#!/bin/bash
# Test script to verify environment variable sanitization
# This tests that invisible Unicode characters are properly removed

set -e

# Define Unicode character constants for testing
readonly U200E_HEX=$'\xE2\x80\x8E'  # U+200E Left-to-Right Mark
readonly U200B_HEX=$'\xE2\x80\x8B'  # U+200B Zero Width Space

echo "Testing environment variable sanitization..."
echo ""

# Create a test file with problematic Unicode characters
TEST_FILE=$(mktemp /tmp/test_env_with_unicode.XXXXXX.sh)

# Create test environment with U+200E (Left-to-Right Mark) after variable names
cat > "$TEST_FILE" << 'EOF'
# Test environment variables with invisible Unicode characters
export ADMIN_USER‎="testuser"
export ADMIN_PASSWORD‎="testpass123"
export OPENROUTER_API_KEY‎="sk-test-key-12345"
export NORMAL_VAR="normalvalue"
EOF

# Add actual invisible U+200E character (E2 80 8E in UTF-8) to the file
# This simulates what happens when users copy-paste from web browsers
printf "export TEST_VAR_WITH_U200E%s=\"value_with_unicode\"\n" "$U200E_HEX" >> "$TEST_FILE"
printf "export TEST_VAR_WITH_U200B%s=\"value_with_zwsp\"\n" "$U200B_HEX" >> "$TEST_FILE"

echo "Original test file (with invisible characters):"
hexdump -C "$TEST_FILE" | grep -E "e2 80" || echo "  (invisible characters present but not shown)"
echo ""

# Test 1: Verify the original file has Unicode characters
echo "Test 1: Checking for invisible Unicode characters in original file..."
if grep -q "$U200E_HEX" "$TEST_FILE" 2>/dev/null; then
    echo "  ✓ U+200E (Left-to-Right Mark) detected"
else
    echo "  ✗ U+200E not found (expected to find it)"
fi

if grep -q "$U200B_HEX" "$TEST_FILE" 2>/dev/null; then
    echo "  ✓ U+200B (Zero Width Space) detected"
else
    echo "  ✗ U+200B not found (expected to find it)"
fi
echo ""

# Test 2: Apply the same sanitization logic from entrypoint.sh
echo "Test 2: Applying sanitization..."
SANITIZED_FILE=$(mktemp /tmp/test_env_sanitized.XXXXXX.sh)
cp "$TEST_FILE" "$SANITIZED_FILE"

# Remove common invisible Unicode characters (same logic as entrypoint.sh)
sed -i \
    -e 's/\xE2\x80\x8E//g' \
    -e 's/\xE2\x80\x8F//g' \
    -e 's/\xE2\x80\x8B//g' \
    -e 's/\xEF\xBB\xBF//g' \
    -e 's/\xE2\x80\xAA//g' \
    -e 's/\xE2\x80\xAB//g' \
    -e 's/\xE2\x80\xAC//g' \
    -e 's/\xE2\x80\xAD//g' \
    -e 's/\xE2\x80\xAE//g' \
    "$SANITIZED_FILE" 2>/dev/null

echo "  Sanitization complete"
echo ""

# Test 3: Verify Unicode characters are removed
echo "Test 3: Verifying invisible characters are removed..."
if grep -q "$U200E_HEX" "$SANITIZED_FILE" 2>/dev/null; then
    echo "  ✗ FAILED: U+200E still present after sanitization"
    exit 1
else
    echo "  ✓ U+200E successfully removed"
fi

if grep -q "$U200B_HEX" "$SANITIZED_FILE" 2>/dev/null; then
    echo "  ✗ FAILED: U+200B still present after sanitization"
    exit 1
else
    echo "  ✓ U+200B successfully removed"
fi
echo ""

# Test 4: Verify the sanitized file is valid bash and can be sourced
echo "Test 4: Testing if sanitized file is valid bash..."
if bash -n "$SANITIZED_FILE" 2>/dev/null; then
    echo "  ✓ Sanitized file has valid bash syntax"
else
    echo "  ✗ FAILED: Sanitized file has syntax errors"
    cat "$SANITIZED_FILE"
    exit 1
fi
echo ""

# Test 5: Try sourcing the sanitized environment
echo "Test 5: Testing if sanitized environment can be sourced..."
(
    source "$SANITIZED_FILE"
    if [ "$ADMIN_USER" = "testuser" ] && [ "$ADMIN_PASSWORD" = "testpass123" ]; then
        echo "  ✓ Environment variables loaded correctly"
    else
        echo "  ✗ FAILED: Environment variables not loaded correctly"
        echo "    ADMIN_USER=$ADMIN_USER (expected: testuser)"
        echo "    ADMIN_PASSWORD=$ADMIN_PASSWORD (expected: testpass123)"
        exit 1
    fi
)
echo ""

# Test 6: Compare file sizes (sanitized should be smaller)
ORIGINAL_SIZE=$(wc -c < "$TEST_FILE")
SANITIZED_SIZE=$(wc -c < "$SANITIZED_FILE")
REMOVED_BYTES=$((ORIGINAL_SIZE - SANITIZED_SIZE))

echo "Test 6: Verifying bytes were removed..."
if [ $REMOVED_BYTES -gt 0 ]; then
    echo "  ✓ Removed $REMOVED_BYTES bytes of invisible Unicode characters"
else
    echo "  ✗ WARNING: No bytes removed (original: $ORIGINAL_SIZE, sanitized: $SANITIZED_SIZE)"
fi
echo ""

# Cleanup
rm -f "$TEST_FILE" "$SANITIZED_FILE"

echo "=================================="
echo "All sanitization tests PASSED! ✓"
echo "=================================="
echo ""
echo "The entrypoint.sh sanitization logic will prevent the Portainer U+200E error."