aboutsummaryrefslogtreecommitdiff
path: root/catgets/gencat.c
diff options
context:
space:
mode:
authorUlrich Drepper <drepper@redhat.com>1999-04-15 08:34:13 +0000
committerUlrich Drepper <drepper@redhat.com>1999-04-15 08:34:13 +0000
commit96eaef36197e8664f6baadd0a1bffe72edd4a4e7 (patch)
tree0ed7fdd0efee2b7a9a9520af8c85bef0fc9f66b8 /catgets/gencat.c
parent670af0d939fa14479b5fc0a3a3ed07bd76935940 (diff)
downloadglibc-96eaef36197e8664f6baadd0a1bffe72edd4a4e7.tar
glibc-96eaef36197e8664f6baadd0a1bffe72edd4a4e7.tar.gz
glibc-96eaef36197e8664f6baadd0a1bffe72edd4a4e7.tar.bz2
glibc-96eaef36197e8664f6baadd0a1bffe72edd4a4e7.zip
Update.
1999-04-15 Ulrich Drepper <drepper@cygnus.com> * catgets/gencat.c (read_input_file): Don't drop white spaces between number/identifier and string. Patch by Dima Barsky <dima@pwd.hp.com>. If no white space at all follows number/identifier remove existing message with the number/ientifier.
Diffstat (limited to 'catgets/gencat.c')
-rw-r--r--catgets/gencat.c67
1 files changed, 49 insertions, 18 deletions
diff --git a/catgets/gencat.c b/catgets/gencat.c
index ccc658570c..eaeb59ca60 100644
--- a/catgets/gencat.c
+++ b/catgets/gencat.c
@@ -492,38 +492,54 @@ this is the first definition"));
{
const char *ident = this_line;
int message_number;
+ int any_space;
do
++this_line;
- while (this_line[0] != '\0' && !isspace (this_line[0]));;
- this_line[0] = '\0'; /* Terminate the identifier. */
+ while (this_line[0] != '\0' && !isspace (this_line[0]));
+ any_space = isspace (*this_line);
+ *this_line++ = '\0'; /* Terminate the identifier. */
- do
- ++this_line;
- while (isspace (this_line[0]));
/* Now we found the beginning of the message itself. */
if (isdigit (ident[0]))
{
struct message_list *runp;
+ struct message_list *lastp;
message_number = atoi (ident);
/* Find location to insert the new message. */
runp = current->current_set->messages;
+ lastp = NULL;
while (runp != NULL)
if (runp->number == message_number)
break;
else
- runp = runp->next;
+ {
+ lastp = runp;
+ runp = runp->next;
+ }
if (runp != NULL)
{
- /* Oh, oh. There is already a message with this
- number is the message set. */
- error_at_line (0, 0, fname, start_line,
- gettext ("duplicated message number"));
- error_at_line (0, 0, runp->fname, runp->line,
- gettext ("this is the first definition"));
+ if (any_space)
+ {
+ /* Oh, oh. There is already a message with this
+ number in the message set. */
+ error_at_line (0, 0, fname, start_line,
+ gettext ("duplicated message number"));
+ error_at_line (0, 0, runp->fname, runp->line,
+ gettext ("this is the first definition"));
+ }
+ else
+ {
+ /* We have to remove this message. */
+ if (lastp != NULL)
+ lastp->next = runp->next;
+ else
+ current->current_set->messages = runp->next;
+ free (runp);
+ }
message_number = 0;
}
ident = NULL; /* We don't have a symbol. */
@@ -535,10 +551,12 @@ this is the first definition"));
else if (ident[0] != '\0')
{
struct message_list *runp;
- runp = current->current_set->messages;
+ struct message_list *lastp;
/* Test whether the symbolic name was not used for
another message in this message set. */
+ runp = current->current_set->messages;
+ lastp = NULL;
while (runp != NULL)
if (runp->symbol != NULL && strcmp (ident, runp->symbol) == 0)
break;
@@ -546,11 +564,24 @@ this is the first definition"));
runp = runp->next;
if (runp != NULL)
{
- /* The name is already used. */
- error_at_line (0, 0, fname, start_line,
- gettext ("duplicated message identifier"));
- error_at_line (0, 0, runp->fname, runp->line,
- gettext ("this is the first definition"));
+ if (any_space)
+ {
+ /* The name is already used. */
+ error_at_line (0, 0, fname, start_line,
+ gettext ("\
+duplicated message identifier"));
+ error_at_line (0, 0, runp->fname, runp->line,
+ gettext ("this is the first definition"));
+ }
+ else
+ {
+ /* We have to remove this message. */
+ if (lastp != NULL)
+ lastp->next = runp->next;
+ else
+ current->current_set->messages = runp->next;
+ free (runp);
+ }
message_number = 0;
}
else