twitools/csvdb.py

33 lines
1.3 KiB
Python
Raw Normal View History

2015-04-13 20:58:32 +00:00
#!/usr/bin/env python3
2016-04-07 22:23:04 +00:00
import dbtools
import sqlite3, csv, sys
2016-04-07 22:23:04 +00:00
def makeDB(dbo=dbtools.dbObject(), infile='tweets.csv'):
2015-10-10 22:10:57 +00:00
try:
infile = open(infile)
2015-10-10 22:10:57 +00:00
except IOError:
raise IOError("Unable to read %s." % infile)
2015-10-10 22:10:57 +00:00
input = list(csv.reader(infile))
try:
2016-04-07 22:23:04 +00:00
dbo.executeQuery("CREATE TABLE tweets(`tweet_id` INTEGER NOT NULL, `in_reply_to_status_id` TEXT, `in_reply_to_user_id` TEXT, `timestamp` TEXT, `source` TEXT, `text` TEXT, `retweeted_status_id` TEXT, `retweeted_status_user_id` TEXT, `retweeted_status_timestamp` TEXT, `expanded_urls` TEXT, PRIMARY KEY(tweet_id));")
except:
2015-10-10 22:10:57 +00:00
raise IOError("%s already exists. Please delete it before trying to create a new one." % path)
for row in input[1:]:
2016-04-07 22:23:04 +00:00
dbo.executeQuery("INSERT INTO tweets VALUES(" + row[0].replace("'","''") + ",'" + row[1].replace("'","''") + "','" + row[2].replace("'","''") + "','" + row[3].replace("'","''") + "','" + row[4].replace("'","''") + "','" + row[5].replace("'","''") + "','" + row[6].replace("'","''") + "','" + row[7].replace("'","''") + "','" + row[8].replace("'","''") + "','" + row[9].replace("'","''") + "');")
2015-10-10 22:10:57 +00:00
2016-04-07 22:23:04 +00:00
dbo.commit()
2015-10-10 22:10:57 +00:00
if __name__ == "__main__":
if len(sys.argv) > 2:
raise ValueError(sys.argv[0] + " only takes one argument, the path of the new database file.")
try:
makeDB(sys.argv[1])
except IndexError:
makeDB()