GitLocker: The Coding Marketplace

Description:

csvinsight 0.4.0

csvinsight

Fast & simple summary for large CSV files

Free software: MIT license
Documentation: https://csvinsight.readthedocs.io.

Features

Calculates basic stats for each column: max, min, mean length; number of non-empty values
Calculates exact number of unique values and the top 20 most frequent values
Supports non-orthogonal data (list fields)
Works with very large files: does not load the entire CSV into memory
Fast splitting of CSVs into columns, one file per column
Multiprocessing-enabled

Example Usage
Given a CSV file:
bash-3.2$ cat tests/sampledata.csv
name|age|fave_color
Alexey|33|red;yellow
Boris|31|blue
Valentina|0|
you can obtain a CsvInsight report with:
bash-3.2$ csvi tests/sampledata.csv --list-fields fave_color
CSV Insight Report
Total # Rows: 3
Column counts:
3 columns -> 3 rows

Report Format:
Column Number. Column Header -> Uniques: # ; Fills: # ; Fill Rate:
Field Length: min #, max #, average:
Top n field values -> Dupe Counts

1. name -> Uniques: 3 ; Fills: 3 ; Fill Rate: 100.0%
Field Length: min 5, max 9, avg 6.67
Counts Percent Field Value
1 33.33 % Valentina
1 33.33 % Boris
1 33.33 % Alexey

2. age -> Uniques: 3 ; Fills: 3 ; Fill Rate: 100.0%
Field Length: min 1, max 2, avg 1.67
Counts Percent Field Value
1 33.33 % 33
1 33.33 % 31
1 33.33 % 0

3. fave_color -> Uniques: 4 ; Fills: 3 ; Fill Rate: 75.0%
Field Length: min 0, max 6, avg 3.25
Counts Percent Field Value
1 25.00 % yellow
1 25.00 % red
1 25.00 % blue
1 25.00 % NULL
Since CSV comes in different flavors, you may need to tweak the underlying CSV parser’s parameters to read your file successfully.
CSVInsight handles this via CSV dialects.
For example, to read a comma-separated file, you would use the following command:
bash-3.2$ csvi your/file.csv --dialect delimiter=,
You may combine as many dialect parameters as needed:
bash-3.2$ csvi your/file.csv --dialect delimiter=, quoting=QUOTE_NONE
For a full list of dialect parameters, see the documentation for Python’s csv module.
Constant values like QUOTE_NONE are resolved automagically.
Once you’ve discovered the winning parameter combination for your file, save it to a YAML file:
list_fields:
- fave_color
- another_field_name
list_separator: ;
dialect:
- "delimiter=|"
- "quoting=QUOTE_NONE"
You can then invoke CSVI as follows:
bash-3.2$ csvi your/file.csv --config your/config.yaml

Credits
This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.

History

Unreleased

0.4.0 (2024-04-09)

Drop support for Py2.7, get rid of dependency on six, futures imports, etc.
Drop support for Py3.7 and below
Get rid of ipynb feature
Correctly handle empty quotechar and escapechar dialect parameters
Correctly handle doublequote dialect parameter

0.3.3 (2020-12-02)

Handle numeric quoting parameter, e.g. “–dialect quoting=3”

0.3.2 (2019-07-01)

Set the field size limit to sys.maxsize

0.3.1 (2019-06-26)

Make Jupyter notebook an optional dependency

0.3.0 (2018-07-11)

Added –most-common parameter (resolved Issue #14)
Added –no-tiny parameter
Refactored temporary file naming
Improve error message when handling empty CSV files
Fixed “Argument list too long” error (Issue #15)
Added –json parameter
Added –ipynb parameter to generate IPython notebook

0.2.3 (2017-12-09)

Fix bug: Unicode column names now work under Py2

0.2.2 (2017-12-04)

Fix bug: Unicode characters no longer break CsvInsight on Py2

0.2.1 (2017-11-27)

Fix bug: opening gzipped files with Py3 now works

0.2.0 (2017-11-25)

Split files using gsplit and process them in parallel for faster processing
No longer work with streams; works exclusively with files
Get rid of csvi_summarize and csvi_split entry points
Integrated plumbum for cleaner pipelines
Fixed issue #11: added support for more CSV parameters via the –dialect option
Fixed issue #10: reading from empty files no longer raises StopIteration
Fixed issue #8: use the correct link to the GitHub project in the documentation
Fixed issue #2: implemented in-memory mode for smaller files

0.1.0 (2017-10-29)

First release on PyPI.