@blamejs/blamejs-shop 0.3.42 → 0.3.44

package/CHANGELOG.md

@@ -8,6 +8,10 @@ upgrading across more than a few patches at a time.
 
 ## v0.3.x
 
+- v0.3.44 (2026-05-31) — **Customers can rate a delivered order, and the admin console moderates and replies.** There was no way for a customer to give feedback on how an order was fulfilled, and no way for an operator to act on it. Customers can now rate one of their own delivered orders on shipping, packaging, and how likely they are to recommend you, with an optional comment. The admin console gets a Ratings screen with a moderation queue of every flagged comment plus the most recent ratings, where an operator can flag a comment or post one public reply that the customer then sees on their order page. A customer can only rate an order they own, and every comment and reply is HTML-escaped wherever it is shown, so feedback text can't inject markup into another shopper's or an operator's page. **Added:** *Order ratings* — A customer viewing one of their own orders can rate it on three axes — shipping, packaging, and likelihood to recommend (1 to 5) — and optionally leave a comment. Each order can be rated once. The rating, and any public reply from the store, appear on the customer's order page. A customer can only rate an order tied to their own account; the rating is always recorded against the signed-in customer, never a value taken from the request. · *Ratings moderation in the admin console* — A new Ratings screen lists recent ratings with their scores and comments, alongside a moderation queue that surfaces every flagged comment regardless of its score, so a flagged comment on an otherwise high rating is never hidden. An operator can flag a comment for moderation and post a single public reply to a rating, which is then shown to the customer. Comments and replies are HTML-escaped everywhere they are rendered.
+
+- v0.3.43 (2026-05-31) — **Export orders for a date range as CSV or NDJSON, with scheduled exports.** There was no way to pull a bulk export of orders over a date range. A new Exports screen in the admin console previews how many orders and how much revenue a date range covers, then downloads them as a CSV or NDJSON file. The export can also be scheduled and managed as a queued job. Customer email addresses are exported only as a hash, never in the clear, and any field that could be interpreted as a spreadsheet formula is neutralized so an exported file can't run code when opened. This release also updates the vendored blamejs runtime to v0.14.16. **Added:** *Order export* — An Exports screen in the admin console exports orders over a chosen date range. It first shows a preview — the order count, revenue, and average order value for the range — then downloads the orders as a CSV or NDJSON file with a standard set of columns. Exports can also be queued as scheduled jobs and cancelled before they run. Customer email is exported as a hash rather than in the clear, and any cell that begins with a character a spreadsheet would treat as a formula is escaped, so an exported file can't execute anything when opened in Excel or Sheets. **Changed:** *Vendored blamejs runtime updated to v0.14.16* — The bundled blamejs runtime is updated to v0.14.16, which validates connection ports at configuration time — a malformed port (a non-integer or out-of-range value) is now rejected with a clear error at startup instead of silently falling back to a default. No operator action is required.
+
 - v0.3.42 (2026-05-31) — **The blog now pages through every post, and the byline shows the shop name.** The blog index showed only the 12 most recent posts with no way to reach the rest, so older posts were a dead end for anyone browsing — even though they stayed in the sitemap. The index now has previous/next pagination, so every published post is reachable. The post byline and the article's structured data now show the shop name instead of an internal author identifier, and a missing blog post or product page no longer returns a body on a HEAD request. **Fixed:** *Blog index pagination* — The blog index listed only the 12 most recent posts and offered no way to reach older ones from /blog. It now has previous/next pagination — the same control used on collection and search pages — so every published post is reachable, with the next link appearing only when there is genuinely another page. · *Blog byline shows the shop name, not an internal id* — The blog post byline and the article's structured data displayed the raw internal author identifier. They now show the shop name, so the public byline and the author Google reads from the structured data are meaningful rather than an internal id. · *Missing pages don't return a body on HEAD* — A missing blog post or product page returned a full page body even for a HEAD request. Those 404 responses now omit the body on HEAD, matching the other pages and the HTTP spec.
 
 - v0.3.41 (2026-05-31) — **The public product API no longer leaks internal error details, and bad order links return Not Found.** A server-side error on the public product API could include the raw internal error message — potentially database or query details — in its 500 response to anonymous callers. Those responses now return a generic message, with the real error recorded server-side only, matching how the admin side already behaves. Separately, opening an order or payment page with a malformed link now returns a clean Not Found instead of a server error. This release also updates the vendored blamejs runtime to v0.14.15. **Changed:** *Vendored blamejs runtime updated to v0.14.15* — The bundled blamejs runtime is updated to v0.14.15, which adds a recognized-purpose vocabulary to its consent ledger (with lawful-basis enforcement for regulated purposes) and a new privacy namespace for periodic third-party vendor-review attestations. These are additive, opt-in capabilities; existing behavior is unchanged and no operator action is required. **Fixed:** *Public product API hides internal error details* — When the public product API (the catalog endpoints) hit a server-side error, its 500 response could carry the raw internal error message — including database or constraint text — to any caller. It now returns a generic "something went wrong" message and records the real error server-side only, the same way the admin console already scrubs its errors. Validation errors still return their helpful 400 message unchanged. · *Malformed order and payment links return Not Found* — Opening an order confirmation or payment page with a malformed (non-identifier) link returned a server error instead of a Not Found. Both pages now return a clean 404 for a malformed or unknown link, matching the order reorder and cancel pages.