From 48b1ddf47a3fca24b2d8d93e29f0e064f11eddfe Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Thibault=20Cl=C3=A9rice?= <leponteineptique@gmail.com>
Date: Sat, 24 Aug 2024 17:56:39 +0200
Subject: [PATCH] Parents are parenting

---
 dapitains/metadata/xml_parser.py         | 11 +++++------
 tests/catalog/example-sub-collection.xml |  1 +
 tests/test_catalog.py                    |  1 +
 3 files changed, 7 insertions(+), 6 deletions(-)

diff --git a/dapitains/metadata/xml_parser.py b/dapitains/metadata/xml_parser.py
index 946312e..c33c8a4 100644
--- a/dapitains/metadata/xml_parser.py
+++ b/dapitains/metadata/xml_parser.py
@@ -15,7 +15,7 @@ class Catalog:
     objects: Dict[str, Collection] = field(default_factory=dict)
 
 
-def parse_metadata(xml: ET.Element) -> Dict[str, Any]:
+def parse_metadata(xml: ET.Element) -> Tuple[Dict[str, Any], List[str]]:
     """ Parse Metadata
 
     :param xml: Collection/Resource tag
@@ -50,9 +50,8 @@ def parse_metadata(xml: ET.Element) -> Dict[str, Any]:
     parents = []
     for node in xml.xpath("./parent/text()"):
         parents.append(str(node))
-    obj["parents"] = parents
 
-    return obj
+    return obj, parents
 
 
 def parse_collection(xml: ET.Element, basedir: str, tree: Catalog) -> Collection:
@@ -62,8 +61,10 @@ def parse_collection(xml: ET.Element, basedir: str, tree: Catalog) -> Collection
     :param basedir: Directory used to resolve filepath, that are relative to the main object
     :param tree: Catalog that is updated with objects.
     """
-    obj = parse_metadata(xml)
+    obj, parents = parse_metadata(xml)
     obj = Collection(**obj, resource=xml.tag == "resource")
+    for parent in parents:
+        tree.relationships.append((parent, obj.identifier))
     tree.objects[obj.identifier] = obj
     if xml.attrib.get("filepath") and obj.resource:
         obj.filepath = os.path.normpath(os.path.join(basedir, xml.attrib["filepath"]))
@@ -74,8 +75,6 @@ def parse_collection(xml: ET.Element, basedir: str, tree: Catalog) -> Collection
         else:
             _, child = ingest_catalog(os.path.join(basedir, member.attrib["filepath"]), tree)
             tree.relationships.append((obj.identifier, child.identifier))
-        for parent in child.parents:
-            tree.relationships.append((parent, child.identifier))
     return obj
 
 
diff --git a/tests/catalog/example-sub-collection.xml b/tests/catalog/example-sub-collection.xml
index e311d53..f5798c4 100644
--- a/tests/catalog/example-sub-collection.xml
+++ b/tests/catalog/example-sub-collection.xml
@@ -9,6 +9,7 @@
         <resource identifier="https://example.org/resource1" filepath="../tei/multiple_tree.xml">
             <title>Historical Document</title>
             <description>A document about historical events.</description>
+            <parent>https://foo.bar/default</parent>
             <dublinCore>
                 <subject xmlns="http://purl.org/dc/terms/">World War II</subject>
                 <language xmlns="http://purl.org/dc/terms/">en</language>
diff --git a/tests/test_catalog.py b/tests/test_catalog.py
index 9776303..c376624 100644
--- a/tests/test_catalog.py
+++ b/tests/test_catalog.py
@@ -60,5 +60,6 @@ def test_ingestion():
     assert sorted(tree.relationships) == [
         ('https://example.org/collection1', 'https://example.org/resource1'),
         ('https://foo.bar/default', 'https://example.org/collection1'),
+        ('https://foo.bar/default', 'https://example.org/resource1',),
         ('https://foo.bar/default', 'https://foo.bar/text')
     ]