@ajbowen249/lait NPM

lait; an AWK-inspired TypeScript Command-Line Utility

Check out the interactive version of these docs here!

lait is a command-line utility designed as an alternative to awk. Like awk, lait matches input line-by-line against a list of regular expressions and runs the first block of code it matches. Unlike awk, lait uses TypeScript as its scripting language.

$ cat demo_input
OrderId Item Quantity
#a2fr5 Socks 16
#d8j38 Avocado 2
#8fh39 Shampoo 1
#qb6ag Candle 3
$ lait '/#\w{5} (?<name>\w+) (?<quantity>\d+)/; { print(g.name, g.q) }' demo_input
16 Socks
2 Avocado
1 Shampoo
3 Candle

A lait program is just a TypeScript script. However, lait picks it apart and puts it into a wrapper program, where top-level blocks preceded by a regular expression literal are registered as handlers for matching lines of input. A top- level block not preceded by a regex will be registered as the default handler if given. Any statements written before the first handler will be run before scanning input, and all code after the first handler definition will be run after processing input. Note that this means there is no need for an explicit BEGIN or END block like in awk.

Let's take a look at a more complicated lait script. Like awk, lait can take a file as its script input:

interface Order {
    id: string;
    item: string;
    quantity: number;
}

const letterOrders: Order[] = [];
const digitOrders: Order[] = [];

function parseOrder(g: RegExpMatchArray['groups']) {
    return {
        id: g.id,
        item: g.item,
        quantity: parseInt(g.q),
    };
}

/(?<id>[a-z]\w{4}) (?<item>\w+) (?<q>\d+)/; {
    letterOrders.push(parseOrder(g));
}

/(?<id>\d\w{4}) (?<item>\w+) (?<q>\d+)/; {
    digitOrders.push(parseOrder(g));
}

function toTable(order: Order) {
    return `${order.id}: ${order.quantity} ${order.item}`;
}

// Note: print is just an alias of console.log
print('Order ID starts with letter:');
print(letterOrders.map(toTable).join('\n'));

print('Order ID starts with digit:');
print(digitOrders.map(toTable).join('\n'));

$ lait -f demo.lait.ts demo_input
Order ID starts with letter:
#a2fr5: 16 Socks
#d8j38: 2 Avocado
#qb6ag: 3 Candle
Order ID starts with digit:
#8fh39: 1 Shampoo

Note that programs of this size are less common. Like awk, lait programs are meant to be small and integrated into shell operations.

The first input given to handlers is $, the array of "fields" on the line. Unlike awk, the array is simply a zero- based array of fields, and there is not a "complete line" entry in $. Like awk, the default field separator is space, and it can be overridden via the FS global:

$ cat demo.csv
Name,Age,ID
Carol,25,2lf8ah
Bob,21,j8efy3g
Jesse,34,j8fhiuad8

$ lait 'FS=`,`; { print($[0], `#${$[2]}`); }' demo.csv
Name #ID
Carol #2lf8ah
Bob #j8efy3g
Jesse #j8fhiuad8

The complete list of args given to matchers is:

$: The set of awk-style fields
$_: The complete matched line
m: The RegExpMatchArray from matching the line
g: m.groups (possibly undefined!)

lait '/(?<id>#\w{5}) (?<name>\w+) (?<amnt>\d+)/; { print($, $_, JSON.stringify(m), JSON.stringify(g)) }' demo_input
#a2fr5,Socks,16 #a2fr5 Socks 16 ["#a2fr5 Socks 16","#a2fr5","Socks","16"] {"id":"#a2fr5","name":"Socks","amnt":"16"}
#d8j38,Avocado,2 #d8j38 Avocado 2 ["#d8j38 Avocado 2","#d8j38","Avocado","2"] {"id":"#d8j38","name":"Avocado","amnt":"2"}
#8fh39,Shampoo,1 #8fh39 Shampoo 1 ["#8fh39 Shampoo 1","#8fh39","Shampoo","1"] {"id":"#8fh39","name":"Shampoo","amnt":"1"}
#qb6ag,Candle,3 #qb6ag Candle 3 ["#qb6ag Candle 3","#qb6ag","Candle","3"] {"id":"#qb6ag","name":"Candle","amnt":"3"}

lait can also accept input from standard in:

$ ls -al
total 14
drwxr-xr-x 1 user 197609   0 Dec  8 22:01 .
drwxr-xr-x 1 user 197609   0 Dec  8 21:47 ..
-rw-r--r-- 1 user 197609  61 Dec  8 19:19 demo.csv
-rw-r--r-- 1 user 197609 699 Dec  8 22:01 demo.lait.ts
-rw-r--r-- 1 user 197609  87 Dec  8 18:46 demo_input

$ ls -al | lait '/^d|-/;{print($[8], `(${$[4]} bytes)`)}'
. (0 bytes)
.. (0 bytes)
demo.csv (61 bytes)
demo.lait.ts (699 bytes)
demo_input (87 bytes)

Globals

In addition to FS, there is also TRIM_EMPTY, which is true by default as is usually convenient for processing space-aligned data like in the ls example. That's less useful when processing CSVs:

$ cat demo2.csv
Name,Favorite Film (Optional),Score
Cathy Smith,A Fistful of Dollars,78
Paulo Henry MacMasterson III,Finding Nemo,4
Bob Nofunpants,,4

$ lait 'FS=`,`;TRIM_EMPTY=false; { print($[0], $[1] || `does not like movies :(`)  }' demo2.csv
Name Favorite Film (Optional)
Cathy Smith A Fistful of Dollars
Paulo Henry MacMasterson III Finding Nemo
Bob Nofunpants does not like movies :(

Without TRIM_EMPTY=false;, the last line of that would have been:

Bob Nofunpants 4

Command-Line Definitions

Variables can be declared via one or more instances of the -d (--define) command:

$ lait -d x=12 -d y=some_string 'print(x, y)'
12 some_string

The FS and TRIM_EMPTY variables can also be overridden via this argument. The Globals example could have also been:

$ lait -d FS=, -d TRIM_EMPTY=false '{ print($[0], $[1] || `does not like movies :(`)  }' demo2.csv
Name Favorite Film (Optional)
Cathy Smith A Fistful of Dollars
Paulo Henry MacMasterson III Finding Nemo
Bob Nofunpants does not like movies :(

Imports

You can import from the Node standard library, as well as any node modules you would expect to be accessible either globally or from a module installed in the working directory's node_modules.

$ lait 'import * as _ from "lodash"; print(_.isNumber(12));'
true

Note: Imports from local files are not yet working

Installation

All you gotta do is

npm i -g @ajbowen249/lait

But Why?

While awk is a very powerful tool, it can have a few rough edges. Those being:

Fragmentation - One must always be wary of which implementation they're using, and which builtins are available.
Regex Syntax - awk has yet another regular expression syntax to remember, and it even varies by vendor.
Scripting Syntax - Like the regex point, awk using its own scripting language means one must play the, "is this how I do x in this language?" game. While TypeScript (and, let's be real, using the type system here will be rare) is certainly not universal, it's much more likely someone willing to install an NPM package will be familiar with it. Standard awk is also light on language features and built-ins, and having built-in array sort and higher-order functions should prove useful in problems typically solved with awk.