feat(checks): report uninteded list conversion in rst

docs/changes.rst

@@ -9,6 +9,7 @@ Weblate 5.15.2
 
 * :ref:`addon-weblate.generate.generate` is now triggered upon installation.
 * Screenshots updated from the repository have proper history.
+* :ref:`check-rst-syntax` now reports unintended list conversion.
 
 .. rubric:: Bug fixes
 

docs/locales/ja/LC_MESSAGES/docs.po

@@ -48921,7 +48921,7 @@ msgstr ""
 
 #: ../../security/passwords.rst:33
 msgid "Social or third-party authentication"
-msgstr "- ソーシャル認証またはサードパーティ認証"
+msgstr "ソーシャル認証またはサードパーティ認証"
 
 #: ../../security/passwords.rst:35
 msgid ""

weblate/checks/markup.py

@@ -114,6 +114,8 @@
     re.compile(r"""Interpreted text role "([^"]*)" not implemented\."""),
 ]
 
+RST_LIST_START = ("- ", "* ", "+ ")
+
 
 def strip_entities(text):
     """Strip all HTML entities (we don't care about them)."""
@@ -418,7 +420,7 @@
 def extract_rst_references(text: str) -> tuple[dict[str, str], Counter, list[str]]:
     memo = SimpleNamespace()
     publisher = get_rst_publisher()
     document = utils.new_document("", publisher.settings)
     memo.reporter = document.reporter
     memo.document = document
     memo.language = languages.get_language(
@@ -492,7 +494,7 @@
             gettext("The following reStructuredText markup is extra: {}"), values
         )
 
     def check_single(
         self, source: str, target: str, unit: Unit
     ) -> bool | MissingExtraDict:
         src_references, src_set, _alltags = extract_rst_references(source)
@@ -531,7 +533,7 @@
         ):
             if isinstance(result, dict):
                 for key, value in result.items():
                     results[key].extend(value)
         if results:
             errors.extend(self.format_result(results))
         if errors:
@@ -577,9 +579,9 @@
 @lru_cache(maxsize=512)
 def validate_rst_snippet(
     snippet: str, source_tags: tuple[str] | None = None
-) -> tuple[list[str], list[str]]:
+) -> tuple[tuple[str, ...], tuple[str, ...]]:
     publisher = get_rst_publisher()
     document = utils.new_document("", publisher.settings)
 
     errors: list[str] = []
     roles: list[str] = []
@@ -632,7 +634,7 @@
         transform = transform_class(transformer.document, startnode=pending)
         transform.apply(**kwargs)
         transformer.applied.append((priority, transform_class, pending, kwargs))
-    return errors, roles
+    return tuple(errors), tuple(roles)
 
 
 class RSTSyntaxCheck(RSTBaseCheck):
@@ -640,11 +642,18 @@
     name = gettext_lazy("reStructuredText syntax error")
     description = gettext_lazy("reStructuredText syntax error in the translation.")
 
     def check_single(
         self, source: str, target: str, unit: Unit
     ) -> bool | MissingExtraDict:
         _errors, source_roles = validate_rst_snippet(source)
-        errors, _target_roles = validate_rst_snippet(target, tuple(source_roles))
+        rst_errors, _target_roles = validate_rst_snippet(target, source_roles)
+        errors = list(rst_errors)
+
+        # This is valid RST, but might mess up the document
+        if not source.startswith(RST_LIST_START) and target.startswith(RST_LIST_START):
+            errors.append(
+                gettext("The translation should not start with a list marker.")
+            )
 
         if errors:
             return {"errors": errors}
@@ -662,7 +671,7 @@
         ):
             if isinstance(result, dict):
                 for key, value in result.items():
                     results[key].extend(value)
         if results:
             errors.extend(self.format_result(results))
         if errors:

weblate/checks/tests/test_markup_checks.py

@@ -93,7 +93,7 @@
     def test_skip_mixed(self) -> None:
         self.do_test(
             False,
             (
                 ["<emphasis>1st</emphasis>", "<invalid>"],
                 "<emphasis>not</ emphasis>",
                 "",
@@ -413,7 +413,7 @@
         )
         check = Check(unit=unit)
         self.assertHTMLEqual(
             self.check.get_description(check),
             """
             The following reStructuredText markup is missing:
             <span class="hlcheck" data-value=":ref:`bar`">:ref:`bar`</span>,
@@ -888,7 +888,7 @@
         )
         check = Check(unit=unit)
         self.assertHTMLEqual(
             self.check.get_description(check),
             """
             The following errors were found:<br>
             Inline interpreted text or phrase reference start-string without end-string.
@@ -904,3 +904,61 @@
                 "rst-text",
             ),
         )
+
+    def test_list(self) -> None:
+        self.do_test(
+            False,
+            (
+                "Text",
+                "Text",
+                "rst-text",
+            ),
+        )
+        self.do_test(
+            True,
+            (
+                "Text",
+                "* Text",
+                "rst-text",
+            ),
+        )
+        self.do_test(
+            True,
+            (
+                "Text",
+                "- Text",
+                "rst-text",
+            ),
+        )
+        self.do_test(
+            True,
+            (
+                "Text",
+                "+ Text",
+                "rst-text",
+            ),
+        )
+        self.do_test(
+            False,
+            (
+                "- Text",
+                "- Text",
+                "rst-text",
+            ),
+        )
+        self.do_test(
+            False,
+            (
+                "- Text",
+                "* Text",
+                "rst-text",
+            ),
+        )
+        self.do_test(
+            False,
+            (
+                "- Text",
+                "+ Text",
+                "rst-text",
+            ),
+        )