support utf8 convertion, but still buggy

git-svn-id: http://opensvn.csie.org/pttbbs/trunk/pttbbs@1456 63ad8ddf-47c3-0310-b6dd-a9e9d9715204

3 files changed, 60 insertions, 91 deletions

diff --git a/mbbsd/convert.c b/mbbsd/convert.c
index 141548d1..75480ab1 100644
--- a/mbbsd/convert.c
+++ b/mbbsd/convert.c
@@ -4,74 +4,58 @@
  * The following code is copied and modified from "autoconvert" with GPL.
  */
 
-#ifdef GB_CONVERT
+#ifdef CONVERT
 
-#include "convert.h"
+extern read_write_type write_type;
+extern read_write_type read_type;
 
-char *hzconvert(char *, int *, char *, void (*dbcvrt)());
-
-extern const unsigned char GtoB[], BtoG[];
-
-#define	c1	(unsigned char)(s[0])
-#define	c2	(unsigned char)(s[1])
-
-static void g2b(char *s)
+static int gb_read(int fd, void *buf, size_t count)
 {
-    unsigned int i;
-
-    if ((c1 >= 0x81) && (c1 <= 0xfe) && (c2 >= 0x40) && (c2 <= 0xfe) && (c2 != 0x7f)) {
-	/*
-	 * c1c2 in the table 
-	 */
-	if (c2 < 0x7f)
-	    i = ((c1 - 0x81) * 190 + (c2 - 0x40)) * 2;
-	else
-	    i = ((c1 - 0x81) * 190 + (c2 - 0x41)) * 2;
-	s[0] = GtoB[i++];
-	s[1] = GtoB[i];
-    } else {                    /* c1c2 not in the table */
-	if ((char) s[1] >= 0)   /* half HANZI-character */
-	    s[0] = '?';
-	else {                  /* invalid gbk-character */
-	    s[0] = GtoB_bad1;
-	    s[1] = GtoB_bad2;
-	}
-    }
+    int len = read(fd, buf, count);
+    if (len > 0)
+	gb2big(buf, len);
+    return len;
 }
 
-static void b2g(char *s)
+static int gb_write(int fd, void *buf, size_t count)
 {
-    unsigned int i;
+    big2gb(buf, count);
+    return write(fd, buf, count);
+}
 
-    if ((c1 >= 0x81) && (c1 <= 0xfe) && (c2 >= 0x40) && (c2 <= 0xfe) && (c2 != 0x7f)) {
-        /*
-         * c1c2 in the table 
-         */
-        if (c2 < 0x7f)
-            i = ((c1 - 0x81) * 190 + (c2 - 0x40)) * 2;
-        else
-            i = ((c1 - 0x81) * 190 + (c2 - 0x41)) * 2;
-        s[0] = BtoG[i++];
-        s[1] = BtoG[i];
-    } else {                    /* c1c2 not in the table */
-        if ((char) s[1] >= 0)   /* half HANZI-character */
-            s[0] = '?';
-        else {                  /* invalid big5-character */
-            s[0] = BtoG_bad1;
-            s[1] = BtoG_bad2;
-        }
+static int utf8_read(int fd, void *buf, size_t count)
+{
+    count = read(fd, buf, count);
+    if (count > 0) {
+	strcpy(buf, utf8_uni(buf, &count, 0));
+	uni2big(buf, &count, 0);
+	((char *)buf)[count] = 0;
     }
+    return count;
 }
 
-unsigned char *gb2big(unsigned char *s, int plen)
+static int utf8_write(int fd, void *buf, size_t count)
 {
-    unsigned char c = 0;
-    return hzconvert(s, &plen, &c, g2b);
+    strcpy(buf, big2uni(buf, &count, 0));
+    uni_utf8(buf, &count, 0);
+    ((char *)buf)[count] = 0;
+    return write(fd, buf, count);
 }
 
-unsigned char *big2gb(unsigned char *s, int plen)
+void set_converting_type(int which)
 {
-    unsigned char c = 0;
-    return hzconvert(s, &plen, &c, b2g);
+    if (which == CONV_NORMAL) {
+	read_type = read;
+	write_type = (read_write_type)write;
+    }
+    else if (which == CONV_GB) {
+	read_type = gb_read;
+	write_type = gb_write;
+    }
+    else if (which == CONV_UTF8) {
+	read_type = utf8_read;
+	write_type = utf8_write;
+    }
 }
+
 #endif
diff --git a/mbbsd/io.c b/mbbsd/io.c
index 41ce6813..97ddb35d 100644
--- a/mbbsd/io.c
+++ b/mbbsd/io.c
@@ -17,37 +17,10 @@ static int      icurrchar = 0;
 /* ----------------------------------------------------- */
 /* convert routines                                      */
 /* ----------------------------------------------------- */
-#ifdef GB_CONVERT
+#ifdef CONVERT
 
-typedef int (* read_write_type)(int, void *, size_t);
-static read_write_type write_type = (read_write_type)write;
-static read_write_type read_type = read;
-
-int converting_read(int fd, void *buf, size_t count)
-{
-    int len = read(fd, buf, count);
-    if (len >= 0)
-	gb2big(buf, len);
-    return len;
-}
-
-int converting_write(int fd, void *buf, size_t count)
-{
-    big2gb(buf, count);
-    return write(fd, buf, count);
-}
-
-void set_converting_type(int which)
-{
-    if (which == 0) {
-	read_type = read;
-	write_type = (read_write_type)write;
-    }
-    else if (which == 1) {
-	read_type = converting_read;
-	write_type = converting_write;
-    }
-}
+read_write_type write_type = (read_write_type)write;
+read_write_type read_type = read;
 
 inline static int read_wrapper(int fd, void *buf, size_t count) {
     return (*read_type)(fd, buf, count);
@@ -65,7 +38,7 @@ void
 oflush()
 {
     if (obufsize) {
-#ifdef GB_CONVERT
+#ifdef CONVERT
 	write_wrapper(1, outbuf, obufsize);
 #else
 	write(1, outbuf, obufsize);
@@ -87,7 +60,7 @@ output(char *s, int len)
 
     assert(len<OBUFSIZE);
     if (obufsize + len > OBUFSIZE) {
-#ifdef GB_CONVERT
+#ifdef CONVERT
 	write_wrapper(1, outbuf, obufsize);
 #else
 	write(1, outbuf, obufsize);
@@ -193,7 +166,7 @@ dogetch()
 	do{
 #endif
 
-#ifdef GB_CONVERT
+#ifdef CONVERT
 	    while ((len = read_wrapper(0, inbuf, IBUFSIZE)) <= 0) {
 #else
 	    while ((len = read(0, inbuf, IBUFSIZE)) <= 0) {
diff --git a/mbbsd/mbbsd.c b/mbbsd/mbbsd.c
index 425406f1..5bd1185b 100644
--- a/mbbsd/mbbsd.c
+++ b/mbbsd/mbbsd.c
@@ -512,7 +512,7 @@ inline static void mkuserdir(char *userid)
 static void
 login_query()
 {
-#ifdef GB_CONVERT
+#ifdef CONVERT
     /* uid 加一位, for gb login */
     char            uid[IDLEN + 2], passbuf[PASSLEN];
     int             attempts, len;
@@ -547,11 +547,15 @@ login_query()
 	getdata(20, 0, "請輸入代號，或以[guest]參觀，以[new]註冊：",
 		uid, sizeof(uid), DOECHO);
 
-#ifdef GB_CONVERT
+#ifdef CONVERT
 	/* switch to gb mode if uid end with '.' */
 	len = strlen(uid);
 	if (uid[0] && uid[len - 1] == '.') {
-	    set_converting_type(1);
+	    set_converting_type(CONV_GB);
+	    uid[len - 1] = 0;
+	}
+	else if (uid[0] && uid[len - 1] == ',') {
+	    set_converting_type(CONV_UTF8);
 	    uid[len - 1] = 0;
 	}
 	else if (len >= IDLEN + 1)
@@ -1337,6 +1341,14 @@ daemon_login(int argc, char *argv[], char *envp[])
     setuid(BBSUID);
     chdir(BBSHOME);
 
+    /* It's better to do something before fork */
+#ifdef CONVERT
+    big2gb_init();
+    gb2big_init();
+    big2uni_init();
+    uni2big_init();
+#endif
+    
 #ifndef NO_FORK
 #ifdef PRE_FORK
     if( listen_port == 23 ){ // only pre-fork in port 23