feat: expose load from csv file via Python driver

[rad-pat]

Currently, only exposed is to load a data array via stream_load method on Python driver which involves reading the data from file and passing into the method
It would be good to have a method to load from CSV file with options, and/or the ability to stage a file with the driver bindings.

[everpcpc]

How about adding support to stream_load a pandas DataFrame?

[rad-pat]

We're looking for the fastest/most optimal way to load data into Databend. If we have a local CSV file, what is the best option? Currently we're pushing the file to cloud storage and executing COPY INTO statement.
With Greenplum/Postgres - we could feed the file through STDIN into COPY command.

[everpcpc]

If you are just seeking to load files, then you may try out with bendsql:

bendsql --query='INSERT INTO http_books_02 VALUES;' --format=csv --data=@cli/tests/data/books.csv

# or with STDIN

bendsql --query='INSERT INTO http_books_01 VALUES;' --format=csv --data=@- <cli/tests/data/books.csv

# or with more options

bendsql \
    --query='INSERT INTO http_ontime_03 VALUES;' \
    --format=csv \
    --format-opt="compression=gzip" \
    --format-opt="skip_header=1" \
    --data=@cli/tests/data/ontime_200.csv.gz

[rad-pat]

Yep, that's good for a test. How can I replicate with Python driver, passing the file name?

[everpcpc]

We will add support for python driver in next release.