comma

View documentation for the original command

$ tldr ,

comma [OPTION...] [mapping]

-h, --help
    display this help and exit

--version
    output version information and exit

--delimiter=CHAR
    set field delimiter (default: ,)

--quote=CHAR
    set quote character (default: ")

--empty-as-zero
    treat empty fields as zero

--fields=FIELDS
    comma-separated list of input fields

--output-fields=FIELDS
    comma-separated list of output fields

--header
    expect and output header line

--sort
    sort records by all fields

--precision=DIGITS
    floating-point output precision

--trailing-delimiter
    output trailing delimiter after last field

--line-number
    add line number as first field

--flatten
    flatten nested structures

The comma command is a powerful Linux utility for serializing and deserializing data streams in CSV format. It excels in command-line pipelines, allowing users to define the structure of tabular data using a simple mapping syntax like field1=int32,field2=float64,time. This enables type-safe conversions, filtering, sorting, and transformation of data on the fly.

Comma reads from stdin or files, processes according to the specified mapping and options, and outputs to stdout. It supports both text CSV input/output and binary formats, making it ideal for high-performance scientific computing, robotics, and data processing workflows. For example, it can convert numeric CSV to binary for efficient storage or vice versa.

Key strengths include precise field selection with --fields and --output-fields, handling of headers, quoting, delimiters, and special types like time (nanoseconds since epoch). It automatically infers missing values and supports sorting without loading entire datasets into memory. Widely used in projects requiring robust stream processing, comma bridges simple CSV handling with advanced data manipulation.

Requires explicit mapping for typed processing; large datasets may need careful memory management with --sort. Not for interactive use.

Specify structure as field=type[,field=type...]. Types: int8,uint16,float32,double,time, etc.
Ex: comma x=float,y=float,time

Convert CSV to binary: cat data.csv | comma x,y | gzip > data.csv.gz
Filter fields: comma --fields=x,y,z input.csv

Developed by Peter Mitrano around 2010 as part of the Comma library for robotics and scientific data pipelines, particularly in ROS ecosystems. Evolved for efficient stream processing; stable since version 2.x.

awk(1), cut(1), mlr(1), paste(1)