excel-2-json

Convert Excel Files to JSON

Install

Expected use is offline translation of Excel data to JSON files, although async facilities are provided.

$ npm install excel-as-json --save-dev

Use

convertExcel = require('excel-as-json').processFile;
convertExcel(<src>, <dst>, isColOriented, callback);

• src: path to source Excel file (xlsx only) - will read sheet 0
• dst: path to destination JSON file. If null, simply return the parsed object tree
• isColOriented: is an Excel row an object, or is a column an object (Default: false)
• callback(err, data): callback for completion notification

With these arguments, you can:

• convertExcel(src, dst) will write a row oriented xlsx to file with no notification
• convertExcel(src, dst, true) will write a col oriented xlsx to file with no notification
• convertExcel(src, dst, true, callback) will write a col oriented xlsx to file and notify with errors and data
• convertExcel(src, null, true, callback) will return errors and the parsed object tree in the callback

Convert a row/col oriented Excel file to JSON as a development task and log errors:

convertExcel = require('excel-as-json').processFile

convertExcel 'row.xlsx', 'row.json', false, (err, data) ->
	if err then console.log "JSON conversion failure: #{err}"
convertExcel 'col.xlsx', 'col.json', true, (err, data) ->
	if err then console.log "JSON conversion failure: #{err}"

Convert Excel file to an object tree and use that tree. Note that properly formatted data will convert to the same object tree whether row or column oriented.

convertExcel = require('excel-as-json').processFile

convertExcel 'row.xlsx', undefined, false, (err, data) ->
	if err throw err
	doSomethingInteresting data
convertExcel 'col.xlsx', undefined, true, (err, data) ->
	if err throw err
	doSomethingInteresting data

Why?

• Your application serves static data obtained as Excel reports from another application
• Whoever manages your static data finds Excel more pleasant than editing JSON
• Your data is the result of calculations or formatting that is more simply done in Excel

What's the challenge?

Excel stores tabular data. Converting that to JSON using only a couple of assumptions is straight-forward. Most interesting JSON contains nested lists and objects. How do you map a flat data square that is easy for anyone to edit into these nested lists and objects?

Solving the challenge

• Use a key row to name JSON keys
• Allow data to be stored in row or column orientation.
• Use javascript notation for keys and arrays
  • Allow dotted key path notation
  • Allow arrays of objects and literals

Excel Data

What is the easiest way to organize and edit your Excel data? Lists of simple objects seem a natural fit for a row oriented sheets. Single objects with more complex structure seem more naturally presented as column oriented sheets. Doesn't really matter which orientation you use, the module allows you to speciy a row or column orientation; basically, where your keys are located: row 0 or column 0.

Keys and values:

• Row or column 0 contains JSON key paths
• Remaining rows/columns contain values for those keys
• Multiple value rows/columns represent multiple objects stored as a list
• Within an object, lists of objects have keys like phones[1].type
• Within an object, flat lists have keys like aliases[]

Examples

A simple, row oriented key

firstName
Jihad

produces

[{
  "firstName": "Jihad"
}]

A dotted key name looks like

address.street
12 Beaver Court

and produces

[{
  "address": {
    "street": "12 Beaver Court"
    }
}]

An indexed array key name looks like

phones[0].number
123.456.7890

and produces

[{
  "phones": [{
      "number": "123.456.7890"
    }]
}]

An embedded array key name looks like this and has ';' delimited values

aliases[]
stormagedden;bob

and produces

[{
  "aliases": [
    "stormagedden",
    "bob"
  ]
}]

A more complete row oriented example

firstName	lastName	address.street	address.city	address.state	address.zip
Jihad	Saladin	12 Beaver Court	Snowmass	CO	81615
Marcus	Rivapoli	16 Vail Rd	Vail	CO	81657

would produce

[{
    "firstName": "Jihad",
    "lastName": "Saladin",
    "address": {
      "street": "12 Beaver Court",
      "city": "Snowmass",
      "state": "CO",
      "zip": "81615"
    }
  },
  {
    "firstName": "Marcus",
    "lastName": "Rivapoli",
    "address": {
      "street": "16 Vail Rd",
      "city": "Vail",
      "state": "CO",
      "zip": "81657"
    }
  }]

You can do something similar in column oriented sheets. Note that indexed and flat arrays are added.

firstName	Jihad	Marcus
lastName	Saladin	Rivapoli
address.street	12 Beaver Court	16 Vail Rd
address.city	Snowmass	Vail
address.state	CO	CO
address.zip	81615	81657
phones[0].type	home	home
phones[0].number	123.456.7890	123.456.7891
phones[1].type	work	work
phones[1].number	098.765.4321	098.765.4322
aliases[]	stormagedden;bob	mac;markie

would produce

[
  {
    "firstName": "Jihad",
    "lastName": "Saladin",
    "address": {
      "street": "12 Beaver Court",
      "city": "Snowmass",
      "state": "CO",
      "zip": "81615"
    },
    "phones": [
      {
        "type": "home",
        "number": "123.456.7890"
      },
      {
        "type": "work",
        "number": "098.765.4321"
      }
    ],
    "aliases": [
      "stormagedden",
      "bob"
    ]
  },
  {
    "firstName": "Marcus",
    "lastName": "Rivapoli",
    "address": {
      "street": "16 Vail Rd",
      "city": "Vail",
      "state": "CO",
      "zip": "81657"
    },
    "phones": [
      {
        "type": "home",
        "number": "123.456.7891"
      },
      {
        "type": "work",
        "number": "098.765.4322"
      }
    ],
    "aliases": [
      "mac",
      "markie"
    ]
  }
]

Data Conversions

All values from the 'excel' package are returned as text. This module detects numbers and booleans and converts them to javascript types. Booleans must be text 'true' or 'false'. Excel FALSE and TRUE are provided from 'excel' as 0 and 1 - just too confusing.

Caveats

During install (mac), you may see compiler warnings while installing the excel dependency - although questionable, they appear to be benign.

TODO

• provide processSync - using 'async' module
• Detect and convert dates
• Make 1 column values a single object?

Change History

1.0.0

• Changed process() to processFile() to avoid name collision with node's process object
• Automatically convert text numbers and booleans to native values
• Create destination directory if it does not exist