def setup_scd(dbname, number):
open(os.path.expanduser(dbname), 'w')
output_conn = sqlite3.connect(dbname)
output_cur = output_conn.cursor()
output_cur.execute("CREATE TABLE scd "
"(key INTEGER PRIMARY KEY, attr1 INTEGER, "
"attr2 INTEGER, version INTEGER)")
output_conn.commit()
output_wrapper = pygrametl.ConnectionWrapper(connection=output_conn)
scd = SlowlyChangingDimension(
name='scd',
key='key',
attributes=['attr1', 'attr2', 'version'],
lookupatts=['attr2'],
versionatt='version'
)
data = []
total_elapsed = 0.
print('Generating scd data')
start = time.monotonic()
for i in range(1, number + 1):
data.append({'attr1': i, 'attr2': number + 1 - i})
end = time.monotonic()
elapsed = end - start
print('Generated: {}{}'.format(round(elapsed, 3), 's'))
total_elapsed += elapsed
print('Inserting data into scd')
start = time.monotonic()
for row in data:
scd.scdensure(row)
end = time.monotonic()
elapsed = end - start
print('Inserted: {}{}'.format(round(elapsed, 3), 's'))
total_elapsed += elapsed
print('DW populated')
print('Total time: {}{}\n'.format(round(total_elapsed, 3), 's'))
output_conn.commit()
cid_map = {}
for row in country_src:
cid = country_dim.ensure(row)
cid_map[row["city"]] = cid
# We populate the authordim and the fact table
for row in author_src:
if row["city"] in ["Hadsten", "Skanderborg", "Kobenhavn"]:
row["cid"] = cid_map[row["city"]]
else:
row["cid"] = None
row["name"] = row["firstname"] + " " + row["lastname"]
row.pop("aid", 0) # Gets rid of aid so that pygrametl can generate them
# Placing new row in author_dim
row["aid"] = author_dim.ensure(row)
# Placing new row in fact_table
fact_table.ensure(row)
# Places books directly into book_dim
for row in book_src:
book_dim.scdensure(row)
wrapper.commit()
wrapper.close()
author_conn.close()
book_conn.close()
country_handle.close()
