PodcastFeedMaker

It all started with a simple need: generate clean, standards-compliant podcast RSS feeds without hacking XML by hand.

Between 2006 and 2012, I was producing electronic music podcasts. Back then, I managed everything with a homemade PHP backoffice — a DIY setup that worked, but required tremendous effort to stay compatible with iTunes, aggregators, and validators.

Years later, while working on audio projects, the need came back. But this time, I wanted to do things right: a modern, typed, tested Swift library that respects all current standards — Apple Podcasts, Podcast Namespace, PSP-1.

It was while developing this library that the idea for Podcast Maker was born. The library first, the app second.

PodcastFeedMaker is a Swift library for generating, parsing, and validating podcast RSS feeds. Not a wrapper, not a template generator — a real, complete modeling of the RSS format with all its namespaces.

The key feature? It's bidirectional. You can build a feed from scratch with Swift models, generate XML, but also parse existing XML and get it back as typed models. Full round-trip, no data loss.

It covers the 7 XML namespaces found in the podcast world: RSS 2.0, iTunes, Podcast Namespace 2.0 (with all 32 tags), Atom, Dublin Core, Content Module, and Podlove Simple Chapters. For validation, five major platforms are supported: Apple Podcasts, Spotify, Amazon Music, Podcast Index, and the PSP-1 standard.

All of this with zero external dependencies in the core library — pure Swift with Foundation, Linux-compatible from day one.

Here's the complete workflow: build a feed, generate XML, re-parse it, and validate it.

The podcast RSS world is a stack of XML namespaces that have accumulated over the years. PodcastFeedMaker handles them all — every tag, every attribute is modeled, generated, parsed, and validated.

The Podcast Namespace led by Podcast Index brings 32 new tags organized in phases. PodcastFeedMaker implements them all:

Every platform has its own requirements. Apple wants HTTPS everywhere and artwork between 1400 and 3000px. Spotify prefers MP3 and limits files to 200 MB. PSP-1 requires a GUID on every episode.

Instead of discovering these issues at submission time, FeedValidator checks everything upfront:

Building a feed with PodcastFeedMaker means chaining modifiers in a fluent way. No XML to write, no magic strings — everything is typed.

18 modifiers are available for channels (author, categories, image, owner, funding, etc.) and 14 for items (description, duration, season, episode, persons, transcripts, soundbites, etc.).

Three factories set the MIME type automatically — no need to remember if it's audio/mpeg or audio/mp3:

FeedGenerator produces complete XML synchronously. For massive catalogs (10,000+ episodes), StreamingFeedGenerator generates in async chunks:

Four modes for handling namespace declarations in generated XML:

The parser handles all 7 namespaces with best-effort error recovery. A malformed feed doesn't crash everything — you get what's readable along with diagnostics about what caused issues.

In the real world, podcast feeds use... creative date formats. The parser handles 12 of them:

Four pre-configured templates depending on your needs. Each activates a set of tags and validations suited to the use case:

Templates are composable with the + operator and configurable via .requiring(), .recommending(), .targeting().

Parse an existing feed, modify it, and regenerate — without losing elements you don't know about. This is crucial when working with third-party feeds that use custom extensions.

FeedDiff detects differences between two feed versions:

Two chapter systems coexist in the podcast ecosystem. PodcastFeedMaker supports both:

podcastfeed exposes all library functionality from the command line. Perfect for automation and CI/CD pipelines.

For CI/CD integration: 0 = success, 1 = error, 2 = warnings without errors. In --strict mode, warnings become errors.

2655 tests across 283 test suites. Complete coverage of all 7 namespaces, 12 date formats, 5 validators. All tests run on macOS and Linux in CI — no mocks, real XMLParser and real Data.

The specs PodcastFeedMaker is built on: