# Compressed data As of [Miller 6](new-in-miller-6.md), Miller supports reading GZIP, BZIP2, and ZLIB formats transparently, and in-process. And (as before Miller 6) you have a more general `--prepipe` option to support other decompression programs. ## Automatic detection on input If your files end in `.gz`, `.bz2`, or `.z` then Miller will autodetect by file extension:

file gz-example.csv.gz

gz-example.csv.gz: gzip compressed data, was "gz-example.csv", last modified: Mon Aug 23 02:04:34 2021, from Unix, original size modulo 2^32 429

mlr --csv sort -f color gz-example.csv.gz

color,shape,flag,k,index,quantity,rate
purple,triangle,false,5,51,81.2290,8.5910
purple,triangle,false,7,65,80.1405,5.8240
purple,square,false,10,91,72.3735,8.2430
red,square,true,2,15,79.2778,0.0130
red,circle,true,3,16,13.8103,2.9010
red,square,false,4,48,77.5542,7.4670
red,square,false,6,64,77.1991,9.5310
yellow,triangle,true,1,11,43.6498,9.8870
yellow,circle,true,8,73,63.9785,4.2370
yellow,circle,true,9,87,63.5058,8.3350

This will decompress the input data on the fly, while leaving the disk file unmodified. This helps you save disk space, at the cost of some additional runtime CPU usage to decompress the data. ## Manual detection on input If the filename doesn't in in `.gz`, `.bz2`, or `.z` then you can use the flags `--gzin`, `--bz2in`, or `--zin` to let Miller know:

mlr --csv --gzin sort -f color myfile.bin # myfile.bin has gzip contents

## External decompressors on input Using the `--prepipe` flag, you can provide the name of any decompression program in your `$PATH` and Miller will run it on each input file, effectively piping the standard output of that program to Miller's standard input. You can, of course, already do without this for single input files, for example:

gunzip < gz-example.csv.gz | mlr --csv sort -f color

color,shape,flag,k,index,quantity,rate
purple,triangle,false,5,51,81.2290,8.5910
purple,triangle,false,7,65,80.1405,5.8240
purple,square,false,10,91,72.3735,8.2430
red,square,true,2,15,79.2778,0.0130
red,circle,true,3,16,13.8103,2.9010
red,square,false,4,48,77.5542,7.4670
red,square,false,6,64,77.1991,9.5310
yellow,triangle,true,1,11,43.6498,9.8870
yellow,circle,true,8,73,63.9785,4.2370
yellow,circle,true,9,87,63.5058,8.3350

The benefit of `--prepipe` is that Miller will run the specified program once per file, respecting file boundaries. The prepipe command can be anything which reads from standard input and produces data acceptable to Miller. Nominally this allows you to use whichever decompression utilities you have installed on your system, on a per-file basis. If the command has flags, quote them: e.g. `mlr --prepipe 'zcat -cf'`. Note that this feature is quite general and is not limited to decompression utilities. You can use it to apply per-file filters of your choice: e.g. `mlr --prepipe head -n 10 ...`, if you like. There is a `--prepipe` and a `--prepipex`: * If the command normally runs with `nameofprogram < filename.ext` (such as `gunzip` or `zcat -cf` or `xz -cd`) then use `--prepipe`. * If the command normally runs with `nameofprogram filename.ext` (such as `unzip -qc`) then use `--prepipex`. Lastly, note that if `--prepipe` or `--prepipex` is specified on the Miller command line, it replaces any autodetect decisions that might have been made based on the filename extension. Likewise, `--gzin`/`--bz2in`/`--zin` are ignored if `--prepipe` or `--prepipex` is also specified. ## Compressed output Everything said so far on this page has to do with compressed input. For compressed output: * Normally Miller output is to stdout, so you can pipe the output: `mlr sort -n quantity foo.csv | gzip > sorted.csv.gz`. * For [`tee` statements](reference-dsl-output-statements.md#tee-statements), which write output to files rather than stdout, use `tee`'s redirect syntax:

mlr --from example.csv --csv put -q '
  filename = $color.".csv.gz";
  tee | "gzip > ".filename, $*
'

file red.csv.gz purple.csv.gz yellow.csv.gz

red.csv.gz:    gzip compressed data, last modified: Mon Aug 23 02:34:05 2021, from Unix, original size modulo 2^32 185
purple.csv.gz: gzip compressed data, last modified: Mon Aug 23 02:34:05 2021, from Unix, original size modulo 2^32 164
yellow.csv.gz: gzip compressed data, last modified: Mon Aug 23 02:34:05 2021, from Unix, original size modulo 2^32 158

mlr --csv cat yellow.csv.gz

color,shape,flag,k,index,quantity,rate
yellow,triangle,true,1,11,43.6498,9.8870
yellow,circle,true,8,73,63.9785,4.2370
yellow,circle,true,9,87,63.5058,8.3350

* Using the [in-place flag](reference-main-io-options.md#in-place-mode) `-I`, as of August 2021 the overwritten file will _not_ be compressed as it was when it was read: e.g. `mlr -I --csv cat gz-example.csv.gz` will write `gz-example.csv.gz` which contains a plain, uncompressed CSV contents. This is a bug and will be fixed.