server

NodeJS Server Deployments

This library is 100% pure JS. This is great for compatibility but tends to lock up long-running processes. In the web browser, Web Workers are used to offload work from the main browser thread. In NodeJS, there are other strategies. This demo shows a few different strategies applied to different server frameworks.

NOTE: these examples merely demonstrate the core concepts and do not include appropriate error checking or other production-level features.

Node Buffer

The read and write functions can handle Buffer data with type:"buffer". For example, the request library returns data in a buffer:

var XLSX = require('xlsx'), request = require('request');
request(url, {encoding: null}, function(err, res, data) {
	if(err || res.statusCode !== 200) return;

	/* data is a node Buffer that can be passed to XLSX.read */
	var workbook = XLSX.read(data, {type:'buffer'});

	/* DO SOMETHING WITH workbook HERE */
});

The readFile / writeFile functions wrap fs.{read,write}FileSync:

/* equivalent to `var wb = XLSX.readFile("sheetjs.xlsx");` */
var buf = fs.readFileSync("sheetjs.xlsx");
var wb = XLSX.read(buf, {type:'buffer'});

Responding to Form Uploads

Using formidable, files uploaded to forms are stored to temporary files that can be read with readFile:

/* within the server callback function(request, response) { */
var form = new formidable.IncomingForm();
form.parse(req, function(err, fields, files) {
	var f = files[Object.keys(files)[0]];
	var workbook = XLSX.readFile(f.path);
	/* DO SOMETHING WITH workbook HERE */
});

The node.js demo shows a plain HTTP server that accepts file uploads and converts data to requested output format.

Example servers

Each example server is expected to hold an array-of-arrays in memory. They are expected to handle:

• POST / accepts an encoded file and updates the internal storage
• GET /?t=<type> returns the internal storage in the specified type
• POST /?f=<name> reads the local file and updates the internal storage
• GET /?f=<name> writes the file to the specified name

Testing with cURL is straightforward:

# upload test.xls and update data
curl -X POST -F "[email protected]" http://localhost:7262/
# download data in SYLK format
curl -X GET http://localhost:7262/?t=slk
# read sheetjs.xlsx from the server directory
curl -X POST http://localhost:7262/?f=sheetjs.xlsx
# write sheetjs.xlsb in the XLSB format
curl -X GET http://localhost:7262/?f=sheetjs.xlsb

Main-process logic with express

The most straightforward approach is to handle the data directly in HTTP event handlers. The buffer type for XLSX.read and XLSX.write work with http module and with express directly. The following snippet generates a workbook based on an array of arrays and sends it to the client:

function send_aoa_to_client(req, res, data, bookType) {
	/* generate workbook */
	var ws = XLSX.utils.aoa_to_sheet(data);
	var wb = XLSX.utils.book_new();
	XLSX.utils.book_append_sheet(wb, ws, "SheetJS");

	/* generate buffer */
	var buf = XLSX.write(wb, {type:'buffer', bookType:bookType || "xlsx"});

	/* send to client */
	res.status(200).send(buf);
}

fork with koa

child_process.fork provides a light-weight and customizable way to offload work from the main server process. This demo passes commands to a custom child process and the child passes back buffers of data.

The main server script is koa.js and the worker script is koasub.js. State is maintained in the worker script.

command-line utility with micro

The npm module ships with the xlsx command line tool. For global installs, the script bin/xlsx.njs is added to a directory in PATH. For local installs, the appropriate script or symbolic link is set up in node_modules/.bin/.

The --arrays option directs xlsx to generate an array of arrays that can be parsed by the server. To generate files, the json2csv module exports the JS array of arrays to a CSV, the server writes the file, and the xlsx command is used to generate files of different formats.

tiny-worker with hapi

tiny-worker provides a Web Worker-like interface. Binary strings and simple objects are readily passed across the Worker divide.

The main server script is hapi.js and the worker script is worker.js. State is maintained in the server script.

Note: due to an issue with hapi payload parsing, the route POST /file is used to handle the case of reading from file, so the cURL test is:

# upload test.xls and update data
curl -X POST -F "[email protected]" http://localhost:7262/
# download data in SYLK format
curl -X GET http://localhost:7262/?t=slk
# read sheetjs.xlsx from the server directory
curl -X POST http://localhost:7262/file?f=sheetjs.xlsx
# write sheetjs.xlsb in the XLSB format
curl -X GET http://localhost:7262/?f=sheetjs.xlsb