Check language of VOT document

If an English translation isn’t yet available, requesting the English translation will return the French original. This checks whether the returned document actually is the English version. If not, it raises a `NoWorkingUrlError` exception, which will be treated in the same way as if the EP website had returned a 404 (i.e. the scraper will be retried later).

Fixes #1167

Author: tillprochaska

backend/howtheyvote/scrapers/votes.py

@@ -19,7 +19,7 @@
     VotePosition,
     VoteResult,
 )
-from .common import BeautifulSoupScraper, RequestCache, ScrapingError
+from .common import BeautifulSoupScraper, NoWorkingUrlError, RequestCache, ScrapingError
 from .helpers import (
     fill_missing_by_reference,
     normalize_name,
@@ -363,6 +363,23 @@ def _url(self) -> str:
         return f"{self.BASE_URL}/PV-{self.term}-{date}-VOT_EN.xml"
 
     def _extract_data(self, doc: BeautifulSoup) -> Iterator[Fragment | None]:
+        root = doc.select_one("file")
+
+        if not root:
+            raise ScrapingError("Missing root element `file` in VOT list")
+
+        # https://github.com/python/typeshed/issues/8755
+        language = cast(str, root["language"]).lower()
+
+        if language != "en":
+            # If an English translation isn’t yet available, requesting the English translation
+            # will return the French original. In case a French document is returned, we raise
+            # `NoWorkingUrlError`. Pipelines catching this exception will usually be re-run
+            # later (rather than being marked as permanently failed).
+            raise NoWorkingUrlError(
+                "Request English version of document, but received language {language}."
+            )
+
         for vote_tag in doc.select("votes vote"):
             # The source data often contains sections with additional information (such as
             # corrections). These are also modeled as "votes" (even though there was no