Table of Contents
A low-level client representing Amazon Kinesis Video Streams Archived Media (Kinesis Video Archived Media):
import boto3
client = boto3.client('kinesis-video-archived-media')
These are the available methods:
Check if an operation can be paginated.
Generate a presigned url given a client, its method, and arguments
The presigned url
Retrieves an HTTP Live Streaming (HLS) URL for the stream. You can then open the URL in a browser or media player to view the stream contents.
You must specify either the StreamName or the StreamARN .
An Amazon Kinesis video stream has the following requirements for providing data through HLS:
Kinesis Video Streams HLS sessions contain fragments in the fragmented MPEG-4 form (also called fMP4 or CMAF), rather than the MPEG-2 form (also called TS chunks, which the HLS specification also supports). For more information about HLS fragment types, see the HLS specification .
The following procedure shows how to use HLS with Kinesis Video Streams:
Note
Don't share or store this token where an unauthorized entity could access it. The token provides access to the content of the stream. Safeguard the token with the same measures that you would use with your AWS credentials.
The media that is made available through the playlist consists only of the requested stream, time range, and format. No other media data (such as frames outside the requested window or alternate bitrates) is made available.
Provide the URL (containing the encrypted session token) for the HLS master playlist to a media player that supports the HLS protocol. Kinesis Video Streams makes the HLS media playlist, initialization fragment, and media fragments available through the master playlist URL. The initialization fragment contains the codec private data for the stream, and other data needed to set up the video or audio decoder and renderer. The media fragments contain H.264-encoded video frames or AAC-encoded audio samples.
The media player receives the authenticated URL and requests stream metadata and media data normally. When the media player requests data, it calls the following actions:
Note
After the first media fragment is made available in a streaming session, any fragments that don't contain the same codec private data cause an error to be returned when those different media fragments are loaded. Therefore, the codec private data should not change between fragments in a session. This also means that the session fails if the fragments in a stream change from having only video to having both audio and video.
Data retrieved with this action is billable. See Pricing for details.
Note
If the ContainerFormat is MPEG_TS , this API is used instead of GetMP4InitFragment and GetMP4MediaFragment to retrieve stream media.
Data retrieved with this action is billable. For more information, see Kinesis Video Streams pricing .
Note
The following restrictions apply to HLS sessions:
You can monitor the amount of data that the media player consumes by monitoring the GetMP4MediaFragment.OutgoingBytes Amazon CloudWatch metric. For information about using CloudWatch to monitor Kinesis Video Streams, see Monitoring Kinesis Video Streams . For pricing information, see Amazon Kinesis Video Streams Pricing and AWS Pricing . Charges for both HLS sessions and outgoing AWS data apply.
For more information about HLS, see HTTP Live Streaming on the Apple Developer site .
See also: AWS API Documentation
Request Syntax
response = client.get_hls_streaming_session_url(
StreamName='string',
StreamARN='string',
PlaybackMode='LIVE'|'ON_DEMAND',
HLSFragmentSelector={
'FragmentSelectorType': 'PRODUCER_TIMESTAMP'|'SERVER_TIMESTAMP',
'TimestampRange': {
'StartTimestamp': datetime(2015, 1, 1),
'EndTimestamp': datetime(2015, 1, 1)
}
},
ContainerFormat='FRAGMENTED_MP4'|'MPEG_TS',
DiscontinuityMode='ALWAYS'|'NEVER',
DisplayFragmentTimestamp='ALWAYS'|'NEVER',
Expires=123,
MaxMediaPlaylistFragmentResults=123
)
The name of the stream for which to retrieve the HLS master playlist URL.
You must specify either the StreamName or the StreamARN .
The Amazon Resource Name (ARN) of the stream for which to retrieve the HLS master playlist URL.
You must specify either the StreamName or the StreamARN .
Whether to retrieve live or archived, on-demand data.
Features of the two types of session include the following:
Note
In LIVE mode, the newest available fragments are included in an HLS media playlist, even if there is a gap between fragments (that is, if a fragment is missing). A gap like this might cause a media player to halt or cause a jump in playback. In this mode, fragments are not added to the HLS media playlist if they are older than the newest fragment in the playlist. If the missing fragment becomes available after a subsequent fragment is added to the playlist, the older fragment is not added, and the gap is not filled.
In both playback modes, if FragmentSelectorType is PRODUCER_TIMESTAMP , and if there are multiple fragments with the same start timestamp, the fragment that has the larger fragment number (that is, the newer fragment) is included in the HLS media playlist. The other fragments are not included. Fragments that have different timestamps but have overlapping durations are still included in the HLS media playlist. This can lead to unexpected behavior in the media player.
The default is LIVE .
The time range of the requested fragment, and the source of the timestamps.
This parameter is required if PlaybackMode is ON_DEMAND . This parameter is optional if PlaybackMode is LIVE . If PlaybackMode is LIVE , the FragmentSelectorType can be set, but the TimestampRange should not be set. If PlaybackMode is ON_DEMAND , both FragmentSelectorType and TimestampRange must be set.
The source of the timestamps for the requested media.
When FragmentSelectorType is set to PRODUCER_TIMESTAMP and GetHLSStreamingSessionURLInput$PlaybackMode is ON_DEMAND , the first fragment ingested with a producer timestamp within the specified FragmentSelector$TimestampRange is included in the media playlist. In addition, the fragments with producer timestamps within the TimestampRange ingested immediately following the first fragment (up to the GetHLSStreamingSessionURLInput$MaxMediaPlaylistFragmentResults value) are included.
Fragments that have duplicate producer timestamps are deduplicated. This means that if producers are producing a stream of fragments with producer timestamps that are approximately equal to the true clock time, the HLS media playlists will contain all of the fragments within the requested timestamp range. If some fragments are ingested within the same time range and very different points in time, only the oldest ingested collection of fragments are returned.
When FragmentSelectorType is set to PRODUCER_TIMESTAMP and GetHLSStreamingSessionURLInput$PlaybackMode is LIVE , the producer timestamps are used in the MP4 fragments and for deduplication. But the most recently ingested fragments based on server timestamps are included in the HLS media playlist. This means that even if fragments ingested in the past have producer timestamps with values now, they are not included in the HLS media playlist.
The default is SERVER_TIMESTAMP .
The start and end of the timestamp range for the requested media.
This value should not be present if PlaybackType is LIVE .
The start of the timestamp range for the requested media.
If the HLSTimestampRange value is specified, the StartTimestamp value is required.
Note
This value is inclusive. Fragments that start before the StartTimestamp and continue past it are included in the session. If FragmentSelectorType is SERVER_TIMESTAMP , the StartTimestamp must be later than the stream head.
The end of the timestamp range for the requested media. This value must be within 3 hours of the specified StartTimestamp , and it must be later than the StartTimestamp value.
If FragmentSelectorType for the request is SERVER_TIMESTAMP , this value must be in the past.
If the HLSTimestampRange value is specified, the EndTimestamp value is required.
Note
This value is inclusive. The EndTimestamp is compared to the (starting) timestamp of the fragment. Fragments that start before the EndTimestamp value and continue past it are included in the session.
Specifies which format should be used for packaging the media. Specifying the FRAGMENTED_MP4 container format packages the media into MP4 fragments (fMP4 or CMAF). This is the recommended packaging because there is minimal packaging overhead. The other container format option is MPEG_TS . HLS has supported MPEG TS chunks since it was released and is sometimes the only supported packaging on older HLS players. MPEG TS typically has a 5-25 percent packaging overhead. This means MPEG TS typically requires 5-25 percent more bandwidth and cost than fMP4.
The default is FRAGMENTED_MP4 .
Specifies when flags marking discontinuities between fragments will be added to the media playlists. The default is ALWAYS when HLSFragmentSelector is SERVER_TIMESTAMP , and NEVER when it is PRODUCER_TIMESTAMP .
Media players typically build a timeline of media content to play, based on the timestamps of each fragment. This means that if there is any overlap between fragments (as is typical if HLSFragmentSelector is SERVER_TIMESTAMP ), the media player timeline has small gaps between fragments in some places, and overwrites frames in other places. When there are discontinuity flags between fragments, the media player is expected to reset the timeline, resulting in the fragment being played immediately after the previous fragment. We recommend that you always have discontinuity flags between fragments if the fragment timestamps are not accurate or if fragments might be missing. You should not place discontinuity flags between fragments for the player timeline to accurately map to the producer timestamps.
Specifies when the fragment start timestamps should be included in the HLS media playlist. Typically, media players report the playhead position as a time relative to the start of the first fragment in the playback session. However, when the start timestamps are included in the HLS media playlist, some media players might report the current playhead as an absolute time based on the fragment timestamps. This can be useful for creating a playback experience that shows viewers the wall-clock time of the media.
The default is NEVER . When HLSFragmentSelector is SERVER_TIMESTAMP , the timestamps will be the server start timestamps. Similarly, when HLSFragmentSelector is PRODUCER_TIMESTAMP , the timestamps will be the producer start timestamps.
The time in seconds until the requested session expires. This value can be between 300 (5 minutes) and 43200 (12 hours).
When a session expires, no new calls to GetHLSMasterPlaylist , GetHLSMediaPlaylist , GetMP4InitFragment , or GetMP4MediaFragment can be made for that session.
The default is 300 (5 minutes).
The maximum number of fragments that are returned in the HLS media playlists.
When the PlaybackMode is LIVE , the most recent fragments are returned up to this value. When the PlaybackMode is ON_DEMAND , the oldest fragments are returned, up to this maximum number.
When there are a higher number of fragments available in a live HLS media playlist, video players often buffer content before starting playback. Increasing the buffer size increases the playback latency, but it decreases the likelihood that rebuffering will occur during playback. We recommend that a live HLS media playlist have a minimum of 3 fragments and a maximum of 10 fragments.
The default is 5 fragments if PlaybackMode is LIVE , and 1,000 if PlaybackMode is ON_DEMAND .
The maximum value of 1,000 fragments corresponds to more than 16 minutes of video on streams with 1-second fragments, and more than 2 1/2 hours of video on streams with 10-second fragments.
dict
Response Syntax
{
'HLSStreamingSessionURL': 'string'
}
Response Structure
(dict) --
HLSStreamingSessionURL (string) --
The URL (containing the session token) that a media player can use to retrieve the HLS master playlist.
Gets media for a list of fragments (specified by fragment number) from the archived data in an Amazon Kinesis video stream.
Note
You must first call the GetDataEndpoint API to get an endpoint. Then send the GetMediaForFragmentList requests to this endpoint using the --endpoint-url parameter .
The following limits apply when using the GetMediaForFragmentList API:
See also: AWS API Documentation
Request Syntax
response = client.get_media_for_fragment_list(
StreamName='string',
Fragments=[
'string',
]
)
[REQUIRED]
The name of the stream from which to retrieve fragment media.
[REQUIRED]
A list of the numbers of fragments for which to retrieve media. You retrieve these values with ListFragments .
dict
Response Syntax
{
'ContentType': 'string',
'Payload': StreamingBody()
}
Response Structure
(dict) --
ContentType (string) --
The content type of the requested media.
Payload (StreamingBody) --
The payload that Kinesis Video Streams returns is a sequence of chunks from the specified stream. For information about the chunks, see PutMedia . The chunks that Kinesis Video Streams returns in the GetMediaForFragmentList call also include the following additional Matroska (MKV) tags:
The following tags will be included if an exception occurs:
Create a paginator for an operation.
Returns an object that can wait for some condition.
Returns a list of Fragment objects from the specified stream and timestamp range within the archived data.
Listing fragments is eventually consistent. This means that even if the producer receives an acknowledgment that a fragment is persisted, the result might not be returned immediately from a request to ListFragments . However, results are typically available in less than one second.
Note
You must first call the GetDataEndpoint API to get an endpoint. Then send the ListFragments requests to this endpoint using the --endpoint-url parameter .
See also: AWS API Documentation
Request Syntax
response = client.list_fragments(
StreamName='string',
MaxResults=123,
NextToken='string',
FragmentSelector={
'FragmentSelectorType': 'PRODUCER_TIMESTAMP'|'SERVER_TIMESTAMP',
'TimestampRange': {
'StartTimestamp': datetime(2015, 1, 1),
'EndTimestamp': datetime(2015, 1, 1)
}
}
)
[REQUIRED]
The name of the stream from which to retrieve a fragment list.
Describes the timestamp range and timestamp origin for the range of fragments to return.
The origin of the timestamps to use (Server or Producer).
The range of timestamps to return.
The starting timestamp in the range of timestamps for which to return fragments.
The ending timestamp in the range of timestamps for which to return fragments.
dict
Response Syntax
{
'Fragments': [
{
'FragmentNumber': 'string',
'FragmentSizeInBytes': 123,
'ProducerTimestamp': datetime(2015, 1, 1),
'ServerTimestamp': datetime(2015, 1, 1),
'FragmentLengthInMilliseconds': 123
},
],
'NextToken': 'string'
}
Response Structure
(dict) --
Fragments (list) --
A list of archived Fragment objects from the stream that meet the selector criteria. Results are in no specific order, even across pages.
(dict) --
Represents a segment of video or other time-delimited data.
FragmentNumber (string) --
The index value of the fragment.
FragmentSizeInBytes (integer) --
The total fragment size, including information about the fragment and contained media data.
ProducerTimestamp (datetime) --
The timestamp from the producer corresponding to the fragment.
ServerTimestamp (datetime) --
The timestamp from the AWS server corresponding to the fragment.
FragmentLengthInMilliseconds (integer) --
The playback duration or other time value associated with the fragment.
NextToken (string) --
If the returned list is truncated, the operation returns this token to use to retrieve the next page of results. This value is null when there are no more results to return.
The available paginators are:
paginator = client.get_paginator('list_fragments')
Creates an iterator that will paginate through responses from KinesisVideoArchivedMedia.Client.list_fragments().
See also: AWS API Documentation
Request Syntax
response_iterator = paginator.paginate(
StreamName='string',
FragmentSelector={
'FragmentSelectorType': 'PRODUCER_TIMESTAMP'|'SERVER_TIMESTAMP',
'TimestampRange': {
'StartTimestamp': datetime(2015, 1, 1),
'EndTimestamp': datetime(2015, 1, 1)
}
},
PaginationConfig={
'MaxItems': 123,
'PageSize': 123,
'StartingToken': 'string'
}
)
[REQUIRED]
The name of the stream from which to retrieve a fragment list.
Describes the timestamp range and timestamp origin for the range of fragments to return.
The origin of the timestamps to use (Server or Producer).
The range of timestamps to return.
The starting timestamp in the range of timestamps for which to return fragments.
The ending timestamp in the range of timestamps for which to return fragments.
A dictionary that provides parameters to control pagination.
The total number of items to return. If the total number of items available is more than the value specified in max-items then a NextToken will be provided in the output that you can use to resume pagination.
The size of each page.
A token to specify where to start paginating. This is the NextToken from a previous response.
dict
Response Syntax
{
'Fragments': [
{
'FragmentNumber': 'string',
'FragmentSizeInBytes': 123,
'ProducerTimestamp': datetime(2015, 1, 1),
'ServerTimestamp': datetime(2015, 1, 1),
'FragmentLengthInMilliseconds': 123
},
],
}
Response Structure
(dict) --
Fragments (list) --
A list of archived Fragment objects from the stream that meet the selector criteria. Results are in no specific order, even across pages.
(dict) --
Represents a segment of video or other time-delimited data.
FragmentNumber (string) --
The index value of the fragment.
FragmentSizeInBytes (integer) --
The total fragment size, including information about the fragment and contained media data.
ProducerTimestamp (datetime) --
The timestamp from the producer corresponding to the fragment.
ServerTimestamp (datetime) --
The timestamp from the AWS server corresponding to the fragment.
FragmentLengthInMilliseconds (integer) --
The playback duration or other time value associated with the fragment.