eslite-spectrum

name	eslite_spectrum
description	Platform rules and field mappings for the 誠品生活日本橋 (eslite spectrum Nihonbashi) scraper

eslite_spectrum — Source Skill

Platform Overview

• Name: 誠品生活日本橋 (eslite spectrum Nihonbashi)
• URL: https://www.eslitespectrum.jp/news
• Type: Taiwanese cultural bookstore and event space in Tokyo
• Location: COREDO室町テラス 2F, 東京都中央区日本橋室町3-2-1
• Rendering: Static HTML (requests + BeautifulSoup)
• Events/month: ~2–4 Taiwan-related events

Scraper Key

--source eslite_spectrum

Class: EsliteSpectrumScraper → key auto-derived as eslite_spectrum (matches SOURCE_NAME).

Strategy

1. Fetch /news listing page (static HTML).
2. Collect all /news/catalog/{id} links with published date and title.
3. For each item, fetch the detail page.
4. Filter: check Taiwan keywords against content_text = f"{title}\n{description}" only (NOT the full page HTML).
5. Extract event date from the detail page body (first YYYY-MM-DD string).

Field Mappings

Field	Source
source_id	eslite_spectrum_{catalog_id} (e.g. /news/catalog/9 → eslite_spectrum_9)
source_url	https://www.eslitespectrum.jp/news/catalog/{id}
name_ja	Article <h1> or equivalent title element
start_date	First YYYY-MM-DD string found in detail page body
location_name	Always "誠品生活日本橋" (hardcoded)
location_address	Always "東京都中央区日本橋室町3-2-1 COREDO室町テラス2F" (hardcoded)
raw_description	"開催日時: YYYY年MM月DD日\n\n" + main body text

Publication Rule Sync

• Pure publication classification is exact-only: normalized event_form must equal ['publication'].
• Pure publication rows keep seven intentional-null fields with sentinel locks: location_address, location_address_zh, location_address_en, business_hours, business_hours_zh, business_hours_en, location_prefectures.
• Preserve real DB prices (is_paid, price_info, price_amount); hide pure-publication prices only in UI and JSON-LD. Price fields, location_name, and location_url are outside the seven-field NULL/clear policy.
• Publisher (organizer) stays required even for pure publication rows.
• Mixed physical rows (book launch/talk/signing/lecture/workshop) must not include publication in event_form.

Taiwan Filter Rules

IMPORTANT: Keywords are checked against main content only:

content_text = f"{title}\n{description}"

Do NOT check page.text or the full HTML — 誠品 appears in every page's navigation and footer, which would cause all articles to match.

TAIWAN_KEYWORDS

["台湾", "Taiwan", "臺灣", "台灣", "台北", "高雄", "台中", "台南", "台日", "日台"]

Note: "誠品" is intentionally excluded — it appears in every page's nav/footer.

Skip patterns (_SKIP_TITLE_RE)

Articles with these title patterns are skipped before detail-page fetch:

会員募集 | メンバーズカード | ワークショップカレンダー | ポイント | お知らせ | 営業時間 | 定休日 | リニューアル

Date Format

Detail pages contain dates in YYYY-MM-DD format embedded in the body text. Example: 2026-06-07 extracted from 【開催日時】2026年6月7日.

The scraper uses _DATE_RE = re.compile(r"\d{4}-\d{2}-\d{2}") to extract the first match.

Known Issues / Edge Cases

• The site has very few news items at a time (~5 per listing page). No pagination needed.
• Some items are glass workshop calendars (non-Taiwan). _SKIP_TITLE_RE filters these.
• Catalog IDs are sequential integers; IDs may be reused if articles are deleted.