refactor: Use a custom autoref HTML tag

PR-48: #48

Author: pawamoy

src/mkdocs_autorefs/plugin.py

@@ -52,6 +52,7 @@ class AutorefsPlugin(BasePlugin):
 
     scan_toc: bool = True
     current_page: str | None = None
+    legacy_refs: bool = True
 
     def __init__(self) -> None:
         """Initialize the object."""
@@ -211,7 +212,7 @@ def on_post_page(self, output: str, page: Page, **kwargs: Any) -> str:  # noqa:
         log.debug(f"Fixing references in page {page.file.src_path}")
 
         url_mapper = functools.partial(self.get_item_url, from_url=page.url, fallback=self.get_fallback_anchor)
-        fixed_output, unmapped = fix_refs(output, url_mapper)
+        fixed_output, unmapped = fix_refs(output, url_mapper, _legacy_refs=self.legacy_refs)
 
         if unmapped and log.isEnabledFor(logging.WARNING):
             for ref in unmapped:

src/mkdocs_autorefs/references.py

@@ -6,6 +6,7 @@
 import re
 import warnings
 from html import escape, unescape
+from html.parser import HTMLParser
 from typing import TYPE_CHECKING, Any, Callable, ClassVar, Match
 from urllib.parse import urlsplit
 from xml.etree.ElementTree import Element
@@ -44,7 +45,12 @@ def __getattr__(name: str) -> Any:
     rf"(?: class=(?P<class>{_ATTR_VALUE}))?(?P<attrs> [^<>]+)?>(?P<title>.*?)</span>",
     flags=re.DOTALL,
 )
-"""A regular expression to match mkdocs-autorefs' special reference markers
+"""Deprecated. Use [`AUTOREF_RE`][mkdocs_autorefs.references.AUTOREF_RE] instead."""
+
+AUTOREF_RE = re.compile(r"<autoref (?P<attrs>.*?)>(?P<title>.*?)</autoref>", flags=re.DOTALL)
+"""The autoref HTML tag regular expression.
+
+A regular expression to match mkdocs-autorefs' special reference markers
 in the [`on_post_page` hook][mkdocs_autorefs.plugin.AutorefsPlugin.on_post_page].
 """
 
@@ -135,8 +141,8 @@ def _make_tag(self, identifier: str, text: str) -> Element:
         Returns:
             A new element.
         """
-        el = Element("span")
-        el.set("data-autorefs-identifier", identifier)
+        el = Element("autoref")
+        el.set("identifier", identifier)
         el.text = text
         return el
 
@@ -167,7 +173,7 @@ def relative_url(url_a: str, url_b: str) -> str:
     return f"{relative}#{anchor}"
 
 
-def fix_ref(url_mapper: Callable[[str], str], unmapped: list[str]) -> Callable:
+def _legacy_fix_ref(url_mapper: Callable[[str], str], unmapped: list[str]) -> Callable:
     """Return a `repl` function for [`re.sub`](https://docs.python.org/3/library/re.html#re.sub).
 
     In our context, we match Markdown references and replace them with HTML links.
@@ -216,7 +222,84 @@ def inner(match: Match) -> str:
     return inner
 
 
-def fix_refs(html: str, url_mapper: Callable[[str], str]) -> tuple[str, list[str]]:
+class _AutorefsAttrs(dict):
+    _handled_attrs: ClassVar[set[str]] = {"identifier", "optional", "hover", "class"}
+
+    @property
+    def remaining(self) -> str:
+        return " ".join(k if v is None else f'{k}="{v}"' for k, v in self.items() if k not in self._handled_attrs)
+
+
+class _HTMLAttrsParser(HTMLParser):
+    def __init__(self):
+        super().__init__()
+        self.attrs = {}
+
+    def parse(self, html: str) -> _AutorefsAttrs:
+        self.attrs.clear()
+        self.feed(html)
+        return _AutorefsAttrs(self.attrs)
+
+    def handle_starttag(self, tag: str, attrs: list[tuple[str, str | None]]) -> None:  # noqa: ARG002
+        self.attrs.update(attrs)
+
+
+_html_attrs_parser = _HTMLAttrsParser()
+
+
+def fix_ref(url_mapper: Callable[[str], str], unmapped: list[str]) -> Callable:
+    """Return a `repl` function for [`re.sub`](https://docs.python.org/3/library/re.html#re.sub).
+
+    In our context, we match Markdown references and replace them with HTML links.
+
+    When the matched reference's identifier was not mapped to an URL, we append the identifier to the outer
+    `unmapped` list. It generally means the user is trying to cross-reference an object that was not collected
+    and rendered, making it impossible to link to it. We catch this exception in the caller to issue a warning.
+
+    Arguments:
+        url_mapper: A callable that gets an object's site URL by its identifier,
+            such as [mkdocs_autorefs.plugin.AutorefsPlugin.get_item_url][].
+        unmapped: A list to store unmapped identifiers.
+
+    Returns:
+        The actual function accepting a [`Match` object](https://docs.python.org/3/library/re.html#match-objects)
+        and returning the replacement strings.
+    """
+
+    def inner(match: Match) -> str:
+        title = match["title"]
+        attrs = _html_attrs_parser.parse(f"<a {match['attrs']}>")
+        identifier: str = attrs["identifier"]
+        optional = "optional" in attrs
+        hover = "hover" in attrs
+
+        try:
+            url = url_mapper(unescape(identifier))
+        except KeyError:
+            if optional:
+                if hover:
+                    return f'<span title="{identifier}">{title}</span>'
+                return title
+            unmapped.append(identifier)
+            if title == identifier:
+                return f"[{identifier}][]"
+            return f"[{title}][{identifier}]"
+
+        parsed = urlsplit(url)
+        external = parsed.scheme or parsed.netloc
+        classes = (attrs.get("class") or "").strip().split()
+        classes = ["autorefs", "autorefs-external" if external else "autorefs-internal", *classes]
+        class_attr = " ".join(classes)
+        if remaining := attrs.remaining:
+            remaining = f" {remaining}"
+        if optional and hover:
+            return f'<a class="{class_attr}" title="{identifier}" href="{escape(url)}"{remaining}>{title}</a>'
+        return f'<a class="{class_attr}" href="{escape(url)}"{remaining}>{title}</a>'
+
+    return inner
+
+
+def fix_refs(html: str, url_mapper: Callable[[str], str], *, _legacy_refs: bool = True) -> tuple[str, list[str]]:
     """Fix all references in the given HTML text.
 
     Arguments:
@@ -228,7 +311,9 @@ def fix_refs(html: str, url_mapper: Callable[[str], str]) -> tuple[str, list[str
         The fixed HTML.
     """
     unmapped: list[str] = []
-    html = AUTO_REF_RE.sub(fix_ref(url_mapper, unmapped), html)
+    html = AUTOREF_RE.sub(fix_ref(url_mapper, unmapped), html)
+    if _legacy_refs:
+        html = AUTO_REF_RE.sub(_legacy_fix_ref(url_mapper, unmapped), html)
     return html, unmapped
 
 

tests/test_references.py

@@ -212,7 +212,7 @@ def test_ignore_reference_with_special_char() -> None:
     )
 
 
-def test_custom_required_reference() -> None:
+def test_legacy_custom_required_reference() -> None:
     """Check that external HTML-based references are expanded or reported missing."""
     url_map = {"ok": "ok.html#ok"}
     source = "<span data-autorefs-identifier=bar>foo</span> <span data-autorefs-identifier=ok>ok</span>"
@@ -221,7 +221,16 @@ def test_custom_required_reference() -> None:
     assert unmapped == ["bar"]
 
 
-def test_custom_optional_reference() -> None:
+def test_custom_required_reference() -> None:
+    """Check that external HTML-based references are expanded or reported missing."""
+    url_map = {"ok": "ok.html#ok"}
+    source = "<autoref identifier=bar>foo</autoref> <autoref identifier=ok>ok</autoref>"
+    output, unmapped = fix_refs(source, url_map.__getitem__)
+    assert output == '[foo][bar] <a class="autorefs autorefs-internal" href="ok.html#ok">ok</a>'
+    assert unmapped == ["bar"]
+
+
+def test_legacy_custom_optional_reference() -> None:
     """Check that optional HTML-based references are expanded and never reported missing."""
     url_map = {"ok": "ok.html#ok"}
     source = '<span data-autorefs-optional="bar">foo</span> <span data-autorefs-optional=ok>ok</span>'
@@ -230,7 +239,16 @@ def test_custom_optional_reference() -> None:
     assert unmapped == []
 
 
-def test_custom_optional_hover_reference() -> None:
+def test_custom_optional_reference() -> None:
+    """Check that optional HTML-based references are expanded and never reported missing."""
+    url_map = {"ok": "ok.html#ok"}
+    source = '<autoref optional identifier="bar">foo</autoref> <autoref identifier=ok optional>ok</autoref>'
+    output, unmapped = fix_refs(source, url_map.__getitem__)
+    assert output == 'foo <a class="autorefs autorefs-internal" href="ok.html#ok">ok</a>'
+    assert unmapped == []
+
+
+def test_legacy_custom_optional_hover_reference() -> None:
     """Check that optional-hover HTML-based references are expanded and never reported missing."""
     url_map = {"ok": "ok.html#ok"}
     source = '<span data-autorefs-optional-hover="bar">foo</span> <span data-autorefs-optional-hover=ok>ok</span>'
@@ -242,7 +260,19 @@ def test_custom_optional_hover_reference() -> None:
     assert unmapped == []
 
 
-def test_external_references() -> None:
+def test_custom_optional_hover_reference() -> None:
+    """Check that optional-hover HTML-based references are expanded and never reported missing."""
+    url_map = {"ok": "ok.html#ok"}
+    source = '<autoref optional hover identifier="bar">foo</autoref> <autoref optional identifier=ok hover>ok</autoref>'
+    output, unmapped = fix_refs(source, url_map.__getitem__)
+    assert (
+        output
+        == '<span title="bar">foo</span> <a class="autorefs autorefs-internal" title="ok" href="ok.html#ok">ok</a>'
+    )
+    assert unmapped == []
+
+
+def test_legacy_external_references() -> None:
     """Check that external references are marked as such."""
     url_map = {"example": "https://example.com"}
     source = '<span data-autorefs-optional="example">example</span>'
@@ -251,6 +281,15 @@ def test_external_references() -> None:
     assert unmapped == []
 
 
+def test_external_references() -> None:
+    """Check that external references are marked as such."""
+    url_map = {"example": "https://example.com"}
+    source = '<autoref optional identifier="example">example</autoref>'
+    output, unmapped = fix_refs(source, url_map.__getitem__)
+    assert output == '<a class="autorefs autorefs-external" href="https://example.com">example</a>'
+    assert unmapped == []
+
+
 def test_register_markdown_anchors() -> None:
     """Check that Markdown anchors are registered when enabled."""
     plugin = AutorefsPlugin()
@@ -333,9 +372,17 @@ def test_register_markdown_anchors_with_admonition() -> None:
     }
 
 
-def test_keep_data_attributes() -> None:
+def test_legacy_keep_data_attributes() -> None:
     """Keep HTML data attributes from autorefs spans."""
     url_map = {"example": "https://e.com"}
     source = '<span data-autorefs-optional="example" class="hi ho" data-foo data-bar="0">e</span>'
     output, _ = fix_refs(source, url_map.__getitem__)
     assert output == '<a class="autorefs autorefs-external hi ho" href="https://e.com" data-foo data-bar="0">e</a>'
+
+
+def test_keep_data_attributes() -> None:
+    """Keep HTML data attributes from autorefs spans."""
+    url_map = {"example": "https://e.com"}
+    source = '<autoref optional identifier="example" class="hi ho" data-foo data-bar="0">e</autoref>'
+    output, _ = fix_refs(source, url_map.__getitem__)
+    assert output == '<a class="autorefs autorefs-external hi ho" href="https://e.com" data-foo data-bar="0">e</a>'