Plugins

Overview

To allow the player to be extended, a system of “plugins” has been added.

Those plugins are often under the form of functions passed as an argument to the loadVideo API call.

segmentLoader

The segmentLoader is a function that can be included as an option of the loadVideo API call.

A segmentLoader allows to define a custom audio/video segment loader (it might on the future work for other types of segments, so always check the type if you only want those two). The segment loader is the part performing the segment request. One usecase where you might want to set your own segment loader is to integrate Peer-to-Peer segment downloading through the player.

Before the complete documentation, let’s write an example which will just use an XMLHttpRequest (it has no use, as our implementation does the same thing and more):

/**
 * @param {Object} segmentInfo - Information about the segment to download
 * @param {Object} callbacks - Object containing several callbacks to indicate
 * that the segment has been loaded, the loading operation has failed or to
 * fallback to our default implementation. More information on this object below
 * this code example.
 * @returns {Function|undefined} - If a function is defined in the return value,
 * it will be called if and when the request is canceled.
 */
const customSegmentLoader = (segmentInfo, callbacks) => {

  // we will only use this custom loader for videos segments.
  // we will also ignore edge cases where the URL is undefined.
  if (segmentInfo.trackType !== "video" || segmentInfo.url === undefined) {
    callbacks.fallback();
    return;
  }

  const xhr = new XMLHttpRequest();
  const sendingTime = performance.now();

  xhr.onload = function onXHRLoaded(r) {
    if (200 <= xhr.status && xhr.status < 300) {
      const duration = performance.now() - sendingTime;
      const size = r.total;
      const data = xhr.response;
      callbacks.resolve({ duration, size, data });
    } else {
      const err = new Error("didn't work");
      callbacks.reject(err);
    }
  };

  xhr.onprogress = function onXHRProgress(event) {
    const currentTime = performance.now();
    callbacks.progress({ duration: currentTime - sendingTime,
                         size: event.loaded,
                         totalSize: event.total });
  };

  xhr.onerror = function onXHRError() {
    const err = new Error("didn't work");
    callbacks.reject(err);
  };

  xhr.open("GET", segmentInfo.url);
  xhr.responseType = "arraybuffer";

  const ranges = segmentInfo.byteRanges;
  if (ranges) {
    // Theoretically, we could have multiple non-contiguous byte ranges, which
    // would imply several requests.
    // In reality, this is extremely rare so we just create a global byte-range
    // encompassing all in this example.
    const range = [ranges[0][0], ranges[ranges.length - 1][1]];
    if (range[1] && range[1] !== Infinity) {
      xhr.setRequestHeader("Range", `bytes=${range[0]}-${range[1]}`);
    } else {
      xhr.setRequestHeader("Range", `bytes=${range[0]}-`);
    }
  }

  xhr.send();

  return () => {
    xhr.abort();
  };
};

As you can see, this function takes two arguments:

1. segmentInfo (object): An Object giving information about the wanted segments. This object contains the following properties:

   • url (string|undefined): The URL the segment request should normally be performed at.

     This property can be undefined in a condition where the segment URL either doesn’t exist or has not been communicated by the Manifest.

   • timeout (number|undefined: Timeout in milliseconds after which a request should preferably be aborted, according to current configuration.

     This property is mainly indicative, you may or may not want to exploit this information depending on your use cases.

   • isInit (boolean|undefined): If true this segment is an initialization segment which contains no decodable data.

     Those types of segment are mainly there for initialization purposes, such as giving initial infos to the decoder on subsequent media segments that will be pushed.

     Note that if isInit is false, it only means that the segment contains decodable media, it can also contain important initialization information.

     If undefined, we could not determine whether this segment was an initialization segment. This case is not currently possible but may be in future versions.

   • byteRanges (Array.<[number, number]>|undefined): If defined, only the corresponding byte-ranges, which are subsets in bytes of the full data concerned, should be loaded.

     If multiple non-contiguous byte-ranges are given, the result should be the concatenation of those byte-ranges, in the same order.

     For example [[0, 100], [150, 180]] means that the bytes of both 0 to 100 (included) and from 150 to 180 (included) should be requested. The communicated result should then be a concatenation of both in the same order.

   • trackType (string): The concerned type of track. Can be "video", "audio", "text" (for subtitles)

2. callbacks: An object containing multiple callbacks to allow this segmentLoader to communicate various events to the RxPlayer.

   This Object contains the following functions:

   • resolve: To call after the segment is loaded, to communicate it to the RxPlayer.

     When called, it should be given an object with the following properties:

     • data (ArrayBuffer|Uint8Array) - the segment data.

     • duration (Number|undefined) - the duration the request took, in milliseconds.

       This value may be used to estimate the ideal user bandwidth.

     • size (Number|undefined) size, in bytes, of the total downloaded response.

       This value may be used to estimate the ideal user bandwidth.

   • progress - Callback to call when progress information is available on the current request. This callback allows to improve our adaptive streaming logic by better predicting the bandwidth before the request is finished and whether a request is stalling.

     When called, it should be given an object with the following properties:

     • duration (Number) - The duration since the beginning of the request, in milliseconds.

     • size (Number) - the current size loaded, in bytes.

     • totalSize (Number|undefined) - the whole size of the wanted data, in bytes. Can be let to undefined when not known.

   • reject: Callback to call when an error is encountered which made loading the segment impossible.

     It is recommended (but not enforced) to give it an Object or error instance with the following properties:

     • canRetry (boolean|undefined): If set to true, the RxPlayer may retry the request (depending on the configuration set by the application).

       If set to false, the RxPlayer will never try to retry this request and will probably just stop the current content.

       If not set or set to undefined, the RxPlayer might retry or fail depending on other factors.

   • fallback: Callback to call if you want to call our default implementation instead for loading this segment. No argument is needed.

The segmentLoader can also return a function, which will be called if/when the request is aborted. You can define one to clean-up or dispose all resources.

manifestLoader

The manifestLoader is a function that can be included as an option of the loadVideo API call.

A manifestLoader allows to define a custom Manifest loader.

Before the complete documentation, let’s write an example which will just use an XMLHttpRequest (it has no use, as our implementation does the same thing and more):

/**
 * @param {Object} manifestInfo - Information about the Manifest to download
 * @param {Object} callbacks - Object containing several callbacks to indicate
 * that the manifest has been loaded, the loading operation has failed or to
 * fallback to our default implementation. More information on this object below
 * this code example.
 * @returns {Function|undefined} - If a function is defined in the return value,
 * it will be called if and when the request is canceled.
 */
const customManifestLoader = (manifestInfo, callbacks) => {
  const { url } = manifestInfo;
  const xhr = new XMLHttpRequest();
  const baseTime = performance.now();

  xhr.onload = (r) => {
    if (200 <= xhr.status && xhr.status < 300) {
      const duration = performance.now() - baseTime;

      const now = Date.now();
      const receivingTime = now;

      // Note: We could have calculated `sendingTime` before the request, but
      // that date would be wrong if the user updated the clock while the
      // request was pending.
      // `performance.now` doesn't depend on the user's clock. It is thus a
      // better candidate here.
      // This is why we re-calculate the sendingTime a posteriori, we are now
      // sure to be aligned with the current clock.
      const sendingTime = now - duration;

      // the request could have been redirected,
      // we have to feed back the real URL
      const _url = xhr.responseURL || url;

      const size = r.total;
      const data = xhr.response;
      callbacks.resolve({
        url: _url,
        sendingTime,
        receivingTime,
        duration,
        size,
        data,
      });
    } else {
      const err = new Error("didn't work");
      err.xhr = xhr;
      callbacks.reject(err);
    }
  };

  xhr.onerror = () => {
    const err = new Error("didn't work");
    err.xhr = xhr;
    callbacks.reject(err);
  };

  xhr.open("GET", url);
  xhr.responseType = "document";

  xhr.send();

  return () => {
    xhr.abort();
  };
};

As you can see, this function takes three arguments:

1. manifestInfo (object): An Object giving information about the wanted Manifest. This object contains the following properties:

   • url (string|undefined): The URL the Manifest request should normally be performed at.

     This argument can be undefined in very rare and specific conditions where the Manifest URL doesn’t exist or has not been communicated by the application.

   • timeout (number|undefined): Timeout in milliseconds after which a request should preferably be aborted, according to current configuration.

     This property is mainly indicative, you may or may not want to exploit this information depending on your use cases.

2. callbacks: An object containing multiple callbacks to allow this manifestLoader to communicate the loaded Manifest or an encountered error to the RxPlayer.

   This Object contains the following functions:

   • resolve: To call after the Manifest is loaded, to communicate it to the RxPlayer.

     When called, it should be given an object with the following properties:

     • data - the Manifest data. Many formats are accepted depending on what makes sense in the current transport: string, Document, ArrayBuffer, Uint8Array, object.

     • duration (Number|undefined) - the duration of the request, in milliseconds.

     • size (Number|undefined) size, in bytes, of the total downloaded response.

     • url (string|undefined) - url of the Manifest (after any potential redirection if one).

     • sendingTime (number|undefined) - Time at which the manifest request was done as a unix timestamp in milliseconds.

     • receivingTime (number|undefined) - Time at which the manifest request was finished as a unix timestamp in milliseconds.

   • reject: Callback to call when an error is encountered which made loading the Manifest impossible.

     It is recommended (but not enforced) to give it an Object or error instance with the following properties:

     • canRetry (boolean|undefined): If set to true, the RxPlayer may retry the request (depending on the configuration set by the application).

       If set to false, the RxPlayer will never try to retry this request and will probably just stop the current content.

       If not set or set to undefined, the RxPlayer might retry or fail depending on other factors.

   • fallback: Callback to call if you want to call our default implementation instead for this Manifest. No argument is needed.

The manifestLoader can also return a function, which will be called if/when the request is aborted. You can define one to clean-up or dispose all resources.

representationFilter

The representationFilter is a function that can be included as an option of the loadVideo API call.

A representationFilter allows you to filter out Representations (i.e. media qualities) based on its attributes.

The representationFilter will be called each time we load a Manifest with two arguments:

• representation {Object}: An object describing the Representation.

  This object contains the following properties:

  • id (string): The id used to identify this Representation.

  • bitrate (Number|undefined): The bitrate of this Representation, in bits per seconds.

    undefined if unknown.

  • width (Number|undefined): If the Representation is from a video track and if its width is known, this is the width of video, in pixels.

  • height (Number|undefined): If the Representation is from a video track and if its height is known, this is the height of video, in pixels.

  • codec (string|undefined): The codec the Representation is in.

  • frameRate (Number|undefined): If the Representation is from a video track and if its frame rate is known, this is the frame rate of video, in image per seconds.

  • hdrInfo (Object|undefined): If the Representation is from a video track and if it has HDR information associated to it, this is set to an object describing the hdr characteristics of the track. (see HDR support documentation)

  • contentProtections (Object|undefined): Encryption information linked to this content.

    If set to an Object, the Representation is known to be encrypted. If unset or set to undefined the Representation is either unencrypted or we don’t know if it is.

    When set to an object, it may contain the following properties:

    • keyIds (Array.<Uint8Array>|undefined): Known key ids linked to that Representation.

• context {Object}: Basic context about this Representation. Contains the following keys:

  • trackType {string}: The concerned type of track. Can be "video", "audio", "text" (for subtitles)

  • language {string|undefined}: The language the Representation is in, as announced by the Manifest.

  • normalizedLanguage {string|undefined}: An attempt to translate the language into an ISO 639-3 code. If the translation attempt fails (no corresponding ISO 639-3 language code is found), it will equal the value of language

  • isClosedCaption {Boolean|undefined}: If set to true and if this is a text track, the Representation links to subtitles with added hints for the hard of hearing.

  • isAudioDescription {Boolean|undefined}: If set to true and if this is an audio track, the Representation links to an audio track with added commentary for the visually impaired.

  • isDub {Boolean|undefined}): If set to true and if this is an audio track, then this audio track is a “dub”, meaning it was recorded in another language than the original. If set to false, we know that this audio track is in an original language.

    This property is undefined if we do not known whether it is in an original language or if does not apply for the track type.

  • isSignInterpreted {Boolean|undefined}): If set to true and if this is a video track, then it contains visual sign interpretation.

This function should then returns true if the Representation should be kept or false if it should be removed.

For example, here is a representationFilter that removes video Representations with a video resolution higher than HD (1920x1080):

/**
 * @param {Object} representationInfo
 * @param {Object} infos - supplementary information about the given
 * Representation.
 * @returns {boolean}
 */
function representationFilter(representationInfo, infos) {
  if (infos.trackType === "video") {
    // If video representation, allows only those for which the height and width
    // is known to be below our 1920x1080 limit
    const { width, height } = representationInfo;
    return width != null && height != null && width <= 1920 && height <= 1080;
  }

  // Otherwise, allow all non-video representations
  return true;
}