strip common punct from around nums

laws-africa · Feb 10, 2025 · 29d3d1f · 29d3d1f
1 parent 90bae07
commit 29d3d1f
Show file tree

Hide file tree

Showing 2 changed files with 80 additions and 1 deletion.
diff --git a/indigo/analysis/refs/provisions.py b/indigo/analysis/refs/provisions.py
@@ -320,16 +320,20 @@ def find_numbered_hier_element(self, root: Element, names: Optional[List[str]],
         names = [f'{{{ns}}}{n}' for n in names]
         dead_ends = [f'{{{ns}}}{n}' for n in ['quotedStructure', 'embeddedStructure', 'content']]
         not_outside_of = None if not_outside_of is None else [f'{{{ns}}}{n}' for n in not_outside_of]
+        clean_num = self.clean_num(num)
 
         # do a breadth-first search, starting at root, and walk upwards, expanding until we find something or reach the top
         for elem in bfs_upward_search(root, names, dead_ends, not_outside_of):
             # ignore matches to the root element, which avoids things like section (1)(a)(a) matching the same (a) twice
             if elem == root:
                 continue
             num_elem = elem.find('a:num', {'a': ns})
-            if num_elem is not None and num_elem.text.rstrip(".") == num:
+            if num_elem is not None and self.clean_num(num_elem.text) == clean_num:
                 return elem
 
+    def clean_num(self, num):
+        return num.strip("()").rstrip(".º")
+
 
 class ProvisionRefsMatcher(CitationMatcher):
     """ Finds internal references to sections in documents, of the form:

diff --git a/indigo/tests/test_provision_refs.py b/indigo/tests/test_provision_refs.py
@@ -494,6 +494,81 @@ def test_local_sections_synonyms(self):
             etree.tostring(actual, encoding='unicode')
         )
 
+    def test_num_punct(self):
+        # nums can be surrounded by () or have . or º at the end
+        doc = AkomaNtosoDocument(document_fixture(xml="""
+            <section eId="sec_7">
+              <num>7.</num>
+              <heading>Section 7</heading>
+              <content>
+                <p>As given in section 26, blah.</p>
+                <p>As given in section 26(a), blah.</p>
+                <p>As given in section 26(a)(1)(iii), blah.</p>
+                <p>As given in section 26B, blah.</p>
+              </content>
+            </section>
+            <section eId="sec_26">
+              <num>26.</num>
+              <heading>Important heading</heading>
+              <subsection eId="sec_26__subsec_a">
+                <num>a)</num>
+                <paragraph eId="sec_26__subsec_a__para_1">
+                  <num>1)</num>
+                </paragraph>
+                <paragraph eId="sec_26__subsec_a__para_2">
+                  <num>(2)</num>
+                </paragraph>
+              </subsection>
+            </section>
+            <section eId="sec_26B">
+              <num>26Bº</num>
+              <heading>Another important heading</heading>
+              <content>
+                <p>Another important provision.</p>
+              </content>
+            </section>
+        """))
+
+        expected = AkomaNtosoDocument(document_fixture(xml="""
+            <section eId="sec_7">
+              <num>7.</num>
+              <heading>Section 7</heading>
+              <content>
+                <p>As given in section <ref href="#sec_26">26</ref>, blah.</p>
+                <p>As given in section <ref href="#sec_26__subsec_a">26(a)</ref>, blah.</p>
+                <p>As given in section <ref href="#sec_26__subsec_a__para_1">26(a)(1)</ref>(iii), blah.</p>
+                <p>As given in section <ref href="#sec_26B">26B</ref>, blah.</p>
+              </content>
+            </section>
+            <section eId="sec_26">
+              <num>26.</num>
+              <heading>Important heading</heading>
+              <subsection eId="sec_26__subsec_a">
+                <num>a)</num>
+                <paragraph eId="sec_26__subsec_a__para_1">
+                  <num>1)</num>
+                </paragraph>
+                <paragraph eId="sec_26__subsec_a__para_2">
+                  <num>(2)</num>
+                </paragraph>
+              </subsection>
+            </section>
+            <section eId="sec_26B">
+              <num>26Bº</num>
+              <heading>Another important heading</heading>
+              <content>
+                <p>Another important provision.</p>
+              </content>
+            </section>
+        """))
+
+        actual = etree.fromstring(doc.to_xml())
+        self.finder.markup_xml_matches(self.frbr_uri, actual)
+        self.assertEqual(
+            expected.to_xml(encoding='unicode'),
+            etree.tostring(actual, encoding='unicode')
+        )
+
     def test_local_ambiugous_levels(self):
         doc = AkomaNtosoDocument(document_fixture(xml="""
             <section eId="sec_7">