#!/usr/bin/python2

import string, os, os.path, random

path = "../dict/tagged-corpora/susanne"
files = map(lambda x: "A%02d" % x, range(1, 15)) \
        + map(lambda x: "G%02d" % x, range(1, 14)) \
        + map(lambda x: "J%02d" % x, range(1, 11)) \
        + map(lambda x: "N%02d" % x, range(1, 16))

pos_to_words = {}
successors = {}

lookback = 3

def scan_file(name):
    print "Scanning %s..." % name
    f = file(name)
    history = []
    for line in f.xreadlines():
        F = line.split("\t")
        pos = F[2]
        word = F[3]
        if word[0] == "+" and word[1] in string.letters:
            word = word[1:]
            
        pos_to_words.setdefault(pos, []).append(word)

        if history:
            key = string.join(history, "//")
            successors.setdefault(key, []).append(pos)

        history.append(pos)
        if len(history) > lookback:
            history.pop(0)


def pick_word(pos):
    return random.choice(pos_to_words.get(pos, ["*error*"]))

def pick_successor(pos):
    L = successors.get(pos)
    return L and random.choice(L)



for f in files:
    name = os.path.join(path, f)
    scan_file(name)


history = []
pos = "YB"
str = ""
while 1:
    if pos == "YB" or pos == "YF":
        if str:
            print str
            print
        str = ""
        history = []
    else:
        word = pick_word(pos)
        if word[0] == "+":
            word = word[1:]
        elif str:
            str += " "
        str += word
        #str += "%s[%s]" % (word, pos)

    history.append(pos)
    if len(history) > lookback:
        history.pop(0)
    key = string.join(history, "//")
    pos = pick_successor(key)

    if not pos:
        str += " [DeadEnd!]"
        pos = "YB"
        history = []