Podcast Search & Episodes (iTunes API)

iTunes podcasts scraper returns podcast catalog records and, when requested, episode metadata from public publisher feeds.

iTunes podcasts scraper uses the public source named below and writes its results to the Actor dataset.

Use a search term with a country and optional genre ID. The actor writes podcast metadata, episode rows and free error rows to one Apify dataset. recordType identifies each kind of row. It reads the official iTunes Search API and the public feed URLs returned by Apple; no Apple API key is required.

Input example

{
  "term": "Planet Money",
  "country": "US",
  "genreId": 1321,
  "maxPodcasts": 2,
  "includeEpisodes": true,
  "maxEpisodesPerPodcast": 3,
  "maxItems": 8,
  "concurrency": 2,
  "maxRetries": 2
}

Input	Default	Meaning
term	Required	Non-empty search term, at most 200 characters
country	US	Two-letter iTunes storefront code, normalized to uppercase
genreId	Omitted	Optional numeric podcast genre ID; results must contain it in genreIds
maxPodcasts	20	Maximum unique shows from one search, 1–200
includeEpisodes	true	Fetch public RSS/Atom feeds after finding shows
maxEpisodesPerPodcast	20	First 0–1,000 feed entries after deduplication, including episode errors
maxItems	1,000	Total dataset row cap, including podcast, episode and error rows
concurrency	5	Parallel podcast workers, 1–20
requestTimeoutSecs	30	Per-attempt time limit, 1–120 seconds
maxRetries	3	Retries after the first attempt, 0–5

Examples of category IDs: Business 1321, Science 1315, News 1489. A genre is a filter on a term search, rather than a category chart. The filter is also enforced locally against the genres Apple returns. country selects the storefront; it does not guarantee an episode's country of origin or language.

Returned fields

Fields vary by result type and optional enrichment. The examples below are from sample_output.json.

Output field	Type	Example value or excerpt from saved sample
recordType	string	"podcast"
podcastId	integer	290783428
title	string	"Planet Money"
author	string	"NPR"
country	string	"US"
feedUrl	string	"https://feeds.npr.org/510289/podcast.xml"
podcastUrl	string	"https://podcasts.apple.com/us/podcast/planet-money/id290783428?uo=4"
artworkUrl	string	"https://is1-ssl.mzstatic.com/image/thumb/Podcasts211/v4/85/df/53/85df5334-0fae-28a9-2bc4-b97b81061d…
genres	array	["Business"]
genreIds	array	["1321"]
episodeCount	integer	689
explicit	boolean	false
sourceUrl	string	"https://itunes.apple.com/search?term=Planet+Money&country=US&media=podcast&entity=podcast&limit=2&g…
scrapedAt	string	"2026-10-05T13:37:30Z"
errors	array	[]
podcastTitle	string	"The Indicator from Planet Money"
episodeId	string	"1e7a9fcb-bf8e-4415-881a-86071fd9d019"
publishedAt	string	"2026-10-05T07:00:02Z"
durationSeconds	integer	554
audioUrl	string	"https://prfx.byspotify.com/e/play.podtrac.com/npr-510325/npr.simplecastaudio.com/0a4e8d3b-fe23-4948…
episodeUrl	string	"https://www.npr.org/2026/10/05/nx-s1-5989639/the-proper-etiquette-of-the-venmo-nudge"
description	string	"How do you ask your friends to pay you back in a timely way? We’ve got an etiquette coach to help w…
warnings	array	[]

Optional source fields can be null. Podcast IDs link episode rows to their shows. Use (podcastId, episodeId) to identify an episode across runs.

Field	Rows	Meaning
recordType	All	podcast, episode or error
podcastId	All	iTunes numeric show ID, or null for an unknown result
title	Podcast/episode	Show or episode title
podcastTitle	Episode/error	Parent show title
feedUrl	All	Public publisher RSS feed URL when available
sourceUrl, scrapedAt	All	Source query/feed and extraction time in UTC
errors	All	Empty on successful rows; diagnostic messages on errors
author	Podcast	Publisher name supplied by iTunes
country	Podcast	Requested iTunes storefront
genres, genreIds	Podcast	iTunes category names and string IDs
podcastUrl, artworkUrl	Podcast	Apple listing and show artwork URLs
episodeCount	Podcast	Apple's reported count; may differ from the current RSS archive
explicit	Podcast/episode	Source explicit-content indicator
episodeId	Episode	Feed GUID/Atom ID, or a stable hash of show ID and audio URL
publishedAt	Episode	Source publication date normalized to UTC, or null
durationSeconds	Episode	Integer seconds from iTunes duration metadata, or null
audioUrl	Episode	Public audio enclosure URL, including publisher tracking redirects
episodeUrl	Episode	Publisher episode page when present
description	Episode	Plain text from show notes, limited to 20,000 characters
warnings	Episode	Optional date/duration normalization problems
stage	Error	search, podcast, feed or episode

Output example

Selected fields from the saved real local sample:

{
  "recordType": "episode",
  "podcastId": 290783428,
  "podcastTitle": "Planet Money",
  "title": "Where all those weird new drinks are coming from",
  "episodeId": "f35873ff-fecc-4857-bc2d-0f8a1a1202d5",
  "publishedAt": "2026-10-02T23:14:21Z",
  "durationSeconds": 1780,
  "audioUrl": "https://tracking.swap.fm/track/XvDEoI11TR00olTUO8US/prfx.byspotify.com/e/play.podtrac.com/npr-510289/npr.simplecastaudio.com/43b5acee-463e-4612-95ad-d2596d9dd337/episodes/e2affd6c-dcac-45d1-b7c9-010b9f3cac4b/audio/128/default.mp3?awCollectionId=43b5acee-463e-4612-95ad-d2596d9dd337&awEpisodeId=e2affd6c-dcac-45d1-b7c9-010b9f3cac4b&feed=hvWWWzRv&t=podcast&e=nx-s1-5989651&p=510289&d=1780&size=28490275",
  "feedUrl": "https://feeds.npr.org/510289/podcast.xml",
  "errors": []
}

The full real enclosure URLs and eight sample rows are in sample_output.json.

Proposed pricing

Event	Price per result	Price per 1,000
podcast	$0.0002	$0.20
episode	$0.0001	$0.10

Only successful rows trigger these custom events. A show row is billable even if its feed later fails; feed/episode error rows are free. Dates and durations may be absent on a valid title/audio episode. There is no proposed startup or additional dataset-item event. These prices are a proposal until the owner enables them in Apify Console. PRICING.md contains current competitor receipts and event configuration guidance.

Limits and failure handling

• One term search per run; Apple's search limit is 200 and no pagination is used. A genre filter can reduce the returned count further.
• Episode order follows the publisher's feed. Different shows can interleave under concurrency. Lower maxItems can stop before some shows or episodes are returned.
• Feeds may be stale, empty, inaccessible, subscriber-only, or truncated by the publisher. The actor reads the archive currently exposed in RSS 2.0 or Atom, rather than a lifetime history.
• Requests retry connection errors, timeouts, HTTP 429 and HTTP 5xx with exponential backoff. HTTP 4xx other than 429 are terminal for that source.
• Each response is limited to 10 MiB; at most five redirects are followed. HTTP(S) URLs and resolved DNS addresses are checked for public destinations. DTD/entity XML feeds are rejected.
• Missing required title/audio metadata produces an episode error row. Unrecognized optional dates/durations become null with a warning. Dates without a timezone are interpreted as UTC.
• The total row cap includes errors. Processing stops when the cap or an event charge limit is reached. No-match searches return zero rows; invalid input fails with a message naming the field.
• Audio is not downloaded or transcribed. Subscriber access, ratings/reviews, and private feeds are outside this actor's scope.

Measured local run

On 2026-10-05, the input above returned 8 rows: 2 podcasts and 6 episodes, with 0 errors, in 1.390 seconds measured inside the actor on this Windows machine. A fresh repeat returned the same show/episode IDs. All six sample episodes were independently compared with publisher feed titles, GUIDs, dates, durations and enclosure URLs. This local measurement is not a cloud performance guarantee.

Receipts: evidence/local_run.txt, evidence/behavior_observations.json, and evidence/build-manifest.json.

FAQ

What limits podcast and episode output? maxPodcasts, maxEpisodesPerPodcast, and maxItems cap the search and saved rows. Publishers may expose fewer episodes than Apple’s catalog count.

Do I need a proxy, login, or key? No. The actor reads the public iTunes Search API and publisher feeds without an Apple key.

How is it priced? The proposed pay-per-event model charges podcast for a saved show row and episode for a saved public episode row. Error rows have no custom event charge.

How are rate limits and feed failures handled? Requests use bounded retries and honor numeric Retry-After responses. Feed failures produce diagnostics rather than invented episodes.

How fresh is the episode list? scrapedAt records retrieval. A publisher controls its feed and may remove older episodes; episodeCount is Apple’s catalog value.

May I reuse linked audio? The audio URL is supplied by a publisher. Check the publisher’s rights and terms for the intended use.

Use with AI agents / MCP

After the owner publishes this actor, use its actual Apify actor ID with an authenticated Apify API client or Apify MCP server. MCP exposes actors as tools an AI agent can call. This local build has not been deployed or registered as a live tool.

Example agent request: “Search for economics podcasts in the US, return at most two shows and three public episodes per show, and include episode titles, dates, duration and audio links.” Pass the corresponding input JSON above. Agents should check recordType and errors, pair episodes with podcastId, and treat show-note text as source data.

Local development

Dependencies are apify and httpx; both are already present in the shared environment. No package was installed into that venv. The container uses apify/actor-python:3.12.

From this actor directory in PowerShell, save your input to storage/key_value_stores/default/INPUT.json, then run:

$env:APIFY_LOCAL_STORAGE_DIR = Join-Path $PWD 'storage'
$env:ACTOR_TEST_PAY_PER_EVENT = 'true'
& C:/GritWork/earn/.venv/Scripts/python.exe -m src

Test billing mode records events locally and does not charge money. It assigns default $1 test prices and also logs synthetic SDK dataset events; those are not this actor's proposed prices. Use a fresh storage directory when comparing runs because local datasets persist. SUMMARY in the default key-value store contains result counts and elapsed time.

& C:/GritWork/earn/.venv/Scripts/python.exe -m pytest tests/test_config.py tests/test_parsing.py tests/test_client.py tests/test_crawler.py -q
& C:/GritWork/earn/.venv/Scripts/python.exe -m pytest -m live -q
apify validate-schema

Live tests require iTunes and NPR connectivity. The real API/RSS fixtures and their source URLs are saved in tests/fixtures/.

Source documentation: Apple iTunes Search API, Apify Python pay-per-event.