Files
athena-oracle/oracle-pipeline.sh
T

48 lines
1.6 KiB
Bash
Executable File

#!/usr/bin/env bash
#
# Oracle daily pipeline — cron entry point (no_agent; stdout is delivered verbatim).
# Runs: fetch+store (pipeline.py) -> summarize (summarize.py) -> soft-cap archive (archive.py).
# run_log captures per-source failure + zero-fetch degradation for visibility.
#
set -u
ORACLE_DIR="/home/vpsadmin/oracle"
LOG_DIR="$ORACLE_DIR/logs"
TS="$(date -u +%Y%m%d-%H%M%S)"
LOG="$LOG_DIR/cron_run_${TS}.log"
mkdir -p "$LOG_DIR"
cd "$ORACLE_DIR" || { echo "FATAL: cannot cd $ORACLE_DIR"; exit 1; }
# Source tokens from .env (GITHUB_TOKEN, HUGGINGFACE_TOKEN)
if [ -f "$ORACLE_DIR/.env" ]; then
set -a
source "$ORACLE_DIR/.env"
set +a
fi
{
echo "=== Oracle pipeline run: $(date -u) ==="
python3 pipeline.py --limit 20
echo
echo "=== Summarization Engine ==="
python3 summarize.py
echo
echo "=== Phase 6 theme trend scan (new arrivals this cycle) ==="
python3 theme_scan.py
echo
echo "=== Soft-cap archive (dry-safe default: 30d / 5000 cap) ==="
python3 archive.py --days 30 --cap 5000
echo
echo "=== run_log tail (failure visibility) ==="
python3 -c "
import sqlite3
c = sqlite3.connect('oracle.db')
for r in c.execute('SELECT id,run_time,total_fetched,total_stored,sources_failed,notes FROM run_log ORDER BY id DESC LIMIT 1'):
print(' run', r[0], '|', r[1], '| fetched', r[2], '| stored', r[3], '| failed', r[4], '| notes:', r[5])
print(' live entries:', c.execute('SELECT COUNT(*) FROM entries').fetchone()[0])
print(' archived entries:', c.execute('SELECT COUNT(*) FROM entries_archive').fetchone()[0])
"
echo "=== Done ==="
} 2>&1 | tee "$LOG"