Use metalink

11_rheinland_pfalz.py

@@ -3,12 +3,9 @@
 # ©GeoBasis-DE / LVermGeoRP <Jahr des Datenbezugs>, dl-de/by-2-0, www.lvermgeo.rlp.de [Daten bearbeitet]
 # {"id":"dl-by-de/2.0","name":"Datenlizenz Deutschland Namensnennung 2.0","url":"https://www.govdata.de/dl-de/by-2-0","quelle":"© GeoBasis-DE / LVermGeoRP<Jahr des Datenbezugs>, dl-de/by-2-0, www.lvermgeo.rlp.de [Daten bearbeitet]"}
 
-import random
-import xml.etree.ElementTree as ET
-from pathlib import Path
-
 # Metalink https://geobasis-rlp.de/data/geb3dlo/current/meta4/geb3dlo_gml_07.meta4
-from citygml_download import TMP_DIR, Bundesland, CityGMLWithHash, download_all_files, download_file
+from citygml_download import Bundesland, download_all_files
+from metalink_download import download_metalink, parse_metalink
 
 # BUG: There are missing certificates on the server :-/
 # https://serverfault.com/questions/1118578/ubuntu-cannot-verify-sectigo-certificate
@@ -18,56 +15,6 @@ RHEINLAND_PFALZ = Bundesland(
     info="https://lvermgeo.rlp.de/produktinformationen/geotopografie/3d-geodaten/3d-gebaeudemodell",
 )
 
-# TODO: DRY WITH BAYERN!
-def download_metalink(metalink_url: str, tmp_path: Path = TMP_DIR) -> Path:
-    """Download metalink file to temporary directory."""
-    print(f"Downloading {metalink_url}")
-    basename = metalink_url.split("/")[-1] or "metalink.xml"
-    metalink_path = tmp_path / basename
-    download_file(metalink_url, metalink_path)
-    return metalink_path
-
-
-def parse_metalink(metalink_path: Path, bundesland: Bundesland) -> list[CityGMLWithHash]:
-    """Parse metalink XML file and return CityGMLWithHash objects."""
-    print(f"Parsing metalink file: {metalink_path}")
-    tree = ET.parse(metalink_path)
-    root = tree.getroot()
-
-    # Handle namespace
-    ns = {"ml": "urn:ietf:params:xml:ns:metalink"}
-
-    files = []
-    for file_elem in root.findall("ml:file", ns):
-        filename = file_elem.get("name")
-        if filename is None:
-            raise ValueError(f"Not enough information from {file_elem}")
-        hash_node = file_elem.find('ml:hash[@type="sha-256"]', ns)
-
-        if hash_node is None or hash_node.text is None:
-            raise ValueError(f"{filename} has no sha256")
-
-        urls = [url.text for url in file_elem.findall("ml:url", ns) if url.text]
-
-        if len(urls) == 0:
-            raise ValueError(f"No URL for {file_elem}")
-        else:
-            # Randomize URL order to distribute load across servers
-            random.shuffle(urls)
-            url = urls[0]
-
-        files.append(
-            CityGMLWithHash(
-                url=url,
-                coordinate_reference_system="EPSG:25832",
-                bundesland=bundesland,
-                expected_sha256=hash_node.text,
-            )
-        )
-
-    return files
-
-
 if __name__ == "__main__":
 
     metalink_path = download_metalink(RHEINLAND_PFALZ.source)