diff options
| author | Hye-Shik Chang <hyeshik@gmail.com> | 2006-03-26 02:34:59 (GMT) |
|---|---|---|
| committer | Hye-Shik Chang <hyeshik@gmail.com> | 2006-03-26 02:34:59 (GMT) |
| commit | e2ac4abd016310b0ce1152ccf65666b98e33a76e (patch) | |
| tree | 7e6e8423a0ee27770c0635196a2ccfada73b33f9 /Tools | |
| parent | a531e5b84c62e90ff0eaed1a61caf5d952409397 (diff) | |
| download | cpython-e2ac4abd016310b0ce1152ccf65666b98e33a76e.zip cpython-e2ac4abd016310b0ce1152ccf65666b98e33a76e.tar.gz cpython-e2ac4abd016310b0ce1152ccf65666b98e33a76e.tar.bz2 | |
Patch #1443155: Add the incremental codecs support for CJK codecs.
(reviewed by Walter Dörwald)
Diffstat (limited to 'Tools')
| -rw-r--r-- | Tools/unicode/Makefile | 5 | ||||
| -rw-r--r-- | Tools/unicode/gencjkcodecs.py | 65 |
2 files changed, 69 insertions, 1 deletions
diff --git a/Tools/unicode/Makefile b/Tools/unicode/Makefile index f266d4d..fbd3557 100644 --- a/Tools/unicode/Makefile +++ b/Tools/unicode/Makefile @@ -15,7 +15,7 @@ RM = /bin/rm all: distclean mappings codecs -codecs: misc windows iso apple ebcdic custom-mappings +codecs: misc windows iso apple ebcdic custom-mappings cjk ### Mappings @@ -72,6 +72,9 @@ ebcdic: build/ $(PYTHON) gencodec.py MAPPINGS/VENDORS/MICSFT/EBCDIC/ build/ $(RM) -f build/readme.* +cjk: build/ + $(PYTHON) gencjkcodecs.py build/ + ### Cleanup clean: diff --git a/Tools/unicode/gencjkcodecs.py b/Tools/unicode/gencjkcodecs.py new file mode 100644 index 0000000..47627c5 --- /dev/null +++ b/Tools/unicode/gencjkcodecs.py @@ -0,0 +1,65 @@ +import os, string + +codecs = { + 'cn': ('gb2312', 'gbk', 'gb18030', 'hz'), + 'tw': ('big5', 'cp950'), + 'hk': ('big5hkscs',), + 'jp': ('cp932', 'shift_jis', 'euc_jp', 'euc_jisx0213', 'shift_jisx0213', + 'euc_jis_2004', 'shift_jis_2004'), + 'kr': ('cp949', 'euc_kr', 'johab'), + 'iso2022': ('iso2022_jp', 'iso2022_jp_1', 'iso2022_jp_2', + 'iso2022_jp_2004', 'iso2022_jp_3', 'iso2022_jp_ext', + 'iso2022_kr'), +} + +TEMPLATE = string.Template("""\ +# +# $encoding.py: Python Unicode Codec for $ENCODING +# +# Written by Hye-Shik Chang <perky@FreeBSD.org> +# + +import _codecs_$owner, codecs +import _multibytecodec as mbc + +codec = _codecs_$owner.getcodec('$encoding') + +class Codec(codecs.Codec): + encode = codec.encode + decode = codec.decode + +class IncrementalEncoder(mbc.MultibyteIncrementalEncoder, + codecs.IncrementalEncoder): + codec = codec +class IncrementalDecoder(mbc.MultibyteIncrementalDecoder, + codecs.IncrementalDecoder): + codec = codec +class StreamReader(Codec, mbc.MultibyteStreamReader, codecs.StreamReader): + codec = codec +class StreamWriter(Codec, mbc.MultibyteStreamWriter, codecs.StreamWriter): + codec = codec + +def getregentry(): + return codecs.CodecInfo( + name='$encoding', + encode=Codec().encode, + decode=Codec().decode, + incrementalencoder=IncrementalEncoder, + incrementaldecoder=IncrementalDecoder, + streamreader=StreamReader, + streamwriter=StreamWriter, + ) +""") + +def gencodecs(prefix): + for loc, encodings in codecs.iteritems(): + for enc in encodings: + code = TEMPLATE.substitute(ENCODING=enc.upper(), + encoding=enc.lower(), + owner=loc) + codecpath = os.path.join(prefix, enc + '.py') + open(codecpath, 'w').write(code) + +if __name__ == '__main__': + import sys + gencodecs(sys.argv[1]) |