In [24]:
import json
import os
import codecs
import operator
feq_words = json.load(codecs.open('./feq_words.json', 'r', 'utf-8'))
out = {}
for filename in os.listdir("./pages_json"):
if filename.endswith(".json"):
arr = {}
file = json.load(codecs.open('./pages_json/'+filename, 'r', 'utf-8'))
#sorted_file = dict(list(reversed(sorted(file.items(), key=lambda x: x[1])))) #sort by value
for key in file:
arr[key] = file[key]/feq_words[key]
out[ filename[:-5] ] = [ key for key,value in reversed(sorted(arr.items(), key=lambda x: x[1])) ]
with open('./summary_keyword.json', 'w', encoding='utf8') as json_file:
json.dump(out, json_file, ensure_ascii=False, sort_keys=True)
json_file.close()
In [ ]: