | ||
504 | new_module = (struct module *) xmalloc (sizeof (struct module) | |
505 | + fromlen + tolen + filelen | |
506 | + need_ext); | |
507 | ||
508 | new_module->fromname = mempcpy (new_module->toname, toname, tolen); | |
509 | ||
510 | new_module->filename = mempcpy (new_module->fromname, fromname, fromlen); | |
511 | ||
512 | new_module->cost = cost; | |
513 | new_module->next = NULL; | |
514 | ||
515 | tmp = mempcpy (new_module->filename, filename, filelen); | |
516 | if (need_ext) | |
517 | { | |
47bd306e UD |
518 | memcpy (tmp - 1, gconv_module_ext, need_ext + 1); |
519 | filelen += need_ext; | |
6b98979f UD |
520 | } |
521 | new_module->directory = directory; | |
522 | ||
523 | /* Now insert the new module data structure in our search tree. */ | |
524 | inserted = (void **) tsearch (new_module, &modules, module_compare); | |
525 | if (inserted == NULL) | |
526 | error (EXIT_FAILURE, errno, "while inserting in search tree"); | |
527 | if (*inserted != new_module) | |
528 | free (new_module); | |
529 | else | |
530 | { | |
531 | new_module->fromname_strent = strtabadd (strtab, new_module->fromname, | |
532 | fromlen); | |
533 | new_module->toname_strent = strtabadd (strtab, new_module->toname, | |
534 | tolen); | |
535 | new_module->filename_strent = strtabadd (strtab, new_module->filename, | |
536 | filelen); | |
537 | new_module->directory_strent = strtabadd (strtab, directory, dirlen); | |
538 | } | |
539 | } | |
540 | ||
541 | ||
542 | /* Add new module. */ | |
543 | static void | |
544 | internal_function | |
545 | add_module (char *rp, const char *directory) | |
546 | { | |
547 | /* We expect now | |
548 | 1. `from' name | |
549 | 2. `to' name | |
550 | 3. filename of the module | |
551 | 4. an optional cost value | |
552 | */ | |
553 | char *from; | |
554 | char *to; | |
555 | char *module; | |
556 | char *wp; | |
557 | int need_ext; | |
558 | int cost; | |
559 | ||
560 | while (isspace (*rp)) | |
561 | ++rp; | |
562 | from = rp; | |
563 | while (*rp != '\0' && !isspace (*rp)) | |
564 | { | |
565 | *rp = toupper (*rp); | |
566 | ++rp; | |
567 | } | |
568 | if (*rp == '\0') | |
569 | return; | |
570 | *rp++ = '\0'; | |
571 | to = wp = rp; | |
572 | while (isspace (*rp)) | |
573 | ++rp; | |
574 | while (*rp != '\0' && !isspace (*rp)) | |
575 | *wp++ = toupper (*rp++); | |
576 | if (*rp == '\0') | |
577 | return; | |
578 | *wp++ = '\0'; | |
579 | do | |
580 | ++rp; | |
581 | while (isspace (*rp)); | |
582 | module = wp; | |
583 | while (*rp != '\0' && !isspace (*rp)) | |
584 | *wp++ = *rp++; | |
585 | if (*rp == '\0') | |
586 | { | |
587 | /* There is no cost, use one by default. */ | |
588 | *wp++ = '\0'; | |
589 | cost = 1; | |
590 | } | |
591 | else | |
592 | { | |
593 | /* There might be a cost value. */ | |
594 | char *endp; | |
595 | ||
596 | *wp++ = '\0'; | |
597 | cost = strtol (rp, &endp, 10); | |
598 | if (rp == endp || cost < 1) | |
599 | /* No useful information. */ | |
600 | cost = 1; | |
601 | } | |
602 | ||
603 | if (module[0] == '\0') | |
604 | /* No module name given. */ | |
605 | return; | |
606 | ||
607 | /* See whether we must add the ending. */ | |
608 | need_ext = 0; | |
6dd67bd5 | 609 | if ((size_t) (wp - module) < sizeof (gconv_module_ext) |
6b98979f UD |
610 | || memcmp (wp - sizeof (gconv_module_ext), gconv_module_ext, |
611 | sizeof (gconv_module_ext)) != 0) | |
612 | /* We must add the module extension. */ | |
613 | need_ext = sizeof (gconv_module_ext) - 1; | |
614 | ||
6dd67bd5 UD |
615 | assert (strlen (from) + 1 == (size_t) (to - from)); |
616 | assert (strlen (to) + 1 == (size_t) (module - to)); | |
617 | assert (strlen (module) + 1 == (size_t) (wp - module)); | |
6b98979f UD |
618 | |
619 | new_module (from, to - from, to, module - to, directory, module, wp - module, | |
620 | cost, need_ext); | |
621 | } | |
622 | ||
623 | ||
624 | /* Read the config file and add the data for this directory to that. */ | |
625 | static int | |
626 | handle_dir (const char *dir) | |
627 | { | |
628 | char *infile; | |
d8d5a647 | 629 | char *cp; |
6b98979f UD |
630 | FILE *fp; |
631 | char *line = NULL; | |
632 | size_t linelen = 0; | |
633 | size_t dirlen = strlen (dir); | |
634 | ||
635 | if (dir[dirlen - 1] != '/') | |
636 | { | |
637 | char *newp = (char *) xmalloc (dirlen + 2); | |
638 | dir = memcpy (newp, dir, dirlen); | |
639 | newp[dirlen++] = '/'; | |
640 | newp[dirlen] = '\0'; | |
641 | } | |
642 | ||
d8d5a647 UD |
643 | cp = infile = (char *) alloca (prefix_len + dirlen + sizeof "gconv-modules"); |
644 | if (dir[0] == '/') | |
645 | cp = mempcpy (cp, prefix, prefix_len); | |
646 | strcpy (mempcpy (cp, dir, dirlen), "gconv-modules"); | |
6b98979f UD |
647 | |
648 | fp = fopen (infile, "r"); | |
649 | if (fp == NULL) | |
650 | { | |
651 | error (0, errno, "cannot open `%s'", infile); | |
652 | return 1; | |
653 | } | |
654 | ||
655 | /* No threads present. */ | |
656 | __fsetlocking (fp, FSETLOCKING_BYCALLER); | |
657 | ||
658 | while (!feof_unlocked (fp)) | |
659 | { | |
660 | char *rp, *endp, *word; | |
661 | ssize_t n = __getdelim (&line, &linelen, '\n', fp); | |
662 | ||
663 | if (n < 0) | |
664 | /* An error occurred. */ | |
665 | break; | |
666 | ||
667 | rp = line; | |
668 | /* Terminate the line (excluding comments or newline) with a NUL | |
669 | byte to simplify the following code. */ | |
670 | endp = strchr (rp, '#'); | |
671 | if (endp != NULL) | |
672 | *endp = '\0'; | |
673 | else | |
674 | if (rp[n - 1] == '\n') | |
675 | rp[n - 1] = '\0'; | |
676 | ||
677 | while (isspace (*rp)) | |
678 | ++rp; | |
679 | ||
680 | /* If this is an empty line go on with the next one. */ | |
681 | if (rp == endp) | |
682 | continue; | |
683 | ||
684 | word = rp; | |
685 | while (*rp != '\0' && !isspace (*rp)) | |
686 | ++rp; | |
687 | ||
688 | if (rp - word == sizeof ("alias") - 1 | |
689 | && memcmp (word, "alias", sizeof ("alias") - 1) == 0) | |
690 | add_alias (rp); | |
691 | else if (rp - word == sizeof ("module") - 1 | |
692 | && memcmp (word, "module", sizeof ("module") - 1) == 0) | |
693 | add_module (rp, dir); | |
694 | /* else */ | |
695 | /* Otherwise ignore the line. */ | |
696 | } | |
697 | ||
698 | free (line); | |
699 | ||
700 | fclose (fp); | |
701 | ||
702 | return 0; | |
703 | } | |
704 | ||
705 | ||
706 | static void | |
707 | append_module (const void *nodep, VISIT value, int level) | |
708 | { | |
709 | struct module *mo; | |
710 | ||
711 | if (value != leaf && value != postorder) | |
712 | return; | |
713 | ||
714 | mo = *(struct module **) nodep; | |
715 | ||
716 | if (nmodule_list > 0 | |
717 | && strcmp (module_list[nmodule_list - 1]->fromname, mo->fromname) == 0) | |
718 | { | |
719 | /* Same name. */ | |
720 | mo->next = module_list[nmodule_list - 1]; | |
721 | module_list[nmodule_list - 1] = mo; | |
722 | ||
723 | return; | |
724 | } | |
725 | ||
726 | if (nmodule_list_max == nmodule_list) | |
727 | { | |
728 | nmodule_list_max += 50; | |
729 | module_list = (struct module **) xrealloc (module_list, | |
730 | (nmodule_list_max | |
731 | * sizeof (struct module *))); | |
732 | } | |
733 | ||
734 | module_list[nmodule_list++] = mo; | |
735 | } | |
736 | ||
737 | ||
738 | static void | |
739 | get_modules (void) | |
740 | { | |
741 | twalk (modules, append_module); | |
742 | } | |
743 | ||
744 | ||
745 | static void | |
746 | add_builtins (void) | |
747 | { | |
748 | size_t cnt; | |
749 | ||
750 | /* Add all aliases. */ | |
751 | for (cnt = 0; cnt < nbuiltin_alias; ++cnt) | |
752 | new_alias (builtin_alias[cnt].from, | |
753 | strlen (builtin_alias[cnt].from) + 1, | |
754 | builtin_alias[cnt].to, | |
755 | strlen (builtin_alias[cnt].to) + 1); | |
756 | ||
757 | /* add the builtin transformations. */ | |
758 | for (cnt = 0; cnt < nbuiltin_trans; ++cnt) | |
759 | new_module (builtin_trans[cnt].from, | |
760 | strlen (builtin_trans[cnt].from) + 1, | |
761 | builtin_trans[cnt].to, | |
762 | strlen (builtin_trans[cnt].to) + 1, | |
763 | "", builtin_trans[cnt].module, | |
764 | strlen (builtin_trans[cnt].module) + 1, | |
765 | builtin_trans[cnt].cost, 0); | |
766 | } | |
767 | ||
768 | ||
769 | static int | |
770 | name_compare (const void *p1, const void *p2) | |
771 | { | |
772 | const struct name *n1 = (const struct name *) p1; | |
773 | const struct name *n2 = (const struct name *) p2; | |
774 | ||
775 | return strcmp (n1->name, n2->name); | |
776 | } | |
777 | ||
778 | ||
779 | static struct name * | |
780 | new_name (const char *str, struct Strent *strent) | |
781 | { | |
782 | struct name *newp = (struct name *) xmalloc (sizeof (struct name)); | |
783 | ||
784 | newp->name = str; | |
785 | newp->strent = strent; | |
786 | newp->module_idx = -1; | |
dd9423a6 | 787 | newp->hashval = __hash_string (str); |
6b98979f UD |
788 | |
789 | ++nnames; | |
790 | ||
791 | return newp; | |
792 | } | |
793 | ||
794 | ||
795 | static void | |
796 | generate_name_list (void) | |
797 | { | |
798 | size_t i; | |
799 | ||
b2e3d177 UD |
800 | /* A name we always need. */ |
801 | tsearch (new_name ("INTERNAL", strtabadd (strtab, "INTERNAL", | |
802 | sizeof ("INTERNAL"))), | |
803 | &names, name_compare); | |
804 | ||
6b98979f UD |
805 | for (i = 0; i < nmodule_list; ++i) |
806 | { | |
807 | struct module *runp; | |
808 | ||
809 | if (strcmp (module_list[i]->fromname, "INTERNAL") != 0) | |
810 | tsearch (new_name (module_list[i]->fromname, | |
811 | module_list[i]->fromname_strent), | |
812 | &names, name_compare); | |
813 | ||
814 | for (runp = module_list[i]; runp != NULL; runp = runp->next) | |
815 | if (strcmp (runp->toname, "INTERNAL") != 0) | |
816 | tsearch (new_name (runp->toname, runp->toname_strent), | |
817 | &names, name_compare); | |
818 | } | |
819 | } | |
820 | ||
821 | ||
822 | static int | |
823 | name_to_module_idx (const char *name, int add) | |
824 | { | |
825 | struct name **res; | |
826 | struct name fake_name = { .name = name }; | |
827 | int idx; | |
828 | ||
829 | res = (struct name **) tfind (&fake_name, &names, name_compare); | |
830 | if (res == NULL) | |
831 | abort (); | |
832 | ||
833 | idx = (*res)->module_idx; | |
834 | if (idx == -1 && add) | |
835 | /* No module index assigned yet. */ | |
836 | idx = (*res)->module_idx = nname_info++; | |
837 | ||
838 | return idx; | |
839 | } | |
840 | ||
841 | ||
842 | static void | |
843 | generate_name_info (void) | |
844 | { | |
845 | size_t i; | |
b2e3d177 | 846 | int idx; |
6b98979f | 847 | |
b2e3d177 | 848 | name_info = (struct name_info *) xcalloc (nmodule_list + 1, |
6b98979f UD |
849 | sizeof (struct name_info)); |
850 | ||
b2e3d177 UD |
851 | /* First add a special entry for the INTERNAL name. This must have |
852 | index zero. */ | |
853 | idx = name_to_module_idx ("INTERNAL", 1); | |
854 | name_info[0].canonical_name = "INTERNAL"; | |
855 | name_info[0].canonical_strent = strtabadd (strtab, "INTERNAL", | |
856 | sizeof ("INTERNAL")); | |
857 | assert (nname_info == 1); | |
858 | ||
6b98979f UD |
859 | for (i = 0; i < nmodule_list; ++i) |
860 | { | |
861 | struct module *runp; | |
862 | ||
863 | for (runp = module_list[i]; runp != NULL; runp = runp->next) | |
864 | if (strcmp (runp->fromname, "INTERNAL") == 0) | |
865 | { | |
b2e3d177 | 866 | idx = name_to_module_idx (runp->toname, 1); |
6b98979f UD |
867 | name_info[idx].from_internal = runp; |
868 | assert (name_info[idx].canonical_name == NULL | |
869 | || strcmp (name_info[idx].canonical_name, | |
870 | runp->toname) == 0); | |
871 | name_info[idx].canonical_name = runp->toname; | |
872 | name_info[idx].canonical_strent = runp->toname_strent; | |
873 | } | |
874 | else if (strcmp (runp->toname, "INTERNAL") == 0) | |
875 | { | |
b2e3d177 | 876 | idx = name_to_module_idx (runp->fromname, 1); |
6b98979f UD |
877 | name_info[idx].to_internal = runp; |
878 | assert (name_info[idx].canonical_name == NULL | |
879 | || strcmp (name_info[idx].canonical_name, | |
880 | runp->fromname) == 0); | |
881 | name_info[idx].canonical_name = runp->fromname; | |
882 | name_info[idx].canonical_strent = runp->fromname_strent; | |
883 | } | |
884 | else | |
885 | { | |
886 | /* This is a transformation not to or from the INTERNAL | |
887 | encoding. */ | |
888 | int from_idx = name_to_module_idx (runp->fromname, 1); | |
889 | int to_idx = name_to_module_idx (runp->toname, 1); | |
890 | struct other_conv_list *newp; | |
891 | ||
892 | newp = (struct other_conv_list *) | |
893 | xmalloc (sizeof (struct other_conv_list)); | |
894 | newp->other_conv.module_idx = to_idx; | |
895 | newp->other_conv.module = runp; | |
896 | newp->other_conv.next = NULL; /* XXX Allow multiple module sequence */ | |
897 | newp->dest_idx = to_idx; | |
898 | newp->next = name_info[from_idx].other_conv_list; | |
899 | name_info[from_idx].other_conv_list = newp; | |
900 | assert (name_info[from_idx].canonical_name == NULL | |
901 | || strcmp (name_info[from_idx].canonical_name, | |
902 | runp->fromname) == 0); | |
903 | name_info[from_idx].canonical_name = runp->fromname; | |
904 | name_info[from_idx].canonical_strent = runp->fromname_strent; | |
905 | ||
906 | ++nextra_modules; | |
907 | } | |
908 | } | |
909 | ||
910 | /* Now add the module index information for all the aliases. */ | |
911 | for (i = 0; i < nalias_list; ++i) | |
912 | { | |
913 | struct name fake_name = { .name = alias_list[i]->toname }; | |
914 | struct name **tonamep; | |
915 | ||
916 | tonamep = (struct name **) tfind (&fake_name, &names, name_compare); | |
917 | if (tonamep != NULL) | |
918 | { | |
919 | struct name *newp = new_name (alias_list[i]->fromname, | |
920 | alias_list[i]->froment); | |
921 | newp->module_idx = (*tonamep)->module_idx; | |
922 | tsearch (newp, &names, name_compare); | |
923 | } | |
924 | } | |
925 | } | |
926 | ||
927 | ||
928 | static int | |
929 | is_prime (unsigned long int candidate) | |
930 | { | |
931 | /* No even number and none less than 10 will be passed here. */ | |
932 | unsigned long int divn = 3; | |
933 | unsigned long int sq = divn * divn; | |
934 | ||
935 | while (sq < candidate && candidate % divn != 0) | |
936 | { | |
937 | ++divn; | |
938 | sq += 4 * divn; | |
939 | ++divn; | |
940 | } | |
941 | ||
942 | return candidate % divn != 0; | |
943 | } | |
944 | ||
945 | ||
946 | static uint32_t | |
947 | next_prime (uint32_t seed) | |
948 | { | |
949 | /* Make it definitely odd. */ | |
950 | seed |= 1; | |
951 | ||
952 | while (!is_prime (seed)) | |
953 | seed += 2; | |
954 | ||
955 | return seed; | |
956 | } | |
957 | ||
958 | ||
959 | /* Format of the output file. | |
960 | ||
961 | Offset Length Description | |
962 | 0000 4 Magic header bytes | |
963 | 0004 4 Offset of string table (stoff) | |
964 | 0008 4 Offset of name hashing table (hoff) | |
965 | 000C 4 Hashing table size (hsize) | |
966 | 0010 4 Offset of module table (moff) | |
967 | 0014 4 Offset of other conversion module table (ooff) | |
968 | ||
969 | stoff ??? String table | |
970 | ||
971 | hoff 8*hsize Array of tuples | |
972 | string table offset | |
973 | module index | |
974 | ||
975 | moff ??? Array of tuples | |
976 | canonical name offset | |
977 | from-internal module dir name offset | |
978 | from-internal module name off | |
979 | to-internal module dir name offset | |
980 | to-internal module name offset | |
981 | offset into other conversion table | |
982 | ||
bbdf8285 UD |
983 | ooff ??? One or more of |
984 | number of steps/modules | |
6b98979f | 985 | one or more of tuple |
bbdf8285 | 986 | canonical name offset for output |
6b98979f UD |
987 | module dir name offset |
988 | module name offset | |
bbdf8285 | 989 | (following last entry with step count 0) |
6b98979f UD |
990 | */ |
991 | static int | |
992 | write_output (void) | |
993 | { | |
994 | int fd; | |
995 | char *string_table; | |
996 | size_t string_table_size; | |
997 | struct gconvcache_header header; | |
998 | struct hash_entry *hash_table; | |
999 | size_t hash_size; | |
1000 | struct module_entry *module_table; | |
1001 | char *extra_table; | |
1002 | char *cur_extra_table; | |
1003 | size_t n; | |
1004 | int idx; | |
1005 | struct iovec iov[6]; | |
1006 | static const gidx_t null_word; | |
1007 | size_t total; | |
d8d5a647 UD |
1008 | char tmpfname[prefix_len + sizeof (GCONV_MODULES_CACHE) |
1009 | + strlen (".XXXXXX")]; | |
1010 | char finalname[prefix_len + sizeof (GCONV_MODULES_CACHE)]; | |
6b98979f UD |
1011 | |
1012 | /* Function to insert the names. */ | |
53e62906 | 1013 | static void name_insert (const void *nodep, VISIT value, int level) |
6b98979f UD |
1014 | { |
1015 | struct name *name; | |
1016 | unsigned int idx; | |
1017 | unsigned int hval2; | |
1018 | ||
1019 | if (value != leaf && value != postorder) | |
1020 | return; | |
1021 | ||
1022 | name = *(struct name **) nodep; | |
1023 | idx = name->hashval % hash_size; | |
1024 | hval2 = 1 + name->hashval % (hash_size - 2); | |
1025 | ||
1026 | while (hash_table[idx].string_offset != 0) | |
1027 | if ((idx += hval2) >= hash_size) | |
1028 | idx -= hash_size; | |
1029 | ||
1030 | hash_table[idx].string_offset = strtaboffset (name->strent); | |
1031 | ||
1032 | assert (name->module_idx != -1); | |
1033 | hash_table[idx].module_idx = name->module_idx; | |
1034 | } | |
1035 | ||
1036 | /* Open the output file. */ | |
d8d5a647 UD |
1037 | assert (GCONV_MODULES_CACHE[0] == '/'); |
1038 | strcpy (stpcpy (mempcpy (tmpfname, prefix, prefix_len), GCONV_MODULES_CACHE), | |
1039 | ".XXXXXX"); | |
9fcddc2b | 1040 | fd = mkstemp (tmpfname); |
6b98979f UD |
1041 | if (fd == -1) |
1042 | return 1; | |
1043 | ||
d8d5a647 UD |
1044 | strcpy (mempcpy (finalname, prefix, prefix_len), GCONV_MODULES_CACHE); |
1045 | ||
6b98979f UD |
1046 | /* Create the string table. */ |
1047 | string_table = strtabfinalize (strtab, &string_table_size); | |
1048 | ||
1049 | /* Create the hashing table. We know how many strings we have. | |
1050 | Creating a perfect hash table is not reasonable here. Therefore | |
1051 | we use open hashing and a table size which is the next prime 40% | |
1052 | larger than the number of strings. */ | |
1053 | hash_size = next_prime (nnames * 1.4); | |
1054 | hash_table = (struct hash_entry *) xcalloc (hash_size, | |
1055 | sizeof (struct hash_entry)); | |
1056 | /* Fill the hash table. */ | |
1057 | twalk (names, name_insert); | |
1058 | ||
1059 | /* Create the section for the module list. */ | |
1060 | module_table = (struct module_entry *) xcalloc (sizeof (struct module_entry), | |
1061 | nname_info); | |
1062 | ||
1063 | /* Allocate memory for the non-INTERNAL conversions. The allocated | |
1064 | memory can be more than is actually needed. */ | |
1065 | extra_table = (char *) xcalloc (sizeof (struct extra_entry) | |
1066 | + sizeof (gidx_t) | |
1067 | + sizeof (struct extra_entry_module), | |
1068 | nextra_modules); | |
1069 | cur_extra_table = extra_table; | |
1070 | ||
1071 | /* Fill in the module information. */ | |
1072 | for (n = 0; n < nname_info; ++n) | |
1073 | { | |
1074 | module_table[n].canonname_offset = | |
1075 | strtaboffset (name_info[n].canonical_strent); | |
1076 | ||
1077 | if (name_info[n].from_internal == NULL) | |
1078 | { | |
1079 | module_table[n].fromdir_offset = 0; | |
1080 | module_table[n].fromname_offset = 0; | |
1081 | } | |
1082 | else | |
1083 | { | |
1084 | module_table[n].fromdir_offset = | |
1085 | strtaboffset (name_info[n].from_internal->directory_strent); | |
1086 | module_table[n].fromname_offset = | |
1087 | strtaboffset (name_info[n].from_internal->filename_strent); | |
1088 | } | |
1089 | ||
1090 | if (name_info[n].to_internal == NULL) | |
1091 | { | |
1092 | module_table[n].todir_offset = 0; | |
1093 | module_table[n].toname_offset = 0; | |
1094 | } | |
1095 | else | |
1096 | { | |
1097 | module_table[n].todir_offset = | |
1098 | strtaboffset (name_info[n].to_internal->directory_strent); | |
1099 | module_table[n].toname_offset = | |
1100 | strtaboffset (name_info[n].to_internal->filename_strent); | |
1101 | } | |
1102 | ||
1103 | if (name_info[n].other_conv_list != NULL) | |
1104 | { | |
1105 | struct other_conv_list *other = name_info[n].other_conv_list; | |
1106 | ||
1107 | /* Store the reference. We add 1 to distinguish the entry | |
1108 | at offset zero from the case where no extra modules are | |
1109 | available. The file reader has to account for the | |
1110 | offset. */ | |
1111 | module_table[n].extra_offset = 1 + cur_extra_table - extra_table; | |
1112 | ||
1113 | do | |
1114 | { | |
1115 | struct other_conv *runp; | |
1116 | struct extra_entry *extra; | |
1117 | ||
1118 | /* Allocate new entry. */ | |
1119 | extra = (struct extra_entry *) cur_extra_table; | |
1120 | cur_extra_table += sizeof (struct extra_entry); | |
1121 | extra->module_cnt = 0; | |
1122 | ||
1123 | runp = &other->other_conv; | |
1124 | do | |
1125 | { | |
1126 | cur_extra_table += sizeof (struct extra_entry_module); | |
1127 | extra->module[extra->module_cnt].outname_offset = | |
1128 | runp->next == NULL | |
1129 | ? other->dest_idx : runp->next->module_idx; | |
1130 | extra->module[extra->module_cnt].dir_offset = | |
1131 | strtaboffset (runp->module->directory_strent); | |
1132 | extra->module[extra->module_cnt].name_offset = | |
1133 | strtaboffset (runp->module->filename_strent); | |
1134 | ++extra->module_cnt; | |
1135 | ||
1136 | runp = runp->next; | |
1137 | } | |
1138 | while (runp != NULL); | |
1139 | ||
1140 | other = other->next; | |
1141 | } | |
1142 | while (other != NULL); | |
1143 | ||
1144 | /* Final module_cnt is zero. */ | |
1145 | *((gidx_t *) cur_extra_table) = 0; | |
1146 | cur_extra_table += sizeof (gidx_t); | |
1147 | } | |
1148 | } | |
1149 | ||
1150 | header.magic = GCONVCACHE_MAGIC; | |
1151 | ||
1152 | iov[0].iov_base = &header; | |
1153 | iov[0].iov_len = sizeof (struct gconvcache_header); | |
1154 | total = iov[0].iov_len; | |
1155 | ||
1156 | header.string_offset = total; | |
1157 | iov[1].iov_base = string_table; | |
1158 | iov[1].iov_len = string_table_size; | |
1159 | total += iov[1].iov_len; | |
1160 | ||
1161 | idx = 2; | |
1162 | if ((string_table_size & (sizeof (gidx_t) - 1)) != 0) | |
1163 | { | |
1164 | iov[2].iov_base = (void *) &null_word; | |
1165 | iov[2].iov_len = (sizeof (gidx_t) | |
1166 | - (string_table_size & (sizeof (gidx_t) - 1))); | |
1167 | total += iov[2].iov_len; | |
1168 | ++idx; | |
1169 | } | |
1170 | ||
1171 | header.hash_offset = total; | |
1172 | header.hash_size = hash_size; | |
1173 | iov[idx].iov_base = hash_table; | |
1174 | iov[idx].iov_len = hash_size * sizeof (struct hash_entry); | |
1175 | total += iov[idx].iov_len; | |
1176 | ++idx; | |
1177 | ||
1178 | header.module_offset = total; | |
1179 | iov[idx].iov_base = module_table; | |
1180 | iov[idx].iov_len = nname_info * sizeof (struct module_entry); | |
1181 | total += iov[idx].iov_len; | |
1182 | ++idx; | |
1183 | ||
6dd67bd5 | 1184 | assert ((size_t) (cur_extra_table - extra_table) |
6b98979f UD |
1185 | <= ((sizeof (struct extra_entry) + sizeof (gidx_t) |
1186 | + sizeof (struct extra_entry_module)) | |
1187 | * nextra_modules)); | |
1188 | header.otherconv_offset = total; | |
1189 | iov[idx].iov_base = extra_table; | |
1190 | iov[idx].iov_len = cur_extra_table - extra_table; | |
1191 | total += iov[idx].iov_len; | |
1192 | ++idx; | |
1193 | ||
6dd67bd5 | 1194 | if ((size_t) TEMP_FAILURE_RETRY (writev (fd, iov, idx)) != total |
9fcddc2b UD |
1195 | /* The file was created with mode 0600. Make it world-readable. */ |
1196 | || fchmod (fd, 0644) != 0 | |
1197 | /* Rename the file, possibly replacing an old one. */ | |
d8d5a647 | 1198 | || rename (tmpfname, finalname) != 0) |
6b98979f UD |
1199 | { |
1200 | int save_errno = errno; | |
1201 | close (fd); | |
9fcddc2b UD |
1202 | unlink (tmpfname); |
1203 | error (EXIT_FAILURE, save_errno, | |
1204 | gettext ("cannot generate output file")); | |
6b98979f UD |
1205 | } |
1206 | ||
1207 | close (fd); | |
1208 | ||
1209 | return 0; | |
1210 | } |