#!/usr/bin/python2
# This is -*- Python -*-

import sys
sys.path.append("../src")
sys.path.append("../engine/build/lib.linux-i686-2.2")

import gnoetics

txt = gnoetics.text_new(sys.argv[1])

tokens = {}
for i in xrange(txt.length()):
    tok = txt.get_token(i)
    tokens[tok] = tok

all = tokens.values()
all.sort(lambda x, y:cmp(x.to_string().lower(), y.to_string().lower()))
decomp_count = 0
for t in all:
    decomp = t.get_decomp()
    if decomp:
        decomp_str = gnoetics.phoneme_decomp_to_string(decomp)
        decomp_count += 1
    else:
        decomp_str = ""
    if not t.pos_mask():
        print "%-20s  %d  %d  %s" % (t.to_string(),
                                     t.syllables(),
                                     t.word_count(),
                                     decomp_str)

print
print "%d of %d tokens have decompositions (%.1f%%)" % \
      (decomp_count, len(all), 100*(decomp_count / float(len(all))))