# THIS PYTHON SCRIPTS USES RAKE METRICS

# DIFFERS FROM RAKE IN CALCULATING THE CANDIDATE KEYWORDS AKA phrase_list

# USES NP_CHUNK TO FIND CANDIDATE KEYWORDS

import nltk

import operator

import sys

from textblob import TextBlob

from rake import RakeKeywordExtractor

from textblob.np_extractors import ConllExtractor

from nltk.stem import WordNetLemmatizer

from nltk.tokenize import WordPunctTokenizer

from textblob.taggers import NLTKTagger

from stopwordList import getList

import codecs

## GLOBAL VARIABLES

top_fraction = 1

LEMMA_OBJ = WordNetLemmatizer()

tokenizer = WordPunctTokenizer()

nltk_tagger = NLTKTagger()

stopwords = getList()

COLL_OBJ = ConllExtractor()

def rake_extract(phrase_list):

return sorted_phrase_scores[0:int(n_phrases)]

#FILE = open(sys.argv[1],"r")

FILE = codecs.open(sys.argv[1],"r","iso8859-15")

CONTENT = FILE.read()

CONTENT = CONTENT.encode('ascii','ignore')

BLOB_OBJ = TextBlob(CONTENT,tokenizer=tokenizer,pos_tagger = nltk_tagger,np_extractor = COLL_OBJ) # OBJECT WITH FAST NP CHUNKER

pos_tags = nltk.pos_tag(BLOB_OBJ.tokens)

BLOB_OBJ.pos_tags = pos_tags

phrase_list = BLOB_OBJ.noun_phrases

newlist =[]

temp =[]

for word in phrase_list:

words = word.split(' ')

for x in words:

x = LEMMA_OBJ.lemmatize(x)

x = x.lower()

if x not in stopwords:

temp.append(x)

newlist.append(temp)

temp = []

phrase_list = newlist

#print COLLBLOB_OBJ.noun_phrases

keywordList = rake_extract(phrase_list)

for keyword in keywordList:

print str(keyword[0])