1 /* Convert text in given files from the specified from-set to the to-set.
2 Copyright (C) 1998,1999,2000,2001,2002,2003 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998.
6 The GNU C Library is free software; you can redistribute it and/or
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.
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
14 Lesser General Public License for more details.
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
37 #ifdef _POSIX_MAPPED_FILES
38 # include <sys/mman.h>
41 #include <gconv_int.h>
42 #include "iconv_prog.h"
43 #include "iconvconfig.h"
45 /* Get libc version number. */
46 #include "../version.h"
48 #define PACKAGE _libc_intl_domainname
51 /* Name and version of program. */
52 static void print_version (FILE *stream
, struct argp_state
*state
);
53 void (*argp_program_version_hook
) (FILE *, struct argp_state
*) = print_version
;
55 #define OPT_VERBOSE 1000
58 /* Definitions of arguments for argp functions. */
59 static const struct argp_option options
[] =
61 { NULL
, 0, NULL
, 0, N_("Input/Output format specification:") },
62 { "from-code", 'f', "NAME", 0, N_("encoding of original text") },
63 { "to-code", 't', "NAME", 0, N_("encoding for output") },
64 { NULL
, 0, NULL
, 0, N_("Information:") },
65 { "list", 'l', NULL
, 0, N_("list all known coded character sets") },
66 { NULL
, 0, NULL
, 0, N_("Output control:") },
67 { NULL
, 'c', NULL
, 0, N_("omit invalid characters from output") },
68 { "output", 'o', "FILE", 0, N_("output file") },
69 { "silent", 's', NULL
, 0, N_("suppress warnings") },
70 { "verbose", OPT_VERBOSE
, NULL
, 0, N_("print progress information") },
71 { NULL
, 0, NULL
, 0, NULL
}
74 /* Short description of program. */
75 static const char doc
[] = N_("\
76 Convert encoding of given files from one encoding to another.");
78 /* Strings for arguments in help texts. */
79 static const char args_doc
[] = N_("[FILE...]");
81 /* Prototype for option handler. */
82 static error_t
parse_opt (int key
, char *arg
, struct argp_state
*state
);
84 /* Function to print some extra text in the help message. */
85 static char *more_help (int key
, const char *text
, void *input
);
87 /* Data structure to communicate with argp functions. */
88 static struct argp argp
=
90 options
, parse_opt
, args_doc
, doc
, NULL
, more_help
93 /* Code sets to convert from and to respectively. An empty string as the
94 default causes the 'iconv_open' function to look up the charset of the
95 currently selected locale and use it. */
96 static const char *from_code
= "";
97 static const char *to_code
= "";
99 /* File to write output to. If NULL write to stdout. */
100 static const char *output_file
;
102 /* Nonzero if verbose ouput is wanted. */
105 /* Nonzero if list of all coded character sets is wanted. */
108 /* If nonzero omit invalid character from output. */
111 /* Prototypes for the functions doing the actual work. */
112 static int process_block (iconv_t cd
, char *addr
, size_t len
, FILE *output
);
113 static int process_fd (iconv_t cd
, int fd
, FILE *output
);
114 static int process_file (iconv_t cd
, FILE *input
, FILE *output
);
115 static void print_known_names (void) internal_function
;
119 main (int argc
, char *argv
[])
121 int status
= EXIT_SUCCESS
;
125 const char *orig_to_code
;
126 struct charmap_t
*from_charmap
= NULL
;
127 struct charmap_t
*to_charmap
= NULL
;
129 /* Set locale via LC_ALL. */
130 setlocale (LC_ALL
, "");
132 /* Set the text message domain. */
133 textdomain (_libc_intl_domainname
);
135 /* Parse and process arguments. */
136 argp_parse (&argp
, argc
, argv
, 0, &remaining
, NULL
);
138 /* List all coded character sets if wanted. */
141 print_known_names ();
145 /* If we have to ignore errors make sure we use the appropriate name for
146 the to-character-set. */
147 orig_to_code
= to_code
;
150 const char *errhand
= strchrnul (to_code
, '/');
158 errhand
= strchrnul (errhand
, '/');
163 errhand
= strchr (errhand
, '\0');
167 newp
= (char *) alloca (errhand
- to_code
+ nslash
+ 7 + 1);
168 cp
= mempcpy (newp
, to_code
, errhand
- to_code
);
173 memcpy (cp
, "IGNORE", sizeof ("IGNORE"));
178 /* POSIX 1003.2b introduces a silly thing: the arguments to -t anf -f
179 can be file names of charmaps. In this case iconv will have to read
180 those charmaps and use them to do the conversion. But there are
181 holes in the specification. There is nothing said that if -f is a
182 charmap filename that -t must be, too. And vice versa. There is
183 also no word about the symbolic names used. What if they don't
185 if (strchr (from_code
, '/') != NULL
)
186 /* The from-name might be a charmap file name. Try reading the
188 from_charmap
= charmap_read (from_code
, /*0, 1*/1, 0, 0);
190 if (strchr (orig_to_code
, '/') != NULL
)
191 /* The to-name might be a charmap file name. Try reading the
193 to_charmap
= charmap_read (orig_to_code
, /*0, 1,*/1,0, 0);
196 /* Determine output file. */
197 if (output_file
!= NULL
&& strcmp (output_file
, "-") != 0)
199 output
= fopen (output_file
, "w");
201 error (EXIT_FAILURE
, errno
, _("cannot open output file"));
206 /* At this point we have to handle two cases. The first one is
207 where a charmap is used for the from- or to-charset, or both. We
208 handle this special since it is very different from the sane way of
209 doing things. The other case allows converting using the iconv()
211 if (from_charmap
!= NULL
|| to_charmap
!= NULL
)
212 /* Construct the conversion table and do the conversion. */
213 status
= charmap_conversion (from_code
, from_charmap
, to_code
, to_charmap
,
214 argc
, remaining
, argv
, output
);
217 /* Let's see whether we have these coded character sets. */
218 cd
= iconv_open (to_code
, from_code
);
219 if (cd
== (iconv_t
) -1)
223 /* Try to be nice with the user and tell her which of the
224 two encoding names is wrong. This is possible because
225 all supported encodings can be converted from/to Unicode,
226 in other words, because the graph of encodings is
229 (iconv_open ("UTF-8", from_code
) == (iconv_t
) -1
232 (iconv_open (to_code
, "UTF-8") == (iconv_t
) -1
234 const char *from_pretty
=
235 (from_code
[0] ? from_code
: nl_langinfo (CODESET
));
236 const char *to_pretty
=
237 (orig_to_code
[0] ? orig_to_code
: nl_langinfo (CODESET
));
242 error (EXIT_FAILURE
, 0,
244 conversion from `%s' and to `%s' are not supported"),
245 from_pretty
, to_pretty
);
247 error (EXIT_FAILURE
, 0,
248 _("conversion from `%s' is not supported"),
254 error (EXIT_FAILURE
, 0,
255 _("conversion to `%s' is not supported"),
258 error (EXIT_FAILURE
, 0,
259 _("conversion from `%s' to `%s' is not supported"),
260 from_pretty
, to_pretty
);
264 error (EXIT_FAILURE
, errno
,
265 _("failed to start conversion processing"));
268 /* Now process the remaining files. Write them to stdout or the file
269 specified with the `-o' parameter. If we have no file given as
270 the parameter process all from stdin. */
271 if (remaining
== argc
)
273 if (process_file (cd
, stdin
, output
) != 0)
274 status
= EXIT_FAILURE
;
279 #ifdef _POSIX_MAPPED_FILES
286 printf ("%s:\n", argv
[remaining
]);
287 if (strcmp (argv
[remaining
], "-") == 0)
291 fd
= open (argv
[remaining
], O_RDONLY
);
295 error (0, errno
, _("cannot open input file `%s'"),
297 status
= EXIT_FAILURE
;
302 #ifdef _POSIX_MAPPED_FILES
303 /* We have possibilities for reading the input file. First try
304 to mmap() it since this will provide the fastest solution. */
305 if (fstat (fd
, &st
) == 0
306 && ((addr
= mmap (NULL
, st
.st_size
, PROT_READ
, MAP_PRIVATE
,
307 fd
, 0)) != MAP_FAILED
))
309 /* Yes, we can use mmap(). The descriptor is not needed
312 error (EXIT_FAILURE
, errno
,
313 _("error while closing input `%s'"),
316 if (process_block (cd
, addr
, st
.st_size
, output
) < 0)
318 /* Something went wrong. */
319 status
= EXIT_FAILURE
;
321 /* We don't need the input data anymore. */
322 munmap ((void *) addr
, st
.st_size
);
324 /* We cannot go on with producing output since it might
325 lead to problem because the last output might leave
326 the output stream in an undefined state. */
330 /* We don't need the input data anymore. */
331 munmap ((void *) addr
, st
.st_size
);
334 #endif /* _POSIX_MAPPED_FILES */
336 /* Read the file in pieces. */
337 if (process_fd (cd
, fd
, output
) != 0)
339 /* Something went wrong. */
340 status
= EXIT_FAILURE
;
342 /* We don't need the input file anymore. */
345 /* We cannot go on with producing output since it might
346 lead to problem because the last output might leave
347 the output stream in an undefined state. */
351 /* Now close the file. */
355 while (++remaining
< argc
);
358 /* Close the output file now. */
360 error (EXIT_FAILURE
, errno
, _("error while closing output file"));
366 /* Handle program arguments. */
368 parse_opt (int key
, char *arg
, struct argp_state
*state
)
382 /* Nothing, for now at least. We are not giving out any information
383 about missing character or so. */
386 /* Omit invalid characters from output. */
396 return ARGP_ERR_UNKNOWN
;
403 more_help (int key
, const char *text
, void *input
)
407 case ARGP_KEY_HELP_EXTRA
:
408 /* We print some extra information. */
409 return strdup (gettext ("\
410 Report bugs using the `glibcbug' script to <bugs@gnu.org>.\n"));
414 return (char *) text
;
418 /* Print the version information. */
420 print_version (FILE *stream
, struct argp_state
*state
)
422 fprintf (stream
, "iconv (GNU %s) %s\n", PACKAGE
, VERSION
);
423 fprintf (stream
, gettext ("\
424 Copyright (C) %s Free Software Foundation, Inc.\n\
425 This is free software; see the source for copying conditions. There is NO\n\
426 warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n\
428 fprintf (stream
, gettext ("Written by %s.\n"), "Ulrich Drepper");
433 process_block (iconv_t cd
, char *addr
, size_t len
, FILE *output
)
435 #define OUTBUF_SIZE 32768
436 const char *start
= addr
;
437 char outbuf
[OUTBUF_SIZE
];
445 outlen
= OUTBUF_SIZE
;
446 n
= iconv (cd
, &addr
, &len
, &outptr
, &outlen
);
448 if (outptr
!= outbuf
)
450 /* We have something to write out. */
451 int errno_save
= errno
;
453 if (fwrite (outbuf
, 1, outptr
- outbuf
, output
)
454 < (size_t) (outptr
- outbuf
)
457 /* Error occurred while printing the result. */
459 conversion stopped due to problem in writing the output"));
466 if (n
!= (size_t) -1)
468 /* All the input test is processed. For state-dependent
469 character sets we have to flush the state now. */
471 outlen
= OUTBUF_SIZE
;
472 (void) iconv (cd
, NULL
, NULL
, &outptr
, &outlen
);
474 if (outptr
!= outbuf
)
476 /* We have something to write out. */
477 int errno_save
= errno
;
479 if (fwrite (outbuf
, 1, outptr
- outbuf
, output
)
480 < (size_t) (outptr
- outbuf
)
483 /* Error occurred while printing the result. */
485 conversion stopped due to problem in writing the output"));
497 /* iconv() ran into a problem. */
502 error (0, 0, _("illegal input sequence at position %ld"),
503 (long int) (addr
- start
));
507 incomplete character or shift sequence at end of buffer"));
510 error (0, 0, _("internal error (illegal descriptor)"));
513 error (0, 0, _("unknown iconv() error %d"), errno
);
526 process_fd (iconv_t cd
, int fd
, FILE *output
)
528 /* we have a problem with reading from a desriptor since we must not
529 provide the iconv() function an incomplete character or shift
530 sequence at the end of the buffer. Since we have to deal with
531 arbitrary encodings we must read the whole text in a buffer and
532 process it in one step. */
533 static char *inbuf
= NULL
;
534 static size_t maxlen
= 0;
538 while (actlen
< maxlen
)
540 ssize_t n
= read (fd
, inptr
, maxlen
- actlen
);
543 /* No more text to read. */
548 /* Error while reading. */
549 error (0, errno
, _("error while reading the input"));
557 if (actlen
== maxlen
)
563 /* Increase the buffer. */
564 new_inbuf
= (char *) realloc (inbuf
, maxlen
+ 32768);
565 if (new_inbuf
== NULL
)
567 error (0, errno
, _("unable to allocate buffer for input"));
572 inptr
= inbuf
+ actlen
;
576 n
= read (fd
, inptr
, maxlen
- actlen
);
579 /* No more text to read. */
584 /* Error while reading. */
585 error (0, errno
, _("error while reading the input"));
592 while (actlen
< maxlen
);
595 /* Break again so we leave both loops. */
599 /* Now we have all the input in the buffer. Process it in one run. */
600 return process_block (cd
, inbuf
, actlen
, output
);
605 process_file (iconv_t cd
, FILE *input
, FILE *output
)
607 /* This should be safe since we use this function only for `stdin' and
608 we haven't read anything so far. */
609 return process_fd (cd
, fileno (input
), output
);
613 /* Print all known character sets/encodings. */
614 static void *printlist
;
615 static size_t column
;
616 static int not_first
;
619 insert_print_list (const void *nodep
, VISIT value
, int level
)
621 if (value
== leaf
|| value
== postorder
)
623 const struct gconv_alias
*s
= *(const struct gconv_alias
**) nodep
;
624 tsearch (s
->fromname
, &printlist
, (__compar_fn_t
) strverscmp
);
629 do_print_human (const void *nodep
, VISIT value
, int level
)
631 if (value
== leaf
|| value
== postorder
)
633 const char *s
= *(const char **) nodep
;
634 size_t len
= strlen (s
);
637 while (len
> 0 && s
[len
- 1] == '/')
640 for (cnt
= 0; cnt
< len
; ++cnt
)
641 if (isalnum (s
[cnt
]))
651 if (column
> 2 && column
+ len
> 77)
653 fputs ("\n ", stdout
);
665 fwrite (s
, len
, 1, stdout
);
671 do_print (const void *nodep
, VISIT value
, int level
)
673 if (value
== leaf
|| value
== postorder
)
675 const char *s
= *(const char **) nodep
;
683 add_known_names (struct gconv_module
*node
)
685 if (node
->left
!= NULL
)
686 add_known_names (node
->left
);
687 if (node
->right
!= NULL
)
688 add_known_names (node
->right
);
691 if (strcmp (node
->from_string
, "INTERNAL"))
692 tsearch (node
->from_string
, &printlist
,
693 (__compar_fn_t
) strverscmp
);
694 if (strcmp (node
->to_string
, "INTERNAL") != 0)
695 tsearch (node
->to_string
, &printlist
, (__compar_fn_t
) strverscmp
);
699 while (node
!= NULL
);
706 const struct gconvcache_header
*header
;
708 const struct hash_entry
*hashtab
;
711 header
= (const struct gconvcache_header
*) __gconv_get_cache ();
712 strtab
= (char *) header
+ header
->string_offset
;
713 hashtab
= (struct hash_entry
*) ((char *) header
+ header
->hash_offset
);
715 for (cnt
= 0; cnt
< header
->hash_size
; ++cnt
)
716 if (hashtab
[cnt
].string_offset
!= 0)
718 const char *str
= strtab
+ hashtab
[cnt
].string_offset
;
720 if (strcmp (str
, "INTERNAL") != 0)
721 tsearch (str
, &printlist
, (__compar_fn_t
) strverscmp
);
728 print_known_names (void)
733 /* We must initialize the internal databases first. */
734 h
= iconv_open ("L1", "L1");
737 /* See whether we have a cache. */
738 cache
= __gconv_get_cache ();
740 /* Yep, use only this information. */
744 struct gconv_module
*modules
;
746 /* No, then use the information read from the gconv-modules file.
747 First add the aliases. */
748 twalk (__gconv_get_alias_db (), insert_print_list
);
750 /* Add the from- and to-names from the known modules. */
751 modules
= __gconv_get_modules_db ();
753 add_known_names (modules
);
757 The following list contain all the coded character sets known. This does\n\
758 not necessarily mean that all combinations of these names can be used for\n\
759 the FROM and TO command line parameters. One coded character set can be\n\
760 listed with several different names (aliases).\n\n "), stdout
);
762 /* Now print the collected names. */
764 if (isatty (fileno (stdout
)))
766 twalk (printlist
, do_print_human
);
772 twalk (printlist
, do_print
);