Insert binary file in SQLite database with Python

Question

I'm trying to write a simple Python script that inserts .odt documents into an SQLite database. Here is what I have done so far, but it doesn't seem to work:

f=open('Loremipsum.odt', 'rb')
k=f.read()
f.close()
cursor.execute="INSERT INTO notes (note) VALUES ('%s')" %(sqlite.Binary(k))
cursor.close()
conn.close()

I don't get any error messages, but as far as I can see the record is not inserted. What am I doing wrong? Also, how can I extract the stored document back? Thanks!

Solution 1

Not sure what is that sqlite.Binary you're using, but, anyway, here's a working example:

import sqlite3

# let's just make an arbitrary binary file...
with open('/tmp/abin', 'wb') as f:
  f.write(''.join(chr(i) for i in range(55)))
# ...and read it back into a blob
with open('/tmp/abin', 'rb') as f:
  ablob = f.read()

# OK, now for the DB part: we make it...:
db = sqlite3.connect('/tmp/thedb')
db.execute('CREATE TABLE t (thebin BLOB)')
db.execute('INSERT INTO t VALUES(?)', [buffer(ablob)])
db.commit()
db.close()

# ...and read it back:
db = sqlite3.connect('/tmp/thedb')
row = db.execute('SELECT * FROM t').fetchone()
print repr(str(row[0]))

When run with Python 2.6, this code shows, as expected and desired: '\x00\x01\x02\x03\x04\x05\x06\x07\x08\t\n\x0b\x0c\r\x0e\x0f\x10\x11\x12\x13\x14\x15\x16\x17\x18\x19\x1a\x1b\x1c\x1d\x1e\x1f !"#$%&\'()*+,-./0123456'

Note the need to use buffer to insert the blob, and str to read it back as a string (since it uses the buffer type as a result as well) -- if you're just going to write it to disk the latter passage would not be needed (since the write method of files does accept buffer objects just as well as it accepts strings).

Solution 2

There are multiple problems with the given example. I will address them one by one.

• There is no error checking. We either need to use the try/except/finally construct or use the with keyword.
• Python methods are not like C# properties. You are not running the execute method, you are assigning some string to an object. (In Python, methods are objects too.)
• Very important is that your code is subject to SQL Injection attacks. We should never build SQL statements using Python string operations. We should always use placeholders.
• The example is incomplete. Which leads to a tricky issue. Supposing, that there was a CREATE TABLE statement then a new implicit transaction would be created. And a commit statement must be issued to save the data to the database file. In SQLite, any statement other than SELECT starts an implicit transaction. (Some databases, like MySQL, are in the autocommit mode by default. This is not true for SQLite.)

Here is a proper working example, which writes a LibreOffice document to a docs table of an SQLite database:

#!/usr/bin/env python

import sqlite3


def readData():

    fl = open('book.odt', 'rb')

    with fl:
        data = fl.read()

    return data


con = sqlite3.connect('test.db')

with con:

    cur = con.cursor()     
    cur.execute("CREATE TABLE IF NOT EXISTS docs(Data BLOB)")

    data = readData()

    sql = "INSERT INTO docs(Data) VALUES (?)" 
    cur.execute(sql, (sqlite3.Binary(data), ))

The book.odt file is located in the current working directory. We did not call the commit method manually, since this is handled by the with keyword behind the scenes.

Edit removed the lite alias

Solution 3

Problems:

1. You didn't show the full code that you ran. You shouldn't leave answerers guessing what things like sqlite.Binary(k).

2. Fundamental problem: You didn't commit your transaction. Use conn.commit() before conn.close().

Comments

[dmpop]

Perfect! Thank you so much for your help!

[Thiago Padilha]

Nice example, I have a similar question (stackoverflow.com/questions/3915888/…), I would be very grateful if you could take a look at it.

[Hadi]

I get the following error "sqlite3.InterfaceError: Error binding parameter 7 - probably unsupported type."

[tshepang]

Why access sqlite3 as lite?

[tshepang]

But you only use it once. It's too strange... I don't see a win. If we do a character count, it does not shorten the code.

[Jan Bodnar]

Twice, actually. This is more a matter of a personal taste. Rather than a strict rule. Say in two years, there will be sqlite4 module. And we don't need to correct all the occurrences of the module name in older code. However, such things are irrelevant in small code examples.

[Jan Bodnar]

Binary is an alias for Python buffer.

[leetNightshade]

Having it as sqlite instead of lite would improve readability and still be "future proof" with your method. Or just have it as 'sql', as 'lite' doesn't really mean anything and 'sql' is a standard for database manipulation.

[Gleno]

I usually shorten it to "db", because having a number in a name feels so wrong somehow.

[toonarmycaptain]

...k is the data from f.read and sqlite.Binary is a function in the sqlite3 module, what's to guess about that? [NB I'm not sure what sqlite.Binary does over Path.read_bytes(), but what OP is running seems evident there.]

[Kennet Celeste]

buffer in python3 is now memoryview