src/reqs_optional/requirements_optional_langchain.txt

# optional for chat with PDF
langchain==0.0.300
pypdf==3.14.0
# avoid textract, requires old six
#textract==1.6.5
pypdfium2==4.19.0

# for HF embeddings
sentence_transformers==2.2.2

# optional: for OpenAI endpoint or embeddings (requires key)
openai==0.27.8
replicate==0.10.0

# local vector db
chromadb==0.4.10

# chroma migration
chroma-migrate==0.0.7
duckdb==0.7.1
https://h2o-release.s3.amazonaws.com/h2ogpt/chromamigdb-0.3.25-py3-none-any.whl
https://h2o-release.s3.amazonaws.com/h2ogpt/hnswmiglib-0.7.0.tgz

# server vector db
#pymilvus==2.2.8

# weak url support, if can't install opencv etc. If comment-in this one, then comment-out unstructured[local-inference]==0.6.6
# unstructured==0.8.1

# strong support for images
# Requires on Ubuntu: sudo apt-get install libmagic-dev poppler-utils tesseract-ocr libtesseract-dev libreoffice
unstructured[local-inference]==0.9.0
#pdf2image==1.16.3
#pytesseract==0.3.10
pillow==9.5.0
posthog==3.0.1

pdfminer.six==20221105
urllib3
requests_file

#pdf2image==1.16.3
#pytesseract==0.3.10
tabulate==0.9.0
# FYI pandoc already part of requirements.txt

# JSONLoader, but makes some trouble for some users
# TRY: apt-get install autoconf libtool
# unclear what happens on windows/mac for now
jq==1.4.1; platform_machine == "x86_64"

# to check licenses
# Run: pip-licenses|grep -v 'BSD\|Apache\|MIT'
pip-licenses==4.3.0

# weaviate vector db
weaviate-client==3.22.1