Add governed Statbel edition probe
GeoIntel CI / docs-smoke (push) Canceled after 0s
GeoIntel CI / contract-smoke (push) Canceled after 0s

This commit is contained in:
Codex
2026-07-16 22:14:48 +02:00
parent 9059db3445
commit da2371d662
23 changed files with 805 additions and 28 deletions
@@ -24,6 +24,11 @@ from app.schemas.source_catalog import (
SourceCatalogProbeReport,
SourceCatalogProbeSummary,
)
from app.services.statbel_catalog_probe import (
StatbelCatalogError,
parse_statbel_population_catalog,
validate_statbel_catalog_url,
)
_GMD = "http://www.isotc211.org/2005/gmd"
@@ -44,6 +49,7 @@ _ALZ_SNAPSHOT = re.compile(
re.IGNORECASE,
)
_ALZ_EDITION = re.compile(r"^(20\d{2})-(?:definitive|v3)$", re.IGNORECASE)
_YEAR_EDITION = re.compile(r"^20\d{2}$")
class CatalogProbeFailure(RuntimeError):
@@ -159,12 +165,19 @@ def _with_capabilities_query(base_url: str, service_type: str) -> str:
return urlunsplit((parsed.scheme, parsed.netloc, parsed.path, query, ""))
def _bounded_fetch(url: str, settings: Settings, opener: Callable[..., Any] | None = None) -> FetchResult:
def _bounded_fetch(
url: str,
settings: Settings,
opener: Callable[..., Any] | None = None,
*,
max_response_mb: int | None = None,
accept: str = "application/xml,text/xml,text/html,application/json",
) -> FetchResult:
request = Request(
url,
headers={"Accept": "application/xml,text/xml,text/html,application/json", "User-Agent": "GeoIntel/0.1 source-catalog-probe"},
headers={"Accept": accept, "User-Agent": "GeoIntel/0.1 source-catalog-probe"},
)
max_bytes = settings.source_catalog_probe_max_response_mb * 1024 * 1024
max_bytes = (max_response_mb or settings.source_catalog_probe_max_response_mb) * 1024 * 1024
try:
with (opener or urlopen)(request, timeout=settings.source_catalog_probe_timeout_seconds) as response:
content_length = _header(response.headers, "Content-Length")
@@ -509,6 +522,56 @@ def _probe_alz_remote(
)
def _probe_statbel_remote(
contract: ProbeContract,
settings: Settings,
*,
opener: Callable[..., Any] | None,
now: datetime,
) -> RemoteProbe:
try:
catalog_url = validate_statbel_catalog_url(contract.endpoint_url)
response = _bounded_fetch(
catalog_url,
settings,
opener,
max_response_mb=settings.source_catalog_statbel_max_response_mb,
accept="text/turtle,application/x-turtle,application/octet-stream,text/plain",
)
validate_statbel_catalog_url(response.final_url)
content_type = response.content_type.lower()
if content_type and not any(token in content_type for token in ("turtle", "octet-stream", "text/plain")):
raise StatbelCatalogError(
"CATALOG_STATBEL_INVALID_CONTENT_TYPE",
"De officiële Statbel DCAT-catalogus heeft geen ondersteund Turtle-contenttype.",
)
release = parse_statbel_population_catalog(response.content)
except StatbelCatalogError as exc:
raise CatalogProbeFailure(exc.code, exc.message) from exc
layout = "nieuwe REDEGEO-sectorindeling" if release.current_distribution_variant == "new" else "actuele sectorindeling"
message = f"De officiële Statbel DCAT-catalogus bevestigt bevolkingseditie {release.version} met de {layout}."
if release.legacy_distribution_available:
message += " De oude 2025-indeling is alleen overgangsevidentie en wordt niet als actuele GeoIntel-editie gebruikt."
return RemoteProbe(
status="available",
reachable=True,
checked_at=now,
expected_layers=contract.expected_layers,
matched_layers=contract.expected_layers,
advertised_layer_count=release.distribution_count,
metadata_url=release.landing_page,
metadata_identifier=release.identifier,
remote_title=f"Bevolking per statistische sector {release.version} ({layout})",
remote_version=release.version,
remote_modified_at=release.catalog_modified_at,
capabilities_sha256=sha256(response.content).hexdigest(),
capabilities_etag=response.etag,
capabilities_last_modified_at=response.last_modified_at,
message=message,
)
def _probe_remote(
contract: ProbeContract,
settings: Settings,
@@ -519,6 +582,8 @@ def _probe_remote(
try:
if contract.service_type == "HTML":
return _probe_alz_remote(contract, settings, opener=opener, now=now)
if contract.service_type == "DCAT":
return _probe_statbel_remote(contract, settings, opener=opener, now=now)
capabilities = _bounded_fetch(_validate_capabilities_url(contract.endpoint_url), settings, opener)
_validate_capabilities_url(capabilities.final_url)
content_type = capabilities.content_type.lower()
@@ -635,6 +700,10 @@ def _latest_local_version(source_name: str, datasets: list[Dataset]) -> str | No
),
)
return latest.source_version
elif source_name == "statbel":
annual = [item for item in candidates if _YEAR_EDITION.fullmatch((item.source_version or "").strip())]
if annual:
return max(annual, key=lambda item: int((item.source_version or "0").strip())).source_version
if not candidates:
return None
latest = max(
@@ -660,6 +729,8 @@ def _normalized_version(source_name: str, version: str | None) -> str | None:
if source_name == _ALZ_SOURCE_NAME:
match = _ALZ_EDITION.fullmatch(value)
return f"{match.group(1)}-v3" if match else None
if source_name == "statbel" and _YEAR_EDITION.fullmatch(value):
return value
return None
@@ -711,6 +782,13 @@ class SourceCatalogProbeService:
endpoint_url=_with_capabilities_query(active_settings.orthophoto_wms_url, "WMS"),
expected_layers=(active_settings.orthophoto_wms_layer, "Vliegdagcontour"),
),
ProbeContract(
source_name="statbel",
display_name="Bevolking per statistische sector (Statbel)",
service_type="DCAT",
endpoint_url=active_settings.source_catalog_statbel_dcat_url,
expected_layers=("population_txt_current", "landing_page", "cc_by_4_0"),
),
ProbeContract(
source_name=_ALZ_SOURCE_NAME,
display_name="Landbouwgebruikspercelen (ALZ)",
@@ -789,8 +867,9 @@ class SourceCatalogProbeService:
summary=summary,
items=items,
limitations=[
"Deze expliciete controle leest alleen allowlisted WFS/WMS-capabilities, gekoppelde ISO 19139 metadata en de officiële ALZ-publicatiepagina.",
"Deze expliciete controle leest alleen allowlisted WFS/WMS-capabilities, gekoppelde ISO 19139 metadata, de officiële Statbel DCAT-catalogus en de officiële ALZ-publicatiepagina.",
"Er worden geen features, rasters of modelbestanden opgehaald en geen datasets aangemaakt of overschreven.",
"Statbel distributielinks worden alleen als release-evidentie gevalideerd; de oude en nieuwe 2025-sectorindeling blijven semantisch gescheiden.",
"Voor ALZ is alleen de nieuwste definitieve v3-editie vergelijkbaar; voorlopige v1/v2-snapshots zijn uitsluitend informatief.",
"Een versieverschil is controlesignaal, geen bewijs dat een lokale dataset onbruikbaar is en geen automatische importopdracht.",
],