correlator: LLM tier reads the scene's transcript, not the whole call (server-26#102)

The last leg of the #80/#95 scene-context leak. llm_correlator._call_block
read call_doc's whole-call transcript for every scene, so on a multi-scene
call every scene's cheap-tier and tiebreaker decision was made against text
that also contained the other scenes.

- intelligence.py: each processed[] scene now carries its own "transcript" —
  transcript_corrected, else this scene's segments joined, else (single scene)
  the whole transcript.
- _build_context / preview_correlation / correlate_call: take a `transcript`
  param; _build_context resolves ctx["scene_transcript"] from it, falling
  back to the call doc (sweep, single-scene, tests) — the fallback is kept
  here, unlike embedding/severity, because a scene always has real text.
- upload.py: both scene loops pass scene["transcript"].
- llm_correlator._call_block: reads ctx["scene_transcript"] (call-doc
  fallback retained for test-built ctx).
- recorrelation_sweep: passes the call doc's text explicitly.
- +1 regression test. Full c2-core suite green (296 passed, sandboxed venv).

NOT for merge until the running correlation measurement window closes and its
dump is analysed — deploying a correlator change mid-window would mix old and
new behaviour in the sample.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
Logan Cusano
2026-09-07 00:06:37 -04:00
co-authored by Claude Sonnet 5
parent b430cf32f2
commit ef1e3d7f9d
6 changed files with 75 additions and 7 deletions
@@ -304,6 +304,39 @@ async def test_a_scene_is_judged_on_its_own_embedding_and_severity():
assert ctx["call_severity"] == "major"
@pytest.mark.asyncio
async def test_the_llm_tier_reads_the_scene_transcript_not_the_whole_call():
"""
server-26#102. intelligence.py writes only the primary scene's corrected
text to calls/{id}. _call_block (the LLM correlation prompt) must reason
over the SCENE being correlated, not a whole-call transcript that also
contains the other scenes. _build_context threads the scene's text in;
with no scene text it falls back to the call doc (sweep / single-scene).
"""
with patch("app.internal.incident_correlator.fstore") as mock_fstore:
mock_fstore.doc_get = AsyncMock(return_value={
"transcript": "scene one about a fire. scene two about a traffic stop.",
})
mock_fstore.collection_list = AsyncMock(return_value=[])
scene = await _build_context(
call_id="call-1", units=None, vehicles=None, cleared_units=None,
location_coords=None, reference_time=NOW,
system_id="sys-1", talkgroup_id=383, talkgroup_name=DISPATCH_TG,
tags=[], incident_type="police", location=None,
reassignment=False, create_if_new=True,
transcript="scene two about a traffic stop.",
)
fallback = await _build_context(
call_id="call-1", units=None, vehicles=None, cleared_units=None,
location_coords=None, reference_time=NOW,
system_id="sys-1", talkgroup_id=383, talkgroup_name=DISPATCH_TG,
tags=[], incident_type="police", location=None,
reassignment=False, create_if_new=True,
)
assert scene["scene_transcript"] == "scene two about a traffic stop."
assert fallback["scene_transcript"] == "scene one about a fire. scene two about a traffic stop."
@pytest.mark.asyncio
async def test_a_bare_number_never_becomes_an_incident_location_or_title():
inc = await _create(tags=["flames"], location="49", coords=None,