mc-search-segv.patch

   1 diff --git a/lib/search/regex.c b/lib/search/regex.c
   2 index f6eb24e..53e3b6f 100644
   3 --- a/lib/search/regex.c
   4 +++ b/lib/search/regex.c
   5 @@ -249,6 +249,60 @@ mc_search__cond_struct_new_regex_ci_str (const char *charset, const GString * as
   6
   7  /* --------------------------------------------------------------------------------------------- */
   8
   9 +#ifdef SEARCH_TYPE_GLIB
  10 +/* Glib doesn't like invalid UTF-8 so sanitize it first: ticket 3449.
  11 + * Be careful: there might be embedded NULs in the strings. */
  12 +static gboolean
  13 +mc_search__g_regex_match_full_safe (const GRegex * regex,
  14 +                                    const gchar * string,
  15 +                                    gssize string_len,
  16 +                                    gint start_position,
  17 +                                    GRegexMatchFlags match_options,
  18 +                                    GMatchInfo ** match_info, GError ** error)
  19 +{
  20 +    char *string_safe, *p, *end;
  21 +    gboolean ret;
  22 +
  23 +    if ((g_regex_get_compile_flags (regex) & G_REGEX_RAW)
  24 +        || g_utf8_validate (string, string_len, NULL))
  25 +    {
  26 +        return g_regex_match_full (regex, string, string_len, start_position, match_options,
  27 +                                   match_info, error);
  28 +    }
  29 +
  30 +    if (string_len < 0)
  31 +    {
  32 +        string_len = strlen (string);
  33 +    }
  34 +    p = string_safe = g_strndup (string, string_len);
  35 +    end = p + string_len;
  36 +
  37 +    while (p < end)
  38 +    {
  39 +        gunichar c = g_utf8_get_char_validated (p, -1);
  40 +        if (c != (gunichar) (-1) && c != (gunichar) (-2))
  41 +        {
  42 +            p = g_utf8_next_char (p);
  43 +        }
  44 +        else
  45 +        {
  46 +            /* U+FFFD would be the proper choice, but then we'd have to
  47 +               maintain mapping between old and new offsets.
  48 +               So rather do a byte by byte replacement. */
  49 +            *p++ = '\0';
  50 +        }
  51 +    }
  52 +
  53 +    ret =
  54 +        g_regex_match_full (regex, string_safe, string_len, start_position, match_options,
  55 +                            match_info, error);
  56 +    g_free (string_safe);
  57 +    return ret;
  58 +}
  59 +#endif /* SEARCH_TYPE_GLIB */
  60 +
  61 +/* --------------------------------------------------------------------------------------------- */
  62 +
  63  static mc_search__found_cond_t
  64  mc_search__regex_found_cond_one (mc_search_t * lc_mc_search, mc_search_regex_t * regex,
  65                                   GString * search_str)
  66 @@ -256,8 +310,9 @@ mc_search__regex_found_cond_one (mc_search_t * lc_mc_search, mc_search_regex_t *
  67  #ifdef SEARCH_TYPE_GLIB
  68      GError *mcerror = NULL;
  69
  70 -    if (!g_regex_match_full (regex, search_str->str, search_str->len, 0, G_REGEX_MATCH_NEWLINE_ANY,
  71 -                             &lc_mc_search->regex_match_info, &mcerror))
  72 +    if (!mc_search__g_regex_match_full_safe
  73 +        (regex, search_str->str, search_str->len, 0, G_REGEX_MATCH_NEWLINE_ANY,
  74 +         &lc_mc_search->regex_match_info, &mcerror))
  75      {
  76          g_match_info_free (lc_mc_search->regex_match_info);
  77          lc_mc_search->regex_match_info = NULL;