refactor: Backup anchors with id and no href, for compatibility with autorefs' Markdown anchors

pawamoy · oprypin · web-flow · commit b5236b4333eb · 2024-02-22T14:58:49.000+01:00
PR-#651: #651 Related-to-mkdocs-autorefs#39: mkdocstrings/autorefs#39 Co-authored-by: Oleh Prypin <oleh@pryp.in>
diff --git a/src/mkdocstrings/handlers/rendering.py b/src/mkdocstrings/handlers/rendering.py
@@ -146,19 +146,35 @@ def __init__(self, md: Markdown, id_prefix: str):
         self.id_prefix = id_prefix
 
     def run(self, root: Element) -> None:  # noqa: D102 (ignore missing docstring)
-        if not self.id_prefix:
-            return
-        for el in root.iter():
-            id_attr = el.get("id")
-            if id_attr:
-                el.set("id", self.id_prefix + id_attr)
+        if self.id_prefix:
+            self._prefix_ids(root)
 
+    def _prefix_ids(self, root: Element) -> None:
+        index = len(root)
+        for el in reversed(root):  # Reversed mainly for the ability to mutate during iteration.
+            index -= 1
+
+            self._prefix_ids(el)
             href_attr = el.get("href")
+
+            if id_attr := el.get("id"):
+                if el.tag == "a" and not href_attr:
+                    # An anchor with id and no href is used by autorefs:
+                    # leave it untouched and insert a copy with updated id after it.
+                    new_el = copy.deepcopy(el)
+                    new_el.set("id", self.id_prefix + id_attr)
+                    root.insert(index + 1, new_el)
+                else:
+                    # Anchors with id and href are not used by autorefs:
+                    # update in place.
+                    el.set("id", self.id_prefix + id_attr)
+
+            # Always update hrefs, names and labels-for:
+            # there will always be a corresponding id.
             if href_attr and href_attr.startswith("#"):
                 el.set("href", "#" + self.id_prefix + href_attr[1:])
 
-            name_attr = el.get("name")
-            if name_attr:
+            if name_attr := el.get("name"):
                 el.set("name", self.id_prefix + name_attr)
 
             if el.tag == "label":
diff --git a/tests/fixtures/markdown_anchors.py b/tests/fixtures/markdown_anchors.py
@@ -0,0 +1,16 @@
+"""Module docstring.
+
+[](){#anchor}
+
+Paragraph.
+
+[](){#heading-anchor-1}
+[](){#heading-anchor-2}
+[](){#heading-anchor-3}
+## Heading
+
+[](#has-href1)
+[](#has-href2){#with-id}
+
+Pararaph.
+"""
diff --git a/tests/test_extension.py b/tests/test_extension.py
@@ -172,3 +172,50 @@ def test_removing_duplicated_headings(ext_markdown: Markdown) -> None:
     assert output.count(">Heading two<") == 1
     assert output.count(">Heading three<") == 1
     assert output.count('class="mkdocstrings') == 0
+
+
+def _assert_contains_in_order(items: list[str], string: str) -> None:
+    index = 0
+    for item in items:
+        assert item in string[index:]
+        index = string.index(item, index) + len(item)
+
+
+@pytest.mark.parametrize("ext_markdown", [{"markdown_extensions": [{"attr_list": {}}]}], indirect=["ext_markdown"])
+def test_backup_of_anchors(ext_markdown: Markdown) -> None:
+    """Anchors with empty `href` are backed up."""
+    output = ext_markdown.convert("::: tests.fixtures.markdown_anchors")
+
+    # Anchors with id and no href have been backed up and updated.
+    _assert_contains_in_order(
+        [
+            'id="anchor"',
+            'id="tests.fixtures.markdown_anchors--anchor"',
+            'id="heading-anchor-1"',
+            'id="tests.fixtures.markdown_anchors--heading-anchor-1"',
+            'id="heading-anchor-2"',
+            'id="tests.fixtures.markdown_anchors--heading-anchor-2"',
+            'id="heading-anchor-3"',
+            'id="tests.fixtures.markdown_anchors--heading-anchor-3"',
+        ],
+        output,
+    )
+
+    # Anchors with href and with or without id have been updated but not backed up.
+    _assert_contains_in_order(
+        [
+            'id="tests.fixtures.markdown_anchors--with-id"',
+        ],
+        output,
+    )
+    assert 'id="with-id"' not in output
+
+    _assert_contains_in_order(
+        [
+            'href="#tests.fixtures.markdown_anchors--has-href1"',
+            'href="#tests.fixtures.markdown_anchors--has-href2"',
+        ],
+        output,
+    )
+    assert 'href="#has-href1"' not in output
+    assert 'href="#has-href2"' not in output
