markbase/tests/archive_perf_test.sh

#!/bin/bash

# MarkBase Archive Module Performance Test Script

echo "========================================="
echo "MarkBase Archive Performance Test Suite"
echo "========================================="

# Configuration
SERVER_URL="http://localhost:11438"
UPLOAD_ENDPOINT="/api/v2/upload"
USER_ID="perf_test"
TEST_DIR="/tmp/markbase_perf_test"
LOG_FILE="/tmp/markbase_perf_results.log"

# Create test directory
mkdir -p $TEST_DIR

# Function to generate test ZIP files
generate_test_zip() {
    size=$1
    name=$2
    
    echo "Generating $size MB test ZIP file..."
    
    # Create temp directory with files
    temp_dir=$(mktemp -d)
    for i in {1..10}; do
        dd if=/dev/urandom of="$temp_dir/file_$i.txt" bs=1M count=$((size / 10)) 2>/dev/null
    done
    
    # Create ZIP
    cd $temp_dir
    zip -r "$TEST_DIR/$name" .  >/dev/null 2>&1
    cd -
    rm -rf $temp_dir
    
    echo "Generated $TEST_DIR/$name ($(du -h $TEST_DIR/$name | cut -f1))"
}

# Function to test upload performance
test_upload() {
    file=$1
    description=$2
    
    echo ""
    echo "Testing: $description"
    echo "File: $file"
    
    # Measure upload time
    start_time=$(date +%s%N)
    
    response=$(curl -s -w "\n%{http_code}\n%{time_total}" \
        -X POST "$SERVER_URL$UPLOAD_ENDPOINT/$USER_ID" \
        -F "file=@$file")
    
    end_time=$(date +%s%N)
    
    # Parse response
    http_code=$(echo "$response" | tail -n 2 | head -n 1)
    time_total=$(echo "$response" | tail -n 1)
    json_response=$(echo "$response" | head -n -2)
    
    # Calculate duration in milliseconds
    duration_ms=$(( (end_time - start_time) / 1000000 ))
    
    # Log results
    echo "HTTP Code: $http_code"
    echo "Upload Time: ${time_total}s"
    echo "Duration: ${duration_ms}ms"
    
    if [ "$http_code" == "201" ]; then
        echo "✅ Upload successful"
        
        # Check if extracted field exists
        if echo "$json_response" | jq -e '.extracted' >/dev/null 2>&1; then
            extracted_count=$(echo "$json_response" | jq -r '.extracted.count')
            extracted_bytes=$(echo "$json_response" | jq -r '.extracted.bytes')
            echo "✅ Extracted: $extracted_count files, $extracted_bytes bytes"
            
            echo "SUCCESS|$description|$http_code|${time_total}s|${duration_ms}ms|$extracted_count|$extracted_bytes" >> $LOG_FILE
        else
            echo "⚠️  No extracted field in response"
            echo "PARTIAL|$description|$http_code|${time_total}s|${duration_ms}ms" >> $LOG_FILE
        fi
    else
        echo "❌ Upload failed"
        echo "FAILED|$description|$http_code|${time_total}s|${duration_ms}ms" >> $LOG_FILE
    fi
}

# Function to test concurrent uploads
test_concurrent() {
    count=$1
    
    echo ""
    echo "Testing concurrent uploads: $count simultaneous"
    
    # Generate small test files
    for i in {1..$count}; do
        temp_dir=$(mktemp -d)
        echo "file $i content" > "$temp_dir/test_$i.txt"
        cd $temp_dir
        zip -r "$TEST_DIR/concurrent_$i.zip" test_$i.txt >/dev/null 2>&1
        cd -
        rm -rf $temp_dir
    done
    
    # Run concurrent uploads
    start_time=$(date +%s%N)
    
    for i in {1..$count}; do
        curl -s -X POST "$SERVER_URL$UPLOAD_ENDPOINT/concurrent_test" \
            -F "file=@$TEST_DIR/concurrent_$i.zip" &
    done
    
    # Wait for all to complete
    wait
    
    end_time=$(date +%s%N)
    duration_ms=$(( (end_time - start_time) / 1000000 ))
    
    echo "Total time for $count concurrent uploads: ${duration_ms}ms"
    echo "CONCURRENT|$count uploads|${duration_ms}ms" >> $LOG_FILE
    
    # Cleanup
    rm -f $TEST_DIR/concurrent_*.zip
}

# Main test execution
echo ""
echo "Starting performance tests..."
echo "Log file: $LOG_FILE"
echo ""

# Test 1: Small file (10MB)
generate_test_zip 10 "test_10mb.zip"
test_upload "$TEST_DIR/test_10mb.zip" "10MB ZIP file"

# Test 2: Medium file (100MB)  
generate_test_zip 100 "test_100mb.zip"
test_upload "$TEST_DIR/test_100mb.zip" "100MB ZIP file"

# Test 3: Large file (1GB) - Optional, comment out if needed
# generate_test_zip 1000 "test_1gb.zip"
# test_upload "$TEST_DIR/test_1gb.zip" "1GB ZIP file"

# Test 4: Concurrent uploads (10)
test_concurrent 10

# Test 5: Concurrent uploads (50)
test_concurrent 50

echo ""
echo "========================================="
echo "Performance tests completed"
echo "Results saved to: $LOG_FILE"
echo "========================================="
echo ""
echo "Summary:"
cat $LOG_FILE

# Cleanup
rm -rf $TEST_DIR