App::td - Manipulate table data


This document describes version 0.100 of App::td (from Perl distribution App-td), released on 2020-05-29.




 td(%args) -> [status, msg, payload, meta]

Manipulate table data.

td receives table data from standard input and performs an action on it. It has functionality similar to some Unix commands like head, tail, wc, cut, sort except that it operates on table rows/columns instead of lines/characters. This is convenient to use with CLI scripts that output table data.

A table data is JSON-encoded data in the form of either: hos (hash of scalars, which is viewed as a two-column table where the columns are key and value), aos (array of scalars, which is viewed as a 1-column array where the column is elem), aoaos (array of arrays of scalars), or aohos (array of hashes of scalars).

The input can also be an enveloped table data, where the envelope is an array: [status, message, content, meta] and content is the actual table data. This kind of data is produced by Perinci::CmdLine-based scripts and can contain more detailed table specification in the meta hash, which td can parse.

First you might want to use the info action to see if the input is a table data:

 % osnames -l --json | td info

If input is not valid JSON, a JSON parse error will be displayed. If input is valid JSON but not a table data, another error will be displayed. Otherwise, information about the table will be displayed (form, number of columns, column names, number of rows, and so on).

Next, you can use these actions:

 # List available actions
 % td actions

 # Convert table data (which might be hash, aos, or aohos) to aoaos form
 % list-files -l --json | td as-aoaos

 # Convert table data (which might be hash, aos, or aoaos) to aohos form
 % list-files -l --json | td as-aohos

 # Convert table data to CSV
 % list-files -l --json | td as-csv

 # Calculate arithmetic average of numeric columns
 % list-files -l --json | td avg

 # Append a row at the end containing arithmetic average of number columns
 % list-files -l --json | td avg-row

 # Count number of columns
 % osnames -l --json | td colcount

 # Append a single-column row at the end containing number of columns
 % osnames -l --json | td colcount-row

 # Return the column names only
 % lcpan related-mods Perinci::CmdLine | td colnames

 # append a row containing column names
 % lcpan related-mods Perinci::CmdLine | td colnames-row

 # Only show first 5 rows
 % osnames -l --json | td head -n5

 # Show all but the last 5 rows
 % osnames -l --json | td head -n -5

 # Check if input is table data and show information about the table
 % osnames -l --json | td info

 # Count number of rows
 % osnames -l --json | td rowcount
 % osnames -l --json | td wc            ;# shorter alias

 # Append a single-column row containing row count
 % osnames -l --json | td rowcount-row
 % osnames -l --json | td wc-row        ;# shorter alias

 # Add a row number column (1, 2, 3, ...)
 % list-files -l --json | td rownum-col

 # Select some columns
 % osnames -l --json | td select value description

 # Select all columns but some
 % osnames -l --json | td select '*' -e value -e description

 # Return the rows in a random order
 % osnames -l --json | td shuf

 # Pick 5 random rows from input
 % osnames -l --json | td shuf -n5

 # Sort by column(s) (add "-" prefix to for descending order)
 % osnames -l --json | td sort value tags
 % osnames -l --json | td sort -- -value

 # Return sum of all numeric columns
 % list-files -l --json | td sum

 # Append a sum row
 % list-files -l --json | td sum-row

 # Only show last 5 rows
 % osnames -l --json | td tail -n5

 # Show rows from the row 5 onwards
 % osnames -l --json | td tail -n +5

 # Transpose table (make first column of rows as column names in the transposed
 # table)

 % osnames -l --json | td transpose

 # Transpose table (make columns named 'row1', 'row2', 'row3', ... in the
 # transposed table)

 % osnames -l --json | td transpose --no-header-column

 # Use Perl code to filter rows. Perl code gets row in $row or $_
 # (scalar/aos/hos) or $rowhash (always a hos) or $rowarray (always aos). There
 # are also $rownum (integer, starts at 0) and $td (table data object). Perl
 # code is eval'ed in the 'main' package with strict/warnings turned off. The
 # example below selects videos that are larger than 480p.

 % media-info *.mp4 | td grep 'use List::Util qw(min); min($I<< ->{video >>height}, $I<< ->{video >>width}) > 480'

 # Use Perl code to filter columns. Perl code gets column name in $colname or
 # $_. There's also $colidx (column index, from 1) and $td (table data object).
 # If table data form is 'hash' or 'aos', it will be transformed into 'aoaos'.
 # The example below only select even columns that match /col/i. Note that most
 # of the time, 'td select' is better. But when you have a lot of columns and
 # want to select them programmatically, you have grep-col.

 % somecd --json | td grep-col '$colidx % 2 == 0 && /col/i'

 # Use Perl code to transform row. Perl code gets row in $row or $_
 # (scalar/hash/array) and is supposed to return the new row. As in 'grep',
 # $rowhash, $rowarray, $rownum, $td are also available as helper. The example
 # below adds a field called 'is_landscape'.

 % media-info *.jpg | td map '$I<< ->{is >>landscape} = $I<< ->{video >>height} < $I<< ->{video >>width} ? 1:0; $_'

 # Use perl code to sort rows. Perl sorter code gets row in $a & $b or $I<[0] &
 # $>[1] (hash/array). Sorter code, like in Perl's standard sort(), is expected
 # to return -1/0/1. The example belows sort videos by height, descendingly then
 # by width, descendingly.

 % media-info *.mp4 | td psort '$b->{video_height} <=> $a->{video_height} || $b->{video_width} <=> $b->{video_width}'

This function is not exported.

Arguments ('*' denotes required arguments):

  • action* => str

    Action to perform on input table.

  • argv => array[str] (default: [])


  • detail => bool

  • exclude_columns => array[str]

  • lines => str

  • no_header_column => true

    Don't make the first column as column names of the transposed table; instead create column named 'row1', 'row2', ...

  • repeat => bool

    Allow duplicates.

Returns an enveloped result (an array).

First element (status) is an integer containing HTTP status code (200 means OK, 4xx caller error, 5xx function error). Second element (msg) is a string containing error message, or 'OK' if status is 200. Third element (payload) is optional, the actual result. Fourth element (meta) is called result metadata and is optional, a hash that contains extra information.

Return value: (any)


Please visit the project's homepage at


Source repository is at


Please report any bugs or feature requests on the bugtracker website

When submitting a bug or request, please include a test-file or a patch to an existing test-file that illustrates the bug or desired feature.


Rinci::function for a more detailed explanation on enveloped result.

TableDef for more detailed explanation of table data definition, which can be specified in enveloped result's `meta` hash in the `table` key (see Perinci::Sub::Property::result::table).




perlancar <>


This software is copyright (c) 2020, 2019, 2017, 2016, 2015 by

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.