fetch

#!/usr/bin/env bash
# Fetch administrative boundary shapefiles from 政府資料開放平臺 (data.gov.tw).
#
# The actual files live on tgos.tw and their names carry a release date that
# changes on every update ( e.g. 鄉(鎮、市、區)界線1140318.zip ), so we resolve the
# current download url from the dataset api instead of hardcoding it:
#
#   county  - https://data.gov.tw/dataset/7442  直轄市、縣市界線(TWD97經緯度)
#   town    - https://data.gov.tw/dataset/7441  鄉鎮市區界線(TWD97經緯度)
#   village - https://data.gov.tw/dataset/7438  村里界圖(TWD97經緯度)
set -e

api="https://data.gov.tw/api/v2/rest/dataset"

# resolve the SHP distribution url of a dataset. prints url, or nothing if not found.
resolve() {
  curl -sfL --max-time 60 "$api/$1" | python3 -c '
import sys, json, urllib.parse
r = json.load(sys.stdin).get("result") or {}
for d in r.get("distribution", []):
    if d.get("resourceFormat", "").upper() == "SHP":
        u = d["resourceDownloadUrl"]
        # the path may contain chinese characters / parentheses. percent-encode it.
        p = urllib.parse.urlsplit(u)
        print(urllib.parse.urlunsplit(p._replace(path=urllib.parse.quote(p.path))))
        break
'
}

get() {
  local lv=$1 dataset=$2 url
  echo "resolving $lv (dataset $dataset) ..."
  url=$(resolve "$dataset")
  if [ -z "$url" ]; then
    echo "  !! cannot resolve SHP url for dataset $dataset. see https://data.gov.tw/dataset/$dataset" >&2
    return 1
  fi
  echo "  url: $url"
  rm -rf "download/$lv"
  mkdir -p "download/$lv"
  echo "$url" > "download/$lv/source.txt"
  curl -fL --max-time 900 -o "download/$lv/shp.zip" "$url"
  # extract map files only: the zips also carry an xlsx whose filename is not valid
  # utf-8, which makes unzip fail on macos.
  ( cd "download/$lv" && unzip -o -q -j shp.zip '*.shp' '*.dbf' '*.prj' '*.shx' )
}

get county  7442
get town    7441
get village 7438
echo "done."