1 @node Message Translation, Searching and Sorting, Locales, Top
2 @c %MENU% How to make the program speak the user's language
3 @chapter Message Translation
5 The program's interface with the human should be designed in a way to
6 ease the human the task. One of the possibilities is to use messages in
7 whatever language the user prefers.
9 Printing messages in different languages can be implemented in different
10 ways. One could add all the different languages in the source code and
11 add among the variants every time a message has to be printed. This is
12 certainly no good solution since extending the set of languages is
13 difficult (the code must be changed) and the code itself can become
14 really big with dozens of message sets.
16 A better solution is to keep the message sets for each language are kept
17 in separate files which are loaded at runtime depending on the language
18 selection of the user.
20 The GNU C Library provides two different sets of functions to support
21 message translation. The problem is that neither of the interfaces is
22 officially defined by the POSIX standard. The @code{catgets} family of
23 functions is defined in the X/Open standard but this is derived from
24 industry decisions and therefore not necessarily based on reasonable
27 As mentioned above the message catalog handling provides easy
28 extendibility by using external data files which contain the message
29 translations. I.e., these files contain for each of the messages used
30 in the program a translation for the appropriate language. So the tasks
31 of the message handling functions are
35 locate the external data file with the appropriate translations.
37 load the data and make it possible to address the messages
39 map a given key to the translated message
42 The two approaches mainly differ in the implementation of this last
43 step. The design decisions made for this influences the whole rest.
46 * Message catalogs a la X/Open:: The @code{catgets} family of functions.
47 * The Uniforum approach:: The @code{gettext} family of functions.
51 @node Message catalogs a la X/Open
52 @section X/Open Message Catalog Handling
54 The @code{catgets} functions are based on the simple scheme:
57 Associate every message to translate in the source code with a unique
58 identifier. To retrieve a message from a catalog file solely the
62 This means for the author of the program that s/he will have to make
63 sure the meaning of the identifier in the program code and in the
64 message catalogs are always the same.
66 Before a message can be translated the catalog file must be located.
67 The user of the program must be able to guide the responsible function
68 to find whatever catalog the user wants. This is separated from what
69 the programmer had in mind.
71 All the types, constants and functions for the @code{catgets} functions
72 are defined/declared in the @file{nl_types.h} header file.
75 * The catgets Functions:: The @code{catgets} function family.
76 * The message catalog files:: Format of the message catalog files.
77 * The gencat program:: How to generate message catalogs files which
78 can be used by the functions.
79 * Common Usage:: How to use the @code{catgets} interface.
83 @node The catgets Functions
84 @subsection The @code{catgets} function family
88 @deftypefun nl_catd catopen (const char *@var{cat_name}, int @var{flag})
89 The @code{catgets} function tries to locate the message data file names
90 @var{cat_name} and loads it when found. The return value is of an
91 opaque type and can be used in calls to the other functions to refer to
94 The return value is @code{(nl_catd) -1} in case the function failed and
95 no catalog was loaded. The global variable @var{errno} contains a code
96 for the error causing the failure. But even if the function call
97 succeeded this does not mean that all messages can be translated.
99 Locating the catalog file must happen in a way which lets the user of
100 the program influence the decision. It is up to the user to decide
101 about the language to use and sometimes it is useful to use alternate
102 catalog files. All this can be specified by the user by setting some
103 environment variables.
105 The first problem is to find out where all the message catalogs are
106 stored. Every program could have its own place to keep all the
107 different files but usually the catalog files are grouped by languages
108 and the catalogs for all programs are kept in the same place.
110 @cindex NLSPATH environment variable
111 To tell the @code{catopen} function where the catalog for the program
112 can be found the user can set the environment variable @code{NLSPATH} to
113 a value which describes her/his choice. Since this value must be usable
114 for different languages and locales it cannot be a simple string.
115 Instead it is a format string (similar to @code{printf}'s). An example
119 /usr/share/locale/%L/%N:/usr/share/locale/%L/LC_MESSAGES/%N
122 First one can see that more than one directory can be specified (with
123 the usual syntax of separating them by colons). The next things to
124 observe are the format string, @code{%L} and @code{%N} in this case.
125 The @code{catopen} function knows about several of them and the
126 replacement for all of them is of course different.
130 This format element is substituted with the name of the catalog file.
131 This is the value of the @var{cat_name} argument given to
135 This format element is substituted with the name of the currently
136 selected locale for translating messages. How this is determined is
140 (This is the lowercase ell.) This format element is substituted with the
141 language element of the locale name. The string describing the selected
142 locale is expected to have the form
143 @code{@var{lang}[_@var{terr}[.@var{codeset}]]} and this format uses the
144 first part @var{lang}.
147 This format element is substituted by the territory part @var{terr} of
148 the name of the currently selected locale. See the explanation of the
152 This format element is substituted by the codeset part @var{codeset} of
153 the name of the currently selected locale. See the explanation of the
157 Since @code{%} is used in a meta character there must be a way to
158 express the @code{%} character in the result itself. Using @code{%%}
159 does this just like it works for @code{printf}.
163 Using @code{NLSPATH} allows arbitrary directories to be searched for
164 message catalogs while still allowing different languages to be used.
165 If the @code{NLSPATH} environment variable is not set, the default value
169 @var{prefix}/share/locale/%L/%N:@var{prefix}/share/locale/%L/LC_MESSAGES/%N
173 where @var{prefix} is given to @code{configure} while installing the GNU
174 C Library (this value is in many cases @code{/usr} or the empty string).
176 The remaining problem is to decide which must be used. The value
177 decides about the substitution of the format elements mentioned above.
178 First of all the user can specify a path in the message catalog name
179 (i.e., the name contains a slash character). In this situation the
180 @code{NLSPATH} environment variable is not used. The catalog must exist
181 as specified in the program, perhaps relative to the current working
182 directory. This situation in not desirable and catalogs names never
183 should be written this way. Beside this, this behavior is not portable
184 to all other platforms providing the @code{catgets} interface.
186 @cindex LC_ALL environment variable
187 @cindex LC_MESSAGES environment variable
188 @cindex LANG environment variable
189 Otherwise the values of environment variables from the standard
190 environment are examined (@pxref{Standard Environment}). Which
191 variables are examined is decided by the @var{flag} parameter of
192 @code{catopen}. If the value is @code{NL_CAT_LOCALE} (which is defined
193 in @file{nl_types.h}) then the @code{catopen} function use the name of
194 the locale currently selected for the @code{LC_MESSAGES} category.
196 If @var{flag} is zero the @code{LANG} environment variable is examined.
197 This is a left-over from the early days where the concept of the locales
198 had not even reached the level of POSIX locales.
200 The environment variable and the locale name should have a value of the
201 form @code{@var{lang}[_@var{terr}[.@var{codeset}]]} as explained above.
202 If no environment variable is set the @code{"C"} locale is used which
203 prevents any translation.
205 The return value of the function is in any case a valid string. Either
206 it is a translation from a message catalog or it is the same as the
207 @var{string} parameter. So a piece of code to decide whether a
208 translation actually happened must look like this:
212 char *trans = catgets (desc, set, msg, input_string);
213 if (trans == input_string)
215 /* Something went wrong. */
221 When an error occurred the global variable @var{errno} is set to
225 The catalog does not exist.
227 The set/message tuple does not name an existing element in the
231 While it sometimes can be useful to test for errors programs normally
232 will avoid any test. If the translation is not available it is no big
233 problem if the original, untranslated message is printed. Either the
234 user understands this as well or s/he will look for the reason why the
235 messages are not translated.
238 Please note that the currently selected locale does not depend on a call
239 to the @code{setlocale} function. It is not necessary that the locale
240 data files for this locale exist and calling @code{setlocale} succeeds.
241 The @code{catopen} function directly reads the values of the environment
245 @deftypefun {char *} catgets (nl_catd @var{catalog_desc}, int @var{set}, int @var{message}, const char *@var{string})
246 The function @code{catgets} has to be used to access the massage catalog
247 previously opened using the @code{catopen} function. The
248 @var{catalog_desc} parameter must be a value previously returned by
251 The next two parameters, @var{set} and @var{message}, reflect the
252 internal organization of the message catalog files. This will be
253 explained in detail below. For now it is interesting to know that a
254 catalog can consists of several set and the messages in each thread are
255 individually numbered using numbers. Neither the set number nor the
256 message number must be consecutive. They can be arbitrarily chosen.
257 But each message (unless equal to another one) must have its own unique
258 pair of set and message number.
260 Since it is not guaranteed that the message catalog for the language
261 selected by the user exists the last parameter @var{string} helps to
262 handle this case gracefully. If no matching string can be found
263 @var{string} is returned. This means for the programmer that
267 the @var{string} parameters should contain reasonable text (this also
268 helps to understand the program seems otherwise there would be no hint
269 on the string which is expected to be returned.
271 all @var{string} arguments should be written in the same language.
275 It is somewhat uncomfortable to write a program using the @code{catgets}
276 functions if no supporting functionality is available. Since each
277 set/message number tuple must be unique the programmer must keep lists
278 of the messages at the same time the code is written. And the work
279 between several people working on the same project must be coordinated.
280 We will see some how these problems can be relaxed a bit (@pxref{Common
283 @deftypefun int catclose (nl_catd @var{catalog_desc})
284 The @code{catclose} function can be used to free the resources
285 associated with a message catalog which previously was opened by a call
286 to @code{catopen}. If the resources can be successfully freed the
287 function returns @code{0}. Otherwise it return @code{@minus{}1} and the
288 global variable @var{errno} is set. Errors can occur if the catalog
289 descriptor @var{catalog_desc} is not valid in which case @var{errno} is
294 @node The message catalog files
295 @subsection Format of the message catalog files
297 The only reasonable way the translate all the messages of a function and
298 store the result in a message catalog file which can be read by the
299 @code{catopen} function is to write all the message text to the
300 translator and let her/him translate them all. I.e., we must have a
301 file with entries which associate the set/message tuple with a specific
302 translation. This file format is specified in the X/Open standard and
307 Lines containing only whitespace characters or empty lines are ignored.
310 Lines which contain as the first non-whitespace character a @code{$}
311 followed by a whitespace character are comment and are also ignored.
314 If a line contains as the first non-whitespace characters the sequence
315 @code{$set} followed by a whitespace character an additional argument
316 is required to follow. This argument can either be:
320 a number. In this case the value of this number determines the set
321 to which the following messages are added.
324 an identifier consisting of alphanumeric characters plus the underscore
325 character. In this case the set get automatically a number assigned.
326 This value is one added to the largest set number which so far appeared.
328 How to use the symbolic names is explained in section @ref{Common Usage}.
330 It is an error if a symbol name appears more than once. All following
331 messages are placed in a set with this number.
335 If a line contains as the first non-whitespace characters the sequence
336 @code{$delset} followed by a whitespace character an additional argument
337 is required to follow. This argument can either be:
341 a number. In this case the value of this number determines the set
342 which will be deleted.
345 an identifier consisting of alphanumeric characters plus the underscore
346 character. This symbolic identifier must match a name for a set which
347 previously was defined. It is an error if the name is unknown.
350 In both cases all messages in the specified set will be removed. They
351 will not appear in the output. But if this set is later again selected
352 with a @code{$set} command again messages could be added and these
353 messages will appear in the output.
356 If a line contains after leading whitespaces the sequence
357 @code{$quote}, the quoting character used for this input file is
358 changed to the first non-whitespace character following the
359 @code{$quote}. If no non-whitespace character is present before the
360 line ends quoting is disable.
362 By default no quoting character is used. In this mode strings are
363 terminated with the first unescaped line break. If there is a
364 @code{$quote} sequence present newline need not be escaped. Instead a
365 string is terminated with the first unescaped appearance of the quote
368 A common usage of this feature would be to set the quote character to
369 @code{"}. Then any appearance of the @code{"} in the strings must
370 be escaped using the backslash (i.e., @code{\"} must be written).
373 Any other line must start with a number or an alphanumeric identifier
374 (with the underscore character included). The following characters
375 (starting after the first whitespace character) will form the string
376 which gets associated with the currently selected set and the message
377 number represented by the number and identifier respectively.
379 If the start of the line is a number the message number is obvious. It
380 is an error if the same message number already appeared for this set.
382 If the leading token was an identifier the message number gets
383 automatically assigned. The value is the current maximum messages
384 number for this set plus one. It is an error if the identifier was
385 already used for a message in this set. It is OK to reuse the
386 identifier for a message in another thread. How to use the symbolic
387 identifiers will be explained below (@pxref{Common Usage}). There is
388 one limitation with the identifier: it must not be @code{Set}. The
389 reason will be explained below.
391 The text of the messages can contain escape characters. The usual bunch
392 of characters known from the @w{ISO C} language are recognized
393 (@code{\n}, @code{\t}, @code{\v}, @code{\b}, @code{\r}, @code{\f},
394 @code{\\}, and @code{\@var{nnn}}, where @var{nnn} is the octal coding of
398 @strong{Important:} The handling of identifiers instead of numbers for
399 the set and messages is a GNU extension. Systems strictly following the
400 X/Open specification do not have this feature. An example for a message
401 catalog file is this:
404 $ This is a leading comment.
409 two " Message with ID \"two\", which gets the value 2 assigned"
412 $ Since the last set got the number 1 assigned this set has number 2.
413 4000 "The numbers can be arbitrary, they need not start at one."
416 This small example shows various aspects:
419 Lines 1 and 9 are comments since they start with @code{$} followed by
422 The quoting character is set to @code{"}. Otherwise the quotes in the
423 message definition would have to be left away and in this case the
424 message with the identifier @code{two} would loose its leading whitespace.
426 Mixing numbered messages with message having symbolic names is no
427 problem and the numbering happens automatically.
431 While this file format is pretty easy it is not the best possible for
432 use in a running program. The @code{catopen} function would have to
433 parser the file and handle syntactic errors gracefully. This is not so
434 easy and the whole process is pretty slow. Therefore the @code{catgets}
435 functions expect the data in another more compact and ready-to-use file
436 format. There is a special program @code{gencat} which is explained in
437 detail in the next section.
439 Files in this other format are not human readable. To be easy to use by
440 programs it is a binary file. But the format is byte order independent
441 so translation files can be shared by systems of arbitrary architecture
442 (as long as they use the GNU C Library).
444 Details about the binary file format are not important to know since
445 these files are always created by the @code{gencat} program. The
446 sources of the GNU C Library also provide the sources for the
447 @code{gencat} program and so the interested reader can look through
448 these source files to learn about the file format.
451 @node The gencat program
452 @subsection Generate Message Catalogs files
455 The @code{gencat} program is specified in the X/Open standard and the
456 GNU implementation follows this specification and so processes
457 all correctly formed input files. Additionally some extension are
458 implemented which help to work in a more reasonable way with the
459 @code{catgets} functions.
461 The @code{gencat} program can be invoked in two ways:
464 `gencat [@var{Option}]@dots{} [@var{Output-File} [@var{Input-File}]@dots{}]`
467 This is the interface defined in the X/Open standard. If no
468 @var{Input-File} parameter is given input will be read from standard
469 input. Multiple input files will be read as if they are concatenated.
470 If @var{Output-File} is also missing, the output will be written to
471 standard output. To provide the interface one is used to from other
472 programs a second interface is provided.
475 `gencat [@var{Option}]@dots{} -o @var{Output-File} [@var{Input-File}]@dots{}`
478 The option @samp{-o} is used to specify the output file and all file
479 arguments are used as input files.
481 Beside this one can use @file{-} or @file{/dev/stdin} for
482 @var{Input-File} to denote the standard input. Corresponding one can
483 use @file{-} and @file{/dev/stdout} for @var{Output-File} to denote
484 standard output. Using @file{-} as a file name is allowed in X/Open
485 while using the device names is a GNU extension.
487 The @code{gencat} program works by concatenating all input files and
488 then @strong{merge} the resulting collection of message sets with a
489 possibly existing output file. This is done by removing all messages
490 with set/message number tuples matching any of the generated messages
491 from the output file and then adding all the new messages. To
492 regenerate a catalog file while ignoring the old contents therefore
493 requires to remove the output file if it exists. If the output is
494 written to standard output no merging takes place.
497 The following table shows the options understood by the @code{gencat}
498 program. The X/Open standard does not specify any option for the
499 program so all of these are GNU extensions.
504 Print the version information and exit.
507 Print a usage message listing all available options, then exit successfully.
509 Do never merge the new messages from the input files with the old content
510 of the output files. The old content of the output file is discarded.
513 This option is used to emit the symbolic names given to sets and
514 messages in the input files for use in the program. Details about how
515 to use this are given in the next section. The @var{name} parameter to
516 this option specifies the name of the output file. It will contain a
517 number of C preprocessor @code{#define}s to associate a name with a
520 Please note that the generated file only contains the symbols from the
521 input files. If the output is merged with the previous content of the
522 output file the possibly existing symbols from the file(s) which
523 generated the old output files are not in the generated header file.
528 @subsection How to use the @code{catgets} interface
530 The @code{catgets} functions can be used in two different ways. By
531 following slavishly the X/Open specs and not relying on the extension
532 and by using the GNU extensions. We will take a look at the former
533 method first to understand the benefits of extensions.
535 @subsubsection Not using symbolic names
537 Since the X/Open format of the message catalog files does not allow
538 symbol names we have to work with numbers all the time. When we start
539 writing a program we have to replace all appearances of translatable
540 strings with something like
543 catgets (catdesc, set, msg, "string")
547 @var{catgets} is retrieved from a call to @code{catopen} which is
548 normally done once at the program start. The @code{"string"} is the
549 string we want to translate. The problems start with the set and
552 In a bigger program several programmers usually work at the same time on
553 the program and so coordinating the number allocation is crucial.
554 Though no two different strings must be indexed by the same tuple of
555 numbers it is highly desirable to reuse the numbers for equal strings
556 with equal translations (please note that there might be strings which
557 are equal in one language but have different translations due to
558 difference contexts).
560 The allocation process can be relaxed a bit by different set numbers for
561 different parts of the program. So the number of developers who have to
562 coordinate the allocation can be reduced. But still lists must be keep
563 track of the allocation and errors can easily happen. These errors
564 cannot be discovered by the compiler or the @code{catgets} functions.
565 Only the user of the program might see wrong messages printed. In the
566 worst cases the messages are so irritating that they cannot be
567 recognized as wrong. Think about the translations for @code{"true"} and
568 @code{"false"} being exchanged. This could result in a disaster.
571 @subsubsection Using symbolic names
573 The problems mentioned in the last section derive from the fact that:
577 the numbers are allocated once and due to the possibly frequent use of
578 them it is difficult to change a number later.
580 the numbers do not allow to guess anything about the string and
581 therefore collisions can easily happen.
584 By constantly using symbolic names and by providing a method which maps
585 the string content to a symbolic name (however this will happen) one can
586 prevent both problems above. The cost of this is that the programmer
587 has to write a complete message catalog file while s/he is writing the
590 This is necessary since the symbolic names must be mapped to numbers
591 before the program sources can be compiled. In the last section it was
592 described how to generate a header containing the mapping of the names.
593 E.g., for the example message file given in the last section we could
594 call the @code{gencat} program as follow (assume @file{ex.msg} contains
598 gencat -H ex.h -o ex.cat ex.msg
602 This generates a header file with the following content:
605 #define SetTwoSet 0x2 /* ex.msg:8 */
607 #define SetOneSet 0x1 /* ex.msg:4 */
608 #define SetOnetwo 0x2 /* ex.msg:6 */
611 As can be seen the various symbols given in the source file are mangled
612 to generate unique identifiers and these identifiers get numbers
613 assigned. Reading the source file and knowing about the rules will
614 allow to predict the content of the header file (it is deterministic)
615 but this is not necessary. The @code{gencat} program can take care for
616 everything. All the programmer has to do is to put the generated header
617 file in the dependency list of the source files of her/his project and
618 to add a rules to regenerate the header of any of the input files
621 One word about the symbol mangling. Every symbol consists of two parts:
622 the name of the message set plus the name of the message or the special
623 string @code{Set}. So @code{SetOnetwo} means this macro can be used to
624 access the translation with identifier @code{two} in the message set
627 The other names denote the names of the message sets. The special
628 string @code{Set} is used in the place of the message identifier.
630 If in the code the second string of the set @code{SetOne} is used the C
631 code should look like this:
634 catgets (catdesc, SetOneSet, SetOnetwo,
635 " Message with ID \"two\", which gets the value 2 assigned")
638 Writing the function this way will allow to change the message number
639 and even the set number without requiring any change in the C source
640 code. (The text of the string is normally not the same; this is only
644 @subsubsection How does to this allow to develop
646 To illustrate the usual way to work with the symbolic version numbers
647 here is a little example. Assume we want to write the very complex and
648 famous greeting program. We start by writing the code as usual:
655 printf ("Hello, world!\n");
660 Now we want to internationalize the message and therefore replace the
661 message with whatever the user wants.
664 #include <nl_types.h>
670 nl_catd catdesc = catopen ("hello.cat", NL_CAT_LOCALE);
671 printf (catgets (catdesc, SetMainSet, SetMainHello,
678 We see how the catalog object is opened and the returned descriptor used
679 in the other function calls. It is not really necessary to check for
680 failure of any of the functions since even in these situations the
681 functions will behave reasonable. They simply will be return a
684 What remains unspecified here are the constants @code{SetMainSet} and
685 @code{SetMainHello}. These are the symbolic names describing the
686 message. To get the actual definitions which match the information in
687 the catalog file we have to create the message catalog source file and
688 process it using the @code{gencat} program.
691 $ Messages for the famous greeting program.
695 Hello "Hallo, Welt!\n"
698 Now we can start building the program (assume the message catalog source
699 file is named @file{hello.msg} and the program source file @file{hello.c}):
703 % gencat -H msgnrs.h -o hello.cat hello.msg
705 #define MainSet 0x1 /* hello.msg:4 */
706 #define MainHello 0x1 /* hello.msg:5 */
707 % gcc -o hello hello.c -I.
708 % cp hello.cat /usr/share/locale/de/LC_MESSAGES
717 The call of the @code{gencat} program creates the missing header file
718 @file{msgnrs.h} as well as the message catalog binary. The former is
719 used in the compilation of @file{hello.c} while the later is placed in a
720 directory in which the @code{catopen} function will try to locate it.
721 Please check the @code{LC_ALL} environment variable and the default path
722 for @code{catopen} presented in the description above.
725 @node The Uniforum approach
726 @section The Uniforum approach to Message Translation
728 Sun Microsystems tried to standardize a different approach to message
729 translation in the Uniforum group. There never was a real standard
730 defined but still the interface was used in Sun's operation systems.
731 Since this approach fits better in the development process of free
732 software it is also used throughout the GNU project and the GNU
733 @file{gettext} package provides support for this outside the GNU C
736 The code of the @file{libintl} from GNU @file{gettext} is the same as
737 the code in the GNU C Library. So the documentation in the GNU
738 @file{gettext} manual is also valid for the functionality here. The
739 following text will describe the library functions in detail. But the
740 numerous helper programs are not described in this manual. Instead
741 people should read the GNU @file{gettext} manual
742 (@pxref{Top,,GNU gettext utilities,gettext,Native Language Support Library and Tools}).
743 We will only give a short overview.
745 Though the @code{catgets} functions are available by default on more
746 systems the @code{gettext} interface is at least as portable as the
747 former. The GNU @file{gettext} package can be used wherever the
748 functions are not available.
752 * Message catalogs with gettext:: The @code{gettext} family of functions.
753 * Helper programs for gettext:: Programs to handle message catalogs
758 @node Message catalogs with gettext
759 @subsection The @code{gettext} family of functions
761 The paradigms underlying the @code{gettext} approach to message
762 translations is different from that of the @code{catgets} functions the
763 basic functionally is equivalent. There are functions of the following
767 * Translation with gettext:: What has to be done to translate a message.
768 * Locating gettext catalog:: How to determine which catalog to be used.
769 * Advanced gettext functions:: Additional functions for more complicated
771 * Charset conversion in gettext:: How to specify the output character set
773 * GUI program problems:: How to use @code{gettext} in GUI programs.
774 * Using gettextized software:: The possibilities of the user to influence
775 the way @code{gettext} works.
778 @node Translation with gettext
779 @subsubsection What has to be done to translate a message?
781 The @code{gettext} functions have a very simple interface. The most
782 basic function just takes the string which shall be translated as the
783 argument and it returns the translation. This is fundamentally
784 different from the @code{catgets} approach where an extra key is
785 necessary and the original string is only used for the error case.
787 If the string which has to be translated is the only argument this of
788 course means the string itself is the key. I.e., the translation will
789 be selected based on the original string. The message catalogs must
790 therefore contain the original strings plus one translation for any such
791 string. The task of the @code{gettext} function is it to compare the
792 argument string with the available strings in the catalog and return the
793 appropriate translation. Of course this process is optimized so that
794 this process is not more expensive than an access using an atomic key
795 like in @code{catgets}.
797 The @code{gettext} approach has some advantages but also some
798 disadvantages. Please see the GNU @file{gettext} manual for a detailed
799 discussion of the pros and cons.
801 All the definitions and declarations for @code{gettext} can be found in
802 the @file{libintl.h} header file. On systems where these functions are
803 not part of the C library they can be found in a separate library named
804 @file{libintl.a} (or accordingly different for shared libraries).
808 @deftypefun {char *} gettext (const char *@var{msgid})
809 The @code{gettext} function searches the currently selected message
810 catalogs for a string which is equal to @var{msgid}. If there is such a
811 string available it is returned. Otherwise the argument string
812 @var{msgid} is returned.
814 Please note that all though the return value is @code{char *} the
815 returned string must not be changed. This broken type results from the
816 history of the function and does not reflect the way the function should
819 Please note that above we wrote ``message catalogs'' (plural). This is
820 a specialty of the GNU implementation of these functions and we will
821 say more about this when we talk about the ways message catalogs are
822 selected (@pxref{Locating gettext catalog}).
824 The @code{gettext} function does not modify the value of the global
825 @var{errno} variable. This is necessary to make it possible to write
829 printf (gettext ("Operation failed: %m\n"));
832 Here the @var{errno} value is used in the @code{printf} function while
833 processing the @code{%m} format element and if the @code{gettext}
834 function would change this value (it is called before @code{printf} is
835 called) we would get a wrong message.
837 So there is no easy way to detect a missing message catalog beside
838 comparing the argument string with the result. But it is normally the
839 task of the user to react on missing catalogs. The program cannot guess
840 when a message catalog is really necessary since for a user who speaks
841 the language the program was developed in does not need any translation.
844 The remaining two functions to access the message catalog add some
845 functionality to select a message catalog which is not the default one.
846 This is important if parts of the program are developed independently.
847 Every part can have its own message catalog and all of them can be used
848 at the same time. The C library itself is an example: internally it
849 uses the @code{gettext} functions but since it must not depend on a
850 currently selected default message catalog it must specify all ambiguous
855 @deftypefun {char *} dgettext (const char *@var{domainname}, const char *@var{msgid})
856 The @code{dgettext} functions acts just like the @code{gettext}
857 function. It only takes an additional first argument @var{domainname}
858 which guides the selection of the message catalogs which are searched
859 for the translation. If the @var{domainname} parameter is the null
860 pointer the @code{dgettext} function is exactly equivalent to
861 @code{gettext} since the default value for the domain name is used.
863 As for @code{gettext} the return value type is @code{char *} which is an
864 anachronism. The returned string must never be modified.
869 @deftypefun {char *} dcgettext (const char *@var{domainname}, const char *@var{msgid}, int @var{category})
870 The @code{dcgettext} adds another argument to those which
871 @code{dgettext} takes. This argument @var{category} specifies the last
872 piece of information needed to localize the message catalog. I.e., the
873 domain name and the locale category exactly specify which message
874 catalog has to be used (relative to a given directory, see below).
876 The @code{dgettext} function can be expressed in terms of
877 @code{dcgettext} by using
880 dcgettext (domain, string, LC_MESSAGES)
887 dgettext (domain, string)
890 This also shows which values are expected for the third parameter. One
891 has to use the available selectors for the categories available in
892 @file{locale.h}. Normally the available values are @code{LC_CTYPE},
893 @code{LC_COLLATE}, @code{LC_MESSAGES}, @code{LC_MONETARY},
894 @code{LC_NUMERIC}, and @code{LC_TIME}. Please note that @code{LC_ALL}
895 must not be used and even though the names might suggest this, there is
896 no relation to the environments variables of this name.
898 The @code{dcgettext} function is only implemented for compatibility with
899 other systems which have @code{gettext} functions. There is not really
900 any situation where it is necessary (or useful) to use a different value
901 but @code{LC_MESSAGES} in for the @var{category} parameter. We are
902 dealing with messages here and any other choice can only be irritating.
904 As for @code{gettext} the return value type is @code{char *} which is an
905 anachronism. The returned string must never be modified.
908 When using the three functions above in a program it is a frequent case
909 that the @var{msgid} argument is a constant string. So it is worth to
910 optimize this case. Thinking shortly about this one will realize that
911 as long as no new message catalog is loaded the translation of a message
912 will not change. This optimization is actually implemented by the
913 @code{gettext}, @code{dgettext} and @code{dcgettext} functions.
916 @node Locating gettext catalog
917 @subsubsection How to determine which catalog to be used
919 The functions to retrieve the translations for a given message have a
920 remarkable simple interface. But to provide the user of the program
921 still the opportunity to select exactly the translation s/he wants and
922 also to provide the programmer the possibility to influence the way to
923 locate the search for catalogs files there is a quite complicated
924 underlying mechanism which controls all this. The code is complicated
927 Basically we have two different tasks to perform which can also be
928 performed by the @code{catgets} functions:
932 Locate the set of message catalogs. There are a number of files for
933 different languages and which all belong to the package. Usually they
934 are all stored in the filesystem below a certain directory.
936 There can be arbitrary many packages installed and they can follow
937 different guidelines for the placement of their files.
940 Relative to the location specified by the package the actual translation
941 files must be searched, based on the wishes of the user. I.e., for each
942 language the user selects the program should be able to locate the
946 This is the functionality required by the specifications for
947 @code{gettext} and this is also what the @code{catgets} functions are
948 able to do. But there are some problems unresolved:
952 The language to be used can be specified in several different ways.
953 There is no generally accepted standard for this and the user always
954 expects the program understand what s/he means. E.g., to select the
955 German translation one could write @code{de}, @code{german}, or
956 @code{deutsch} and the program should always react the same.
959 Sometimes the specification of the user is too detailed. If s/he, e.g.,
960 specifies @code{de_DE.ISO-8859-1} which means German, spoken in Germany,
961 coded using the @w{ISO 8859-1} character set there is the possibility
962 that a message catalog matching this exactly is not available. But
963 there could be a catalog matching @code{de} and if the character set
964 used on the machine is always @w{ISO 8859-1} there is no reason why this
965 later message catalog should not be used. (We call this @dfn{message
969 If a catalog for a wanted language is not available it is not always the
970 second best choice to fall back on the language of the developer and
971 simply not translate any message. Instead a user might be better able
972 to read the messages in another language and so the user of the program
973 should be able to define an precedence order of languages.
976 We can divide the configuration actions in two parts: the one is
977 performed by the programmer, the other by the user. We will start with
978 the functions the programmer can use since the user configuration will
981 As the functions described in the last sections already mention separate
982 sets of messages can be selected by a @dfn{domain name}. This is a
983 simple string which should be unique for each program part with uses a
984 separate domain. It is possible to use in one program arbitrary many
985 domains at the same time. E.g., the GNU C Library itself uses a domain
986 named @code{libc} while the program using the C Library could use a
987 domain named @code{foo}. The important point is that at any time
988 exactly one domain is active. This is controlled with the following
993 @deftypefun {char *} textdomain (const char *@var{domainname})
994 The @code{textdomain} function sets the default domain, which is used in
995 all future @code{gettext} calls, to @var{domainname}. Please note that
996 @code{dgettext} and @code{dcgettext} calls are not influenced if the
997 @var{domainname} parameter of these functions is not the null pointer.
999 Before the first call to @code{textdomain} the default domain is
1000 @code{messages}. This is the name specified in the specification of
1001 the @code{gettext} API. This name is as good as any other name. No
1002 program should ever really use a domain with this name since this can
1003 only lead to problems.
1005 The function returns the value which is from now on taken as the default
1006 domain. If the system went out of memory the returned value is
1007 @code{NULL} and the global variable @var{errno} is set to @code{ENOMEM}.
1008 Despite the return value type being @code{char *} the return string must
1009 not be changed. It is allocated internally by the @code{textdomain}
1012 If the @var{domainname} parameter is the null pointer no new default
1013 domain is set. Instead the currently selected default domain is
1016 If the @var{domainname} parameter is the empty string the default domain
1017 is reset to its initial value, the domain with the name @code{messages}.
1018 This possibility is questionable to use since the domain @code{messages}
1019 really never should be used.
1024 @deftypefun {char *} bindtextdomain (const char *@var{domainname}, const char *@var{dirname})
1025 The @code{bindtextdomain} function can be used to specify the directory
1026 which contains the message catalogs for domain @var{domainname} for the
1027 different languages. To be correct, this is the directory where the
1028 hierarchy of directories is expected. Details are explained below.
1030 For the programmer it is important to note that the translations which
1031 come with the program have be placed in a directory hierarchy starting
1032 at, say, @file{/foo/bar}. Then the program should make a
1033 @code{bindtextdomain} call to bind the domain for the current program to
1034 this directory. So it is made sure the catalogs are found. A correctly
1035 running program does not depend on the user setting an environment
1038 The @code{bindtextdomain} function can be used several times and if the
1039 @var{domainname} argument is different the previously bound domains
1040 will not be overwritten.
1042 If the program which wish to use @code{bindtextdomain} at some point of
1043 time use the @code{chdir} function to change the current working
1044 directory it is important that the @var{dirname} strings ought to be an
1045 absolute pathname. Otherwise the addressed directory might vary with
1048 If the @var{dirname} parameter is the null pointer @code{bindtextdomain}
1049 returns the currently selected directory for the domain with the name
1052 The @code{bindtextdomain} function returns a pointer to a string
1053 containing the name of the selected directory name. The string is
1054 allocated internally in the function and must not be changed by the
1055 user. If the system went out of core during the execution of
1056 @code{bindtextdomain} the return value is @code{NULL} and the global
1057 variable @var{errno} is set accordingly.
1061 @node Advanced gettext functions
1062 @subsubsection Additional functions for more complicated situations
1064 The functions of the @code{gettext} family described so far (and all the
1065 @code{catgets} functions as well) have one problem in the real world
1066 which have been neglected completely in all existing approaches. What
1067 is meant here is the handling of plural forms.
1069 Looking through Unix source code before the time anybody thought about
1070 internationalization (and, sadly, even afterwards) one can often find
1071 code similar to the following:
1074 printf ("%d file%s deleted", n, n == 1 ? "" : "s");
1078 After the first complains from people internationalizing the code people
1079 either completely avoided formulations like this or used strings like
1080 @code{"file(s)"}. Both look unnatural and should be avoided. First
1081 tries to solve the problem correctly looked like this:
1085 printf ("%d file deleted", n);
1087 printf ("%d files deleted", n);
1090 But this does not solve the problem. It helps languages where the
1091 plural form of a noun is not simply constructed by adding an `s' but
1092 that is all. Once again people fell into the trap of believing the
1093 rules their language is using are universal. But the handling of plural
1094 forms differs widely between the language families. There are two
1095 things we can differ between (and even inside language families);
1099 The form how plural forms are build differs. This is a problem with
1100 language which have many irregularities. German, for instance, is a
1101 drastic case. Though English and German are part of the same language
1102 family (Germanic), the almost regular forming of plural noun forms
1103 (appending an `s') is hardly found in German.
1106 The number of plural forms differ. This is somewhat surprising for
1107 those who only have experiences with Romanic and Germanic languages
1108 since here the number is the same (there are two).
1110 But other language families have only one form or many forms. More
1111 information on this in an extra section.
1114 The consequence of this is that application writers should not try to
1115 solve the problem in their code. This would be localization since it is
1116 only usable for certain, hardcoded language environments. Instead the
1117 extended @code{gettext} interface should be used.
1119 These extra functions are taking instead of the one key string two
1120 strings and an numerical argument. The idea behind this is that using
1121 the numerical argument and the first string as a key, the implementation
1122 can select using rules specified by the translator the right plural
1123 form. The two string arguments then will be used to provide a return
1124 value in case no message catalog is found (similar to the normal
1125 @code{gettext} behavior). In this case the rules for Germanic language
1126 is used and it is assumed that the first string argument is the singular
1127 form, the second the plural form.
1129 This has the consequence that programs without language catalogs can
1130 display the correct strings only if the program itself is written using
1131 a Germanic language. This is a limitation but since the GNU C library
1132 (as well as the GNU @code{gettext} package) are written as part of the
1133 GNU package and the coding standards for the GNU project require program
1134 being written in English, this solution nevertheless fulfills its
1139 @deftypefun {char *} ngettext (const char *@var{msgid1}, const char *@var{msgid2}, unsigned long int @var{n})
1140 The @code{ngettext} function is similar to the @code{gettext} function
1141 as it finds the message catalogs in the same way. But it takes two
1142 extra arguments. The @var{msgid1} parameter must contain the singular
1143 form of the string to be converted. It is also used as the key for the
1144 search in the catalog. The @var{msgid2} parameter is the plural form.
1145 The parameter @var{n} is used to determine the plural form. If no
1146 message catalog is found @var{msgid1} is returned if @code{n == 1},
1147 otherwise @code{msgid2}.
1149 An example for the us of this function is:
1152 printf (ngettext ("%d file removed", "%d files removed", n), n);
1155 Please note that the numeric value @var{n} has to be passed to the
1156 @code{printf} function as well. It is not sufficient to pass it only to
1162 @deftypefun {char *} dngettext (const char *@var{domain}, const char *@var{msgid1}, const char *@var{msgid2}, unsigned long int @var{n})
1163 The @code{dngettext} is similar to the @code{dgettext} function in the
1164 way the message catalog is selected. The difference is that it takes
1165 two extra parameter to provide the correct plural form. These two
1166 parameters are handled in the same way @code{ngettext} handles them.
1171 @deftypefun {char *} dcngettext (const char *@var{domain}, const char *@var{msgid1}, const char *@var{msgid2}, unsigned long int @var{n}, int @var{category})
1172 The @code{dcngettext} is similar to the @code{dcgettext} function in the
1173 way the message catalog is selected. The difference is that it takes
1174 two extra parameter to provide the correct plural form. These two
1175 parameters are handled in the same way @code{ngettext} handles them.
1178 @subsubheading The problem of plural forms
1180 A description of the problem can be found at the beginning of the last
1181 section. Now there is the question how to solve it. Without the input
1182 of linguists (which was not available) it was not possible to determine
1183 whether there are only a few different forms in which plural forms are
1184 formed or whether the number can increase with every new supported
1187 Therefore the solution implemented is to allow the translator to specify
1188 the rules of how to select the plural form. Since the formula varies
1189 with every language this is the only viable solution except for
1190 hardcoding the information in the code (which still would require the
1191 possibility of extensions to not prevent the use of new languages). The
1192 details are explained in the GNU @code{gettext} manual. Here only a a
1193 bit of information is provided.
1195 The information about the plural form selection has to be stored in the
1196 header entry (the one with the empty (@code{msgid} string). There should
1200 nplurals=2; plural=n == 1 ? 0 : 1
1203 The @code{nplurals} value must be a decimal number which specifies how
1204 many different plural forms exist for this language. The string
1205 following @code{plural} is an expression which is using the C language
1206 syntax. Exceptions are that no negative number are allowed, numbers
1207 must be decimal, and the only variable allowed is @code{n}. This
1208 expression will be evaluated whenever one of the functions
1209 @code{ngettext}, @code{dngettext}, or @code{dcngettext} is called. The
1210 numeric value passed to these functions is then substituted for all uses
1211 of the variable @code{n} in the expression. The resulting value then
1212 must be greater or equal to zero and smaller than the value given as the
1213 value of @code{nplurals}.
1216 The following rules are known at this point. The language with families
1217 are listed. But this does not necessarily mean the information can be
1218 generalized for the whole family (as can be easily seen in the table
1219 below).@footnote{Additions are welcome. Send appropriate information to
1220 @email{bug-glibc-manual@@gnu.org}.}
1223 @item Only one form:
1224 Some languages only require one single form. There is no distinction
1225 between the singular and plural form. And appropriate header entry
1226 would look like this:
1229 nplurals=1; plural=0
1233 Languages with this property include:
1236 @item Finno-Ugric family
1240 @item Turkic/Altaic family
1244 @item Two forms, singular used for one only
1245 This is the form used in most existing programs since it is what English
1246 is using. A header entry would look like this:
1249 nplurals=2; plural=n != 1
1252 (Note: this uses the feature of C expressions that boolean expressions
1253 have to value zero or one.)
1256 Languages with this property include:
1259 @item Germanic family
1260 Danish, Dutch, English, German, Norwegian, Swedish
1261 @item Finno-Ugric family
1263 @item Latin/Greek family
1265 @item Semitic family
1267 @item Romance family
1273 @item Two forms, singular used for zero and one
1274 Exceptional case in the language family. The header entry would be:
1277 nplurals=2; plural=n>1
1281 Languages with this property include:
1284 @item Romanic family
1288 @item Three forms, special cases for one and two
1289 The header entry would be:
1292 nplurals=3; plural=n==1 ? 0 : n==2 ? 1 : 2
1296 Languages with this property include:
1303 @item Three forms, special case for one and all numbers ending in 2, 3, or 4
1304 The header entry would look like this:
1307 nplurals=3; plural=n==1 ? 0 : n%10>=2 && n%10<=4 ? 1 : 2
1311 Languages with this property include:
1318 @item Three forms, special case for one and some numbers ending in 2, 3, or 4
1319 The header entry would look like this:
1322 nplurals=3; plural=n==1 ? 0 : \
1323 n%10>=2 && n%10<=4 && (n%100<10 || n%100>=20) ? 1 : 2
1326 (Continuation in the next line is possible.)
1329 Languages with this property include:
1336 @item Four forms, special case for one and all numbers ending in 2, 3, or 4
1337 The header entry would look like this:
1340 nplurals=4; plural=n==1 ? 0 : n%10==2 ? 1 : n%10==3 || n%10==4 ? 2 : 3
1344 Languages with this property include:
1353 @node Charset conversion in gettext
1354 @subsubsection How to specify the output character set @code{gettext} uses
1356 @code{gettext} not only looks up a translation in a message catalog. It
1357 also converts the translation on the fly to the desired output character
1358 set. This is useful if the user is working in a different character set
1359 than the translator who created the message catalog, because it avoids
1360 distributing variants of message catalogs which differ only in the
1363 The output character set is, by default, the value of @code{nl_langinfo
1364 (CODESET)}, which depends on the @code{LC_CTYPE} part of the current
1365 locale. But programs which store strings in a locale independent way
1366 (e.g. UTF-8) can request that @code{gettext} and related functions
1367 return the translations in that encoding, by use of the
1368 @code{bind_textdomain_codeset} function.
1370 Note that the @var{msgid} argument to @code{gettext} is not subject to
1371 character set conversion. Also, when @code{gettext} does not find a
1372 translation for @var{msgid}, it returns @var{msgid} unchanged --
1373 independently of the current output character set. It is therefore
1374 recommended that all @var{msgid}s be US-ASCII strings.
1378 @deftypefun {char *} bind_textdomain_codeset (const char *@var{domainname}, const char *@var{codeset})
1379 The @code{bind_textdomain_codeset} function can be used to specify the
1380 output character set for message catalogs for domain @var{domainname}.
1381 The @var{codeset} argument must be a valid codeset name which can be used
1382 for the @code{iconv_open} function, or a null pointer.
1384 If the @var{codeset} parameter is the null pointer,
1385 @code{bind_textdomain_codeset} returns the currently selected codeset
1386 for the domain with the name @var{domainname}. It returns @code{NULL} if
1387 no codeset has yet been selected.
1389 The @code{bind_textdomain_codeset} function can be used several times.
1390 If used multiple times with the same @var{domainname} argument, the
1391 later call overrides the settings made by the earlier one.
1393 The @code{bind_textdomain_codeset} function returns a pointer to a
1394 string containing the name of the selected codeset. The string is
1395 allocated internally in the function and must not be changed by the
1396 user. If the system went out of core during the execution of
1397 @code{bind_textdomain_codeset}, the return value is @code{NULL} and the
1398 global variable @var{errno} is set accordingly. @end deftypefun
1401 @node GUI program problems
1402 @subsubsection How to use @code{gettext} in GUI programs
1404 One place where the @code{gettext} functions, if used normally, have big
1405 problems is within programs with graphical user interfaces (GUIs). The
1406 problem is that many of the strings which have to be translated are very
1407 short. They have to appear in pull-down menus which restricts the
1408 length. But strings which are not containing entire sentences or at
1409 least large fragments of a sentence may appear in more than one
1410 situation in the program but might have different translations. This is
1411 especially true for the one-word strings which are frequently used in
1414 As a consequence many people say that the @code{gettext} approach is
1415 wrong and instead @code{catgets} should be used which indeed does not
1416 have this problem. But there is a very simple and powerful method to
1417 handle these kind of problems with the @code{gettext} functions.
1420 As as example consider the following fictional situation. A GUI program
1421 has a menu bar with the following entries:
1424 +------------+------------+--------------------------------------+
1425 | File | Printer | |
1426 +------------+------------+--------------------------------------+
1429 +----------+ | Connect |
1433 To have the strings @code{File}, @code{Printer}, @code{Open},
1434 @code{New}, @code{Select}, and @code{Connect} translated there has to be
1435 at some point in the code a call to a function of the @code{gettext}
1436 family. But in two places the string passed into the function would be
1437 @code{Open}. The translations might not be the same and therefore we
1438 are in the dilemma described above.
1440 One solution to this problem is to artificially enlengthen the strings
1441 to make them unambiguous. But what would the program do if no
1442 translation is available? The enlengthened string is not what should be
1443 printed. So we should use a little bit modified version of the functions.
1445 To enlengthen the strings a uniform method should be used. E.g., in the
1446 example above the strings could be chosen as
1455 Menu|Printer|Connect
1458 Now all the strings are different and if now instead of @code{gettext}
1459 the following little wrapper function is used, everything works just
1465 sgettext (const char *msgid)
1467 char *msgval = gettext (msgid);
1468 if (msgval == msgid)
1469 msgval = strrchr (msgid, '|') + 1;
1474 What this little function does is to recognize the case when no
1475 translation is available. This can be done very efficiently by a
1476 pointer comparison since the return value is the input value. If there
1477 is no translation we know that the input string is in the format we used
1478 for the Menu entries and therefore contains a @code{|} character. We
1479 simply search for the last occurrence of this character and return a
1480 pointer to the character following it. That's it!
1482 If one now consistently uses the enlengthened string form and replaces
1483 the @code{gettext} calls with calls to @code{sgettext} (this is normally
1484 limited to very few places in the GUI implementation) then it is
1485 possible to produce a program which can be internationalized.
1487 With advanced compilers (such as GNU C) one can write the
1488 @code{sgettext} functions as an inline function or as a macro like this:
1492 #define sgettext(msgid) \
1493 (@{ const char *__msgid = (msgid); \
1494 char *__msgstr = gettext (__msgid); \
1495 if (__msgval == __msgid) \
1496 __msgval = strrchr (__msgid, '|') + 1; \
1500 The other @code{gettext} functions (@code{dgettext}, @code{dcgettext}
1501 and the @code{ngettext} equivalents) can and should have corresponding
1502 functions as well which look almost identical, except for the parameters
1503 and the call to the underlying function.
1505 Now there is of course the question why such functions do not exist in
1506 the GNU C library? There are two parts of the answer to this question.
1510 They are easy to write and therefore can be provided by the project they
1511 are used in. This is not an answer by itself and must be seen together
1512 with the second part which is:
1515 There is no way the C library can contain a version which can work
1516 everywhere. The problem is the selection of the character to separate
1517 the prefix from the actual string in the enlenghtened string. The
1518 examples above used @code{|} which is a quite good choice because it
1519 resembles a notation frequently used in this context and it also is a
1520 character not often used in message strings.
1522 But what if the character is used in message strings. Or if the chose
1523 character is not available in the character set on the machine one
1524 compiles (e.g., @code{|} is not required to exist for @w{ISO C}; this is
1525 why the @file{iso646.h} file exists in @w{ISO C} programming environments).
1528 There is only one more comment to make left. The wrapper function above
1529 require that the translations strings are not enlengthened themselves.
1530 This is only logical. There is no need to disambiguate the strings
1531 (since they are never used as keys for a search) and one also saves
1532 quite some memory and disk space by doing this.
1535 @node Using gettextized software
1536 @subsubsection User influence on @code{gettext}
1538 The last sections described what the programmer can do to
1539 internationalize the messages of the program. But it is finally up to
1540 the user to select the message s/he wants to see. S/He must understand
1543 The POSIX locale model uses the environment variables @code{LC_COLLATE},
1544 @code{LC_CTYPE}, @code{LC_MESSAGES}, @code{LC_MONETARY}, @code{NUMERIC},
1545 and @code{LC_TIME} to select the locale which is to be used. This way
1546 the user can influence lots of functions. As we mentioned above the
1547 @code{gettext} functions also take advantage of this.
1549 To understand how this happens it is necessary to take a look at the
1550 various components of the filename which gets computed to locate a
1551 message catalog. It is composed as follows:
1554 @var{dir_name}/@var{locale}/LC_@var{category}/@var{domain_name}.mo
1557 The default value for @var{dir_name} is system specific. It is computed
1558 from the value given as the prefix while configuring the C library.
1559 This value normally is @file{/usr} or @file{/}. For the former the
1560 complete @var{dir_name} is:
1566 We can use @file{/usr/share} since the @file{.mo} files containing the
1567 message catalogs are system independent, so all systems can use the same
1568 files. If the program executed the @code{bindtextdomain} function for
1569 the message domain that is currently handled, the @code{dir_name}
1570 component is exactly the value which was given to the function as
1571 the second parameter. I.e., @code{bindtextdomain} allows overwriting
1572 the only system dependent and fixed value to make it possible to
1573 address files anywhere in the filesystem.
1575 The @var{category} is the name of the locale category which was selected
1576 in the program code. For @code{gettext} and @code{dgettext} this is
1577 always @code{LC_MESSAGES}, for @code{dcgettext} this is selected by the
1578 value of the third parameter. As said above it should be avoided to
1579 ever use a category other than @code{LC_MESSAGES}.
1581 The @var{locale} component is computed based on the category used. Just
1582 like for the @code{setlocale} function here comes the user selection
1583 into the play. Some environment variables are examined in a fixed order
1584 and the first environment variable set determines the return value of
1585 the lookup process. In detail, for the category @code{LC_xxx} the
1586 following variables in this order are examined:
1595 This looks very familiar. With the exception of the @code{LANGUAGE}
1596 environment variable this is exactly the lookup order the
1597 @code{setlocale} function uses. But why introducing the @code{LANGUAGE}
1600 The reason is that the syntax of the values these variables can have is
1601 different to what is expected by the @code{setlocale} function. If we
1602 would set @code{LC_ALL} to a value following the extended syntax that
1603 would mean the @code{setlocale} function will never be able to use the
1604 value of this variable as well. An additional variable removes this
1605 problem plus we can select the language independently of the locale
1606 setting which sometimes is useful.
1608 While for the @code{LC_xxx} variables the value should consist of
1609 exactly one specification of a locale the @code{LANGUAGE} variable's
1610 value can consist of a colon separated list of locale names. The
1611 attentive reader will realize that this is the way we manage to
1612 implement one of our additional demands above: we want to be able to
1613 specify an ordered list of language.
1615 Back to the constructed filename we have only one component missing.
1616 The @var{domain_name} part is the name which was either registered using
1617 the @code{textdomain} function or which was given to @code{dgettext} or
1618 @code{dcgettext} as the first parameter. Now it becomes obvious that a
1619 good choice for the domain name in the program code is a string which is
1620 closely related to the program/package name. E.g., for the GNU C
1621 Library the domain name is @code{libc}.
1624 A limit piece of example code should show how the programmer is supposed
1629 setlocale (LC_ALL, "");
1630 textdomain ("test-package");
1631 bindtextdomain ("test-package", "/usr/local/share/locale");
1632 puts (gettext ("Hello, world!"));
1636 At the program start the default domain is @code{messages}, and the
1637 default locale is "C". The @code{setlocale} call sets the locale
1638 according to the user's environment variables; remember that correct
1639 functioning of @code{gettext} relies on the correct setting of the
1640 @code{LC_MESSAGES} locale (for looking up the message catalog) and
1641 of the @code{LC_CTYPE} locale (for the character set conversion).
1642 The @code{textdomain} call changes the default domain to
1643 @code{test-package}. The @code{bindtextdomain} call specifies that
1644 the message catalogs for the domain @code{test-package} can be found
1645 below the directory @file{/usr/local/share/locale}.
1647 If now the user set in her/his environment the variable @code{LANGUAGE}
1648 to @code{de} the @code{gettext} function will try to use the
1649 translations from the file
1652 /usr/local/share/locale/de/LC_MESSAGES/test-package.mo
1655 From the above descriptions it should be clear which component of this
1656 filename is determined by which source.
1658 In the above example we assumed that the @code{LANGUAGE} environment
1659 variable to @code{de}. This might be an appropriate selection but what
1660 happens if the user wants to use @code{LC_ALL} because of the wider
1661 usability and here the required value is @code{de_DE.ISO-8859-1}? We
1662 already mentioned above that a situation like this is not infrequent.
1663 E.g., a person might prefer reading a dialect and if this is not
1664 available fall back on the standard language.
1666 The @code{gettext} functions know about situations like this and can
1667 handle them gracefully. The functions recognize the format of the value
1668 of the environment variable. It can split the value is different pieces
1669 and by leaving out the only or the other part it can construct new
1670 values. This happens of course in a predictable way. To understand
1671 this one must know the format of the environment variable value. There
1672 are to more or less standardized forms:
1676 @code{language[_territory[.codeset]][@@modifier]}
1678 @item CEN Format (European Community Standard)
1679 @code{language[_territory][+audience][+special][,[sponsor][_revision]]}
1682 The functions will automatically recognize which format is used. Less
1683 specific locale names will be stripped of in the order of the following
1696 @code{normalized codeset}
1700 @code{audience}/@code{modifier}
1703 From the last entry one can see that the meaning of the @code{modifier}
1704 field in the X/Open format and the @code{audience} format have the same
1705 meaning. Beside one can see that the @code{language} field for obvious
1706 reasons never will be dropped.
1708 The only new thing is the @code{normalized codeset} entry. This is
1709 another goodie which is introduced to help reducing the chaos which
1710 derives from the inability of the people to standardize the names of
1711 character sets. Instead of @w{ISO-8859-1} one can often see @w{8859-1},
1712 @w{88591}, @w{iso8859-1}, or @w{iso_8859-1}. The @code{normalized
1713 codeset} value is generated from the user-provided character set name by
1714 applying the following rules:
1718 Remove all characters beside numbers and letters.
1720 Fold letters to lowercase.
1722 If the same only contains digits prepend the string @code{"iso"}.
1726 So all of the above name will be normalized to @code{iso88591}. This
1727 allows the program user much more freely choosing the locale name.
1729 Even this extended functionality still does not help to solve the
1730 problem that completely different names can be used to denote the same
1731 locale (e.g., @code{de} and @code{german}). To be of help in this
1732 situation the locale implementation and also the @code{gettext}
1733 functions know about aliases.
1735 The file @file{/usr/share/locale/locale.alias} (replace @file{/usr} with
1736 whatever prefix you used for configuring the C library) contains a
1737 mapping of alternative names to more regular names. The system manager
1738 is free to add new entries to fill her/his own needs. The selected
1739 locale from the environment is compared with the entries in the first
1740 column of this file ignoring the case. If they match the value of the
1741 second column is used instead for the further handling.
1743 In the description of the format of the environment variables we already
1744 mentioned the character set as a factor in the selection of the message
1745 catalog. In fact, only catalogs which contain text written using the
1746 character set of the system/program can be used (directly; there will
1747 come a solution for this some day). This means for the user that s/he
1748 will always have to take care for this. If in the collection of the
1749 message catalogs there are files for the same language but coded using
1750 different character sets the user has to be careful.
1753 @node Helper programs for gettext
1754 @subsection Programs to handle message catalogs for @code{gettext}
1756 The GNU C Library does not contain the source code for the programs to
1757 handle message catalogs for the @code{gettext} functions. As part of
1758 the GNU project the GNU gettext package contains everything the
1759 developer needs. The functionality provided by the tools in this
1760 package by far exceeds the abilities of the @code{gencat} program
1761 described above for the @code{catgets} functions.
1763 There is a program @code{msgfmt} which is the equivalent program to the
1764 @code{gencat} program. It generates from the human-readable and
1765 -editable form of the message catalog a binary file which can be used by
1766 the @code{gettext} functions. But there are several more programs
1769 The @code{xgettext} program can be used to automatically extract the
1770 translatable messages from a source file. I.e., the programmer need not
1771 take care for the translations and the list of messages which have to be
1772 translated. S/He will simply wrap the translatable string in calls to
1773 @code{gettext} et.al and the rest will be done by @code{xgettext}. This
1774 program has a lot of option which help to customize the output or do
1775 help to understand the input better.
1777 Other programs help to manage development cycle when new messages appear
1778 in the source files or when a new translation of the messages appear.
1779 here it should only be noted that using all the tools in GNU gettext it
1780 is possible to @emph{completely} automize the handling of message
1781 catalog. Beside marking the translatable string in the source code and
1782 generating the translations the developers do not have anything to do