flyer-crawler.projectium.com/src/services/flyerDataTransformer.ts

// src/services/flyerDataTransformer.ts
import path from 'path';
import type { z } from 'zod';
import type { Logger } from 'pino';
import type { FlyerInsert, FlyerItemInsert } from '../types';
import type { AiProcessorResult } from './flyerAiProcessor.server'; // Keep this import for AiProcessorResult
import { AiFlyerDataSchema } from '../types/ai'; // Import consolidated schema
import { TransformationError } from './processingErrors';
import { parsePriceToCents } from '../utils/priceParser';
import { getBaseUrl } from '../utils/serverUtils';

/**
 * This class is responsible for transforming the validated data from the AI service
 * into the structured format required for database insertion (FlyerInsert and FlyerItemInsert).
 */
export class FlyerDataTransformer {
  /**
   * Normalizes a single raw item from the AI, providing default values for nullable fields.
   * @param item The raw item object from the AI.
   * @returns A normalized item object ready for database insertion.
   */
  private _normalizeItem(
    item: z.infer<typeof AiFlyerDataSchema>['items'][number],
  ): FlyerItemInsert {
    // If the AI fails to provide `price_in_cents` but provides a parsable `price_display`,
    // we can use our own parser as a fallback to improve data quality.
    const priceFromDisplay = parsePriceToCents(item.price_display ?? '');

    // Prioritize the AI's direct `price_in_cents` value, but use the parsed value if the former is null.
    const finalPriceInCents = item.price_in_cents ?? priceFromDisplay;

    return {
      ...item,
      // Use nullish coalescing and trim for robustness.
      // An empty or whitespace-only name falls back to 'Unknown Item'.
      item: (String(item.item ?? '')).trim() || 'Unknown Item',
      // Default null/undefined to an empty string and trim.
      price_display: (String(item.price_display ?? '')).trim(),
      quantity: (String(item.quantity ?? '')).trim(),
      // An empty or whitespace-only category falls back to 'Other/Miscellaneous'.
      category_name: (String(item.category_name ?? '')).trim() || 'Other/Miscellaneous',
      // Overwrite price_in_cents with our calculated value.
      price_in_cents: finalPriceInCents,
      // Use nullish coalescing to convert null to undefined for the database.
      master_item_id: item.master_item_id ?? undefined,
      view_count: 0,
      click_count: 0,
    };
  }

  /**
   * Constructs the full public URLs for the flyer image and its icon.
   * @param imageFileName The filename of the main processed image.
   * @param iconFileName The filename of the generated icon.
   * @param baseUrl The base URL from the job payload.
   * @param logger The logger instance.
   * @returns An object containing the full image_url and icon_url.
   */
  private _buildUrls(
    imageFileName: string,
    iconFileName: string,
    baseUrl: string | undefined,
    logger: Logger,
  ): { imageUrl: string; iconUrl: string } {
    logger.debug({ imageFileName, iconFileName, baseUrl }, 'Building URLs');
    const finalBaseUrl = baseUrl || getBaseUrl(logger);
    const imageUrl = `${finalBaseUrl}/flyer-images/${imageFileName}`;
    const iconUrl = `${finalBaseUrl}/flyer-images/icons/${iconFileName}`;
    logger.debug({ imageUrl, iconUrl }, 'Constructed URLs');
    return { imageUrl, iconUrl };
  }

  /**
   * Transforms AI-extracted data into database-ready flyer and item records.
   * @param extractedData The validated data from the AI.
   * @param imagePaths The paths to the flyer images.
   * @param originalFileName The original name of the uploaded file.
   * @param checksum The checksum of the file.
   * @param userId The ID of the user who uploaded the file, if any.
   * @param logger The request-scoped or job-scoped logger instance.
   * @returns A promise that resolves to an object containing the prepared flyer and item data.
   */
  async transform(
    aiResult: AiProcessorResult,
    originalFileName: string,
    imageFileName: string,
    iconFileName: string,
    checksum: string,
    userId: string | undefined,
    logger: Logger,
    baseUrl?: string,
  ): Promise<{ flyerData: FlyerInsert; itemsForDb: FlyerItemInsert[] }> {
    logger.info('Starting data transformation from AI output to database format.');

    try {
      const { data: extractedData, needsReview } = aiResult;

      const { imageUrl, iconUrl } = this._buildUrls(imageFileName, iconFileName, baseUrl, logger);

      const itemsForDb: FlyerItemInsert[] = extractedData.items.map((item) => this._normalizeItem(item));

      const storeName = extractedData.store_name || 'Unknown Store (auto)';
      if (!extractedData.store_name) {
        logger.warn('AI did not return a store name. Using fallback "Unknown Store (auto)".');
      }

      const flyerData: FlyerInsert = {
        file_name: originalFileName,
        image_url: imageUrl,
        icon_url: iconUrl,
        checksum,
        store_name: storeName,
        valid_from: extractedData.valid_from,
        valid_to: extractedData.valid_to,
        store_address: extractedData.store_address,
        item_count: itemsForDb.length,
        uploaded_by: userId ? userId : null,
        status: needsReview ? 'needs_review' : 'processed',
      };

      logger.info(
        { itemCount: itemsForDb.length, storeName: flyerData.store_name },
        'Data transformation complete.',
      );

      return { flyerData, itemsForDb };
    } catch (err) {
      logger.error({ err }, 'Transformation process failed');
      // Wrap and rethrow with the new error class
      throw new TransformationError('Flyer Data Transformation Failed');
    }
  }
}