Correct the transcript before anything reads it

Correction existed, but as a line in intelligence.py's EXTRACTION_PROMPT --
which put it in the wrong place twice over. The same model call that extracted
units, location and severity emitted the correction afterwards, so extraction
reasoned over text already known to be wrong; and it sat behind
correlation_enabled, so during a cost-controlled STT-only window nothing was
ever corrected at all. That is the normal state during development.

internal/transcript_correction.py is now its own pass, between the degenerate
filter and the Firestore write. It receives an already-produced transcript plus
a reference list, so unlike a Whisper prompt it has no series to extend -- the
distinction that keeps vocabulary out of the recogniser's prompt, where an
enumerated ten-code list once made it hallucinate ten-code runs.

Reference data is merged from the talkgroup and the system, TALKGROUP FIRST. A
system spanning several counties can have a talkgroup covering one
municipality, and that municipality's streets must not be buried under a
county-wide list. A single-municipality system is the degenerate case: populate
the system level and every talkgroup inherits it. Area context is now SET --
municipality, county, roads, landmarks, on both scopes -- rather than guessed
from talkgroup names, which is what vocabulary_learner did and which is close
to useless across multiple counties.

Segments are corrected too, not just the joined text. extract_scenes builds its
prompt from numbered segments whenever there is more than one, so a correction
that only fixed the transcript would have been discarded on exactly the
multi-transmission calls carrying the most content. Alignment is enforced: an
array of the wrong length or type is dropped whole, because scenes map back to
transmissions by index and a shifted array would misattribute audio silently.

Whisper is also retried once on degenerate output. Call e49ea32c produced a
56-word ten-code counting run on one attempt and ordinary speech on the next --
same clip, same temperature=0 -- so a hallucination is a coin-flip, and
discarding on the first bad roll threw away a recoverable transcript.

Two things found on the way:

PUT /systems/{id} wiped ten_codes on every save. The systems form sends only
{name, type, config}, and model_dump() wrote every omitted field as its default
over the top. Now exclude_unset. area_context would have been the next victim,
which is why it gets its own route alongside ten-codes rather than a field on
that payload.

Closes server-26#36.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

drb-c2-core/app/routers/systems.py

@@ -1,7 +1,7 @@
 import uuid
 from fastapi import APIRouter, HTTPException, Depends, Query
 from pydantic import BaseModel
-from typing import Dict, Optional
+from typing import Dict, List, Optional
 from app.models import SystemCreate, SystemRecord
 from app.internal import firestore as fstore
 from app.internal.auth import (
@@ -23,6 +23,14 @@ class TenCodesBody(BaseModel):
     ten_codes: Dict[str, str]
 
 
+class AreaContextBody(BaseModel):
+    """Ground truth about the area a system covers — see PUT /{id}/area-context."""
+    municipality: Optional[str] = None
+    county: Optional[str] = None
+    roads: List[str] = []
+    landmarks: List[str] = []
+
+
 class AiFlagsBody(BaseModel):
     stt_enabled: Optional[bool] = None
     correlation_enabled: Optional[bool] = None
@@ -64,8 +72,14 @@ async def update_system(system_id: str, body: SystemCreate, _: dict = Depends(re
     existing = await fstore.doc_get("systems", system_id)
     if not existing:
         raise HTTPException(404, f"System '{system_id}' not found.")
-    await fstore.doc_update("systems", system_id, body.model_dump())
-    return {**existing, **body.model_dump()}
+    # exclude_unset, or every field the caller omitted gets written as its
+    # default and silently erases what was there. The systems page PUTs only
+    # {name, type, config}, so a plain model_dump() wiped ten_codes on every
+    # save — they are edited through PUT /{id}/ten-codes and were never in this
+    # payload. area_context (server-26#36) would have been the second casualty.
+    patch = body.model_dump(exclude_unset=True)
+    await fstore.doc_update("systems", system_id, patch)
+    return {**existing, **patch}
 
 
 @router.delete("/{system_id}", status_code=204)
@@ -129,6 +143,42 @@ async def update_ten_codes(
     return {"ok": True, "ten_codes": body.ten_codes}
 
 
+# ── Area context ──────────────────────────────────────────────────────────────
+
+@router.get("/{system_id}/area-context")
+async def get_area_context(system_id: str, _: dict = Depends(require_admin_token)):
+    system = await fstore.doc_get("systems", system_id)
+    if not system:
+        raise HTTPException(404, f"System '{system_id}' not found.")
+    return {"area_context": system.get("area_context") or {}}
+
+
+@router.put("/{system_id}/area-context")
+async def update_area_context(
+    system_id: str,
+    body: AreaContextBody,
+    _: dict = Depends(require_admin_token),
+):
+    """
+    Replace the system-wide area context used by the transcript corrector.
+
+    Ground truth about where this system operates — municipality, county, the
+    roads and landmarks whose names Whisper mangles. Per-talkgroup overrides
+    live inside config.talkgroups[] and rank ABOVE this (server-26#36), so a
+    multi-county system narrows per channel rather than replacing this wholesale.
+
+    Its own route rather than a field on PUT /systems/{id} for the same reason
+    ten-codes has one: the systems form does not carry it, and folding it into
+    that payload is how ten_codes kept getting wiped.
+    """
+    existing = await fstore.doc_get("systems", system_id)
+    if not existing:
+        raise HTTPException(404, f"System '{system_id}' not found.")
+    area = body.model_dump(exclude_none=True)
+    await fstore.doc_update("systems", system_id, {"area_context": area})
+    return {"ok": True, "area_context": area}
+
+
 # ── Vocabulary endpoints ───────────────────────────────────────────────────────
 
 @router.get("/{system_id}/vocabulary")