How to specify required System Dependencies when deploying via Streamlit Cloud

[nhtkid]

I love unstructured.io. It is so powerful and so easy to use.
I literally only need two lines of code to load different type of data.

from langchain.document_loaders import UnstructuredFileLoader
loader = UnstructuredFileLoader("state_of_the_union.txt")
loader.load()

But when I deploy my code via Streamlit Cloud, how do I specify these following system dependencies?

Install the following system dependencies if they are not already available on your system. Depending on what document types you're parsing, you may not need all of these.
libmagic-dev (filetype detection)
poppler-utils (images and PDFs)
tesseract-ocr(images and PDFs)
libreoffice (MS Office docs)
pandoc (EPUBs)

Thanks!

[nhtkid]

I have created a packages.txt based on the streamlit recommendation below with all the required packages.
https://docs.streamlit.io/streamlit-community-cloud/deploy-your-app/app-dependencies

ibmagic-dev poppler-utils tesseract-ocr libreoffice pandoc
It looks like the deployment is successfully. However, I could only upload pdf/docx/txt.
If I upload pptx or png, I keeps getting (IndexError: list index out of range).

I am not sure whether the loader is not able to load, or after loading, it was not able to convert to embeddings.
FYI I am using OpenAI embeddings and ChromaDB.

[cragwolfe]

If I upload pptx or png, I keeps getting (IndexError: list index out of range).

My best guess is file detection went awry. Also, typo in your first line, missing the l inlibmagic-dev.

To test theory, I'd try unstructured/partition/pptx.py::partition_pptx and unstructured/partition/image.py::partition_image directly.