]>
Commit | Line | Data |
---|---|---|
6b98979f | 1 | /* Generate fastloading iconv module configuration files. |
581c785b | 2 | Copyright (C) 2000-2022 Free Software Foundation, Inc. |
6b98979f | 3 | This file is part of the GNU C Library. |
6b98979f | 4 | |
43bc8ac6 | 5 | This program is free software; you can redistribute it and/or modify |
2e2efe65 RM |
6 | it under the terms of the GNU General Public License as published |
7 | by the Free Software Foundation; version 2 of the License, or | |
8 | (at your option) any later version. | |
6b98979f | 9 | |
43bc8ac6 | 10 | This program is distributed in the hope that it will be useful, |
6b98979f | 11 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
43bc8ac6 UD |
12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
13 | GNU General Public License for more details. | |
6b98979f | 14 | |
43bc8ac6 | 15 | You should have received a copy of the GNU General Public License |
5a82c748 | 16 | along with this program; if not, see <https://www.gnu.org/licenses/>. */ |
6b98979f UD |
17 | |
18 | #include <argp.h> | |
19 | #include <assert.h> | |
20 | #include <error.h> | |
21 | #include <errno.h> | |
22 | #include <fcntl.h> | |
23 | #include <libintl.h> | |
24 | #include <locale.h> | |
25 | #include <mcheck.h> | |
26 | #include <search.h> | |
27 | #include <stdint.h> | |
ed2bca56 | 28 | #include <stdbool.h> |
6b98979f UD |
29 | #include <stdio.h> |
30 | #include <stdio_ext.h> | |
31 | #include <stdlib.h> | |
32 | #include <string.h> | |
33 | #include <unistd.h> | |
0588dfe7 | 34 | #include <sys/cdefs.h> |
6b98979f UD |
35 | #include <sys/uio.h> |
36 | ||
37 | #include "iconvconfig.h" | |
eeac390e | 38 | #include <gconv_parseconfdir.h> |
6b98979f UD |
39 | |
40 | /* Get libc version number. */ | |
41 | #include "../version.h" | |
42 | ||
43 | #define PACKAGE _libc_intl_domainname | |
44 | ||
45 | ||
46 | /* The hashing function we use. */ | |
47 | #include "../intl/hash-string.h" | |
48 | ||
49 | ||
50 | /* Types used. */ | |
51 | struct module | |
52 | { | |
53 | char *fromname; | |
54 | struct Strent *fromname_strent; | |
55 | char *filename; | |
56 | struct Strent *filename_strent; | |
57 | const char *directory; | |
58 | struct Strent *directory_strent; | |
59 | struct module *next; | |
60 | int cost; | |
61 | struct Strent *toname_strent; | |
62 | char toname[0]; | |
63 | }; | |
64 | ||
65 | struct alias | |
66 | { | |
67 | char *fromname; | |
68 | struct Strent *froment; | |
69 | struct module *module; | |
70 | struct Strent *toent; | |
71 | char toname[0]; | |
72 | }; | |
73 | ||
74 | struct name | |
75 | { | |
76 | const char *name; | |
77 | struct Strent *strent; | |
78 | int module_idx; | |
79 | uint32_t hashval; | |
80 | }; | |
81 | ||
82 | struct name_info | |
83 | { | |
84 | const char *canonical_name; | |
85 | struct Strent *canonical_strent; | |
86 | ||
87 | struct module *from_internal; | |
88 | struct module *to_internal; | |
89 | ||
90 | struct other_conv_list | |
91 | { | |
92 | int dest_idx; | |
93 | struct other_conv | |
94 | { | |
95 | gidx_t module_idx; | |
96 | struct module *module; | |
97 | struct other_conv *next; | |
98 | } other_conv; | |
99 | struct other_conv_list *next; | |
100 | } *other_conv_list; | |
101 | }; | |
102 | ||
103 | ||
104 | /* Name and version of program. */ | |
105 | static void print_version (FILE *stream, struct argp_state *state); | |
106 | void (*argp_program_version_hook) (FILE *, struct argp_state *) = print_version; | |
107 | ||
6b98979f UD |
108 | /* Short description of program. */ |
109 | static const char doc[] = N_("\ | |
110 | Create fastloading iconv module configuration file."); | |
111 | ||
112 | /* Strings for arguments in help texts. */ | |
113 | static const char args_doc[] = N_("[DIR...]"); | |
114 | ||
d8d5a647 UD |
115 | /* Prototype for option handler. */ |
116 | static error_t parse_opt (int key, char *arg, struct argp_state *state); | |
117 | ||
6b98979f UD |
118 | /* Function to print some extra text in the help message. */ |
119 | static char *more_help (int key, const char *text, void *input); | |
120 | ||
d8d5a647 UD |
121 | /* Definitions of arguments for argp functions. */ |
122 | #define OPT_PREFIX 300 | |
ed2bca56 | 123 | #define OPT_NOSTDLIB 301 |
d8d5a647 UD |
124 | static const struct argp_option options[] = |
125 | { | |
0e2b9cdd RM |
126 | { "prefix", OPT_PREFIX, N_("PATH"), 0, |
127 | N_("Prefix used for all file accesses") }, | |
128 | { "output", 'o', N_("FILE"), 0, N_("\ | |
ed2bca56 RM |
129 | Put output in FILE instead of installed location\ |
130 | (--prefix does not apply to FILE)") }, | |
131 | { "nostdlib", OPT_NOSTDLIB, NULL, 0, | |
132 | N_("Do not search standard directories, only those on the command line") }, | |
d8d5a647 UD |
133 | { NULL, 0, NULL, 0, NULL } |
134 | }; | |
135 | ||
6b98979f UD |
136 | /* Data structure to communicate with argp functions. */ |
137 | static struct argp argp = | |
138 | { | |
d8d5a647 | 139 | options, parse_opt, args_doc, doc, NULL, more_help |
6b98979f UD |
140 | }; |
141 | ||
142 | ||
143 | /* The function doing the actual work. */ | |
144 | static int handle_dir (const char *dir); | |
145 | ||
146 | /* Add all known builtin conversions and aliases. */ | |
147 | static void add_builtins (void); | |
148 | ||
149 | /* Create list of all aliases without circular aliases. */ | |
150 | static void get_aliases (void); | |
151 | ||
152 | /* Create list of all modules. */ | |
153 | static void get_modules (void); | |
154 | ||
155 | /* Get list of all the names and thereby indexing them. */ | |
156 | static void generate_name_list (void); | |
157 | ||
158 | /* Collect information about all the names. */ | |
159 | static void generate_name_info (void); | |
160 | ||
161 | /* Write the output file. */ | |
162 | static int write_output (void); | |
163 | ||
164 | ||
d8d5a647 UD |
165 | /* Prefix to be used for all file accesses. */ |
166 | static const char *prefix = ""; | |
167 | /* Its length. */ | |
168 | static size_t prefix_len; | |
169 | ||
ed2bca56 RM |
170 | /* Directory to place output file in. */ |
171 | static const char *output_file; | |
172 | /* Its length. */ | |
173 | static size_t output_file_len; | |
174 | ||
175 | /* If true, omit the GCONV_PATH directories and require some arguments. */ | |
176 | static bool nostdlib; | |
177 | ||
6b98979f UD |
178 | /* Search tree of the modules we know. */ |
179 | static void *modules; | |
180 | ||
181 | /* Search tree of the aliases we know. */ | |
182 | static void *aliases; | |
183 | ||
184 | /* Search tree for name to index mapping. */ | |
185 | static void *names; | |
186 | ||
187 | /* Number of names we know about. */ | |
188 | static int nnames; | |
189 | ||
190 | /* List of all aliases. */ | |
191 | static struct alias **alias_list; | |
192 | static size_t nalias_list; | |
193 | static size_t nalias_list_max; | |
194 | ||
195 | /* List of all modules. */ | |
196 | static struct module **module_list; | |
197 | static size_t nmodule_list; | |
198 | static size_t nmodule_list_max; | |
199 | ||
200 | /* Names and information about them. */ | |
201 | static struct name_info *name_info; | |
202 | static size_t nname_info; | |
203 | ||
204 | /* Number of translations not from or to INTERNAL. */ | |
205 | static size_t nextra_modules; | |
206 | ||
207 | ||
208 | /* Names and aliases for the builtin transformations. */ | |
209 | static struct | |
210 | { | |
211 | const char *from; | |
212 | const char *to; | |
213 | } builtin_alias[] = | |
214 | { | |
215 | #define BUILTIN_ALIAS(alias, real) \ | |
216 | { .from = alias, .to = real }, | |
f9ad060c UD |
217 | #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, BtowcFct, \ |
218 | MinF, MaxF, MinT, MaxT) | |
6b98979f UD |
219 | #include <gconv_builtin.h> |
220 | }; | |
874d8461 UD |
221 | #undef BUILTIN_ALIAS |
222 | #undef BUILTIN_TRANSFORMATION | |
6b98979f UD |
223 | #define nbuiltin_alias (sizeof (builtin_alias) / sizeof (builtin_alias[0])) |
224 | ||
225 | static struct | |
226 | { | |
227 | const char *from; | |
228 | const char *to; | |
229 | const char *module; | |
230 | int cost; | |
231 | } builtin_trans[] = | |
232 | { | |
233 | #define BUILTIN_ALIAS(alias, real) | |
f9ad060c UD |
234 | #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, BtowcFct, \ |
235 | MinF, MaxF, MinT, MaxT) \ | |
6b98979f UD |
236 | { .from = From, .to = To, .module = Name, .cost = Cost }, |
237 | #include <gconv_builtin.h> | |
238 | }; | |
f9ad060c UD |
239 | #undef BUILTIN_ALIAS |
240 | #undef BUILTIN_TRANSFORMATION | |
6b98979f UD |
241 | #define nbuiltin_trans (sizeof (builtin_trans) / sizeof (builtin_trans[0])) |
242 | ||
243 | ||
244 | /* Filename extension for the modules. */ | |
245 | #ifndef MODULE_EXT | |
246 | # define MODULE_EXT ".so" | |
247 | #endif | |
248 | static const char gconv_module_ext[] = MODULE_EXT; | |
249 | ||
250 | ||
6ff444c4 | 251 | #include <programs/xmalloc.h> |
9429049c | 252 | #include <programs/xasprintf.h> |
6b98979f UD |
253 | |
254 | ||
255 | /* C string table handling. */ | |
256 | struct Strtab; | |
257 | struct Strent; | |
258 | ||
259 | /* Create new C string table object in memory. */ | |
260 | extern struct Strtab *strtabinit (void); | |
261 | ||
262 | /* Free resources allocated for C string table ST. */ | |
263 | extern void strtabfree (struct Strtab *st); | |
264 | ||
265 | /* Add string STR (length LEN is != 0) to C string table ST. */ | |
266 | extern struct Strent *strtabadd (struct Strtab *st, const char *str, | |
267 | size_t len); | |
268 | ||
269 | /* Finalize string table ST and store size in *SIZE and return a pointer. */ | |
270 | extern void *strtabfinalize (struct Strtab *st, size_t *size); | |
271 | ||
272 | /* Get offset in string table for string associated with SE. */ | |
273 | extern size_t strtaboffset (struct Strent *se); | |
274 | ||
275 | /* String table we construct. */ | |
276 | static struct Strtab *strtab; | |
277 | ||
278 | ||
279 | ||
280 | int | |
281 | main (int argc, char *argv[]) | |
282 | { | |
283 | int remaining; | |
284 | int status = 0; | |
6b98979f UD |
285 | |
286 | /* Enable memory use testing. */ | |
47bd306e UD |
287 | /* mcheck_pedantic (NULL); */ |
288 | mtrace (); | |
6b98979f UD |
289 | |
290 | /* Set locale via LC_ALL. */ | |
291 | setlocale (LC_ALL, ""); | |
292 | ||
293 | /* Set the text message domain. */ | |
294 | textdomain (_libc_intl_domainname); | |
295 | ||
296 | /* Parse and process arguments. */ | |
297 | argp_parse (&argp, argc, argv, 0, &remaining, NULL); | |
298 | ||
ed2bca56 RM |
299 | if (nostdlib && remaining == argc) |
300 | error (2, 0, _("Directory arguments required when using --nostdlib")); | |
301 | ||
6b98979f UD |
302 | /* Initialize the string table. */ |
303 | strtab = strtabinit (); | |
304 | ||
305 | /* Handle all directories mentioned. */ | |
306 | while (remaining < argc) | |
307 | status |= handle_dir (argv[remaining++]); | |
308 | ||
ed2bca56 | 309 | if (! nostdlib) |
6b98979f | 310 | { |
ed2bca56 RM |
311 | /* In any case also handle the standard directory. */ |
312 | char *path = strdupa (GCONV_PATH), *tp = strsep (&path, ":"); | |
313 | while (tp != NULL) | |
314 | { | |
315 | status |= handle_dir (tp); | |
6b98979f | 316 | |
ed2bca56 RM |
317 | tp = strsep (&path, ":"); |
318 | } | |
6b98979f UD |
319 | } |
320 | ||
321 | /* Add the builtin transformations and aliases without overwriting | |
322 | anything. */ | |
323 | add_builtins (); | |
324 | ||
325 | /* Store aliases in an array. */ | |
326 | get_aliases (); | |
327 | ||
328 | /* Get list of all modules. */ | |
329 | get_modules (); | |
330 | ||
331 | /* Generate list of all the names we know to handle in some way. */ | |
332 | generate_name_list (); | |
333 | ||
334 | /* Now we know all the names we will handle, collect information | |
335 | about them. */ | |
336 | generate_name_info (); | |
337 | ||
ba915a38 UD |
338 | /* Write the output file, but only if we haven't seen any error. */ |
339 | if (status == 0) | |
340 | status = write_output (); | |
341 | else | |
c69136ae | 342 | error (1, 0, _("no output file produced because warnings were issued")); |
6b98979f UD |
343 | |
344 | return status; | |
345 | } | |
346 | ||
347 | ||
d8d5a647 UD |
348 | /* Handle program arguments. */ |
349 | static error_t | |
350 | parse_opt (int key, char *arg, struct argp_state *state) | |
351 | { | |
352 | switch (key) | |
353 | { | |
354 | case OPT_PREFIX: | |
355 | prefix = arg; | |
356 | prefix_len = strlen (prefix); | |
357 | break; | |
ed2bca56 RM |
358 | case 'o': |
359 | output_file = arg; | |
360 | output_file_len = strlen (output_file); | |
361 | break; | |
362 | case OPT_NOSTDLIB: | |
363 | nostdlib = true; | |
364 | break; | |
d8d5a647 UD |
365 | default: |
366 | return ARGP_ERR_UNKNOWN; | |
367 | } | |
368 | return 0; | |
369 | } | |
370 | ||
371 | ||
6b98979f UD |
372 | static char * |
373 | more_help (int key, const char *text, void *input) | |
374 | { | |
8b748aed | 375 | char *tp = NULL; |
6b98979f UD |
376 | switch (key) |
377 | { | |
378 | case ARGP_KEY_HELP_EXTRA: | |
379 | /* We print some extra information. */ | |
8b748aed | 380 | if (asprintf (&tp, gettext ("\ |
d40eb37a | 381 | For bug reporting instructions, please see:\n\ |
8b748aed JM |
382 | %s.\n"), REPORT_BUGS_TO) < 0) |
383 | return NULL; | |
384 | return tp; | |
6b98979f UD |
385 | default: |
386 | break; | |
387 | } | |
388 | return (char *) text; | |
389 | } | |
390 | ||
391 | ||
392 | /* Print the version information. */ | |
393 | static void | |
394 | print_version (FILE *stream, struct argp_state *state) | |
395 | { | |
8b748aed | 396 | fprintf (stream, "iconvconfig %s%s\n", PKGVERSION, VERSION); |
6b98979f UD |
397 | fprintf (stream, gettext ("\ |
398 | Copyright (C) %s Free Software Foundation, Inc.\n\ | |
399 | This is free software; see the source for copying conditions. There is NO\n\ | |
400 | warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n\ | |
634b5eba | 401 | "), "2022"); |
6b98979f UD |
402 | fprintf (stream, gettext ("Written by %s.\n"), "Ulrich Drepper"); |
403 | } | |
404 | ||
405 | ||
406 | static int | |
407 | alias_compare (const void *p1, const void *p2) | |
408 | { | |
409 | const struct alias *a1 = (const struct alias *) p1; | |
410 | const struct alias *a2 = (const struct alias *) p2; | |
411 | ||
412 | return strcmp (a1->fromname, a2->fromname); | |
413 | } | |
414 | ||
415 | ||
416 | static void | |
417 | new_alias (const char *fromname, size_t fromlen, const char *toname, | |
418 | size_t tolen) | |
419 | { | |
420 | struct alias *newp; | |
421 | void **inserted; | |
422 | ||
423 | newp = (struct alias *) xmalloc (sizeof (struct alias) + fromlen + tolen); | |
424 | ||
425 | newp->fromname = mempcpy (newp->toname, toname, tolen); | |
426 | memcpy (newp->fromname, fromname, fromlen); | |
427 | newp->module = NULL; | |
428 | ||
429 | inserted = (void **) tsearch (newp, &aliases, alias_compare); | |
430 | if (inserted == NULL) | |
431 | error (EXIT_FAILURE, errno, gettext ("while inserting in search tree")); | |
432 | if (*inserted != newp) | |
433 | /* Something went wrong, free this entry. */ | |
434 | free (newp); | |
435 | else | |
436 | { | |
437 | newp->froment = strtabadd (strtab, newp->fromname, fromlen); | |
438 | newp->toent = strtabadd (strtab, newp->toname, tolen); | |
439 | } | |
440 | } | |
441 | ||
442 | ||
443 | /* Add new alias. */ | |
444 | static void | |
445 | add_alias (char *rp) | |
446 | { | |
447 | /* We now expect two more string. The strings are normalized | |
448 | (converted to UPPER case) and strored in the alias database. */ | |
449 | char *from; | |
450 | char *to; | |
451 | char *wp; | |
452 | ||
453 | while (isspace (*rp)) | |
454 | ++rp; | |
455 | from = wp = rp; | |
456 | while (*rp != '\0' && !isspace (*rp)) | |
457 | *wp++ = toupper (*rp++); | |
458 | if (*rp == '\0') | |
459 | /* There is no `to' string on the line. Ignore it. */ | |
460 | return; | |
461 | *wp++ = '\0'; | |
462 | to = ++rp; | |
463 | while (isspace (*rp)) | |
464 | ++rp; | |
465 | while (*rp != '\0' && !isspace (*rp)) | |
466 | *wp++ = toupper (*rp++); | |
467 | if (to == wp) | |
468 | /* No `to' string, ignore the line. */ | |
469 | return; | |
470 | *wp++ = '\0'; | |
471 | ||
6dd67bd5 UD |
472 | assert (strlen (from) + 1 == (size_t) (to - from)); |
473 | assert (strlen (to) + 1 == (size_t) (wp - to)); | |
47bd306e | 474 | |
6b98979f UD |
475 | new_alias (from, to - from, to, wp - to); |
476 | } | |
477 | ||
478 | ||
479 | static void | |
480 | append_alias (const void *nodep, VISIT value, int level) | |
481 | { | |
482 | if (value != leaf && value != postorder) | |
483 | return; | |
484 | ||
485 | if (nalias_list_max == nalias_list) | |
486 | { | |
487 | nalias_list_max += 50; | |
488 | alias_list = (struct alias **) xrealloc (alias_list, | |
489 | (nalias_list_max | |
490 | * sizeof (struct alias *))); | |
491 | } | |
492 | ||
493 | alias_list[nalias_list++] = *(struct alias **) nodep; | |
494 | } | |
495 | ||
496 | ||
497 | static void | |
498 | get_aliases (void) | |
499 | { | |
500 | twalk (aliases, append_alias); | |
501 | } | |
502 | ||
503 | ||
504 | static int | |
505 | module_compare (const void *p1, const void *p2) | |
506 | { | |
507 | const struct module *m1 = (const struct module *) p1; | |
508 | const struct module *m2 = (const struct module *) p2; | |
509 | int result; | |
510 | ||
511 | result = strcmp (m1->fromname, m2->fromname); | |
512 | if (result == 0) | |
513 | result = strcmp (m1->toname, m2->toname); | |
514 | ||
515 | return result; | |
516 | } | |
517 | ||
518 | ||
519 | /* Create new module record. */ | |
520 | static void | |
521 | new_module (const char *fromname, size_t fromlen, const char *toname, | |
9429049c | 522 | size_t tolen, const char *dir_in, |
6b98979f UD |
523 | const char *filename, size_t filelen, int cost, size_t need_ext) |
524 | { | |
525 | struct module *new_module; | |
9429049c SP |
526 | size_t dirlen = strlen (dir_in) + 1; |
527 | const char *directory = xstrdup (dir_in); | |
6b98979f UD |
528 | char *tmp; |
529 | void **inserted; | |
530 | ||
531 | new_module = (struct module *) xmalloc (sizeof (struct module) | |
532 | + fromlen + tolen + filelen | |
533 | + need_ext); | |
534 | ||
535 | new_module->fromname = mempcpy (new_module->toname, toname, tolen); | |
536 | ||
537 | new_module->filename = mempcpy (new_module->fromname, fromname, fromlen); | |
538 | ||
539 | new_module->cost = cost; | |
540 | new_module->next = NULL; | |
541 | ||
542 | tmp = mempcpy (new_module->filename, filename, filelen); | |
543 | if (need_ext) | |
544 | { | |
47bd306e UD |
545 | memcpy (tmp - 1, gconv_module_ext, need_ext + 1); |
546 | filelen += need_ext; | |
6b98979f UD |
547 | } |
548 | new_module->directory = directory; | |
549 | ||
550 | /* Now insert the new module data structure in our search tree. */ | |
551 | inserted = (void **) tsearch (new_module, &modules, module_compare); | |
552 | if (inserted == NULL) | |
553 | error (EXIT_FAILURE, errno, "while inserting in search tree"); | |
554 | if (*inserted != new_module) | |
555 | free (new_module); | |
556 | else | |
557 | { | |
558 | new_module->fromname_strent = strtabadd (strtab, new_module->fromname, | |
559 | fromlen); | |
560 | new_module->toname_strent = strtabadd (strtab, new_module->toname, | |
561 | tolen); | |
562 | new_module->filename_strent = strtabadd (strtab, new_module->filename, | |
563 | filelen); | |
564 | new_module->directory_strent = strtabadd (strtab, directory, dirlen); | |
565 | } | |
566 | } | |
567 | ||
568 | ||
569 | /* Add new module. */ | |
570 | static void | |
eeac390e SP |
571 | add_module (char *rp, const char *directory, |
572 | size_t dirlen __attribute__ ((__unused__)), | |
573 | int modcount __attribute__ ((__unused__))) | |
6b98979f UD |
574 | { |
575 | /* We expect now | |
576 | 1. `from' name | |
577 | 2. `to' name | |
578 | 3. filename of the module | |
579 | 4. an optional cost value | |
580 | */ | |
581 | char *from; | |
582 | char *to; | |
583 | char *module; | |
584 | char *wp; | |
585 | int need_ext; | |
586 | int cost; | |
587 | ||
588 | while (isspace (*rp)) | |
589 | ++rp; | |
590 | from = rp; | |
591 | while (*rp != '\0' && !isspace (*rp)) | |
592 | { | |
593 | *rp = toupper (*rp); | |
594 | ++rp; | |
595 | } | |
596 | if (*rp == '\0') | |
597 | return; | |
598 | *rp++ = '\0'; | |
599 | to = wp = rp; | |
600 | while (isspace (*rp)) | |
601 | ++rp; | |
602 | while (*rp != '\0' && !isspace (*rp)) | |
603 | *wp++ = toupper (*rp++); | |
604 | if (*rp == '\0') | |
605 | return; | |
606 | *wp++ = '\0'; | |
607 | do | |
608 | ++rp; | |
609 | while (isspace (*rp)); | |
610 | module = wp; | |
611 | while (*rp != '\0' && !isspace (*rp)) | |
612 | *wp++ = *rp++; | |
613 | if (*rp == '\0') | |
614 | { | |
615 | /* There is no cost, use one by default. */ | |
616 | *wp++ = '\0'; | |
617 | cost = 1; | |
618 | } | |
619 | else | |
620 | { | |
621 | /* There might be a cost value. */ | |
622 | char *endp; | |
623 | ||
624 | *wp++ = '\0'; | |
625 | cost = strtol (rp, &endp, 10); | |
626 | if (rp == endp || cost < 1) | |
627 | /* No useful information. */ | |
628 | cost = 1; | |
629 | } | |
630 | ||
631 | if (module[0] == '\0') | |
632 | /* No module name given. */ | |
633 | return; | |
634 | ||
635 | /* See whether we must add the ending. */ | |
636 | need_ext = 0; | |
6dd67bd5 | 637 | if ((size_t) (wp - module) < sizeof (gconv_module_ext) |
6b98979f UD |
638 | || memcmp (wp - sizeof (gconv_module_ext), gconv_module_ext, |
639 | sizeof (gconv_module_ext)) != 0) | |
640 | /* We must add the module extension. */ | |
641 | need_ext = sizeof (gconv_module_ext) - 1; | |
642 | ||
6dd67bd5 UD |
643 | assert (strlen (from) + 1 == (size_t) (to - from)); |
644 | assert (strlen (to) + 1 == (size_t) (module - to)); | |
645 | assert (strlen (module) + 1 == (size_t) (wp - module)); | |
6b98979f UD |
646 | |
647 | new_module (from, to - from, to, module - to, directory, module, wp - module, | |
648 | cost, need_ext); | |
649 | } | |
650 | ||
0c78b0bb SP |
651 | /* Read config files and add the data for this directory to cache. */ |
652 | static int | |
653 | handle_dir (const char *dir) | |
654 | { | |
43cea6d5 | 655 | char *newp = NULL; |
0c78b0bb SP |
656 | size_t dirlen = strlen (dir); |
657 | bool found = false; | |
658 | ||
43cea6d5 SP |
659 | /* End directory path with a '/' if it doesn't already. */ |
660 | if (dir[dirlen - 1] != '/') | |
661 | { | |
662 | newp = xmalloc (dirlen + 2); | |
663 | memcpy (newp, dir, dirlen); | |
664 | newp[dirlen++] = '/'; | |
665 | newp[dirlen] = '\0'; | |
666 | dir = newp; | |
667 | } | |
eeac390e | 668 | |
43cea6d5 | 669 | found = gconv_parseconfdir (dir[0] == '/' ? prefix : NULL, dir, dirlen); |
e3217c7f | 670 | |
0c78b0bb SP |
671 | if (!found) |
672 | { | |
3979c3e1 | 673 | error (0, errno, "failed to open gconv configuration files in `%s'", |
0c78b0bb SP |
674 | dir); |
675 | error (0, 0, | |
3979c3e1 SP |
676 | "ensure that the directory contains either a valid " |
677 | "gconv-modules file or a gconv-modules.d directory with " | |
678 | "configuration files with names ending in .conf."); | |
0c78b0bb SP |
679 | } |
680 | ||
43cea6d5 | 681 | free (newp); |
9429049c | 682 | |
0c78b0bb | 683 | return found ? 0 : 1; |
6b98979f UD |
684 | } |
685 | ||
686 | ||
687 | static void | |
688 | append_module (const void *nodep, VISIT value, int level) | |
689 | { | |
690 | struct module *mo; | |
691 | ||
692 | if (value != leaf && value != postorder) | |
693 | return; | |
694 | ||
695 | mo = *(struct module **) nodep; | |
696 | ||
697 | if (nmodule_list > 0 | |
698 | && strcmp (module_list[nmodule_list - 1]->fromname, mo->fromname) == 0) | |
699 | { | |
700 | /* Same name. */ | |
701 | mo->next = module_list[nmodule_list - 1]; | |
702 | module_list[nmodule_list - 1] = mo; | |
703 | ||
704 | return; | |
705 | } | |
706 | ||
707 | if (nmodule_list_max == nmodule_list) | |
708 | { | |
709 | nmodule_list_max += 50; | |
710 | module_list = (struct module **) xrealloc (module_list, | |
711 | (nmodule_list_max | |
712 | * sizeof (struct module *))); | |
713 | } | |
714 | ||
715 | module_list[nmodule_list++] = mo; | |
716 | } | |
717 | ||
718 | ||
719 | static void | |
720 | get_modules (void) | |
721 | { | |
722 | twalk (modules, append_module); | |
723 | } | |
724 | ||
725 | ||
726 | static void | |
727 | add_builtins (void) | |
728 | { | |
729 | size_t cnt; | |
730 | ||
731 | /* Add all aliases. */ | |
732 | for (cnt = 0; cnt < nbuiltin_alias; ++cnt) | |
733 | new_alias (builtin_alias[cnt].from, | |
734 | strlen (builtin_alias[cnt].from) + 1, | |
735 | builtin_alias[cnt].to, | |
736 | strlen (builtin_alias[cnt].to) + 1); | |
737 | ||
738 | /* add the builtin transformations. */ | |
739 | for (cnt = 0; cnt < nbuiltin_trans; ++cnt) | |
740 | new_module (builtin_trans[cnt].from, | |
741 | strlen (builtin_trans[cnt].from) + 1, | |
742 | builtin_trans[cnt].to, | |
743 | strlen (builtin_trans[cnt].to) + 1, | |
744 | "", builtin_trans[cnt].module, | |
745 | strlen (builtin_trans[cnt].module) + 1, | |
746 | builtin_trans[cnt].cost, 0); | |
747 | } | |
748 | ||
749 | ||
750 | static int | |
751 | name_compare (const void *p1, const void *p2) | |
752 | { | |
753 | const struct name *n1 = (const struct name *) p1; | |
754 | const struct name *n2 = (const struct name *) p2; | |
755 | ||
756 | return strcmp (n1->name, n2->name); | |
757 | } | |
758 | ||
759 | ||
760 | static struct name * | |
761 | new_name (const char *str, struct Strent *strent) | |
762 | { | |
763 | struct name *newp = (struct name *) xmalloc (sizeof (struct name)); | |
764 | ||
765 | newp->name = str; | |
766 | newp->strent = strent; | |
767 | newp->module_idx = -1; | |
dd9423a6 | 768 | newp->hashval = __hash_string (str); |
6b98979f UD |
769 | |
770 | ++nnames; | |
771 | ||
772 | return newp; | |
773 | } | |
774 | ||
775 | ||
776 | static void | |
777 | generate_name_list (void) | |
778 | { | |
779 | size_t i; | |
780 | ||
b2e3d177 UD |
781 | /* A name we always need. */ |
782 | tsearch (new_name ("INTERNAL", strtabadd (strtab, "INTERNAL", | |
783 | sizeof ("INTERNAL"))), | |
784 | &names, name_compare); | |
785 | ||
6b98979f UD |
786 | for (i = 0; i < nmodule_list; ++i) |
787 | { | |
788 | struct module *runp; | |
789 | ||
790 | if (strcmp (module_list[i]->fromname, "INTERNAL") != 0) | |
791 | tsearch (new_name (module_list[i]->fromname, | |
792 | module_list[i]->fromname_strent), | |
793 | &names, name_compare); | |
794 | ||
795 | for (runp = module_list[i]; runp != NULL; runp = runp->next) | |
796 | if (strcmp (runp->toname, "INTERNAL") != 0) | |
797 | tsearch (new_name (runp->toname, runp->toname_strent), | |
798 | &names, name_compare); | |
799 | } | |
800 | } | |
801 | ||
802 | ||
803 | static int | |
804 | name_to_module_idx (const char *name, int add) | |
805 | { | |
806 | struct name **res; | |
807 | struct name fake_name = { .name = name }; | |
808 | int idx; | |
809 | ||
810 | res = (struct name **) tfind (&fake_name, &names, name_compare); | |
811 | if (res == NULL) | |
812 | abort (); | |
813 | ||
814 | idx = (*res)->module_idx; | |
815 | if (idx == -1 && add) | |
816 | /* No module index assigned yet. */ | |
817 | idx = (*res)->module_idx = nname_info++; | |
818 | ||
819 | return idx; | |
820 | } | |
821 | ||
822 | ||
823 | static void | |
824 | generate_name_info (void) | |
825 | { | |
826 | size_t i; | |
b2e3d177 | 827 | int idx; |
6b98979f | 828 | |
b2e3d177 | 829 | name_info = (struct name_info *) xcalloc (nmodule_list + 1, |
6b98979f UD |
830 | sizeof (struct name_info)); |
831 | ||
b2e3d177 UD |
832 | /* First add a special entry for the INTERNAL name. This must have |
833 | index zero. */ | |
834 | idx = name_to_module_idx ("INTERNAL", 1); | |
835 | name_info[0].canonical_name = "INTERNAL"; | |
836 | name_info[0].canonical_strent = strtabadd (strtab, "INTERNAL", | |
837 | sizeof ("INTERNAL")); | |
838 | assert (nname_info == 1); | |
839 | ||
6b98979f UD |
840 | for (i = 0; i < nmodule_list; ++i) |
841 | { | |
842 | struct module *runp; | |
843 | ||
844 | for (runp = module_list[i]; runp != NULL; runp = runp->next) | |
845 | if (strcmp (runp->fromname, "INTERNAL") == 0) | |
846 | { | |
b2e3d177 | 847 | idx = name_to_module_idx (runp->toname, 1); |
6b98979f UD |
848 | name_info[idx].from_internal = runp; |
849 | assert (name_info[idx].canonical_name == NULL | |
850 | || strcmp (name_info[idx].canonical_name, | |
851 | runp->toname) == 0); | |
852 | name_info[idx].canonical_name = runp->toname; | |
853 | name_info[idx].canonical_strent = runp->toname_strent; | |
854 | } | |
855 | else if (strcmp (runp->toname, "INTERNAL") == 0) | |
856 | { | |
b2e3d177 | 857 | idx = name_to_module_idx (runp->fromname, 1); |
6b98979f UD |
858 | name_info[idx].to_internal = runp; |
859 | assert (name_info[idx].canonical_name == NULL | |
860 | || strcmp (name_info[idx].canonical_name, | |
861 | runp->fromname) == 0); | |
862 | name_info[idx].canonical_name = runp->fromname; | |
863 | name_info[idx].canonical_strent = runp->fromname_strent; | |
864 | } | |
865 | else | |
866 | { | |
867 | /* This is a transformation not to or from the INTERNAL | |
868 | encoding. */ | |
869 | int from_idx = name_to_module_idx (runp->fromname, 1); | |
870 | int to_idx = name_to_module_idx (runp->toname, 1); | |
871 | struct other_conv_list *newp; | |
872 | ||
873 | newp = (struct other_conv_list *) | |
874 | xmalloc (sizeof (struct other_conv_list)); | |
875 | newp->other_conv.module_idx = to_idx; | |
876 | newp->other_conv.module = runp; | |
877 | newp->other_conv.next = NULL; /* XXX Allow multiple module sequence */ | |
878 | newp->dest_idx = to_idx; | |
879 | newp->next = name_info[from_idx].other_conv_list; | |
880 | name_info[from_idx].other_conv_list = newp; | |
881 | assert (name_info[from_idx].canonical_name == NULL | |
882 | || strcmp (name_info[from_idx].canonical_name, | |
883 | runp->fromname) == 0); | |
884 | name_info[from_idx].canonical_name = runp->fromname; | |
885 | name_info[from_idx].canonical_strent = runp->fromname_strent; | |
886 | ||
887 | ++nextra_modules; | |
888 | } | |
889 | } | |
890 | ||
891 | /* Now add the module index information for all the aliases. */ | |
892 | for (i = 0; i < nalias_list; ++i) | |
893 | { | |
894 | struct name fake_name = { .name = alias_list[i]->toname }; | |
895 | struct name **tonamep; | |
896 | ||
897 | tonamep = (struct name **) tfind (&fake_name, &names, name_compare); | |
898 | if (tonamep != NULL) | |
899 | { | |
900 | struct name *newp = new_name (alias_list[i]->fromname, | |
901 | alias_list[i]->froment); | |
902 | newp->module_idx = (*tonamep)->module_idx; | |
903 | tsearch (newp, &names, name_compare); | |
904 | } | |
905 | } | |
906 | } | |
907 | ||
908 | ||
909 | static int | |
910 | is_prime (unsigned long int candidate) | |
911 | { | |
912 | /* No even number and none less than 10 will be passed here. */ | |
913 | unsigned long int divn = 3; | |
914 | unsigned long int sq = divn * divn; | |
915 | ||
916 | while (sq < candidate && candidate % divn != 0) | |
917 | { | |
918 | ++divn; | |
919 | sq += 4 * divn; | |
920 | ++divn; | |
921 | } | |
922 | ||
923 | return candidate % divn != 0; | |
924 | } | |
925 | ||
926 | ||
927 | static uint32_t | |
928 | next_prime (uint32_t seed) | |
929 | { | |
930 | /* Make it definitely odd. */ | |
931 | seed |= 1; | |
932 | ||
933 | while (!is_prime (seed)) | |
934 | seed += 2; | |
935 | ||
936 | return seed; | |
937 | } | |
938 | ||
939 | ||
940 | /* Format of the output file. | |
941 | ||
942 | Offset Length Description | |
943 | 0000 4 Magic header bytes | |
2e2b6e26 UD |
944 | 0004 2 Offset of string table (stoff) |
945 | 0006 2 Offset of name hashing table (hoff) | |
946 | 0008 2 Hashing table size (hsize) | |
947 | 000A 2 Offset of module table (moff) | |
948 | 000C 2 Offset of other conversion module table (ooff) | |
6b98979f UD |
949 | |
950 | stoff ??? String table | |
951 | ||
952 | hoff 8*hsize Array of tuples | |
561470e0 UD |
953 | string table offset |
954 | module index | |
6b98979f UD |
955 | |
956 | moff ??? Array of tuples | |
561470e0 UD |
957 | canonical name offset |
958 | from-internal module dir name offset | |
959 | from-internal module name off | |
960 | to-internal module dir name offset | |
961 | to-internal module name offset | |
962 | offset into other conversion table | |
6b98979f | 963 | |
bbdf8285 | 964 | ooff ??? One or more of |
561470e0 UD |
965 | number of steps/modules |
966 | one or more of tuple | |
967 | canonical name offset for output | |
968 | module dir name offset | |
969 | module name offset | |
970 | (following last entry with step count 0) | |
6b98979f | 971 | */ |
b5f13526 UD |
972 | |
973 | static struct hash_entry *hash_table; | |
974 | static size_t hash_size; | |
975 | ||
976 | /* Function to insert the names. */ | |
977 | static void name_insert (const void *nodep, VISIT value, int level) | |
978 | { | |
979 | struct name *name; | |
980 | unsigned int idx; | |
981 | unsigned int hval2; | |
982 | ||
983 | if (value != leaf && value != postorder) | |
984 | return; | |
985 | ||
986 | name = *(struct name **) nodep; | |
987 | idx = name->hashval % hash_size; | |
988 | hval2 = 1 + name->hashval % (hash_size - 2); | |
989 | ||
990 | while (hash_table[idx].string_offset != 0) | |
991 | if ((idx += hval2) >= hash_size) | |
992 | idx -= hash_size; | |
993 | ||
994 | hash_table[idx].string_offset = strtaboffset (name->strent); | |
995 | ||
996 | assert (name->module_idx != -1); | |
997 | hash_table[idx].module_idx = name->module_idx; | |
998 | } | |
999 | ||
6b98979f UD |
1000 | static int |
1001 | write_output (void) | |
1002 | { | |
1003 | int fd; | |
1004 | char *string_table; | |
1005 | size_t string_table_size; | |
1006 | struct gconvcache_header header; | |
6b98979f UD |
1007 | struct module_entry *module_table; |
1008 | char *extra_table; | |
1009 | char *cur_extra_table; | |
1010 | size_t n; | |
1011 | int idx; | |
1012 | struct iovec iov[6]; | |
1013 | static const gidx_t null_word; | |
1014 | size_t total; | |
ed2bca56 RM |
1015 | char finalname[prefix_len + sizeof GCONV_MODULES_CACHE]; |
1016 | char tmpfname[(output_file == NULL ? sizeof finalname : output_file_len + 1) | |
d8d5a647 | 1017 | + strlen (".XXXXXX")]; |
6b98979f | 1018 | |
6b98979f | 1019 | /* Open the output file. */ |
ed2bca56 RM |
1020 | if (output_file == NULL) |
1021 | { | |
1022 | assert (GCONV_MODULES_CACHE[0] == '/'); | |
1023 | strcpy (stpcpy (mempcpy (tmpfname, prefix, prefix_len), | |
1024 | GCONV_MODULES_CACHE), | |
1025 | ".XXXXXX"); | |
1026 | strcpy (mempcpy (finalname, prefix, prefix_len), GCONV_MODULES_CACHE); | |
1027 | } | |
1028 | else | |
1029 | strcpy (mempcpy (tmpfname, output_file, output_file_len), ".XXXXXX"); | |
9fcddc2b | 1030 | fd = mkstemp (tmpfname); |
6b98979f UD |
1031 | if (fd == -1) |
1032 | return 1; | |
1033 | ||
1034 | /* Create the string table. */ | |
1035 | string_table = strtabfinalize (strtab, &string_table_size); | |
1036 | ||
1037 | /* Create the hashing table. We know how many strings we have. | |
1038 | Creating a perfect hash table is not reasonable here. Therefore | |
ac64195c | 1039 | we use open hashing and a table size which is the next prime 50% |
6b98979f | 1040 | larger than the number of strings. */ |
5abcddd7 | 1041 | hash_size = next_prime (nnames + (nnames >> 1)); |
6b98979f UD |
1042 | hash_table = (struct hash_entry *) xcalloc (hash_size, |
1043 | sizeof (struct hash_entry)); | |
1044 | /* Fill the hash table. */ | |
1045 | twalk (names, name_insert); | |
1046 | ||
1047 | /* Create the section for the module list. */ | |
1048 | module_table = (struct module_entry *) xcalloc (sizeof (struct module_entry), | |
1049 | nname_info); | |
1050 | ||
1051 | /* Allocate memory for the non-INTERNAL conversions. The allocated | |
1052 | memory can be more than is actually needed. */ | |
1053 | extra_table = (char *) xcalloc (sizeof (struct extra_entry) | |
1054 | + sizeof (gidx_t) | |
1055 | + sizeof (struct extra_entry_module), | |
1056 | nextra_modules); | |
1057 | cur_extra_table = extra_table; | |
1058 | ||
1059 | /* Fill in the module information. */ | |
1060 | for (n = 0; n < nname_info; ++n) | |
1061 | { | |
1062 | module_table[n].canonname_offset = | |
1063 | strtaboffset (name_info[n].canonical_strent); | |
1064 | ||
1065 | if (name_info[n].from_internal == NULL) | |
1066 | { | |
1067 | module_table[n].fromdir_offset = 0; | |
1068 | module_table[n].fromname_offset = 0; | |
1069 | } | |
1070 | else | |
1071 | { | |
1072 | module_table[n].fromdir_offset = | |
1073 | strtaboffset (name_info[n].from_internal->directory_strent); | |
1074 | module_table[n].fromname_offset = | |
1075 | strtaboffset (name_info[n].from_internal->filename_strent); | |
1076 | } | |
1077 | ||
1078 | if (name_info[n].to_internal == NULL) | |
1079 | { | |
1080 | module_table[n].todir_offset = 0; | |
1081 | module_table[n].toname_offset = 0; | |
1082 | } | |
1083 | else | |
1084 | { | |
1085 | module_table[n].todir_offset = | |
1086 | strtaboffset (name_info[n].to_internal->directory_strent); | |
1087 | module_table[n].toname_offset = | |
1088 | strtaboffset (name_info[n].to_internal->filename_strent); | |
1089 | } | |
1090 | ||
1091 | if (name_info[n].other_conv_list != NULL) | |
1092 | { | |
1093 | struct other_conv_list *other = name_info[n].other_conv_list; | |
1094 | ||
1095 | /* Store the reference. We add 1 to distinguish the entry | |
1096 | at offset zero from the case where no extra modules are | |
1097 | available. The file reader has to account for the | |
1098 | offset. */ | |
1099 | module_table[n].extra_offset = 1 + cur_extra_table - extra_table; | |
1100 | ||
1101 | do | |
1102 | { | |
1103 | struct other_conv *runp; | |
1104 | struct extra_entry *extra; | |
1105 | ||
1106 | /* Allocate new entry. */ | |
1107 | extra = (struct extra_entry *) cur_extra_table; | |
1108 | cur_extra_table += sizeof (struct extra_entry); | |
1109 | extra->module_cnt = 0; | |
1110 | ||
1111 | runp = &other->other_conv; | |
1112 | do | |
1113 | { | |
1114 | cur_extra_table += sizeof (struct extra_entry_module); | |
1115 | extra->module[extra->module_cnt].outname_offset = | |
1116 | runp->next == NULL | |
1117 | ? other->dest_idx : runp->next->module_idx; | |
1118 | extra->module[extra->module_cnt].dir_offset = | |
1119 | strtaboffset (runp->module->directory_strent); | |
1120 | extra->module[extra->module_cnt].name_offset = | |
1121 | strtaboffset (runp->module->filename_strent); | |
1122 | ++extra->module_cnt; | |
1123 | ||
1124 | runp = runp->next; | |
1125 | } | |
1126 | while (runp != NULL); | |
1127 | ||
1128 | other = other->next; | |
1129 | } | |
1130 | while (other != NULL); | |
1131 | ||
1132 | /* Final module_cnt is zero. */ | |
1133 | *((gidx_t *) cur_extra_table) = 0; | |
1134 | cur_extra_table += sizeof (gidx_t); | |
1135 | } | |
1136 | } | |
1137 | ||
2e2b6e26 UD |
1138 | /* Clear padding. */ |
1139 | memset (&header, 0, sizeof (struct gconvcache_header)); | |
1140 | ||
6b98979f UD |
1141 | header.magic = GCONVCACHE_MAGIC; |
1142 | ||
1143 | iov[0].iov_base = &header; | |
1144 | iov[0].iov_len = sizeof (struct gconvcache_header); | |
1145 | total = iov[0].iov_len; | |
1146 | ||
1147 | header.string_offset = total; | |
1148 | iov[1].iov_base = string_table; | |
1149 | iov[1].iov_len = string_table_size; | |
1150 | total += iov[1].iov_len; | |
1151 | ||
1152 | idx = 2; | |
1153 | if ((string_table_size & (sizeof (gidx_t) - 1)) != 0) | |
1154 | { | |
1155 | iov[2].iov_base = (void *) &null_word; | |
1156 | iov[2].iov_len = (sizeof (gidx_t) | |
1157 | - (string_table_size & (sizeof (gidx_t) - 1))); | |
1158 | total += iov[2].iov_len; | |
1159 | ++idx; | |
1160 | } | |
1161 | ||
1162 | header.hash_offset = total; | |
1163 | header.hash_size = hash_size; | |
1164 | iov[idx].iov_base = hash_table; | |
1165 | iov[idx].iov_len = hash_size * sizeof (struct hash_entry); | |
1166 | total += iov[idx].iov_len; | |
1167 | ++idx; | |
1168 | ||
1169 | header.module_offset = total; | |
1170 | iov[idx].iov_base = module_table; | |
1171 | iov[idx].iov_len = nname_info * sizeof (struct module_entry); | |
1172 | total += iov[idx].iov_len; | |
1173 | ++idx; | |
1174 | ||
6dd67bd5 | 1175 | assert ((size_t) (cur_extra_table - extra_table) |
6b98979f UD |
1176 | <= ((sizeof (struct extra_entry) + sizeof (gidx_t) |
1177 | + sizeof (struct extra_entry_module)) | |
1178 | * nextra_modules)); | |
1179 | header.otherconv_offset = total; | |
1180 | iov[idx].iov_base = extra_table; | |
1181 | iov[idx].iov_len = cur_extra_table - extra_table; | |
1182 | total += iov[idx].iov_len; | |
1183 | ++idx; | |
1184 | ||
6dd67bd5 | 1185 | if ((size_t) TEMP_FAILURE_RETRY (writev (fd, iov, idx)) != total |
9fcddc2b UD |
1186 | /* The file was created with mode 0600. Make it world-readable. */ |
1187 | || fchmod (fd, 0644) != 0 | |
1188 | /* Rename the file, possibly replacing an old one. */ | |
ed2bca56 | 1189 | || rename (tmpfname, output_file ?: finalname) != 0) |
6b98979f UD |
1190 | { |
1191 | int save_errno = errno; | |
1192 | close (fd); | |
9fcddc2b UD |
1193 | unlink (tmpfname); |
1194 | error (EXIT_FAILURE, save_errno, | |
1195 | gettext ("cannot generate output file")); | |
6b98979f UD |
1196 | } |
1197 | ||
1198 | close (fd); | |
1199 | ||
1200 | return 0; | |
1201 | } |