fix(classifiers): guard against content=None in DocumentLanguageClassifier (fixes #11418)

[devteamaegis]

Summary

Fixes #11418 — DocumentLanguageClassifier crashes with TypeError when a Document has content=None.

Before:

def _detect_language(self, document: Document) -> str | None:
    language = None
    try:
        language = langdetect.detect(document.content)  # TypeError if content is None
    except langdetect.LangDetectException:
        ...

langdetect.detect(None) raises TypeError, which is not caught by the LangDetectException handler. The exception propagates to the caller, crashing the pipeline. This affects any blob-only Document (e.g. images, PDFs loaded without text extraction) since Document.content is explicitly allowed to be None.

After: an explicit None guard is added before calling langdetect.detect(). Documents with content=None log a warning and return None (which causes run() to route them to "unmatched"), consistent with existing behaviour for text that langdetect fails to detect.

Changes

haystack/components/classifiers/document_language_classifier.py

• Added if document.content is None: guard at the top of _detect_language
• Logs a warning including the document ID (same pattern as the LangDetectException branch)
• Returns None so the caller routes the document to "unmatched"

test/components/classifiers/test_document_language_classifier.py
Three new tests:

Test	Assertion
test_content_none_does_not_raise	run([Document(content=None)]) must not raise; document gets language="unmatched"
test_content_none_emits_warning	A warning containing the document ID is logged
test_mixed_none_and_text_content	Batch with a None-content doc and a text doc both classified correctly

Test plan

• All 10 tests in test_document_language_classifier.py pass: uv run --with langdetect --with pytest python -m pytest test/components/classifiers/test_document_language_classifier.py -v → 10 passed

[vercel]

@devteamaegis is attempting to deploy a commit to the deepset Team on Vercel.

A member of the Team first needs to authorize it.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[sjrl]

Closing as duplicate of #11419