stranger

name	stranger
description	Scraper conventions for Stranger cinema (stranger.jp) — Eigaland JSON API, Taiwan movie filter via countries field
applyTo	scraper/sources/stranger.py

Stranger Scraper — Source-Specific Skill

Platform Profile

Property	Value
Site URL	https://stranger.jp/
API / Rendering	Eigaland JSON API (no auth)
Auth	None required
Rate limit	Conservative: 0.3s between calls
Source name	stranger
Source ID format	stranger_{movieId} (MongoDB ObjectId string)
Venue	Stranger, 東京都墨田区菊川3丁目6-13

API Endpoints

Endpoint	Purpose
GET /api/website/show/listByDomainAndDate?domain=stranger.jp&date=YYYY-MM-DD	Daily show list
GET /api/website/movie/detail/{movieId}?domain=stranger.jp	Full movie detail

Field Mappings

Event field	API source
raw_title / name_ja	movieDetail.movieName
start_date	Earliest date in 90-day window where movie appears
end_date	Latest date in 90-day window where movie appears
director	detail.directors joined with ", "
raw_description	Decoded base64 detail.synopsis + slogan + cast etc.
official_url	detail.officialPageUrl (movie's own website)
source_url	https://stranger.jp/movie/{movieId}
location_name	"Stranger" (fixed)
location_address	"東京都墨田区菊川3丁目6-13" (fixed)

Taiwan Relevance Filter

• Check movieDetail.countries array for "台湾" or "台灣".
• Filter applied in the list API loop — skip detail API call for non-Taiwan movies.
• Do not rely on keyword matching in title or synopsis (countries field is authoritative).

Synopsis Decoding

detail.synopsis is base64-encoded HTML. Decode via:

1. base64.b64decode(b64_str).decode("utf-8")
2. Strip HTML tags with html.parser.HTMLParser

The resulting plain text may contain non-breaking spaces; strip() each part.

Dedup Strategy

One Event per movieId — not one per screening day. The scraper loops all 90 days to find min_date / max_date, then calls detail API once per unique movieId.

Date Notes

• openDate in the list API is the movie's release date in Japan, not the screening date.
• Use the calendar date of the API query (listByDomainAndDate?date=...) as the screening date.
• start_date = earliest date the movie appears across the 90-day window.
• end_date = latest date the movie appears across the 90-day window.
• Both are datetime with tzinfo=JST (required by dedup_events).

Troubleshooting

Symptom	Cause	Fix
No Taiwan events found	Film schedule not yet published	Normal; API only returns published shows
director is empty list	No directors in API data	No fix needed — annotator will try to fill
synopsis decode error	API returned non-base64 data	_decode_synopsis() returns "" gracefully
91 events (one per day)	Bug: forgot to dedup by movieId	Ensure taiwan_movies dict deduplication in loop

Pending Rules

Add lessons learned here as they arise.