feat(postcodes/TW): 371 Chunghwa Post codes (#1039)

[dr5hn]

Summary

• Imports Taiwan's 3-digit Chunghwa Post postal-area codes (371 rows) for Can we add a postcode for this? #1039
• 100% state FK resolution across all 22 CSC TW states
• Fixes TW regex ^\d{5}$ → ^\d{3}(\d{2,3})?$ to accept the three Taiwan postcode generations

Source

• flying-itmen-eagle/eagle-tw-open-data — community mirror of Taiwan government open data feeds (GPL-3 — flagged per Can we add a postcode for this? #1039 license-tier policy)
• File: taiwan_postal_code_information.csv (CP950 / Big5 — UTF-8 read produces mojibake)
• Upstream: Chunghwa Post / Ministry of Interior

Regex fix

Taiwan has used three postcode generations:

Era	Form	Example
pre-2020	3-digit area code	100
intermediate	3+2 = 5-digit	10001
2020+ (Chunghwa current)	3+3 = 6-digit	100002

This dataset is the 3-digit form. Old regex would have failed all three modern forms.

Edge cases handled

Source label	Mapped to	Reason
'Taoyuan City City'	TAO	source typo
'Diaoyutai'	ILA Yilan	Senkaku/Diaoyu — ROC administers under Yilan County
'Dongsha Islands, Nanhai Islands'	KHH Kaohsiung	Pratas — administered by Kaohsiung
'Nansha Islands, Nanhai Islands'	KHH Kaohsiung	Spratlys — administered by Kaohsiung

Distribution (top 5)

iso2	division	rows
KHH	Kaohsiung	40
TNN	Tainan	37
PIF	Pingtung County	33
NWT	New Taipei	29
TXG	Taichung	29

All 22 CSC TW states covered.

Test plan

• python3 -m py_compile bin/scripts/sync/import_taiwan_postcodes.py
• All 371 codes match ^\d{3}(\d{2,3})?$
• 100% state_id valid; state.country_id == 216; state_code == state.iso2
• No auto-managed fields (id, created_at, updated_at, flag)
• Idempotent merge (re-run produces no diff)

🤖 Generated with Claude Code

[github-actions]

CSC Validation Report

PR Format

• ✅ Description provided
• ✅ Data source linked
• ✅ Issue linked (recommended for data changes)
• ✅ Justification / context provided

Labels applied: data:countries, data:postcodes, large-contribution

⚠️ Large Contribution

This PR contains 621 records. Large contributions require manual review.

Schema Validation (621 records)

Errors (blocking):

• ❌ contributions/countries/countries.json: Record 1 ("Afghanistan"): "id" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 1 ("Afghanistan"): "created_at" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 1 ("Afghanistan"): "updated_at" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 1 ("Afghanistan"): "flag" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 2 ("Aland Islands"): "id" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 2 ("Aland Islands"): "created_at" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 2 ("Aland Islands"): "updated_at" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 2 ("Aland Islands"): "flag" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 3 ("Albania"): "id" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 3 ("Albania"): "created_at" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 3 ("Albania"): "updated_at" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 3 ("Albania"): "flag" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 4 ("Algeria"): "id" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 4 ("Algeria"): "created_at" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 4 ("Algeria"): "updated_at" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 4 ("Algeria"): "flag" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 5 ("American Samoa"): "id" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 5 ("American Samoa"): "created_at" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 5 ("American Samoa"): "updated_at" must not be included (auto-managed)
• ❌ contributions/countries/countries.json: Record 5 ("American Samoa"): "flag" must not be included (auto-managed)
• ...and 980 more errors

Warnings:

• ⚠️ contributions/countries/countries.json: Record 1 ("Afghanistan"): unknown field "population"
• ⚠️ contributions/countries/countries.json: Record 1 ("Afghanistan"): unknown field "gdp"
• ⚠️ contributions/countries/countries.json: Record 1 ("Afghanistan"): unknown field "area_sq_km"
• ⚠️ contributions/countries/countries.json: Record 1 ("Afghanistan"): unknown field "postal_code_format"
• ⚠️ contributions/countries/countries.json: Record 1 ("Afghanistan"): unknown field "postal_code_regex"
• ⚠️ contributions/countries/countries.json: Record 2 ("Aland Islands"): unknown field "population"
• ⚠️ contributions/countries/countries.json: Record 2 ("Aland Islands"): unknown field "gdp"
• ⚠️ contributions/countries/countries.json: Record 2 ("Aland Islands"): unknown field "area_sq_km"
• ⚠️ contributions/countries/countries.json: Record 2 ("Aland Islands"): unknown field "postal_code_format"
• ⚠️ contributions/countries/countries.json: Record 2 ("Aland Islands"): unknown field "postal_code_regex"
• ...and 1240 more warnings

Cross-Reference Validation

✅ 742 reference(s) verified

Duplicate Detection

• ⚠️ contributions/countries/countries.json: Record 1 ("Afghanistan") appears to be a duplicate of existing "Afghanistan" (id: 1, distance: 0.0km)
• ⚠️ contributions/countries/countries.json: Record 2 ("Aland Islands") appears to be a duplicate of existing "Aland Islands" (id: 2, distance: 0.0km)
• ⚠️ contributions/countries/countries.json: Record 3 ("Albania") appears to be a duplicate of existing "Albania" (id: 3, distance: 0.0km)
• ⚠️ contributions/countries/countries.json: Record 4 ("Algeria") appears to be a duplicate of existing "Algeria" (id: 4, distance: 0.0km)
• ⚠️ contributions/countries/countries.json: Record 5 ("American Samoa") appears to be a duplicate of existing "American Samoa" (id: 5, distance: 0.0km)
• ⚠️ contributions/countries/countries.json: Record 6 ("Andorra") appears to be a duplicate of existing "Andorra" (id: 6, distance: 0.0km)
• ⚠️ contributions/countries/countries.json: Record 7 ("Angola") appears to be a duplicate of existing "Angola" (id: 7, distance: 0.0km)
• ⚠️ contributions/countries/countries.json: Record 8 ("Anguilla") appears to be a duplicate of existing "Anguilla" (id: 8, distance: 0.0km)
• ⚠️ contributions/countries/countries.json: Record 9 ("Antarctica") appears to be a duplicate of existing "Antarctica" (id: 9, distance: 0.0km)
• ⚠️ contributions/countries/countries.json: Record 10 ("Antigua and Barbuda") appears to be a duplicate of existing "Antigua and Barbuda" (id: 10, distance: 0.0km)

Source URL Verification

✅ 2 source URL(s) accessible

---

❌ 1000 error(s), 1500 warning(s) | Status: Changes required

Please fix the errors above and push a new commit. Refer to our Contribution Guidelines for details.

[dr5hn]

Weekly data-quality review (2026-05-04)

Verdict: needs-discussion

Checks

• Schema: ✅ Change to contributions/countries/countries.json updates TW postal_code_format and postal_code_regex. No forbidden fields introduced. New contributions/postcodes/TW.json records correctly omit id, flag, created_at, updated_at. (CI errors are false positives — see note below.)
• FK integrity: ✅ All 371 TW postcode records resolve to valid TW states (per CI cross-reference pass).
• Coordinates: ✅ No out-of-bounds coordinates reported by CI.
• Wikidata: N/A.
• Naming convention: N/A.

Discussion item

postal_code_format value — the PR sets this to "###/#####/######". Taiwan postcodes don't include a literal / separator; the three forms (100, 10001, 100002) are just different-length numeric strings. Using / as a format string delimiter is non-standard (other entries use # for digit, @ for letter, ? for alphanumeric, and literal chars for separators that actually appear in the code). Consider either:

• Three separate values separated by a conventional indicator, e.g. "###|#####|######" if the repo adopts a pipe notation for alternatives, or
• Just the shortest form "###" with a note in the PR, since the regex covers all three generations.

The regex itself (^(\d{3}(?:\d{2,3})?)$) is correct and accepts all three generations.

Note on CI "needs-changes" label

Same false-positive issue as the other postcode PRs: validator checks all 250 existing countries in countries.json for forbidden fields that legitimately exist on pre-existing records.

🤖 Automated weekly review — Claude (sonnet-4-6).

---

Generated by Claude Code

[dr5hn]

Merging as-is. The postal_code_format value "###/#####/######" is non-standard (TW codes don't include a literal /). Will normalise the convention across all multi-format-tier countries in a follow-up PR — either pipe notation ###|#####|###### or just the shortest form ###. The regex ^(\d{3}(?:\d{2,3})?)$ correctly accepts all three generations either way.