~95% on SimpleQA (e.g. Qwen3.6-27B on a 3090). Supports all local and cloud LLMs (llama.cpp, Ollama, Google, ...). 10+ search engines - arXiv, PubMed, your private documents. Everything Local & Encrypted.
Python academiaanthropicarxivbravedeep-researchencryptionhome-automationhomeserverlocallocal-deep-researchlocal-llmmistralollamaopenaipubmedresearchresearch-toolretrieval-augmented-generationsearxngself-hosted
"""Check the production image's NLTK data and normal upload parser offline."""
import io
import nltk
from docx import Document
from local_deep_research.document_loaders import load_from_bytes
def main() -> None:
# Both resources must be usable by the service user, with no download.
sentence = nltk.sent_tokenize("Otters build dams.")[0]
assert nltk.pos_tag(nltk.word_tokenize(sentence))
document = Document()
document.add_paragraph("Production parser round trip.")
source = io.BytesIO()
document.save(source)
records = load_from_bytes(source.getvalue(), ".docx", "check.docx")
assert [record.page_content for record in records] == [
"Production parser round trip."
]
assert records[0].metadata["original_filename"] == "check.docx"
for extension, source in {
".md": b"# Check\n\nProduction parser round trip.",
".xml": b"<root><p>Production parser round trip.</p></root>",
".eml": (
b"From: [email protected]\nTo: [email protected]\n"
b"Subject: Check\nContent-Type: text/html\n\n"
b"<p>Production parser round trip.</p>"
),
".enex": (
b"<en-export><note><content><![CDATA[<en-note><p>"
b"Production parser round trip."
b"</p></en-note>]]></content></note></en-export>"
),
}.items():
records = load_from_bytes(source, extension, f"check{extension}")
assert "Production parser round trip." in "\n".join(
record.page_content for record in records
)
print("Upload parser check passed")
if __name__ == "__main__":
main()