Encode::HanConvert - Traditional and Simplified Chinese mappings
This document describes version 0.29 of Encode::HanConvert, released June 4, 2004.
As command line utilities:
b2g.pl [ -p ] [ -u ] [ inputfile ...] > outputfile
g2b.pl [ -p ] [ -u ] [[ inputfile ...] > outputfile
In your program:
# The XS-based implementation needs Encode.pm 1.41; # otherwise, autoloads the Perl-based Encode::HanConvert::Perl use Encode::HanConvert; # Conversion between Chinese encodings $gbk = big5_to_gb($big5); # Big5 to GBK $big5 = gb_to_big5($gbk); # GBK to Big5 # Conversion between Perl's Unicode strings $simp = trad_to_simp($trad); # Traditional to Simplified $trad = simp_to_trad($simp); # Simplified to Traditional # Conversion between Chinese encoding and Unicode strings $simp = big5_to_simp($big5); # Big5 to Simplified $big5 = simp_to_big5($simp); # Simplified to Big5 $trad = gb_to_trad($gbk); # GBK to Traditional $gbk = trad_to_gb($trad); # Traditional to GBK # For completeness' sake... (no conversion, just encode/decode) $simp = gb_to_simp($gbk); # GBK to Simplified $gbk = simp_to_gb($simp); # Simplified to GBK $trad = big5_to_trad($big5); # Big5 to Traditional $big5 = trad_to_big5($trad); # Traditional to Big5 # All functions may be used in void context to transform $_ big5_to_gb($string); # convert $string from Big5 to GBK # Drop-in replacement functions for Lingua::ZH::HanConvert use Encode::HanConvert qw(trad simple); # not exported by default $simp = simple($trad); # Traditional to Simplified $trad = trad($simp); # Simplified to Traditional
This module is an attempt to solve most common problems occured in Traditional vs. Simplified Chinese conversion, in an efficient, flexible way, without resorting to external tools or modules.
If you are using perl 5.7.2 or earlier, all Unicode-related functions are disabled, and Encode::HanConvert::Perl is automagically loaded and used instead. In that case, please consult Encode::HanConvert::Perl instead.
After installing this module, you'll have two additional encoding formats:
big5-simp maps Big5 into Unicode's Simplified Chinese (and vice versa), and
gbk-trad maps GBK (also known as CP936) into Unicode's Traditional Chinese and back.
The module exports various
xxx_to_yyy functions by default, where xxx and yyy are one of
gb (i.e. GBK/CP936),
simp (simplified Chinese unicode), or
trad (traditional Chinese unicode).
You may also import
trad, which are aliases for
trad_to_simp; this is provided as a drop-in replacement for programs using Lingua::ZH::HanConvert.
Since this is built on Encode's architecture, you may also use the line discipline syntax to perform the conversion implicitly (before 5.7.3, you need to use 'cp936' in place of 'gbk'):
require Encode::CN; open BIG5, ':encoding(big5-simp)', 'big5.txt'; # as simplified open EUC, '>:encoding(gbk)', 'gbk.txt'; # as gbk print EUC, <BIG5>; require Encode::TW; open EUC, ':encoding(gbk-trad)', 'gbk.txt'; # as traditional open BIG5, '>:encoding(big5)', 'big5.txt'; # as big-5 print BIG5, <EUC>;
Or, more interestingly:
use encoding 'big5-simp'; print "¤¤¤å"; # prints simplified Chinese in unicode
Although Lingua::ZH::HanConvert module already provides mapping between Simplified and Traditional Unicode characters, it depend on other modules (Text::Iconv or Encode) to provide the necessary mapping with Big5 and GBK encodings.
Also, Encode::HanConvert loads up much faster:
0.148u 0.046s 0:00.19 94.7% # Encode::HanConvert 1.192u 0.046s 0:01.26 97.6% # Encode::HanConvert::Perl 7.096u 0.015s 0:07.23 98.2% # Lingua::ZH::HanConvert (v0.12)
The difference in actual conversion is much more significant. Use 32k text of trad => simp as an example:
0.082u 0.031s 0:00.12 91.6% # iconv | b2g | iconv 0.263u 0.038s 0:00.30 96.6% # Encode::HanConvert 23.715u 0.054s 0:24.51 96.9% # Lingua::ZH::HanConvert (v0.12)
b2g above refers to Yeung and Lee's HanZi Converter, a C-based program that maps big5 to gb2312 and back;
iconv refers to GNU libiconv. If you don't mind the overhead of calling an external process, their result is nearly identical with this module; however, their map falls short on rarely-used characters and box-drawing symbols.
Please note that from version 0.03 and above, this module support the more expressive range GBK instead of EUC-CN. This may cause incompatibilities with older fonts. Programs using an earlier version of this module should rename
gbk-trad; sorry for the inconvenience.
This module does not preserve one-to-many mappings; it blindly chooses the most frequently used substitutions, instead of presenting the user multiple choices. This can be remedied by a dictionary-based post processor that restores the correct character.
As of version 0.05, the mapping from Big5 to GBK is complete: All displayable Big5 characters are mapped, although substitute characters are used where there's no direct corresponding characters.
However, there are numerous GBK characters without its Big5 counterparts:
grep ¡¼ map/g2b_map.txt from the distribution directory should show all of them. Any help on completing this mapping are very appreciated.
The conversion table used in this module comes from various sources, including Lingua::ZH::HanConvert by David Chan, hc by Ricky Yeung & Fung F. Lee, and Doggy Chinese Big5-GB Conversion Master from Doggy Digital Creative Inc. (http://www.miniasp.com/), Rei-Li Chen (rexchen), as well as mappings used in Microsoft Word 2000, Far East edition.
The *.ucm files are checked against test files generated by GNU libiconv with kind permission from Bruno Haible.
Kudos to Nick Ing-Simmons, Dan Kogai and Jarkko Hietaniemi for showing me how to use Encode and PerlIO. Thanks!
Autrijus Tang <email@example.com>
Copyright 2002, 2003, 2004 by Autrijus Tang <firstname.lastname@example.org>.
This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.