encoding.py

import sys,os,string,random
if len(sys.argv)==1:
    print 'warn: encoding name missing'
    print 'python %s <encoding name>'%(os.path.basename(__file__))
    sys.exit(0)

encoding=sys.argv[1]
charset={}
n=0xff+1#skip 0x00-0xFF
while n<0xfffd:
    try:
        u=unichr(n)
        c=u.encode(encoding)
        #output 2 bytes char, skip others
        if len(c)==2:
            charset[int(c.encode('hex'),16)]=n
        else:
            #print c.encode('hex')
            pass
    except UnicodeEncodeError:
       pass 
    except LookupError as e:
        print e.message
        sys.exit(1)
    except:
        print 'Unexpected error:', sys.exc_info()[0]
        raise
    finally:
        n+=1

tableTmpl='''// auto generated mapping table, ${encoding} ==> unicode
var iconv=require('iconv-lite')
var encoding='${encoding}',_encoding=encoding.replace(/[- ]/g,'')
/** 
 * set aliase name for ${encoding}
 * for big5, iconv.encoding['cp950']=_encoding
 */
//iconv.encodings['aliase1']=_encoding
//iconv.encodings['aliase2']=_encoding
iconv.encodings[encoding]=_encoding
iconv.encodings[_encoding]={
    type:'table',
    table:{
'''
with open(encoding+'.js','wb') as f:
    t=string.Template(tableTmpl)
    f.write(t.substitute(encoding=encoding))
    for k in sorted(charset.keys()):
        v=charset[k]
        f.write('        0x%04X:0x%04X,//%s\n'%(k,v,unichr(v).encode('utf8')))
    f.write('    }\n}')
print 'generated %s.js'%encoding

testTmpl='''
var iconv=require('iconv-lite'),
    assert=require('assert');
require('./${encoding}.js');

var testStr='${testStr}abc',
    testBuf=new Buffer('${testStringHex}616263','hex');

var resultBuf=iconv.encode(testStr,'${encoding}'),
    resultStr=iconv.decode(testBuf,'${encoding}');

assert.strictEqual(testStr,resultStr);
assert.strictEqual(testBuf.toString('hex'),resultBuf.toString('hex'));
console.log('test passed');
'''
with open(encoding+'Test.js','wb') as f:
    chars=random.sample(charset.keys(),5);
    testStr=''.join([unichr(charset[c]) for c in chars]).encode('utf8')
    testStringHex=''.join(['%X'%c for c in chars])
    t=string.Template(testTmpl)
    f.write(t.substitute(
        encoding=encoding,testStr=testStr,testStringHex=testStringHex))
print 'generated %sTest.js'%encoding