NN - The Node CLI Helper

What is this?

It's a tool designed for use in the command line to process streams or files. If you're a data scientist or hobbyist and find yourself working with JSON or CSV files a lot, but don't have the time or will to learn how to use jq, awk, cut, tr, grep, or csvtool, this is what you need!

Note: JSON support is currently much more developed than CSV, though both formats are usable.
Note 2: Since this is driven by NodeJS and not buffered by line (due to JSON / CSV parsing), avoid processing files larger than a Gigabyte or so if you value your memory and CPU cycles.

Features:

• Automatically recognizes files as JSON, CSV, or plaintext, allowing you to use the same command for all file types.
• Lets you use familiar JS syntax to accomplish much with little effort
• Full NodeJS API support for those cases where you need to go all out

How to use

Install via yarn or npm

• yarn global add node-cli-processing
• npm install -g node-cli-processing

This will make the nn command available globally.

Use the command either with a file or via shell pipe

• nn ./fruits.json "data = data.filter(fruit => fruit.type === 'citrus')"
• cat ./fruits.json | nn "data = data.filter(fruit => fruit.type === 'citrus')"

The commands accepts the input and a string of JS code that is used to process the data.
The file or STDIN content is made available as the data variable, after being automatically turned into a JS object or array if the input file is deteced as being a JSON or CSV file respectively.
Afterwards executing the code provided, the content of the data variable is then piped to STDOUT, after being serialized as JSON or CSV if the type matches.

More examples

Assuming fruits.json is a JSON file like so:

[
  { "name": "Orange", "weight": 100, "type": "citrus" },
  { "name": "Apple", "weight": 80, "type": "boring fruit" },
  { "name": "Lemon", "weight": 60, "type": "citrus" },
  { "name": "Pear", "weight": 85, "type": "boring fruit" },
  { "name": "Sugar melon", "weight": 2500, "type": "melon" },
  { "name": "Honeydew", "weight": 2400, "type": "melon" }
]

and fruits.csv is a CSV file like so:

name,weight,type
orange,100,citrus
apple,80,boring fruit
lemon,60,citrus
pear,85,boring fruit
sugar melon,2500,melon
honeydew,2400,melon

Here are some possible use-cases:

• Filtering by json attribute: nn ./fruits.json "data = data.map(fruit => fruit.weight > 2000)"
• Filtering by csv column: nn ./fruits.csv "data = data.map(fruit => +fruit[1] > 2000)"
• Sorting by json attribute: nn ./fruits.json "data = data.sort((a,b) => a.weight - b.weight)"
• Sorting by csv column: nn ./fruits.csv "data = data.sort((a,b) => +[1] - +b[1])"

Since this is driven directly by NodeJS, everything is technically possible.

Planned Features

• Add file ending detection to prioritize certain file type parsing
• Allow manual choice between file types
• Alternate CSV parsing that produces a json array
• Parse numbers where possible in CSV format
• Parse and serialize CSV via third-party package to include proper escaping
• Add a flag to automatically keep a log of commands executed
• Separate flags for disabling input or output parsing