fix: ignore IRIs in stashing algorithm (DEV-2885) (#603)

src/dsp_tools/analyse_xml_data/construct_and_analyze_graph.py

@@ -1,6 +1,6 @@
 from __future__ import annotations
 
-from typing import Any
+from typing import Any, cast
 
 import regex
 import rustworkx as rx
@@ -53,11 +53,12 @@ def _create_info_from_xml_for_graph_from_one_resource(
 def _create_resptr_link_objects(subject_id: str, resptr_prop: etree._Element) -> list[ResptrLink]:
     resptr_links = []
     for resptr in resptr_prop.getchildren():
-        if r_text := resptr.text:
-            instance = ResptrLink(subject_id, r_text)
+        resptr.text = cast(str, resptr.text)
+        if not regex.search(r"https?://rdfh.ch/[a-fA-F0-9]{4}/\w{22}", resptr.text):
+            link_object = ResptrLink(subject_id, resptr.text)
             # this UUID is so that the links that were stashed can be identified in the XML data file
-            resptr.attrib["stashUUID"] = instance.link_uuid
-            resptr_links.append(instance)
+            resptr.attrib["stashUUID"] = link_object.link_uuid
+            resptr_links.append(link_object)
     return resptr_links
 
 
@@ -79,9 +80,8 @@ def _extract_ids_from_one_text_value(text: etree._Element) -> set[str]:
     all_links = set()
     for ele in text.iterdescendants():
         if href := ele.attrib.get("href"):
-            searched = regex.search(r"IRI:(.*):IRI", href)
-            if searched:
-                all_links.add(searched.group(1))
+            if internal_id := regex.search(r"IRI:(.*):IRI", href):
+                all_links.add(internal_id.group(1))
     return all_links
 
 

test/unittests/test_analyse_xml_data/test_construct_and_analyze_graph.py

@@ -117,8 +117,8 @@ def test_extract_id_one_text_with_several_id() -> None:
     test_ele = etree.fromstring(
         """<text permissions="prop-default" encoding="xml">
             <a class="salsah-link" href="IRI:res_A_11:IRI">res_A_11</a>
-            <a class="salsah-link" href="IRI:res_B_11:IRI">res_A_11</a>
-            <a class="salsah-link" href="IRI:res_B_11:IRI">res_A_11</a>
+            <a class="salsah-link" href="IRI:res_B_11:IRI">res_B_11</a>
+            <a class="salsah-link" href="IRI:res_B_11:IRI">res_B_11</a>
         </text>"""
     )
     res = _extract_ids_from_one_text_value(test_ele)
@@ -141,6 +141,25 @@ def test_extract_ids_from_text_prop_with_several_text_links() -> None:
     assert unordered(res_ids) == [{"res_A_18"}, {"res_B_18"}]
 
 
+def test_extract_ids_from_text_prop_with_iris_and_ids() -> None:
+    test_ele = etree.fromstring(
+        """<text-prop name=":hasRichtext">
+            <text permissions="prop-default" encoding="xml">
+                <a class="salsah-link" href="http://rdfh.ch/4123/vEpjk7zAQBC2j3pvTGSxcw">foo</a>
+            </text>
+            <text permissions="prop-default" encoding="xml">
+                <a class="salsah-link" href="IRI:res_B_18:IRI">res_B_18</a>
+            </text>
+        </text-prop>"""
+    )
+    res = _create_text_link_objects("foo", test_ele)
+    assert len(res) == 1
+    assert res[0].target_ids == {"res_B_18"}
+    children = list(test_ele.iterchildren())
+    assert not children[0].attrib.get("stashUUID")
+    assert children[1].attrib.get("stashUUID")
+
+
 def test_create_class_instance_resptr_link_one_link() -> None:
     test_ele = etree.fromstring(
         """<resptr-prop xmlns="https://dasch.swiss/schema" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" 
@@ -167,6 +186,24 @@ def test_create_class_instance_resptr_link_several() -> None:
     assert res[2].target_id == "res_C_13"
 
 
+def test_create_class_instance_resptr_link_with_iris() -> None:
+    test_ele = etree.fromstring(
+        """<resptr-prop name=":hasResource1">
+            <resptr permissions="prop-default">res_A_13</resptr>
+            <resptr permissions="prop-default">res_B_13</resptr>
+            <resptr permissions="prop-default">http://rdfh.ch/4123/vEpjk7zAQBC2j3pvTGSxcw</resptr>
+        </resptr-prop>"""
+    )
+    res = _create_resptr_link_objects("foo", test_ele)
+    assert len(res) == 2
+    assert res[0].target_id == "res_A_13"
+    assert res[1].target_id == "res_B_13"
+    children = list(test_ele.iterchildren())
+    assert children[0].attrib.get("stashUUID")
+    assert children[1].attrib.get("stashUUID")
+    assert not children[2].attrib.get("stashUUID")
+
+
 def test_create_info_from_xml_for_graph_check_UUID_in_root() -> None:
     root = etree.fromstring(
         """<knora shortcode="0700" default-ontology="simcir">