simplify?

[ltalirz]

sqlite2duckdb/sqlite2duckdb/sqlite_to_duckdb.py

Lines 32 to 46 in c318ead

	## Get sqlite Names
	conn.sql("USE __other")
	tables = [i[0] for i in conn.sql("SHOW tables").fetchall()]
	print(f"{len(tables)} tables found(s)")
	conn.sql(f"USE {db_name}")
	# Create tables
	for table in tables:
	print(f"Create duckdb table {table}")
	conn.sql(f"CREATE TABLE {table} AS select * FROM __other.{table}")
	conn.sql(f"DETACH __other")
	conn.close()
	end_time = time.perf_counter()
	execution_time = (end_time - start_time) * 1000

This seems overly complicated. What about something like

# Attach the DuckDB database (named)
conn.execute(f"ATTACH '{duckdb_file}' AS duck_db;")
# Attach the SQLITE database (named)
conn.execute(f"ATTACH '{sqlite_db}' as sqlite_db;")

conn.execute("COPY FROM DATABASE sqlite_db TO duck_db;")

[dridk]

Ok ! Looks easier indeed .
My package is obsolete now :)

[danbirken]

I have a very half-hearted fork that does this (it needs a couple more tweaks from above): https://github.com/danbirken/sqlite2duckdb

I'm not going to do a pull request because I honestly don't know enough about what minor differences there might be between the "COPY" method and the existing method in this library.

But if the author or somebody else wants to figure that out, my fork uses the "COPY" command and I'm using it for my needs (though the duckdb database is much, much larger... I guess all those keys add up!)

[dridk]

Sorry guy,

I didn't check my GitHub notification.
I m going to fix that.