From 005427cd2a462c63a98cf9d4da3f27c83aa0a030 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Wilfried=20G=C3=B6esgens?= <willi@citadel.org>
Date: Mon, 12 Dec 2005 00:53:19 +0000
Subject: [PATCH] * another rework of the locale parser. Hopefully feature
 complete now.

---
 webcit/ChangeLog |   3 +
 webcit/gettext.c | 175 +++++++++++++++++++++++++++++++++++++++--------
 2 files changed, 148 insertions(+), 30 deletions(-)

diff --git a/webcit/ChangeLog b/webcit/ChangeLog
index ce8738500..589ef13fd 100644
--- a/webcit/ChangeLog
+++ b/webcit/ChangeLog
@@ -1,5 +1,8 @@
 $Id$
 
+Mon Dec 12 1:40:03 CET 2005 dothebart
+* another rework of the locale parser. Hopefully feature complete now.
+	
 Sun Dec 11 00:03:04 EST 2005 ajc
 * Mercilessly ripped out all of the gratuitously complex GNU libintl
   detection m4 crapola and replaced it with some simple autoconf directives
diff --git a/webcit/gettext.c b/webcit/gettext.c
index c99574110..36e7548b3 100644
--- a/webcit/gettext.c
+++ b/webcit/gettext.c
@@ -4,8 +4,11 @@
 
 #ifdef ENABLE_NLS
 
-#define NUM_LANGS 3
+#define NUM_LANGS 4
+#define SEARCH_LANG 20
+
 char *AvailLang[NUM_LANGS] = {
+	"C",
 	"en_US",
 	"de_DE",
 	"it_IT"
@@ -16,47 +19,159 @@ locale_t wc_locales[NUM_LANGS];
 typedef struct _lang_pref{
 	char lang[16];
 	char region[16];
-	char *match;
-	double Priority;
+	long priority;
+	int availability;
+	int selectedlang;
 } LangStruct;
 
-/* TODO: we skip the language weightening so far. */
+/* seems as most browsers just do a one after coma value even if more than 10 locales are available. */
+/* opera: */
+/* Accept-Language: sq;q=1.0,de;q=0.9,as;q=0.8,ar;q=0.7,bn;q=0.6,zh-cn;q=0.5,kn;q=0.4,ch;q=0.3,fo;q=0.2,gn;q=0.1,ce;q=0.1,ie;q=0.1 */
+/* Firefox */
 /* Accept-Language: 'de-de,en-us;q=0.7,en;q=0.3' */
 /* Accept-Language: de,en-ph;q=0.8,en-us;q=0.5,de-at;q=0.3 */
+/* Accept-Language: de,en-us;q=0.9,it;q=0.9,de-de;q=0.8,en-ph;q=0.7,de-at;q=0.7,zh-cn;q=0.6,cy;q=0.5,ar-om;q=0.5,en-tt;q=0.4,xh;q=0.3,nl-be;q=0.3,cs;q=0.2,sv;q=0.1,tk;q=0.1 */
+
 void httplang_to_locale(char *LocaleString)
 {
-	char selected_locale[16];
-	int i, j;
-	char lang[64];
-	int num_accept = 0;
-
-	lprintf(9, "languageAccept: %s\n", LocaleString);
-
-	strcpy(selected_locale, "C");
-	num_accept = num_tokens(LocaleString, ',');
+	LangStruct wanted_locales[SEARCH_LANG];
+	LangStruct *ls;
 
-	for (i=num_accept-1; i>=0; --i) {
-		extract_token(lang, LocaleString, i, ',', sizeof lang);
+	int i = 0;
+	int j = 0;
+	size_t len = strlen(LocaleString);
+	long prio;
+	int av;
+	int nBest;
+	int nParts;
+	char *search = (char *) malloc(len);
+	// locale_t my_Locale;
+	// locale_t my_Empty_Locale;
+	
+	memcpy(search, LocaleString, len);
+	search[len] = '\0';
+	nParts=num_tokens(search,',');
+	for (i=0; ((i<nParts)&&(i<SEARCH_LANG)); i++)
+        {
+			char buf[16];
+			char sbuf[16];
+			char lbuf[16];
+			int blen;
+			
+			ls=&wanted_locales[i];
 
-		/* Strip out the weights; we don't use them.  Also convert
-		 * hyphens to underscores.
-		 */
-		for (j=0; j<strlen(lang); ++j) {
-			if (lang[j] == '-') lang[j] = '_';
-			if (lang[j] == ';') lang[j] = 0;
-		}
-
-		for (j=0; j<NUM_LANGS; ++j) {
-			if (!strncasecmp(lang, AvailLang[j], strlen(lang))) {
-				strcpy(selected_locale, AvailLang[j]);
+			extract_token(&buf[0],search, i,',',16);
+			/* we are searching, if this list item has something like ;q=n*/
+			if (num_tokens(&buf[0],'=')>1) {
+				int sbuflen, k;
+				extract_token(&sbuf[0],&buf[0], 1,'=',16);
+				sbuflen=strlen(&sbuf[0]);
+				for (k=0; k<sbuflen; k++) if (sbuf[k]=='.') sbuf[k]='0';
+				ls->priority=atol(&sbuf[0]);
+			}
+			else {
+				ls->priority=1000;
+			}
+			/* get the locale part */
+			extract_token(&sbuf[0],&buf[0],0,';',16);
+			/* get the lang part, which should be allways there */
+			extract_token(&ls->lang[0],&sbuf[0],0,'-',16);
+			/* get the area code if any. */
+			if (num_tokens(&sbuf[0],'-')>1) {
+				extract_token(&ls->region[0],&sbuf[0],1,'-',16);
+			}
+			else { /* no ara code? use lang code */
+				blen=strlen(&ls->lang[0]);
+				memcpy(&ls->region[0], ls->lang,blen);
+				ls->region[blen]='\0';
+			} /* area codes are uppercase */
+			blen=strlen(&ls->region[0]);
+			for (j=0; j<blen; j++)
+				{
+					int chars=toupper(ls->region[j]);
+					ls->region[j]=(char)chars;/*todo ?! */
+				}
+			sprintf(&lbuf[0],"%s_%s",&ls->lang[0],&ls->region[0]);
+			
+			/* check if we have this lang */
+			ls->availability=1;
+			ls->selectedlang=-1;
+			for (j=0; j<NUM_LANGS; j++) {
+				int result;
+				/* match against the LANG part */
+				result=strcasecmp(&ls->lang[0], AvailLang[j]);
+				if ((result<0)&&(result<ls->availability)){
+					ls->availability=result;
+					ls->selectedlang=j;
+				}
+				/* match against lang and locale */
+				if (0==strcasecmp(&lbuf[0], AvailLang[j])){
+					ls->availability=0;
+					ls->selectedlang=j;
+					j=NUM_LANGS;
+				}
+			}
+        }
+	
+	prio=0;
+	av=-1000;
+	nBest=-1;
+	for (i=0; ((i<nParts)&&(i<SEARCH_LANG)); i++)
+		{
+			ls=&wanted_locales[i];
+			if ((ls->availability<=0)&& 
+				(av<ls->availability)&&
+				(prio<ls->priority)&&
+				(ls->selectedlang!=-1)){
+				nBest=ls->selectedlang;
+				av=ls->availability;
+				prio=ls->priority;
 			}
 		}
-	}
-
-	lprintf(9, "language found: %s\n", selected_locale);
-	set_selected_language(selected_locale);
+	if (nBest==-1) /* fall back to C */
+		nBest=0;
+	WC->selected_language=nBest;
+	lprintf(9, "language found: %s\n", AvailLang[WC->selected_language]);
+	//	set_selected_language(selected_locale);
 }
 
+/* TODO: we skip the language weightening so far. */
+/* Accept-Language: 'de-de,en-us;q=0.7,en;q=0.3' */
+/* Accept-Language: de,en-ph;q=0.8,en-us;q=0.5,de-at;q=0.3 */
+//void httplang_to_locale(char *LocaleString)
+//{
+//	char selected_locale[16];
+//	int i, j;
+//	char lang[64];
+//	int num_accept = 0;
+//
+//	lprintf(9, "languageAccept: %s\n", LocaleString);
+//
+//	strcpy(selected_locale, "C");
+//	num_accept = num_tokens(LocaleString, ',');
+//
+//	for (i=num_accept-1; i>=0; --i) {
+//		extract_token(lang, LocaleString, i, ',', sizeof lang);
+//
+//		/* Strip out the weights; we don't use them.  Also convert
+//		 * hyphens to underscores.
+//		 */
+//		for (j=0; j<strlen(lang); ++j) {
+//			if (lang[j] == '-') lang[j] = '_';
+//			if (lang[j] == ';') lang[j] = 0;
+//		}
+//
+//		for (j=0; j<NUM_LANGS; ++j) {
+//			if (!strncasecmp(lang, AvailLang[j], strlen(lang))) {
+//				strcpy(selected_locale, AvailLang[j]);
+//			}
+//		}
+//	}
+//
+//	lprintf(9, "language found: %s\n", selected_locale);
+//	set_selected_language(selected_locale);
+//}
+
 
 void offer_languages(void) {
 	int i;
-- 
2.39.2