tvst / plost Goto Github PK

View Code? Open in Web Editor NEW

219.0 3.0 13.0 47 KB

A deceptively simple plotting library for Streamlit

License: Apache License 2.0

Python 98.70% Makefile 1.30%

plost's Introduction

🍅 Plost

A deceptively simple plotting library for Streamlit.

Because you've been writing plots wrong all this time.

👇 THE REAL README IS ACTUALLY HERE:

👆 You can find interactive examples, documentation, and much more in the app above.

Our goal

What you need 99% of the time is insanely easy
The other 1% is impossible. Use Vega-Lite instead!

Getting started

pip install plost

Basics

Plost makes it easy to build common plots using the Vega-Lite library but without having to delve into Vega-Lite specs (unless you're doing something tricky), and without having to melt your DataFrame from long format to wide format (the bane of most Vega-Lite plots!)

For example, let's say you have a "long-format" table like this:

time	stock_name	stock_value
...	stock1	1
...	stock2	2
...	stock1	100
...	stock2	200

Then you can draw a line chart by simply calling line_chart() with some column names:

import plost

plost.line_chart(
  my_dataframe,
  x='time',  # The name of the column to use for the x axis.
  y='stock_value',  # The name of the column to use for the data itself.
  color='stock_name', # The name of the column to use for the line colors.
)

Simple enough! But what if you instead have a "wide-format" table like this, which is super common in reality:

time	stock1	stock2
...	1	100
...	2	200

Normally you'd have to melt() the table with Pandas first or create a complex Vega-Lite layered plot. But with Plost, you can just specify what you're trying to accomplish and it will melt the data internally for you:

import plost

plost.line_chart(
  my_dataframe,
  x='time',
  y=('stock1', 'stock2'),  # 👈 This is magic!
)

Ok, now let's add a mini-map to make panning/zooming even easier:

import plost

plost.line_chart(
  my_dataframe,
  x='time',
  y=('stock1', 'stock2'),
  pan_zoom='minimap',  # 👈 This is magic!
)

But we're just scratching the surface. Basically the idea is that Plost allows you to make beautiful Vega-Lite-driven charts for your most common needs, without having to learn about the powerful yet complex language behind Vega-Lite.

Check out the the sample app / docs for a taste of other amazing things you can do!

Juicy examples

Check out the documentation app!

Documentation

This is in the documentation app too!

plost's People

Contributors

Stargazers

Watchers

Forkers

techventurebuilder tanglespace jrieke techthiyanes algonacci alsou-eqn charlywargnier edisplay lukasmasuch devicemxl i6213546 kokkulaumakanth wonchul

plost's Issues

Bug Report Bar Chart

Jo @tvst ,

think i found a bug.
What i tried is:

plost.bar_chart(
    data=datasets['stocks'],
    bar='company',
    value=['q2', 'q3'],
    direction='horizontal'
)

But values of q3 get stacked on top of q2 even though the
values of q3 should be plotted as part of q2 as the value of q2 is higher.
For vertical it works as expected.

Could you double check and in case fix that?

Best regards
Chris

Feature request

Hi @tvst ,

could you please add an easier way to pass individual color values?
Why I am asking is, i have bar_chart and i want to add a color for each value in the bar chart:

plost.bar_chart(
data=df, 
bar="epic", 
value=["time_1", "time_2", "time_3"],
direction="horizontal", 
colors=["#0001","#0002","#0003"]
)

So each bar has the three times in a different color for all unquie values in epic.
Also i saw you used the param legend=None but never passed any value, how could i use it to specifiy the x-axis value?

Best Regards
Chris

color attribute not supported

In the current version of plost 0.2.5, color attribute of every chart type is not supported.

Looking in the code, there seem to be a lack of passing through the value of the color argument.

One easy fix is to replace every

spec = D(
        mark=D(type='circle', tooltip=True),
        encoding=D(
            x=_clean_encoding(data, x),
            y=y_enc,
            color=color_enc,
            size=_clean_encoding(data, size, legend=legend),
            opacity=_clean_encoding(data, opacity, legend=legend),
        ),
        selection=_get_selection(pan_zoom),
    )

spec = D(
        mark=D(type='circle', tooltip=True),
        encoding=D(
            x=_clean_encoding(data, x),
            y=y_enc,
            color=_clean_encoding(data, color),
            size=_clean_encoding(data, size, legend=legend),
            opacity=_clean_encoding(data, opacity, legend=legend),
        ),
        selection=_get_selection(pan_zoom),
    )

The change is in the line color=color_enc to color=_clean_encoding(data, color).

Over resource limits on Streamlit Cloud

Hey there 👋 Just wanted to let you know that your app on Streamlit Cloud deployed from this repo has gone over its resource limits. Access to the app is temporarily limited. Visit the app to see more details and possible solutions.

Over resource limits on Streamlit Cloud

how tricky is it to implement two-way communication between plots and python?

Thank you for this great library 👏 !
I'm wondering how difficult it is to add two-way communication between chart objects and streamlit in order to update streamlit applications based upon click events, selections, etc... on charts. From some preliminary digging it seems like it's pretty non-trivial but want to be sure I'm not missing something obvious.
I am interested in working on this but it would be great to get some understanding of the level of challenge first.

Line Chart with Dots

Hi @tvst ,

can i somehow add dots to the linechart?
Plus, can i ushed dashed lines?

Best regards
Chris

Failed to open tvst/post

https://share.streamlit.io/tvst/plost

Oh no.
Error running app. If this keeps happening, please contact support.

Certain legends are black-on-dark in dark mode

Hi, not a user, just a passerby, but I noticed that on dark mode of the demo site some of the legends/labels are black (whereas some are, correctly, white).

For example, the q2 and q3 in the below screenshot (whereas the individual bar legends are correct).

Why do I need to explictly assign bar/value/x/y args for every plost funcs

I'm trying to plot a chart via plost instead of streamlit, this code is fine:
st.bar_chart(df)
it uses index as bar index, each columns as values, however this code raises an error:
plost.bar_chart(df)
TypeError: bar_chart() missing 2 required positional arguments: 'bar' and 'value'
I need to change the code like this:
plost.bar_chart(df.reset_index(names=['index']),bar='index',value=list(df.columns))
It is weird and no need to copy and change the df object at the first place, why do I need to explictly assign bar/value/x/y args for every plost funcs, why don't you just use the index and whole columns as value content to run the render as default?