Refactor

georss_download.py

@@ -5,6 +5,19 @@ from urllib.parse import urljoin
 from citygml_download import TMP_DIR, Bundesland, download_file
 from zipfile_download import ZipFile
 
+NAMESPACE = {"atom": "http://www.w3.org/2005/Atom"}
+
+
+def get_updated_date(node: ET.Element) -> datetime | None:
+    updated_node = node.find("atom:updated", NAMESPACE)
+    if updated_node is None:
+        return None
+    updated_date = updated_node.text
+    if updated_date is None:
+        return None
+    source_date = datetime.strptime(updated_date, "%Y-%m-%dT%H:%M:%SZ")
+    return source_date
+
 
 def parse_atom_georss(bundesland: Bundesland) -> list[ZipFile]:
     atom_path = TMP_DIR / f"{bundesland}.atom"
@@ -14,25 +27,21 @@ def parse_atom_georss(bundesland: Bundesland) -> list[ZipFile]:
     root = tree.getroot()
     zip_files = []
 
-    ns = {"atom": "http://www.w3.org/2005/Atom"}
+    global_updated_date = get_updated_date(root)
 
-    updated_node = root.find("atom:updated", ns)
-    if updated_node is None:
-        raise ValueError("No date has been found")
-    updated_date = updated_node.text
-    if updated_date is None:
-        raise ValueError("No date has been found")
-    source_date = datetime.strptime(updated_date, "%Y-%m-%dT%H:%M:%SZ")
-    for file_elem in root.findall("atom:entry/atom:link", ns):
-        href = file_elem.attrib["href"]
+    for link_node in root.findall("atom:entry/atom:link", NAMESPACE):
+        href = link_node.attrib["href"]
         href = urljoin(bundesland.source, href)
-        # TODO: Check if source_date is specified for each ZipFile
+        gml_updated_date = get_updated_date(link_node)
+        updated_date = gml_updated_date or global_updated_date
+        if updated_date is None:
+            raise ValueError(f"Unknown date for {href}")
         if href.endswith(".zip"):
             zip_files.append(
                 ZipFile(
                     url=href,
                     bundesland=bundesland,
-                    source_date=source_date,
+                    source_date=updated_date,
                 )
             )
     return zip_files