sql语句在python中的运行速度比命令行sqlite3慢得多

Question

I have a sql script that creates 2 temp tables, then creates 2 permanent tables based on the temp tables (4 steps). No rows are returned. When that script is run from the command line using $ sqlite3 MyDB.db -init Summarize.sql it takes less than a minute.
When the sql is run as a big string in python (via the sqlite3 standard module) connection.executescript("""<contents of Summarize.sql here>""") it takes 15 minutes. (The first 3 steps take 1 minute, the last step takes the remaining 14.)

Given that it's the exact same sql, what could python be doing that would slow it down so much? How can I speed this up? (The last step is a drop table if exists xxx; create table xxx as select ... It's a join, but it's the same join that the command line runs quickly.) A friend suggested the apsw module, but I'm loathe to switch without strong cause.

Answer 1

Calling SQLite from Python is about 100 times slower versus a command line call in my tests. Here is a solution. I just output my query to a text file sql.txt and invoke a command line SQLite call from Python:

import subprocess
import sys

print subprocess.Popen("sqlite3 reviews.db < sql.txt", shell=True, stdout=subprocess.PIPE).stdout.read()

You can add multiple queries to sql.txt and call other files by using .read filename command