Extended CSV Parser for TypeScript

This project began as a way to write JSON files inside Excel spreadsheets.
It allows for nested object properties, strongly typed values, arrays, and comments.

---

It transforms text inputs like this:

id:int,user.name,user.age:float,user.isVip:bool,user.gamesPlayed:[],message.content,message.dateSend:int,lucky:random,nested.property.number.array:number[],special:emoji
1,Alice,28.5,true,Tetris;Pac-Man,Hello how are you?,1648872000,-this-can-be-anything-,1;2;3;4,:D
2,Bob,42.0,false,,Hey there!,1648882800,-1,5;6;7,:party:
3,Charlie,19.3,true,Chess;Checkers;Monopoly,Howdy!,1648915200,---,21;22;23,:beer:
5,Hannah,31.6,false,Animal Crossing,What's new?,1648926000,random,24;25;26;27,:flower:
# 6,Isaac,19.4,true,League of Legends;Dota 2
# This is a comment

Into beautiful objects that are ready to use in a JavaScript or TypeScript project.

[
  {
    "id": 1,
    "user": {
      "name": "Alice",
      "age": 28.5,
      "isVip": true,
      "gamesPlayed": ["Tetris", "Pac-Man"]
    },
    "message": {
      "content": "Hello how are you?",
      "dateSend": 1648872000
    },
    "lucky": 0.849495919176418,
    "nested": { "property": { "number": { "array": [1, 2, 3, 4] } } },
    "special": "😃"
  },
  {
    "id": 2,
    "user": {
      "name": "Bob",
      "age": 42,
      "isVip": false,
      "gamesPlayed": []
    },
    "message": {
      "content": "Hey there!",
      "dateSend": 1648882800
    },
    "lucky": 0.027607189531068554,
    "nested": { "property": { "number": { "array": [5, 6, 7] } } },
    "special": "🎉"
  },
  {
    "id": 3,
    "user": {
      "name": "Charlie",
      "age": 19.3,
      "isVip": true,
      "gamesPlayed": ["Chess", "Checkers", "Monopoly"]
    },
    "message": {
      "content": "Howdy!",
      "dateSend": 1648915200
    },
    "lucky": 0.31084866168836345,
    "nested": { "property": { "number": { "array": [21, 22, 23] } } },
    "special": "🍺"
  }
]

---

Usage

To use it, simply import the parseCSV function and pass a string with CSV data like this:

import { type CustomTypeDefinition, parseCSV } from "extended-csv-parser"

const customTypes = [] as CustomTypeDefinition[]

const file = Deno.readTextFileSync("demo.csve")
const data = parseCSV(file, customTypes)
const json = JSON.stringify(data, null, 2)

console.log(json)

---

Supported built-in types

• string ( default )
• float
• number
• int
• bool
• boolean
• all above +[]

---

Custom Types

You may be wondering what those random and emoji types are for.
This parser allows for custom type definitions, as shown here:

import type { CustomTypeDefinition } from "extended-csv-parser"

const customTypes: CustomTypeDefinition[] = [
  {
    name: "random",
    parse: () => Math.random(),
  },
  {
    name: "emoji",
    parse: input => {
      switch (input) {
        case ":D":
          return "😃"
        case ":flower:":
          return "🌺"
        case ":party:":
          return "🎉"
        case ":beer:":
          return "🍺"
        default:
          return input
      }
    },
  },
]

---

Syntax Explained

• , - separates header names and row values
• . - indicates that a header name is a nested object property
• : - used to add a optional type to a header entry
• ; - array item separator

To do

• Escaping , and " characters
  ( right now parser uses .split(",") method, which is problematic for some files. )
• NPM package
• Vscode Extension for better syntax highlighting
• Rollup plugin
• readStream version of parseCSV for better performance