watch-finished-turbo/scripts/example-duplicate-detection.js

#!/usr/bin/env node

/**
 * Example script demonstrating the new duplicate detection optimization
 *
 * This shows how to:
 * 1. Index destination files for fast duplicate detection
 * 2. Query duplicate statistics
 * 3. Run duplicate scans using the database
 */

const API_BASE = process.env.API_BASE || "http://localhost:3000";

async function indexDestination(dataset, destination) {
  console.log(`\n📁 Indexing ${dataset} destination: ${destination}`);

  const response = await fetch(`${API_BASE}/maintenance/index/destination`, {
    method: "POST",
    headers: { "Content-Type": "application/json" },
    body: JSON.stringify({
      dataset,
      destination,
      reindex: false, // Set to true to rebuild index
      batchSize: 100,
    }),
  });

  const result = await response.json();
  console.log(
    `✅ Indexed: ${result.indexed}, Skipped: ${result.skipped}, Errors: ${result.errors}`
  );
  return result;
}

async function getIndexCount(dataset, destination) {
  console.log(`\n📊 Getting index count for ${dataset}`);

  const params = new URLSearchParams({ dataset });
  if (destination) params.append("destination", destination);

  const response = await fetch(`${API_BASE}/maintenance/index/count?${params}`);
  const result = await response.json();

  console.log(`📈 Indexed files: ${result.count}`);
  return result;
}

async function getDuplicateStats(dataset) {
  console.log(
    `\n🔍 Getting duplicate statistics for ${dataset || "all datasets"}`
  );

  const params = dataset ? `?dataset=${dataset}` : "";
  const response = await fetch(`${API_BASE}/maintenance/index/stats${params}`);
  const result = await response.json();

  console.log(`🔄 Total duplicate groups: ${result.totalDuplicates}`);

  if (result.duplicatesByDataset.length > 0) {
    console.log("\nDuplicate groups:");
    result.duplicatesByDataset.slice(0, 5).forEach((dup, idx) => {
      console.log(`\n  Group ${idx + 1}:`);
      console.log(`    Hash: ${dup.hash.substring(0, 16)}...`);
      console.log(`    Size: ${(dup.file_size / 1024 / 1024).toFixed(2)} MB`);
      console.log(`    Count: ${dup.file_count} files`);
      console.log(`    Files:`);
      dup.files.forEach((file) => {
        console.log(`      - ${file}`);
      });
    });

    if (result.duplicatesByDataset.length > 5) {
      console.log(
        `\n  ... and ${result.duplicatesByDataset.length - 5} more groups`
      );
    }
  }

  return result;
}

async function scanDuplicates(resetExisting = false) {
  console.log(`\n🔎 Scanning for duplicates (reset: ${resetExisting})`);

  const response = await fetch(`${API_BASE}/maintenance/duplicates/scan`, {
    method: "POST",
    headers: { "Content-Type": "application/json" },
    body: JSON.stringify({ resetExisting }),
  });

  const result = await response.json();
  console.log("✅ Duplicate scan completed");
  return result;
}

async function clearIndex(dataset, destination) {
  console.log(`\n🗑️  Clearing index for ${dataset}`);

  const params = destination ? `?destination=${destination}` : "";
  const response = await fetch(
    `${API_BASE}/maintenance/index/${dataset}${params}`,
    {
      method: "DELETE",
    }
  );

  const result = await response.json();
  console.log(`🗑️  Cleared ${result.cleared} entries`);
  return result;
}

async function reindexDestination(dataset, destination) {
  console.log(`\n🔄 Re-indexing ${dataset} destination: ${destination}`);

  const response = await fetch(`${API_BASE}/maintenance/index/destination`, {
    method: "POST",
    headers: { "Content-Type": "application/json" },
    body: JSON.stringify({
      dataset,
      destination,
      reindex: true, // Force rebuild
      batchSize: 100,
    }),
  });

  const result = await response.json();
  console.log(
    `✅ Re-indexed: ${result.indexed}, Skipped: ${result.skipped}, Errors: ${result.errors}`
  );
  return result;
}

// Example workflow
async function exampleWorkflow() {
  console.log("🚀 Duplicate Detection Optimization - Example Workflow\n");
  console.log(`Using API: ${API_BASE}\n`);

  try {
    // Example 1: Index a destination directory
    console.log("═".repeat(60));
    console.log("Example 1: Index destination files");
    console.log("═".repeat(60));

    // Uncomment and modify these lines with your actual paths:
    // await indexDestination('movies', '/path/to/movies/destination');
    // await indexDestination('tvshows', '/path/to/tvshows/destination');

    console.log(
      "\nℹ️  Uncomment the indexDestination calls in the script to run this example"
    );

    // Example 2: Check index count
    console.log("\n" + "═".repeat(60));
    console.log("Example 2: Check index count");
    console.log("═".repeat(60));

    // await getIndexCount('movies');
    console.log(
      "\nℹ️  Uncomment the getIndexCount call in the script to run this example"
    );

    // Example 3: Get duplicate statistics
    console.log("\n" + "═".repeat(60));
    console.log("Example 3: Get duplicate statistics");
    console.log("═".repeat(60));

    // await getDuplicateStats('movies');
    console.log(
      "\nℹ️  Uncomment the getDuplicateStats call in the script to run this example"
    );

    // Example 4: Run duplicate scan (uses database)
    console.log("\n" + "═".repeat(60));
    console.log("Example 4: Run duplicate scan");
    console.log("═".repeat(60));

    // await scanDuplicates(false);
    console.log(
      "\nℹ️  Uncomment the scanDuplicates call in the script to run this example"
    );

    // Example 5: Re-index (clear and rebuild)
    console.log("\n" + "═".repeat(60));
    console.log("Example 5: Re-index destination");
    console.log("═".repeat(60));

    // await clearIndex('movies');
    // await indexDestination('movies', '/path/to/movies/destination');
    console.log(
      "\nℹ️  Uncomment the clearIndex and indexDestination calls in the script to run this example"
    );

    console.log("\n" + "═".repeat(60));
    console.log("✨ Workflow complete!");
    console.log("═".repeat(60));
  } catch (error) {
    console.error("\n❌ Error:", error.message);
    if (error.cause) {
      console.error("Cause:", error.cause);
    }
  }
}

// Run the workflow
exampleWorkflow().catch(console.error);