`README.md`

26 lines
# dw-vendor-scrapers-local

Local working copy of select Kamatera DW vendor scrapers
(`/root/DW-Agents/vendor-scrapers/`) for editing + version control on Mac2.

- `scrape-coordonne.js`, `scrape-hollywood.js`, `recrawl-osborne.js` — read-only
  reference pulls of the deployed Kamatera scrapers.
- `recrawl-coordonne.js` — NEW (2026-06-10). Additive `all_images` re-crawl for
  the coordonne onboarding cohort. The original WP-API scraper captured only one
  hero image; this re-crawls each product PAGE and fills `all_images` + `specs`
  on existing rows (UPDATE only — never inserts, never publishes, never touches
  SKUs). Run `--canary 5` first to confirm the gallery selector against live HTML,
  then `--all --net-new` for the full pass. Deploy + run is Steve-gated.
- `recrawl-designers-guild.js` — NEW (2026-06-10). Additive `all_images` re-crawl
  for designers_guild. The deployed scraper grabs only the first `imageId`
  (`/image-nonwebp/600/<id>`); this re-crawls each product page and collects the
  product's FULL gallery via the `/image-nonwebp/1024/<id>` scoping heuristic
  (related-product cards only appear at 156/239 thumb sizes, so 1024 = own gallery).
  Verified 1→5 images on a sample, URLs HTTP 200. UPDATEs `vendor_catalog`
  (vendor_code='designers_guild') — never inserts/publishes/touches SKUs.
  `--net-new` = the 985 clean net-net-new rows (excludes 87 SKU-collisions incl.
  79 Ralph Lauren `PRL*`). Run `--canary 5` first. Deploy + run is Steve-gated
  (parked at `~/.claude/yolo-queue/pending-approval/designers_guild-recrawl.md`).

Local commits only. No push, no deploy.