amponce/archive-movie-browser

What happens Descriptions from Archive.org are turned into plain text, but only a handful of entities are decoded, so Caf&eacute;, &copy; 1956, &mdash; and &hellip; show exactly like that. A > inside a quoted attribute (<a href="x>y">) also leaks y"> into the text. What should happen Café © 1956 — link … Steps to reproduce In the repo: node --input-type=module -e "import {plainText} from './src/services/plainText.js'; console.log(plainText('Caf&eacute; &copy; 1956 &mdash; <a href=\"x>y\">link</a> &hellip;'))" Where to look src/services/plainText.js (a bigger entity table, or numeric decoding plus the common named ones; keep it working in node, since the MCP server uses it too). How to check Add these cases to the plainText test file; npm test passes.

What happens When a filter removes every film in the first batch but Archive.org still has more pages, /browse says "No movies found matching your criteria" and shows a Load more button underneath. Clicking Load more can fetch several more pages, still find nothing, and the page looks exactly the same, so it feels broken. What should happen One clear state. Either "Nothing yet in the uploads checked so far" with a "Keep looking" button, or just the empty state. Steps to reproduce Open https://www.orphanedfilms.com/browse and pick a collection that is mostly short uploads, then set the runtime filter to 90+ minutes. See the empty message and Load more together. Click Load more: nothing visible changes. (Easiest locally: mock Archive.org's search to return only 1-minute uploads with a large numFound.) Where to look src/components/browse/FilmGrid.jsx: the empty-state block keys on movies.length === 0 and the Load more block keys on nextPage, so both render. How to check With the mock ab

A round of small, well-scoped fixes found while testing the site end to end. Each sub-issue says what happens, what should happen, how to reproduce it on the live site, and where the fix likely lives. Most are one or two files. New here? Pick one, say you're taking it, and open a draft PR (see CONTRIBUTING.md for the claim rules). Questions are welcome on any of them. More will be added here as these get picked up.

What happens README "Feeds" says /api/tv/guide.xml is "XMLTV for the next 24 hours"; it runs from six hours ago to three days ahead (the /iptv page says this correctly). The table also leaves out /api/tv/channels.m3u and /api/tv/live/<channel-id>. The "How it is built" map lists five pages, four API functions and four scripts; the repo now has eleven pages, eight API functions and ten scripts, plus shelves.json/channels.json in src/programme and new hooks and ui primitives (also missing from src/ui/README.md). The "poster index" link at the top of mcp/README.md points to ../README.md#poster-index, which doesn't exist; the section is "How films get identified". What should happen The docs match the repo. Steps to reproduce Compare README "Feeds" with https://www.orphanedfilms.com/api/tv/guide.xml (last <programme> dates) and the /iptv page; compare "How it is built" with ls src/pages api scripts src/programme src/hooks src/ui; click "poster index" in mcp/README.md on GitHub. Where to

What happens In grid view on /browse, every card whose upload has no description asks our TMDB proxy for the film's full details (with cast) just to get the plot summary. The summary only shows in list view, so in the default grid those 24 requests per page (24 more per Load more) are wasted and slow down the lookups the page does need. What should happen No detail requests in grid view. Steps to reproduce Open https://www.orphanedfilms.com/browse in a private window with DevTools > Network filtered to api/tmdb: about 24 path=movie/…&append_to_response=credits requests, while no card shows a description. Where to look The overview effect in src/components/MovieCard.jsx: only run it when viewMode === 'list'. How to check Grid view: 0 of those requests. Switch to list view: descriptions still appear.