fix: Raise error if duplicate parts are found in the control statement (

#1351)

* fix: Raise error if duplicate parts are found in statement

Signed-off-by: Ekaterina Nikonova <enikonovad@gmail.com>

* fix: Make lint happy

Signed-off-by: Ekaterina Nikonova <enikonovad@gmail.com>

* Remove unnecessary logging

Signed-off-by: Ekaterina Nikonova <enikonovad@gmail.com>

---------

Signed-off-by: Ekaterina Nikonova <enikonovad@gmail.com>

tests/trestle/core/commands/author/catalog_test.py

@@ -661,3 +661,35 @@ def test_catalog_multiline_statement(tmp_trestle_dir: pathlib.Path, monkeypatch:
     assert catalog.groups[0].controls[1].parts[0].parts[
         0
     ].prose == 'Define and document the types of accounts allowed and specifically prohibited for use within the system;'  # noqa E501
+
+
+def test_catalog_duplicate_parts_statement(tmp_trestle_dir: pathlib.Path, monkeypatch: MonkeyPatch, capsys) -> None:
+    """Test catalog-assemble with duplicate parts for control statement."""
+    catalog = cat.Catalog.oscal_read(test_utils.JSON_TEST_DATA_PATH / test_utils.SIMPLIFIED_NIST_CATALOG_NAME)
+    ModelUtils.save_top_level_model(catalog, tmp_trestle_dir, 'my_catalog', FileContentType.JSON)
+
+    catalog_generate = 'trestle author catalog-generate -n my_catalog -o md_catalog'
+    test_utils.execute_command_and_assert(catalog_generate, 0, monkeypatch)
+    md_path = tmp_trestle_dir / 'md_catalog/ac/ac-2.md'
+    assert md_path.exists()
+
+    control_statement_prose_with_parts = """The organization:
+
+- \[a\] Part A
+- \[a\] Part A Duplicate.
+  - \[a.1\] Documents 1
+  - \[a.2\] Documents 2
+    - \[a.2.1\] SubDocuments 1
+    - \[a.2.2\] SubDocuments 2
+- \[b\] Part 2.
+
+
+"""
+
+    file_utils.insert_text_in_file(md_path, '## Control Statement', control_statement_prose_with_parts)
+
+    catalog_assemble = 'trestle author catalog-assemble -o my_catalog -m md_catalog'
+    test_utils.execute_command_and_assert(catalog_assemble, 1, monkeypatch)
+
+    _, error = capsys.readouterr()
+    assert 'Duplicate part id ac-2_smt.a' in error

trestle/core/markdown/base_markdown_node.py

@@ -70,7 +70,7 @@ def get_all_headers_for_level(self, level: int) -> Iterable[str]:
     def get_node_for_key(self, key: str, strict_matching: bool = True) -> Optional[BaseMarkdownNode]:
         """Return a first node for the given key, substring matching is supported. The method is case insensitive."""
         if not strict_matching:
-            if not any([key.lower() in el.lower() for el in self.content.subnodes_keys]):
+            if not any(key.lower() in el.lower() for el in self.content.subnodes_keys):
                 return None
             elif len(as_filtered_list(self.content.subnodes_keys, lambda el: key.lower() in el.lower())) > 1:
                 logger.warning(f'Multiple nodes for {key} were found, only the first one will be returned.')
@@ -99,7 +99,7 @@ def get_all_nodes_for_keys(
         Returns: List of found markdown nodes
         """
         if not strict_matching:
-            if not any([key in el for el in self.content.subnodes_keys for key in keys]):
+            if not any(key in el for el in self.content.subnodes_keys for key in keys):
                 return []
         elif not set(keys).intersection(self.content.subnodes_keys):
             return []
@@ -256,10 +256,10 @@ def _rec_traverse(self, node: BaseMarkdownNode, key: str, strict_matching: bool)
         """
         if key.lower() == node.key.lower() or (not strict_matching and key.lower() in node.key.lower()):
             return node
-        if (not strict_matching and any([key.lower() in el.lower()
-                                         for el in node.content.subnodes_keys])) or (key.lower() in [
-                                             el.lower() for el in node.content.subnodes_keys
-                                         ]):
+        if (not strict_matching and any(key.lower() in el.lower()
+                                        for el in node.content.subnodes_keys)) or (key.lower() in [
+                                            el.lower() for el in node.content.subnodes_keys
+                                        ]):
             for subnode in node.subnodes:
                 matched_node = self._rec_traverse(subnode, key, strict_matching)
                 if matched_node is not None:

trestle/core/markdown/control_markdown_node.py

@@ -462,6 +462,11 @@ def _read_parts(self, indent: int, ii: int, lines: List[str], parent_id: str,
                 name = 'objective' if id_.find('_obj') > 0 else 'item'
                 prop = common.Property(name='label', value=id_text)
                 part = common.Part(name=name, id=id_, prose=prose, props=[prop])
+                if id_ in [p.id for p in parts]:
+                    raise TrestleError(
+                        f'Duplicate part id {id_} is found in markdown '
+                        f'{tree_context.control_id}. Please correct the part label in line {line}.'
+                    )
                 parts.append(part)
                 ii += 1
             elif new_indent > indent:

trestle/core/markdown/markdown_processor.py

@@ -81,6 +81,7 @@ def process_control_markdown(
             _ = self.render_gfm_to_html(markdown_wo_header)
 
             lines = markdown_wo_header.split('\n')
+            tree_context.reset()
             tree_context.section_to_part_name_map = section_to_part_name_map
             tree_context.part_label_to_id_map = part_label_to_id_map
             tree = ControlMarkdownNode.build_tree_from_markdown(lines)