#!/usr/bin/env bash
#
# Validate that compressed and decompressed archives produce the same results.
# Run this after run_all.sh to smoke test the outputs.
#

set -euo pipefail

SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PROJECT_DIR="$(dirname "$(dirname "$SCRIPT_DIR")")"
BINARY="$PROJECT_DIR/target/debug/json-archive"

echo "=== Validation ==="

errors=0

for format in gzip brotli zstd; do
    dir="$SCRIPT_DIR/out/$format"

    if [ ! -d "$dir" ]; then
        echo "SKIP: $format (no output directory)"
        continue
    fi

    # Find the compressed and uncompressed files
    compressed=$(find "$dir" -name "*.gz" -o -name "*.br" -o -name "*.zst" | head -1)
    uncompressed="$dir/test.json.archive"

    if [ ! -f "$compressed" ] || [ ! -f "$uncompressed" ]; then
        echo "SKIP: $format (missing files)"
        continue
    fi

    # Compare state output
    state_compressed=$("$BINARY" state "$compressed")
    state_uncompressed=$("$BINARY" state "$uncompressed")

    if [ "$state_compressed" = "$state_uncompressed" ]; then
        echo "OK: $format - state matches"
    else
        echo "FAIL: $format - state differs"
        errors=$((errors + 1))
    fi

    # Compare observation count from info
    count_compressed=$("$BINARY" info "$compressed" --output json | python3 -c "import sys,json; print(json.load(sys.stdin)['observation_count'])")
    count_uncompressed=$("$BINARY" info "$uncompressed" --output json | python3 -c "import sys,json; print(json.load(sys.stdin)['observation_count'])")

    if [ "$count_compressed" = "$count_uncompressed" ]; then
        echo "OK: $format - observation count matches ($count_compressed)"
    else
        echo "FAIL: $format - observation count differs ($count_compressed vs $count_uncompressed)"
        errors=$((errors + 1))
    fi
done

echo ""
if [ $errors -eq 0 ]; then
    echo "All validations passed."
else
    echo "$errors validation(s) failed."
    exit 1
fi