X11R6 Sample Implementation Frame Work

This document proposes to define the structures, methods and their signatures that are expected to be common to all locale dependent functions within the Xlib sample implementation. The following illustration (Fig.1) is proposed to outline the separating of the components within the sample implementation.

Preface drawing.

Generally speaking, the internationalized portion of Xlib (Locale Dependent X, LDX) consists of three objects; locale (LC) , input method (IM) and output method (OM). The LC provides a set of information that depends on user's language environment. The IM manages text inputing, and the OM manages text drawing. Both IM and OM highly depend on LC data.

In X11R5, there are two sample implementations, Ximp and Xsi, for Xlib internationalization. But in both implementations, IM and OM actually refer the private extension of LC. It breaks coexistence of these two sample implementations. For example, if a user creates a new OM for special purpose as a part of Ximp, it will not work with Xsi.

As a solution of this problem, we propose to define the standard APIs between these three objects, and define the structure that are common to these objects.

This chapter describes functions related locale object binding for implementing the pluggable layer.

A locale loader is an entry point for locale object, which instantiates XLCd object and binds locale methods with specified locale name. The behavior of loader is implementation dependent. And, what kind of loaders are available is also implementation dependent.

The loader is called in _XOpenLC, but caller of _XOpenLC does not need to care about its inside. For example, if the loader is implemented with dynamic load functions, and the dynamic module is expected to be unloaded when the corresponding XLCd is freed, close methods of XLCdMethods should handle unloading.

Initializing a locale loader list

void _XlcInitLoader

The _XlcInitLoader function initializes the locale loader list with vendor specific manner. Each loader is registered with calling _XlcAddLoader. The number of loaders and their order in the loader list is implementation dependent.

Add a loader

The _XlcAddLoader function registers the specified locale loader "proc" to the internal loader list. The position specifies that the loader "proc" should be placed in the top of the loader list(XlcHead) or last(XlcTail).

The object loader is called from the top of the loader list in order, when calling time.

Remove a loader

The _XlcRemoveLoader function removes the locale loader specified by "proc" from the loader list.

Current implementation provides following locale loaders;

This chapter describes the locale method API, which is a set of accessible functions from both IM and OM parts. The locale method API provides the functionalities; obtaining locale dependent information, handling charset, converting text, etc.

As a result of using these APIs instead of accessing vender private extension of the locale object, we can keep locale, IM and OM independently each other.

Open a Locale Method

The _XOpenLC function opens a locale method which corresponds to the specified locale name. _XOpenLC calls a locale object loader, which is registered via _XlcAddLoader into the internal loader list. If the called loader is valid and successfully opens a locale, _XOpenLC returns the XLCd. If the loader is invalid or failed to open a locale, _XOpenLC calls the next loader. If all registered loaders cannot open a locale, _XOpenLC returns NULL.

XLCd _XlcCurrentLC

The _XlcCurrentLC function returns an XLCd that are bound to current locale.

Close a Locale Method

The _XCloseLC function close a locale method the specified lcd.

Obtain Locale Method values

The _XGetLCValues function returns NULL if no error occurred; otherwise, it returns the name of the first argument that could not be obtained. The following values are defined as standard arguments. Other values are implementation dependent.

The XlcCharSet is an identifier which represents a subset of characters (character set) in the locale object.

typedef enum {
      XlcUnknown, XlcC0, XlcGL, XlcC1, XlcGR, XlcGLGR, XlcOther
} XlcSide;

typedef struct _XlcCharSetRec *XlcCharSet;

typedef struct {
      char *name;
      XPointer value;
} XlcArg, *XlcArgList;

typedef char* (*XlcGetCSValuesProc)(charset, args, num_args);
      XlcCharSet charset;
      XlcArgList args;
      int num_args;

typedef struct _XlcCharSetRec {
      char *name;
      XrmQuark xrm_name;
      char *encoding_name;
      XrmQuark xrm_encoding_name;
      XlcSide side;
      int char_size;
      int set_size;
      char *ct_sequence;
      XlcGetCSValuesProc get_values;
} XlcCharSetRec;

Get an XlcCharSet

The _XlcGetCharSet function gets an XlcCharSet which corresponds to the charset name specified by "name". _XlcGetCharSet returns NULL, if no XlcCharSet bound to specified "name".

The following character sets are pre-registered.

Add an XlcCharSet

The _XlcAddCharSet function registers XlcCharSet specified by "charset".

Obtain Character Set values

The _XlcGetCSValues function returns NULL if no error occurred; otherwise, it returns the name of the first argument that could not be obtained. The following values are defined as standard arguments. Other values are implementation dependent.

We provide a set of the common converter APIs, that are independent from both of source and destination text type.

typedef struct _XlcConvRec *XlcConv;

typedef void (*XlcCloseConverterProc)(conv);
      XlcConv conv;

typedef int (*XlcConvertProc)(conv, from, from_left, to, to_left, args, num_args);
      XlcConv conv;
      XPointer *from;
      int *from_left;
      XPointer *to;
      int *to_left;
      XPointer *args;
      int num_args;

typedef void (*XlcResetConverterProc)(conv);
      XlcConv conv;

typedef struct _XlcConvMethodsRec {
      XlcCloseConverterProc close;
      XlcConvertProc convert;
      XlcResetConverterProc reset;
} XlcConvMethodsRec, *XlcConvMethods;

typedef struct _XlcConvRec {
    XlcConvMethods methods;
    XPointer state;
} XlcConvRec;

Open a converter

_XlcOpenConverter function opens the converter which converts a text from specified "from_type" to specified "to_type" encoding. If the function cannot find proper converter or cannot open a corresponding converter, it returns NULL. Otherwise, it returns the conversion descriptor.

The following types are pre-defined. Other types are implementation dependent.

Close a converter

The _XlcCloseConverter function closes the specified converter "conv".

Code conversion

The _XlcConvert function converts a sequence of characters from one type, in the array specified by "from", into a sequence of corresponding characters in another type, in the array specified by "to". The types are those specified in the _XlcOpenConverter() call that returned the conversion descriptor, "conv". The arguments "from", "from_left", "to" and "to_left" have the same specification of XPG4 iconv function.

For state-dependent encodings, the conversion descriptor "conv" is placed into its initial shift state by a call for which "from" is a NULL pointer, or for which "from" points to a null pointer.

The following 2 converters prepared by locale returns appropriate charset (XlcCharSet) in an area pointed by args[0].

The conversion, from XlcNMultiByte/XlcNWideChar to XlcNCharSet, extracts a segment which has same charset encoding characters. More than one segment cannot be converted in a call.

Reset a converter

The _XlcResetConverter function reset the specified converter "conv".

Register a converter

typedef XlcConv (*XlcOpenConverterProc)(from_lcd, from_type, to_lcd, to_type);
      XLCd from_lcd;
      char *from_type;
      XLCd to_lcd;
      char *to_type;

The XlcSetConverter function registers a converter which convert from "from_type" to "to_type" into the converter list (in the specified XLCd).

X Locale Database contains the subset of user's environment that depends on language. The following APIs are provided for accessing X Locale Database and other locale relative files.

For more detail about X Locale Database, please refer X Locale Database Definition document.

Get a resource from database

The _XlcGetResource function obtains a locale dependent data which is associated with the locale of specified "lcd". The locale data is provided by system locale or by X Locale Database file, and what kind of data is available is implementation dependent.

The specified "category" and "class" are used for finding out the objective locale data.

The returned value is returned in value argument in string list form, and the returned count shows the number of strings in the value.

The returned value is owned by locale method, and should not be modified or freed by caller.

Get a locale relative file name

The _XlcFileName functions returns a file name which is bound to the specified "lcd" and "category", as a null-terminated string. If no file name can be found, or there is no readable file for the found file name, _XlcFileName returns NULL. The returned file name should be freed by caller.

The rule for searching a file name is implementation dependent. In current implementation, _XlcFileName uses "{category}.dir" file as mapping table, which has pairs of strings, a full locale name and a corresponding file name.

Compare Latin-1 strings

The _XlcCompareIsoLatin1 function to compares two ISO-8859-1 strings. Bytes representing ASCII lower case letters are converted to upper case before making the comparison. The value returned is an integer less than, equal to, or greater than zero, depending on whether "str1" is lexicographicly less than, equal to, or greater than "str2".

The _XlcNCompareIsoLatin1 function is identical to _XlcCompareISOLatin1, except that at most "len" bytes are compared.

Resource Utility

Similar to XtNumber.

Similar to _XtCopyFromArg and _XtCopyToArg.

Similar to _XtCountVaList.

Similar to _XtVaToArgList.

typedef struct _XlcResource {
      char *name;
      XrmQuark xrm_name;
      int size;
      int offset;
      unsigned long mask;
} XlcResource, *XlcResourceList;

#define    XlcCreateMask   (1L<<0)
#define    XlcDefaultMask  (1L<<1)
#define    XlcGetMask      (1L<<2)
#define    XlcSetMask      (1L<<3)
#define    XlcIgnoreMask   (1L<<4)

Similar to _XtCompileResourceList.

Similar to XtGetSubvalues.

Similar to XtSetSubvalues.

ANSI C Compatible Functions

The following are ANSI C/MSE Compatible Functions for non-ANSI C environment.

The _Xmblen function returns the number of characters pointed to by "str". Only "len" bytes in "str" are used in determining the character count returned. "Str" may point at characters from any valid codeset in the current locale.

The call _Xmblen is equivalent to _Xmbtowc(_Xmbtowc((wchar_t*)NULL, str, len))

The _Xmbtowc function converts the character(s) pointed to by "str" to their wide character representation(s) pointed to by "wstr". "Len" is the number of bytes in "str" to be converted. The return value is the number of characters converted.

The call _Xmbtowc is equivalent to _Xlcmbtowc((XLCd)NULL, wstr, str, len)

The _Xlcmbtowc function is identical to _Xmbtowc, except that it requires the "lcd" argument. If "lcd" is (XLCd) NULL, _Xlcmbtowc, calls _XlcCurrentLC to determine the current locale.

The _Xwctomb function converts a single wide character pointed to by "wc" to its multibyte representation pointed to by "str". On success, the return value is 1.

The call _Xwctomb is equivalent to _Xlcwctomb((XLCd)NULL, str, wstr)

The _Xlcwctomb function is identical to _Xwctomb, except that it requires the "lcd" argument. If "lcd" is (XLCd) NULL, _Xlcwctomb, calls _XlcCurrentLC to determine the current locale.

The _Xmbstowcs function converts the NULL-terminated string pointed to by "str" to its wide character string representation pointed to by "wstr". "Len" is the number of characters in "str" to be converted.

The call _Xmbstowcs is equivalent to _Xlcmbstowcs((XLCd)NULL, wstr, str, len)

The _Xlcmbstowcs function is identical to _Xmbstowcs, except that it requires the "lcd" argument. If "lcd" is (XLCd) NULL, _Xlcmbstowcs, calls _XlcCurrentLC to determine the current locale.

The _Xwcstombs function converts the (wchar_t) NULL terminated wide character string pointed to by "wstr" to the NULL terminated multibyte string pointed to by "str".

The call _Xwcstombs is equivalent to _Xlcwcstombs((XLCd)NULL, str, wstr, len)

The _Xlcwcstombs function is identical to _Xwcstombs, except that it requires the "lcd" argument. If "lcd" is (XLCd) NULL, _Xlcwcstombs, calls _XlcCurrentLC to determine the current locale.

The _Xwcslen function returns the count of wide characters in the (wchar_t) NULL terminated wide character string pointed to by "wstr".

The _Xwcscpy function copies the (wchar_t) NULL terminated wide character string pointed to by "wstr2" to the object pointed at by "wstr1". "Wstr1" is (wchar_t) NULL terminated. The return value is a pointer to "wstr1".

The _Xwcsncpy function is identical to _Xwcscpy, except that it copies "len" wide characters from the object pointed to by "wstr2" to the object pointed to "wstr1".

The _Xwcscmp function compares two (wchar_t) NULL terminated wide character strings. The value returned is an integer less than, equal to, or greater than zero, depending on whether "wstr1" is lexicographicly less then, equal to, or greater than "str2".

The _Xwcsncmp function is identical to _XlcCompareISOLatin1, except that at most "len" wide characters are compared.