# ColdFilm project memory

## Project summary
- Project name: ColdFilm
- Type: static browser-only frontend, no build step, no backend.
- Goal: monitor new ColdFilm releases, show posters, and let user open torrent/magnet links from browser.
- Intended use: local network only; not public-facing.
- Requirement: VPN enabled for accessing coldfilm.ink from Russia.

## File structure
- index.html — page shell with header, status area, refresh button, film list container, quality modal.
- style.css — dark theme, responsive card layout, modal styling, buttons, status colors.
- script.js — main app logic.
- README.md — project description and operational notes.
- project-memory.md — persistent project architecture and change log.

## Core architecture
### Frontend-only architecture
- HTML renders skeleton and DOM containers.
- CSS supplies layout and visual design.
- JavaScript performs all runtime behavior.

### Runtime data flow
1. On page load, `loadFilms()` runs immediately.
2. `fetchWithProxy()` retrieves the main ColdFilm page HTML using a list of public CORS proxies.
3. `parseColdfilm(html)` extracts release titles and page URLs, and tries to pair them with posters.
4. Results are rendered into the DOM as list items with poster, title, and a “Download” button.
5. When user presses “Download”, `showQualityModal(filmName)` loads the film's detail page, extracts `.torrent` and `magnet` links, and shows quality buttons in a modal.
6. `downloadTorrent(url)` opens the chosen torrent/magnet link in a new browser tab.

## State and cache
- `currentFilmName` stores the currently selected film title.
- `availableTorrents` stores extracted torrent options for the modal.
- `filmData` is a `Map` cache keyed by film title, storing `{ url, poster }`.

## Parsing strategy
- Parsing is based on regex matching against HTML.
- Release links are detected via `href="/news/..."` patterns and titles from `class="kino-h"` and `title="... [Смотреть Онлайн]"`.
- Posters are extracted using `<img>` `src`, `alt`, or `title` attributes.
- The parser filters out Telegram-related entries.

## Quality detection logic
- Torrent links are classified by filename substring:
  - `720p` / `hd720` → `720p`
  - `1080p` / `hd1080` → `1080p`
  - `4k` / `2160p` → `4K`
  - otherwise → `Стандарт`
- Magnet links are labeled `Magnet`.
- Quality buttons are sorted by priority: `4K`, `1080p`, `720p`, `Стандарт`, `Magnet`.

## Reliability constraints
- Uses third-party public CORS proxies; if proxies fail, the app cannot fetch content reliably.
- Parser is fragile because it depends heavily on the remote site's HTML structure.
- Browser fetches are subject to CORS, proxy availability, and remote host changes.

## Auto-refresh behavior
- `loadFilms()` is called on page load.
- `setInterval(loadFilms, 15 * 60 * 1000)` refreshes the list every 15 minutes.

## UI behavior
- Status panel displays loading, success, error, and downloading states.
- The refresh button triggers `loadFilms()` manually.
- Modal allows quality choice before opening torrent.

## Known implementation notes
- `fetchWithProxy()` loops through `PROXIES` and returns the first successful response.
- The `PROXIES` list includes public proxies; it should be reviewed and cleaned periodically.
- The code is intentionally simple and dependency-free to keep deployment easy.

## Current project principles
- Minimal dependencies.
- No backend/server-side logic.
- Direct browser execution only.
- Keep it easy to host on static web server / Synology Web Station.

## Recent implementation update
- Removed the broken proxy entry with an invalid host string.
- Reworked proxy fetching to use an `AbortController` timeout instead of the unsupported `fetch({ timeout })` option.
- Added `normalizePosterUrl()` to centralize poster URL normalization.
- Broadened torrent-link regex to support query strings in `.torrent` URLs.
- Verified JavaScript syntax with `node --check script.js` and confirmed exit code `0`.

## Best practice for future edits
- Preserve the current static browser-only architecture.
- If changing parser logic, keep it regex-based and resilient to HTML variations.
- Prefer small, local changes over large rewrites.
- Record any new dependency, proxy change, or URL structure change here.