问题
So I've read the documentation and it says:
Cursor.rowcount¶
Although the Cursor class of the sqlite3 module implements this attribute, the database engine’s own support for the determination of “rows affected”/”rows selected” is quirky.
This includes SELECT statements because we cannot determine the number of rows a query produced until all rows were fetched.
How come, immediately after executing a SELECT, I can't use this to determine the number of rows returned. Surely the query, having been executed, has now fetched all the rows?
回答1:
SQLite does not "fetch" all the rows when initially processing the query. It returns a single row each time the prepared query is stepped.
From the SQLite3 FAQ here
Q) Which func could get the number of rows?
A) There is no function to retrieve the number of rows in a result set. SQLite doesn't know the number in advance, but returns row by row while iterating through the tables. The application can increment a row counter as needed at every successful sqlite3_step() .
Some wrappers are able to collect all rows in a resultset in a in-memory table, so they can return the number of rows.
You can always get the number of rows that a certain SELECT statement would return at the cost of some performance:
BEGIN IMMEDIATE TRANSACTION;
SELECT COUNT(*) FROM x WHERE y;
SELECT a,b,c FROM x WHERE y;
ROLLBACK TRANSACTION;
You have to wrap this in a transaction to prevent other connections from inserting / deleting rows between the two SELECT statements.
来源：https://stackoverflow.com/questions/4911404/in-pythons-sqlite3-module-why-cant-cursor-rowcount-tell-me-the-number-of-ro