def check_load_file(filename, files):
if not os.path.exists(filename) or not os.path.isfile(filename):
print 'Error can not find the specified file'
return
filetype = filename.split('.')[1]
if filetype == 'csv':
csv_data = CSVData(filename)
csv_data.parse_vectors()
files.append(csv_data)
elif filetype == 'txt':
txt_data = TXTData(filename)
txt_data.read_document()
files.append(txt_data)
else:
print 'Error unrecognized file type ', filetype
return
print 'Parsed file: ', filename
def restore_state(self):
if self.params_set == False:
print "Error connection parameters not set"
return None
else:
statment = "SELECT id, text, words FROM documents"
self.cursor.execute(statment)
rows = self.cursor.fetchall()
ndict = {}
data = JokerData("no.fn")
for row in rows:
txt = TXTData("no.fn")
txt.set_text(row[1])
txt.set_words(row[2].split(' '))
txt.unique_word_frequency()
ndict[row[0]] = txt
data.set_docs(ndict)
return data