def get_word_term_dic(self):
lines = tool.get_file_lines("./data/word_term_dic.txt")
for line in lines:
temp = line.split("@@@@")
t = Term(int(temp[0]), int(temp[1]), int(temp[2]))
t.append_location([int(x) for x in temp[3].split("##")])
self.word_term_dic[int(
temp[0])] = self.word_term_dic.get(int(temp[0]), []) + [t]
def get_word_index_dic(self):
lines = tool.get_file_lines("./data/word_index_dic.txt")
for line in lines:
temp = line.split("\t")
try:
self.word_index_dic[temp[0].decode("utf-8")] = int(temp[1])
except:
continue
def get_lasted_doc_id():
""" 获取最新的文档编号
:return:
"""
lines = tool.get_file_lines("./data/doc.txt")
try:
lasted_id = int(lines[-1].split("@@@@")[0])
except IndexError or TypeError:
lasted_id = 0
return lasted_id
def get_word_freq_dic(self):
lines = tool.get_file_lines("./data/word_freq_dic.txt")
for line in lines:
temp = line.split("\t")
self.word_freq_dic[int(temp[0])] = int(temp[1])
def get_doc_dic(self):
lines = tool.get_file_lines("./data/doc.txt")
for line in lines:
temp = line.split("@@@@")
self.doc_dic[int(temp[0])] = temp[1]