strong-log-transformer
    DefinitelyTyped icon, indicating that this package has TypeScript declarations provided by the separate @types/strong-log-transformer package

    2.1.0 • Public • Published

    strong-log-transformer

    A stream filter for performing common log stream transformations like timestamping and joining multi-line messages.

    This is not a logger! But it may be useful for rolling your own logger.

    Usage

    Install strong-log-transformer and add it to your dependencies list.

    npm install --save strong-log-transformer

    CLI

    When installed globally the sl-log-transformer CLI utility is exposed. It is primarily used for testing, but it can also be used as an alternative to awk or sed for jobs such as timestamping every line of another process's output. This can be useful for cron jobs, for example.

    $ npm install -g strong-log-transformer
    $ sl-log-tranformer --help
    Usage: sl-log-transformer [options]
     
    Stream transformer that prefixes lines with timestamps and other things.
     
    OPTIONS:
       --format FORMAT        default: "text"
       --tag TAG              default: ""
       --mergeMultiline       default: off
       --timeStamp            default: off

    Line Merging

    In order to keep things flowing when line merging is enabled (disabled by default) there is a sliding 10ms timeout for flushing the buffer. This means that whitespace leading lines are only considered part of the previous line if they arrive within 10ms of the previous line, which should be reasonable considering the lines were likely written in the same write().

    Example

    Here's an example using the transformer to annotate log messages from cluster workers.

    var cluster = require('cluster');
     
    if (cluster.isMaster) {
      // Make sure workers get their own stdout/stderr streams
      cluster.setupMaster({silent: true});
     
      // require log transformer module
      var transformer = require('strong-log-transformer');
     
      // Following the 12-factor app model, we pipe to stdout, but we could easily
      // pipe to any other stream(s), such as a FileStream for a log file.
     
      // stdout is plain line-oriented logs, but we want to add timestamps
      var info = transformer({ timeStamp: true,
                               tag: 'INFO' });
      // stderr will only be used for strack traces on crash, which are multi-line
      var error = transformer({ timeStamp: true,
                                tag: 'ERROR',
                                mergeMultiline: true });
     
      // Each worker's stdout/stderr gets piped into our info and erro transformers
      cluster.on('fork', function(worker) {
        console.error('connecting worker');
        worker.process.stdout.pipe(info).pipe(process.stdout);
        worker.process.stderr.pipe(error).pipe(process.stdout);
      });
     
      //... cluster fork logic goes here ...
      cluster.fork();
     
    } else {
      //... worker code here ...
     
      console.log('new worker, this line will be timestamped!');
      throw new Error('This will generate a multi-line message!');
    }
     

    When we run the example code as example.js we get:

    $ node example.js
    connecting worker
    2014-06-08T18:54:00.920Z INFO new worker, this line will be timestamped!
    2014-06-08T18:54:00.926Z ERROR /Users/ryan/work/strong-log-transformer/e.js:33\n    throw new Error('This will generate a multi-line message!');\n          ^
    2014-06-08T18:54:00.926Z ERROR Error: This will generate a multi-line message!\n    at null._onTimeout (/Users/ryan/work/strong-log-transformer/e.js:33:11)\n    at Timer.listOnTimeout [as ontimeout] (timers.js:110:15)

    Install

    npm i strong-log-transformer

    DownloadsWeekly Downloads

    1,232,699

    Version

    2.1.0

    License

    Apache-2.0

    Unpacked Size

    16.4 kB

    Total Files

    9

    Last publish

    Collaborators

    • rmg
    • ritch
    • rfeng
    • bajtos
    • 0candy
    • amir-61
    • hacksparrow
    • superkhau
    • kraman
    • thegman
    • davidcheung
    • tonyf-ibm
    • qpresley
    • kjdelisle
    • jannyhou2016
    • b-admike