CODEWITHSUNDEEP
pythonsqlite
user2489252
user2489252

Reputation:

Using `executemany` to update entries in an existing SQLite3 database (using Python sqlite3)

I know that executemany can be used to conveniently add new entries to a database; useful to reduce the Python method-call overheads compared to single executes in a for loop. However, I am wondering if this could work with SQLite's UPDATE.

More concretely, consider the following setup:

cnx = sqlite3.connect(DATABASE)
c = cnx.cursor()
for path in paths:
    for data in some_computation(path):
        c.execute("UPDATE TABLENAME SET cont=? WHERE id=?", (data[1], data[0]))
cnx.commit()
cnx.close()

I am not even sure if the approach below would be any faster (would have to benchmark it), but the problem is that it doesn't work, because I am doing it incorrectly I assume. Any tips to use executemany in the code snippet below to accomplish the task that I posted above?

cnx = sqlite3.connect(DATABASE)
c = cnx.cursor()

for path in paths:
    data_ids, data_conts = [], []
    for data in some_computation(path):
        if len(data_ids) >= CHUNKSIZE:
            c.executemany("UPDATE TABLENAME SET cont=? WHERE id=?", (data_conts, data_ids))
            cnx.commit()
            data_ids, data_conts = [], []
        data_ids.append(data[0])
        data_conts.append(data[1])
    c.executemany("UPDATE TABLENAME SET cont=? WHERE id=?", (data_conts, data_ids))      
    cnx.commit()

cnx.commit()
cnx.close()

Many thanks for tips and insights!

EDIT 1:

The problem with the bottom example:

ProgrammingError: Incorrect number of bindings supplied. The current statement uses 2, and there are 50000 supplied.

(where CHUNKSIZE=50000)

EDIT 2:

The same error occurs

cnx = sqlite3.connect(DATABASE)
c = cnx.cursor()

for path in paths:
    data_conts = []
    for data in some_computation(path):
        if len(data_ids) >= CHUNKSIZE:
            c.executemany("UPDATE TABLENAME SET cont=? WHERE id=?", (data_conts,))
            cnx.commit()
            data_conts = []

        data_conts.append([data[1], data[0]])
    c.executemany("UPDATE TABLENAME SET cont=? WHERE id=?", (data_conts,))   
    cnx.commit()

cnx.commit()
cnx.close()

but thanks to @falsetru I then noticed my error, It should be 

... WHERE id=?", data_conts)

and not 

... WHERE id=?", (data_conts,))

Upvotes: 14

Views: 21133

Answers (2)

Eamonn Kenny
Eamonn Kenny

Reputation: 2072

what you have is perfect except that you should use zip(conts,ids) where conts and ids are lists. This does the rearrangement automatically for you.

Upvotes: 0

falsetru
falsetru

Reputation: 369494

You need to pass a sequence of sequences ([[cont,id], [cont,id], [cont,id], ...], not [cont, cont, cont, ...], [id, id, id, ..]):

for path in paths:
    params = []
    for data in some_computation(path):
        if len(data_ids) >= CHUNKSIZE:
            c.executemany("UPDATE TABLENAME SET cont=? WHERE id=?", params)
            cnx.commit()
            params = []
        params.append([data[1], data[0]])
    if params:
        c.executemany("UPDATE TABLENAME SET cont=? WHERE id=?", params)
    cnx.commit()

Upvotes: 25

Related Questions