- January 8, 2009
- sql db db-api fetchmany
- 0 (after 0 ratings)
When executing custom sql, the temptation is to use fetchall or fetchone, since the API for fetchmany is a bit awkward. (fetchall makes all records resident in client memory at once; fetchone takes a network round-trip to the DB for each record.)
This snippet, hoisted from django.db.models.sql.query.results_iter, presents a nice, simple iterator over multiple fetchmany calls which hits a sweet spot of minimizing memory and network usage.
1 2 3 4 5 6 7 8 9 10 11 12
def iter_fetchmany(cursor, size=100): """ Handles the awkwardness of iterating using fetchmany, and exposes all records as a (generated) iterable. Assumes django.db is the connection for the given cursor. FIXME for multi-db support. """ from django.db import connection rowset_iter = iter((lambda: cursor.fetchmany(size)), connection.features.empty_fetchmany_value) for rows in rowset_iter: for row in rows: yield row