Lightweight plotting to the terminal. 4x resolution via Unicode.
When working with production data science code it can be handy to have plotting tool that does not rely on graphics dependencies or works only in a Jupyter notebook.
The use case that this was built for is to have plots as part of your data science / machine learning CI pipeline - that way whenever something goes wrong, you get not only the error and backtrace but also plots that show what the problem was.
- Unicode drawing, so 4x the resolution (pixels) of usual ASCII plots
- Super simple API
- Interactive mode (pass
interactive=True) - Color mode (pass
color=True) useful in particular when plotting multiple series - It's fast: Plotting 1M data points takes 100ms thanks to NumPy magic
- Only one dependency: NumPy (but you have that anyway don't you)
Please note that Unicode drawing will work correctly only when using a font that fully supports the Box-drawing character set. Please refer to this page for a (incomplete) list of supported fonts.
Note that all the examples are without color and plotting only a single series od data. For using color see the GIF example above.
import math
x = [math.sin(i/20)+i/300 for i in range(600)]
from uniplot import plot
plot(x, title="Sine wave")
Result:
Sine wave
┌────────────────────────────────────────────────────────────┐
│ ▟▀▚ │
│ ▗▘ ▝▌ │
│ ▗▛▜▖ ▞ ▐ │
│ ▞ ▜ ▗▌ ▌ │ 2
│ ▟▀▙ ▗▘ ▝▌ ▐ ▜ │
│ ▐▘ ▝▖ ▞ ▜ ▌ ▝▌ │
│ ▗▛▜▖ ▛ ▜ ▗▌ ▝▌ ▐▘ ▜ │
│ ▛ ▙ ▗▘ ▝▖ ▐ ▚ ▞ ▝▌ │
│ ▟▀▖ ▐▘ ▝▖ ▟ ▚ ▌ ▝▖ ▗▌ ▜▄│ 1
│ ▐▘ ▐▖ ▛ ▙ ▌ ▐▖ ▗▘ ▚ ▞ │
│ ▛ ▙ ▗▘ ▐▖ ▐ ▙ ▞ ▝▙▟▘ │
│▐▘ ▐▖ ▐ ▌ ▛ ▐▖ ▗▘ │
│▞ ▌ ▌ ▐ ▗▘ ▜▄▛ │
│▌─────▐────▐▘───────▙──▞────────────────────────────────────│ 0
│ ▌ ▛ ▝▙▟▘ │
│ ▜ ▐▘ │
│ ▙▄▛ │
└────────────────────────────────────────────────────────────┘
100 200 300 400 500 600
Here we're using Pandas to load and prepare gloabl temperature data from the Our World in Data GitHub repository.
First we load the data, rename a column and and filter the data:
import pandas as pd
uri = "https://github.com/owid/owid-datasets/raw/master/datasets/Global%20average%20temperature%20anomaly%20-%20Hadley%20Centre/Global%20average%20temperature%20anomaly%20-%20Hadley%20Centre.csv"
data = pd.read_csv(uri)
data = data.rename(columns={"Global average temperature anomaly (Hadley Centre)": "Global"})
data = data[data.Entity == "median"]
Then we can plot it:
from uniplot import plot
plot(xs=data.Year, ys=data.Global, lines=True, title="Global normalized land-sea temperature anomaly", y_unit=" °C")
Result:
Global normalized land-sea temperature anomaly
┌────────────────────────────────────────────────────────────┐
│ ▞▀│
│ ▐ │
│ ▐ │
│ ▗ ▌ │ 0.6 °C
│ ▙ ▗▄ ▛▄▖▗▘▌ ▞ │
│ ▗▜ ▌ ▜ ▚▞ ▚▞ │
│ ▐▝▖▐ ▘ │
│ ▗ ▗ ▌ ▙▌ │ 0.3 °C
│ ▛▖ ▞▙▘ ▘ │
│ ▖ ▗▄▗▘▐ ▐▘▜ │
│ ▟ █ ▞ ▜ ▝▄▘ │
│ ▗▚ ▗ ▖ ▗ ▖▗▞ █▐ ▌ ▘ │
│▁▁▁▞▐▁▁▗▘▜▗▀▀▌▁▁▁▁▙▁▁▟▁▁▁▙▐▁▁▜▁▌▞▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁│ 0 °C
│▚ ▐ ▝▖ ▐ ▛ ▌ ▗▄▐ ▌▗▘▌ ▐▝▌ ▝▘ │
│ ▌▌ ▌ ▞ ▐▗▘ ▛ ▐▞ ▌ ▐ │
│ ▝ ▝▖▌ ▐▞ ▝▌ ▚▜▐ │
│ ▗▌ ▝ ▝ ▌ │
└────────────────────────────────────────────────────────────┘
1,950 1,960 1,970 1,980 1,990 2,000 2,010
The plot function accepts a number of parameters, all listed below. Note that only
ys is required, all others are optional.
There is also a plot_to_string function with the same signature, if you want the result as a list of strings, to include the output elsewhere.
xs- The x coordinates of the points to plot. Can either beNone, or a list or NumPy array for plotting a single series, or a list of those for plotting multiple series. Defaults toNone, meaning that the x axis will be just the sample index ofys.ys- The y coordinates of the points to plot. Can either be a list or NumPy array for plotting a single series, or a list of those for plotting multiple series.
In alphabetical order:
color- Draw series in color. Defaults toFalsewhen plotting a single series, and toTruewhen plotting multiple.height- The height of the plotting region, in characters. Default is17.interactive- Enable interactive mode. Defaults toFalse.legend_labels- Labels for the series. Can beNoneor a list of strings. Defaults toNone.lines- Enable lines between points. Can either beTrueorFalse, or a list of those values for plotting multiple series. Defaults toFalse.line_length_hard_cap- Enforce a hard limit on the number of characters per line of the plot area. This may override thewidthoption if there is not enough space. Defaults toNone.title- The title of the plot. Defaults toNone.width- The width of the plotting region, in characters. Default is60. Note that if theline_length_hard_capoption is used and there is not enough space, the actual width may be smaller.x_gridlines- A list of x values that have a vertical line for better orientation. Defaults to[0].x_max- Maximum x value of the view. Defaults to a value that shows all data points.x_min- Minimum x value of the view. Defaults to a value that shows all data points.x_unit- Unit of the x axis. This is a string that is appended to the axis labels. Defaults to"".y_gridlines- A list of y values that have a horizontal line for better orientation. Defaults to[0].y_max- Maximum y value of the view. Defaults to a value that shows all data points.y_min- Minimum y value of the view. Defaults to a value that shows all data points.y_unit- Unit of the y axis. This is a string that is appended to the axis labels. Defaults to"".
For convenience there is also a histogram function that accepts one or more series and
plots bar-chart like histograms. It will automatically discretize the series into a
number of bins given by the bins option and display the result.
When calling the histogram function, the lines option is True by default.
Example:
import numpy as np
x = np.sin(np.linspace(1, 1000))
from uniplot import histogram
histogram(x)
Result:
┌────────────────────────────────────────────────────────────┐
│ ▛▀▀▌ │ ▐▀▀▜ │ 5
│ ▌ ▌ │ ▐ ▐ │
│ ▌ ▌ │ ▐ ▐ │
│ ▌ ▀▀▀▌ │ ▐▀▀▀ ▝▀▀▜ │ 4
│ ▌ ▌ │ ▐ ▐ │
│ ▌ ▌ │ ▐ ▐ │
│ ▌ ▙▄▄▄▄▄▖ │ ▗▄▄▄ ▐ ▐ │ 3
│ ▌ ▌ │ ▐ ▐ ▐ ▐ │
│ ▌ ▌ │ ▐ ▐ ▐ ▐ │
│ ▌ ▌ │ ▐ ▐ ▐ ▐ │
│ ▌ ▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▜ ▐▀▀▀ ▝▀▀▀ ▐ │ 2
│ ▌ │ ▐ ▐ ▐ │
│ ▌ │ ▐ ▐ ▐ │
│ ▌ │ ▐▄▄▟ ▐ │ 1
│ ▌ │ ▐ │
│ ▌ │ ▐ │
│▄▄▄▌▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁│▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▐▄▄▄│ 0
└────────────────────────────────────────────────────────────┘
-1 0 1
Install via pip using:
pip install uniplot
Coming up:
- Fill area under curve
- Log scales
- Add generated page with list of supported fonts
- Auto-detect color mode depending on terminal capabilities
- Possibly: Fallback to ASCII characters
Input is always welcome, let me know what is most needed for this to be as useful as possible.
Clone this repository, and install dependecies via pip3 install -e .\[dev\].
You can run the tests vie ./run_tests to make sure your setup is good. Then
proceed with issues, PRs etc. the usual way.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent