HTML Search: omit anchor reference from document titles in the search index.

CHANGES.rst

@@ -304,6 +304,10 @@ Bugs fixed
   download cache.
   Patch by James Addison and Adam Turner.
 
+* #11961: Omit anchor references from document title entries in the search index,
+  removing duplication of search results.
+  Patch by James Addison.
+
 Testing
 -------
 

sphinx/search/__init__.py

@@ -504,7 +504,7 @@ def _visit_nodes(node):
             elif isinstance(node, nodes.title):
                 title = node.astext()
                 ids = node.parent['ids']
-                word_store.titles.append((title, ids[0] if ids else None))
+                word_store.titles.append((title, ids[0] if ids and word_store.titles else None))  # NoQA: E501
                 word_store.title_words.extend(split(title))
             for child in node.children:
                 _visit_nodes(child)

tests/test_search.py

@@ -67,6 +67,9 @@ def is_registered_term(index, keyword):
 
 .. test that comments are not indexed: boson
 
+another_title
+===============
+
 test that non-comments are indexed: fermion
 '''
 
@@ -172,7 +175,10 @@ def test_IndexBuilder():
         'index': {'docname1_1', 'docname1_2', 'docname2_1', 'docname2_2'},
         'test': {'docname1_1', 'docname1_2', 'docname2_1', 'docname2_2'},
     }
-    assert index._title_mapping == {'section_titl': {'docname1_1', 'docname1_2', 'docname2_1', 'docname2_2'}}
+    assert index._title_mapping == {
+        'another_titl': {'docname1_1', 'docname1_2', 'docname2_1', 'docname2_2'},
+        'section_titl': {'docname1_1', 'docname1_2', 'docname2_1', 'docname2_2'},
+    }
     assert index._objtypes == {}
     assert index._objnames == {}
 
@@ -192,8 +198,14 @@ def test_IndexBuilder():
                   'non': [0, 1, 2, 3],
                   'test': [0, 1, 2, 3]},
         'titles': ('title1_1', 'title1_2', 'title2_1', 'title2_2'),
-        'titleterms': {'section_titl': [0, 1, 2, 3]},
-        'alltitles': {'section_title': [(0, 'section-title'), (1, 'section-title'), (2, 'section-title'), (3, 'section-title')]},
+        'titleterms': {
+            'another_titl': [0, 1, 2, 3],
+            'section_titl': [0, 1, 2, 3],
+        },
+        'alltitles': {
+            'another_title': [(0, 'another-title'), (1, 'another-title'), (2, 'another-title'), (3, 'another-title')],
+            'section_title': [(0, None), (1, None), (2, None), (3, None)],
+        },
         'indexentries': {},
     }
     assert index._objtypes == {('dummy1', 'objtype1'): 0, ('dummy2', 'objtype1'): 1}
@@ -234,7 +246,10 @@ def test_IndexBuilder():
         'index': {'docname1_2', 'docname2_2'},
         'test': {'docname1_2', 'docname2_2'},
     }
-    assert index._title_mapping == {'section_titl': {'docname1_2', 'docname2_2'}}
+    assert index._title_mapping == {
+        'another_titl': {'docname1_2', 'docname2_2'},
+        'section_titl': {'docname1_2', 'docname2_2'},
+    }
     assert index._objtypes == {('dummy1', 'objtype1'): 0, ('dummy2', 'objtype1'): 1}
     assert index._objnames == {0: ('dummy1', 'objtype1', 'objtype1'), 1: ('dummy2', 'objtype1', 'objtype1')}
 
@@ -253,8 +268,14 @@ def test_IndexBuilder():
                   'non': [0, 1],
                   'test': [0, 1]},
         'titles': ('title1_2', 'title2_2'),
-        'titleterms': {'section_titl': [0, 1]},
-        'alltitles': {'section_title': [(0, 'section-title'), (1, 'section-title')]},
+        'titleterms': {
+            'another_titl': [0, 1],
+            'section_titl': [0, 1],
+        },
+        'alltitles': {
+            'another_title': [(0, 'another-title'), (1, 'another-title')],
+            'section_title': [(0, None), (1, None)],
+        },
         'indexentries': {},
     }
     assert index._objtypes == {('dummy1', 'objtype1'): 0, ('dummy2', 'objtype1'): 1}
@@ -343,6 +364,7 @@ def assert_is_sorted(item, path: str):
             assert_is_sorted(value, f'{path}.{key}')
     elif isinstance(item, list):
         if not is_title_tuple_type(item) and path not in lists_not_to_sort:
-            assert item == sorted(item), f'{err_path} is not sorted'
+            # sort nulls last; http://stackoverflow.com/questions/19868767/
+            assert item == sorted(item, key=lambda x: (x is None, x)), f'{err_path} is not sorted'
         for i, child in enumerate(item):
             assert_is_sorted(child, f'{path}[{i}]')