How do I create a CSV file from database in Python?

You can easily create it manually, writing a file with a chosen separator. You can also use csv module.

If it’s from database you can alo just use a query from your sqlite client :

sqlite <db params> < queryfile.sql > output.csv

Which will create a csv file with tab separator.

unless i’m missing something, you just want to do something like so…

f = open("somefile.csv")
f.writelines("header_row")

logic to write lines to file (you may need to organize values and add comms or pipes etc…)

f.close()

Using the csv module is very straight forward and made for this task.

import csv
writer = csv.writer(open("out.csv", 'w'))
writer.writerow(['name', 'address', 'phone', 'etc'])
writer.writerow(['bob', '2 main st', '703', 'yada'])
writer.writerow(['mary', '3 main st', '704', 'yada'])

Creates exactly the format you’re expecting.

How to extract the column headings from an existing table:

You don’t need to parse an SQL “create table” statement. This is fortunate, as the “create table” syntax is neither nice nor clean, it is warthog-ugly.

You can use the table_info pragma. It gives you useful information about each column in a table, including the name of the column.

Example:

>>> #coding: ascii
... import sqlite3
>>>
>>> def get_col_names(cursor, table_name):
...     results = cursor.execute("PRAGMA table_info(%s);" % table_name)
...     return [row[1] for row in results]
...
>>> def wrong_way(cur, table):
...     import re
...     cur.execute("SELECT sql FROM sqlite_master WHERE name=?;", (table, ))
...     sql = cur.fetchone()[0]
...     column_defs = re.findall("[(](.*)[)]", sql)[0]
...     first_words = (line.split()[0].strip() for line in column_defs.split(','))
...     columns = [word for word in first_words if word.upper() != "CONSTRAINT"]
...     return columns
...
>>> conn = sqlite3.connect(":memory:")
>>> curs = conn.cursor()
>>> _ignored = curs.execute(
...     "create table foo (id integer, name text, [haha gotcha] text);"
...     )
>>> print get_col_names(curs, "foo")
[u'id', u'name', u'haha gotcha']
>>> print wrong_way(curs, "foo")
[u'id', u'name', u'[haha'] <<<<<===== WHOOPS!
>>>

Other problems with the now-deleted “parse the create table SQL” answer:

1. Stuffs up with e.g. create table test (id1 text, id2 int, msg text, primary key(id1, id2)) … needs to ignore not only CONSTRAINT, but also keywords PRIMARY, UNIQUE, CHECK and FOREIGN (see the create table docs).

2. Needs to specify re.DOTALL in case there are newlines in the SQL.

3. In line.split()[0].strip() the strip is redundant.

This is simple and works fine for me.

Lets say you have already connected to your database table and also got a cursor object. So following on on from that point.

import csv
curs = conn.cursor()
curs.execute("select * from oders")
m_dict = list(curs.fetchall())

with open("mycsvfile.csv", "wb") as f:
    w = csv.DictWriter(f, m_dict[0].keys())
    w.writerow(dict((fn,fn) for fn in m_dict[0].keys()))
    w.writerows(m_dict)

It can be easily done using pandas and sqlite3. In extension to the answer from Cristian Ciupitu.

import sqlite3
from glob import glob; from os.path import expanduser
conn = sqlite3.connect(glob(expanduser('data/clients_data.sqlite'))[0])
cursor = conn.cursor()

Now use pandas to read the table and write to csv.

clients = pd.read_sql('SELECT * FROM clients' ,conn)
clients.to_csv('data/Clients100914.csv', index=False)

This is more direct and works all the time.

The below code works for Oracle with Python 3.6 :

import cx_Oracle
import csv

# Create tns
dsn_tns=cx_Oracle.makedsn('<host>', '<port>', service_name='<service_name>')

# Connect to DB using user, password and tns settings
conn=cx_Oracle.connect(user='<user>', password='<pass>',dsn=dsn_tns)
c=conn.cursor()

#Execute the Query
c.execute("select * from <table>")

# Write results into CSV file
with open("<file>.csv", "w", newline='') as csv_file:
    csv_writer = csv.writer(csv_file)
    csv_writer.writerow([i[0] for i in c.description]) # write headers
    csv_writer.writerows(c)