CSV file format

csv is the most basic file format to store tabular data, where all the values are strings and are separated by a delimiter (typically comma). dlt uses it for specific use cases - mostly for the performance and compatibility reasons.

Internally we use two implementations:

• pyarrow csv writer - very fast, multithreaded writer for the arrow tables
• python stdlib writer - a csv writer included in the Python standard library for Python objects

Supported Destinations

The csv format is supported by the following destinations: Postgres, Filesystem, Snowflake

How to configure

There are several ways of configuring dlt to use csv file format for normalization step and to store your data at the destination:

1. You can set the loader_file_format argument to csv in the run command:

info = pipeline.run(some_source(), loader_file_format="csv")

2. You can set the loader_file_format in config.toml or secrets.toml:

[normalize]
loader_file_format="csv"

3. You can set the loader_file_format via environment variable:

export NORMALIZE__LOADER_FILE_FORMAT="csv"

4. You can set the file type directly in the resource decorator.

@dlt.resource(file_format="csv")
def generate_rows(nr):
    pass

Default Settings

dlt attempts to make both writers to generate similarly looking files

• separators are commas
• quotes are " and are escaped as ""
• NULL values both are empty strings and empty tokens as in the example below
• UNIX new lines are used
• dates are represented as ISO 8601
• quoting style is "when needed"

Example of NULLs:

text1,text2,text3
A,B,C
A,,""

In the last row both text2 and text3 values are NULL. Python csv writer is not able to write unquoted None values so we had to settle for ""

Note: all destinations capable of writing csvs must support it.

Change settings

You can change basic csv settings, this may be handy when working with filesystem destination. Other destinations are tested with standard settings:

• delimiter: change the delimiting character (default: ',')
• include_header: include the header row (default: True)
• quoting: quote_all - all values are quoted, quote_needed - quote only values that need quoting (default: quote_needed)

When quote_needed is selected: in case of Python csv writer all non-numeric values are quoted. In case of pyarrow csv writer, the exact behavior is not described in the documentation. We observed that in some cases, strings are not quoted as well.

[normalize.data_writer]
delimiter="|"
include_header=false
quoting="quote_all"

Or using environment variables:

NORMALIZE__DATA_WRITER__DELIMITER=|
NORMALIZE__DATA_WRITER__INCLUDE_HEADER=False
NORMALIZE__DATA_WRITER__QUOTING=quote_all

Destination settings

A few additional settings are available when copying csv to destination tables:

• on_error_continue - skip lines with errors (only Snowflake)
• encoding - encoding of the csv file

tip

You'll need those setting when importing external files

Limitations

arrow writer

• binary columns are supported only if they contain valid UTF-8 characters
• json (nested, struct) types are not supported

csv writer

• binary columns are supported only if they contain valid UTF-8 characters (easy to add more encodings)
• json columns dumped with json.dumps
• None values are always quoted