jimbly-tail-stream v0.1.0

node-tail-stream

Continuous file tail as a stream. Handles file truncation, but not yet file rollover (old file being deleted/moved and a new file with the same name replacing it).

Differences from tail-stream

Somehow, despite searching, I failed to find this existing module before authoring my own. The only key difference is that this new module handles broken pipes slightly more elegantly, with some automatic cleanup and a manual close to prevent leaking of FSWatchers. Other than that, the existing module has a lot more functionality and robustness.

Usage

Use tail_stream.createReadStream just like fs.createReadStream except the stream will not end but continue emitting data as the backing file grows.

Example

// Pipe the tail stream to stdout
var tail_stream = require('tail-stream');
var stream = tail_stream.createReadStream('example.txt', {interval: 100});
stream.pipe(process.stdout);

For an example of using this to tail stream a file to an http request, see the example under Caveats.

Caveats

Due to the Streams API in node, we do not know if our Readable stream was piped into something which has now been closed, and our TailStream will stay active (leaving the program running and also with an active FSWatcher which may be consuming significant CPU) unless explicitly closed. We automatically recover if the stream was not at the end of the file we're streaming or if the file we're streaming continues to be modified, but a stale, unchanging file will leave a FSWatcher (due to an outstanding Readable Stream ._read call) unless explicitly closed.

Example:

require('http').createServer(function (req, res) {
  var stream = require('tail-stream').createReadStream('example.txt');
  stream.pipe(res);
  res.on('end', function() {
    stream.close();
  });
}).listen(1337, '127.0.0.1');