Whenever you feel like it @blackary!

Neither pip install non pip upgrade install the new version.

I have la last Verdon of streamlit installéd 1.16

As we implemented #24, it is now not straightforward to test the gallery while adding a new extra! Let's tackle that.

Make it be a Literal so it doesn't break in the playground

Would be great to have an extra that makes an app play well with GitHub, and:

• adds a sticky github button to reach the source code of an app Open in GitHub sorta
• optionally adds the . trigger to open github.dev, too

API:

• add_sticky_github_button(label="Open in GitHub", owner="arnaudmiribel", repository="streamlit-extras")
• add_sticky_github_button(label="Open in GitHub", repository="arnaudmiribel/streamlit-extras")

Hi, i have recently installed streamlit_extras, but after importing the library and try to run the switch page function, i encountered a problem where it did not run. Checking the source code return nothing in init.py file. I have uninstalled then installed the package again, but the problem is still there.

Im currently running

• Python 3.7
• streamlit 1.15.2
• Windows 11

Could solve streamlit/streamlit#949 with a simple extra -- returns None if you select the "SELECT AN ITEM" thing, customizable label, otherwise just vanilla selectbox

Maybe deprecate + better advocate for markdownlit

Figured it could be nice to add this kind of badge on top of repositories readme/forum posts we eventually integrate inside streamlit-extras!

Since <style> blocks take vertical space in the app, avoid loading CSS each and every time a mention() is called! Call it only once.

Fun little addition to Altex to have mini charts

https://playground.streamlit.app?q=minispark

Code

import yfinance
import pandas as pd 
import yfinance as yf
from streamlit_extras.altex import sparkline_chart
import altair as alt 

msft = yf.Ticker("MSFT")
data = msft.history().reset_index()

def minispark(data: pd.DataFrame, x: str = "record_date:T", y: str = "cost") -> None:
    sparkline_chart(
        data=data,
        x=x,
        y=alt.Y(
            y + ":Q",
            scale=alt.Scale(domain=[data[y].min(), data[y].max()]),
        ),
        height=80,
    )
    
left, middle, right = st.columns(3)
with left:
    st.metric("Average Open", int(data["Open"].mean()))
    minispark(data, x="Date", y="Open")
with middle:
    st.metric("Average Close", int(data["Close"].mean()))
    minispark(data, x="Date", y="Close")
with right:
    st.metric("Average Volume", int(data["Volume"].mean()))
    minispark(data, x="Date", y="Open")

@sfc-gh-brianhess looks like this example has a bug!

Hi,

when I check the checkbox "I want to explore", the site reruns but nothing more happens.

Reproducible even on your demo page:
https://extras.streamlit.app/Dataframe%20explorer%20UI

I am sure version 0.2.2 worked fine.

It would be nice to have an option to use a local file in the project as the app_logo.

It is described how to encode the data for use at stackoverflow:
https://stackoverflow.com/questions/73251012/put-logo-and-title-above-on-top-of-page-navigation-in-sidebar-of-streamlit-multi

Right now I have used a hack like this:

def get_base64_of_bin_file(png_file):
    with open(png_file, "rb") as f:
        data = f.read()
    return base64.b64encode(data).decode()

logo = "data:image/png;base64," + get_base64_of_bin_file("project/data/logo.png")

add_logo(logo)

Could have two folders, one for actual extras, one for the gallery app (and the gallery app could have a separate, tighter set of dependencies).

That way we can have much looser requirements on the package itself.

See https://www.notion.so/streamlit/Diagrams-Streamlit-870ea95f1f3742f3a200054c94cadc6a

As seen in https://twitter.com/andrejusb/status/1578099977036779520?s=46&t=MFCVxxXAuuu-icifGbq5KA and see slack thread

Add a use_checkbox: bool = True parameter to the dataframe explorer to optionally omit the checkbox and directly unlock the filtering

https://facelessuser.github.io/pymdown-extensions

Hi,

I get following exception:
Traceback (most recent call last):
File "c:/Users/Public/Documents/BankAccount/test_drop.py", line 2, in
from streamlit_extras.dataframe_explorer import dataframe_explorer
File "C:\Python38\lib\site-packages\streamlit_extras\dataframe_explorer_init_.py", line 16, in
def dataframe_explorer(df: pd.DataFrame) -> pd.DataFrame:
File "C:\Python38\lib\site-packages\streamlit_extras_init_.py", line 41, in extra
submodule = filename.split("/")[-2]
IndexError: list index out of range

My code:

from streamlit_extras.dataframe_explorer import dataframe_explorer

Yes, 1 line.

Used versions:
Python 3.8.5

streamlit 1.15.1
streamlit-extras 0.2.2

App: https://github.com/streamlit/example-app-time-series-annotation

See staging app https://arnaudmiribel-streamlit-extras-gallerystreamlit-app-2n3dja.streamlitapp.com/Badges

Replace this

__func__ = get_streamlit_faker
__title__ = "Streamlit Faker"
__desc__ = "Fake Streamlit commands at the speed of light! Great for prototyping apps."
__icon__ = "🥷"
__examples__ = [example]
__author__ = "Arnaud Miribel"
__github_repo__ = "arnaudmiribel/streamlit-faker"
__pypi_name__ = "streamlit-faker"
__package_name__ = "streamlit_faker"
__experimental_playground__ = False

With

from common import set_extra_gallery_config

extra_gallery_config: dict = set_extra_gallery_config(
  func=get_streamlit_faker,
  title="Streamlit Faker",
  desc="Fake Streamlit commands at the speed of light! Great for prototyping apps.",
  ...
)

And consume that extra_gallery_config variable in gallery/streamlit_app.py.

Main perks:

• canonical definition of all potential config keys in set_extra_gallery_config
• auto-complete unlocked for all extra configs config
• makes it clear that this part in the init.py is for the gallery and has no impact whatsoever in the inner workings of the extra.

As in https://playground.streamlitapp.com/?q=github-mention
Worth trying with other pages, other social networks too

Just takes markdown, or perhaps markdownlit

• Handle unhandled arguments by skipping them, rather than defaulting to a string
• Allow ability to st.write outputs of functions

e.g. json.loads should be usable :)

See thread

Potentials ideas:

• Content
• Some sort of dictionary specifying input elements (e.g. {"username": "text", "password": "password"})
• Returns the value of input elements on submit

(It requires streamlit-extras, not sure if that's gonna raise troubles)

Hi all,

Is there a way to pass an argument with the switch_page function?
Given the fact that the function can't be inserted in the "on-click" parameter of a button it would be quite useful to have the possibility to pass some arguments. Could be this a possible ER? Or is there already a way to perform this action?

Thank you
Regards,
Lorenzo

Which always returns a tuple of dates

There is a problem with the Rerun function. It has been renamed in the new version and it's raising problems with the switch_page functionality (That has not been updated in the new version of streamlit-extras). Is there any plan to fix this issue? Thank you.

[Workaround for the moment is to downgrade streamlit at 1.13.0 and keep working with that one]

gather_metrics() got an unexpected keyword argument 'name'

• Code: https://github.com/pytorch/captum/blob/ff2b403965f99f7ff9eb64cdf9370c1044390080/captum/attr/_utils/visualization.py#L815
• App: https://sgoede-sentiment-explained-sentiment-explained-0n8uqj.streamlitapp.com/

Thanks for the great package!

I came across the following error when using switch_page_button, e.g. switch_page("Page2")

ModuleNotFoundError: No module named 'streamlit.runtime'

Traceback: from streamlit.runtime.scriptrunner import RerunData, RerunException

See in/out examples echo in https://markdownlit.streamlitapp.com/ - I often find it useful to disambiguate what's in and what's out! Also looks familiar for Jupyter users

Maybe better to just use a same set of datasets throughout all extras and have them locally in the repo, instead of fetching all around

Would be good to at least test that every component has the required __x__ attributes

Hi All, fantastic addons! I try them on Streamlit Cloud, but unable to install on local Python 3.10, seems like 3.9 works. Is there a solution? Thanks!