Fix gettext skipping MyST table cells inside tab-item

CHANGELOG.md

@@ -2,6 +2,9 @@
 
 ## Unreleased
 
+- 🐛 FIX: Table cells inside a MyST ``tab-item`` are now extracted for gettext
+  and translated, rather than being omitted from the ``.pot`` because they
+  lacked ``source`` after nested parse ({issue}`234`)
 - ♻️ IMPROVE: Replace the Sass/Node build with a dependency-free Python CSS
   generator (`tools/generate_css.py` driven by `style/design.toml` and
   hand-authored `style/*.css`); `package.json` is gone. The compiled

sphinx_design/tabs.py

@@ -132,6 +132,7 @@ def run_with_defaults(self) -> list[nodes.Node]:
             classes=["sd-tab-content", *self.options.get("class-content", [])],
         )
         self.state.nested_parse(self.content, self.content_offset, tab_content)
+        _propagate_table_source(tab_content)
         tab_item += tab_content
 
         return [tab_item]
@@ -197,6 +198,27 @@ def run_with_defaults(self) -> list[nodes.Node]:
         return [tab_set]
 
 
+def _propagate_table_source(root: nodes.Node) -> None:
+    """Copy source/line from a table onto descendants that lack them.
+
+    MyST's table renderer sets source on the ``table`` node but not on cell
+    paragraphs. At document level those cells inherit
+    ``document.current_source``; during ``nested_parse`` that is ``None``,
+    so Sphinx's gettext extractor skips them (``is_translatable`` requires
+    ``node.source``). See https://github.com/executablebooks/sphinx-design/issues/234
+    """
+    for table in root.findall(nodes.table):
+        source = table.source
+        if not source:
+            continue
+        line = table.line
+        for child in table.findall(nodes.Element):
+            if child.source is None:
+                child.source = source
+            if child.line is None:
+                child.line = line
+
+
 class sd_tab_input(nodes.Element, nodes.General):  # noqa: N801
     pass
 

tests/test_misc.py

@@ -504,6 +504,78 @@ def test_button_i18n_translated(sphinx_builder):
     assert badges[0].astext() == "stable"
 
 
+TAB_TABLE_MYST = """\
+# Heading
+
+::::{tab-set}
+
+:::{tab-item} Label1
+Content1
+
+| UniqueHeaderA | UniqueHeaderB |
+| --- | --- |
+| UniqueCellC | UniqueCellD |
+:::
+
+::::
+"""
+
+
+@pytest.mark.skipif(not MYST_INSTALLED, reason="myst-parser not installed")
+def test_tab_table_i18n_gettext(sphinx_builder):
+    """Table cells inside a MyST tab-item must be extracted for gettext.
+
+    See https://github.com/executablebooks/sphinx-design/issues/234
+    """
+    builder = sphinx_builder("gettext")
+    builder.src_path.joinpath("index.md").write_text(TAB_TABLE_MYST, encoding="utf8")
+    builder.build()
+    pot = (builder.out_path / "index.pot").read_text(encoding="utf8")
+    for msgid in (
+        "Heading",
+        "Label1",
+        "Content1",
+        "UniqueHeaderA",
+        "UniqueHeaderB",
+        "UniqueCellC",
+        "UniqueCellD",
+    ):
+        assert f'msgid "{msgid}"' in pot, pot
+
+
+@pytest.mark.skipif(not MYST_INSTALLED, reason="myst-parser not installed")
+def test_tab_table_i18n_translated(sphinx_builder):
+    """Translated table cells inside a tab-item must appear in the HTML output.
+
+    See https://github.com/executablebooks/sphinx-design/issues/234
+    """
+    builder = sphinx_builder(
+        conf_kwargs={
+            "extensions": ["myst_parser", "sphinx_design"],
+            "myst_enable_extensions": ["colon_fence"],
+            "language": "de",
+            "locale_dirs": ["locales"],
+        }
+    )
+    builder.src_path.joinpath("index.md").write_text(TAB_TABLE_MYST, encoding="utf8")
+    catalog = Catalog(locale="de", domain="index")
+    catalog.add("UniqueHeaderA", "KopfA")
+    catalog.add("UniqueHeaderB", "KopfB")
+    catalog.add("UniqueCellC", "ZelleC")
+    catalog.add("UniqueCellD", "ZelleD")
+    mo_dir = builder.src_path / "locales" / "de" / "LC_MESSAGES"
+    mo_dir.mkdir(parents=True)
+    with (mo_dir / "index.mo").open("wb") as handle:
+        write_mo(handle, catalog)
+
+    builder.build()
+    html = (builder.out_path / "index.html").read_text(encoding="utf8")
+    for translated in ("KopfA", "KopfB", "ZelleC", "ZelleD"):
+        assert translated in html, html
+    for original in ("UniqueHeaderA", "UniqueHeaderB", "UniqueCellC", "UniqueCellD"):
+        assert original not in html, html
+
+
 INVALID_CONFIG_VALUES = {
     "custom_directives": (["not", "a", "dict"], "must be a dictionary"),
     "fontawesome_source": ("invalid", "must be one of"),