refactor(xmlupload): optimize stash links (DEV-2847)

src/dsp_tools/analyse_xml_data/models.py

@@ -1,17 +1,24 @@
 from __future__ import annotations
 
+import uuid
 from dataclasses import dataclass, field
 
 
-@dataclass
+@dataclass(frozen=True)
 class ResptrLink:
     """This class represents a link between two resources."""
 
     subject_id: str
     object_id: str
+    link_uuid: str = field(default_factory=lambda: str(uuid.uuid4()))
 
+    @property
+    def cost_links(self) -> float:
+        """The cost of this outgoing is consistently 1"""
+        return 1
 
-@dataclass
+
+@dataclass(frozen=True)
 class XMLLink:
     """
     This class represents a link between a resource and an XML text
@@ -20,6 +27,7 @@ class XMLLink:
 
     subject_id: str
     object_link_ids: set[str]
+    link_uuid: str = field(default_factory=lambda: str(uuid.uuid4()))
 
     @property
     def cost_links(self) -> float:
@@ -28,13 +36,13 @@ def cost_links(self) -> float:
 
 
 @dataclass
-class UploadResource:
+class ResourceStashInfo:
     """
     Holds information about a resource that can be uploaded to the DSP.
 
     May hold information about the links that need to be stashed from this resource before it can be uploaded.
-    A ordered list of UploadResources can be used to determine the order in which resources need to be uploaded.
+    An ordered list of UploadResources can be used to determine the order in which resources need to be uploaded.
     """
 
     res_id: str
-    stash_links_to: list[str] = field(default_factory=list)
+    stash_links_to: list[XMLLink | ResptrLink] = field(default_factory=list)

src/dsp_tools/utils/xmlupload/stash/stash_models.py

@@ -91,7 +91,7 @@ def make(standoff_stash: StandoffStash | None, link_value_stash: LinkValueStash
             link_value_stash: A LinkValueStash object or None.
 
         Returns:
-            Stash: A Stash object, or None if both iunputs are None.
+            Stash: A Stash object, or None if both inputs are None.
         """
         if standoff_stash or link_value_stash:
             return Stash(standoff_stash, link_value_stash)

test/unittests/test_analyse_xml_data/test_extract_links_from_XML.py

@@ -5,10 +5,11 @@
 from pytest_unordered import unordered
 
 from dsp_tools.analyse_xml_data.extract_links_from_XML import (
+    _create_class_instance_resptr_link,
+    _create_class_instance_text_prop,
+    _create_info_from_xml_for_graph,
     _create_info_from_xml_for_graph_from_one_resource,
-    _extract_ids_from_one_resptr_prop,
     _extract_ids_from_one_text_value,
-    _extract_ids_from_text_prop,
     _get_all_links_from_one_resource,
 )
 
@@ -25,8 +26,7 @@ def test_create_info_from_xml_for_graph_from_one_resource() -> None:
     res_resptr_links, res_xml_links, res_all_used_ids = _create_info_from_xml_for_graph_from_one_resource(test_ele)
     res_B_19 = [obj.object_id for obj in res_resptr_links]
     assert "res_B_19" in res_B_19
-    res_C_19 = [obj.object_id for obj in res_resptr_links]
-    assert "res_C_19" in res_C_19
+    assert "res_C_19" in res_B_19
     assert "res_A_19" == res_all_used_ids
     assert res_xml_links[0].subject_id == "res_A_19"
     assert res_xml_links[0].object_link_ids == {"res_B_19", "res_C_19"}
@@ -40,17 +40,17 @@ def test_get_all_links_from_one_resource() -> None:
         'href="IRI:res_B_11:IRI">res_B_11</a></text></text-prop><resptr-prop name=":hasResource1"><resptr '
         'permissions="prop-default">res_B_11</resptr></resptr-prop></resource>'
     )
-    res_resptr, res_xml = _get_all_links_from_one_resource(test_ele)
-    expected_resptr, expected_xml = ["res_B_11"], [{"res_B_11"}]
-    assert expected_resptr == res_resptr
-    assert unordered(res_xml) == expected_xml
+    res_resptr, res_xml = _get_all_links_from_one_resource("res_A_11", test_ele)
+    assert res_resptr[0].object_id == "res_B_11"
+    res_xml_ids = res_xml[0]
+    assert res_xml_ids.object_link_ids == {"res_B_11"}
 
 
 def test_get_all_links_from_one_resource_no_links() -> None:
     test_ele = etree.fromstring(
         '<resource label="res_B_18" restype=":TestThing" id="res_B_18" permissions="res-default"/>'
     )
-    res = _get_all_links_from_one_resource(test_ele)
+    res = _get_all_links_from_one_resource("res_B_18", test_ele)
     assert res == ([], [])
 
 
@@ -62,9 +62,10 @@ def test_text_only_get_all_links_from_one_resource() -> None:
         'href="IRI:res_A_18:IRI">res_A_18</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></resource>'
     )
-    res_resptr, res_xml = _get_all_links_from_one_resource(test_ele)
+    res_resptr, res_xml = _get_all_links_from_one_resource("res_C_18", test_ele)
     assert not res_resptr
-    assert unordered(res_xml) == [{"res_A_18"}, {"res_B_18"}]
+    res_xml_ids = [x.object_link_ids for x in res_xml]
+    assert unordered(res_xml_ids) == [{"res_A_18"}, {"res_B_18"}]
 
 
 def test_extract_id_one_text_with_one_id() -> None:
@@ -102,28 +103,54 @@ def test_extract_ids_from_text_prop_with_several_text_links() -> None:
         'href="IRI:res_A_18:IRI">res_A_18</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 = _extract_ids_from_text_prop(test_ele)
-    assert unordered(res) == [{"res_A_18"}, {"res_B_18"}]
+    res = _create_class_instance_text_prop("res_C_18", test_ele)
+    res_ids = [x.object_link_ids for x in res]
+    assert unordered(res_ids) == [{"res_A_18"}, {"res_B_18"}]
 
 
-def test_extract_one_id_resptr_prop() -> None:
+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" name=":hasResource1"><resptr '
         'permissions="prop-default">res_C_15</resptr></resptr-prop>'
     )
-    res = _extract_ids_from_one_resptr_prop(test_ele)
-    assert unordered(res) == ["res_C_15"]
+    res = _create_class_instance_resptr_link("res_A_15", test_ele)
+    assert res[0].object_id == "res_C_15"
 
 
-def test_extract_several_id_resptr_prop() -> None:
+def test_create_class_instance_resptr_link_several() -> None:
     test_ele = etree.fromstring(
         '<resptr-prop xmlns="https://dasch.swiss/schema" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" '
         'name=":hasResource1"><resptr permissions="prop-default">res_A_13</resptr><resptr '
         'permissions="prop-default">res_B_13</resptr><resptr permissions="prop-default">res_C_13</resptr></resptr-prop>'
     )
-    res = _extract_ids_from_one_resptr_prop(test_ele)
-    assert unordered(res) == ["res_A_13", "res_B_13", "res_C_13"]
+    res = _create_class_instance_resptr_link("res_D_13", test_ele)
+    assert res[0].object_id == "res_A_13"
+    assert res[1].object_id == "res_B_13"
+    assert res[2].object_id == "res_C_13"
+
+
+def test_create_info_from_xml_for_graph_check_UUID_in_root() -> None:
+    root = etree.fromstring(
+        b'<?xml version="1.0" encoding="UTF-8"?><knora xmlns="https://dasch.swiss/schema" '
+        b'xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="https://dasch.swiss/schema '
+        b'https://raw.githubusercontent.com/dasch-swiss/dsp-tools/main/src/dsp_tools/resources/schema/data.xsd" '
+        b'shortcode="0700" default-ontology="simcir"><resource label="res_A_11" restype=":TestThing" id="res_A_11" '
+        b'permissions="res-default"><resptr-prop name=":hasResource1"><resptr '
+        b'permissions="prop-default">res_B_11</resptr></resptr-prop></resource><resource label="res_B_11" '
+        b'restype=":TestThing" id="res_B_11" permissions="res-default"><text-prop name=":hasRichtext"><text '
+        b'permissions="prop-default" encoding="xml">Start text<a class="salsah-link" '
+        b'href="IRI:res_C_11:IRI">res_C_11</a>end text.</text></text-prop></resource><resource label="res_C_11" '
+        b'restype=":TestThing" id="res_C_11" permissions="res-default"></resource></knora>'
+    )
+    res_root, res_resptr_li, res_xml_li, res_all_ids = _create_info_from_xml_for_graph(root)
+    res_resptr = res_resptr_li[0]
+    res_xml = res_xml_li[0]
+    assert unordered(res_all_ids) == ["res_A_11", "res_B_11", "res_C_11"]
+    xml_res_resptr = res_root.find(".//{https://dasch.swiss/schema}resptr")
+    assert xml_res_resptr.attrib["stashUUID"] == res_resptr.link_uuid  # type: ignore[union-attr]
+    xml_res_text = res_root.find(".//{https://dasch.swiss/schema}text")
+    assert xml_res_text.attrib["stashUUID"] == res_xml.link_uuid  # type: ignore[union-attr]
 
 
 if __name__ == "__main__":

testdata/xml-data/circular-references/create_complex_circular_references.py

@@ -70,8 +70,9 @@ def _make_xml_text_prop(target_res: etree._Element | list[etree._Element]) -> et
             salsah_link = _make_salsah_link(target_res.attrib["id"])
             # one resource with many targets
         case list():
-            salsah_link = "".join([_make_salsah_link(x.attrib["id"]) for x in target_res])
-    return excel2xml.make_text_prop(name=":hasRichtext", value=excel2xml.PropertyElement(salsah_link, encoding="xml"))
+            salsah_link = "blabla".join([_make_salsah_link(x.attrib["id"]) for x in target_res])
+    text_value = "Start text" + salsah_link + "end text."
+    return excel2xml.make_text_prop(name=":hasRichtext", value=excel2xml.PropertyElement(text_value, encoding="xml"))
 
 
 def _make_resptr_prop(target_id: list[str] | str, property_name: str = ":hasResource1") -> etree._Element:
@@ -257,7 +258,8 @@ def _make_two_resource_circle_plus_non_circle_link(replication_counter: str) ->
 
 def _make_single_text_ele_for_text_prop(target_resource_id: list[str]) -> etree._Element:
     salsa_links = [_make_salsah_link(x) for x in target_resource_id]
-    xml_props = [excel2xml.PropertyElement(salsah_link, encoding="xml") for salsah_link in salsa_links]
+    text_values = [f"Start text{link}end text." for link in salsa_links]
+    xml_props = [excel2xml.PropertyElement(text_value, encoding="xml") for text_value in text_values]
     return excel2xml.make_text_prop(name=":hasRichtext", value=xml_props)
 
 
@@ -297,3 +299,7 @@ def _make_complex_circle_with_leaf_nodes(replication_counter: str) -> list[etree
     resource_list[-1].append(xml_prop)
     resource_list.extend(leaf_resources)
     return resource_list
+
+
+if __name__ == "__main__":
+    create_and_save_circular_references_test_graph()