Add access to base_href and url_rewriter in tag rewriting

Author: benoit74

src/zimscraperlib/rewriting/html.py

@@ -178,7 +178,11 @@ def handle_starttag(self, tag: str, attrs: AttrsList, *, auto_close: bool = Fals
 
         if (
             rewritten := rules.do_tag_rewrite(
-                tag=tag, attrs=attrs, auto_close=auto_close
+                tag=tag,
+                attrs=attrs,
+                url_rewriter=self.url_rewriter,
+                base_href=self.base_href,
+                auto_close=auto_close,
             )
         ) is not None:
             self.send(rewritten)
@@ -468,6 +472,8 @@ def do_tag_rewrite(
         self,
         tag: str,
         attrs: AttrsList,
+        url_rewriter: ArticleUrlRewriter,
+        base_href: str | None,
         *,
         auto_close: bool,
     ) -> str | None:
@@ -484,6 +490,8 @@ def do_tag_rewrite(
                         for arg_name, arg_value in {  # pyright: ignore[reportUnknownVariableType]
                             "tag": tag,
                             "attrs": attrs,
+                            "url_rewriter": url_rewriter,
+                            "base_href": base_href,
                             "auto_close": auto_close,
                         }.items()
                         if arg_name in _cached_signature(rule.func).parameters

tests/rewriting/test_html_rewriting.py

@@ -1269,6 +1269,20 @@ def rewrite2_tag(
     )
 
 
+@rules.rewrite_tag()
+def rewrite3_tag(
+    tag: str,
+    base_href: str | None,
+    url_rewriter: ArticleUrlRewriter,
+) -> str | None:
+    if tag != "rewrite3":
+        return
+    rewriten_url = url_rewriter(
+        "https://www.acme.com/foo.img", base_href=base_href
+    ).rewriten_url
+    return f'<rewriten src="{rewriten_url}" />'
+
+
 @pytest.mark.parametrize(
     "tag, attrs, auto_close, expected_result",
     [
@@ -1300,6 +1314,13 @@ def rewrite2_tag(
             '<rewriten attr2="value2"/>',
             id="rewrite2_tag_auto_close",
         ),
+        pytest.param(
+            "rewrite3",
+            [("attr2", "value2")],
+            True,
+            '<rewriten src="https://www.acme.com/foo.img?queryparam" />',
+            id="rewrite3_use_url_rewriter",
+        ),
     ],
 )
 def test_html_tag_rewrite_rules(
@@ -1308,11 +1329,16 @@ def test_html_tag_rewrite_rules(
     *,
     auto_close: bool,
     expected_result: str | None,
+    simple_url_rewriter_gen: Callable[[str, str], ArticleUrlRewriter],
 ):
     assert (
         rules.do_tag_rewrite(
             tag=tag,
             attrs=attrs,
+            url_rewriter=simple_url_rewriter_gen(
+                "http://www.example.com", "?queryparam"
+            ),
+            base_href=None,
             auto_close=auto_close,
         )
         == expected_result