#!/bin/bash
# simple_surnames_validation.sh - Robust and simple

DBLP_FILE="${1:-dblp.xml.gz}"
TEMP_DIR="/tmp/surnames_val_$$"

echo "=== Simple Surname Validation ==="
echo "Input: $DBLP_FILE"

mkdir -p "$TEMP_DIR"

# Sequential execution to avoid pipe issues
echo "Running C program..."
gunzip -c "$DBLP_FILE" | ./surnames > "$TEMP_DIR/c_results.txt"
c_count=$(wc -l < "$TEMP_DIR/c_results.txt")
c_wang=$(head -1 "$TEMP_DIR/c_results.txt")

echo "Running Python validation..."
gunzip -c "$DBLP_FILE" | python3 surnames.py > "$TEMP_DIR/python_results.txt"
python_count=$(wc -l < "$TEMP_DIR/python_results.txt")
python_wang=$(head -1 "$TEMP_DIR/python_results.txt")

echo "Results:"
echo "  C program: $c_count surnames"
echo "  Python script: $python_count surnames"
echo "  C top result: $c_wang"
echo "  Python top result: $python_wang"

if diff -q "$TEMP_DIR/c_results.txt" "$TEMP_DIR/python_results.txt" > /dev/null; then
    echo "✓ Results are identical!"
else
    echo "⚠ Results differ - checking details..."
    diff "$TEMP_DIR/c_results.txt" "$TEMP_DIR/python_results.txt" | head -10
fi

echo "Memory check..."
echo "test" | valgrind --leak-check=yes ./surnames > /dev/null 2>&1
if [ $? -eq 0 ]; then
    echo "✓ No memory leaks detected"
fi

# Cleanup
rm -rf "$TEMP_DIR"
echo "Validation complete!"