Files
daqbuffer/crates/httpret/static/documentation/api4.html
2023-07-10 14:45:25 +02:00

336 lines
10 KiB
HTML

<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8" />
<title>Databuffer API 4 Documentation</title>
<meta name="keywords" content="PSI, DAQ, Databuffer">
<meta name="author" content="Dominik Werder">
<link rel="shortcut icon" href="about:blank" />
<link rel="stylesheet" href="style.css" />
</head>
<body>
<h1>Databuffer API 4 Documentation</h1>
<p>Documented here is the databuffer http api 4. The "original" unversioned api is documented at
<a href="https://git.psi.ch/sf_daq/ch.psi.daq.databuffer/blob/master/ch.psi.daq.queryrest/Readme.md">this
location</a>.
</p>
<p>API version 1:
<a href="https://data-api.psi.ch/api/1/documentation/">https://data-api.psi.ch/api/1/documentation/</a>
</p>
<p>In order to keep the api surface as small as possible in comparison to api 0, we add functionality on demand,
so please feel free to create some Jira ticket!</p>
<h2>API functions</h2>
<p>Currently available functionality:</p>
<ul>
<li><a href="#timestamp-format">A note on how timestamps are encoded.</a></li>
<li><a href="#list-backends">List available backends.</a></li>
<li><a href="#api-version">More version information about the running service.</a></li>
<li><a href="#search-channel">Search channel.</a></li>
<li><a href="#map-pulse">Map pulse.</a></li>
<li><a href="#query-events">Query unbinned event data.</a></li>
<li><a href="#query-binned">Query binned data.</a></li>
</ul>
<a id="timestamp-format"></a>
<h2>Timestamp format</h2>
<p>The result encodes timestamps in the form:</p>
<pre>{
"tsAnchor": 1623909860, // Time-anchor of this result in UNIX epoch seconds.
"tsOffMs": [173, 472, 857, ...], // Millisecond-offset to tsAnchor for each event/bin-edge.
"tsOffNs": [422901, 422902, 422903, ...], // Nanosecond-offset to tsAnchor in addition to tsOffMs for each event/bin-edge.
}</pre>
<p>which results in these nanosecond-timestamps:</p>
<pre>1623909860573422901
1623909875671422902
1623909897932422903</pre>
<p>Formally: tsAbsolute = tsAnchor * 10<sup>9</sup> + tsOffMs * 10<sup>6</sup> + tsOffNs</p>
<p>Two reasons lead to this choice of timestamp format:</p>
<ul>
<li>Javascript can not represent the full nanosecond-resolution timestamps in a single numeric variable.</li>
<li>The lowest 6 digits of the nanosecond timestamp are anyway abused by the timing system to emit a pulse-id.
</li>
</ul>
<a id="list-backends"></a>
<h2>List available backends</h2>
<p><strong>Method:</strong> GET</p>
<p><strong>URL:</strong> https://data-api.psi.ch/api/4/backends</p>
<p><strong>Request header:</strong> "Accept" should be "application/json" for forward compatibility.</p>
<h4>CURL example:</h4>
<pre>
curl -H 'Accept: application/json' 'https://data-api.psi.ch/api/4/backends'
</pre>
<h4>Example response</h4>
<pre>{
"backends": [
"sf-databuffer",
"hipa-archive",
"gls-archive",
"proscan-archive",
"sls-archive"
]
}</pre>
<a id="api-version"></a>
<h2>Get version details</h2>
<p><strong>Method:</strong> GET</p>
<p><strong>URL:</strong> https://data-api.psi.ch/api/4/version</p>
<p><strong>Request header:</strong> "Accept" should be "application/json" for forward compatibility.</p>
<h4>CURL example:</h4>
<pre>
curl -H 'Accept: application/json' 'https://data-api.psi.ch/api/4/version'
</pre>
<h4>Example response</h4>
<pre>{ "data_api_version": { "major": 4, "minor": 0} }</pre>
<a id="search-channel"></a>
<h2>Search channel</h2>
<p><strong>Method:</strong> GET</p>
<p><strong>URL:</strong> https://data-api.psi.ch/api/4/search/channel</p>
<p><strong>Query parameters:</strong> (all optional)</p>
<ul>
<li>backend (e.g. "sf-databuffer", "sls-archive", ... any from list-backends API)</li>
<li>nameRegex (e.g. "LSCP.*6")</li>
<li>sourceRegex (e.g. "178:9999")</li>
<li>descriptionRegex (e.g. "celsius")</li>
</ul>
<p><strong>Request header:</strong> "Accept" should be "application/json" for forward-compatibility but can be
omitted for e.g. a quick manual search using CURL.</p>
<p>Full channel list is long, so it's encouraged to provide a search string of some minimal length.</p>
<h4>CURL example:</h4>
<pre>
curl -H 'Accept: application/json' 'https://data-api.psi.ch/api/4/search/channel?sourceRegex=CV.E.+37&nameRegex=120.+y2$'
</pre>
<h4>Example response:</h4>
<p><strong>Keys always present: </strong>name, backend.</p>
<p>Other keys are optional, it depends on the data found on disk: sometimes there is an empty string on disk,
sometimes
that key is missing.</p>
<pre>{
"channels": [
{
"name": "S10MA01-DBPM120:Y2",
"backend": "sf-databuffer",
"source": "tcp://S20-CVME-DBPM2371:9000",
"type": "Float32",
"shape": [],
"unit": "",
"description": ""
},
{
"name": "S20SY02-DBPM120:Y2",
"backend": "sf-databuffer",
"source": "tcp://S20-CVME-DBPM2371:9000",
"type": "Float32",
"shape": [],
"unit": "",
"description": ""
},
{
"isApi0": true,
"name": "EXAMPLE-CHANNEL-FROM-API-0-BACKEND",
"backend": "twlha-databuffer",
"source": "tcp://.....",
"type": "int32",
"shape": [],
"unit": "",
"description": ""
}
]
}</pre>
<p>The search constraints are AND'd.</p>
<a id="map-pulse"></a>
<h2>Map pulse</h2>
<p><strong>Method:</strong> GET</p>
<p><strong>URL:</strong> https://data-api.psi.ch/api/4/map/pulse/BACKEND/PULSE</p>
<p><strong>Request header:</strong> "Accept" should be "application/json" for forward-compatibility but can be
omitted for e.g. a quick manual search using CURL.</p>
<h4>CURL example:</h4>
<pre>
curl -H 'Accept: application/json' 'https://data-api.psi.ch/api/4/map/pulse/sf-databuffer/424242'
</pre>
<a id="query-events"></a>
<h2>Query event data</h2>
<p>Returns the full event values in a given time range.</p>
<p><strong>Method:</strong> GET</p>
<p><strong>URL:</strong> https://data-api.psi.ch/api/4/events</p>
<p><strong>Query parameters:</strong></p>
<ul>
<li>backend (e.g. "sf-databuffer")</li>
<li>channelName (e.g. "S10CB02-RBOC-DCP10:FOR-AMPLT-AVG")</li>
<li>begDate (e.g. "2021-05-26T07:10:00.000Z")</li>
<li>endDate (e.g. "2021-05-26T07:16:00.000Z")</li>
</ul>
<p><strong>Request header:</strong> "Accept" should be "application/json" for forward compatibility.</p>
<h4>Timeout</h4>
<p>If the requested range takes too long to retrieve, then the flags <strong>timedOut: true</strong> will be set.
</p>
<h4>CURL example:</h4>
<pre>
curl -H 'Accept: application/json' 'https://data-api.psi.ch/api/4/events?backend=sf-databuffer
&channelName=S10CB02-RBOC-DCP10:FOR-AMPLT-AVG&begDate=2021-05-26T07:10:00.000Z&endDate=2021-05-26T07:16:00.000Z'
</pre>
<p>Example response:</p>
<pre>
{
"rangeFinal": true,
"tsAnchor": 1623763172,
"tsMs": [
5,
15,
25,
35
],
"tsNs": [
299319,
299320,
299321,
299322
],
"values": [
0.6080216765403748,
0.6080366969108582,
0.6080275177955627,
0.6080636382102966
]
}
</pre>
<h4>Finalised range</h4>
<p>If the server can determine that no more data will be added to the requested time range
then it will add the flag <strong>rangeFinal: true</strong> to the response.</p>
<a id="query-binned"></a>
<h2>Query binned data</h2>
<p><strong>Method:</strong> GET</p>
<p><strong>URL:</strong> https://data-api.psi.ch/api/4/binned</p>
<p><strong>Query parameters:</strong></p>
<ul>
<li>backend (e.g. "sf-databuffer")</li>
<li>channelName (e.g. "SLAAR-LSCP4-LAS6891:CH7:1")</li>
<li>begDate (e.g. "2021-05-26T07:10:00.000Z")</li>
<li>endDate (e.g. "2021-05-26T07:16:00.000Z")</li>
<li>binCount (number of requested bins in time-dimension, e.g. "6". The actual number of returned bins depends on
bin-cache-grid-resolution. The server tries to find a reasonable match.)</li>
<li>binningScheme (optional)</li>
<ul>
<li>if not specified: default is "binningScheme=timeWeightedScalar".</li>
<li>"binningScheme=unweightedScalar": non-weighted binning, waveform gets first averaged to a scalar.</li>
<li>"binningScheme=timeWeightedScalar": time-weighted binning, waveform gets first averaged to a scalar.
</li>
<li>"binningScheme=binnedX&binnedXcount=13": waveform gets first binned to 13 bins in X-dimension
(waveform-dimension).</li>
<li>"binningScheme=binnedX&binnedXcount=0": waveform is not binned in X-dimension but kept at full length.
</li>
</ul>
</ul>
<p><strong>Request header:</strong> "Accept" should be "application/json" for forward compatibility.</p>
<h4>CURL example:</h4>
<pre>
curl -H 'Accept: application/json' 'https://data-api.psi.ch/api/4/binned?backend=sf-databuffer
&channelName=SLAAR-LSCP4-LAS6891:CH7:1&begDate=2021-05-25T00:00:00.000Z&endDate=2021-05-26T00:00:00.000Z&binCount=3'
</pre>
<h4>Partial result</h4>
<p>If the requested range takes longer time to retrieve, then a partial result with at least one bin is returned.
The partial result will contain the necessary information to send another request with a range that
starts with the first not-yet-retrieved bin.
This information is provided by the <strong>continueAt</strong> and <strong>missingBins</strong> fields.
This enables the user agent to start the presentation to the user while updating the user interface
as new bins are received.</p>
<h4>Example response (without usage of binningScheme):</h4>
<pre>{
"tsAnchor": 1623769850,
"ts1Ms": [
0,
10000,
20000
],
"ts1Ns": [
0,
0,
0
],
"ts2Ms": [
10000,
20000,
30000
],
"ts2Ns": [
0,
0,
0
],
"counts": [
1000,
999,
1000
],
"avgs": [
16204.087890625,
16204.3798828125,
16203.9296875
],
"maxs": [
48096,
48100,
48094
],
"mins": [
0,
0,
0
],
"rangeFinal": true
}
</pre>
<h4>Complete result</h4>
<p>If the result does not contain a <strong>continueAt</strong> key then the result is complete.</p>
<h4>Finalised range</h4>
<p>If the server can determine that no more data will be added to the requested time range
then it will add the flag <strong>rangeFinal: true</strong> to the response.</p>
<h2>Feedback and comments very much appreciated!</h2>
<p>dominik.werder@psi.ch</p>
<p>or please assign me a JIRA ticket.</p>
<div id="footer"></div>
</body>
</html>