The goal here is to be able to use this in connection with jsdom to create test utilities,
possibly scraping utilities, ultimately an API-driven browser written in Node.
Since HTML5 FileAPI has been described by the W3C (terse, technical) and Mozilla Developer Center (understandable, end-user-oriented) in detail, I'll just highlight the differences:
File is not (yet) a subclass of Blob
FileError and FileException are not yet implemented (they use Error instead)
blob: scheme and remote URI schemes are not yet implemented
In the browser, File has no constructor. In Node, it does.
node-mime is used for extension-based automatic ContentType detection (uses name if available, or path if not)
File(StringUriPath)
var file = new File("./files/myfile.txt");
File({ buffer: Node.Buffer })
var file = new File({ name: "abc-song.txt", // required type: "text/plain", // optional buffer: new Buffer("abcdefg,hijklmnop, qrs, tuv, double-u, x, y and z")});
File({ stream: Node.ReadStream })
var file = new File({ name: "abc-song.txt", // required type: "text/plain", // optional stream: new EventEmitter() // a read stream (emits `error`, `data`, `end`)});process.nextTick(function () { file.stream.emit('data', "abcdefg,hijklmnop, qrs, tuv, double-u, x, y and z"); file.stream.emit('end');});
File(Object)
var file = new File({ path: "./files/myfile.txt", // path of file to read buffer: Node.Buffer, // use this Buffer instead of reading file stream: Node.ReadStream, // use this ReadStream instead of reading file name: "SomeAwesomeFile.txt", // optional when using `path` // must be supplied when using `Node.Buffer` or `Node.ReadStream` type: "text/plain", // generated based on the extension of `name` or `path` jsdom: true, // be DoM-like and immediately get `size` and `lastModifiedDate` // [default: false] async: true, // use `fs.stat` instead of `fs.statSync` for getting // the `jsdom` info // [default: false] lastModifiedDate: fileStat.mtime.toISOString(), size: fileStat.size || Buffer.length);
FileReader
FileReader.setNodeChunkedEncoding() is a non-standard method which hints that the FileReader should chunk if possible
I.E. The file will be sent with the header Transfer-Encoding: chunked
The default is false since many webservers do not correctly implement the standard correctly,
and hence do not expect or accept Transfer-Encoding: chunked from clients.
FileReader.on is a non-standard alias of addEventListener
EventTarget.target.nodeBufferResult is a non-standard property which is a Node.Buffer instance of the data.
FileReader.on('data', fn) is a non-standard event which passes a Node.Buffer chunk each time the progress event is fired.
var fileReader = new FileReader();fileReader.setNodeChunkedEncoding(true || false);fileReader.readAsDataURL(new File('./files/my-file.txt'));// non-standard alias of `addEventListener` listening to non-standard `data` eventfileReader.on('data', function (data) { console.log("chunkSize:", data.length);});// `onload` as listenerfileReader.addEventListener('load', function (ev) { console.log("dataUrlSize:", ev.target.result.length);});// `onloadend` as propertyfileReader.onloadend', function () { console.log("Success");});
FileList
The browser has no constructor for this. Node has two.
new FileList(f1, f2, ...)
var fileList = new FileList(file1, file2, file3);
new FileList([f1, f2])
var files = [ new File('./files/blob.bin'), new File('./files/image.jpg'), new File('./files/readme.txt') ], fileList;fileList = new FileList(files);
//// TODO//// HTML5 File URI should be implemented// will need non-ahr 301-handling requester not prevent circular dep//// File should be a subclass of Blob//// jsdom EventTarget // http://aptana.com/reference/html/api/EventTarget.html// target.result// // jsdom ProgressEvent // http://www.w3.org/TR/progress-events/// lengthComputable// loaded// total// initProgressEvent