Line data    Source code

       1              : /* vim: set sw=8: -*- Mode: C; tab-width: 8; indent-tabs-mode: t; c-basic-offset: 8 -*- */
       2              : /*
       3              :  * utils.c:  Various utility routines that should have been in glib.
       4              :  *
       5              :  * Authors:
       6              :  *    Miguel de Icaza (miguel@gnu.org)
       7              :  *    Jukka-Pekka Iivonen (iivonen@iki.fi)
       8              :  *    Zbigniew Chyla (cyba@gnome.pl)
       9              :  *    Morten Welinder (terra@gnome.org)
      10              :  */
      11              : #include <config.h>
      12              : #include "go-glib-extras.h"
      13              : 
      14              : #include <glib/gi18n-lib.h>
      15              : #include <libxml/encoding.h>
      16              : 
      17              : #include <stdlib.h>
      18              : #include <math.h>
      19              : #include <stdio.h>
      20              : #include <string.h>
      21              : #include <sys/types.h>
      22              : #include <sys/stat.h>
      23              : #ifdef HAVE_UNISTD_H
      24              : #include <unistd.h>
      25              : #endif
      26              : #include <fcntl.h>
      27              : #include <errno.h>
      28              : 
      29            0 : gint go_ascii_strcase_equal(gconstpointer v1, gconstpointer v2)
      30              : {
      31            0 :     return g_ascii_strcasecmp((char const *) v1, (char const *) v2) == 0;
      32              : }
      33              : 
      34              : /* a char* hash function from ASU */
      35            0 : guint go_ascii_strcase_hash(gconstpointer v)
      36              : {
      37            0 :     unsigned const char *s = (unsigned const char *) v;
      38              :     unsigned const char *p;
      39            0 :     guint h = 0, g;
      40              : 
      41            0 :     for (p = s; *p != '\0'; p += 1)
      42              :     {
      43            0 :         h = (h << 4) + g_ascii_tolower(*p);
      44            0 :         if ((g = h & 0xf0000000))
      45              :         {
      46            0 :             h = h ^ (g >> 24);
      47            0 :             h = h ^ g;
      48              :         }
      49              :     }
      50              : 
      51            0 :     return h /* % M */;
      52              : }
      53              : 
      54              : const char *
      55            3 : go_guess_encoding(const char *raw, size_t len, const char *user_guess,
      56              :         char **utf8_str)
      57              : {
      58              :     int try;
      59            3 :     gboolean debug = FALSE;
      60              : 
      61            3 :     g_return_val_if_fail(raw != NULL, NULL);
      62              : 
      63            0 :     for (try = 1; 1; try++)
      64            0 :     {
      65            3 :         char const *guess = NULL;
      66            3 :         GError *error = NULL;
      67              :         char *utf8_data;
      68              : 
      69            3 :         switch (try)
      70              :         {
      71            3 :         case 1:
      72            3 :             guess = user_guess;
      73            3 :             break;
      74            0 :         case 2:
      75            0 :             g_get_charset(&guess);
      76            0 :             break;
      77            0 :         case 3:
      78              :         {
      79            0 :             xmlCharEncoding enc = xmlDetectCharEncoding(
      80              :                     (const unsigned char*) raw, len);
      81            0 :             switch (enc)
      82              :             {
      83            0 :             case XML_CHAR_ENCODING_ERROR:
      84              :             case XML_CHAR_ENCODING_NONE:
      85            0 :                 break;
      86            0 :             case XML_CHAR_ENCODING_UTF16LE:
      87              :                 /* Default would give "UTF-16".  */
      88            0 :                 guess = "UTF-16LE";
      89            0 :                 break;
      90            0 :             case XML_CHAR_ENCODING_UTF16BE:
      91              :                 /* Default would give "UTF-16".  */
      92            0 :                 guess = "UTF-16BE";
      93            0 :                 break;
      94            0 :             default:
      95            0 :                 guess = xmlGetCharEncodingName(enc);
      96              :             }
      97            0 :             break;
      98              :         }
      99            0 :         case 4:
     100            0 :             guess = "ASCII";
     101            0 :             break;
     102            0 :         case 5:
     103            0 :             guess = "ISO-8859-1";
     104            0 :             break;
     105            0 :         case 6:
     106            0 :             guess = "UTF-8";
     107            0 :             break;
     108            0 :         default:
     109            0 :             return NULL;
     110              :         }
     111              : 
     112            3 :         if (!guess)
     113            0 :             continue;
     114              : 
     115            3 :         if (debug)
     116            0 :             g_print("Trying %s as encoding.\n", guess);
     117              : 
     118            3 :         utf8_data = g_convert(raw, len, "UTF-8", guess,
     119              :         NULL, NULL, &error);
     120            3 :         if (!error)
     121              :         {
     122              :             /*
     123              :              * We can actually fail this test when guess is UTF-8,
     124              :              * see #401588.
     125              :              */
     126            3 :             if (!g_utf8_validate(utf8_data, -1, NULL))
     127            0 :                 continue;
     128            3 :             if (debug)
     129            0 :                 g_print("Guessed %s as encoding.\n", guess);
     130            3 :             if (utf8_str)
     131            0 :                 *utf8_str = utf8_data;
     132              :             else
     133            3 :                 g_free(utf8_data);
     134            3 :             return guess;
     135              :         }
     136              : 
     137            0 :         g_error_free(error);
     138              :     }
     139              : }