# This is -*- Python -*-

# The first haiku
#
# The defensive heat
# participated --- to be
# termed from most negros.
#

import string, random, os, sys

import gnoetics
from grammar_model import *

model = gnoetics.seq_model_new(4)
word_dict = {}

final_tags = ("YB", "YF", "YQ", "YX")

punct_map = {
    "<apos>": "'",
    "<hyphen>": "-",
    "<mdash>": "---",
    }

def scan_file(filename):
    f = file(filename)

    sentence = []
    
    for line in f.xreadlines():
        F = line.split()
        tag_str = F[2]
        word = string.lower(F[3])

        for key, punct in punct_map.items():
            i = word.find(key)
            if i >= 0:
                word = word[:i] + punct + word[i+len(key):]
                
        if tag_str != "YB" \
               and word.find("ldquo") == -1 \
               and word.find("lsquo") == -1 \
               and word.find("rdquo") == -1 \
               and word.find("rsquo") == -1:
            tag = gnoetics.pos_from_string(tag_str)
            tok = gnoetics.token_lookup(word)
            if word_dict.has_key(tag):
                c = word_dict[tag]
            else:
                c = gnoetics.TokenChooser()
                word_dict[tag] = c
            c.add(tok)
            sentence.append(tag)

        if tag_str in final_tags:
            model.add_sentence(sentence)
            sentence = []

    if sentence:
        model.add_sentence(sentence)

path = "../dict/tagged-corpora/susanne"
for name in os.listdir(path):
    if len(name) == 3:
        sys.stderr.write(name + " ")
        name = os.path.join(path, name)
        scan_file(name)

sys.stderr.write("\n")

terminals = model.terminals()


def gen_haiku():

    t = None

    while 1:
        if t is None:
            pattern = [5, 7, 5]
            t = model.get_break_tuple()
            sentence = []

        x = model.step_next(t)
        if not pattern:
            if x in terminals: # we ended right on time
                tok = word_dict[x].choice(syllables_eq=0)
                if tok:
                    sentence[-1] = tok.to_string()
                    return sentence
            t = None
        else:
            if x != 0: # we ended early
                n = pattern[0]
                c = word_dict[x]
                if c.choice_exists(syllables_lteq=n):
                    tok = c.choice(syllables_lteq=n)
                    sentence.append(tok.to_string())
                    pattern[0] -= tok.syllables()
                    if pattern[0] == 0:
                        pattern.pop(0)
                        sentence.append("\n")
                else:
                    t = None

        if t:
            t = t[1:] + (x,)


while 1:
    sentence = gen_haiku()
    print string.join(sentence)
    print
    print