feat(ebay): search.searchByImage — eBay's own visual search

[0xGhostCasper]

Adds client.ebay.search.searchByImage(...) for POST /v1/ebay/search/by-image — what eBay's camera icon does. 10 credits.

The method

const visual = await client.ebay.search.searchByImage({
  image_url: "https://example.com/sneaker.jpg",
  domain: "com",
  condition: "used",
  max_price: 200,
});

// or upload the picture yourself
const byUpload = await client.ebay.search.searchByImage({
  image_base64: readFileSync("./photo.jpg").toString("base64"),
});

The SDK's first eBay POST, so it follows the request(path, { method: "POST", body }) shape the web-scraping and YouTube-batch endpoints already use, and keeps the eBay module's snake_case parameter names.

Returns the same SearchResponse as search(), with query null.

Notes

• Exactly one of image_url / image_base64. Neither or both throws an Error client-side, so the wrong shape costs no round trip and no credits — mirroring the server's own validator.
• No sort_by. eBay ignores _sop on a visual results page, and a parameter that silently does nothing is worse than no parameter.
• location and language ride along. The visual SRP honours both, so EbaySearchByImageParams exposes them via two new unions, EbayLocation and EbayLanguage.
• EbayCondition gains graded / ungraded — eBay's trading-card conditions (slabbed vs raw). The scraper has accepted them on search, completed and by-image alike; the union here was stale, so this fixes every eBay method, not just the new one.
• image_base64 takes a bare payload or a data:image/jpeg;base64,... URL. JPEG and PNG, at most 10 MB decoded.

Test plan

• npm run typecheck — clean
• npm run lint — clean
• npm test — 279 passed (21 files); tests/ebay.test.ts 14 → 17
• npm run build — ESM + CJS + DTS build success
• New tests cover the POST target and JSON body, every filter passed through, sort_by never sent, and both bad-input shapes rejected before fetch is called

Docs: README.md eBay row and docs/ebay.md (sub-client table, param enums, a Search by Image section).

🤖 Generated with Claude Code

https://claude.ai/code/session_01RaAGmbLG3FotV7LFRkGhr3