README file from
GithubXHS Product Search
Collect data from Xiaohongshu (小红书) product and shop pages into structured JSON — works on desktop and mobile Obsidian.
What it does
A product link (or a share text / product ID) → a normalized product record:
- Title, price, final price (after coupon), shipping origin, shipping fee, dispatch time
- Shop name, shop rating, shop total sales, follower count, shop link
- Listing date / days on shelf (only when the page exposes it)
- SKU count, selected SKU, product badges, sales
A shop link → shop info + the product list visible on the shop homepage (shop name, total sales, followers, positive-review rate, dispatch time; each product with title, price, sales, listing date/age, badges).
Sales figures are kept exactly as the page states them: if the page only shows a lower bound ("1.5万+") or a per-SKU figure, the record keeps the raw text and a scope marker instead of pretending it is exact.
It works standalone (ribbon icon or command palette) and exposes a small API (checkDeps() / collect() / mode() / setMode()) so other plugins can use it as a data backend — e.g. the author's goods-trading workbench plugin.
Save as Markdown note
When used standalone (no workbench), click “保存为 Markdown 笔记” after collecting:
- Data goes into the note's frontmatter (properties) — visible in Obsidian and queryable;
- The body holds only: title, product link, and a “每日跟踪（变动）” table (date / sales);
- Re-collecting the same product updates the existing note (refreshes properties + updates that day's row, appends a new row on another day) — no duplicate notes;
- The save folder is configurable (default
小红书商品采集).
How it works
- The collector is built in. It calls two public Xiaohongshu endpoints with Obsidian's cross-platform
requestUrl(which bypasses CORS on both desktop and mobile):- Product detail:
mall.xiaohongshu.com/api/store/jpd/edith/detail/h5/toc - Shop page:
www.xiaohongshu.com/shop/<sellerId>(server-rendered, product JSON embedded)
- Product detail:
- No Python, no external script, no
pip install. No login state, no cookies, no account data. - Requests to
xiaohongshu.comare made only when you run a collection, from your device. - Results are returned to you inside Obsidian; the plugin does no telemetry and writes nowhere else.
Requirements
- Obsidian 1.4.0+ — desktop (Windows / macOS / Linux) or mobile (iOS / Android)
- Nothing else. No Python, no Node, no external tools.
Usage
- Install and enable the plugin.
- Click the shopping-bag ribbon icon (or run “采集一个商品” from the command palette).
- Paste a product link / share text / product ID (a shop homepage link collects the shop instead).
- The result modal shows the collected fields.
- Click “保存为 Markdown 笔记” to save the result as a Markdown note in your vault (folder configurable), or “复制为 JSON” to copy it.
Standalone vs. workbench: on its own, use “保存为 Markdown 笔记” to write a .md into the vault (same idea as the note-collector plugin). When used with the author's goods-trading workbench, the confirm dialog decides which multi-dimensional tables get the data — this plugin does not duplicate that.
Commands:
- 采集一个商品（链接或商品 ID） — collect one product
- 采集器自检 — check that the built-in engine is ready
Short share links (
xhslink.com) are auto-expanded on desktop. On mobile, paste the full product link or the raw product ID instead.
Notes & limitations
- Fields that only exist in the Xiaohongshu mobile app (e.g. shop opening date) are reported as empty rather than guessed.
- Shop collection returns only the products visible on the shop homepage, not the full catalogue.
- Public pages only; the collector intentionally has no login capability.
- Sales numbers are shown exactly as the page states them, together with their raw text and scope marker.