Merge pull request 'intelligence: a plate read on a patrol channel is a traffic stop' (#181) from feat/plate-read-stops into main

drb-c2-core/app/internal/intelligence.py

@@ -517,12 +517,33 @@ _NO_BACKSTOP_TG = re.compile(r"\b(mta|rail|railroad|train|transit|bridges? and t
                              r"rescue|ambulance|dpw|public works)\b", re.IGNORECASE)
 
 
+# A plate read aloud — two or more phonetic letters then 3-7 digits:
+# "Frank David Boy, 4514", "Lincoln, Charlie, Robert, 7-4-0-7". On a patrol
+# channel that is a unit running a car it has stopped. Held-out replay of
+# 09-21 (server-26#170): Ossining's Post 4 stops were read out only as plates
+# and never became incidents.
+_PHONETIC = (r"(?:adam|alpha|baker|boy|bravo|charlie|charles|david|delta|eddie|edward|echo|frank|"
+             r"george|golf|henry|hotel|ida|india|john|juliet|king|kilo|lincoln|lima|larry|mary|"
+             r"michael|mike|nora|nancy|november|ocean|oscar|peter|paul|papa|queen|robert|romeo|"
+             r"sam|sierra|tom|tango|union|uniform|victor|william|whiskey|x-ray|xray|young|yankee|zebra|zulu)")
+_PLATE_READ = re.compile(rf"\b{_PHONETIC}(?:[,\s]+{_PHONETIC}){{1,3}}[,\s]+\d(?:[\s-]?\d){{2,6}}\b",
+                         re.IGNORECASE)
+
+
 def _self_initiated_backstop(
     text: str, tags: list, incident_type: Optional[str], severity: str,
     talkgroup_name: Optional[str] = None,
 ) -> tuple[list, Optional[str], str]:
     if talkgroup_name and _NO_BACKSTOP_TG.search(talkgroup_name):
         return tags, incident_type, severity
+    # A plate read only stands for a stop when extraction found no other
+    # event in the call: the plate on an MVA, a tow or a parked-car complaint
+    # belongs to that event, not to a new stop.
+    if not tags and _PLATE_READ.search(text or ""):
+        tags = ["traffic-stop"]
+        incident_type = incident_type or "police"
+        if severity == "routine":
+            severity = "minor"
     for pattern, tag in _SELF_INITIATED:
         m = pattern.search(text or "")
         if not m or _NEGATED.search(text[: m.start()]):

drb-c2-core/tests/test_clearance_tracking.py

@@ -140,3 +140,17 @@ def test_short_stop_report_opens_a_scene():
             return await intelligence.extract_scenes("c1", "Adam 3 on a stop.", "Ch 1 (Patched with 155.310)")
     scenes = asyncio.run(run())
     assert len(scenes) == 1 and scenes[0]["tags"] == ["traffic-stop"] and scenes[0]["incident_type"] == "police"
+
+
+def test_plate_read_on_a_patrol_channel_is_a_stop():
+    from app.internal.intelligence import _self_initiated_backstop as b
+    ch = "Ossining - Police Dispatch"
+    for t in ("Post 4. 52-62, 3-3. Hemlock Circle. Frank David Boy, 4514. 10-8.",
+              "4, Ossining. 52-22, Ramapo, New York. Lincoln, Charlie, Robert, 7-4-0-7 on a Chevy.",
+              "New York, Mary, Charlie, Nora, 5-8-6-7."):
+        assert b(t, [], None, "routine", ch) == (["traffic-stop"], "police", "minor"), t
+    # a plate on a call that is already about something else stays with it
+    assert b("MVA, plate Mary George Sam 2740", ["mva"], "accident", "moderate", ch) == (["mva"], "accident", "moderate")
+    # not on rail/bridge channels, and not without digits
+    assert b("Frank David Boy 4514", [], None, "routine", "MTA Bridges and Tunnels - Whitestone") == ([], None, "routine")
+    assert b("Charlie, David, go ahead.", [], None, "routine", ch) == ([], None, "routine")