SchoolList

JSON fixture for CycleUni's admin School bulk-import feature.

• Target API: POST /api/v1/admin/schools/bulk/ (adminapi.views.AdminSchoolBulkImportView, staff-only)
• Request body: {"action": "preview" | "apply", "items": [...]} — send schools.json's items array as-is
• Matching key: email_domain (unique on the School model) — re-importing is idempotent; unchanged rows are skipped, differing rows are reported as modified
• Field shapes: name is the canonical English name, region is the CycleUni Region.code, translations holds localized fields keyed by language code

Contents: 101 real universities — 88 Taiwanese (region: "TW", zh-TW names) and 13 Hong Kong (region: "HK", zh-HK names) — with their real student-email domains. Tested end-to-end against the live endpoint in CycleUni-BE/tests/test_bulk_import_fixtures.py.

The build script also outputs region-specific files (schools.TW.json, schools.HK.json).

• Use schools.json if you are a superuser and want to import everything at once.
• Use schools.<region>.json if you are a regional admin managing a specific area, to avoid importing data outside your jurisdiction (which will be marked as forbidden/skipped).

Region is required

School.region is non-null, and the import writes item["region"] straight into it. An item without a region fails the whole batch; an item with the wrong one lets students of one market verify with another market's campus domain. Every item therefore carries a region, and 4_build_schools.py stamps it from --region.

Regions are also why translations are not keyed to a fixed language: Taiwan uses zh-TW, Hong Kong zh-HK. Both are Traditional Chinese but differ in vocabulary, and CycleUni serves them as separate languages.

Pipeline

# 1. Filter the world dataset by ISO 3166-1 alpha-2 code
python3 1_filter_universities.py --country TW     # -> tw_universities.json
python3 1_filter_universities.py --country HK     # -> hk_universities.json

# 2-3. Add Chinese names (Taiwan: LLM-translated, then applied)
python3 2_translate_schools_zhTW.py
python3 3_apply_zh_name.py

# 4. Merge into the bulk-import fixture, stamping the region
python3 4_build_schools.py --input tw_universities.json --region TW
python3 4_build_schools.py --input hk_universities.json --region HK

Steps 2-3 are for Taiwan only. Hong Kong's Chinese names are registered institutional names, not translations — 香港理工大學 is the university's actual name, and an LLM asked to translate "The Hong Kong Polytechnic University" will happily return a plausible-but-wrong variant like 香港理工學院. They are curated in step 1's output instead.

Source-data caveats

The upstream world_universities_and_domains.json needs filtering beyond the country code, handled in 1_filter_universities.py:

• Non-universities: cdnis.edu.hk (Canadian International School of Hong Kong) is K-12. Left in, its domain would pass campus-email verification.
• Campuses in another region: cuhk.edu.cn is CUHK-Shenzhen — same name, mainland China, different domain. It would give one Hong Kong university a second "valid" campus domain.

It also lists some universities twice under different English spellings sharing one domain (nthu.edu.tw, nchu.edu.tw, nsysu.edu.tw). Since email_domain is the unique key, the duplicate would be reported as modified on every import and flip the stored name back and forth. 4_build_schools.py collapses them to the first occurrence and reports what it dropped.