Skip to main content

A thread safe queue worker that executes SQL for multi-threaded applications.

Project description

Sqlite3 Execution Queue

version license pyversions
donate powered made

A thread safe queue worker that executes SQL for multi-threaded applications.

Hierarchy

sqlqueue
'---- SqlQueue()
    |---- sql()
    |---- _sql()
    |---- commit()
    '---- stop()

Example

python

from sqlq import *

# specify the db file, relative or absolute path
# set server=True
sqlqueue = SqlQueue(server=True, db="db.db")

# SQL execution modes
# all will return the executed SQL result immediately
SqlQueue(server=True, db="default", timeout_commit=1000).sql("SELECT * FROM table;")
SqlQueue(server=True, db="commit per 1ms", timeout_commit=1).sql("INSERT INTO table VALUES (?);", (0,))
SqlQueue(server=True, db=r"C:\somewhere\db.db").sql("INSERT INTO table VALUES (?);", ((0,),(0,)))
SqlQueue(server=True, db="../../data/db.db").sql('''
CREATE TABLE "tablea" ("a" TEXT);
DELETE TABLE "table";
''')

# stop the worker
# use it at your own risk
# otherwise data will be lost
# always commit before stopping it
sqlqueue.commit()
sqlqueue.stop()

# using sqlq with encryptedsocket
# server
from encryptedsocket import SS
from easyrsa import *
sqlqueueserver = SqlQueue(server=True, db=r"db.db")
threading.Thread(target=SS(EasyRSA(bits=1024).gen_key_pair(), sqlqueueserver.functions).start).start()
# client
sqlqueue = SqlQueue()
for i in range(10):
    # SqlQueue._sql() must not be used in socket mode
    sqlqueue.sql("INSERT INTO test VALUES (?);", (str(i)))
# server
sqlqueueserver.commit()
sqlqueueserver.stop()

# SQL execution speed
# # server mode _sql() without built-in ThreadWrapper() handler
# SqlQueue(server=True, ...)._sql() <
# # server mode sql() with built-in ThreadWrapper() handler
# SqlQueue(server=True, ...).sql() <
# # client mode sql() with two nested ThreadWrapper() handlers
# SqlQueue().sql()

# this example shows how sqlq is used
# SQL should not be executed frequently
r = (1, 5, 10, 50, 100, 200)
r = (50,)
for l in r:
    tw = ThreadWrapper(threading.Semaphore(l))
    starttime = time.time()
    result = {}  # result pool
    for i in range(l):
        def job(i):
            # return SQL execution result to result pool
            return sqlqueue._sql(threading.get_ident(), "INSERT INTO test VALUES (?);", (str(i),))
        tw.add(job, args=args(i), result=result, key=i)  # pass the pool and uid in
    tw.wait()
    # p(result)
    p(l, (time.time()-starttime)/l, time.time()-starttime)
    p(sqlqueue.sql("SELECT * FROM test;"))
    tw = ThreadWrapper(threading.Semaphore(l))
    starttime = time.time()
    for i in range(l):
        def job(i):
            sqlqueue._sql(threading.get_ident(), f"DELETE FROM test WHERE a = ?;", (str(i),))
        tw.add(job, args=args(i))
    tw.wait()
    p(l, (time.time()-starttime)/l, time.time()-starttime)
    # in order to use SqlQueue()._sql(), ThreadWrapper() is
    # recommended to queue threads, check threadwrapper for more info
    # SqlQueue()._sql() will raise execution error
    # it will only report it to the result
    # you should handle the errors separately


    starttime = time.time()
    for i in range(l):
        sqlqueue.sql("INSERT INTO test VALUES (?);", (str(i),))
    p(l, (time.time()-starttime)/l, time.time()-starttime)
    p(sqlqueue.sql("SELECT * FROM test;"))
    starttime = time.time()
    for i in range(l):
        sqlqueue.sql(f"DELETE FROM test WHERE a = ?;", (str(i),))
    p(l, (time.time()-starttime)/l, time.time()-starttime)
    p()
    sqlqueue.commit()  # manual commit
    # both manual and timeout commit always wait until
    # the current SQL execution is completed. 
    # the worker will not raise any error
    # however SqlQueue().sql() will re-raise execution error

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

sqlq-0.1.1.tar.gz (4.7 kB view details)

Uploaded Source

Built Distribution

sqlq-0.1.1-py3-none-any.whl (16.7 kB view details)

Uploaded Python 3

File details

Details for the file sqlq-0.1.1.tar.gz.

File metadata

  • Download URL: sqlq-0.1.1.tar.gz
  • Upload date:
  • Size: 4.7 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.1.1 pkginfo/1.5.0.1 requests/2.23.0 setuptools/46.1.3 requests-toolbelt/0.9.1 tqdm/4.45.0 CPython/3.7.0

File hashes

Hashes for sqlq-0.1.1.tar.gz
Algorithm Hash digest
SHA256 40fb60af34e8ae38b357e2f57d64caacf13fc26e421855dc98f705b48bc8463d
MD5 1d3c9d8cc93e8a5caa0e11ae2e80a19e
BLAKE2b-256 fb058c7fd1369ba0ee19f589171ec83cc1b5ea0a837d15b558fa3f1dd271b7fe

See more details on using hashes here.

File details

Details for the file sqlq-0.1.1-py3-none-any.whl.

File metadata

  • Download URL: sqlq-0.1.1-py3-none-any.whl
  • Upload date:
  • Size: 16.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.1.1 pkginfo/1.5.0.1 requests/2.23.0 setuptools/46.1.3 requests-toolbelt/0.9.1 tqdm/4.45.0 CPython/3.7.0

File hashes

Hashes for sqlq-0.1.1-py3-none-any.whl
Algorithm Hash digest
SHA256 cd6e3b6e688f715e80804bb7d0f0ed056532c327138203c5bf946f27e5cba664
MD5 5be71afb3693c7478f90f7a7efd39b66
BLAKE2b-256 21e79d90a94354496c4c285a303540b208b1e4e95bdf9e90eccbfa50ea118d53

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page