cdsl-cad/cadfs_to_cdsl/run_full_rebuild.sh

#!/usr/bin/env bash
# Run the complete CADFS corpus while retaining a snapshot of the prior run.
set -Eeuo pipefail

ROOT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")/.." && pwd)"
INPUT_DIR="${ROOT_DIR}/data/cadfs-sample/CADFS_test"
OUTPUT_DIR="${ROOT_DIR}/cadfs_to_cdsl/output"
HISTORY_DIR="${ROOT_DIR}/cadfs_to_cdsl/output-history"
WORKERS=4
TIMEOUT_SECONDS=60
COMPARE_MODE=rp
FORCE=true

usage() {
    cat <<'EOF'
Usage: ./cadfs_to_cdsl/run_full_rebuild.sh [options]

Runs conversion, CDSL rebuild, STEP comparison, and the Markdown report for
every sample in data/cadfs-sample/CADFS_test.

By default the existing cadfs_to_cdsl/output directory is moved to
cadfs_to_cdsl/output-history/<timestamp>/ before a forced full rerun.

Options:
  --resume                     Reuse cached results in output; do not archive or force rerun.
  --workers N                  Concurrent samples (1-8, default: 4).
  --timeout-seconds N          Per-model rebuild timeout in seconds (default: 60).
  --compare-mode rp|strict     Comparison acceptance mode (default: rp).
  -h, --help                   Show this help message.
EOF
}

while [[ $# -gt 0 ]]; do
    case "$1" in
        --resume)
            FORCE=false
            shift
            ;;
        --workers)
            [[ $# -ge 2 ]] || { echo "--workers requires a value" >&2; exit 2; }
            WORKERS="$2"
            shift 2
            ;;
        --timeout-seconds)
            [[ $# -ge 2 ]] || { echo "--timeout-seconds requires a value" >&2; exit 2; }
            TIMEOUT_SECONDS="$2"
            shift 2
            ;;
        --compare-mode)
            [[ $# -ge 2 ]] || { echo "--compare-mode requires a value" >&2; exit 2; }
            COMPARE_MODE="$2"
            shift 2
            ;;
        -h|--help)
            usage
            exit 0
            ;;
        *)
            echo "Unknown option: $1" >&2
            usage >&2
            exit 2
            ;;
    esac
done

[[ "$WORKERS" =~ ^[1-8]$ ]] || { echo "--workers must be an integer between 1 and 8" >&2; exit 2; }
[[ "$TIMEOUT_SECONDS" =~ ^[0-9]+([.][0-9]+)?$ ]] && (( $(awk "BEGIN { print (${TIMEOUT_SECONDS} > 0) }") )) || {
    echo "--timeout-seconds must be positive" >&2
    exit 2
}
[[ "$COMPARE_MODE" == "rp" || "$COMPARE_MODE" == "strict" ]] || {
    echo "--compare-mode must be rp or strict" >&2
    exit 2
}
[[ -d "$INPUT_DIR" ]] || { echo "CADFS input directory not found: $INPUT_DIR" >&2; exit 1; }
[[ -f "$INPUT_DIR/CADFS_text_test.jsonl" ]] || { echo "CADFS text manifest not found: $INPUT_DIR/CADFS_text_test.jsonl" >&2; exit 1; }

if [[ -x "${ROOT_DIR}/backend/.venv/bin/python" ]]; then
    PYTHON_BIN="${PYTHON_BIN:-${ROOT_DIR}/backend/.venv/bin/python}"
else
    PYTHON_BIN="${PYTHON_BIN:-python3}"
fi
command -v "$PYTHON_BIN" >/dev/null 2>&1 || { echo "Python executable not found: $PYTHON_BIN" >&2; exit 1; }

archive_output() {
    [[ -d "$OUTPUT_DIR" ]] || return
    [[ -n "$(find "$OUTPUT_DIR" -mindepth 1 -maxdepth 1 -print -quit)" ]] || return

    mkdir -p "$HISTORY_DIR"
    local timestamp archive_dir suffix=1
    timestamp="$(date +%Y%m%d-%H%M%S)"
    archive_dir="${HISTORY_DIR}/${timestamp}"
    while [[ -e "$archive_dir" ]]; do
        archive_dir="${HISTORY_DIR}/${timestamp}-${suffix}"
        ((suffix += 1))
    done
    mv "$OUTPUT_DIR" "$archive_dir"
    echo "Previous output archived at: $archive_dir"
}

on_error() {
    local code=$?
    echo "Run stopped with exit code ${code}. Partial artifacts remain in: $OUTPUT_DIR" >&2
    exit "$code"
}
trap on_error ERR

if [[ "$FORCE" == true ]]; then
    archive_output
    echo "Starting forced full corpus rebuild."
else
    echo "Resuming existing full corpus rebuild without archiving output."
fi
mkdir -p "$OUTPUT_DIR"

cd "$ROOT_DIR"
pipeline_args=(
    pipeline
    --input "$INPUT_DIR"
    --output "$OUTPUT_DIR"
    --workers "$WORKERS"
    --timeout-seconds "$TIMEOUT_SECONDS"
    --compare-mode "$COMPARE_MODE"
)
if [[ "$FORCE" == true ]]; then
    pipeline_args+=(--force)
fi

PYTHONPATH="${ROOT_DIR}/backend:${ROOT_DIR}${PYTHONPATH:+:${PYTHONPATH}}" \
    "$PYTHON_BIN" -m cadfs_to_cdsl "${pipeline_args[@]}"
PYTHONPATH="${ROOT_DIR}/backend:${ROOT_DIR}${PYTHONPATH:+:${PYTHONPATH}}" \
    "$PYTHON_BIN" -m cadfs_to_cdsl report --input "$INPUT_DIR" --output "$OUTPUT_DIR" --markdown

echo "Current results: $OUTPUT_DIR"
echo "Full report:     $OUTPUT_DIR/full_run_report.md"