#!/usr/bin/env bash
# Full pipeline: extract frames + OCR → heuristic fixes → sew + dedupe → result.csv
# Usage: ./run.sh [video] [output_dir]
# Default: video=video-data.webm, output_dir=frames, result=result.csv
set -e
VIDEO="${1:-video-data.webm}"
OUT_DIR="${2:-frames}"
RESULT="result.csv"

echo "1. Extract frames (every 10) + OCR with template.svg → ${OUT_DIR}/"
python3 extract_frames_and_tables.py "$OUT_DIR" "$VIDEO" template.svg 10

echo "2. Heuristic fixes on ${OUT_DIR}/*.csv"
python3 fix_all_csvs.py "$OUT_DIR"

echo "3. Sew CSVs + dedupe → ${RESULT}"
python3 sew_csvs.py "$OUT_DIR" "$RESULT"

echo "Done. Output: ${RESULT}"