#!/usr/bin/env bash
# OpenStreetMap protected areas outside the US and Canada → geo.osm_public_lands → public-lands-osm.pmtiles
# (docs/07 §5 and §12, roadmap 3.2b). Run through the geotools image:
#
#   tools/geodata/run public-land/osm                                            # Slovenia (dev default)
#   HG_OSM_EXTRACT=europe/france HG_OSM_COUNTRIES=FR tools/geodata/run public-land/osm
#   HG_OSM_PBF=/data/downloads/osm/planet.osm.pbf HG_OSM_COUNTRIES=ALL tools/geodata/run public-land/osm
#
# HG_OSM_COUNTRIES names the countries the file fully covers: only they are kept and replaced (a country extract
# also holds slivers of its neighbours, which must not wipe a full import of those countries).
#
# ODbL: the rows live in their own table and tileset, never mixed with other licences (CLAUDE.md). One public-land
# source per region: anything whose country is covered by PAD-US or CPCAD is dropped. Countries come from Natural
# Earth (public domain).
set -euo pipefail
HERE="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
source "$HERE/../lib/common.sh"

EXTRACT="${HG_OSM_EXTRACT:-europe/slovenia}"
PBF="${HG_OSM_PBF:-}"
COUNTRIES="${HG_OSM_COUNTRIES:-$([[ -z "${HG_OSM_EXTRACT:-}${PBF}" ]] && echo SI)}"
[[ -n "$COUNTRIES" ]] || die "set HG_OSM_COUNTRIES to the ISO codes the file fully covers (or ALL for the planet)"
countries_sql="null"
if [[ "$COUNTRIES" != "ALL" ]]; then
  for c in $COUNTRIES; do [[ "$c" =~ ^[A-Z]{2}$ ]] || die "bad country code: $c"; done
  list="$(printf "'%s'," $COUNTRIES)"
  countries_sql="array[${list%,}]"
fi
COUNTRIES_URL="${HG_NATURAL_EARTH_COUNTRIES:-https://raw.githubusercontent.com/nvkelso/natural-earth-vector/v5.1.2/geojson/ne_10m_admin_0_countries.geojson}"

if [[ -z "$PBF" ]]; then
  [[ "$EXTRACT" =~ ^[a-z0-9/-]+$ ]] || die "bad extract name: $EXTRACT"
  PBF="/data/downloads/osm/${EXTRACT//\//-}-latest.osm.pbf"
  SOURCE_URL="https://download.geofabrik.de/$EXTRACT-latest.osm.pbf"
  mkdir -p /data/downloads/osm
  # Geofabrik refreshes extracts daily; reuse a download from the last day.
  if [[ -z "$(find "$PBF" -mtime -1 2>/dev/null)" ]]; then
    log "Downloading $SOURCE_URL"
    curl -fsSL --retry 3 -A "HuntGrid geodata pipeline" -o "$PBF.part" "$SOURCE_URL"
    mv -f "$PBF.part" "$PBF"
  fi
else
  SOURCE_URL="file://$PBF"
fi
[[ -f "$PBF" ]] || die "no OSM file at $PBF"
VERSION="OpenStreetMap $(date -u -r "$PBF" +%Y-%m-%d)"

psql_run -c "create schema if not exists staging"
log "Countries (Natural Earth)"
ogr2ogr -f PostgreSQL "PG:$HG_DB_URL" "/vsicurl/$COUNTRIES_URL" -nln staging.ne_countries -overwrite \
  -lco GEOMETRY_NAME=geom -nlt PROMOTE_TO_MULTI -t_srs EPSG:4326 -select iso_a2_eh --config OGR_TRUNCATE YES

log "Protected areas from $(basename "$PBF")"
ogr2ogr -f PostgreSQL "PG:$HG_DB_URL" "$PBF" multipolygons -oo CONFIG_FILE="$HERE/osmconf.ini" \
  -where "boundary IN ('protected_area','national_park') OR leisure = 'nature_reserve'" \
  -nln staging.osm_protected -overwrite -lco GEOMETRY_NAME=geom -lco SPATIAL_INDEX=NONE -nlt PROMOTE_TO_MULTI \
  --config OSM_MAX_TMPFILE_SIZE 1024 --config OGR_TRUNCATE YES
fetched="$(psql_run -tA -c "select count(*) from staging.osm_protected")"
log "Found $fetched areas"
[[ "$fetched" -gt 0 ]] || die "no protected areas found; refusing to replace existing data"

log "Catalog layer public-lands-osm"
upsert_layer "$HERE/public-lands-osm.layer.json"

log "Normalising into geo.osm_public_lands"
psql_run <<SQL
drop table if exists staging.osm_run;
create table staging.osm_run (version text not null, source_url text not null, countries text[]);
insert into staging.osm_run values ('$VERSION', '$SOURCE_URL', $countries_sql);
SQL
psql_run -f "$HERE/osm.sql"
psql_run -c "drop table staging.osm_protected; drop table staging.osm_run"
psql_run -tA -c "select country_code, palette, count(*) from geo.osm_public_lands group by 1, 2 order by 1, 2" | sed 's/^/    /'

build_public_land_tiles public-lands-osm "source = 'OSM'" "© OpenStreetMap contributors (ODbL)" geo.osm_public_lands