Functional programming in human-readable form
Speak-r - it's a ramda pipeline without using R.(.
Instead of writing unreadable pipelines full of syntax sugar you can simply write ramda methods like as if they were ordinary words
const R = require('ramda')
const pipe = R.pipe(R.when(R.equals(1),R.add(10)),R.objOf('value'))
let result = pipe(1)
//=> { value: 11 }const say = require('speaker')
const pipe = say("when [ == 1 , + 10 ] objOf 'value'")
let result = pipe(1)
//=> { value: 11 }Every expression is sequence of words separated by spaces, representing a functional pipe. Lean more on functional pipes.
Every ramda method can be written using the corresponding keyword. All these ways are identical.
const Say = require('speak-r')
let fn = Say(`inc of`)
fn = R.pipe( R.inc, R.of )
fn = num => R.of( R.inc(num) )
fn(0)
// => [ 1 ]Numbers are written as is, strings are enclosed in single quotes. Based on the fact that ordinary variables are non-callable, every string, number or argument refers to the nearest preceding function.
Type tokens (Array, String...) also can be written as is.
let fn = Say(`add 10`)
fn(0)
// => 10
fn = Say(`objOf 'name'`)
fn('Anna')
// => { name: 'Anna' }A function can be followed by any number of arguments.
const fn = Say(`assoc 'age' 21`)
fn({ name: 'Anna' })
// => { name: 'Anna', age: 21 }Functions arguments are automatically grouped, so you can safely write functions successively.
const isAdult = Say(`prop 'age' lt 18`)
isAdult({ name: 'Anna', age: 21 })
// => trueOften you want to send more than one argument.
Arguments can be listed at the beginning before :: and then refer to their names as a value or as a function.
By default, the argument is used as a function. To refer to a variable, use symbol @
const getHash = user => {
user.hash = 4 // chosen by fair dice roll
// guaranted to be random
return user
}
const setBalance = Say(`user money hashFunc :: assoc 'balance' @money hashFunc`)
setBalance({ name: 'Anna', age: 21 }, 100, getHash)
// => { name: 'Anna', age: 21, balance: 100, hash: 4 }Now broken.
Nested pipes
Say(`when [ is Array, append 'new element' ]`)
// => { name: 'Anna', age: 21, balance: 100, hash: 4 }set DEBUG=* for watching tree parsing
Basic syntax is defined in app/lang/syntax.js
Grammar rules are written as finite state machine in app/core/convolve.js
const states = {
empty: 0,
pipe : 1,
open : 2,
mid : 3,
close: 4
}
const actions = {
next : 0,
child : 1,
parent: -1,
error : NaN
}
const switches = [
[NaN, 1, 1, NaN, 1], // empty
[NaN, 0, 1, NaN, 0], // pipe
[NaN, -1, 1, 0, -1], // open
[NaN, -1, 1, 0, -1], // mid
[NaN, 0, 1, NaN, 0] // close
]This is the transition matrix from state to state Each language token can affect the current position in the stack (opening bracket - increases, closing bracket - decreases, the function leaves unchanged). transition matrix receives the current state and the type of token, then returns the new state of the translator.
If someone is interested, I can describe this part more