The big thing is locale dependent collation support.

Also added outdigit support and (legacy) YESSTR/NOSTR support.

1 files changed, 18 insertions, 187 deletions

diff --git a/extra/locale/gen_ldc.c b/extra/locale/gen_ldc.c
index 17e6c6875..399587d96 100644
--- a/extra/locale/gen_ldc.c
+++ b/extra/locale/gen_ldc.c
@@ -39,6 +39,7 @@
  #define WCuplow_diff_TBL_LEN (2 * WCuplow_diffs)
 /*  #define WCcomb_TBL_LEN		(WCcomb_II_LEN + WCcomb_TI_LEN + WCcomb_UT_LEN) */
 
+#include "locale_collate.h"
 #include "locale_tables.h"
 
 #include "locale_mmap.h"
@@ -69,7 +70,7 @@
 
 
 static const size_t common_tbl_offsets[CATEGORIES*4] = {
-	0, 0, 0, 0,					/* ctype */
+	COMMON_OFFSETS(ctype),
 	COMMON_OFFSETS(numeric),
 	COMMON_OFFSETS(monetary),
 	COMMON_OFFSETS(time),
@@ -190,6 +191,7 @@ int main(void)
 /* 	const unsigned char tblwcomb[WCcomb_TBL_LEN]; */
 	/* width?? */
 #endif /* __WCHAR_ENABLED */
+	out_uc(lso, __lc_ctype_data, __lc_ctype_data_LEN, "lc_ctype_data");
 	out_uc(lso, __lc_numeric_data, __lc_numeric_data_LEN, "lc_numeric_data");
 	out_uc(lso, __lc_monetary_data, __lc_monetary_data_LEN, "lc_monetary_data");
 	out_uc(lso, __lc_time_data, __lc_time_data_LEN, "lc_time_data");
@@ -209,6 +211,10 @@ int main(void)
 	fprintf(lso, "},\n");
 #endif /* __CTYPE_HAS_8_BIT_LOCALES */
 
+	out_uc(lso, __lc_ctype_rows, __lc_ctype_rows_LEN, "lc_ctype_rows");
+	out_u16(lso, __lc_ctype_item_offsets, __lc_ctype_item_offsets_LEN, "lc_ctype_item_offsets");
+	out_u16(lso, __lc_ctype_item_idx, __lc_ctype_item_idx_LEN, "lc_ctype_item_idx");
+
 	out_uc(lso, __lc_numeric_rows, __lc_numeric_rows_LEN, "lc_numeric_rows");
 	out_u16(lso, __lc_numeric_item_offsets, __lc_numeric_item_offsets_LEN, "lc_numeric_item_offsets");
 	out_u16(lso, __lc_numeric_item_idx, __lc_numeric_item_idx_LEN, "lc_numeric_item_idx");
@@ -221,16 +227,23 @@ int main(void)
 	out_u16(lso, __lc_time_item_offsets, __lc_time_item_offsets_LEN, "lc_time_item_offsets");
 	out_u16(lso, __lc_time_item_idx, __lc_time_item_idx_LEN, "lc_time_item_idx");
 
-	/* TODO -- collate*/
-
 	out_uc(lso, __lc_messages_rows, __lc_messages_rows_LEN, "lc_messages_rows");
 	out_u16(lso, __lc_messages_item_offsets, __lc_messages_item_offsets_LEN, "lc_messages_item_offsets");
 	out_u16(lso, __lc_messages_item_idx, __lc_messages_item_idx_LEN, "lc_messages_item_idx");
 
+	/* collate should be last*/
+	assert(sizeof(__locale_collate_tbl)/sizeof(__locale_collate_tbl[0]) == __lc_collate_data_LEN) ;
+	out_u16(lso, __locale_collate_tbl, __lc_collate_data_LEN, "collate_data");
+	
+
 	{
 		unsigned char co_buf[CATEGORIES] = {
-			1, __lc_numeric_item_offsets_LEN, __lc_monetary_item_offsets_LEN,
-			__lc_time_item_offsets_LEN, 0, __lc_messages_item_offsets_LEN
+			__lc_ctype_item_offsets_LEN,
+			__lc_numeric_item_offsets_LEN,
+			__lc_monetary_item_offsets_LEN,
+			__lc_time_item_offsets_LEN,
+			0,
+			__lc_messages_item_offsets_LEN
 		};
 		out_uc(lso, co_buf, CATEGORIES, "lc_common_item_offsets_LEN");
 	}
@@ -267,190 +280,8 @@ int main(void)
 	return EXIT_SUCCESS;
 }
 
-
-
 /* ---------------------------------------------------------------------- */
 
-#if 0
-#define WRITE_COMMON_MMAP(X) \
-	fseek(fp, (long) offsetof(__locale_mmap_t, __PASTE3(lc_,X,_data)), SEEK_SET); \
-	for (i=0 ; i < __PASTE3(__lc_,X,_data_LEN) ; i++) { \
-		putc(__PASTE3(__lc_,X,_data)[i], fp); \
-	}
-
-#define WRITE_COMMON_MMIDX(X) \
-	fseek(fp, (long) offsetof(__locale_mmap_t, __PASTE3(lc_,X,_rows)), SEEK_SET); \
-	for (i=0 ; i < __PASTE3(__lc_,X,_rows_LEN) ; i++) { \
-		putc(__PASTE3(__lc_,X,_rows)[i], fp); \
-	} \
-	fseek(fp, (long) offsetof(__locale_mmap_t, __PASTE3(lc_,X,_item_offsets)), SEEK_SET); \
-	for (i=0 ; i < __PASTE3(__lc_,X,_item_offsets_LEN) ; i++) { \
-		putc( ((unsigned char *) &(__PASTE3(__lc_,X,_item_offsets)[i]))[0], fp); \
-		putc( ((unsigned char *) &(__PASTE3(__lc_,X,_item_offsets)[i]))[1], fp); \
-	} \
-	fseek(fp, (long) offsetof(__locale_mmap_t, __PASTE3(lc_,X,_item_idx)), SEEK_SET); \
-	for (i=0 ; i < __PASTE3(__lc_,X,_item_idx_LEN) ; i++) { \
-		putc( ((unsigned char *) &(__PASTE3(__lc_,X,_item_idx)[i]))[0], fp); \
-		putc( ((unsigned char *) &(__PASTE3(__lc_,X,_item_idx)[i]))[1], fp); \
-	}
-
-#define WRITE_WC_DATA(X) \
-	fseek(fp, (long) offsetof(__locale_mmap_t, __PASTE2(tblw,X)), SEEK_SET); \
-	for (i=0 ; i < __PASTE3(WC,X,_TBL_LEN) ; i++) { \
-		putc(__PASTE3(WC,X,_data)[i], fp); \
-	}
-
-#define WRITE_WC_I16_DATA(X) \
-	fseek(fp, (long) offsetof(__locale_mmap_t, __PASTE2(tblw,X)), SEEK_SET); \
-	for (i=0 ; i < __PASTE3(WC,X,_TBL_LEN) ; i++) { \
-		putc( ((unsigned char *) &(__PASTE3(WC,X,_data)[i]))[0], fp); \
-		putc( ((unsigned char *) &(__PASTE3(WC,X,_data)[i]))[1], fp); \
-	}
-
-#define WRITE_C_DATA(X) \
-	fseek(fp, (long) offsetof(__locale_mmap_t, __PASTE2(tbl8,X)), SEEK_SET); \
-	for (i=0 ; i < __PASTE3(C,X,_TBL_LEN) ; i++) { \
-		putc(__PASTE3(C,X,_data)[i], fp); \
-	}
-
-#define WRITE_C_U16_DATA(X) \
-	fseek(fp, (long) offsetof(__locale_mmap_t, __PASTE2(tbl8,X)), SEEK_SET); \
-	for (i=0 ; i < __PASTE3(C,X,_TBL_LEN) ; i++) { \
-		putc( ((unsigned char *) &(__PASTE3(C,X,_data)[i]))[0], fp); \
-		putc( ((unsigned char *) &(__PASTE3(C,X,_data)[i]))[1], fp); \
-	}
-
-/**********************************************************************/
-
-#define COMMON_OFFSETS(X) \
-	offsetof(__locale_mmap_t, __PASTE3(lc_,X,_rows)), \
-	offsetof(__locale_mmap_t, __PASTE3(lc_,X,_item_offsets)), \
-	offsetof(__locale_mmap_t, __PASTE3(lc_,X,_item_idx)), \
-	offsetof(__locale_mmap_t, __PASTE3(lc_,X,_data)) \
-
-
-static const size_t common_tbl_offsets[CATEGORIES*4] = {
-	0, 0, 0, 0,					/* ctype */
-	COMMON_OFFSETS(numeric),
-	COMMON_OFFSETS(monetary),
-	COMMON_OFFSETS(time),
-	0, 0, 0, 0,					/* collate */
-	COMMON_OFFSETS(messages),
-};
-
-
-
-
-
-int main(void)
-{
-	FILE *fp;
-	size_t i;
-	unsigned char *p;
-
-	if (!(fp = fopen("locale.mmap", "w"))) {
-		printf("error - can't open locale.mmap for writing!");
-		return EXIT_FAILURE;
-	}
-
-	for (i=0 ; i < sizeof(__locale_mmap_t) ; i++) {
-		putc(0, fp);			/* Zero out the file. */
-	}
-
-#ifdef __CTYPE_HAS_8_BIT_LOCALES
-	WRITE_C_DATA(ctype);
-	WRITE_C_DATA(uplow);
-#ifdef __WCHAR_ENABLED
-	WRITE_C_U16_DATA(c2wc);
-	WRITE_C_DATA(wc2c);
-	/* translit  */
-#endif /* __WCHAR_ENABLED */
-#endif /* __CTYPE_HAS_8_BIT_LOCALES */
-
-#ifdef __WCHAR_ENABLED
-	WRITE_WC_DATA(ctype);
-	WRITE_WC_DATA(uplow);
-	WRITE_WC_I16_DATA(uplow_diff);
-/* 	WRITE_WC_DATA(comb); */
-	/* width?? */
-#endif /* __WCHAR_ENABLED */
-
-	WRITE_COMMON_MMAP(numeric);
-	WRITE_COMMON_MMAP(monetary);
-	WRITE_COMMON_MMAP(time);
-	/* TODO -- collate*/
-	WRITE_COMMON_MMAP(messages);
-
-#ifdef __CTYPE_HAS_8_BIT_LOCALES
-	fseek(fp, (long) offsetof(__locale_mmap_t, codeset_8_bit), SEEK_SET); \
-	p = (unsigned char *) codeset_8_bit;
-	for (i=0 ; i < sizeof(codeset_8_bit) ; i++) {
-		putc(p[i], fp);
-	}
-#endif /* __CTYPE_HAS_8_BIT_LOCALES */
-
-	WRITE_COMMON_MMIDX(numeric);
-	WRITE_COMMON_MMIDX(monetary);
-	WRITE_COMMON_MMIDX(time);
-	/* TODO -- collate*/
-	WRITE_COMMON_MMIDX(messages);
-
-	fseek(fp, (long) offsetof(__locale_mmap_t, lc_common_item_offsets_LEN), SEEK_SET);
-	putc(1, fp);				/* ctype -- (codeset) handled specially */
-	putc(__lc_numeric_item_offsets_LEN, fp);
-	putc(__lc_monetary_item_offsets_LEN, fp);
-	putc(__lc_time_item_offsets_LEN, fp);
-	putc(0, fp);				/* collate */
-	putc(__lc_messages_item_offsets_LEN, fp);
-
-	fseek(fp, (long) offsetof(__locale_mmap_t, lc_common_tbl_offsets), SEEK_SET);
-	for (i=0 ; i < sizeof(common_tbl_offsets) ; i++) {
-		putc(((unsigned char *)common_tbl_offsets)[i], fp);
-	}
-
-#ifdef NUM_LOCALES
-	fseek(fp, (long) offsetof(__locale_mmap_t, locales), SEEK_SET);
-	for (i=0 ; i < (NUM_LOCALES * WIDTH_LOCALES) ; i++) {
-		putc(__locales[i], fp);
-	}
-
-	fseek(fp, (long) offsetof(__locale_mmap_t, locale_names5), SEEK_SET);
-	for (i=0 ; i < 5 * NUM_LOCALE_NAMES ; i++) {
-		putc(__locale_names5[i], fp);
-	}
-
-#ifdef LOCALE_AT_MODIFIERS_LENGTH
-	fseek(fp, (long) offsetof(__locale_mmap_t, locale_at_modifiers), SEEK_SET);
-	for (i=0 ; i < LOCALE_AT_MODIFIERS_LENGTH ; i++) {
-		putc(__locale_at_modifiers[i], fp);
-	}
-#endif /* LOCALE_AT_MODIFIERS_LENGTH */
-#endif /* NUM_LOCALES */
-
-	fseek(fp, (long) offsetof(__locale_mmap_t, lc_names), SEEK_SET);
-	for (i=0 ; i < lc_names_LEN ; i++) {
-		putc(lc_names[i], fp);
-	}
-
-#ifdef __CTYPE_HAS_8_BIT_LOCALES
-	fseek(fp, (long) offsetof(__locale_mmap_t, codeset_list), SEEK_SET);
-	for (i=0 ; i < sizeof(CODESET_LIST) ; i++) {
-		putc((unsigned char)(CODESET_LIST[i]), fp);
-	}
-#endif /* __CTYPE_HAS_8_BIT_LOCALES */
-
-
-	if (ferror(fp) || (fclose(fp) == EOF)) {
-		printf("error - stream in error state or fclose failed!");
-		return EXIT_FAILURE;
-	}
-
-	printf("sizeof(__locale_mmap_t) = %zd\n", sizeof(__locale_mmap_t));
-
-	return EXIT_SUCCESS;
-}
-#endif
-
 /* TODO:
  * collate data (8-bit weighted single char only)
  * @ mappings!