Running Python apps from R

You too can run Python apps from R. I did the following in RStudio:

  • Install the reticulate package.
  • Run one of the scripts below.

It will install everything required in a separate Python virtual environment, i.e. isolated from any other Python setup on your computer.

pyspread

Link: Welcome to pyspread | pyspread

Want to use Python in your spreadsheet and tired of waiting for your corporate IT to roll that out? Try pyspread.

first_run <- TRUE  # change to FALSE after first run

library(reticulate)

if (first_run)
  virtualenv_create('pyspread')
	
use_virtualenv('pyspread')

if (first_run) {
  install_python()
  py_install('pyspread', pip = TRUE)
}

python_path <- py_exe()
system2(python_path, '-m pyspread')

datasette

Link: Datasette: An open source multi-tool for exploring and publishing data

An interactive Web app for exploring a SQLite database (might use this to improve my SQL)

Download and unzip the training database chinook.db (to the same folder as the R script) from here.

The app should open in your browser.

first_run <- TRUE  # change to FALSE after first run

library(reticulate)

if (first_run)
  virtualenv_create('datasette')

use_virtualenv('datasette')

if (first_run) {
  install_python()
  py_install('datasette', pip = TRUE)
}

python_path <- py_exe()
browseURL('http://127.0.0.1:8001')
system2(python_path, '-m datasette serve chinook.db')

howdoi

Link: gleitz/howdoi: instant coding answers via the command line

Search for help with programming tasks from the command line

Only partial success with this one - search engines block automated queries in Europe because cookies have not been accepted (I will have to try this one with a VPN)

first_run <- TRUE  # change to FALSE after first run

library(reticulate)

if (first_run)
  virtualenv_create('howdoi')

use_virtualenv('howdoi')

if (first_run) {
  install_python()
  py_install('howdoi', pip = TRUE)
}

python_path <- py_exe()
Sys.setenv(HOWDOI_SEARCH_ENGINE = 'bing')  # or google, duckduckgo
howdoi <- function(query) {
  cmd <- paste('-m howdoi', query)
  system2(python_path, cmd)
}

Example usage:

howdoi('howdoi')
howdoi('format bash date')

httpie

Link: HTTPie – API testing client that flows with you

Interact with Web APIs from the command line (this could be very useful)

first_run <- TRUE  # change to FALSE after first run

library(reticulate)

if (first_run)
  virtualenv_create('httpie')
	
use_virtualenv('httpie')

if (first_run) {
  install_python()
  py_install('httpie', pip = TRUE)
}

https <- function(query) {
  https_path <- file.path(py_config()$pythonhome, 'Scripts', 'https.exe')
  system2(https_path, query)
}

Example usage:

https('httpie.io/hello')  # their test server

Returns JSON:

{"ahoy":["Hello, World! 👋 Thank you for trying out HTTPie 🥳","We hope this will become a friendship."],"links":{"homepage":"https://httpie.io","twitter":"https://twitter.com/httpie","discord":"https://httpie.io/discord","github":"https://github.com/httpie"}}

yt-dlp

Link: yt-dlp/yt-dlp: A feature-rich command-line audio/video downloader

Download YouTube videos

  • not adapted for Linux but you would probably just use yt-dlp directly anyway
first_run <- TRUE  # change to FALSE after first run

library(reticulate)

if (first_run)
  virtualenv_create('yt-dlp')

use_virtualenv('yt-dlp')

if (first_run) {
  install_python()
  py_install('yt-dlp', pip = TRUE)
}

yt_dlp <- function(query) {
  yt_dlp_path <- file.path(py_config()$pythonhome, 'Scripts', 'yt-dlp.exe')
  system2(yt_dlp_path, '-o', query)
}

Usage:

yt_dlp('https://www.youtube.com/watch?v=6541wEO_04o')

Output:

[youtube] Extracting URL: https://www.youtube.com/watch?v=6541wEO_04o
[youtube] 6541wEO_04o: Downloading webpage
[youtube] 6541wEO_04o: Downloading ios player API JSON
[youtube] 6541wEO_04o: Downloading web creator player API JSON
[youtube] 6541wEO_04o: Downloading m3u8 information
WARNING: ffmpeg not found. The downloaded format may not be the best available. Installing ffmpeg is strongly recommended: https://github.com/yt-dlp/yt-dlp#dependencies
[info] 6541wEO_04o: Downloading 1 format(s): 18
[download] Destination: How Best to Run Python in RStudio Using the Reticulate Library in R [6541wEO_04o].mp4
[download] 100% of    9.85MiB in 00:00:04 at 1.98MiB/s

hawk-eye

Link: rohitcoder/hawk-eye: A powerful scanner to scan your Filesystem, S3, MySQL, Redis, Google Cloud Storage and Firebase storage for PII and sensitive data.

Scan your filesystem (and much more) for sensitive credentials (identifiable by regular expressions).

Instructions for installation and simple use via R below. Put the following two configuration files in your working directory:

  • fingerprint.yml from hawk-eye/fingerprint.yml at main · rohitcoder/hawk-eye - the regular expressions it will look for
  • connection.yaml: I used the configuration below (to scan my Desktop)
    notify:
  redacted: True
  suppress_duplicates: True

sources:
  fs:
    mydesktop:
      path: "C:/Users/fred.bloggs/Desktop"
      exclude_patterns:
        - .pdf
        - .docx
        - private
        - venv
        - node_modules
first_run <- TRUE  # change to FALSE after first run

library(reticulate)

if (first_run)
  virtualenv_create('hawk-scanner')

use_virtualenv('hawk-scanner')

if (first_run) {
  install_python()
  py_install('hawk-scanner', pip = TRUE)
}
hs_path <- file.path(py_config()$pythonhome, 'Scripts', 'hawk_scanner.exe')
system2(hs_path, 'fs --connection connection.yaml --fingerprint fingerprint.yml')

If there are lots of files it will run for ages and possibly max out your CPU.

Troubleshooting