Added UTF-8 functions to plugins API

7 files changed, 331 insertions, 46 deletions

diff --git a/src/core/wee-utf8.c b/src/core/wee-utf8.c
index 085e6e6dc..9e52dd6b3 100644
--- a/src/core/wee-utf8.c
+++ b/src/core/wee-utf8.c
@@ -226,7 +226,7 @@ utf8_next_char (char *string)
 }
 
 /*
- * utf8_char_size: return UTF-8 char size
+ * utf8_char_size: return UTF-8 char size (in bytes)
  */
 
 int
@@ -455,43 +455,3 @@ utf8_pos (char *string, int real_pos)
     }
     return count;
 }
-
-/*
- * utf8_get_wide_char: get wide char from string (first char)
- */
-
-wint_t
-utf8_get_wide_char (char *string)
-{
-    int char_size;
-    wint_t result;
-    
-    if (!string || !string[0])
-        return WEOF;
-    
-    char_size = utf8_char_size (string);
-    switch (char_size)
-    {
-        case 1:
-            result = (wint_t)string[0];
-            break;
-        case 2:
-            result = ((wint_t)((unsigned char)string[0])) << 8
-                |  ((wint_t)((unsigned char)string[1]));
-            break;
-        case 3:
-            result = ((wint_t)((unsigned char)string[0])) << 16
-                |  ((wint_t)((unsigned char)string[1])) << 8
-                |  ((wint_t)((unsigned char)string[2]));
-            break;
-        case 4:
-            result = ((wint_t)((unsigned char)string[0])) << 24
-                |  ((wint_t)((unsigned char)string[1])) << 16
-                |  ((wint_t)((unsigned char)string[2])) << 8
-                |  ((wint_t)((unsigned char)string[3]));
-            break;
-        default:
-            result = WEOF;
-    }
-    return result;
-}
diff --git a/src/core/wee-utf8.h b/src/core/wee-utf8.h
index b66996864..f16733535 100644
--- a/src/core/wee-utf8.h
+++ b/src/core/wee-utf8.h
@@ -47,6 +47,5 @@ extern int utf8_char_size_screen (char *);
 extern char *utf8_add_offset (char *, int);
 extern int utf8_real_pos (char *, int);
 extern int utf8_pos (char *, int);
-extern wint_t utf8_get_wide_char (char *);
 
 #endif /* wee-utf8.h */
diff --git a/src/plugins/irc/irc-protocol.c b/src/plugins/irc/irc-protocol.c
index a9efb3245..f69a6c0b1 100644
--- a/src/plugins/irc/irc-protocol.c
+++ b/src/plugins/irc/irc-protocol.c
@@ -24,6 +24,16 @@
 #include "config.h"
 #endif
 
+#ifndef __USE_XOPEN
+#define __USE_XOPEN
+#endif
+
+#if defined(__OpenBSD__)
+#include <utf8/wchar.h>
+#else
+#include <wchar.h>
+#endif
+
 #include <stdlib.h>
 #include <string.h>
 #include <ctype.h>
@@ -161,13 +171,53 @@ char *irc_message = NULL;
 
 
 /*
+ * irc_protocol_get_wide_char: get wide char from string (first char)
+ */
+
+wint_t
+irc_protocol_get_wide_char (char *string)
+{
+    int char_size;
+    wint_t result;
+    
+    if (!string || !string[0])
+        return WEOF;
+    
+    char_size = weechat_utf8_char_size (string);
+    switch (char_size)
+    {
+        case 1:
+            result = (wint_t)string[0];
+            break;
+        case 2:
+            result = ((wint_t)((unsigned char)string[0])) << 8
+                |  ((wint_t)((unsigned char)string[1]));
+            break;
+        case 3:
+            result = ((wint_t)((unsigned char)string[0])) << 16
+                |  ((wint_t)((unsigned char)string[1])) << 8
+                |  ((wint_t)((unsigned char)string[2]));
+            break;
+        case 4:
+            result = ((wint_t)((unsigned char)string[0])) << 24
+                |  ((wint_t)((unsigned char)string[1])) << 16
+                |  ((wint_t)((unsigned char)string[2])) << 8
+                |  ((wint_t)((unsigned char)string[3]));
+            break;
+        default:
+            result = WEOF;
+    }
+    return result;
+}
+
+/*
  * irc_protocol_is_word_char: return 1 if given character is a "word character"
  */
 
 int
 irc_protocol_is_word_char (char *str)
 {
-    wint_t c = utf8_get_wide_char (str);
+    wint_t c = irc_protocol_get_wide_char (str);
 
     if (c == WEOF)
         return 0;
@@ -222,7 +272,7 @@ irc_protocol_is_highlight (char *message, char *nick)
     match = strstr (message, nick);
     if (match)
     {
-        match_pre = utf8_prev_char (message, match);
+        match_pre = weechat_utf8_prev_char (message, match);
         if (!match_pre)
             match_pre = match - 1;
         match_post = match + strlen(nick);
diff --git a/src/plugins/plugin-api.c b/src/plugins/plugin-api.c
index 5b14f0efe..85d87b952 100644
--- a/src/plugins/plugin-api.c
+++ b/src/plugins/plugin-api.c
@@ -251,6 +251,205 @@ plugin_api_string_free_splitted_command (struct t_weechat_plugin *plugin,
 }
 
 /*
+ * plugin_api_utf8_has_8bits: return 1 if string has 8-bits chars, 0 if only
+ *                            7-bits chars
+ */
+
+int
+plugin_api_utf8_has_8bits (struct t_weechat_plugin *plugin, char *string)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_has_8bits (string);
+}
+
+/*
+ * plugin_api_utf8_is_valid: return 1 if UTF-8 string is valid, 0 otherwise
+ *                           if error is not NULL, it's set with first non
+ *                           valid UTF-8 char in string, if any
+ */
+
+int
+plugin_api_utf8_is_valid (struct t_weechat_plugin *plugin, char *string,
+                          char **error)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_is_valid (string, error);
+}
+
+/*
+ * plugin_api_utf8_normalize: normalize UTF-8 string: remove non UTF-8 chars
+ *                            and replace them by a char
+ */
+
+void
+plugin_api_utf8_normalize (struct t_weechat_plugin *plugin, char *string,
+                           char replacement)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    utf8_normalize (string, replacement);
+}
+
+/*
+ * plugin_api_utf8_prev_char: return previous UTF-8 char in a string
+ */
+
+char *
+plugin_api_utf8_prev_char (struct t_weechat_plugin *plugin, char *string_start,
+                           char *string)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_prev_char (string_start, string);
+}
+
+/*
+ * plugin_api_utf8_next_char: return next UTF-8 char in a string
+ */
+
+char *
+plugin_api_utf8_next_char (struct t_weechat_plugin *plugin, char *string)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_next_char (string);
+}
+
+/*
+ * plugin_api_utf8_char_size: return UTF-8 char size (in bytes)
+ */
+
+int
+plugin_api_utf8_char_size (struct t_weechat_plugin *plugin, char *string)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_char_size (string);
+}
+
+/*
+ * plugin_api_utf8_strlen: return length of an UTF-8 string (<= strlen(string))
+ */
+
+int
+plugin_api_utf8_strlen (struct t_weechat_plugin *plugin, char *string)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_strlen (string);
+}
+
+/*
+ * plugin_api_utf8_strnlen: return length of an UTF-8 string, for N bytes max
+ *                          in string
+ */
+
+int
+plugin_api_utf8_strnlen (struct t_weechat_plugin *plugin, char *string,
+                         int bytes)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_strnlen (string, bytes);
+}
+
+/*
+ * plugin_api_utf8_strlen_screen: return number of chars needed on screen to
+ *                                display UTF-8 string
+ */
+
+int
+plugin_api_utf8_strlen_screen (struct t_weechat_plugin *plugin, char *string)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_strlen_screen (string);
+}
+
+/*
+ * plugin_api_utf8_charcasecmp: compare two utf8 chars (case is ignored)
+ */
+
+int
+plugin_api_utf8_charcasecmp (struct t_weechat_plugin *plugin, char *string1,
+                             char *string2)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_charcasecmp (string1, string2);
+}
+
+/*
+ * plugin_api_utf8_char_size_screen: return number of chars needed on screen
+ *                                   to display UTF-8 char
+ */
+
+int
+plugin_api_utf8_char_size_screen (struct t_weechat_plugin *plugin, char *string)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_char_size_screen (string);
+}
+
+/*
+ * plugin_api_utf8_add_offset: moves forward N chars in an UTF-8 string
+ */
+
+char *
+plugin_api_utf8_add_offset (struct t_weechat_plugin *plugin, char *string,
+                            int offset)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_add_offset (string, offset);
+}
+
+/*
+ * plugin_api_utf8_real_pos: get real position in UTF-8
+ *                           for example: ("aébc", 2) returns 3
+ */
+
+int
+plugin_api_utf8_real_pos (struct t_weechat_plugin *plugin, char *string,
+                          int pos)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_real_pos (string, pos);
+}
+
+/*
+ * plugin_api_utf8_pos: get position in UTF-8
+ *                      for example: ("aébc", 3) returns 2
+ */
+
+int
+plugin_api_utf8_pos (struct t_weechat_plugin *plugin, char *string,
+                     int real_pos)
+{
+    /* make C compiler happy */
+    (void) plugin;
+    
+    return utf8_real_pos (string, real_pos);
+}
+
+/*
  * plugin_api_mkdir_home: create a directory in WeeChat home
  */
 
diff --git a/src/plugins/plugin-api.h b/src/plugins/plugin-api.h
index 2f7505e55..1f7a227eb 100644
--- a/src/plugins/plugin-api.h
+++ b/src/plugins/plugin-api.h
@@ -44,6 +44,22 @@ extern char **plugin_api_string_split_command (struct t_weechat_plugin *,
 extern void plugin_api_string_free_splitted_command (struct t_weechat_plugin *,
                                                      char **);
 
+/* UTF-8 strings */
+extern int plugin_api_utf8_has_8bits (struct t_weechat_plugin *, char *);
+extern int plugin_api_utf8_is_valid (struct t_weechat_plugin *, char *, char **);
+extern void plugin_api_utf8_normalize (struct t_weechat_plugin *, char *, char);
+extern char *plugin_api_utf8_prev_char (struct t_weechat_plugin *, char *, char *);
+extern char *plugin_api_utf8_next_char (struct t_weechat_plugin *, char *);
+extern int plugin_api_utf8_char_size (struct t_weechat_plugin *, char *);
+extern int plugin_api_utf8_strlen (struct t_weechat_plugin *, char *);
+extern int plugin_api_utf8_strnlen (struct t_weechat_plugin *, char *, int);
+extern int plugin_api_utf8_strlen_screen (struct t_weechat_plugin *, char *);
+extern int plugin_api_utf8_charcasecmp (struct t_weechat_plugin *, char *, char *);
+extern int plugin_api_utf8_char_size_screen (struct t_weechat_plugin *, char *);
+extern char *plugin_api_utf8_add_offset (struct t_weechat_plugin *, char *, int);
+extern int plugin_api_utf8_real_pos (struct t_weechat_plugin *, char *, int);
+extern int plugin_api_utf8_pos (struct t_weechat_plugin *, char *, int);
+
 /* directories */
 extern int plugin_api_mkdir_home (struct t_weechat_plugin *, char *, int);
 extern int plugin_api_mkdir (struct t_weechat_plugin *, char *, int);
diff --git a/src/plugins/plugin.c b/src/plugins/plugin.c
index 034d18e78..b085195e5 100644
--- a/src/plugins/plugin.c
+++ b/src/plugins/plugin.c
@@ -236,6 +236,21 @@ plugin_load (char *filename)
         new_plugin->string_free_exploded = &plugin_api_string_free_exploded;
         new_plugin->string_split_command = &plugin_api_string_split_command;
         new_plugin->string_free_splitted_command = &plugin_api_string_free_splitted_command;
+
+        new_plugin->utf8_has_8bits = &plugin_api_utf8_has_8bits;
+        new_plugin->utf8_is_valid = &plugin_api_utf8_is_valid;
+        new_plugin->utf8_normalize = &plugin_api_utf8_normalize;
+        new_plugin->utf8_prev_char = &plugin_api_utf8_prev_char;
+        new_plugin->utf8_next_char = &plugin_api_utf8_next_char;
+        new_plugin->utf8_char_size = &plugin_api_utf8_char_size;
+        new_plugin->utf8_strlen = &plugin_api_utf8_strlen;
+        new_plugin->utf8_strnlen = &plugin_api_utf8_strnlen;
+        new_plugin->utf8_strlen_screen = &plugin_api_utf8_strlen_screen;
+        new_plugin->utf8_charcasecmp = &plugin_api_utf8_charcasecmp;
+        new_plugin->utf8_char_size_screen = &plugin_api_utf8_char_size_screen;
+        new_plugin->utf8_add_offset = &plugin_api_utf8_add_offset;
+        new_plugin->utf8_real_pos = &plugin_api_utf8_real_pos;
+        new_plugin->utf8_pos = &plugin_api_utf8_pos;
         
         new_plugin->mkdir_home = &plugin_api_mkdir_home;
         new_plugin->mkdir = &plugin_api_mkdir;
diff --git a/src/plugins/weechat-plugin.h b/src/plugins/weechat-plugin.h
index bed73a4f6..ae43acfde 100644
--- a/src/plugins/weechat-plugin.h
+++ b/src/plugins/weechat-plugin.h
@@ -71,15 +71,31 @@ struct t_weechat_plugin
     char **(*string_split_command) (struct t_weechat_plugin *, char *, char);
     void (*string_free_splitted_command) (struct t_weechat_plugin *, char **);
     
+    /* UTF-8 strings */
+    int (*utf8_has_8bits) (struct t_weechat_plugin *, char *);
+    int (*utf8_is_valid) (struct t_weechat_plugin *, char *, char **);
+    void (*utf8_normalize) (struct t_weechat_plugin *, char *, char);
+    char *(*utf8_prev_char) (struct t_weechat_plugin *, char *, char *);
+    char *(*utf8_next_char) (struct t_weechat_plugin *, char *);
+    int (*utf8_char_size) (struct t_weechat_plugin *, char *);
+    int (*utf8_strlen) (struct t_weechat_plugin *, char *);
+    int (*utf8_strnlen) (struct t_weechat_plugin *, char *, int);
+    int (*utf8_strlen_screen) (struct t_weechat_plugin *, char *);
+    int (*utf8_charcasecmp) (struct t_weechat_plugin *, char *, char *);
+    int (*utf8_char_size_screen) (struct t_weechat_plugin *, char *);
+    char *(*utf8_add_offset) (struct t_weechat_plugin *, char *, int);
+    int (*utf8_real_pos) (struct t_weechat_plugin *, char *, int);
+    int (*utf8_pos) (struct t_weechat_plugin *, char *, int);
+    
     /* directories */
     int (*mkdir_home) (struct t_weechat_plugin *, char *, int);
     int (*mkdir) (struct t_weechat_plugin *, char *, int);
     void (*exec_on_files) (struct t_weechat_plugin *, char *,
                            int (*)(char *));
-
+    
     /* util */
     long (*timeval_diff) (struct t_weechat_plugin *, void *, void *);
-
+    
     /* sorted list */
     struct t_weelist *(*list_new) (struct t_weechat_plugin *);
     char *(*list_add) (struct t_weechat_plugin *, void *, char *,
@@ -240,6 +256,36 @@ struct t_weechat_plugin
     weechat_plugin->string_free_splitted_command(weechat_plugin,        \
                                                  __array_str)
 
+/* UTF-8 strings */
+#define weechat_utf8_has_8bits(__string)                        \
+    weechat_plugin->utf8_has_8bits(weechat_plugin, __string)
+#define weechat_utf8_is_valid(__string, __error)                        \
+    weechat_plugin->utf8_is_valid(weechat_plugin, __string, __error)
+#define weechat_utf8_normalise(__string, __char)                        \
+    weechat_plugin->utf8_normalize(weechat_plugin, __string, __char)
+#define weechat_utf8_prev_char(__start, __string)                       \
+    weechat_plugin->utf8_has_8bits(weechat_plugin, __start, __string)
+#define weechat_utf8_next_char(__string)                        \
+    weechat_plugin->utf8_next_char(weechat_plugin, __string)
+#define weechat_utf8_char_size(__string)                        \
+    weechat_plugin->utf8_char_size(weechat_plugin, __string)
+#define weechat_utf8_strlen(__string)                           \
+    weechat_plugin->utf8_strlen(weechat_plugin, __string)
+#define weechat_utf8_strnlen(__string, __bytes)                         \
+    weechat_plugin->utf8_strnlen(weechat_plugin, __string, __bytes)
+#define weechat_utf8_strlen_screen(__string)                            \
+    weechat_plugin->utf8_strlen_screen(weechat_plugin, __string)
+#define weechat_utf8_charcasecmp(__string)                      \
+    weechat_plugin->utf8_charcasecmp(weechat_plugin, __string)
+#define weechat_utf8_char_size_screen(__string)                         \
+    weechat_plugin->utf8_char_size_screen(weechat_plugin, __string)
+#define weechat_utf8_add_offset(__string, __offset)                     \
+    weechat_plugin->utf8_add_offset(weechat_plugin, __string, __offset)
+#define weechat_utf8_real_pos(__string, __pos)                          \
+    weechat_plugin->utf8_real_pos(weechat_plugin, __string, __pos)
+#define weechat_utf8_pos(__string, __real_pos)                          \
+    weechat_plugin->utf8_pos(weechat_plugin, __string, __real_pos)
+
 /* directories */
 #define weechat_mkdir_home(__directory, __mode)                         \
     weechat_plugin->mkdir_home(weechat_plugin, __directory, __mode)