@proveanything/smartlinks 2.1.10 → 2.1.12

package/dist/docs/tag-ranges.md

@@ -0,0 +1,77 @@
+# NFC tag ranges and partners
+
+For partners and enterprise clients who run their own tag programmes: mint a batch of tags, send them
+to a manufacturer with instructions, get the chip UIDs back, then put the tags to work for a client or
+product. Ordinary business collections don't use this.
+
+## The words
+
+| Word | Means |
+|---|---|
+| **Claim set** | A pool of codes on one tap domain and one tag type (NTAG 424 DNA, NTAG 213/215/216, or QR). Most partners need one. |
+| **Range** | A named, contiguous block of tags inside a set: "Tap to donate - February", 100 tags, indexes 201-300. |
+| **Index** | A tag's place in its set. Ranges never overlap and indexes are never reused. |
+| **Partner** | An organisation above collections that owns sets and assigns their tags into collections its people administer. Set up by SmartLinks staff. |
+
+## Who can do what
+
+`scope` is `{ partnerId }` for a partner's sets, or `{ collectionId }` for a collection's own sets.
+
+| Role | Can |
+|---|---|
+| Partner owner / admin, or a collection admin | Make sets, mint ranges, download the printer file and instructions, import UIDs, program tags, assign |
+| Partner assigner | See ranges and tags (never keys), assign |
+
+Assigning always needs you to be an admin of the collection the tags go to.
+
+## The flow
+
+```ts
+import { partner, claimSet } from '@proveanything/smartlinks'
+
+const [p] = await partner.list()                       // the partners you belong to
+const scope = { partnerId: p.id }
+const [set] = await partner.claimSets(p.id)            // or partner.createClaimSet(p.id, { name, tagType: 'ntag424', tapDomain })
+
+// 1. Mint
+const range = await claimSet.ranges.create(scope, set.id, { name: 'Tap to donate - February', quantity: 100 })
+
+// 2. Send to print: the CSV (carries each tag's key) and the instruction sheet
+const file = await claimSet.ranges.exportFile(scope, set.id, range.id)          // { filename, csv }
+const sheet = await claimSet.ranges.instructions(scope, set.id, range.id, 'pdf') // { filename, base64 }
+await claimSet.ranges.update(scope, set.id, range.id, { stage: 'sent', manufacturer: 'Print Co', orderRef: 'PO-123' })
+
+// 3. Receive: the manufacturer sends back the same file with UID filled in
+const check = await claimSet.ranges.importUids(scope, set.id, range.id, { csv: returnedCsv })
+// check.rows: ok / index_mismatch / bad_uid / uid_in_use / … ; nothing written yet
+await claimSet.ranges.importUids(scope, set.id, range.id, { csv: returnedCsv, commit: true })
+
+// 4. Assign all of it, or from-to
+await claimSet.ranges.assign(scope, set.id, range.id, { collectionId: 'reading-fc', productId: 'scarf', start: 201, end: 250 })
+```
+
+**Programming on a phone instead of a manufacturer** (Sidekick): `ranges.next()` gives the lowest
+tag without a chip, with its URL and key; write it, then `ranges.recordTag(scope, set.id, range.id, code, { uid })` (add `password` for a plain tag's write password).
+
+To find where a chip or code went, `partner.lookupTags(partner.id, { uid })` or `{ code }` searches every set the partner owns and returns each hit's set, tag and range, never a key.
+
+## The printer file
+
+One row per tag, in index order. The manufacturer fills in `UID` and sends the same file back; rows are
+matched by `Code`, and `Index` is checked.
+
+| Tag type | Columns |
+|---|---|
+| NTAG 424 DNA | `Index, Code, URL, PiccOffset, Key, UID` |
+| NTAG 213/215/216 | `Index, Code, URL, MirrorOffset, MirrorPage, MirrorByte, UID` |
+| QR | `Index, Code, URL` |
+
+`URL` is `https://{tap domain}/t/{set}/{code}` plus the placeholder the chip fills on every read
+(`?p=` + 32 zeros for 424 SUN, `?m=00000000000000x000000` for the 21x UID and counter mirror). The
+offsets are byte positions in the NDEF data, assuming the `https://` URI prefix code and a short record.
+
+## Notes
+
+- Range responses are never cached by the SDK: tags change as they're programmed and several carry keys.
+- A tag bound to a proof (minted or claimed) is never moved or cleared by `assign`; it's reported in `skipped`.
+- `status` is the stage (`minted`, `sent`, `received`), or `assigned` once every tag is assigned.

package/dist/http.js

@@ -219,6 +219,9 @@ function shouldSkipCache(path) {
     // Bookings change by the second (places left, busy times, a booking's status) — never serve them stale.
     if (/\/app\/[^/]+\/bookings(\/|\?|$)/i.test(path))
         return true;
+    // Tag ranges change as tags are programmed, and carry the tags' keys — never cached, never persisted.
+    if (/\/claimSet\/[^/]+\/ranges(\/|\?|$)/i.test(path))
+        return true;
     return false;
 }
 /** Evict the oldest (LRU) entry when the cache is at capacity. */