Question

I use the following python3 code to add and update entries in an sqlite3 database:

def increment_person_counts(count_per_person):
   with sqlite3.connect(r'./people_database') as connection:
      cursor = connection.cursor()
      for person in count_per_person:
         if cursor.execute('select * from personCounts where person = ?', [person]).fetchone()==None:
            cursor.execute('insert into personCounts(person, count) values (?, ?)', [person, count_per_person[person]])
         else:
            cursor.execute('update personCounts SET count=count + ? WHERE person=?', [count_per_person[person], person])
      connection.commit()

count_per_person contains 4 million entries, and I seem to be able to add/update around 100 entries per second which means it will take half a day to add these values. Is there a better/faster approach to doing this that I should consider?

Thanks for your help,

Barry

Answer 1

You can read your whole 'select * from personCounts' into a python set() at the beginning and then check just against this set.

def increment_person_counts(count_per_person):
   with sqlite3.connect(r'./people_database') as connection:
      cursor = connection.cursor()
      cursor.execute('select person from personCounts')
      known_persons = set(row[0] for row in cursor.fetchall())
      for person, count in count_per_person.iteritems():
         if person in known_persons:
            cursor.execute('insert into personCounts(person, count) values (?, ?)', [person, count])
         else:
            cursor.execute('update personCounts SET count=count + ? WHERE person=?', [count, person])
      connection.commit()

UPDATE: after my comment, here is the update with executemany:

def increment_person_counts(count_per_person):
    with sqlite3.connect(r'./people_database') as connection:
        cursor = connection.cursor()
        cursor.execute('select person from personCounts')
        known_persons = set(row[0] for row in cursor.fetchall())
        cursor.executemany('insert into personCounts(person, count) values (?, ?)', ((person, count) for count_per_person.iteritems() if person in known_persons))
        for person, count in count_per_person.iteritems():
            if person not in known_persons:
                cursor.execute('update personCounts SET count=count + ? WHERE person=?', [count, person])
        connection.commit()