fix: handle Document with content=None in DocumentLanguageClassifier

haystack/components/classifiers/document_language_classifier.py

@@ -110,11 +110,15 @@ def run(self, documents: list[Document]) -> dict[str, list[Document]]:
         return {"documents": new_documents}
 
     def _detect_language(self, document: Document) -> str | None:
-        language = None
+        if document.content is None:
+            logger.warning(
+                "Langdetect cannot detect the language of Document with id: {document_id}", document_id=document.id
+            )
+            return None
         try:
-            language = langdetect.detect(document.content)
+            return langdetect.detect(document.content)
         except langdetect.LangDetectException:
             logger.warning(
                 "Langdetect cannot detect the language of Document with id: {document_id}", document_id=document.id
             )
-        return language
+        return None

releasenotes/notes/document-language-classifier-none-content-a245d044b02a19b0.yaml

@@ -0,0 +1,5 @@
+---
+fixes:
+  - |
+    Prevent DocumentLanguageClassifier from crashing when ``Document.content=None`` by
+    marking them as unmatched and logging a warning.

test/components/classifiers/test_document_language_classifier.py

@@ -50,3 +50,10 @@ def test_warning_if_no_language_detected(self, caplog):
             classifier = DocumentLanguageClassifier()
             classifier.run(documents=[Document(content=".")])
             assert "Langdetect cannot detect the language of Document with id" in caplog.text
+
+    def test_none_content_is_unmatched(self, caplog):
+        with caplog.at_level(logging.WARNING):
+            classifier = DocumentLanguageClassifier()
+            result = classifier.run(documents=[Document(content=None)])
+            assert result["documents"][0].meta["language"] == "unmatched"
+            assert "Langdetect cannot detect the language of Document with id" in caplog.text