Have you ever wanted to use Python functions directly in your Unix shell? Mario can read and write csv, json, and yaml; traverse trees, and even do xpath queries. Plus, it supports async commands right out of the box. Build your own commands with a simple configuration file, and install plugins for even more!
Mario is the plumbing snake ππ§ helping you build data pipelines in your shell π’.
- Execute Python code in your shell.
- Pass Python objects through multi-stage pipelines.
- Read and write csv, json, yaml, toml, xml.
- Run async functions natively.
- Define your own commands in a simple configuration file or by writing Python code.
- Install plugins to get more commands.
- Enjoy high test coverage, continuous integration, and nightly releases.
Windows support is hopefully coming soon. Linux and MacOS are supported now.
Get Mario with pip:
python3.7 -m pip install mario
If you're not inside a virtualenv, you might get a PermissionsError
. In that case, try using:
python3.7 -m pip install --user mario
or for more isolation, use pipx:
pipx install --python python3.7 mario
The mario-addons package provides a number of useful commands not found in the base collection.
Get Mario addons with pip:
python3.7 -m pip install mario-addons
If you're not inside a virtualenv, you might get a PermissionsError
. In that case, try using:
python3.7 -m pip install --user mario-addons
or for more isolation, use pipx:
pipx install --python python3.7 mario
pipx inject mario mario-addons
Invoke with mario
at the command line.
$ mario eval 1+1
2
Given a csv like this:
$ cat <<EOF > hackers.csv
name,age
Alice,21
Bob,22
Carol,23
EOF
Use read-csv-dicts
to read each row into a dict:
$ mario read-csv-dicts < hackers.csv
{'name': 'Alice', 'age': '21'}
{'name': 'Bob', 'age': '22'}
{'name': 'Carol', 'age': '23'}
Use map
to act on each input item x
:
$ mario read-csv-dicts map 'x["name"]' < hackers.csv
Alice
Bob
Carol
Chain Python functions together with !
:
$ mario read-csv-dicts map 'x["name"] ! len' < hackers.csv
5
3
5
or by adding another command
$ mario read-csv-dicts map 'x["name"]' map len < hackers.csv
5
3
5
Use x
as a placeholder for the input at each stage:
$ mario read-csv-dicts map 'x["age"] ! int ! x*2' < hackers.csv
42
44
46
Automatically import modules you need:
$ mario map 'collections.Counter ! dict' <<<mississippi
{'m': 1, 'i': 4, 's': 4, 'p': 2}
You don't need to explicitly call the function with some_function(x)
; just use the function's name, some_function
. For example, instead of
$ mario map 'len(x)' <<EOF
a
bb
EOF
1
2
try
$ mario map len <<EOF
a
bb
EOF
1
2
Here are a few commands. See Command reference for the complete set, and get even more from mario-addons.
Use eval
to evaluate a Python expression.
% mario eval 'datetime.datetime.utcnow()'
2019-01-01 01:23:45.562736
Use map
to act on each input item.
$ mario map 'x * 2' <<EOF
a
bb
EOF
aa
bbbb
Use filter
to evaluate a condition on each line of input and exclude false values.
$ mario filter 'len(x) > 1' <<EOF
a
bb
ccc
EOF
bb
ccc
Use apply
to act on the sequence of items.
$ mario apply 'len(x)' <<EOF
a
bb
EOF
2
Use chain
to flatten a list of lists into a single list, like itertools.chain.from_iterable.
For example, after generating a several rows of items,
$ mario read-csv-tuples <<EOF
a,b,c
d,e,f
g,h,i
EOF
('a', 'b', 'c')
('d', 'e', 'f')
('g', 'h', 'i')
use chain
to put each item on its own row:
$ mario read-csv-tuples chain <<EOF
a,b,c
d,e,f
g,h,i
EOF
a
b
c
d
e
f
g
h
i
Making sequential requests is slow. These requests take 16 seconds to complete.
% time mario map 'await asks.get ! x.json()["url"]' <<EOF
http://httpbin.org/delay/5
http://httpbin.org/delay/1
http://httpbin.org/delay/2
http://httpbin.org/delay/3
http://httpbin.org/delay/4
EOF
https://httpbin.org/delay/5
https://httpbin.org/delay/1
https://httpbin.org/delay/2
https://httpbin.org/delay/3
https://httpbin.org/delay/4
0.51s user
0.02s system
16.460 total
Concurrent requests can go much faster. The same requests now take only 6 seconds. Use async-map
, or async-filter
, or reduce
with await some_async_function
to get concurrency out of the box.
% time mario async-map 'await asks.get ! x.json()["url"]' <<EOF
http://httpbin.org/delay/5
http://httpbin.org/delay/1
http://httpbin.org/delay/2
http://httpbin.org/delay/3
http://httpbin.org/delay/4
EOF
https://httpbin.org/delay/5
https://httpbin.org/delay/1
https://httpbin.org/delay/2
https://httpbin.org/delay/3
https://httpbin.org/delay/4
0.49s user
0.03s system
5.720 total
Define new commands and set default options. See Configuration reference for details.
Add new commands like map
and reduce
by installing Mario plugins. You can try them out without installing by adding them to any .py
file in your ~/.config/mario/modules/
.
Share popular commands by installing the mario-addons package.
- This package is experimental and is subject to change without notice.
- Check the issues page for open tickets.
A number of cool projects have pioneered in the Python-in-shell space. I wrote Mario because I didn't know these existed at the time, but now Mario has a bunch of features the others don't (user configuration, multi-stage pipelines, async, plugins, etc).