Ticket #1838: enca.2.patch

src/ext.c

@@ -301 +301 @@
 #endif
 
 /*
- * Run the "file" command on the local file.
- * Return 1 if the data is valid, 0 otherwise, -1 for fatal errors.
- */
+ * Run cmd_file with args, put in buf[buflen] result.
+ * If error, put in buf[0] = '\0';
+ *
+ * NOTES: buf is string, always end '\0' char.
+ *  */
+
 static int
-get_file_type_local (const char *filename, char *buf, int buflen)
+get_popen_information(const char *cmd_file, const char *args,  char *buf, int buflen)
 {
     int read_bytes = 0;
 
-    char *tmp = name_quote (filename, 0);
-    char *command = g_strconcat (FILE_CMD, tmp, " 2>/dev/null", (char *) 0);
+    char *command = g_strconcat (cmd_file, args, " 2>/dev/null", (char *) 0);
     FILE *f = popen (command, "r");
 
-    g_free (tmp);
     g_free (command);
     if (f != NULL) {
 #ifdef __QNXNTO__
@@ -325 +326 @@
         read_bytes = (fgets (buf, buflen, f)
                       != NULL);
         if (read_bytes == 0)
+/* if(buflen > 0) // ;-) */
             buf[0] = 0;
         pclose (f);
     } else {
+/* if(buflen > 0) // ;-) */
+        buf[0] = 0; /* Paranoid termination */
         return -1;
     }
+    /* Paranoid termination */
+    buf[buflen - 1] = 0;
 
     return (read_bytes > 0);
 }
 
+/*
+ * Run the "file" command on the local file.
+ * Return 1 if the data is valid, 0 otherwise, -1 for fatal errors.
+ */
+static int
+get_file_type_local (const char *filename, char *buf, int buflen)
+{
+    int read_bytes = 0;
+    char *tmp = name_quote (filename, 0);
+
+    return get_popen_information(FILE_CMD, tmp, buf, buflen);
+}
+
+/*
+ * Run the "enca" command on the local file.
+ * Return 1 if the data is valid, 0 otherwise, -1 for fatal errors.
+ */
+static int
+get_file_encoding_local (const char *filename, char *buf, int buflen)
+{
+    if(is_autodetect_codeset_enabled == TRUE) {
+        int read_bytes = 0;
+        char *tmp = name_quote (filename, 0);
+        char *lang = name_quote (autodetect_codeset, 0);
+        char *args= g_strconcat (" -L", lang, " -i ", tmp, (char *) 0);
+        int ret = get_popen_information("enca", args, buf, buflen);
+        g_free (args);
+        g_free (lang);
+        g_free (tmp);
+        return ret;
+    }
+    return 0;
+}
+
 
 /*
  * Invoke the "file" command on the file and match its output against PTR.
@@ -348 +388 @@
 
     /* Following variables are valid if *have_type is 1 */
     static char content_string[2048];
+    static char encoding_id[21]; /* example: CSISO51INISCYRILLIC -- 20 */
+    static int got_encoding_data = 0;
     static int content_shift = 0;
     static int got_data = 0;
 
@@ -366 +408 @@
             return -1;
 
         realname = localfile;
-        got_data =
-            get_file_type_local (localfile, content_string,
-                                 sizeof (content_string));
+
+        got_encoding_data =
+            get_file_encoding_local (localfile, encoding_id,
+                sizeof (encoding_id));
         mc_ungetlocalcopy (filename, localfile, 0);
 
-        if (got_data > 0) {
+        if( got_encoding_data > 0 )
+        {
             char *pp;
+            if ((pp = strchr (encoding_id, '\n')) != 0)
+                *pp = 0;
 
-            /* Paranoid termination */
-            content_string[sizeof (content_string) - 1] = 0;
+        source_codepage = get_codepage_index( encoding_id );
+        if(source_codepage == -1)
+            source_codepage = default_source_codepage;
+        }
 
-            if ((pp = strchr (content_string, '\n')) != 0)
-                *pp = 0;
 
-            if (!strncmp (content_string, realname, strlen (realname))) {
-                /* Skip "realname: " */
-                content_shift = strlen (realname);
-                if (content_string[content_shift] == ':') {
-                    /* Solaris' file prints tab(s) after ':' */
-                    for (content_shift++;
-                         content_string[content_shift] == ' '
-                         || content_string[content_shift] == '\t';
-                         content_shift++);
-                }
-            }
-        } else {
-            /* No data */
-            content_string[0] = 0;
+  got_data =
+    get_file_type_local (localfile, content_string,
+                                 sizeof (content_string));
+
+  if (got_data > 0) {
+    char *pp;
+
+    if ((pp = strchr (content_string, '\n')) != 0)
+        *pp = 0;
+
+    if (!strncmp (content_string, realname, strlen (realname))) {
+        /* Skip "realname: " */
+        content_shift = strlen (realname);
+        if (content_string[content_shift] == ':') {
+            /* Solaris' file prints tab(s) after ':' */
+            for (content_shift++;
+                 content_string[content_shift] == ' '
+                 || content_string[content_shift] == '\t';
+                 content_shift++);
         }
+    }
+  } else {
+    /* No data */
+    content_string[0] = 0;
+  }
         g_free (realname);
     }
 

src/main.c

@@ -2257 +2257 @@
     done_key ();
 #ifdef HAVE_CHARSET
     free_codepages_list ();
+    g_free(autodetect_codeset);
 #endif
     str_uninit_strings ();
 

src/main.h

@@ -44 +44 @@
 
 #ifdef HAVE_CHARSET
 extern int source_codepage;
+extern int default_source_codepage;
 extern int display_codepage;
+extern char* autodetect_codeset;
+gboolean is_autodetect_codeset_enabled;
 #else
 extern int eight_bit_clean;
 extern int full_eight_bits;

src/selcodepage.c

@@ -40 +40 @@
 
 /* Numbers of (file I/O) and (input/display) codepages. -1 if not selected */
 int source_codepage = -1;
+int default_source_codepage = -1;
 int display_codepage = -1;
+char* autodetect_codeset = 0;
+gboolean is_autodetect_codeset_enabled=FALSE;
 
 static unsigned char
 get_hotkey (int n)
@@ -119 +122 @@
         return FALSE;
 
     source_codepage = r;
+    default_source_codepage = source_codepage;
 
     errmsg = init_translation_table (r == SELECT_CHARSET_NO_TRANSLATE ?
                                         display_codepage : source_codepage,

src/setup.c

@@ -387 +387 @@
     mc_config_set_string(mc_main_config, "Misc" , "display_codepage",
                  get_codepage_id( display_codepage ));
     mc_config_set_string(mc_main_config, "Misc" , "source_codepage",
-                 get_codepage_id( source_codepage ));
+                 get_codepage_id( default_source_codepage ));
+    mc_config_set_string(mc_main_config, "Misc" , "autodetect_codeset",
+                 autodetect_codeset );
 #endif /* HAVE_CHARSET */
     tmp_profile = g_build_filename (home_dir, MC_USERCONF_DIR, MC_CONFIG_FILE, NULL);
     ret = mc_config_save_to_file (mc_main_config, tmp_profile, NULL);
@@ -813 +815 @@
         buffer = mc_config_get_string(mc_main_config, "Misc", "source_codepage", "");
         if ( buffer[0] != '\0' )
         {
-            source_codepage = get_codepage_index( buffer );
+            default_source_codepage = get_codepage_index( buffer );
+            source_codepage = default_source_codepage; /* Mabye source_codepage don't needed this */
             cp_source = get_codepage_id (source_codepage);
         }
         g_free(buffer);
-    }
+   }
+
+  autodetect_codeset = mc_config_get_string(mc_main_config, "Misc", "autodetect_codeset", "");
+        if ( (autodetect_codeset[0] != '\0') && ( strcmp(autodetect_codeset, "off") ) )
+      is_autodetect_codeset_enabled=TRUE;
+
     init_translation_table( source_codepage, display_codepage );
     if ( get_codepage_id( display_codepage ) )
         utf8_display = str_isutf8 (get_codepage_id( display_codepage ));

src/viewer/display.c

@@ -174 +174 @@
                 tty_printf (_(">= %s bytes"), size_trunc (filesize));
             }
         }
+        if (width > 62 + 11){
+            char* cp = get_codepage_id (source_codepage);
+            if(cp)
+                tty_printf ("    %s", cp);
+        }
         if (width > 26) {
             mcview_percent (view, view->hex_mode ? view->hex_cursor : view->dpy_end);
         }