morphemes.py

import codecs

def file2ms( path, ws=None, bs=[u'記号'] ): # bs filters punctuation
    #inp = unicode( open( path, 'r' ).read(), 'utf-8' )
    #return getMorphemes1( inp, ws, bs)
    f = codecs.open( path, 'r', 'utf-8' )
    inp = f.readlines()
    f.close()

    #return getMorphemes( mecab(None), e, ws, bs )
    mcb = mecab(None)

    s = set()
    for i in inp:
        ms = getMorphemes( mcb, i.strip(), ws, bs)
        for m in ms:
            s.add(m)

    mcb.terminate()

    return list(s)