format tests for atom, docx, odt, and xlsx

formats/atom_format.py

+import xml.etree.ElementTree as ET
+
+
+def is_valid(resource, file):
+    """Check if the HTTP response is an ATOM feed."""
+
+    with open(file.name, "rb") as f:
+        try:
+            xml = ET.parse(f).getroot()
+
+            if xml.tag == "{http://www.w3.org/2005/Atom}feed":
+                return True
+            else:
+                resource["error"] = (
+                    "Root element is not {http://www.w3.org/2005/Atom}feed"
+                )
+                return False
+        except Exception as e:
+            resource["error"] = str(e)
+            return False

formats/docx_format.py

+import zipfile
+
+
+def is_valid(resource, file):
+    """Check if the content is a DOCX file."""
+
+    if not zipfile.is_zipfile(file.name):
+        resource["error"] = "Not a ZIP file."
+        return False
+
+    with zipfile.ZipFile(file.name, "r") as zip_ref:
+        zip_contents = zip_ref.namelist()
+
+        required_files = ["word/document.xml", "word/styles.xml"]
+
+        if not all(file in zip_contents for file in required_files):
+            resource["error"] = "That does not look like an DOCX file."
+            return False
+
+        return True

formats/odt_format.py

+import zipfile
+
+
+def is_valid(resource, file):
+    """Check if the content is a ODT file."""
+
+    if not zipfile.is_zipfile(file.name):
+        resource["error"] = "Not a ZIP file."
+        return False
+
+    with zipfile.ZipFile(file.name, "r") as zip_ref:
+        zip_contents = zip_ref.namelist()
+
+        required_files = ["mimetype", "content.xml", "meta.xml", "styles.xml"]
+
+        if not all(file in zip_contents for file in required_files):
+            resource["error"] = "That does not look like an ODT file."
+            return False
+
+        with zip_ref.open("mimetype") as mimetype_file:
+            mimetype_content = mimetype_file.read().decode("utf-8").strip()
+
+        if mimetype_content != "application/vnd.oasis.opendocument.text":
+            resource["error"] = f"Incorrect MIME type: {mimetype_content}"
+            return False
+
+        return True

formats/xlsx_format.py

+import zipfile
+
+
+def is_valid(resource, file):
+    """Check if the content is a XLSX file."""
+
+    if not zipfile.is_zipfile(file.name):
+        resource["error"] = "Not a ZIP file."
+        return False
+
+    with zipfile.ZipFile(file.name, "r") as zip_ref:
+        zip_contents = zip_ref.namelist()
+
+        required_files = ["xl/workbook.xml", "xl/styles.xml"]
+
+        if not all(file in zip_contents for file in required_files):
+            resource["error"] = "That does not look like an XLSX file."
+            return False
+
+        return True

tests/data/Atom_SH_Feldblockfinder_OpenGBD.xml

+<?xml version="1.0" encoding="UTF-8"?>
+<feed xmlns:georss="http://www.georss.org/georss" xmlns:inspire_dls="http://inspire.ec.europa.eu/schemas/inspire_dls/1.0" xmlns:lang="ger" xmlns="http://www.w3.org/2005/Atom">
+	<title>ATOMFeed SH Feldblockfinder Schleswig-Holstein (Downloaddienst)</title>
+	<subtitle>Downloaddienst zur Bereitstellung der Feldblöcke, Landschaftselemente, GLÖZ 2- und Nitratkulisse in Schleswig-Holstein.</subtitle>
+	<link href="http://sh-mis.schleswig-holstein.de/soapServices/CSWStartup?service=CSW&amp;version=2.0.2&amp;request=GetRecordById&amp;outputschema=http://www.isotc211.org/2005/gmd&amp;elementsetname=full&amp;ID=cc68aa82-d71b-42bb-b5ce-7b850486a842" rel="describedby" type="application/xml"/>
+	<link href="https://service.gdi-sh.de/SH_OpenGBD/feeds/Atom_SH_Feldblockfinder_OpenGBD/Atom_SH_Feldblockfinder_OpenGBD.xml" rel="self" type="application/atom+xml" hreflang="ger" title="Selbstreferenz"/>
+	<link href="https://service.gdi-sh.de/SH_OpenGBD/feeds/Atom_SH_Feldblockfinder_OpenGBD/Atom_SH_Feldblockfinder_OpenGBD_OSD.xml" rel="search" type="application/opensearchdescription+xml" title="Open Search Beschreibung ATOMFeed SH Feldblockfinder Schleswig-Holstein (Downloaddienst)"/>
+	<id>https://service.gdi-sh.de/SH_OpenGBD/feeds/Atom_SH_Feldblockfinder_OpenGBD/Atom_SH_Feldblockfinder_OpenGBD.xml</id>
+	<updated>2023-08-22T12:00:00+01:00</updated>
+	<author>
+		<name>Landesamt für Vermessung und Geoinformation Schleswig-Holstein (Servicestelle DANord)</name>
+		<email>DANord@LVermGeo.landsh.de</email>
+	</author>
+	<entry>
+		<title>Feldblöcke SH 2024</title>
+		<inspire_dls:spatial_dataset_identifier_code>21f67269-780f-4f3c-8f66-03dde27acfe7</inspire_dls:spatial_dataset_identifier_code>
+		<inspire_dls:spatial_dataset_identifier_namespace>http://registry.gdi.de.org/id/de.sh/</inspire_dls:spatial_dataset_identifier_namespace>
+		<link rel="describedby" href="http://sh-mis.schleswig-holstein.de/soapServices/CSWStartup?service=CSW&amp;version=2.0.2&amp;request=GetRecordById&amp;outputschema=http://www.isotc211.org/2005/gmd&amp;elementsetname=full&amp;ID=21f67269-780f-4f3c-8f66-03dde27acfe7" type="application/xml"/>
+		<link rel="alternate" href="https://service.gdi-sh.de/SH_OpenGBD/feeds/Atom_SH_Feldblockfinder_OpenGBD/Atom_SH_Feldblockfinder_OpenGBD_21f67269-780f-4f3c-8f66-03dde27acfe7.xml" type="application/atom+xml" hreflang="ger" title="Feldblöcke SH 2024"/>
+		<id>https://service.gdi-sh.de/SH_OpenGBD/feeds/Atom_SH_Feldblockfinder_OpenGBD/Atom_SH_Feldblockfinder_OpenGBD_21f67269-780f-4f3c-8f66-03dde27acfe7.xml</id>
+		<updated>2023-08-22T12:00:00+01:00</updated>
+		<summary>Feldblöcke in Schleswig-Holstein für den Stichtag 01.01.2024.</summary>
+		<georss:polygon>53.35 7.86 55.05 7.86 55.05 11.35 53.35 11.35 53.35 7.86</georss:polygon>
+		<category term="http://www.opengis.net/def/crs/EPSG/0/4258" label="ETRS89"/>
+	</entry>
+	<entry>
+		<title>Landschaftselemente SH 2024</title>
+		<inspire_dls:spatial_dataset_identifier_code>c4ddc5b7-036c-4670-b5ed-445d1aa20a2a</inspire_dls:spatial_dataset_identifier_code>
+		<inspire_dls:spatial_dataset_identifier_namespace>http://registry.gdi.de.org/id/de.sh/</inspire_dls:spatial_dataset_identifier_namespace>
+		<link rel="describedby" href="http://sh-mis.schleswig-holstein.de/soapServices/CSWStartup?service=CSW&amp;version=2.0.2&amp;request=GetRecordById&amp;outputschema=http://www.isotc211.org/2005/gmd&amp;elementsetname=full&amp;ID=c4ddc5b7-036c-4670-b5ed-445d1aa20a2a" type="application/xml"/>
+		<link rel="alternate" href="https://service.gdi-sh.de/SH_OpenGBD/feeds/Atom_SH_Feldblockfinder_OpenGBD/Atom_SH_Feldblockfinder_OpenGBD_c4ddc5b7-036c-4670-b5ed-445d1aa20a2a.xml" type="application/atom+xml" hreflang="ger" title="Landschaftselemente SH 2024"/>
+		<id>https://service.gdi-sh.de/SH_OpenGBD/feeds/Atom_SH_Feldblockfinder_OpenGBD/Atom_SH_Feldblockfinder_OpenGBD_c4ddc5b7-036c-4670-b5ed-445d1aa20a2a.xml</id>
+		<updated>2023-08-22T12:00:00+01:00</updated>
+		<summary>Landschaftselemente in Schleswig Holstein zum Stichtag 01.01.2024.</summary>
+		<georss:polygon>53.35 7.86 55.05 7.86 55.05 11.35 53.35 11.35 53.35 7.86</georss:polygon>
+		<category term="http://www.opengis.net/def/crs/EPSG/0/4258" label="ETRS89"/>
+	</entry>
+	<entry>
+		<title>Feuchtgebiete und Moore ab 2 ha (GLÖZ 2) SH</title>
+		<inspire_dls:spatial_dataset_identifier_code>176bcfa1-7af2-4bd5-b4e3-2a6212b8fcd2</inspire_dls:spatial_dataset_identifier_code>
+		<inspire_dls:spatial_dataset_identifier_namespace>http://registry.gdi.de.org/id/de.sh/</inspire_dls:spatial_dataset_identifier_namespace>
+		<link rel="describedby" href="http://sh-mis.schleswig-holstein.de/soapServices/CSWStartup?service=CSW&amp;version=2.0.2&amp;request=GetRecordById&amp;outputschema=http://www.isotc211.org/2005/gmd&amp;elementsetname=full&amp;ID=176bcfa1-7af2-4bd5-b4e3-2a6212b8fcd2" type="application/xml"/>
+		<link rel="alternate" href="https://service.gdi-sh.de/SH_OpenGBD/feeds/Atom_SH_Feldblockfinder_OpenGBD/Atom_SH_Feldblockfinder_OpenGBD_176bcfa1-7af2-4bd5-b4e3-2a6212b8fcd2.xml" type="application/atom+xml" hreflang="ger" title="Feuchtgebiete und Moore ab 2 ha (GLÖZ 2) SH"/>
+		<id>https://service.gdi-sh.de/SH_OpenGBD/feeds/Atom_SH_Feldblockfinder_OpenGBD/Atom_SH_Feldblockfinder_OpenGBD_176bcfa1-7af2-4bd5-b4e3-2a6212b8fcd2.xml</id>
+		<updated>2023-08-18T12:00:00+01:00</updated>
+		<summary>Feuchtgebiete und Moore ab 2 ha (GLÖZ 2) in Schleswig-Holstein.</summary>
+		<georss:polygon>11.35 7.86 55.05 7.86 55.05 53.35 11.35 53.35 11.35 7.86</georss:polygon>
+		<category term="http://www.opengis.net/def/crs/EPSG/0/4258" label="ETRS89"/>
+	</entry>
+	<entry>
+		<title>Nitrat-belastete Gebiete nach LDüV in SH</title>
+		<inspire_dls:spatial_dataset_identifier_code>694bd2d9-3bea-4818-961c-25024acc0588</inspire_dls:spatial_dataset_identifier_code>
+		<inspire_dls:spatial_dataset_identifier_namespace>http://registry.gdi.de.org/id/de.sh/</inspire_dls:spatial_dataset_identifier_namespace>
+		<link rel="describedby" href="http://sh-mis.schleswig-holstein.de/soapServices/CSWStartup?service=CSW&amp;version=2.0.2&amp;request=GetRecordById&amp;outputschema=http://www.isotc211.org/2005/gmd&amp;elementsetname=full&amp;ID=694bd2d9-3bea-4818-961c-25024acc0588" type="application/xml"/>
+		<link rel="alternate" href="https://service.gdi-sh.de/SH_OpenGBD/feeds/Atom_SH_Feldblockfinder_OpenGBD/Atom_SH_Feldblockfinder_OpenGBD_694bd2d9-3bea-4818-961c-25024acc0588.xml" type="application/atom+xml" hreflang="ger" title="Nitrat-belastete Gebiete nach LDüV in SH"/>
+		<id>https://service.gdi-sh.de/SH_OpenGBD/feeds/Atom_SH_Feldblockfinder_OpenGBD/Atom_SH_Feldblockfinder_OpenGBD_694bd2d9-3bea-4818-961c-25024acc0588.xml</id>
+		<updated>2023-08-18T12:00:00+01:00</updated>
+		<summary>Nitrat-belastete Gebiete für das Jahr 2022 nach LDüV in SH.</summary>
+		<georss:polygon>11.35 7.86 55.05 7.86 55.05 53.35 11.35 53.35 11.35 7.86</georss:polygon>
+		<category term="http://www.opengis.net/def/crs/EPSG/0/4258" label="ETRS89"/>
+	</entry>
+</feed>

tests/test_atom_format.py

+import unittest
+from formats.atom_format import is_valid
+
+
+class TestAtomFormat(unittest.TestCase):
+    def test_is_valid__valid(self):
+        resource = {}
+        with open("tests/data/Atom_SH_Feldblockfinder_OpenGBD.xml", "r") as file:
+            self.assertTrue(is_valid(resource, file))
+            self.assertIsNone(resource.get("error"))
+
+    def test_is_valid__other_xml(self):
+        resource = {}
+        with open("tests/data/correct.xml", "r") as file:
+            self.assertFalse(is_valid(resource, file))
+            self.assertIsNotNone(resource.get("error"))
+
+    def test_is_valid__invalid_xml(self):
+        resource = {}
+        with open("tests/data/incorrect.xml", "r") as file:
+            self.assertFalse(is_valid(resource, file))
+            self.assertIsNotNone(resource.get("error"))
+
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_docx_format.py

+import unittest
+from formats.docx_format import is_valid
+
+
+class TestDocxFormat(unittest.TestCase):
+    def test_is_valid__valid(self):
+        resource = {}
+        with open("tests/data/valid.docx", "r") as file:
+            self.assertTrue(is_valid(resource, file))
+            self.assertIsNone(resource.get("error"))
+
+    def test_is_valid__invalid_no_zip(self):
+        resource = {}
+        with open("tests/data/correct.json", "r") as file:
+            self.assertFalse(is_valid(resource, file))
+            self.assertIsNotNone(resource.get("error"))
+
+    def test_is_valid__invalid_no_excel(self):
+        resource = {}
+        with open("tests/data/valid.xlsx", "r") as file:
+            self.assertFalse(is_valid(resource, file))
+            self.assertIsNotNone(resource.get("error"))
+
+    def test_is_valid__invalid_zip(self):
+        resource = {}
+        with open("tests/data/valid.odt", "r") as file:
+            self.assertFalse(is_valid(resource, file))
+            self.assertIsNotNone(resource.get("error"))
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_odt_format.py

+import unittest
+from formats.odt_format import is_valid
+
+
+class TestOdsFormat(unittest.TestCase):
+    def test_is_valid__valid(self):
+        resource = {}
+        with open("tests/data/valid.odt", "r") as file:
+            self.assertTrue(is_valid(resource, file))
+            self.assertIsNone(resource.get("error"))
+
+    def test_is_valid__invalid_no_zip(self):
+        resource = {}
+        with open("tests/data/correct.json", "r") as file:
+            self.assertFalse(is_valid(resource, file))
+            self.assertIsNotNone(resource.get("error"))
+
+    def test_is_valid__invalid_no_odt(self):
+        resource = {}
+        with open("tests/data/valid.ods", "r") as file:
+            self.assertFalse(is_valid(resource, file))
+            self.assertIsNotNone(resource.get("error"))
+            self.assertEqual(
+                "Incorrect MIME type: application/vnd.oasis.opendocument.spreadsheet",
+                resource["error"],
+            )
+
+    def test_is_valid__invalid_zip(self):
+        resource = {}
+        with open("tests/data/valid.docx", "r") as file:
+            self.assertFalse(is_valid(resource, file))
+            self.assertIsNotNone(resource.get("error"))
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_xlsx_format.py

+import unittest
+from formats.xlsx_format import is_valid
+
+
+class TestXlsxFormat(unittest.TestCase):
+    def test_is_valid__valid(self):
+        resource = {}
+        with open("tests/data/valid.xlsx", "r") as file:
+            self.assertTrue(is_valid(resource, file))
+            self.assertIsNone(resource.get("error"))
+
+    def test_is_valid__invalid_no_zip(self):
+        resource = {}
+        with open("tests/data/correct.json", "r") as file:
+            self.assertFalse(is_valid(resource, file))
+            self.assertIsNotNone(resource.get("error"))
+
+    def test_is_valid__invalid_no_excel(self):
+        resource = {}
+        with open("tests/data/valid.docx", "r") as file:
+            self.assertFalse(is_valid(resource, file))
+            self.assertIsNotNone(resource.get("error"))
+
+    def test_is_valid__invalid_zip(self):
+        resource = {}
+        with open("tests/data/valid.ods", "r") as file:
+            self.assertFalse(is_valid(resource, file))
+            self.assertIsNotNone(resource.get("error"))
+
+
+if __name__ == "__main__":
+    unittest.main()