performance: faster autorefs URL membership check

Signed-off-by: Timothée Mazzucotelli <dev@pawamoy.fr>

python/tests/unit/extensions/test_autorefs.py

@@ -117,6 +117,34 @@ class TestStore:
             "external": "https://example.com/external"
         }
 
+    def test_urls_are_ordered_and_unique(self) -> None:
+        """Candidate URLs preserve registration order without duplicates."""
+        store = get_autorefs_store()
+        page = Page("page", "page.html")
+
+        store.register_anchor(page, "identifier", "first")
+        store.register_anchor(page, "identifier", "second")
+        store.register_anchor(page, "identifier", "first")
+
+        assert get_autorefs_page_data("page")["primary"] == {
+            "identifier": ["page#first", "page#second"]
+        }
+
+    def test_page_registrations_remove_urls(self) -> None:
+        """Reprocessing a page removes its previously registered URLs."""
+        store = get_autorefs_store()
+        first_page = Page("first", "first.html")
+        second_page = Page("second", "second.html")
+        store.register_anchor(first_page, "identifier", "anchor")
+        store.register_anchor(second_page, "identifier", "anchor")
+
+        store.set_page(first_page)
+
+        assert get_autorefs_page_data("first")["primary"] == {}
+        assert get_autorefs_page_data("second")["primary"] == {
+            "identifier": ["second#anchor"]
+        }
+
 
 # ---------------------------------------------------------------------------
 # Inline processor
@@ -415,8 +443,7 @@ class TestAnchorsTreeprocessor:
                 [](){#alias10}
             """),
         )
-        store = get_autorefs_store()
-        assert store._primary_url_map == {
+        assert get_autorefs_page_data("page")["primary"] == {
             "foo": ["page#heading-foo"],
             "heading-foo": ["page#heading-foo"],
             "bar": ["page#bar"],
@@ -474,8 +501,7 @@ class TestAnchorsTreeprocessor:
                 ## Heading baz
             """),
         )
-        store = get_autorefs_store()
-        assert store._primary_url_map == {
+        assert get_autorefs_page_data("page")["primary"] == {
             "heading-foo": ["page#heading-foo"],
             "heading-bar": ["page#heading-bar"],
             "heading-baz": ["page#heading-baz"],
@@ -510,9 +536,9 @@ class TestHeadingsTreeprocessor:
     def test_register_heading(self, md: Markdown) -> None:
         """A single heading is registered under its toc-generated slug."""
         md.convert("## Foo")
-        store = get_autorefs_store()
-        assert "foo" in store._primary_url_map
-        assert store._primary_url_map["foo"] == ["page#foo"]
+        assert get_autorefs_page_data("page")["primary"]["foo"] == [
+            "page#foo"
+        ]
 
     @pytest.mark.parametrize(
         "md",

python/zensical/extensions/autorefs.py

@@ -27,7 +27,7 @@ import re
 from abc import ABC, abstractmethod
 from dataclasses import dataclass
 from html import escape
-from typing import TYPE_CHECKING, Any
+from typing import TYPE_CHECKING, Any, TypeAlias
 from xml.etree.ElementTree import Element
 
 from markdown.core import Markdown
@@ -59,6 +59,11 @@ if TYPE_CHECKING:
 
 HTAGS = {"h1", "h2", "h3", "h4", "h5", "h6"}
 
+# URL order determines which target wins when several are equally
+# suitable. Dict keys give us that order plus constant-time lookups.
+OrderedSet: TypeAlias = dict[str, None]
+URLMap: TypeAlias = dict[str, OrderedSet]
+
 
 # ----------------------------------------------------------------------------
 # Globals
@@ -81,8 +86,8 @@ class AutorefsStore:
         self.scan_toc: bool = True
         self.record_backlinks: bool = False
 
-        self._primary_url_map: dict[str, list[str]] = {}
-        self._secondary_url_map: dict[str, list[str]] = {}
+        self._primary_url_map: URLMap = {}
+        self._secondary_url_map: URLMap = {}
         self._abs_url_map: dict[str, str] = {}
         self._title_map: dict[str, str] = {}
         self._page_registrations: dict[str, set[tuple[bool, str, str]]] = {}
@@ -118,7 +123,7 @@ class AutorefsStore:
 
     @staticmethod
     def _pop_urls(
-        url_map: dict[str, list[str]],
+        url_map: URLMap,
         registrations: set[tuple[bool, str, str]],
         primary: bool,
     ) -> dict[str, list[str]]:
@@ -130,12 +135,11 @@ class AutorefsStore:
 
         result: dict[str, list[str]] = {}
         for identifier, selected_urls in selected.items():
-            urls = url_map.get(identifier, [])
+            urls = url_map.get(identifier, {})
             result[identifier] = [url for url in urls if url in selected_urls]
-            remaining = [url for url in urls if url not in selected_urls]
-            if remaining:
-                url_map[identifier] = remaining
-            else:
+            for url in selected_urls:
+                urls.pop(url, None)
+            if not urls:
                 url_map.pop(identifier, None)
         return result
 
@@ -150,11 +154,7 @@ class AutorefsStore:
     ) -> None:
         url = f"{page.url}#{anchor or identifier}"
         url_map = self._primary_url_map if primary else self._secondary_url_map
-        if identifier in url_map:
-            if url not in url_map[identifier]:
-                url_map[identifier].append(url)
-        else:
-            url_map[identifier] = [url]
+        url_map.setdefault(identifier, {})[url] = None
         self._page_registrations.setdefault(page.url, set()).add(
             (primary, identifier, url)
         )