API Reference

Base URL

https://dvpx.algsoft.net.tr/api.php

GET Last Heard


        api.php?action=lastheard&limit=5

Call history reported by the reflectors. A row is opened when a transmission STARTS (active=true, duration_ms=0 at that point) and updated with the final length when it ends — use the active field for a "currently talking" indicator.

Parameter	Default	Description
`tg`	`0`	talkgroup number. 0/omitted = all talkgroups.
`limit`	`50`	number of records, 1-200.
`since`	`—`	only records NEWER than this. ISO-8601 or unix seconds. For incremental polling, send back the last started_at you saw.

GET Online Stations


        api.php?action=online

Stations currently connected to the reflectors. Apps pick call targets from this list. Reflectors refresh it about every 10 seconds; entries older than 90 seconds disappear. The talkgroup field returns 0 when the station is not on a talkgroup.

GET Talkgroups


        api.php?action=talkgroups

Talkgroups that are active on the panel. When dmr_id is given the list is personalised for that identity; without the parameter you get the public list.

Parameter	Default	Description
`dmr_id`	`—`	supply your identity to receive the talkgroup list assigned to you (an ESSID-suffixed id is accepted too).

GET Reflectors


        api.php?action=servers

Reflectors you can connect to, with their TCP signalling and UDP voice ports. To appear here a reflector must be online, approved by the administrator AND have contacted the panel within the last 2 minutes — so every server listed is genuinely alive.

GET Everything (default)


        api.php

Servers + talkgroups + DVPX Voice Engine settings. This is what the app fetches on startup.

GET Health Check


        api.php?action=ping

Responds without touching the database; suitable for uptime monitoring.

lastheard response fields

Field	Type	Description
`id`	`int`	unique id of the call record
`source_id`	`int`	DMR ID of the transmitting station
`callsign`	`string`	callsign (empty if unknown)
`target_id`	`int`	target: TG number, or the peer DMR ID for a private call
`tg_name`	`string`	talkgroup name (empty for private calls)
`call_type`	`int`	0 = talkgroup, 1 = private call
`private`	`bool`	convenience flag for call_type == 1
`frames`	`int`	number of voice frames sent (20 ms per frame)
`duration_ms`	`int`	transmission length in milliseconds. It is 0 while the transmission is still running.
`active`	`bool`	true means this transmission is STILL RUNNING — use it for a "currently talking" indicator.
`server_id`	`int\|null`	panel id of the reflector that logged it
`server_name`	`string`	reflector name
`started_at`	`string`	start of transmission, ISO-8601 UTC
`started_at_unix`	`int`	same instant as unix seconds — easier to parse

Records are returned by started_at descending (newest first). A station that transmitted several times appears several times — de-duplicate by source_id if your list shows one row per station.

Examples

curl — last 10 calls on TG 90

curl -s "https://dvpx.algsoft.net.tr/api.php?action=lastheard&tg=90&limit=10"

JavaScript — straight from the browser (CORS is open)

const r = await fetch("https://dvpx.algsoft.net.tr/api.php?action=lastheard&tg=90");
const { calls } = await r.json();
calls.forEach(c => console.log(c.callsign, c.tg_name, c.duration_ms));

Python — incremental polling (new records only)

import requests, time

url, since = "https://dvpx.algsoft.net.tr/api.php", None
while True:
    p = {"action": "lastheard", "limit": 200}
    if since:
        p["since"] = since
    calls = requests.get(url, params=p, timeout=10).json()["calls"]
    for c in reversed(calls):          # eskiden yeniye
        print(c["started_at"], c["callsign"], "->", c["target_id"])
    if calls:
        since = calls[0]["started_at"] # en yeni kayıt
    time.sleep(5)

📞 Full-Duplex Calling TCP signalling

Full duplex is a private call where two stations talk at the same time without holding PTT. Nothing changes in AUDIO TRANSPORT — the reflector already forwards packets regardless of direction, so both ways flow at once. The only addition is CALL SETUP: who is calling whom, whether the other side accepted, and when the call ended.

These lines travel on the TCP signalling channel between client and reflector; they are unrelated to api.php.

→ FDX INVITE <hedefDmrId>
← FDX RING <kaynakId> <cagriIsareti>
→ FDX ACCEPT | REJECT | END
← FDX ACCEPTED | REJECTED | ENDED
← FDX FAIL  NO_TARGET | BUSY | SELF | NO_CALL

ACCEPTED goes to BOTH sides — if one waited for PTT the conversation would stay one-way. A reflector that has not been updated returns ERR UNKNOWN_COMMAND and the client shows "this reflector does not support full duplex". Older clients never send FDX, so nothing changes for them.

📡 Reflector Control API token required

This section is NOT for app developers; it is for reflector operators. Reflectors never touch the database: they fetch the talkgroup list, the block list and the user grants from this endpoint, and report what happens (online list, call records) back to it.

Address reflectors connect to — config.json → dashboard.url

https://dvpx.algsoft.net.tr/reflector.php

Method	Endpoint	Description
GET	`?action=ping`	No token needed. A "did I get the address right?" test — you can open it in a browser.
GET	`?action=snapshot`	Policy only: talkgroup list, blocked ids, user grants. For debugging.
POST	`?action=heartbeat`	The real endpoint. Sends telemetry and receives the policy (only if it CHANGED) in one request.

Authentication — any one of these

Authorization: Bearer dvpx_xxxxxxxx
X-DVPX-Token: dvpx_xxxxxxxx
{"token": "dvpx_xxxxxxxx", ...}

Both headers are sent because some shared hosts never pass Authorization through to PHP. The token is shown once in the panel; only its SHA-256 digest is stored in the database.

Testing the connection

# Is the address right? (no token)
curl -sS "https://dvpx.algsoft.net.tr/reflector.php?action=ping"

# Is the token valid? It should return your talkgroup list
curl -sS -H "X-DVPX-Token: dvpx_xxxx" "https://dvpx.algsoft.net.tr/reflector.php?action=snapshot"

Error contract — this drives the reflector's behaviour

HTTP	`error`	`fatal`	What the reflector does
401	`unauthorized`	`true`	LOCKS: refuses new logins and drops connected users.
403	`pending_approval`	`true`	Waits locked; enters service AUTOMATICALLY the moment the admin approves.
503	`server_error`	`false`	Treats it as TEMPORARY: keeps carrying voice and accepting logins from its cached policy, queues its reports.

This distinction is deliberate: a temporary database or network fault NEVER closes the network to logins. Only an explicit rejection (401/403) locks a reflector. Reflector setup: dvpx-reflector/KURULUM.md

Usage notes

Responses are cached for 5 seconds (Cache-Control: max-age=5). Polling faster than that gains nothing.
If you track a busy feed, use the since parameter instead of pulling 200 rows every time — it spares both the server and your bandwidth.
Timestamps are UTC. Convert to local time on the client side.
A record appears the moment a transmission STARTS, with active=true; when it ends the same record is updated with the final length. Seeing the same id twice is an update, not a duplicate.
On failure the body is {"ok": false, "error": "...", "message": "..."} with a 4xx/5xx status code.

🔌 DVPX Public API