summaryrefslogtreecommitdiff
path: root/extra/locale/gen_locale.c
diff options
context:
space:
mode:
Diffstat (limited to 'extra/locale/gen_locale.c')
-rw-r--r--extra/locale/gen_locale.c1230
1 files changed, 605 insertions, 625 deletions
diff --git a/extra/locale/gen_locale.c b/extra/locale/gen_locale.c
index 268c3d61b..1d50480ee 100644
--- a/extra/locale/gen_locale.c
+++ b/extra/locale/gen_locale.c
@@ -47,6 +47,7 @@ typedef struct {
unsigned char lc_numeric_row;
unsigned char lc_monetary_row;
unsigned char lc_messages_row;
+ unsigned char lc_ctype_row;
#if CATEGORIES != 6
#error unsupported CATEGORIES value
#endif
@@ -68,6 +69,7 @@ static void do_lc_numeric(void);
static void do_lc_monetary(void);
static void do_lc_messages(void);
+static void do_lc_ctype(void);
static FILE *fp;
@@ -119,11 +121,23 @@ static void do_locale_names(void)
fprintf(ofp, "\t%#4x, ", (int)((unsigned char) locales[i].dot_cs));
fprintf(ofp, "\t%#4x, ", (int)((unsigned char) locales[i].cs));
/* lc_ctype would store translit flags and turkish up/low flag. */
- fprintf(ofp, "%#4x, ", 0); /* place holder for lc_ctype */
+ fprintf(ofp, "%#4x, ", (int)((unsigned char) locales[i].lc_ctype_row));
fprintf(ofp, "%#4x, ", (int)((unsigned char) locales[i].lc_numeric_row));
fprintf(ofp, "%#4x, ", (int)((unsigned char) locales[i].lc_monetary_row));
fprintf(ofp, "%#4x, ", (int)((unsigned char) locales[i].lc_time_row));
+#if 1
+ /* lc_collate */
+ if (strlen(locales[i].glibc_name) >= 5) {
+ fprintf(ofp, "COL_IDX_%.2s_%.2s, ", locales[i].glibc_name, locales[i].glibc_name+3);
+ } else if (!strcmp(locales[i].glibc_name, "C")) {
+ fprintf(ofp, "COL_IDX_C , ");
+ } else {
+ printf("don't know how to handle COL_IDX_ for %s\n", locales[i].glibc_name);
+ exit(EXIT_FAILURE);
+ }
+#else
fprintf(ofp, "%#4x, ", 0); /* place holder for lc_collate */
+#endif
fprintf(ofp, "%#4x, ", (int)((unsigned char) locales[i].lc_messages_row));
fprintf(ofp, "\t/* %s */\n", locales[i].glibc_name);
}
@@ -213,6 +227,332 @@ static void do_locale_names(void)
}
+static void read_at_mappings(void)
+{
+ char *p;
+ char *m;
+ int mc = 0;
+
+ do {
+ if (!(p = strtok(line_buf, " \t\n")) || (*p == '#')) {
+ if (!fgets(line_buf, sizeof(line_buf), fp)) {
+ if (ferror(fp)) {
+ printf("error reading file\n");
+ exit(EXIT_FAILURE);
+ }
+ return; /* EOF */
+ }
+ if ((*line_buf == '#') && (line_buf[1] == '-')) {
+ break;
+ }
+ continue;
+ }
+ if (*p == '@') {
+ if (p[1] == 0) {
+ printf("error: missing @modifier name\n");
+ exit(EXIT_FAILURE);
+ }
+ m = p; /* save the modifier name */
+ if (!(p = strtok(NULL, " \t\n")) || p[1] || (((unsigned char) *p) > 0x7f)) {
+ printf("error: missing or illegal @modifier mapping char\n");
+ exit(EXIT_FAILURE);
+ }
+ if (at_mappings[(int)((unsigned char) *p)]) {
+ printf("error: reused @modifier mapping char\n");
+ exit(EXIT_FAILURE);
+ }
+ at_mappings[(int)((unsigned char) *p)] = 1;
+ at_mapto[mc] = *p;
+ ++mc;
+ *at_strings_end = (char)( (unsigned char) (strlen(m)) );
+ strcpy(++at_strings_end, m+1);
+ at_strings_end += (unsigned char) at_strings_end[-1];
+
+ printf("@mapping: \"%s\" to '%c'\n", m, *p);
+
+ if (((p = strtok(NULL, " \t\n")) != NULL) && (*p != '#')) {
+ printf("ignoring trailing text: %s...\n", p);
+ }
+ *line_buf = 0;
+ continue;
+ }
+ break;
+ } while (1);
+
+#if 0
+ {
+ p = at_strings;
+
+ if (!*p) {
+ printf("no @ strings\n");
+ return;
+ }
+
+ do {
+ printf("%s\n", p+1);
+ p += 1 + (unsigned char) *p;
+ } while (*p);
+ }
+#endif
+}
+
+static void read_enable_disable(void)
+{
+ char *p;
+
+ do {
+ if (!(p = strtok(line_buf, " =\t\n")) || (*p == '#')) {
+ if (!fgets(line_buf, sizeof(line_buf), fp)) {
+ if (ferror(fp)) {
+ printf("error reading file\n");
+ exit(EXIT_FAILURE);
+ }
+ return; /* EOF */
+ }
+ if ((*line_buf == '#') && (line_buf[1] == '-')) {
+ break;
+ }
+ continue;
+ }
+ if (!strcmp(p, "UTF-8")) {
+ if (!(p = strtok(NULL, " =\t\n"))
+ || ((toupper(*p) != 'Y') && (toupper(*p) != 'N'))) {
+ printf("error: missing or illegal UTF-8 setting\n");
+ exit(EXIT_FAILURE);
+ }
+ default_utf8 = (toupper(*p) == 'Y');
+ printf("UTF-8 locales are %sabled\n", "dis\0en"+ (default_utf8 << 2));
+ } else if (!strcmp(p, "8-BIT")) {
+ if (!(p = strtok(NULL, " =\t\n"))
+ || ((toupper(*p) != 'Y') && (toupper(*p) != 'N'))) {
+ printf("error: missing or illegal 8-BIT setting\n");
+ exit(EXIT_FAILURE);
+ }
+ default_8bit = (toupper(*p) == 'Y');
+ printf("8-BIT locales are %sabled\n", "dis\0en" + (default_8bit << 2));
+ } else {
+ break;
+ }
+
+ if (((p = strtok(NULL, " \t\n")) != NULL) && (*p != '#')) {
+ printf("ignoring trailing text: %s...\n", p);
+ }
+ *line_buf = 0;
+ continue;
+
+ } while (1);
+}
+
+#ifdef CODESET_LIST
+
+static int find_codeset_num(const char *cs)
+{
+ int r = 2;
+ char *s = CODESET_LIST;
+
+ /* 7-bit is 1, UTF-8 is 2, 8-bits are > 2 */
+
+ if (strcmp(cs, "UTF-8") != 0) {
+ ++r;
+ while (*s && strcmp(CODESET_LIST+ ((unsigned char) *s), cs)) {
+/* printf("tried %s\n", CODESET_LIST + ((unsigned char) *s)); */
+ ++r;
+ ++s;
+ }
+ if (!*s) {
+ printf("error: unsupported codeset %s\n", cs);
+ exit(EXIT_FAILURE);
+ }
+ }
+ return r;
+}
+
+#else
+
+static int find_codeset_num(const char *cs)
+{
+ int r = 2;
+
+ /* 7-bit is 1, UTF-8 is 2, 8-bits are > 2 */
+
+ if (strcmp(cs, "UTF-8") != 0) {
+ printf("error: unsupported codeset %s\n", cs);
+ exit(EXIT_FAILURE);
+ }
+ return r;
+}
+
+#endif
+
+static int find_at_string_num(const char *as)
+{
+ int i = 0;
+ char *p = at_strings;
+
+ while (*p) {
+ if (!strcmp(p+1, as)) {
+ return i;
+ }
+ ++i;
+ p += 1 + (unsigned char) *p;
+ }
+
+ printf("error: unmapped @string %s\n", as);
+ exit(EXIT_FAILURE);
+}
+
+static void read_locale_list(void)
+{
+ char *p;
+ char *s;
+ char *ln; /* locale name */
+ char *ls; /* locale name ll_CC */
+ char *as; /* at string */
+ char *ds; /* dot string */
+ char *cs; /* codeset */
+ int i;
+
+ typedef struct {
+ char *glibc_name;
+ char name[5];
+ char dot_cs; /* 0 if no codeset specified */
+ char cs;
+ } locale_entry;
+
+ /* First the C locale. */
+ locales[0].glibc_name = locales[0].name;
+ strncpy(locales[0].name,"C",5);
+ locales[0].dot_cs = 0;
+ locales[0].cs = 1; /* 7-bit encoding */
+ ++num_locales;
+
+ do {
+ if (!(p = strtok(line_buf, " \t\n")) || (*p == '#')) {
+ if (!fgets(line_buf, sizeof(line_buf), fp)) {
+ if (ferror(fp)) {
+ printf("error reading file\n");
+ exit(EXIT_FAILURE);
+ }
+ return; /* EOF */
+ }
+ if ((*line_buf == '#') && (line_buf[1] == '-')) {
+ break;
+ }
+ continue;
+ }
+
+ s = glibc_locale_names;
+ for (i=0 ; i < num_locales ; i++) {
+ if (!strcmp(s+1, p)) {
+ break;
+ }
+ s += 1 + ((unsigned char) *s);
+ }
+ if (i < num_locales) {
+ printf("ignoring dulplicate locale name: %s", p);
+ *line_buf = 0;
+ continue;
+ }
+
+ /* New locale, but don't increment num until codeset verified! */
+ *s = (char)((unsigned char) (strlen(p) + 1));
+ strcpy(s+1, p);
+ locales[num_locales].glibc_name = s+1;
+ ln = p; /* save locale name */
+
+ if (!(p = strtok(NULL, " \t\n"))) {
+ printf("error: missing codeset for locale %s\n", ln);
+ exit(EXIT_FAILURE);
+ }
+ cs = p;
+ i = find_codeset_num(p);
+ if ((i == 2) && !default_utf8) {
+ printf("ignoring UTF-8 locale %s\n", ln);
+ *line_buf = 0;
+ continue;
+ } else if ((i > 2) && !default_8bit) {
+ printf("ignoring 8-bit codeset locale %s\n", ln);
+ *line_buf = 0;
+ continue;
+ }
+ locales[num_locales].cs = (char)((unsigned char) i);
+
+ if (((p = strtok(NULL, " \t\n")) != NULL) && (*p != '#')) {
+ printf("ignoring trailing text: %s...\n", p);
+ }
+
+ /* Now go back to locale string for .codeset and @modifier */
+ as = strtok(ln, "@");
+ if (as) {
+ as = strtok(NULL, "@");
+ }
+ ds = strtok(ln, ".");
+ if (ds) {
+ ds = strtok(NULL, ".");
+ }
+ ls = ln;
+
+ if ((strlen(ls) != 5) || (ls[2] != '_')) {
+ printf("error: illegal locale name %s\n", ls);
+ exit(EXIT_FAILURE);
+ }
+
+ i = 0; /* value for unspecified codeset */
+ if (ds) {
+ i = find_codeset_num(ds);
+ if ((i == 2) && !default_utf8) {
+ printf("ignoring UTF-8 locale %s\n", ln);
+ *line_buf = 0;
+ continue;
+ } else if ((i > 2) && !default_8bit) {
+ printf("ignoring 8-bit codeset locale %s\n", ln);
+ *line_buf = 0;
+ continue;
+ }
+ }
+ locales[num_locales].dot_cs = (char)((unsigned char) i);
+
+ if (as) {
+ i = find_at_string_num(as);
+ ls[2] = at_mapto[i];
+ }
+ memcpy(locales[num_locales].name, ls, 5);
+/* printf("locale: %5.5s %2d %2d %s\n", */
+/* locales[num_locales].name, */
+/* locales[num_locales].cs, */
+/* locales[num_locales].dot_cs, */
+/* locales[num_locales].glibc_name */
+/* ); */
+ ++num_locales;
+ *line_buf = 0;
+ } while (1);
+}
+
+static int le_cmp(const void *a, const void *b)
+{
+ const locale_entry *p;
+ const locale_entry *q;
+ int r;
+
+ p = (const locale_entry *) a;
+ q = (const locale_entry *) b;
+
+ if (!(r = p->name[0] - q->name[0])
+ && !(r = p->name[1] - q->name[1])
+ && !(r = p->name[3] - q->name[3])
+ && !(r = p->name[4] - q->name[4])
+ && !(r = p->name[2] - q->name[2])
+ && !(r = -(p->cs - q->cs))
+ ) {
+ r = -(p->dot_cs - q->dot_cs);
+ /* Reverse the ordering of the codesets so UTF-8 comes last.
+ * Work-around (hopefully) for glibc bug affecting at least
+ * the euro currency symbol. */
+ }
+
+ return r;
+}
+
int main(int argc, char **argv)
{
if ((argc != 2) || (!(fp = fopen(*++argv, "r")))) {
@@ -251,6 +591,7 @@ int main(int argc, char **argv)
do_lc_numeric();
do_lc_monetary();
do_lc_messages();
+ do_lc_ctype();
do_locale_names();
@@ -267,6 +608,36 @@ static char buf[100000];
static char *last;
static int uniq;
+static int addblock(const char *s, size_t n) /* l includes nul terminator */
+{
+ int j;
+
+ if (!s) {
+ ++null_count;
+ return 0;
+ }
+
+ for (j=0 ; (j < uniq) && (idx[j] + n < last) ; j++) {
+ if (!memcmp(s, idx[j], n)) {
+ return idx[j] - buf;
+ }
+ }
+ if (uniq >= sizeof(idx)) {
+ printf("too many uniq strings!\n");
+ exit(EXIT_FAILURE);
+ }
+ if (last + n >= buf + sizeof(buf)) {
+ printf("need to increase size of buf!\n");
+ exit(EXIT_FAILURE);
+ }
+
+ idx[uniq] = last;
+ ++uniq;
+ memcpy(last, s, n);
+ last += n;
+ return idx[uniq - 1] - buf;
+}
+
static int addstring(const char *s)
{
int j;
@@ -299,20 +670,178 @@ static int addstring(const char *s)
return idx[uniq - 1] - buf;
}
-static int buf16[50*256];
+#define DO_LC_COMMON(CATEGORY) \
+ printf("buf-size=%d uniq=%d rows=%d\n", \
+ (int)(last - buf), uniq, lc_##CATEGORY##_uniq); \
+ printf("total = %d + %d * %d + %d = %d\n", \
+ num_locales, lc_##CATEGORY##_uniq, NUM_NL_##CATEGORY, (int)(last - buf), \
+ i = num_locales + lc_##CATEGORY##_uniq*NUM_NL_##CATEGORY + (int)(last - buf)); \
+ total_size += i; \
+ dump_table8c("__lc_" #CATEGORY "_data", buf, (int)(last - buf)); \
+ for (i=0 ; i < lc_##CATEGORY##_uniq ; i++) { \
+ m = locales[i].lc_##CATEGORY##_row; \
+ for (k=0 ; k < NUM_NL_##CATEGORY ; k++) { \
+ buf[NUM_NL_##CATEGORY*i + k] = (char)((unsigned char) lc_##CATEGORY##_uniq_X[i][k]); \
+ } \
+ } \
+ dump_table8("__lc_" #CATEGORY "_rows", buf, lc_##CATEGORY##_uniq * NUM_NL_##CATEGORY); \
+ buf16[0] =0; \
+ for (i=0 ; i < NUM_NL_##CATEGORY - 1 ; i++) { \
+ buf16[i+1] = buf16[i] + lc_##CATEGORY##_count[i]; \
+ } \
+ dump_table16("__lc_" #CATEGORY "_item_offsets", buf16, NUM_NL_##CATEGORY); \
+ m = 0; \
+ for (k=0 ; k < NUM_NL_##CATEGORY ; k++) { \
+ for (i=0 ; i < lc_##CATEGORY##_count[k] ; i++) { \
+ buf16[m] = lc_##CATEGORY##_item[k][i]; \
+ ++m; \
+ } \
+ } \
+ dump_table16("__lc_" #CATEGORY "_item_idx", buf16, m);
+
+
+#define DL_LC_LOOPTAIL(CATEGORY) \
+ if (k > NUM_NL_##CATEGORY) { \
+ printf("error -- lc_" #CATEGORY " nl_item count > %d!\n", NUM_NL_##CATEGORY); \
+ exit(EXIT_FAILURE); \
+ } \
+ { \
+ int r; \
+ for (r=0 ; r < lc_##CATEGORY##_uniq ; r++) { \
+ if (!memcmp(lc_##CATEGORY##_uniq_X[lc_##CATEGORY##_uniq], \
+ lc_##CATEGORY##_uniq_X[r], NUM_NL_##CATEGORY)) { \
+ break; \
+ } \
+ } \
+ if (r == lc_##CATEGORY##_uniq) { /* new locale row */ \
+ ++lc_##CATEGORY##_uniq; \
+ if (lc_##CATEGORY##_uniq > 255) { \
+ printf("too many unique lc_" #CATEGORY " rows!\n"); \
+ exit(EXIT_FAILURE); \
+ } \
+ } \
+ locales[i].lc_##CATEGORY##_row = r; \
+ }
+
+
+
+static int buf16[100*256];
+
+static void dump_table8(const char *name, const char *tbl, int len)
+{
+ int i;
+
+ fprintf(ofp, "#define %s_LEN\t\t%d\n", name, len);
+ fprintf(ofp, "static const unsigned char %s[%d] = {", name, len);
+ for (i=0 ; i < len ; i++) {
+ if ((i % 12) == 0) {
+ fprintf(ofp, "\n\t");
+ }
+ fprintf(ofp, "%#4x, ", (int)((unsigned char) tbl[i]));
+ }
+ fprintf(ofp, "\n};\n\n");
+}
+
+#define __C_isdigit(c) \
+ ((sizeof(c) == sizeof(char)) \
+ ? (((unsigned char)((c) - '0')) < 10) \
+ : (((unsigned int)((c) - '0')) < 10))
+#define __C_isalpha(c) \
+ ((sizeof(c) == sizeof(char)) \
+ ? (((unsigned char)(((c) | 0x20) - 'a')) < 26) \
+ : (((unsigned int)(((c) | 0x20) - 'a')) < 26))
+#define __C_isalnum(c) (__C_isalpha(c) || __C_isdigit(c))
+
+static void dump_table8c(const char *name, const char *tbl, int len)
+{
+ int i;
+
+ fprintf(ofp, "#define %s_LEN\t\t%d\n", name, len);
+ fprintf(ofp, "static const unsigned char %s[%d] = {", name, len);
+ for (i=0 ; i < len ; i++) {
+ if ((i % 12) == 0) {
+ fprintf(ofp, "\n\t");
+ }
+ if (__C_isalnum(tbl[i]) || (tbl[i] == ' ')) {
+ fprintf(ofp, " '%c', ", (int)((unsigned char) tbl[i]));
+ } else {
+ fprintf(ofp, "%#4x, ", (int)((unsigned char) tbl[i]));
+ }
+ }
+ fprintf(ofp, "\n};\n\n");
+}
+
+static void dump_table16(const char *name, const int *tbl, int len)
+{
+ int i;
+
+ fprintf(ofp, "#define %s_LEN\t\t%d\n", name, len);
+ fprintf(ofp, "static const uint16_t %s[%d] = {", name, len);
+ for (i=0 ; i < len ; i++) {
+ if ((i % 8) == 0) {
+ fprintf(ofp, "\n\t");
+ }
+ if (tbl[i] != (uint16_t) tbl[i]) {
+ printf("error - falls outside uint16 range!\n");
+ exit(EXIT_FAILURE);
+ }
+ fprintf(ofp, "%#6x, ", tbl[i]);
+ }
+ fprintf(ofp, "\n};\n\n");
+}
+
+#define NUM_NL_time 50
-static int lc_time_item[50][256];
-static int lc_time_count[50];
-static unsigned char lc_time_uniq_50[700][50];
+static int lc_time_item[NUM_NL_time][256];
+static int lc_time_count[NUM_NL_time];
+static unsigned char lc_time_uniq_X[700][NUM_NL_time];
static int lc_time_uniq;
#define DO_NL_S(X) lc_time_S(X, k++)
static void lc_time_S(int X, int k)
{
+ size_t len;
int j, m;
- j = addstring(nl_langinfo(X));
+ const char *s = nl_langinfo(X);
+ const char *p;
+ static const char nulbuf[] = "";
+
+ if (X == ALT_DIGITS) {
+ len = 1;
+ if (!s) {
+ s = nulbuf;
+ }
+ if (*s) {
+ p = s;
+ for (j = 0 ; j < 100 ; j++) {
+ while (*p) {
+ ++p;
+ }
+ ++p;
+ }
+ len = p - s;
+ }
+ j = addblock(s, len);
+/* if (len > 1) fprintf(stderr, "alt_digit: called addblock with len %zd\n", len); */
+ } else if (X == ERA) {
+ if (!s) {
+ s = nulbuf;
+ }
+ p = s;
+ while (*p) {
+ while (*p) {
+ ++p;
+ }
+ ++p;
+ }
+ ++p;
+ j = addblock(s, p - s);
+/* if (p-s > 1) fprintf(stderr, "era: called addblock with len %d\n", p-s); */
+ } else {
+ j = addstring(s);
+ }
for (m=0 ; m < lc_time_count[k] ; m++) {
if (lc_time_item[k][m] == j) {
break;
@@ -326,8 +855,7 @@ static void lc_time_S(int X, int k)
lc_time_item[k][m] = j;
++lc_time_count[k];
}
-/* printf("\\x%02x", m); */
- lc_time_uniq_50[lc_time_uniq][k] = m;
+ lc_time_uniq_X[lc_time_uniq][k] = m;
}
static void do_lc_time(void)
@@ -339,11 +867,9 @@ static void do_lc_time(void)
*buf = 0;
*idx = buf;
-/* printf("processing lc_time..."); */
for (i=0 ; i < num_locales ; i++) {
k = 0;
-/* printf(" %d", i); fflush(stdout); */
if (!setlocale(LC_ALL, locales[i].glibc_name)) {
printf("setlocale(LC_ALL,%s) failed!\n",
locales[i].glibc_name);
@@ -406,143 +932,19 @@ static void do_lc_time(void)
DO_NL_S(ERA_D_T_FMT);
DO_NL_S(ERA_T_FMT);
- if (k > 50) {
- printf("error -- lc_time nl_item count > 50!\n");
- exit(EXIT_FAILURE);
- }
-
- {
- int r;
- for (r=0 ; r < lc_time_uniq ; r++) {
- if (!memcmp(lc_time_uniq_50[lc_time_uniq],
- lc_time_uniq_50[r], 50)) {
- break;
- }
- }
- if (r == lc_time_uniq) { /* new locale row */
- ++lc_time_uniq;
- if (lc_time_uniq > 255) {
- printf("too many unique lc_time rows!\n");
- exit(EXIT_FAILURE);
- }
- }
- locales[i].lc_time_row = r;
- }
-/* printf("\n"); */
- }
-/* printf(" done\n"); */
-
- m = k = 0;
- for (i=0 ; i < 50 ; i++) {
- k += lc_time_count[i];
- if (m < lc_time_count[i]) {
- m = lc_time_count[i];
- }
- }
- printf("buf-size=%d uniq=%d item_offsets=%d max=%d rows=%d\n",
- (int)(last - buf), uniq, k, m, lc_time_uniq);
-/* printf("total = %d * 50 + 2 * (50 + %d) + %d = %d\n", */
-/* num_locales, k, (int)(last - buf), */
-/* num_locales*50 + 2*(50 + k) + (int)(last - buf)); */
- printf("total = %d + %d * 50 + 2 * (50 + %d) + %d = %d\n",
- num_locales, lc_time_uniq, k, (int)(last - buf),
- i = num_locales + lc_time_uniq*50 + 2*(50 + k) + (int)(last - buf));
- total_size += i;
-
- dump_table8c("__lc_time_data", buf, (int)(last - buf));
-
- for (i=0 ; i < lc_time_uniq ; i++) {
- m = locales[i].lc_time_row;
- for (k=0 ; k < 50 ; k++) {
- buf[50*i + k] = (char)((unsigned char) lc_time_uniq_50[i][k]);
- }
- }
- dump_table8("__lc_time_rows", buf, lc_time_uniq * 50);
-
- buf16[0] =0;
- for (i=0 ; i < 50 - 1 ; i++) {
- buf16[i+1] = buf16[i] + lc_time_count[i];
- }
- dump_table16("__lc_time_item_offsets", buf16, 50);
-
- m = 0;
- for (k=0 ; k < 50 ; k++) {
- for (i=0 ; i < lc_time_count[k] ; i++) {
- buf16[m] = lc_time_item[k][i];
- ++m;
- }
- }
- dump_table16("__lc_time_item_idx", buf16, m);
-}
-
-static void dump_table8(const char *name, const char *tbl, int len)
-{
- int i;
-
- fprintf(ofp, "#define %s_LEN\t\t%d\n", name, len);
- fprintf(ofp, "static const unsigned char %s[%d] = {", name, len);
- for (i=0 ; i < len ; i++) {
- if ((i % 12) == 0) {
- fprintf(ofp, "\n\t");
- }
- fprintf(ofp, "%#4x, ", (int)((unsigned char) tbl[i]));
+ DL_LC_LOOPTAIL(time)
}
- fprintf(ofp, "\n};\n\n");
-}
-
-#define __C_isdigit(c) \
- ((sizeof(c) == sizeof(char)) \
- ? (((unsigned char)((c) - '0')) < 10) \
- : (((unsigned int)((c) - '0')) < 10))
-#define __C_isalpha(c) \
- ((sizeof(c) == sizeof(char)) \
- ? (((unsigned char)(((c) | 0x20) - 'a')) < 26) \
- : (((unsigned int)(((c) | 0x20) - 'a')) < 26))
-#define __C_isalnum(c) (__C_isalpha(c) || __C_isdigit(c))
-
-static void dump_table8c(const char *name, const char *tbl, int len)
-{
- int i;
- fprintf(ofp, "#define %s_LEN\t\t%d\n", name, len);
- fprintf(ofp, "static const unsigned char %s[%d] = {", name, len);
- for (i=0 ; i < len ; i++) {
- if ((i % 12) == 0) {
- fprintf(ofp, "\n\t");
- }
- if (__C_isalnum(tbl[i]) || (tbl[i] == ' ')) {
- fprintf(ofp, " '%c', ", (int)((unsigned char) tbl[i]));
- } else {
- fprintf(ofp, "%#4x, ", (int)((unsigned char) tbl[i]));
- }
- }
- fprintf(ofp, "\n};\n\n");
-}
-
-static void dump_table16(const char *name, const int *tbl, int len)
-{
- int i;
-
- fprintf(ofp, "#define %s_LEN\t\t%d\n", name, len);
- fprintf(ofp, "static const uint16_t %s[%d] = {", name, len);
- for (i=0 ; i < len ; i++) {
- if ((i % 8) == 0) {
- fprintf(ofp, "\n\t");
- }
- if (tbl[i] != (uint16_t) tbl[i]) {
- printf("error - falls outside uint16 range!\n");
- exit(EXIT_FAILURE);
- }
- fprintf(ofp, "%#6x, ", tbl[i]);
- }
- fprintf(ofp, "\n};\n\n");
+ DO_LC_COMMON(time)
}
#undef DO_NL_S
-static int lc_numeric_item[3][256];
-static int lc_numeric_count[3];
-static unsigned char lc_numeric_uniq_3[700][3];
+#define NUM_NL_numeric 3
+
+static int lc_numeric_item[NUM_NL_numeric][256];
+static int lc_numeric_count[NUM_NL_numeric];
+static unsigned char lc_numeric_uniq_X[700][NUM_NL_numeric];
static int lc_numeric_uniq;
#define DO_NL_S(X) lc_numeric_S(X, k++)
@@ -606,7 +1008,7 @@ static void lc_numeric_S(int X, int k)
++lc_numeric_count[k];
}
/* printf("\\x%02x", m); */
- lc_numeric_uniq_3[lc_numeric_uniq][k] = m;
+ lc_numeric_uniq_X[lc_numeric_uniq][k] = m;
}
static void do_lc_numeric(void)
@@ -630,74 +1032,19 @@ static void do_lc_numeric(void)
DO_NL_S(THOUSEP); /* THOUSANDS_SEP */
DO_NL_S(GROUPING);
- if (k > 3) {
- printf("error -- lc_numeric nl_item count > 3!\n");
- exit(EXIT_FAILURE);
- }
-
- {
- int r;
- for (r=0 ; r < lc_numeric_uniq ; r++) {
- if (!memcmp(lc_numeric_uniq_3[lc_numeric_uniq],
- lc_numeric_uniq_3[r], 3)) {
- break;
- }
- }
- if (r == lc_numeric_uniq) { /* new locale row */
- ++lc_numeric_uniq;
- if (lc_numeric_uniq > 255) {
- printf("too many unique lc_numeric rows!\n");
- exit(EXIT_FAILURE);
- }
- }
- locales[i].lc_numeric_row = r;
- }
- }
-
- printf("buf-size=%d uniq=%d rows=%d\n",
- (int)(last - buf), uniq, lc_numeric_uniq);
- printf("total = %d + %d * 3 + %d = %d\n",
- num_locales, lc_numeric_uniq, (int)(last - buf),
- i = num_locales + lc_numeric_uniq*3 + (int)(last - buf));
- total_size += i;
-
-/* printf("buf-size=%d uniq=%d\n", (int)(last - buf), uniq); */
-
- dump_table8c("__lc_numeric_data", buf, (int)(last - buf));
-
-
- for (i=0 ; i < lc_numeric_uniq ; i++) {
- m = locales[i].lc_numeric_row;
- for (k=0 ; k < 3 ; k++) {
- buf[3*i + k] = (char)((unsigned char) lc_numeric_uniq_3[i][k]);
- }
- }
- dump_table8("__lc_numeric_rows", buf, lc_numeric_uniq * 3);
-
- buf16[0] =0;
- for (i=0 ; i < 3 - 1 ; i++) {
- buf16[i+1] = buf16[i] + lc_numeric_count[i];
+ DL_LC_LOOPTAIL(numeric)
}
- dump_table16("__lc_numeric_item_offsets", buf16, 3);
- m = 0;
- for (k=0 ; k < 3 ; k++) {
- for (i=0 ; i < lc_numeric_count[k] ; i++) {
- buf16[m] = lc_numeric_item[k][i];
- ++m;
- }
- }
- dump_table16("__lc_numeric_item_idx", buf16, m);
+ DO_LC_COMMON(numeric)
}
#undef DO_NL_S
-/* #define NUM_NL_MONETARY 7 */
-#define NUM_NL_MONETARY (7+14+1)
+#define NUM_NL_monetary (7+14+1)
-static int lc_monetary_item[NUM_NL_MONETARY][256];
-static int lc_monetary_count[NUM_NL_MONETARY];
-static unsigned char lc_monetary_uniq_X[700][NUM_NL_MONETARY];
+static int lc_monetary_item[NUM_NL_monetary][256];
+static int lc_monetary_count[NUM_NL_monetary];
+static unsigned char lc_monetary_uniq_X[700][NUM_NL_monetary];
static int lc_monetary_uniq;
#define DO_NL_S(X) lc_monetary_S(X, k++)
@@ -838,71 +1185,20 @@ static void do_lc_monetary(void)
DO_NL_S(CRNCYSTR); /* CURRENCY_SYMBOL */
-/* printf("\n"); */
-
- if (k > NUM_NL_MONETARY) {
- printf("error -- lc_monetary nl_item count > %d!\n", NUM_NL_MONETARY);
- exit(EXIT_FAILURE);
- }
-
- {
- int r;
- for (r=0 ; r < lc_monetary_uniq ; r++) {
- if (!memcmp(lc_monetary_uniq_X[lc_monetary_uniq],
- lc_monetary_uniq_X[r], NUM_NL_MONETARY)) {
- break;
- }
- }
- if (r == lc_monetary_uniq) { /* new locale row */
- ++lc_monetary_uniq;
- if (lc_monetary_uniq > 255) {
- printf("too many unique lc_monetary rows!\n");
- exit(EXIT_FAILURE);
- }
- }
- locales[i].lc_monetary_row = r;
- }
- }
-
- printf("buf-size=%d uniq=%d rows=%d\n",
- (int)(last - buf), uniq, lc_monetary_uniq);
- printf("total = %d + %d * %d + %d = %d\n",
- num_locales, lc_monetary_uniq, NUM_NL_MONETARY, (int)(last - buf),
- i = num_locales + lc_monetary_uniq*NUM_NL_MONETARY + (int)(last - buf));
- total_size += i;
-
- dump_table8c("__lc_monetary_data", buf, (int)(last - buf));
-
- for (i=0 ; i < lc_monetary_uniq ; i++) {
- m = locales[i].lc_monetary_row;
- for (k=0 ; k < NUM_NL_MONETARY ; k++) {
- buf[NUM_NL_MONETARY*i + k] = (char)((unsigned char) lc_monetary_uniq_X[i][k]);
- }
- }
- dump_table8("__lc_monetary_rows", buf, lc_monetary_uniq * NUM_NL_MONETARY);
-
- buf16[0] =0;
- for (i=0 ; i < NUM_NL_MONETARY - 1 ; i++) {
- buf16[i+1] = buf16[i] + lc_monetary_count[i];
+ DL_LC_LOOPTAIL(monetary)
}
- dump_table16("__lc_monetary_item_offsets", buf16, NUM_NL_MONETARY);
- m = 0;
- for (k=0 ; k < NUM_NL_MONETARY ; k++) {
- for (i=0 ; i < lc_monetary_count[k] ; i++) {
- buf16[m] = lc_monetary_item[k][i];
- ++m;
- }
- }
- dump_table16("__lc_monetary_item_idx", buf16, m);
+ DO_LC_COMMON(monetary)
}
#undef DO_NL_S
-static int lc_messages_item[2][256];
-static int lc_messages_count[2];
-static unsigned char lc_messages_uniq_2[700][2];
+#define NUM_NL_messages 4
+
+static int lc_messages_item[NUM_NL_messages][256];
+static int lc_messages_count[NUM_NL_messages];
+static unsigned char lc_messages_uniq_X[700][NUM_NL_messages];
static int lc_messages_uniq;
#define DO_NL_S(X) lc_messages_S(X, k++)
@@ -925,7 +1221,7 @@ static void lc_messages_S(int X, int k)
++lc_messages_count[k];
}
/* printf("\\x%02x", m); */
- lc_messages_uniq_2[lc_messages_uniq][k] = m;
+ lc_messages_uniq_X[lc_messages_uniq][k] = m;
}
static void do_lc_messages(void)
@@ -947,393 +1243,77 @@ static void do_lc_messages(void)
DO_NL_S(YESEXPR);
DO_NL_S(NOEXPR);
+ DO_NL_S(YESSTR);
+ DO_NL_S(NOSTR);
- if (k > 2) {
- printf("error -- lc_messages nl_item count > 2!\n");
- exit(EXIT_FAILURE);
- }
-
- {
- int r;
- for (r=0 ; r < lc_messages_uniq ; r++) {
- if (!memcmp(lc_messages_uniq_2[lc_messages_uniq],
- lc_messages_uniq_2[r], 2)) {
- break;
- }
- }
- if (r == lc_messages_uniq) { /* new locale row */
- ++lc_messages_uniq;
- if (lc_messages_uniq > 255) {
- printf("too many unique lc_messages rows!\n");
- exit(EXIT_FAILURE);
- }
- }
- locales[i].lc_messages_row = r;
- }
- }
-
- printf("buf-size=%d uniq=%d rows=%d\n",
- (int)(last - buf), uniq, lc_messages_uniq);
- printf("total = %d + %d * 2 + %d = %d\n",
- num_locales, lc_messages_uniq, (int)(last - buf),
- i = num_locales + lc_messages_uniq*2 + (int)(last - buf));
- total_size += i;
-
-/* printf("buf-size=%d uniq=%d\n", (int)(last - buf), uniq); */
-
- dump_table8c("__lc_messages_data", buf, (int)(last - buf));
-
- for (i=0 ; i < lc_messages_uniq ; i++) {
- m = locales[i].lc_messages_row;
- for (k=0 ; k < 2 ; k++) {
- buf[2*i + k] = (char)((unsigned char) lc_messages_uniq_2[i][k]);
- }
- }
- dump_table8("__lc_messages_rows", buf, lc_messages_uniq * 2);
-
- buf16[0] =0;
- for (i=0 ; i < 2 - 1 ; i++) {
- buf16[i+1] = buf16[i] + lc_messages_count[i];
+ DL_LC_LOOPTAIL(messages)
}
- dump_table16("__lc_messages_item_offsets", buf16, 2);
- m = 0;
- for (k=0 ; k < 2 ; k++) {
- for (i=0 ; i < lc_messages_count[k] ; i++) {
- buf16[m] = lc_messages_item[k][i];
- ++m;
- }
- }
- dump_table16("__lc_messages_item_idx", buf16, m);
+ DO_LC_COMMON(messages)
}
+#undef DO_NL_S
+#define NUM_NL_ctype 10
+static int lc_ctype_item[NUM_NL_ctype][256];
+static int lc_ctype_count[NUM_NL_ctype];
+static unsigned char lc_ctype_uniq_X[700][NUM_NL_ctype];
+static int lc_ctype_uniq;
+#define DO_NL_S(X) lc_ctype_S(X, k++)
-static void read_at_mappings(void)
-{
- char *p;
- char *m;
- int mc = 0;
-
- do {
- if (!(p = strtok(line_buf, " \t\n")) || (*p == '#')) {
- if (!fgets(line_buf, sizeof(line_buf), fp)) {
- if (ferror(fp)) {
- printf("error reading file\n");
- exit(EXIT_FAILURE);
- }
- return; /* EOF */
- }
- if ((*line_buf == '#') && (line_buf[1] == '-')) {
- break;
- }
- continue;
- }
- if (*p == '@') {
- if (p[1] == 0) {
- printf("error: missing @modifier name\n");
- exit(EXIT_FAILURE);
- }
- m = p; /* save the modifier name */
- if (!(p = strtok(NULL, " \t\n")) || p[1] || (((unsigned char) *p) > 0x7f)) {
- printf("error: missing or illegal @modifier mapping char\n");
- exit(EXIT_FAILURE);
- }
- if (at_mappings[(int)((unsigned char) *p)]) {
- printf("error: reused @modifier mapping char\n");
- exit(EXIT_FAILURE);
- }
- at_mappings[(int)((unsigned char) *p)] = 1;
- at_mapto[mc] = *p;
- ++mc;
- *at_strings_end = (char)( (unsigned char) (strlen(m)) );
- strcpy(++at_strings_end, m+1);
- at_strings_end += (unsigned char) at_strings_end[-1];
-
- printf("@mapping: \"%s\" to '%c'\n", m, *p);
-
- if (((p = strtok(NULL, " \t\n")) != NULL) && (*p != '#')) {
- printf("ignoring trailing text: %s...\n", p);
- }
- *line_buf = 0;
- continue;
- }
- break;
- } while (1);
-
-#if 0
- {
- p = at_strings;
-
- if (!*p) {
- printf("no @ strings\n");
- return;
- }
-
- do {
- printf("%s\n", p+1);
- p += 1 + (unsigned char) *p;
- } while (*p);
- }
-#endif
-}
-
-static void read_enable_disable(void)
+static void lc_ctype_S(int X, int k)
{
- char *p;
-
- do {
- if (!(p = strtok(line_buf, " =\t\n")) || (*p == '#')) {
- if (!fgets(line_buf, sizeof(line_buf), fp)) {
- if (ferror(fp)) {
- printf("error reading file\n");
- exit(EXIT_FAILURE);
- }
- return; /* EOF */
- }
- if ((*line_buf == '#') && (line_buf[1] == '-')) {
- break;
- }
- continue;
- }
- if (!strcmp(p, "UTF-8")) {
- if (!(p = strtok(NULL, " =\t\n"))
- || ((toupper(*p) != 'Y') && (toupper(*p) != 'N'))) {
- printf("error: missing or illegal UTF-8 setting\n");
- exit(EXIT_FAILURE);
- }
- default_utf8 = (toupper(*p) == 'Y');
- printf("UTF-8 locales are %sabled\n", "dis\0en"+ (default_utf8 << 2));
- } else if (!strcmp(p, "8-BIT")) {
- if (!(p = strtok(NULL, " =\t\n"))
- || ((toupper(*p) != 'Y') && (toupper(*p) != 'N'))) {
- printf("error: missing or illegal 8-BIT setting\n");
- exit(EXIT_FAILURE);
- }
- default_8bit = (toupper(*p) == 'Y');
- printf("8-BIT locales are %sabled\n", "dis\0en" + (default_8bit << 2));
- } else {
+ int j, m;
+ j = addstring(nl_langinfo(X));
+ for (m=0 ; m < lc_ctype_count[k] ; m++) {
+ if (lc_ctype_item[k][m] == j) {
break;
}
-
- if (((p = strtok(NULL, " \t\n")) != NULL) && (*p != '#')) {
- printf("ignoring trailing text: %s...\n", p);
- }
- *line_buf = 0;
- continue;
-
- } while (1);
-}
-
-#ifdef CODESET_LIST
-
-static int find_codeset_num(const char *cs)
-{
- int r = 2;
- char *s = CODESET_LIST;
-
- /* 7-bit is 1, UTF-8 is 2, 8-bits are > 2 */
-
- if (strcmp(cs, "UTF-8") != 0) {
- ++r;
- while (*s && strcmp(CODESET_LIST+ ((unsigned char) *s), cs)) {
-/* printf("tried %s\n", CODESET_LIST + ((unsigned char) *s)); */
- ++r;
- ++s;
- }
- if (!*s) {
- printf("error: unsupported codeset %s\n", cs);
- exit(EXIT_FAILURE);
- }
- }
- return r;
-}
-
-#else
-
-static int find_codeset_num(const char *cs)
-{
- int r = 2;
-
- /* 7-bit is 1, UTF-8 is 2, 8-bits are > 2 */
-
- if (strcmp(cs, "UTF-8") != 0) {
- printf("error: unsupported codeset %s\n", cs);
- exit(EXIT_FAILURE);
}
- return r;
-}
-
-#endif
-
-static int find_at_string_num(const char *as)
-{
- int i = 0;
- char *p = at_strings;
-
- while (*p) {
- if (!strcmp(p+1, as)) {
- return i;
+ if (m == lc_ctype_count[k]) { /* new for this nl_item */
+ if (m > 255) {
+ printf("too many nl_item %d entries in lc_ctype\n", k);
+ exit(EXIT_FAILURE);
}
- ++i;
- p += 1 + (unsigned char) *p;
+ lc_ctype_item[k][m] = j;
+ ++lc_ctype_count[k];
}
-
- printf("error: unmapped @string %s\n", as);
- exit(EXIT_FAILURE);
+/* printf("\\x%02x", m); */
+ lc_ctype_uniq_X[lc_ctype_uniq][k] = m;
}
-static void read_locale_list(void)
+static void do_lc_ctype(void)
{
- char *p;
- char *s;
- char *ln; /* locale name */
- char *ls; /* locale name ll_CC */
- char *as; /* at string */
- char *ds; /* dot string */
- char *cs; /* codeset */
- int i;
-
- typedef struct {
- char *glibc_name;
- char name[5];
- char dot_cs; /* 0 if no codeset specified */
- char cs;
- } locale_entry;
-
- /* First the C locale. */
- locales[0].glibc_name = locales[0].name;
- strncpy(locales[0].name,"C",5);
- locales[0].dot_cs = 0;
- locales[0].cs = 1; /* 7-bit encoding */
- ++num_locales;
-
- do {
- if (!(p = strtok(line_buf, " \t\n")) || (*p == '#')) {
- if (!fgets(line_buf, sizeof(line_buf), fp)) {
- if (ferror(fp)) {
- printf("error reading file\n");
- exit(EXIT_FAILURE);
- }
- return; /* EOF */
- }
- if ((*line_buf == '#') && (line_buf[1] == '-')) {
- break;
- }
- continue;
- }
-
- s = glibc_locale_names;
- for (i=0 ; i < num_locales ; i++) {
- if (!strcmp(s+1, p)) {
- break;
- }
- s += 1 + ((unsigned char) *s);
- }
- if (i < num_locales) {
- printf("ignoring dulplicate locale name: %s", p);
- *line_buf = 0;
- continue;
- }
-
- /* New locale, but don't increment num until codeset verified! */
- *s = (char)((unsigned char) (strlen(p) + 1));
- strcpy(s+1, p);
- locales[num_locales].glibc_name = s+1;
- ln = p; /* save locale name */
-
- if (!(p = strtok(NULL, " \t\n"))) {
- printf("error: missing codeset for locale %s\n", ln);
- exit(EXIT_FAILURE);
- }
- cs = p;
- i = find_codeset_num(p);
- if ((i == 2) && !default_utf8) {
- printf("ignoring UTF-8 locale %s\n", ln);
- *line_buf = 0;
- continue;
- } else if ((i > 2) && !default_8bit) {
- printf("ignoring 8-bit codeset locale %s\n", ln);
- *line_buf = 0;
- continue;
- }
- locales[num_locales].cs = (char)((unsigned char) i);
-
- if (((p = strtok(NULL, " \t\n")) != NULL) && (*p != '#')) {
- printf("ignoring trailing text: %s...\n", p);
- }
-
- /* Now go back to locale string for .codeset and @modifier */
- as = strtok(ln, "@");
- if (as) {
- as = strtok(NULL, "@");
- }
- ds = strtok(ln, ".");
- if (ds) {
- ds = strtok(NULL, ".");
- }
- ls = ln;
+ int i, k, m;
- if ((strlen(ls) != 5) || (ls[2] != '_')) {
- printf("error: illegal locale name %s\n", ls);
- exit(EXIT_FAILURE);
- }
+ last = buf+1;
+ uniq = 1;
+ *buf = 0;
+ *idx = buf;
- i = 0; /* value for unspecified codeset */
- if (ds) {
- i = find_codeset_num(ds);
- if ((i == 2) && !default_utf8) {
- printf("ignoring UTF-8 locale %s\n", ln);
- *line_buf = 0;
- continue;
- } else if ((i > 2) && !default_8bit) {
- printf("ignoring 8-bit codeset locale %s\n", ln);
- *line_buf = 0;
- continue;
- }
- }
- locales[num_locales].dot_cs = (char)((unsigned char) i);
+ for (i=0 ; i < num_locales ; i++) {
+ k = 0;
- if (as) {
- i = find_at_string_num(as);
- ls[2] = at_mapto[i];
+ if (!setlocale(LC_ALL, locales[i].glibc_name)) {
+ printf("setlocale(LC_ALL,%s) failed!\n",
+ locales[i].glibc_name);
}
- memcpy(locales[num_locales].name, ls, 5);
-/* printf("locale: %5.5s %2d %2d %s\n", */
-/* locales[num_locales].name, */
-/* locales[num_locales].cs, */
-/* locales[num_locales].dot_cs, */
-/* locales[num_locales].glibc_name */
-/* ); */
- ++num_locales;
- *line_buf = 0;
- } while (1);
-}
-static int le_cmp(const void *a, const void *b)
-{
- const locale_entry *p;
- const locale_entry *q;
- int r;
-
- p = (const locale_entry *) a;
- q = (const locale_entry *) b;
+ DO_NL_S(_NL_CTYPE_OUTDIGIT0_MB);
+ DO_NL_S(_NL_CTYPE_OUTDIGIT1_MB);
+ DO_NL_S(_NL_CTYPE_OUTDIGIT2_MB);
+ DO_NL_S(_NL_CTYPE_OUTDIGIT3_MB);
+ DO_NL_S(_NL_CTYPE_OUTDIGIT4_MB);
+ DO_NL_S(_NL_CTYPE_OUTDIGIT5_MB);
+ DO_NL_S(_NL_CTYPE_OUTDIGIT6_MB);
+ DO_NL_S(_NL_CTYPE_OUTDIGIT7_MB);
+ DO_NL_S(_NL_CTYPE_OUTDIGIT8_MB);
+ DO_NL_S(_NL_CTYPE_OUTDIGIT9_MB);
- if (!(r = p->name[0] - q->name[0])
- && !(r = p->name[1] - q->name[1])
- && !(r = p->name[3] - q->name[3])
- && !(r = p->name[4] - q->name[4])
- && !(r = p->name[2] - q->name[2])
- && !(r = -(p->cs - q->cs))
- ) {
- r = -(p->dot_cs - q->dot_cs);
- /* Reverse the ordering of the codesets so UTF-8 comes last.
- * Work-around (hopefully) for glibc bug affecting at least
- * the euro currency symbol. */
+ DL_LC_LOOPTAIL(ctype)
}
- return r;
+ DO_LC_COMMON(ctype)
}
-
generated by cgit v1.2.3 (git 2.25.1) at 2024-05-18 16:56:27 +0000