duckdb – 李于跃龙门

Base:
1. import duckdb
  1. duckdb.sql(« SELECT 42 »).show()
2. r1 = duckdb.sql(« SELECT 42 AS i »)
  1. duckdb.sql(« SELECT i * 2 AS k FROM r1 »).show()
Data Input
1. import duckdb
2. duckdb.read_csv(« example.csv ») # read a CSV file into a Relation
3. duckdb.read_parquet(« example.parquet ») # read a Parquet file into a Relation
4. duckdb.read_json(« example.json ») # read a JSON file into a Relation
5. duckdb.sql(« SELECT * FROM ‘example.csv' ») # directly query a CSV file
6. duckdb.sql(« SELECT * FROM ‘example.parquet' ») # directly query a Parquet file
7. duckdb.sql(« SELECT * FROM ‘example.json' ») # directly query a JSON file
Pandas
1. pandas_df = pd.DataFrame({« a »: [42]})
2. duckdb.sql(« SELECT * FROM pandas_df »)
Polars
1. polars_df = pl.DataFrame({« a »: [42]})
2. duckdb.sql(« SELECT * FROM polars_df »)
PyArrow
1. arrow_table = pa.Table.from_pydict({« a »: [42]})
2. duckdb.sql(« SELECT * FROM arrow_table »)
Result Conversion
1. duckdb.sql(« SELECT 42 »).fetchall() # Python objects
2. duckdb.sql(« SELECT 42 »).df() # Pandas DataFrame
3. duckdb.sql(« SELECT 42 »).pl() # Polars DataFrame
4. duckdb.sql(« SELECT 42 »).arrow() # Arrow Table
5. duckdb.sql(« SELECT 42 »).fetchnumpy() # NumPy Arrays
writing data to disk
1. duckdb.sql(« SELECT 42 »).write_parquet(« out.parquet ») # Write to a Parquet file
2. duckdb.sql(« SELECT 42 »).write_csv(« out.csv ») # Write to a CSV file
3. duckdb.sql(« COPY (SELECT 42) TO ‘out.parquet' ») # Copy to a Parquet file
Using In Memory
1. con = duckdb.connect()
2. con.sql(« SELECT 42 AS x »).show()
persistent Storage
1. # create a connection to a file called ‘file.db’
2. con = duckdb.connect(« file.db »)
3. # create a table and load data into it
4. con.sql(« CREATE TABLE test (i INTEGER) »)
5. con.sql(« INSERT INTO test VALUES (42) »)
6. # query the table
7. con.table(« test »).show()
8. # explicitly close the connection
9. con.close()
10. # Note: connections also closed implicitly when they go out of scope
Configuration
1. con = duckdb.connect(config = {‘threads’: 1})

Laisser un commentaire