philadams-zz / pluckr

pluck columns from csv files in the commandline. like cut or awk, but don't choke on delimiter escaping.

pluckr

Pluck columns from csv files in the commandline. Like cut or awk, but without choking on delimiter escaping.

install

pip install pluckr. Done.

usage

Grab columns 2 and 3 from stdin (-f is one-indexed):

pluckr -f 2,3 < sample1.csv

Or straight from a file:

pluckr -f 2,1 sample1.csv

Your requested order is retained:

pluckr -f 1,2 sample1.csv != pluckr -f 2,1 sample1.csv

You can use Python's negative indices... grab the first and last cols:

pluckr -f 1,-1 sample1.csv

And you can mix in ranges for column selection:

pluckr -f 2,4-6,9 sample1.csv

Can't be sure which columns exist or what their indices are?

pluckr --names sample1.csv

Fields that don't exist (e.g. -f11111) will be ignored.

Preprended line numbers can be helpful:

cat sample1.csv | pluckr --line-numbers

Skip header row(s):

cat sample1.csv | pluckr -s1

Grab all but the 2nd column (drop columns instead of selecting them):

pluckr -f 1 -i < sample2.csv

Read (and write) various delimiters:

pluckr -d, --out-delimiter=\| < with-commas.csv > with-pipes.csv

Note that passing tabs as arguments can be awkward; in a pinch, use:

pluckr -d $'\t'

help

Via --help:

usage: pluckr [-h] [-f FIELDS] [-i] [-d DELIMITER] [-q QUOTECHAR] [-s SKIP]
              [-l] [-n]
              [infile]

Grab columns from csv input. http://github.com/philadams/pluckr

positional arguments:
  infile                input file (.csv)

optional arguments:
  -h, --help            show this help message and exit
  -f FIELDS             ordered list of columns to retain; one-indexed
  -i, --inverse         invert column retention: drop those in -f
  -d DELIMITER, --delimiter DELIMITER
                        field delimiter when reading infile
  -q QUOTECHAR, --quotechar QUOTECHAR
                        field quotechar when reading infile
  -s SKIP, --skip SKIP  number of rows to skip
  -l, --line-numbers    prepend line numbers to output
  -n, --names           print column names; assumes one header row

tests

This project uses roundup for testing. Run tests with make test.

future

• allow -f to take columns by name when there's a header row?
• implement --out-x for output delimiters etc.
• json output? or, contribute a cli to kennethreitz/tablib
• add out delimiter support
• add out quotechar support