skk-jisyo-cdbmake.py

import sys
import re, io
import subprocess
import concurrent.futures
from pathlib import Path
from hashlib import md5

_source_enc = 'euc-jp'
_cdbmake = '/opt/local/bin/cdbmake'


def _genCdbFormat(target):
    re_commentline = re.compile("^;")
    re_keyvalue = re.compile("^([^ ]+) +(.+)$")
    fmt = "+{key_len},{val_len}:{key}->{val}"
    
    def _gen():
        try:
            while True:
                line = yield
                if not re_commentline.match(line):
                    matched = re_keyvalue.match(line)
                    if matched:
                        key = matched.group(1)
                        val = matched.group(2)
                        print(fmt.format(key=key, key_len=len(key.encode()),
                                         val=val, val_len=len(val.encode())),
                              file=target)
        finally:
            print('', file=target)

    gen = _gen()
    next(gen)
    return gen
    
        
def _runCdbMake(filename, buff):
    src_p = Path(filename)
    tar_p = src_p.parent / (src_p.name + '.cdb')
    tmp_name = md5(bytes(src_p.name, 'utf-8')).hexdigest()
    tmp_p = src_p.parent / tmp_name
    
    if tar_p.exists():
        return ('[Error: '+str(tar_p)+' already exists]', 1)

    done = subprocess.run([_cdbmake, str(tar_p), str(tmp_p)],
                          input=buff.getvalue(), text=True)
    return (str(tar_p), done.returncode)
    
    
def make(filename):
    with open(filename, "r", encoding=_source_enc) as source:
        with io.StringIO() as target:
            gen = _genCdbFormat(target)
            for line in source:
                gen.send(line)
            gen.close()
            return _runCdbMake(filename, target)


if __name__ == '__main__':
    with concurrent.futures.ProcessPoolExecutor() as exe:
        futures = {exe.submit(make, arg): arg for arg in sys.argv[1:]}

        for future in concurrent.futures.as_completed(futures):
            src = futures[future]
            print('%s -> %s, exit %d' % (src, *future.result()))