How can I use variables/dataframe generated in streamlit app to be used by Quarto in generating html report in streamlit ?

[johnsnow09]

Description

I have a python streamlit webapp link running and I want to add a feature to it to allow users to download html reports from it. The downloadable html should be able to use variables, dataframe, plotly charts created in streamlit using .py files running in the vm. Is it possible to use Quarto for this and how can I use those variables from streamlit to Quarto. Will there be any thing extra I need to do to use those variables, dataframe by name.
For example if there is pandas dataframe st.session_state.data_final (st represents streamlit, session_state is to pass variables between pages in session and data_final is the dataframe name) will this be directly loaded/read/utilized within Quarto by mentioning its name st.session_state.data_final ?

Is there any blog or demo or document available related to this ?

[mcanouil]

To be sure I understood what you want to do:
You want to deploy a Streamlit app that can run Quarto CLI using some values/parameters coming straight from your app. This means your app is running on a server that has Quarto CLI installed and all possible dependencies (Jupyter engine?). Then your app needs to be able to execute Quarto CLI and retrieve the rendered file.

I'm afraid, you need to develop this yourself and it is unlikely you'll find anything about this specifically, but you might find things online about the key principle to execute OS command from the app (likely via subprocess)

process = subprocess.Popen(["quarto", "..."], ...)

[johnsnow09]

I understand I can send parameter value to the qmd doc along with output flag quarto render test-6.qmd -P dummy_var:5 --output test6_cmd_out.html but how can I send dataframe object created in streamlit to qmd. I need to send dataframe object to qmd without saving it in directory due to session constrains and If I am able to do this then it is all sorted out for me otherwise I will have stop giving any more time to Quarto and move to alternatives.

Thanks for the help !!!

[mcanouil]

You need to figure out a way to represent the data frame in another way that works in Quarto CLI flags (YAML,json,base64,etc).

[johnsnow09]

Ok, converting to json or YAML shouldn't be a problem.

I am using polars dataframe so I can do
df_json = df.write_json() or df_yaml = yaml.dump(df.to_dict(orient='records'),default_flow_style=None)

So now can I just pass json and yaml file object with it as shown in below commands or do i need to save them as files first? Because saving as file is an issue for me so I cant do that.

quarto render test-6.qmd --execute-param df_json --output test6_cmd_out.html
or
quarto render test-6.qmd --execute-param df_yaml --output test6_cmd_out.html

---

2. If 1st step is done then can I just use below line of code to access the json file contents in qmd ?

df_json

But Getting error: ERROR: YAML file df_json not found.

If this is not correct then can you show me the correct way. Would really Appreciate your help !!

[mcanouil]

Again that's up to you to implement what you want as you want.

Parameters can be pass to Quarto as file or string from the command line.

Note that you are now aware of what Quarto itself can do with the flags or YAML frontmatter in a document.

If you use Python, you get access to everything that Python allows you to do.

Now your question has nothing to do with Quarto and is more about coding what you want.
I'm afraid I don't have the time to think and code any solution for your very specific problem which is not really a Quarto problem.

[johnsnow09]

Sure,thanks !!

First in my earlier post I was told about Anaconda problems when I was facing Quarto CLI installation issues through pip but was not informed about installation through Conda forge . I deleted Anaconda after that and then later got to know it could have also been installed through Conda forge which I eventually did. Deleting Anaconda messed up my system and environments which I am still fixing after days.

Now when I was expecting some support in getting to see some example or 1 liner of code on how it can actually be implemented unfortunately it was not there. Even in below post for R exactly same issue was raised and 1 liner code was provided to show how it can be done as a demo but not in my case.
https://forum.posit.co/t/how-to-access-r-sessions-variables-from-inside-quarto-file/163019/2

If that's how the community support is provided where I end up wasting days then I think its better to stop at the beginning of its use where learning curve is at peak but point of no support means to better look for alternatives with more community support.

Thanks for all the replies !!!