Add option aspell.check.real_time (off by default), fix bugs with aspell and utf-8 chars, add function utf8_char_int to API

10 files changed, 131 insertions, 92 deletions

diff --git a/src/core/wee-utf8.c b/src/core/wee-utf8.c
index 7c5b5cf06..bae3693fe 100644
--- a/src/core/wee-utf8.c
+++ b/src/core/wee-utf8.c
@@ -226,6 +226,61 @@ utf8_next_char (const char *string)
 }
 
 /*
+ * utf8_char_int: return UTF-8 char as integer
+ */
+
+int
+utf8_char_int (const char *string)
+{
+    const unsigned char *ptr_string;
+    
+    if (!string)
+        return 0;
+    
+    ptr_string = (unsigned char *)string;
+    
+    /* UTF-8, 2 bytes: 110vvvvv 10vvvvvv */
+    if ((ptr_string[0] & 0xE0) == 0xC0)
+    {
+        if (!ptr_string[1])
+            return (int)(ptr_string[0] & 0x1F);
+        return ((int)(ptr_string[0] & 0x1F) << 6) +
+            ((int)(ptr_string[1] & 0x3F));
+    }
+    /* UTF-8, 3 bytes: 1110vvvv 10vvvvvv 10vvvvvv */
+    else if ((ptr_string[0] & 0xF0) == 0xE0)
+    {
+        if (!ptr_string[1])
+            return (int)(ptr_string[0] & 0x0F);
+        if (!ptr_string[2])
+            return (((int)(ptr_string[0] & 0x0F)) << 6) +
+                ((int)(ptr_string[1] & 0x3F));
+        return (((int)(ptr_string[0] & 0x0F)) << 12) +
+            (((int)(ptr_string[1] & 0x3F)) << 6) +
+            ((int)(ptr_string[2] & 0x3F));
+    }
+    /* UTF-8, 4 bytes: 11110vvv 10vvvvvv 10vvvvvv 10vvvvvv */
+    else if ((ptr_string[0] & 0xF8) == 0xF0)
+    {
+        if (!ptr_string[1])
+            return (int)ptr_string[0] & 0x07;
+        if (!ptr_string[2])
+            return (((int)(ptr_string[0] & 0x07)) << 6) +
+                ((int)(ptr_string[1] & 0x3F));
+        if (!ptr_string[3])
+            return (((int)(ptr_string[0] & 0x07)) << 12) +
+                (((int)(ptr_string[1] & 0x3F)) << 6) +
+                ((int)(ptr_string[2] & 0x3F));
+        return (((int)(ptr_string[0] & 0x07)) << 18) +
+            (((int)(ptr_string[1] & 0x3F)) << 12) +
+            (((int)(ptr_string[2] & 0x3F)) << 6) +
+            ((int)(ptr_string[3] & 0x3F));
+    }
+    /* UTF-8, 1 byte: 0vvvvvvv */
+    return (int)ptr_string[0];
+}
+
+/*
  * utf8_char_size: return UTF-8 char size (in bytes)
  */
 
diff --git a/src/core/wee-utf8.h b/src/core/wee-utf8.h
index 60511e607..d6fc8f927 100644
--- a/src/core/wee-utf8.h
+++ b/src/core/wee-utf8.h
@@ -38,6 +38,7 @@ extern int utf8_is_valid (const char *string, char **error);
 extern void utf8_normalize (const char *string, char replacement);
 extern char *utf8_prev_char (const char *string_start, const char *string);
 extern char *utf8_next_char (const char *string);
+extern int utf8_char_int (const char *string);
 extern int utf8_char_size (const char *string);
 extern int utf8_strlen (const char *string);
 extern int utf8_strnlen (const char *string, int bytes);
diff --git a/src/plugins/aspell/weechat-aspell-config.c b/src/plugins/aspell/weechat-aspell-config.c
index d61d96bf5..43032b564 100644
--- a/src/plugins/aspell/weechat-aspell-config.c
+++ b/src/plugins/aspell/weechat-aspell-config.c
@@ -16,7 +16,7 @@
  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
-/* aspell-config.c: aspell configuration options */
+/* weechat-aspell-config.c: aspell configuration options */
 
 
 #include <stdlib.h>
@@ -40,8 +40,9 @@ struct t_config_option *weechat_aspell_config_look_color;
 
 struct t_config_option *weechat_aspell_config_check_commands;
 struct t_config_option *weechat_aspell_config_check_default_dict;
-struct t_config_option *weechat_aspell_config_check_word_min_length;
 struct t_config_option *weechat_aspell_config_check_during_search;
+struct t_config_option *weechat_aspell_config_check_real_time;
+struct t_config_option *weechat_aspell_config_check_word_min_length;
 
 
 char **weechat_aspell_commands_to_check = NULL;
@@ -304,17 +305,23 @@ weechat_aspell_config_init ()
            "aspell on buffers for which you didn't explicitely enabled it)"),
         NULL, 0, 0, "", NULL, 0,
         NULL, NULL, &weechat_aspell_config_change_default_dict, NULL, NULL, NULL);
+    weechat_aspell_config_check_during_search = weechat_config_new_option (
+        weechat_aspell_config_file, ptr_section,
+        "during_search", "boolean",
+        N_("check words during text search in buffer"),
+        NULL, 0, 0, "off", NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL);
+    weechat_aspell_config_check_real_time = weechat_config_new_option (
+        weechat_aspell_config_file, ptr_section,
+        "real_time", "boolean",
+        N_("real-time spell checking of words (slower, disabled by default: "
+           "words are checked only if there's delimiter after)"),
+        NULL, 0, 0, "off", NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL);
     weechat_aspell_config_check_word_min_length = weechat_config_new_option (
         weechat_aspell_config_file, ptr_section,
         "word_min_length", "integer",
         N_("minimum length for a word to be spell checked (use 0 to check all "
            "words)"),
         NULL, 0, INT_MAX, "2", NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL);
-    weechat_aspell_config_check_during_search = weechat_config_new_option (
-        weechat_aspell_config_file, ptr_section,
-        "during_search", "boolean",
-        N_("check words during text search in buffer"),
-        NULL, 0, 0, "off", NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL);
     
     /* dict */
     ptr_section = weechat_config_new_section (weechat_aspell_config_file, "dict",
diff --git a/src/plugins/aspell/weechat-aspell-config.h b/src/plugins/aspell/weechat-aspell-config.h
index efb4590ab..6f8fc3ab2 100644
--- a/src/plugins/aspell/weechat-aspell-config.h
+++ b/src/plugins/aspell/weechat-aspell-config.h
@@ -27,8 +27,9 @@ extern struct t_config_option *weechat_aspell_config_look_color;
 
 extern struct t_config_option *weechat_aspell_config_check_commands;
 extern struct t_config_option *weechat_aspell_config_check_default_dict;
-extern struct t_config_option *weechat_aspell_config_check_word_min_length;
 extern struct t_config_option *weechat_aspell_config_check_during_search;
+extern struct t_config_option *weechat_aspell_config_check_real_time;
+extern struct t_config_option *weechat_aspell_config_check_word_min_length;
 
 extern char **weechat_aspell_commands_to_check;
 extern int weechat_aspell_count_commands_to_check;
@@ -41,4 +42,4 @@ extern int weechat_aspell_config_read ();
 extern int weechat_aspell_config_write ();
 extern void weechat_aspell_config_free ();
 
-#endif /* aspell-config.h */
+#endif /* weechat-aspell-config.h */
diff --git a/src/plugins/aspell/weechat-aspell-speller.c b/src/plugins/aspell/weechat-aspell-speller.c
index 44f19aed7..2dd7e2a0c 100644
--- a/src/plugins/aspell/weechat-aspell-speller.c
+++ b/src/plugins/aspell/weechat-aspell-speller.c
@@ -16,7 +16,7 @@
  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
-/* aspell-speller.c: speller management for aspell plugin */
+/* weechat-aspell-speller.c: speller management for aspell plugin */
 
 
 #include <stdlib.h>
diff --git a/src/plugins/aspell/weechat-aspell-speller.h b/src/plugins/aspell/weechat-aspell-speller.h
index 8e12cfdd4..646640902 100644
--- a/src/plugins/aspell/weechat-aspell-speller.h
+++ b/src/plugins/aspell/weechat-aspell-speller.h
@@ -38,4 +38,4 @@ extern struct t_aspell_speller *weechat_aspell_speller_new (const char *lang);
 extern void weechat_aspell_speller_free (struct t_aspell_speller *speller);
 extern void weechat_aspell_speller_free_all ();
 
-#endif /* aspell-speller.h */
+#endif /* weechat-aspell-speller.h */
diff --git a/src/plugins/aspell/weechat-aspell.c b/src/plugins/aspell/weechat-aspell.c
index dc6f1a2f9..18f961e1f 100644
--- a/src/plugins/aspell/weechat-aspell.c
+++ b/src/plugins/aspell/weechat-aspell.c
@@ -16,14 +16,14 @@
  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
-/* aspell.c: aspell plugin for WeeChat */
+/* weechat-aspell.c: aspell plugin for WeeChat */
 
 
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 #include <sys/time.h>
-#include <ctype.h>
+#include <wctype.h>
 
 #include "../weechat-plugin.h"
 #include "weechat-aspell.h"
@@ -551,45 +551,6 @@ weechat_aspell_command_authorized (const char *command)
 }
 
 /*
- * weechat_aspell_string_strip_punctuation: strip punctuation chars at the
- *                                          begining and at the end of a word
- */
-
-char *
-weechat_aspell_string_strip_punctuation (const char *word)
-{
-    const char *ptr_start, *ptr_end;
-    
-    if (!word)
-        return NULL;
-    
-    ptr_start = word;
-    while (ptr_start[0])
-    {
-        if (!ispunct (ptr_start[0]))
-            break;
-        ptr_start++;
-    }
-
-    if (!ptr_start[0])
-        return strdup ("");
-    
-    ptr_end = ptr_start + strlen (ptr_start) - 1;
-    
-    while (ptr_end >= ptr_start)
-    {
-        if (!ispunct (ptr_end[0]))
-            break;
-        ptr_end--;
-    }
-    
-    if (ptr_end < ptr_start)
-        return strdup ("");
-    
-    return weechat_strndup (ptr_start, ptr_end - ptr_start + 1);
-}
-
-/*
  * weechat_aspell_string_is_url: detect if a word is an url
  */
 
@@ -629,17 +590,17 @@ weechat_aspell_string_is_url (const char *word)
 int
 weechat_aspell_string_is_simili_number (const char *word)
 {
-    const char *ptr_word;
+    int utf8_char_int;
     
-    if (!word)
+    if (!word || !word[0])
         return 0;
     
-    ptr_word = word;
-    while (ptr_word[0])
+    while (word && word[0])
     {
-        if (!ispunct (ptr_word[0]) && !isdigit (ptr_word[0]))
+        utf8_char_int = weechat_utf8_char_int (word);
+        if (!iswpunct (utf8_char_int) && !iswdigit (utf8_char_int))
             return 0;
-        ptr_word++;
+        word = weechat_utf8_next_char (word);
     }
     
     /* there's only digit or punctuation */
@@ -654,35 +615,29 @@ weechat_aspell_string_is_simili_number (const char *word)
 int
 weechat_aspell_check_word (struct t_gui_buffer *buffer, const char *word)
 {
-    char *clean_word;
     struct t_aspell_speller *ptr_speller;
     int rc;
     
-    clean_word = weechat_aspell_string_strip_punctuation (word);
-    
-    if (!clean_word)
-        return 1;
-    
     rc = 0;
     
     /* word too small? then do not check word */
     if ((weechat_config_integer (weechat_aspell_config_check_word_min_length) > 0)
-        && ((int)strlen (clean_word) < weechat_config_integer (weechat_aspell_config_check_word_min_length)))
+        && ((int)strlen (word) < weechat_config_integer (weechat_aspell_config_check_word_min_length)))
         rc = 1;
     else
     {
         /* word is URL? then do not check word */
-        if (weechat_aspell_string_is_url (clean_word))
+        if (weechat_aspell_string_is_url (word))
             rc = 1;
         else
         {
             /* word is a number? then do not check word */
-            if (weechat_aspell_string_is_simili_number (clean_word))
+            if (weechat_aspell_string_is_simili_number (word))
                 rc = 1;
             else
             {
                 /* word is a nick of nicklist on this buffer? then do not check word */
-                if (weechat_nicklist_search_nick (buffer, NULL, clean_word))
+                if (weechat_nicklist_search_nick (buffer, NULL, word))
                     rc = 1;
                 else
                 {
@@ -690,7 +645,7 @@ weechat_aspell_check_word (struct t_gui_buffer *buffer, const char *word)
                     for (ptr_speller = weechat_aspell_spellers; ptr_speller;
                          ptr_speller = ptr_speller->next_speller)
                     {
-                        if (aspell_speller_check (ptr_speller->speller, clean_word, -1) == 1)
+                        if (aspell_speller_check (ptr_speller->speller, word, -1) == 1)
                         {
                             rc = 1;
                             break;
@@ -701,8 +656,6 @@ weechat_aspell_check_word (struct t_gui_buffer *buffer, const char *word)
         }
     }
     
-    free (clean_word);
-    
     return rc;
 }
 
@@ -716,8 +669,9 @@ weechat_aspell_modifier_cb (void *data, const char *modifier,
 {
     long unsigned int value;
     struct t_gui_buffer *buffer;
-    char *result, *ptr_string, *pos_space;
+    char *result, *ptr_string, *pos_space, *ptr_end, save_end;
     const char *color_normal, *color_error;
+    int utf8_char_int, char_size;
     int length, index_result, length_word, word_ok;
     int length_color_normal, length_color_error;
     
@@ -784,13 +738,17 @@ weechat_aspell_modifier_cb (void *data, const char *modifier,
             && (ptr_string[1] != ' '))
         {
             ptr_string++;
-            pos_space = strchr (ptr_string, ' ');
-            if (!pos_space)
+            pos_space = ptr_string;
+            while (pos_space && pos_space[0] && (pos_space[0] != ' '))
+            {
+                pos_space = weechat_utf8_next_char (pos_space);
+            }
+            if (!pos_space || !pos_space[0])
             {
                 free (result);
                 return NULL;
             }
-
+            
             pos_space[0] = '\0';
             
             /* exit if command is not authorized for spell checking */
@@ -809,24 +767,39 @@ weechat_aspell_modifier_cb (void *data, const char *modifier,
         
         while (ptr_string[0])
         {
-            while (ptr_string[0] == ' ')
+            /* find start of word */
+            utf8_char_int = weechat_utf8_char_int (ptr_string);
+            while (!iswalnum (utf8_char_int) || iswspace (utf8_char_int))
             {
-                result[index_result++] = ' ';
-                ptr_string++;
+                char_size = weechat_utf8_char_size (ptr_string);
+                memcpy (result + index_result, ptr_string, char_size);
+                index_result += char_size;
+                ptr_string += char_size;
+                if (!ptr_string[0])
+                    break;
+                utf8_char_int = weechat_utf8_char_int (ptr_string);
             }
             if (!ptr_string[0])
                 break;
             
-            pos_space = strchr (ptr_string, ' ');
-            if (pos_space)
+            ptr_end = weechat_utf8_next_char (ptr_string);
+            utf8_char_int = weechat_utf8_char_int (ptr_end);
+            while (iswalnum (utf8_char_int))
             {
-                pos_space[0] = '\0';
-                length_word = pos_space - ptr_string;
+                ptr_end = weechat_utf8_next_char (ptr_end);
+                if (!ptr_end[0])
+                    break;
+                utf8_char_int = weechat_utf8_char_int (ptr_end);
             }
-            else
-                length_word = strlen (ptr_string);
+            save_end = ptr_end[0];
+            ptr_end[0] = '\0';
+            length_word = ptr_end - ptr_string;
             
-            word_ok = weechat_aspell_check_word (buffer, ptr_string);
+            if ((save_end != '\0')
+                || (weechat_config_integer (weechat_aspell_config_check_real_time)))
+                word_ok = weechat_aspell_check_word (buffer, ptr_string);
+            else
+                word_ok = 1;
             
             /* add error color */
             if (!word_ok)
@@ -846,13 +819,11 @@ weechat_aspell_modifier_cb (void *data, const char *modifier,
                 index_result += length_color_normal;
             }
             
-            if (pos_space)
-            {
-                pos_space[0] = ' ';
-                ptr_string = pos_space;
-            }
-            else
+            if (save_end == '\0')
                 break;
+
+            ptr_end[0] = save_end;
+            ptr_string = ptr_end;
         }
         
         result[index_result] = '\0';
diff --git a/src/plugins/aspell/weechat-aspell.h b/src/plugins/aspell/weechat-aspell.h
index 13728ed47..763652b64 100644
--- a/src/plugins/aspell/weechat-aspell.h
+++ b/src/plugins/aspell/weechat-aspell.h
@@ -38,4 +38,4 @@ extern struct t_aspell_code countries_avail[];
 
 extern void weechat_aspell_create_spellers (struct t_gui_buffer *buffer);
 
-#endif /* aspell.h */
+#endif /* weechat-aspell.h */
diff --git a/src/plugins/plugin.c b/src/plugins/plugin.c
index 40f6a8f76..b373c5f39 100644
--- a/src/plugins/plugin.c
+++ b/src/plugins/plugin.c
@@ -420,6 +420,7 @@ plugin_load (const char *filename)
         new_plugin->utf8_normalize = &utf8_normalize;
         new_plugin->utf8_prev_char = &utf8_prev_char;
         new_plugin->utf8_next_char = &utf8_next_char;
+        new_plugin->utf8_char_int = &utf8_char_int;
         new_plugin->utf8_char_size = &utf8_char_size;
         new_plugin->utf8_strlen = &utf8_strlen;
         new_plugin->utf8_strnlen = &utf8_strnlen;
diff --git a/src/plugins/weechat-plugin.h b/src/plugins/weechat-plugin.h
index 96c1e79d7..880fd45bb 100644
--- a/src/plugins/weechat-plugin.h
+++ b/src/plugins/weechat-plugin.h
@@ -175,6 +175,7 @@ struct t_weechat_plugin
     void (*utf8_normalize) (const char *string, char replacement);
     char *(*utf8_prev_char) (const char *string_start, const char *string);
     char *(*utf8_next_char) (const char *string);
+    int (*utf8_char_int) (const char *string);
     int (*utf8_char_size) (const char *string);
     int (*utf8_strlen) (const char *string);
     int (*utf8_strnlen) (const char *string, int bytes);
@@ -708,6 +709,8 @@ extern int weechat_plugin_end (struct t_weechat_plugin *plugin);
     weechat_plugin->utf8_prev_char(__start, __string)
 #define weechat_utf8_next_char(__string)                                \
     weechat_plugin->utf8_next_char(__string)
+#define weechat_utf8_char_int(__string)                                 \
+    weechat_plugin->utf8_char_int(__string)
 #define weechat_utf8_char_size(__string)                                \
     weechat_plugin->utf8_char_size(__string)
 #define weechat_utf8_strlen(__string)                                   \