Extract managed form elements from XML

packages/one-c-normalizer/src/one_c_normalizer/__init__.py

@@ -171,7 +171,7 @@ def parse_one_c_xml_file(path: str | Path) -> list[OneCXmlObject]:
     source_path = normalize_source_path(path)
     root = ET.fromstring(_read_text_file(Path(path)))
     result: list[OneCXmlObject] = []
-    _walk_xml_objects(source_path, root, result, current_role=None, parent_qualified_name=None)
+    _walk_xml_objects(source_path, root, result, current_role=None, parent_qualified_name=None, parent_object_kind=None)
     return result
 
 
@@ -351,10 +351,12 @@ def _walk_xml_objects(
     *,
     current_role: OneCXmlObject | None,
     parent_qualified_name: str | None,
+    parent_object_kind: str | None,
 ) -> None:
     role_context = current_role
     child_parent_qualified_name = parent_qualified_name
-    object_kind = _xml_object_kind(element)
+    child_parent_object_kind = parent_object_kind
+    object_kind = _xml_object_kind(element, parent_object_kind=parent_object_kind)
     if object_kind == "RIGHT":
         right = _xml_right_object(source_path, element, role_context)
         if right is not None:
@@ -371,6 +373,7 @@ def _walk_xml_objects(
             )
             result.append(xml_object)
             child_parent_qualified_name = xml_object.qualified_name
+            child_parent_object_kind = object_kind
             if object_kind == "ROLE":
                 role_context = xml_object
 
@@ -381,6 +384,7 @@ def _walk_xml_objects(
             result,
             current_role=role_context,
             parent_qualified_name=child_parent_qualified_name,
+            parent_object_kind=child_parent_object_kind,
         )
 
 
@@ -695,6 +699,25 @@ _OBJECT_KIND_BY_TAG = {
     "предопределенный": "PREDEFINED",
 }
 
+_FORM_ELEMENT_TAGS = {
+    "item",
+    "items",
+    "element",
+    "elements",
+    "formitem",
+    "formitems",
+    "field",
+    "fields",
+    "group",
+    "groups",
+    "table",
+    "tables",
+    "button",
+    "buttons",
+    "page",
+    "pages",
+}
+
 
 _QUALIFIED_PREFIX_BY_KIND = {
     "CATALOG": "Справочник",
@@ -1288,8 +1311,10 @@ _PATH_METADATA_ALIASES = {
 }
 
 
-def _xml_object_kind(element: ET.Element) -> str | None:
+def _xml_object_kind(element: ET.Element, *, parent_object_kind: str | None = None) -> str | None:
     tag = _local_name(element.tag).lower()
+    if parent_object_kind in {"FORM", "ELEMENT"} and tag in _FORM_ELEMENT_TAGS and _xml_name(element):
+        return "ELEMENT"
     if tag in {"metadataobject", "object"}:
         type_name = _xml_type_name(element)
         if type_name:
@@ -1384,6 +1409,11 @@ def _xml_qualified_name(
 
 def _xml_attributes(element: ET.Element) -> dict:
     attributes = dict(element.attrib)
+    for key, value in element.attrib.items():
+        local_key = _local_name(key)
+        attributes.setdefault(local_key, value)
+        if local_key.lower() == "type":
+            attributes.setdefault("control_kind", value.split(":")[-1].split(".")[-1])
     attribute_role = _xml_attribute_role(element)
     if attribute_role:
         attributes.setdefault("attribute_role", attribute_role)

packages/semantic-kernel/src/semantic_kernel/__init__.py

@@ -983,7 +983,9 @@ def _xml_edge_kind(kind: NodeKind) -> EdgeKind:
         return EdgeKind.HAS_TABULAR_SECTION
     if kind == NodeKind.ROLE:
         return EdgeKind.HAS_ROLE
-    return EdgeKind.HAS_ELEMENT
+    if kind == NodeKind.FORM_ELEMENT:
+        return EdgeKind.HAS_ELEMENT
+    return EdgeKind.CONTAINS
 
 
 def _find_xml_parent(parents: dict[str, SemanticNode], qualified_name: str) -> SemanticNode | None:

packages/semantic-kernel/tests/test_xml_indexing.py

@@ -346,6 +346,33 @@ def test_index_project_links_form_command_to_handler(tmp_path: Path):
     assert any(edge.kind == EdgeKind.HANDLES for edge in snapshot.edges)
 
 
+def test_index_project_extracts_managed_form_items_without_layouts(tmp_path: Path):
+    xml = tmp_path / "form.xml"
+    xml.write_text(
+        """
+<Form name="ФормаДокумента" qualifiedName="Документ.Заказ.ФормаДокумента">
+  <items xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:type="form:FormGroup" name="Основное">
+    <caption>Основное</caption>
+    <items xsi:type="form:FormField" name="Номер">
+      <caption>Номер</caption>
+      <dataPath>Объект.Номер</dataPath>
+    </items>
+  </items>
+  <Layout name="ПечатнаяФорма" qualifiedName="Документ.Заказ.ФормаДокумента.ПечатнаяФорма" />
+</Form>
+""",
+        encoding="utf-8",
+    )
+
+    snapshot = index_project(tmp_path, project_id="ui-form-items")
+
+    form = form_semantics(snapshot)[0]
+    assert [element.name for element in form.elements] == ["Основное", "Номер"]
+    assert all(element.kind == NodeKind.FORM_ELEMENT for element in form.elements)
+    assert form.elements[1].attributes["dataPath"] == "Объект.Номер"
+    assert not any(element.name == "ПечатнаяФорма" for element in form.elements)
+
+
 def test_index_project_links_form_events_to_handlers(tmp_path: Path):
     xml = tmp_path / "form.xml"
     xml.write_text(

packages/ui-semantics/src/ui_semantics/__init__.py

@@ -19,15 +19,20 @@ def form_semantics(snapshot: SirSnapshot) -> list[FormSemantics]:
         for node in snapshot.nodes
         if node.kind == NodeKind.FORM
     }
+    element_children: dict[str, list[SemanticNode]] = {}
     for edge in snapshot.edges:
         form = forms.get(edge.source_lineage)
         target = nodes.get(edge.target_lineage)
-        if form is None or target is None:
+        if target is None:
+            continue
+        if edge.kind == EdgeKind.HAS_ELEMENT and target.kind == NodeKind.FORM_ELEMENT:
+            element_children.setdefault(edge.source_lineage, []).append(target)
+        if form is None:
             continue
         if edge.kind == EdgeKind.HAS_COMMAND:
             form.commands.append(target)
-        elif edge.kind == EdgeKind.HAS_ELEMENT:
-            form.elements.append(target)
+    for form in forms.values():
+        form.elements.extend(_flatten_form_elements(form.form.lineage_id, element_children))
     command_to_form = {
         command.lineage_id: form
         for form in forms.values()
@@ -43,4 +48,20 @@ def form_semantics(snapshot: SirSnapshot) -> list[FormSemantics]:
     return sorted(forms.values(), key=lambda item: item.form.qualified_name)
 
 
+def _flatten_form_elements(root_lineage: str, element_children: dict[str, list[SemanticNode]]) -> list[SemanticNode]:
+    result: list[SemanticNode] = []
+    seen: set[str] = set()
+
+    def visit(parent_lineage: str) -> None:
+        for element in element_children.get(parent_lineage, []):
+            if element.lineage_id in seen:
+                continue
+            seen.add(element.lineage_id)
+            result.append(element)
+            visit(element.lineage_id)
+
+    visit(root_lineage)
+    return result
+
+
 __all__ = ["FormSemantics", "form_semantics"]