fix(scripts): kevin-analyze + reanalyze — self-contained + append-only
All checks were successful
ci/woodpecker/push/woodpecker Pipeline was successful

Two small follow-ups from the session:

- kevin-analyze.sh: always cp the local analyze_kevin_video.py into
  the pod before exec, so the wrapper works even before CI ships an
  image with scripts/ baked in (saw this hit the user with
  ModuleNotFoundError on stale images).

- reanalyze_kevin_videos.py: switch from DELETE+INSERT to
  append-only. The old behaviour violated the FK constraint on
  kevin_signal_bridge_state.mention_id; the new behaviour adds a
  fresh v2 analysis alongside the v1 row and lets the API surface
  the latest. Old v1 rows preserved for audit.
This commit is contained in:
Viktor Barzin 2026-05-28 22:55:14 +00:00
parent b82014995c
commit b4e1c5cd12
2 changed files with 23 additions and 12 deletions

View file

@ -5,8 +5,9 @@
# ./scripts/kevin-analyze.sh <video-id-or-url>
#
# Picks the running meet-kevin-watcher container (which already has
# yt-dlp + ffmpeg + the Anthropic token + the right Python env) and
# runs scripts/analyze_kevin_video.py inside it.
# yt-dlp + ffmpeg + the Anthropic token + the right Python env), copies
# the local script in, and runs it. Works regardless of whether the
# image has scripts/ baked in yet.
set -euo pipefail
@ -15,6 +16,14 @@ if [[ $# -ne 1 ]]; then
exit 1
fi
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
LOCAL_SCRIPT="$SCRIPT_DIR/analyze_kevin_video.py"
if [[ ! -f "$LOCAL_SCRIPT" ]]; then
echo "missing $LOCAL_SCRIPT" >&2
exit 1
fi
POD=$(kubectl -n trading-bot get pod -l app=trading-bot-workers \
-o jsonpath='{.items[0].metadata.name}')
@ -23,5 +32,10 @@ if [[ -z "$POD" ]]; then
exit 1
fi
# Copy the latest local script into the pod, then exec it. This way the
# wrapper works even before CI ships an image with scripts/ baked in.
kubectl -n trading-bot cp "$LOCAL_SCRIPT" \
"$POD:/tmp/kevin-analyze.py" -c meet-kevin-watcher >/dev/null
exec kubectl -n trading-bot exec "$POD" -c meet-kevin-watcher -- \
python -m scripts.analyze_kevin_video "$1"
python /tmp/kevin-analyze.py "$1"

View file

@ -29,7 +29,7 @@ from anthropic import AsyncAnthropic
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
from sqlalchemy import delete, select
from sqlalchemy import select
from sqlalchemy.ext.asyncio import AsyncSession
from services.meet_kevin_watcher.caption_extractor import extract_captions
@ -86,14 +86,11 @@ async def _reanalyze_one(
if dry_run:
return True, result.cost_usd, "dry_run_skipped_write"
# Replace old analysis + mentions for this video atomically.
await session.execute(
delete(KevinStockMention).where(KevinStockMention.video_id == video.id)
)
await session.execute(
delete(KevinAnalysis).where(KevinAnalysis.video_id == video.id)
)
# Append-only: insert new analysis + mentions for this video. We do
# NOT delete the old v1 rows because kevin_signal_bridge_state has
# FKs into kevin_stock_mentions (would cascade-break the audit
# trail). The API surfaces the newest analysis per video and the
# bridge picks up the new mentions by ID > cursor.
db_analysis = KevinAnalysis(
video_id=video.id,
model=analyzer._model,