Add locale-enabled strcoll function from vodz, plus supporting tool.

5 files changed, 283 insertions, 76 deletions

diff --git a/libc/misc/ctype/ctype.c b/libc/misc/ctype/ctype.c
index 8d6a1dba7..18ffed4a5 100644
--- a/libc/misc/ctype/ctype.c
+++ b/libc/misc/ctype/ctype.c
@@ -157,7 +157,7 @@ toupper( int c )
 #else   /* __UCLIBC_HAS_LOCALE__ */
 
 #include <limits.h>
-#include "./ctype.h"
+#include "../locale/_locale.h"
 
 #define _UC_ISCTYPE(c, type) \
 ((c != -1) && ((_uc_ctype_b[(int)((unsigned char)c)] & type) != 0))
diff --git a/libc/misc/ctype/ctype.h b/libc/misc/locale/_locale.h
index f9a34cb18..139a862f9 100644
--- a/libc/misc/ctype/ctype.h
+++ b/libc/misc/locale/_locale.h
@@ -19,3 +19,4 @@ enum
   ISxdigit = ISbit (7),        /* 128 Hexnumeric.        */
 };
 
+extern const unsigned char *_uc_collate_b;
diff --git a/libc/misc/locale/locale.c b/libc/misc/locale/locale.c
index 2abdde34e..d978ae37c 100644
--- a/libc/misc/locale/locale.c
+++ b/libc/misc/locale/locale.c
@@ -1,5 +1,5 @@
 /* setlocale.c
- * Load LC_CTYPE locale only special for uclibc
+ * Load LC_CTYPE and LC_COLLATE locale only special for uclibc
  *
  * Written by Vladimir Oleynik (c) vodz@usa.net
  *
@@ -8,103 +8,162 @@
  * used ideas is part of the GNU C Library.
  */
 
-/*
- * No-locale-support setlocale() added.
- */
-
 #include <locale.h>
 #include <stdio.h>      /* NULL, fopen */
 #include <stdlib.h>     /* malloc */
 #include <string.h>
 #include <limits.h>     /* PATH_MAX */
+#include <errno.h>      /* EINVAL */
+#include <unistd.h>     /* get(e)[u|g]id */
 
-#include "../ctype/ctype.h"
-
-#undef TEST_LOCALE
+#include "_locale.h"
 
+static char C_LOCALE_NAME    []="C";
+static char POSIX_LOCALE_NAME[]="POSIX";
+static char composite_name_C []=
+"LC_CTYPE=C;LC_NUMERIC=C;LC_TIME=C;LC_COLLATE=C;LC_MONETARY=C;LC_MESSAGES=C";
 
 #ifdef __UCLIBC_HAS_LOCALE__
 
-static char C_LOCALE_NAME[]="C";
-
 #ifdef TEST_LOCALE
 static const char PATH_LOCALE[]="./";
 #else
 static const char PATH_LOCALE[]=__UCLIBC_LOCALE_DIR;
 #endif
 
-static const char LC_CTYPE_STR[]="/LC_CTYPE";
-
 struct SAV_LOADED_LOCALE {
+	int category;
 	char *locale;
 	const unsigned char *buf;
 	struct SAV_LOADED_LOCALE *next;
 };
 
+static struct SAV_LOADED_LOCALE sll_C_LC_MESSAGES = {
+  LC_MESSAGES, C_LOCALE_NAME,  0, 0
+};
 
-static struct SAV_LOADED_LOCALE sav_loaded_locale [1] = {
-	{ C_LOCALE_NAME, _uc_ctype_b_C, 0 }
+static struct SAV_LOADED_LOCALE sll_C_LC_MONETARY = {
+  LC_MONETARY, C_LOCALE_NAME, 0, &sll_C_LC_MESSAGES
 };
 
-static struct SAV_LOADED_LOCALE * old_locale = sav_loaded_locale;
+static struct SAV_LOADED_LOCALE sll_C_LC_COLLATE = {
+  LC_COLLATE,  C_LOCALE_NAME, 0, &sll_C_LC_MONETARY
+};
 
-static char *set_new_locale(struct SAV_LOADED_LOCALE * s_locale)
-{
-	_uc_ctype_b     = s_locale->buf;
-	_uc_ctype_trans = s_locale->buf+LOCALE_BUF_SIZE/2;
-	old_locale      = s_locale;
-	return            s_locale->locale;
-}
+static struct SAV_LOADED_LOCALE sll_C_LC_TIME = {
+  LC_TIME,     C_LOCALE_NAME, 0, &sll_C_LC_COLLATE
+};
 
-/* Current support only LC_CTYPE or LC_ALL category */
+static struct SAV_LOADED_LOCALE sll_C_LC_NUMERIC = {
+  LC_NUMERIC,  C_LOCALE_NAME, 0, &sll_C_LC_TIME
+};
 
-char *setlocale(int category, const char *locale)
+static struct SAV_LOADED_LOCALE sll_C_LC_CTYPE = {
+  LC_CTYPE,    C_LOCALE_NAME, _uc_ctype_b_C, &sll_C_LC_NUMERIC
+};
+
+static struct SAV_LOADED_LOCALE *sll = &sll_C_LC_CTYPE;
+
+
+#endif /* __UCLIBC_HAS_LOCALE__ */
+
+
+static char *nl_current[LC_ALL+1] = {
+	C_LOCALE_NAME, C_LOCALE_NAME, C_LOCALE_NAME,
+	C_LOCALE_NAME, C_LOCALE_NAME, C_LOCALE_NAME,
+	composite_name_C
+};
+
+static const char * const LC_strs[LC_ALL+1] = {
+	"/LC_CTYPE",
+	"/LC_NUMERIC",
+	"/LC_TIME",
+	"/LC_COLLATE",
+	"/LC_MONETARY",
+	"/LC_MESSAGES",
+	"/LC_ALL"
+};
+
+static char *find_locale(int c, const char **plocale)
 {
-	FILE * fl;
+#ifdef __UCLIBC_HAS_LOCALE__
 	struct SAV_LOADED_LOCALE *cur;
-	struct SAV_LOADED_LOCALE *bottom;
-	char   full_path[PATH_MAX];
-	char * buf = 0;
-	int    l;
+#endif
+	const char *name = *plocale;
+
+	if (name[0] == '\0') {
+	/* The user decides which locale to use by setting environment
+								variables. */
+		name = getenv (&LC_strs[LC_ALL][1]);
+		if (name == NULL || name[0] == '\0')
+			name = getenv (&LC_strs[c][1]);
+		if (name == NULL || name[0] == '\0')
+			name = getenv ("LANG");
+		if (name == NULL || name[0] == '\0')
+			name = C_LOCALE_NAME;
+	}
 
-	if(category!=LC_CTYPE && category!=LC_ALL)
-		return NULL;
+	if (strcmp (name, C_LOCALE_NAME) == 0 ||
+				strcmp (name, POSIX_LOCALE_NAME) == 0 ||
+		/* TODO! */     (c!=LC_CTYPE && c!=LC_COLLATE))
+	    name = C_LOCALE_NAME;
 
-	if(locale==0)
-		return set_new_locale(old_locale);
+	*plocale = name;
 
-	if(strcmp(locale, "POSIX")==0)
-		return set_new_locale(sav_loaded_locale);
-	 else if(*locale == '\0') {
+#ifdef __UCLIBC_HAS_LOCALE__
+	for(cur = sll; cur; cur = cur->next)
+		if(cur->category == c && strcmp(cur->locale, name)==0)
+			return cur->locale;
+#else
+	if(name == C_LOCALE_NAME)
+		return C_LOCALE_NAME;
+#endif
+	return NULL;
+}
 
-		locale = getenv(LC_CTYPE_STR+1);
-		if(locale == 0 || *locale == 0)
-			locale = getenv("LANG");
-		if(locale == 0 || *locale == '\0')
-			return set_new_locale(old_locale);
-		if(strcmp(locale, "POSIX")==0)
-			return set_new_locale(sav_loaded_locale);
-	}
 
-	for(cur = sav_loaded_locale; cur; cur = cur->next)
-		if(strcmp(cur->locale, locale)==0)
-			return set_new_locale(cur);
+#ifdef __UCLIBC_HAS_LOCALE__
+static char *load_locale(int category, const char *locale)
+{
+	FILE * fl;
+	char   full_path[PATH_MAX];
+	char * buf = 0;
+	struct SAV_LOADED_LOCALE *cur;
+	struct SAV_LOADED_LOCALE *bottom;
+	int    bufsize;
+	int    l = strlen(locale);
 
-	l = strlen(locale);
-	if((l+sizeof(PATH_LOCALE)+sizeof(LC_CTYPE_STR))>=PATH_MAX)
+	if((l+sizeof(PATH_LOCALE)+strlen(LC_strs[category]))>=PATH_MAX)
 		return NULL;
 
+	/* Not allow acces suid/sgid binaries to outside PATH_LOCALE */
+	if((geteuid()!=getuid() || getegid()!=getgid()) &&
+		strchr(locale, '/')!=NULL)
+			return NULL;
+
 	strcpy(full_path, PATH_LOCALE);
 	strcat(full_path, locale);
-	strcat(full_path, LC_CTYPE_STR);
+	strcat(full_path, LC_strs[category]);
 	fl = fopen(full_path, "r");
 	if(fl==0)
 		return NULL;
 
-	cur = malloc(sizeof(struct SAV_LOADED_LOCALE)+LOCALE_BUF_SIZE+l);
+	switch(category) {
+		case LC_CTYPE:
+			bufsize = LOCALE_BUF_SIZE;
+			break;
+		case LC_COLLATE:
+			bufsize = 256;
+			break;
+		default:        /* TODO */
+			bufsize = 0;
+			break;
+	}
+
+	cur = malloc(sizeof(struct SAV_LOADED_LOCALE)+bufsize+l+2);
 	if(cur) {
 		buf = (char *)(cur+1);
-		if(fread(buf, 1, LOCALE_BUF_SIZE+1, fl)!=(LOCALE_BUF_SIZE)) {
+		if(bufsize!=0 && fread(buf, 1, bufsize+1, fl)!=(bufsize)) {
 			/* broken locale file */
 			free(cur);
 			buf = 0;
@@ -117,45 +176,157 @@ char *setlocale(int category, const char *locale)
 	fclose(fl);
 	if(cur==0)      /* not enough memory */
 		return NULL;
-	if(buf==0)      /* broken locale file, set to "C" */
-		return set_new_locale(sav_loaded_locale);
+	if(buf==0) {    /* broken locale file, set to "C" */
+		return C_LOCALE_NAME;
+	}
 
-	cur->next = 0;
-	strcpy(buf+LOCALE_BUF_SIZE, locale);
+	cur->next     = 0;
+	cur->buf      = buf;
+	cur->category = category;
+	cur->locale   = buf+bufsize;
+	strcpy(cur->locale, locale);
 
-	bottom = sav_loaded_locale;
+	bottom = sll;
 	while(bottom->next!=0)
 		bottom = bottom->next;
 	bottom->next = cur;
 
-	/* next two line only pedantic */
-	cur->buf    = buf;
-	cur->locale = buf+LOCALE_BUF_SIZE;
+	return cur->locale;
+}
+
+static char *set_composite(int category, char *locale)
+{
+	int i, l;
+	char *old_composite_name = nl_current[LC_ALL];
+	char *new_composite_name;
+	struct SAV_LOADED_LOCALE *cur;
 
-	return set_new_locale(cur);
+	for(l=i=0; i<LC_ALL; i++) {
+		new_composite_name = i == category ? locale : nl_current[i];
+						/* '=' + ';' or '\0' */
+		l += strlen(&LC_strs[i][1])+strlen(new_composite_name)+2;
+	}
+
+	new_composite_name = malloc(l);
+	if(new_composite_name==NULL)
+		return NULL;
+	if(old_composite_name!=composite_name_C)
+		free(old_composite_name);
+	nl_current[category] = locale;  /* change after malloc */
+
+	*new_composite_name = 0;
+	for(i=0; i<LC_ALL; i++) {
+		if(i)
+			strcat(new_composite_name, ";");
+		strcat(new_composite_name, &LC_strs[i][1]);
+		strcat(new_composite_name, "=");
+		strcat(new_composite_name, nl_current[i]);
+	}
+	nl_current[LC_ALL] = new_composite_name;
+
+	/* set locale data for ctype and strcollate functions */
+	for(cur = sll; ; cur = cur->next)
+		if(cur->category == category && cur->locale == locale)
+			break;
+
+	switch(category) {
+		case LC_CTYPE:
+			_uc_ctype_b     = cur->buf;
+			_uc_ctype_trans = cur->buf+LOCALE_BUF_SIZE/2;
+			break;
+		case LC_COLLATE:
+			_uc_collate_b   = cur->buf;
+			break;
+		default:        /* TODO */
+			break;
+	}
+	return locale;
 }
 
-#else /* no locale support */
+#endif /* __UCLIBC_HAS_LOCALE__ */
 
 char *setlocale(int category, const char *locale)
 {
-	/* Allow locales "C" and "" (native).  Both are "C" for our purposes. */
-	if (locale) {
-		if (*locale == 'C') {
-			++locale;
-		}
-		if (*locale) {				/* locale wasn't "C" or ""!! */
-			return NULL;
-		}
-	}
+	char * tl;
+#ifdef __UCLIBC_HAS_LOCALE__
+	int    i;
+#endif
 
-	/* Allow any legal category for "C" or "" (native) locale. */
-	if((category < LC_CTYPE) || (category > LC_ALL)) { /* Illegal category! */
+	if (category < 0 || category > LC_ALL) {
+#ifdef __UCLIBC_HAS_LOCALE__
+einval:
+#endif
+		errno = EINVAL;
 		return NULL;
 	}
 
-	return "C";
-}
+	if(locale==NULL)
+		return nl_current[category];
 
+	if(category!=LC_ALL) {
+		tl = find_locale(category, &locale);
+#ifdef __UCLIBC_HAS_LOCALE__
+		if(tl==NULL)
+			tl = load_locale(category, locale);
+		if(tl) {
+			if(nl_current[category] != tl)
+				tl = set_composite(category, tl);
+		}
 #endif
+		return tl;
+	}
+	/* LC_ALL */
+#ifdef __UCLIBC_HAS_LOCALE__
+	/* The user wants to set all categories.  The desired locales
+	 for the individual categories can be selected by using a
+	 composite locale name.  This is a semi-colon separated list
+	 of entries of the form `CATEGORY=VALUE'.  */
+	tl = strchr(locale, ';');
+	if(tl==NULL) {
+	/* This is not a composite name. Load the data for each category. */
+		for(i=0; i<LC_ALL; i++)
+			setlocale(i, locale);   /* recursive */
+	} else {
+	/* This is a composite name.  Make a copy and split it up. */
+	  const char *newnames[LC_ALL];
+	  char *np;
+	  char *cp;
+
+	  i  = strlen(locale);
+	  np = alloca (i);
+	  if(np)
+		strcpy(np, locale);
+	  else
+		return NULL;
+	  for (i = 0; i < LC_ALL; ++i)
+		newnames[i] = 0;
+
+	  while ((cp = strchr (np, '=')) != NULL) {
+	      for (i = 0; i < LC_ALL; ++i)
+		if ((size_t) (cp - np) == strlen(&LC_strs[i][1])
+		    && memcmp (np, &LC_strs[i][1], (cp - np)) == 0)
+		  break;
+
+	      if (i == LC_ALL)
+		/* Bogus category name.  */
+		goto einval;
+
+	      /* Found the category this clause sets.  */
+	      newnames[i] = ++cp;
+	      cp = strchr (cp, ';');
+	      if (cp != NULL) {
+		  /* Examine the next clause.  */
+		  *cp = '\0';
+		  np = cp + 1;
+	      } else
+		/* This was the last clause.  We are done.  */
+		break;
+	  }
+
+	  for (i = 0; i < LC_ALL; ++i)
+		setlocale(i, newnames[i]);   /* recursive */
+	}
 
+#endif
+	return nl_current[LC_ALL];
+}
diff --git a/libc/string/Makefile b/libc/string/Makefile
index e2d04c3f9..48d3a3426 100644
--- a/libc/string/Makefile
+++ b/libc/string/Makefile
@@ -28,6 +28,10 @@ MOBJ=strlen.o strcat.o strcpy.o strchr.o strcmp.o strncat.o strncpy.o \
 	strncmp.o strrchr.o strdup.o memcpy.o memccpy.o memset.o \
 	memmove.o memcmp.o memchr.o ffs.o strnlen.o strxfrm.o
 
+ifeq ($(HAS_LOCALE),true)
+	MOBJ += strcoll.o
+endif
+
 MSRC1=strsignal.c
 MOBJ1=strsignal.o psignal.o
 
diff --git a/libc/string/string.c b/libc/string/string.c
index b3070907c..0e2df303b 100644
--- a/libc/string/string.c
+++ b/libc/string/string.c
@@ -76,7 +76,38 @@ int strcmp(const char *s1, const char *s2)
 	return c1 - c2;
 }
 
+#ifndef __UCLIBC_HAS_LOCALE__
 __asm__(".weak strcoll; strcoll = strcmp");
+#endif /* __UCLIBC_HAS_LOCALE__ */
+#endif
+
+/***** Function strcoll (locale only, as non-locale is alias of strcmp *****/
+
+#ifdef L_strcoll
+#ifdef __UCLIBC_HAS_LOCALE__
+
+#include "../misc/locale/_locale.h"
+
+const unsigned char *_uc_collate_b;  /* NULL for no collate, strcoll->strcmp */
+
+int strcoll(const char *s1, const char *s2)
+{
+	unsigned char c1, c2;
+
+	while(1) {
+		c1 = (unsigned char) *s1;
+		c2 = (unsigned char) *s2;
+		if(_uc_collate_b) {     /* setuped non-C locale? */
+			c1 = _uc_collate_b[c1];
+			c2 = _uc_collate_b[c2];
+		}
+		if (*s1 == '\0' || c1 != c2)
+			return c1 - c2;
+		s1++;
+		s2++;
+	}
+}
+#endif /* __UCLIBC_HAS_LOCALE__ */
 #endif
 
 /********************** Function strncat ************************************/