Re: [Mailman-i18n] Automatic translation between GB and BIG5 code
Hi all, Michael Fang contributed a Big5 mailman.po file, which I've just checked into cvs. Thanks Michael! However, I've had trouble running it through msgfmt. See the errors below. If anybody can lend assistance, it would be greatly appreciated. -Barry -------------------- snip snip -------------------- % make catalogs msgfmt -o big5/LC_MESSAGES/mailman.mo big5/LC_MESSAGES/mailman.po big5/LC_MESSAGES/mailman.po:57: end-of-line within string big5/LC_MESSAGES/mailman.po:93: illegal control sequence big5/LC_MESSAGES/mailman.po:100: illegal control sequence big5/LC_MESSAGES/mailman.po:114: illegal control sequence big5/LC_MESSAGES/mailman.po:122: illegal control sequence big5/LC_MESSAGES/mailman.po:126: illegal control sequence big5/LC_MESSAGES/mailman.po:134: illegal control sequence big5/LC_MESSAGES/mailman.po:154: illegal control sequence big5/LC_MESSAGES/mailman.po:158: illegal control sequence big5/LC_MESSAGES/mailman.po:171: illegal control sequence big5/LC_MESSAGES/mailman.po:172: illegal control sequence big5/LC_MESSAGES/mailman.po:173: illegal control sequence big5/LC_MESSAGES/mailman.po:197: illegal control sequence big5/LC_MESSAGES/mailman.po:215: illegal control sequence big5/LC_MESSAGES/mailman.po:216: end-of-line within string big5/LC_MESSAGES/mailman.po:223: illegal control sequence big5/LC_MESSAGES/mailman.po:234: `msgid' and `msgstr' entries do not both end with '\n' big5/LC_MESSAGES/mailman.po:254: illegal control sequence big5/LC_MESSAGES/mailman.po:278: illegal control sequence big5/LC_MESSAGES/mailman.po:287: end-of-line within string too many errors, aborting make: *** [big5/LC_MESSAGES/mailman.mo] Error 1
On Sun, 29 Jul 2001, Barry A. Warsaw wrote:
Hi all,
Michael Fang contributed a Big5 mailman.po file, which I've just checked into cvs. Thanks Michael!
However, I've had trouble running it through msgfmt. See the errors below. If anybody can lend assistance, it would be greatly appreciated.
If you are using a new gettext (> 0.10.35), then you may need to update the PO file. I think you can update the po file with the following command: OLD_PO_FILE_INPUT=1 msgmerge filename.po mailman.pot > fixed.po They decided that 8 bit characters in .po files were bad in the newer versions of gettext. James.
"JH" == James Henstridge <james@daa.com.au> writes:
>> Hi all, Michael Fang contributed a Big5 mailman.po file, which >> I've just checked into cvs. Thanks Michael! However, I've had >> trouble running it through msgfmt. See the errors below. If >> anybody can lend assistance, it would be greatly appreciated. JH> If you are using a new gettext (> 0.10.35), then you may need JH> to update the PO file. I think you can update the po file JH> with the following command: OLD_PO_FILE_INPUT=1 msgmerge JH> filename.po mailman.pot > fixed.po JH> They decided that 8 bit characters in .po files were bad in JH> the newer versions of gettext. Okay. My RH6.1 system got gettext 0.10.35. I just tried to update gettext from cvs but ran into some problems (posted on bug-gnu-utils@gnu.org). I'll give this a try when I get a working version of a newer gettext (yeah, I could try to find a newer RPM, but for some dinosauric reason I find building from source much faster and easier... usually). Thanks, -Barry
"Barry A. Warsaw" wrote:
However, I've had trouble running it through msgfmt. See the errors below. If anybody can lend assistance, it would be greatly appreciated.
-Barry
-------------------- snip snip -------------------- % make catalogs msgfmt -o big5/LC_MESSAGES/mailman.mo big5/LC_MESSAGES/mailman.po big5/LC_MESSAGES/mailman.po:57: end-of-line within string big5/LC_MESSAGES/mailman.po:93: illegal control sequence
Hi Barry, I am not very sure but Big5 encoding may contain \ (backslash) in some of the double byte characters. In Japanese encoding, Shift_JIS has such an encoding scheme; while the first byte of Shift_JIS is 8th bit set, the following second byte may contain 7 or 8 bit code. There is a patch/module for Python to distiguish the second byte '\' in Japanese but I don't know if such is there for Chinese. -- tkikuchi@ is.kochi-u.ac.jp http://weather.is.kochi-u.ac.jp/
However, I've had trouble running it through msgfmt. See the errors below. If anybody can lend assistance, it would be greatly appreciated.
I could not get this file to work with OLD_PO_FILE_INPUT. Instead, it seems that you need a Big5 reader (i.e. mbstowc and friends). Glibc 2.2 has such support in /usr/lib/gconv/BIG5.so. With that, I only get mailman.po:234: `msgid' and `msgstr' entries do not both end with '\n' mailman.po:597: `msgid' and `msgstr' entries do not both end with '\n' mailman.po:625: `msgid' and `msgstr' entries do not both end with '\n' mailman.po:812: `msgid' and `msgstr' entries do not both end with '\n' mailman.po:1195: `msgid' and `msgstr' entries do not both end with '\n' mailman.po:1424: `msgid' and `msgstr' entries do not both end with '\n' mailman.po:1664: `msgid' and `msgstr' entries do not both end with '\n' found 7 fatal errors All these seem to be genuine errors. Yet, msgfmt will still produce a binary catalog; I'll send it in a separate attachment. I would consider it as a bug in msgfmt that it cannot format the file without proper gconv support. Bruno, would you like to have a look at the input file? Regards, Martin
Martin von Loewis wrote:
I could not get this file to work with OLD_PO_FILE_INPUT. Instead, it seems that you need a Big5 reader (i.e. mbstowc and friends). Glibc 2.2 has such support in /usr/lib/gconv/BIG5.so. With that, I only get
mailman.po:234: `msgid' and `msgstr' entries do not both end with '\n' mailman.po:597: `msgid' and `msgstr' entries do not both end with '\n' mailman.po:625: `msgid' and `msgstr' entries do not both end with '\n' mailman.po:812: `msgid' and `msgstr' entries do not both end with '\n' mailman.po:1195: `msgid' and `msgstr' entries do not both end with '\n' mailman.po:1424: `msgid' and `msgstr' entries do not both end with '\n' mailman.po:1664: `msgid' and `msgstr' entries do not both end with '\n' found 7 fatal errors
All these seem to be genuine errors. Yet, msgfmt will still produce a binary catalog; I'll send it in a separate attachment.
I have exprerience that these messages appear when msgid finish with "\n" and msgstr does not. Finish msgstr with "\n". Cheers -- ___ / F \ [[[]]]] ( O O ) #----------------0000--(_)--0000---------------# | Juan Carlos Rey Anaya (jcrey@uma.es) | | Servicio Central de informática | | Universidad de Málaga - España | #----------------------------------------------#
I have exprerience that these messages appear when msgid finish with "\n" and msgstr does not. Finish msgstr with "\n".
It is actually vice versa: msgstr ends with \n, msgid doesn't. However, in these cases, there are a number of commented-out lines in Chinese that I do not understand, so I cannot really propose to change them. Regards, Martin
participants (5)
-
barry＠zope.com -
James Henstridge -
Juan Carlos Rey Anaya -
Martin von Loewis -
Tokio Kikuchi