.. _fwflag_gzip_csv:
============
--gzip_csv
============
Switch
======

--gzip_csv FILENAME

Description
===========

Extracts n-gram features directly from a gzipped CSV file of messages, instead of a message table given by :doc:`fwflag_t`.

Argument and Default Value
==========================

Path to a ``.csv.gz`` file. Default: ``''`` (not used).

Details
=======

Verified from source: this switch is only checked *inside* the :doc:`fwflag_add_ngrams` code path (``if args.addngrams: ... elif args.gzipcsv: ...``), so you must pass ``--add_ngrams`` together with ``--gzip_csv`` — it is not a standalone alternative to it. Routes to ``FeatureExtractor.addNGramTableGzipCsv()`` with a hardcoded column layout (message-id column 3, group-id column 0, message column 19) — this is a fixed-format bulk-import path for a specific corpus layout, not a general-purpose CSV reader. Use :doc:`fwflag_n` to control which n-gram sizes are extracted.

Other Switches
==============

Required Switches:

* :doc:`fwflag_d`, :doc:`fwflag_c`
* :doc:`fwflag_add_ngrams`

Optional Switches:

* :doc:`fwflag_n`

Example Commands
================

.. code-block:: bash

	dlatkInterface.py -d dla_tutorial -c user_id --add_ngrams --gzip_csv /path/to/messages.csv.gz -n 1