Neural Processing Mechanisms


    0.3.1 • Public • Published

    Node library for building a service to pull Live HLS and push to a WebDAV endpoint


    npm install --save @eyevinn/hls-pull-push


    import { HLSPullPush, MediaPackageOutput, S3BucketOutput } from "@eyevinn/hls-pull-push";
    const pullPushService = new HLSPullPush();
    pullPushService.registerPlugin("mediapackage", new MediaPackageOutput());
    pullPushService.registerPlugin("s3", new S3BucketOutput());
    pullPushService.listen(process.env.PORT || 8080);


    /api/docs GET Live API documentation
    /api/v1/fetcher POST Create a fetcher and start to pull and push
    /api/v1/fetcher GET List of active fetchers
    /api/v1/fetcher/{fetcherId} DELETE  Stop an active fetcher

    POST JSON Template

      "name": <string>,       // Name of session
      "url": <string>,        // Reachable HTTP url to HLS live stream
      "output": <string>,     // Output plugin name 
      "payload": <json>,      // Output plugin specific destination payload
      "concurrency": <number> // [OPTIONAL] Number of parallel downloads & uploads, default is 16
      "windowSize": <number>  // [OPTIONAL] Window size (seconds) for Media Playlist uploaded to output, default is 120

    Example MediaPackage:

      "name": "eyevinn-mp",
      "url": "",
      "output": "mediapackage",
      "payload": {
        "ingestUrls": [ { 
          "url": "",
          "username": "***",
          "password": "***"
        } ]

    Example S3:

      "name": "eyevinn-s3",
      "url": "",
      "output": "s3",
      "payload": {
        "bucket": "test-live-output",
        "folder": "S3_PLUGIN_CONTENT",

    Note: To use the S3 plugin you need to set your AWS environment variables including AWS_REGION

    Environment Variables

    • DEFAULT_LIVE_WINDOW_SIZE: The size of desired default window size (in seconds) for output stream when input is a live playlist type. Default 120
    • REMOVED_SEGMENT_TTL: For output plugins which support segment deletion, this variable determines the max age (in seconds) for a segment at an output destination that is no longer included in the output manifest before it is permanently deleted. Default 60


    • nodejs >= 12.0.0
    • In case of using the MediaPackageOutput plugin, you need to have an aws mediapackage channel set up expecting HLS as input.


    The service uses the debugging utility debug. To see relavant logs from the service, run with env DEBUG set to hls-*.

    Input HLS: Supported Types and Expected Behaviours

    HLS streams using fMP4 and encryption are not supported in this service currently. As well as multitrack (demuxed) streams. With that in mind, the service can take in HLS streams that are of the playlist types LIVE, EVENT or VOD.

    When the HLS stream is a LIVE type, then the default windowSize will be set to 120 seconds, unless windowSize is present in the POST JSON. If so then the default will be overwritten by the value found in the POST JSON.

    When the HLS stream is an EVENT type, then the default windowSize is set to -1 (infinite). It will try to have a growing window matching the source. The service will assume that the HLS EVENT stream will eventually end/become a VOD. However, if windowSize is present in the POST JSON, then the default will be overwritten by the value found in the POST JSON.

    When the HLS stream is a VOD type, then the fetcher session will only need to pull and push once. The service will try to push all segments to the output at once (limited by the set concurrency number). Therefor, the fetcher session will not count as an active fetcher, and will not show up in the list obtained from the GET/api/v1/fetcher endpoint.

    Limitations and Future Work

    • Currently, there are no operations in place to check if the request destination is already in use by another session. Allowing for multiple sessions uploading to the same destination and overwriting each others files.
    • Until fixed, when HLS source is type EVENT but still has a sliding window, no fetching occurs.
    • Fetcher List in GET endpoint do not communicate if fetcher sessions are faulty or not, only if they are active.

    About Eyevinn Technology

    Eyevinn Technology is an independent consultant firm specialized in video and streaming. Independent in a way that we are not commercially tied to any platform or technology vendor.

    At Eyevinn, every software developer consultant has a dedicated budget reserved for open source development and contribution to the open source community. This give us room for innovation, team building and personal competence development. And also gives us as a company a way to contribute back to the open source community.

    Want to know more about Eyevinn and how it is to work here. Contact us at!




    npm i @eyevinn/hls-pull-push

    DownloadsWeekly Downloads






    Unpacked Size

    213 kB

    Total Files


    Last publish


    • martinstark
    • jobi
    • slowmove
    • bwallberg