import { chunk } from 'lodash';
import { addJob, JobKind } from '@spokenio/jobqueue-producer';
import { subHours } from 'date-fns';
import express, { Express, Request } from 'express';
import { getRepository, In, IsNull, LessThan } from 'typeorm';
import logger from '@/logger';
import { batchProcessEntityStream } from '@/entities/utils';
import { Page } from '@/entities';

const scrapableRetailers = [
  'urbanoutfitters',
  'homedepot',
  'houzz',
  'amazon',
  'franceandson',
  'luluandgeorgia',
  'pier1',
  'burkedecor',
  'bedbathandbeyond',
  'walmart',
  'dotandbo',
  'onekingslane',
  'amara',
  'worldmarket',
  'kathykuohome',
  'afastores',
  'crateandbarrel',
  'etsy',
  'wayfair',
  'allmodern',
  'jossandmain',
  'birchlane',
  'perigold',
];

const setup = (): Express => {
  const app = express();
  app.put('/update', async (req: Request<{ force?: string }>, res) => {
    const forceQuery = req.query.force;
    if (forceQuery && typeof forceQuery !== 'string') {
      return res
        .status(400)
        .send("Query param 'force' has to be of type string");
    }
    const forceAsString = forceQuery?.toLowerCase().trim();
    const force =
      forceAsString && forceAsString !== 'false' && forceAsString !== '0';

    batchProcessEntityStream(
      Page,
      async (pages) => {
        logger.info(
          `Queuing bulk scrape latest page for ${pages.length} pages.`
        );
        const urls = pages
          .map(
            ({
              urlOriginal: url,
              id: pageId,
              productFirebaseId: productId,
            }) => ({ url, pageId, productId })
          )
          .filter(
            (
              page
            ): page is { url: string; pageId: string; productId: string } => {
              return !!(page.url && page.pageId && page.productId);
            }
          );
        await Promise.all(
          chunk(urls, 10).map(async (urlsChunk) => {
            await addJob(JobKind.BULK_FETCH_LATEST_PAGE_1, {
              pages: urlsChunk,
            });
          })
        );
        await getRepository(Page).update(
          { id: In(pages.map(({ id }) => id)) },
          {
            lastScrapedAt: new Date(),
          }
        );
      },
      {
        where: force
          ? {
              retailer: In(scrapableRetailers),
            }
          : [
              {
                // @TODO: we should probably run this once per hour and check
                // last 23 hours.
                // Instead today, we run this once per day.
                lastScrapedAt: LessThan(subHours(new Date(), 23)),
                retailer: In(scrapableRetailers),
              },
              {
                lastScrapedAt: IsNull(),
                retailer: In(scrapableRetailers),
              },
            ],
      }
    );
    return res.status(200).json({
      status: 'queued',
    });
  });
  return app;
};
export default setup;