$ show sjdonado/idonthavespotify
This is a production-ready web application and API for converting music streaming service links across platforms. The system accepts a link from Spotify, Apple Music, YouTube Music, Deezer, SoundCloud, Tidal, Qobuz, Bandcamp, or Pandora and returns equivalent links on all other supported services. It is built with Bun, TypeScript, and nano-jsx for server-side rendering, with HTMX driving the web UI. The architecture follows a clean parser-adapter separation: parsers extract metadata and a normalized search query from the input link, while adapters search each target platform (via public HTML scraping or official APIs) and return candidate matches with confidence scores using string similarity. A notable implementation detail is the Spotify adapter, which circumvents API restrictions by extracting a TOTP-based anonymous access token from the Spotify web player JavaScript bundle—a clever workaround that avoids requiring a premium developer account. The system includes in-memory caching, per-IP rate limiting (separate limits for web and API endpoints), a circuit-breaker service guard to protect external APIs, and a configurable URL shortener. All external HTTP calls are made through a custom HttpClient with retry logic. Tests cover integration scenarios with mocked snapshots.
The LLM identified 5 files worth highlighting for reuse:
Provides `findBestMatch` and `scoreMatch` functions that use the `string-similarity` library to compute cosine similarity between a candidate text and a query. The `findBestMatch` function iterates over an array of `MatchCandidate` objects (title, artist, URL) and returns the best match with a normalized score, along with `isVerified` and `notAvailable` flags based on configurable thresholds. This is cleanly abstracted and could be extracted into any project that needs to rank search results by textual similarity.
Exports `cleanSearchQuery`, a pure function that normalizes a search query by removing or replacing a wide range of problematic Unicode characters: special quotation marks (German, French, CJK), fancy apostrophes, middle dots, special dashes, zero-width characters, and collapsing whitespace. It preserves emojis, standard punctuation, and accented characters. This is a drop-in utility for any project that needs to sanitize user input or API responses before comparison.
Implements a per-service circuit breaker (`ServiceGuard`) that tracks request count within a sliding window and failure rate within a separate window. When failures exceed a threshold, the circuit opens and blocks requests for a cooldown period before allowing a probe. It also enforces a maximum request budget per window. This pattern is reusable for any project that needs to protect external API dependencies from overuse or cascading failures.
Provides an in-memory sliding-window rate limiter (`RateLimiter` class) along with middleware wrappers `withRateLimit` and `withRateLimitHTML` for Bun HTTP servers. The middleware extracts the client IP from common headers, checks the rate limit, and sets `X-RateLimit-*` response headers. The `RateLimiter` handles cleanup of stale entries and is configurable via `windowMs` and `maxRequests`. This is a self-contained, reusable module for adding rate limiting to any Bun-based HTTP application.
Implements a generic in-memory cache with TTL (`InMemoryCache`) and provides typed helper functions for storing and retrieving search metadata, search result links, and access tokens. The cache automatically cleans up expired entries every 5 minutes. The `cacheStore` object exposes a simple async API (`get`, `set`, `del`, `reset`) that could be used as a lightweight caching layer in any TypeScript project without external dependencies.