website-enchun-mgr/apps/backend/scripts/migration/deduplicator.ts

/**
 * Deduplication Module
 * Story 1.3: Content Migration Script
 *
 * Checks for existing items to prevent duplicates
 */

import type { Payload } from 'payload'

export interface DuplicateCheckOptions {
  force?: boolean // Skip deduplication check
}

// ============================================================
// FIND EXISTING BY SLUG
// ============================================================

/**
 * Check if a document exists by slug
 */
export async function findBySlug(
  payload: Payload,
  collection: string,
  slug: string,
): Promise<{ exists: boolean; id?: string }> {
  try {
    const result = await payload.find({
      collection,
      where: {
        slug: { equals: slug },
      },
      limit: 1,
      depth: 0,
    })

    if (result.docs && result.docs.length > 0) {
      return { exists: true, id: result.docs[0].id }
    }

    return { exists: false }
  } catch (error) {
    console.error(`Error checking for duplicate ${collection} with slug "${slug}":`, error)
    return { exists: false }
  }
}

/**
 * Check if post exists by slug and published date
 */
export async function findBySlugAndDate(
  payload: Payload,
  slug: string,
  publishedAt: Date,
): Promise<{ exists: boolean; id?: string }> {
  try {
    const result = await payload.find({
      collection: 'posts',
      where: {
        and: [
          {
            slug: { equals: slug },
          },
          {
            publishedAt: { equals: publishedAt },
          },
        ],
      },
      limit: 1,
      depth: 0,
    })

    if (result.docs && result.docs.length > 0) {
      return { exists: true, id: result.docs[0].id }
    }

    return { exists: false }
  } catch (error) {
    console.error(`Error checking for duplicate post with slug "${slug}":`, error)
    return { exists: false }
  }
}

// ============================================================
// BULK EXISTENCE CHECK
// ============================================================

/**
 * Get all existing slugs for a collection
 */
export async function getAllSlugs(payload: Payload, collection: string): Promise<Set<string>> {
  try {
    const result = await payload.find({
      collection,
      limit: 1000, // Adjust based on expected data size
      depth: 0,
      select: { slug: true },
    })

    const slugs = new Set<string>()
    if (result.docs) {
      for (const doc of result.docs) {
        if ('slug' in doc && typeof doc.slug === 'string') {
          slugs.add(doc.slug)
        }
      }
    }

    return slugs
  } catch (error) {
    console.error(`Error getting existing slugs for ${collection}:`, error)
    return new Set()
  }
}

/**
 * Get existing posts by slug + date combination
 */
export async function getExistingPostIdentifiers(
  payload: Payload,
): Promise<Map<string, Date>> {
  try {
    const result = await payload.find({
      collection: 'posts',
      limit: 1000,
      depth: 0,
      select: { slug: true, publishedAt: true },
    })

    const identifiers = new Map<string, Date>()
    if (result.docs) {
      for (const doc of result.docs) {
        if ('slug' in doc && 'publishedAt' in doc) {
          const key = `${doc.slug}-${doc.publishedAt}`
          identifiers.set(key, doc.publishedAt as Date)
        }
      }
    }

    return identifiers
  } catch (error) {
    console.error('Error getting existing post identifiers:', error)
    return new Map()
  }
}