aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorFlorian Weimer <fweimer@redhat.com>2022-05-17 11:38:29 +0200
committerFlorian Weimer <fweimer@redhat.com>2022-05-17 11:38:29 +0200
commit0c34593491e4ea2de79ae85fedb26252529b5f35 (patch)
treeba9e4e0bd4d6c8b1ea4d215ae717d299fcc85405
parentde0b9d66446c553bdbae2c15a63ef8eb5f819d1d (diff)
downloadglibc-fw/localedef-utf8.zip
glibc-fw/localedef-utf8.tar.gz
glibc-fw/localedef-utf8.tar.bz2
locale: localdef input files are now encoded in UTF-8fw/localedef-utf8
-rw-r--r--locale/programs/linereader.c32
1 files changed, 29 insertions, 3 deletions
diff --git a/locale/programs/linereader.c b/locale/programs/linereader.c
index ca4abb0..485ccaf 100644
--- a/locale/programs/linereader.c
+++ b/locale/programs/linereader.c
@@ -688,7 +688,11 @@ get_string (struct linereader *lr, const struct charmap_t *charmap,
buf2 = NULL;
while ((ch = lr_getc (lr)) != '"' && ch != '\n' && ch != EOF)
- addc (&lrb, ch);
+ {
+ if (ch >= 0x80)
+ lr_error (lr, _("illegal 8-bit character in untranslated string"));
+ addc (&lrb, ch);
+ }
/* Catch errors with trailing escape character. */
if (lrb.act > 0 && lrb.buf[lrb.act - 1] == lr->escape_char
@@ -733,13 +737,35 @@ get_string (struct linereader *lr, const struct charmap_t *charmap,
if (ch == lr->escape_char)
{
ch = lr_getc (lr);
+ if (ch >= 0x80)
+ {
+ lr_error (lr, _("illegal 8-bit escape sequence"));
+ illegal_string = true;
+ break;
+ }
if (ch == '\n' || ch == EOF)
break;
}
+ else if (ch < 0x80)
+ {
+ wch = ch;
+ addc (&lrb, ch);
+ }
+ else /* UTF-8 sequence. */
+ {
+ if (!get_string_decode_utf8 (lr, ch, &wch))
+ {
+ illegal_string = true;
+ break;
+ }
+ get_string_U_char (locale, charmap, repertoire, wch,
+ &lrb, &illegal_string);
+ if (illegal_string)
+ break;
+ }
- addc (&lrb, ch);
if (return_widestr)
- ADDWC ((uint32_t) ch);
+ ADDWC (wch);
continue;
}