8:56 a.m.
Hi,
today, while testing the Docbook module on new documents, I've seen that
some docbook documents are indented with tabs. I've noticed that the
canonizing function doesn't handle tabs, and such msgids look like this:
"line1\t\t\tline2\t\t\tline3..."
It's so annoying to translate. Can we also canonize tabs to spaces or
would it break something?
This simple patch works for me:
--- lib/Locale/Po4a/Po.pm 27 Feb 2005 22:53:53 -0000 1.37
+++ lib/Locale/Po4a/Po.pm 21 Mar 2005 14:53:19 -0000
@@ -1065,6 +1065,7 @@
$text =~ s/^ *//s;
# if ($text eq "\n"), it messed up the first string (header)
$text =~ s/\n/ /gm if ($text ne "\n");
+ $text =~ s/\t/ /gm;
$text =~ s/([.)]) +/$1 /gm;
$text =~ s/([^.)]) */$1 /gm;
$text =~ s/ *$//s;
Regards,
Jordi Vilalta
4:45 p.m.
On Mon, Mar 21, 2005 at 03:56:37PM +0100, Jordi Vilalta wrote:
today, while testing the Docbook module on new documents, I've
seen that
some docbook documents are indented with tabs. I've noticed that the
canonizing function doesn't handle tabs, and such msgids look like this:
"line1\t\t\tline2\t\t\tline3..."
It's so annoying to translate. Can we also canonize tabs to spaces or
would it break something?
For the TeX format, there shouldn't be any problem (IIRC, spaces and tabs
are quite equivalent, except in verbatim blocks, which are not canonized
anyway).
For the Man and Pod modules, I can think of possible issues, because a
space is not be equivalent to a tab (e.g. at the beginning of a line, or
when the tab is used for alignment). However, all these are theoric, and I
could not make pages (that would not have been considered buggy) showing
any issue (in these cases, the document should have use verbatim blocks).
This simple patch works for me:
I will try the exhaustive test suite (but I think it is not suitable to
detect possible issues triggered by this patch).
+ $text =~ s/\t/ /gm;
Note that this change all tabs, not only trailing and leading tabs.
Best Regards,
--
Nekral
5:02 p.m.
Hi,
On Mon, 21 Mar 2005, Nicolas François wrote:
On Mon, Mar 21, 2005 at 03:56:37PM +0100, Jordi Vilalta wrote:
> + $text =~ s/\t/ /gm;
Note that this change all tabs, not only trailing and leading tabs.
When canonizing, all the newlines disappear, and all the tabs get to the
middle of the line. Is there some format that uses tabs in the middle of a
string for some function?
Regards,
Jordi Vilalta
5:23 p.m.
Hi,
On Tue, Mar 22, 2005 at 12:02:40AM +0100, Jordi Vilalta wrote:
On Mon, 21 Mar 2005, Nicolas François wrote:
>On Mon, Mar 21, 2005 at 03:56:37PM +0100, Jordi Vilalta wrote:
>>+ $text =~ s/\t/ /gm;
>
>Note that this change all tabs, not only trailing and leading tabs.
When canonizing, all the newlines disappear, and all the tabs get to the
middle of the line.
That's right.
Is there some format that uses tabs in the middle of a
string for some function?
The only thing I could imagine is:
.SH 1\tfoo
.SH 10\tbar
And the author willing to align foo and bar. (I admit this would be a
quite buggy man page)
--
Nekral
1:27 a.m.
On Mon, Mar 21, 2005 at 11:45:49PM +0100, Nicolas François wrote:
For the Man and Pod modules, I can think of possible issues, because
a
space is not be equivalent to a tab (e.g. at the beginning of a line, or
when the tab is used for alignment). However, all these are theoric, and I
could not make pages (that would not have been considered buggy) showing
any issue (in these cases, the document should have use verbatim blocks).
From perlpod(1):
A verbatim paragraph is distinguished by having its first character be a
space or a tab. (And commonly, all its lines begin with spaces and/or
tabs.)
And from memory, the groff behaviour is to consider as verbatim any
paragraph containing a white char (space or tab) at the first position of
any line (not only the first one).
But as Jordi said, such paragraph won't get canonized anyway.
I think that this patch won't break anything.
Bye, Mt.
5:55 p.m.
On Tue, Mar 22, 2005 at 08:27:58AM +0100, Martin Quinson wrote:
And from memory, the groff behaviour is to consider as verbatim any
paragraph containing a white char (space or tab) at the first position of
any line (not only the first one).
I could not find any reference in the info groff, but experimentally, I
would say it's only spaces, and on any line except the first one.
I fact, I'm not really concerned by tabs at the beginning of lines, but in
their middle.
I think that this patch won't break anything.
I launched the testsuite/check script before and after the patch.
The patch changes the result for 5 pages:
metamail.1
mimencode.1
This pages are buggy.
totem.1
There were already a po4a bug for this one (due to tabs)
syslog-ng.conf.5
sane-avision.5
Not an important change (I would say these page is buggy)
So, that's what I feared and hoped: there are some differences, but on
buggy (or quite) man pages.
I'm quite uncomfortable with totem, because the author tried very hard to
use tabs as a formatting tool (using .ta).
But that makes only one page out of 5000. However, I may try to detect
when these tabs are used for formatting.
Given the advantages for the other formats, I now think it is better to use
it.
Regards,
--
Nekral
6:14 p.m.
On Wed, Mar 23, 2005 at 12:55:24AM +0100, Nicolas François wrote:
On Tue, Mar 22, 2005 at 08:27:58AM +0100, Martin Quinson wrote:
> And from memory, the groff behaviour is to consider as verbatim any
> paragraph containing a white char (space or tab) at the first position of
> any line (not only the first one).
I could not find any reference in the info groff, but experimentally, I
would say it's only spaces, and on any line except the first one.
I fact, I'm not really concerned by tabs at the beginning of lines, but in
their middle.
If so, you can move the s/\t/ / before the s/\n/ / and change it to a
s/\n\t+/ /. It will do the trick for Jordi's xml document, and may make
your life easier.
> I think that this patch won't break anything.
I launched the testsuite/check script before and after the patch.
The patch changes the result for 5 pages:
metamail.1
mimencode.1
This pages are buggy.
totem.1
There were already a po4a bug for this one (due to tabs)
syslog-ng.conf.5
sane-avision.5
Not an important change (I would say these page is buggy)
So, that's what I feared and hoped: there are some differences, but on
buggy (or quite) man pages.
I'm quite uncomfortable with totem, because the author tried very hard to
use tabs as a formatting tool (using .ta).
But that makes only one page out of 5000. However, I may try to detect
when these tabs are used for formatting.
Given the advantages for the other formats, I now think it is better to use
it.
Cool, thanks for testing.
Mt.
2:38 p.m.
On Wed, Mar 23, 2005 at 01:14:14AM +0100, Martin Quinson wrote:
On Wed, Mar 23, 2005 at 12:55:24AM +0100, Nicolas François wrote:
> I fact, I'm not really concerned by tabs at the beginning of lines, but in
> their middle.
If so, you can move the s/\t/ / before the s/\n/ / and change it to a
s/\n\t+/ /. It will do the trick for Jordi's xml document, and may make
your life easier.
I tried with 's/^[ \t]+/ /gm;' (still before the 's/\n/ /gm') and only
the sane-avision page was changed (its usage of a tab is anyway buggy).
Jordi, would this patch be sufficient for you?
--
Nekral
4:20 p.m.
On Wed, 23 Mar 2005, Nicolas François wrote:
On Wed, Mar 23, 2005 at 01:14:14AM +0100, Martin Quinson wrote:
> On Wed, Mar 23, 2005 at 12:55:24AM +0100, Nicolas François wrote:
>> I fact, I'm not really concerned by tabs at the beginning of lines, but in
>> their middle.
>
> If so, you can move the s/\t/ / before the s/\n/ / and change it to a
> s/\n\t+/ /. It will do the trick for Jordi's xml document, and may make
> your life easier.
I tried with 's/^[ \t]+/ /gm;' (still before the 's/\n/ /gm') and only
the sane-avision page was changed (its usage of a tab is anyway buggy).
Jordi, would this patch be sufficient for you?
Yes, it seems to be enough for my case.
Regards,
Jordi Vilalta
7183
days inactive
7185
days old
8 comments
3 participants
participants (3)
-
Jordi Vilalta -
Martin Quinson -
Nicolas François