From 5b5d942d95dddd56b3bacb465c69098913688d1f Mon Sep 17 00:00:00 2001 From: Trygve Hemminghytt <274629477+trygvehhh@users.noreply.github.com> Date: Fri, 16 Oct 2026 22:23:02 +0200 Subject: [PATCH] fix(data): make the Norway datacenter script portable and re-runnable Read and write files as UTF-8 with LF line endings so the script and --verify work on Windows, where the default codec is cp1252 (thanks to quanticsoul4772 for the report). Geofabrik keeps only its newest dated extract, so --extract now defaults to the current one and accepts a dated name for --verify while it is listed; the dataset README says so. Downloads time out instead of hanging. --- CHANGELOG.md | 1 + scripts/build-datacenters-norway.py | 52 +++++++++++++++-------- src/data/local_data/datacenters/README.md | 8 +++- 3 files changed, 42 insertions(+), 19 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index 20a9da4e6..067ca16b6 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -184,6 +184,7 @@ `scripts/build-datacenters-norway.py` reproduces or `--verify`s the merge, touching only features inside Geofabrik's Norway boundary, and the dataset README records the extract, checksum and filter. + - Add a keyless Air quality layer to the Weather group, with Environment and Climate Change Canada's AQHI as the first registered provider. Readings carry `{ provider, scale, value, band }` so another network can be added without diff --git a/scripts/build-datacenters-norway.py b/scripts/build-datacenters-norway.py index 01a891a8b..687395f7b 100644 --- a/scripts/build-datacenters-norway.py +++ b/scripts/build-datacenters-norway.py @@ -1,10 +1,13 @@ #!/usr/bin/env python3 """Maintainer-only Norway refresh of the bundled datacenter extract (osmium-tool). -Downloads a pinned Geofabrik Norway extract, verifies its MD5, filters +Downloads a dated Geofabrik Norway extract, verifies its MD5, filters telecom=data_center / building=data_center with osmium, and merges the result into datacenters.geojsonl: features inside the Geofabrik Norway boundary are added, updated or removed; every other line is kept byte-for-byte. + +Geofabrik keeps only its newest dated extract, so --extract defaults to the +current one and a recorded extract can be re-verified only while it is listed. """ import argparse import hashlib @@ -16,17 +19,28 @@ import subprocess import urllib.request ROOT = Path(__file__).resolve().parent.parent -EXTRACT = 'norway-261005' GEOFABRIK = 'https://download.geofabrik.de/europe/' DATASET = ROOT / 'src/data/local_data/datacenters/datacenters.geojsonl' CONTACT_TAGS = {'email', 'phone', 'fax', 'mobile', 'whatsapp', 'contact:email', 'contact:phone', 'contact:fax', 'contact:mobile', 'contact:whatsapp'} SAME_GEOMETRY_METERS = 0.05 +TIMEOUT_SECONDS = 60 + + +def latest_extract(): + """Resolve norway-latest to the dated name Geofabrik redirects it to.""" + request = urllib.request.Request(GEOFABRIK + 'norway-latest.osm.pbf', method='HEAD') + with urllib.request.urlopen(request, timeout=TIMEOUT_SECONDS) as response: + name = response.geturl().rsplit('/', 1)[-1] + if not name.startswith('norway-') or not name.endswith('.osm.pbf') or name == 'norway-latest.osm.pbf': + raise RuntimeError('Unexpected Geofabrik latest target: %s' % name) + return name[:-len('.osm.pbf')] def fetch(url, target): if not target.exists(): - with urllib.request.urlopen(url) as response, open(str(target) + '.part', 'wb') as out: + with urllib.request.urlopen(url, timeout=TIMEOUT_SECONDS) as response, \ + open(str(target) + '.part', 'wb') as out: shutil.copyfileobj(response, out) Path(str(target) + '.part').rename(target) return target @@ -43,7 +57,7 @@ def md5(path): def read_poly(path): """Geofabrik .poly -> list of (ring, is_hole).""" rings, ring, hole = [], None, False - for raw in path.read_text().splitlines()[1:]: + for raw in path.read_text(encoding='utf-8').splitlines()[1:]: line = raw.strip() if ring is None: if line and line != 'END': @@ -80,15 +94,15 @@ def osm_key(feature): return ('n' if feature['geometry']['type'] == 'Point' else 'w') + str(osm_id) -def export(osmium, pbf, cache): - filtered = cache / (EXTRACT + '-datacenters.osm.pbf') - exported = cache / (EXTRACT + '-datacenters.geojsonseq') +def export(osmium, pbf, cache, extract): + filtered = cache / (extract + '-datacenters.osm.pbf') + exported = cache / (extract + '-datacenters.geojsonseq') subprocess.run([osmium, 'tags-filter', '-O', '-o', str(filtered), str(pbf), 'nwr/telecom=data_center', 'nwr/building=data_center'], check=True) subprocess.run([osmium, 'export', '-O', '-f', 'geojsonseq', '--add-unique-id=type_id', '-o', str(exported), str(filtered)], check=True) fresh = {} - for line in exported.read_text().splitlines(): + for line in exported.read_text(encoding='utf-8').splitlines(): line = line.lstrip('\x1e').strip() if not line: continue @@ -169,26 +183,30 @@ def main(): parser = argparse.ArgumentParser(description=__doc__) parser.add_argument('--cache', type=Path, required=True) parser.add_argument('--osmium', default=shutil.which('osmium') or 'osmium') + parser.add_argument('--extract', default='latest', + help="dated Geofabrik name such as norway-261005, or 'latest'") parser.add_argument('--verify', action='store_true') args = parser.parse_args() args.cache.mkdir(parents=True, exist_ok=True) - pbf = fetch(GEOFABRIK + EXTRACT + '.osm.pbf', args.cache / (EXTRACT + '.osm.pbf')) - expected = fetch(GEOFABRIK + EXTRACT + '.osm.pbf.md5', - args.cache / (EXTRACT + '.osm.pbf.md5')).read_text().split()[0] + extract = latest_extract() if args.extract == 'latest' else args.extract + pbf = fetch(GEOFABRIK + extract + '.osm.pbf', args.cache / (extract + '.osm.pbf')) + expected = fetch(GEOFABRIK + extract + '.osm.pbf.md5', + args.cache / (extract + '.osm.pbf.md5')).read_text(encoding='utf-8').split()[0] actual = md5(pbf) if actual != expected: raise RuntimeError('MD5 mismatch for %s: %s != %s' % (pbf.name, actual, expected)) rings = read_poly(fetch(GEOFABRIK + 'norway.poly', args.cache / 'norway.poly')) - fresh = export(args.osmium, pbf, args.cache) - current = [line for line in DATASET.read_text().splitlines() if line.strip()] + fresh = export(args.osmium, pbf, args.cache, extract) + current = [line for line in DATASET.read_text(encoding='utf-8').splitlines() if line.strip()] merged, report = merge(current, fresh, rings) payload = '\n'.join(merged) + '\n' if args.verify: - if payload != DATASET.read_text(): - raise SystemExit('datacenters.geojsonl is not the %s merge' % EXTRACT) + if payload != DATASET.read_text(encoding='utf-8'): + raise SystemExit('datacenters.geojsonl is not the %s merge' % extract) else: - DATASET.write_text(payload) - print(json.dumps({'extract': EXTRACT, 'md5': actual, 'norway': len(fresh), + with open(DATASET, 'w', encoding='utf-8', newline='\n') as handle: + handle.write(payload) + print(json.dumps({'extract': extract, 'md5': actual, 'norway': len(fresh), 'features': len(merged), **report})) diff --git a/src/data/local_data/datacenters/README.md b/src/data/local_data/datacenters/README.md index ef0720ae4..8b803dcc7 100644 --- a/src/data/local_data/datacenters/README.md +++ b/src/data/local_data/datacenters/README.md @@ -28,5 +28,9 @@ reflects OSM as of the extract below, everything else the original snapshot. - Filter: `osmium tags-filter nwr/telecom=data_center nwr/building=data_center` - Contact tags are dropped as above; unchanged features keep their original line, so geometry is only rewritten when tags or vertices actually changed. -- Rebuild or check: `python3 scripts/build-datacenters-norway.py --cache ` - (add `--verify` to check without writing; requires `osmium-tool`) +- Refresh: `python3 scripts/build-datacenters-norway.py --cache ` uses + Geofabrik's current Norway extract and prints the dated name it used; record + that name, its OSM timestamp and MD5 here. Requires `osmium-tool`. +- Check: add `--extract --verify` to compare without writing. + Geofabrik keeps only its newest dated extract, so a recorded extract can be + re-verified only while it is still listed; the MD5 above identifies it.