Changeset c608891

Timestamp:

2013-04-23T16:20:06Z (12 years ago)

Author:

Wilmer van der Gaast <wilmer@…>

Branches:

master

Children:

5cb9461

Parents:

e277e80

Message:

Simple (and possibly still fragile) support for UTF-8 nicknames.

Files:

• irc.c (modified) (3 diffs)
• irc.h (modified) (1 diff)
• nick.c (modified) (9 diffs)
• nick.h (modified) (1 diff)

irc.c

@@ -35 +35 @@
 static char *set_eval_password( set_t *set, char *value );
 static char *set_eval_bw_compat( set_t *set, char *value );
+static char *set_eval_utf8_nicks( set_t *set, char *value );
 
 irc_t *irc_new( int fd )
@@ -134 +135 @@
         s = set_add( &b->set, "to_char", ": ", set_eval_to_char, irc );
         s = set_add( &b->set, "typing_notice", "false", set_eval_bool, irc );
+        s = set_add( &b->set, "utf8_nicks", "false", set_eval_utf8_nicks, irc );
 
         irc->root = iu = irc_user_new( irc, ROOT_NICK );
@@ -962 +964 @@
 }
 
+static char *set_eval_utf8_nicks( set_t *set, char *value )
+{
+        irc_t *irc = set->data;
+        gboolean val = bool2int( value );
+        
+        /* Do *NOT* unset this flag in the middle of a session. There will
+           be UTF-8 nicks around already so if we suddenly disable support
+           for them, various functions might behave strangely. */
+        if( val )
+                irc->status |= IRC_UTF8_NICKS;
+        else if( irc->status & IRC_UTF8_NICKS )
+                irc_rootmsg( irc, "You need to reconnect to BitlBee for this "
+                                  "change to take effect." );
+        
+        return set_eval_bool( set, value );
+}
+
 void register_irc_plugin( const struct irc_plugin *p )
 {

irc.h

@@ -62 +62 @@
         OPER_HACK_ACCOUNT_ADD = 0x400,
         OPER_HACK_ANY = 0x3700, /* To check for them all at once. */
+        
+        IRC_UTF8_NICKS = 0x10000, /* Disable ASCII restrictions on buddy nicks. */
 } irc_status_t;
 

nick.c

@@ -112 +112 @@
 {
         gboolean ok = FALSE; /* Set to true once the nick contains something unique. */
-        GString *ret = g_string_new( "" );
+        GString *ret = g_string_sized_new( MAX_NICK_LENGTH + 1 );
+        char *rets;
+        irc_t *irc = (irc_t *) bu->bee->ui_data;
         char *fmt = set_getstr( &bu->ic->acc->set, "nick_format" ) ? :
                     set_getstr( &bu->bee->set, "nick_format" );
@@ -119 +121 @@
         {
                 char *part = NULL, chop = '\0', *asc = NULL;
-                int len = MAX_NICK_LENGTH;
                 
                 if( *fmt != '%' )
@@ -142 +143 @@
                                 fmt += 2;
                         }
-                        else if( isdigit( *fmt ) )
-                        {
-                                len = 0;
-                                /* Grab a number. */
-                                while( isdigit( *fmt ) )
-                                        len = len * 10 + ( *(fmt++) - '0' );
-                        }
                         else if( g_strncasecmp( fmt, "nick", 4 ) == 0 )
                         {
@@ -197 +191 @@
                 }
                 
+                if( !part )
+                        continue;
+                
                 /* Credits to Josay_ in #bitlbee for this idea. //TRANSLIT
                    should do lossy/approximate conversions, so letters with
                    accents don't just get stripped. Note that it depends on
                    LC_CTYPE being set to something other than C/POSIX. */
-                if( part )
+                if( !( irc && irc->status & IRC_UTF8_NICKS ) )
                         part = asc = g_convert_with_fallback( part, -1, "ASCII//TRANSLIT",
                                                               "UTF-8", "", NULL, NULL, NULL );
                 
-                if( ret->len == 0 && part && isdigit( *part ) )
-                        g_string_append_c( ret, '_' );
-                
-                while( part && *part && *part != chop && len > 0 )
-                {
-                        if( strchr( nick_lc_chars, *part ) ||
-                            strchr( nick_uc_chars, *part ) )
-                                g_string_append_c( ret, *part );
-                        
-                        part ++;
-                        len --;
-                }
+                if( part )
+                        g_string_append( ret, part );
                 g_free( asc );
         }
         
-        /* This returns NULL if the nick is empty or otherwise not ok. */
-        return g_string_free( ret, ret->len == 0 || !ok );
+        rets = g_string_free( ret, FALSE );
+        if( ok && rets && *rets )
+        {
+                nick_strip( irc, rets );
+                rets[MAX_NICK_LENGTH] = '\0';
+                return rets;
+        }
+        g_free( rets );
+        return NULL;
 }
 
@@ -247 +241 @@
                 if( inf_protection-- == 0 )
                 {
-                        int i;
+                        g_snprintf( nick, MAX_NICK_LENGTH + 1, "xx%x", rand() );
                         
-                        irc_rootmsg( irc, "Warning: Almost had an infinite loop in nick_get()! "
-                                          "This used to be a fatal BitlBee bug, but we tried to fix it. "
-                                          "This message should *never* appear anymore. "
-                                          "If it does, please *do* send us a bug report! "
-                                          "Please send all the following lines in your report:" );
-                        
-                        irc_rootmsg( irc, "Trying to get a sane nick for handle %s", bu->handle );
-                        for( i = 0; i < MAX_NICK_LENGTH; i ++ )
-                                irc_rootmsg( irc, "Char %d: %c/%d", i, nick[i], nick[i] );
-                        
-                        irc_rootmsg( irc, "FAILED. Returning an insane nick now. Things might break. "
-                                          "Good luck, and please don't forget to paste the lines up here "
-                                          "in #bitlbee on OFTC or in a mail to wilmer@gaast.net" );
-                        
-                        g_snprintf( nick, MAX_NICK_LENGTH + 1, "xx%x", rand() );
+                        irc_rootmsg( irc, "Warning: Something went wrong while trying "
+                                          "to generate a nickname for contact %s on %s.",
+                                          bu->handle, bu->ic->acc->tag );
+                        irc_rootmsg( irc, "This might be a bug in BitlBee, or the result "
+                                          "of a faulty nick_format setting. Will use %s "
+                                          "instead.", nick );
                         
                         break;
@@ -291 +276 @@
 void nick_strip( irc_t *irc, char *nick )
 {
-        int i, j;
-        
-        for( i = j = 0; nick[i] && j < MAX_NICK_LENGTH; i++ )
-        {
-                if( strchr( nick_lc_chars, nick[i] ) || 
-                    strchr( nick_uc_chars, nick[i] ) )
-                {
-                        nick[j] = nick[i];
-                        j++;
+        int len = 0;
+        
+        if( irc && ( irc->status & IRC_UTF8_NICKS ) )
+        {
+                gunichar c;
+                char *p = nick, *n, tmp[strlen(nick)+1];
+                
+                while( p && *p )
+                {
+                        c = g_utf8_get_char_validated( p, -1 );
+                        n = g_utf8_find_next_char( p, NULL );
+                        
+                        if( ( c < 0x7f && !( strchr( nick_lc_chars, c ) ||
+                                             strchr( nick_uc_chars, c ) ) ) ||
+                            !g_unichar_isgraph( c ) )
+                        {
+                                strcpy( tmp, n );
+                                strcpy( p, tmp );
+                        }
+                        else
+                                p = n;
+                }
+                if( p )
+                        len = p - nick;
+        }
+        else
+        {
+                int i;
+                
+                for( i = len = 0; nick[i] && len < MAX_NICK_LENGTH; i++ )
+                {
+                        if( strchr( nick_lc_chars, nick[i] ) || 
+                            strchr( nick_uc_chars, nick[i] ) )
+                        {
+                                nick[len] = nick[i];
+                                len++;
+                        }
                 }
         }
@@ -306 +319 @@
                 char *orig;
                 
+                /* First character of a nick can't be a digit, so insert an
+                   underscore if necessary. */
                 orig = g_strdup( nick );
                 g_snprintf( nick, MAX_NICK_LENGTH, "_%s", orig );
                 g_free( orig );
-                j ++;
-        }
-        while( j <= MAX_NICK_LENGTH )
-                nick[j++] = '\0';
-}
-
-int nick_ok( irc_t *irc, const char *nick )
+                len ++;
+        }
+        while( len <= MAX_NICK_LENGTH )
+                nick[len++] = '\0';
+}
+
+gboolean nick_ok( irc_t *irc, const char *nick )
 {
         const char *s;
@@ -321 +336 @@
         /* Empty/long nicks are not allowed, nor numbers at [0] */
         if( !*nick || isdigit( nick[0] ) || strlen( nick ) > MAX_NICK_LENGTH )
-                return( 0 );
-        
-        for( s = nick; *s; s ++ )
-                if( !strchr( nick_lc_chars, *s ) && !strchr( nick_uc_chars, *s ) )
-                        return( 0 );
-        
-        return( 1 );
+                return 0;
+        
+        if( irc && ( irc->status & IRC_UTF8_NICKS ) )
+        {
+                gunichar c;
+                const char *p = nick, *n;
+                
+                while( p && *p )
+                {
+                        c = g_utf8_get_char_validated( p, -1 );
+                        n = g_utf8_find_next_char( p, NULL );
+                        
+                        if( ( c < 0x7f && !( strchr( nick_lc_chars, c ) ||
+                                             strchr( nick_uc_chars, c ) ) ) ||
+                            !g_unichar_isgraph( c ) )
+                        {
+                                return FALSE;
+                        }
+                        p = n;
+                }
+        }
+        else
+        {
+                for( s = nick; *s; s ++ )
+                        if( !strchr( nick_lc_chars, *s ) && !strchr( nick_uc_chars, *s ) )
+                                return FALSE;
+        }
+        
+        return TRUE;
 }
 
@@ -342 +379 @@
                 }
         
+        if( irc && ( irc->status & IRC_UTF8_NICKS ) )
+        {
+                gchar *down = g_utf8_strdown( nick, -1 );
+                if( strlen( down ) > strlen( nick ) )
+                {
+                        /* Well crap. Corrupt it if we have to. */
+                        down[strlen(nick)] = '\0';
+                }
+                strcpy( nick, down );
+                g_free( down );
+        }
+        
         for( i = 0; nick[i]; i ++ )
-        {
-                if( !tab[(int)nick[i]] )
-                        return( 0 );
-                
                 nick[i] = tab[(int)nick[i]];
-        }
-        
-        return( 1 );
-}
-
-int nick_uc( irc_t *irc, char *nick )
-{
-        static char tab[128] = { 0 };
-        int i;
-        
-        if( tab['A'] == 0 )
-                for( i = 0; nick_lc_chars[i]; i ++ )
-                {
-                        tab[(int)nick_uc_chars[i]] = nick_uc_chars[i];
-                        tab[(int)nick_lc_chars[i]] = nick_uc_chars[i];
-                }
-        
-        for( i = 0; nick[i]; i ++ )
-        {
-                if( !tab[(int)nick[i]] )
-                        return( 0 );
-                
-                nick[i] = tab[(int)nick[i]];
-        }
-        
-        return( 1 );
+        
+        return nick_ok( irc, nick );
 }
 

nick.h

@@ -33 +33 @@
 
 void nick_strip( irc_t *irc, char *nick );
-int nick_ok( irc_t *irc, const char *nick );
+gboolean nick_ok( irc_t *irc, const char *nick );
 int nick_lc( irc_t *irc, char *nick );
 int nick_uc( irc_t *irc, char *nick );