'''
Created on Nov 30, 2016

@author: David
'''

import compound
import lexin
import pattern
import syllable
import pipeline
import codecs
import sys
from collections import Counter

cefr_levels = ["A1","A2","B1","B2","C1"]
file = "f:/parll data/data/svalex-ttm.csv"
outfile = "f:/parll data/data/scowl-summary.csv"

f = codecs.open(file, "r", "utf-8")
dummy = {}
syllable_summary = {}
unigram_summary = {}
bigram_summary = {}
trigram_summary = {}
suffix_summary = {}
suffix_len_summary = {}
compound_summary = {}
compound_count_summary = {}
gender_summary = {}
polysemy_summary = {}
homonymy_summary = {}
pos_summary = {}

summaries = [syllable_summary,unigram_summary,dummy,bigram_summary,dummy,trigram_summary,dummy,suffix_summary,suffix_len_summary,compound_summary,compound_count_summary,gender_summary,polysemy_summary,homonymy_summary]

def reduce_to_value(something):
    #print(type(something))
    if (type(something) is list):
        if not something:
            return 0
        something = something[0]
        if (type(something) is list):
            return something[0]
        else:
            return something
    if (isinstance(something, Counter)):
        return 0
        
    return something

for cefr_level in cefr_levels:
    pos_summary[cefr_level] = {}
    for summary in summaries:
        summary[cefr_level] = {}
        
count = 0

for line in f:
    lemma,pos,cefr = line.rstrip().split("\t")
    
    result=pipeline.run(lemma, pos, cefr)
    
    for i in range(len(summaries)):
        summary = summaries[i]
        
        value = reduce_to_value(result[i])
        if (pos not in pos_summary[cefr]):
            pos_summary[cefr][pos] = 0
        if (value not in summary[cefr]):
            summary[cefr][value] = 0
            
        summary[cefr][value] += 1
        pos_summary[cefr][pos] += 1
#     if (count == 50):
#         break
#     count += 1
#     
summaries.append(pos_summary)

o = codecs.open(outfile,"w","utf-8")

for i in range(len(summaries)):
    
    for cefr in cefr_levels:
    
        print("{} {} {}".format(i,cefr,summaries[i][cefr]))
        o.write("{} {} {}".format(i,cefr,summaries[i][cefr]))    
    o.write("--------------------\n")