Create videos from Wikipedia articles

Turn Wikipedia articles into videos with computer-generated overdubbing, and the option to upload directly to YouTube. Includes helpful parsers for Wikipedia API requests, as well as a module for downloading pictures from Google image searches.

Installation:

Required: Firefox, Geckodriver, Imagemagick, Ghostscript, Python>=3.9

Linux:

sudo apt-get install firefox firefox-geckodriver imagemagick ghostscript libsndfile1-dev espeak

Homebrew:

brew install --cask firefox; brew install geckodriver imagemagick ghostscript

Python dependencies/environment

Create and activate a virtual environment before next steps, such as venv, conda, pipenv.

pipenv install --skip-lock (recommended) or pip install -r requirements.txt

Optional: pip install pygame, if you want to run in-line video previews in jupyter notebooks

Usage

Note: To be able to upload to a YouTube account, you must first obtain a client_secrets.json file for the YouTube API, and store it in the wiki_movie/ directory. See Obtaining authorization credentials.

Command Line

Simple usage:

python main.py (-s SINGLE_PAGE | -t | --url URL) -o

Examples:

python main.py -s Computer Science --narrator py_tts -o

python main.py --top25 --voice Alex --rate 250 --overwrite --upload

python main.py --url https://en.wikipedia.org/wiki/Wikipedia:Top_25_Report/February_7_to_13,_2021

Required arguments:

[-s | -t | --url]
-s --single_page      Title of the article to process (ok to have spaces in title)
-t --top25            Use the top25 articles of the week
--url                 URL of an article, which must contain a list/table of articles            

Optional arguments:

Parameter	Default	Description
-n --narrator	sys_tts	Speech generation engine (string). Accepted values: py_tts, google_tts, dc_tts
-o --overwrite	True	Flag, which if set, will overwrite any data of an article of the same name
-u --upload	False	Flag, upload created video to Youtube
-p --private	False	Flag, upload with 'private' viewing status
-d --delete_all	False	Flag, delete assets(images, audio) after movie is created
--n_pages	25	Integer, use with --url arg. Sets limit on number of article links to process from --url.

Narrator arguments:

So far, only arguments for the SystemNarrator have been implemented for the command line.

Parameter	Default	Description
-v, --voice	MacOS: Alex, Linux: na	narrator voice - available voices dependant on platform
-r --rate	MacOS: 200, Linux: na	words spoken per minute

ImageDownloader object arguments:

Parameter	Default	Description
-i, --image_count	5	Number of images to loop per article section
-c, connect_speed	'medium'	How quickly to timeout scraping search results page
-w, --watch	False	Flag, run browser in non-headless mode

• The first run of main.py may prompt on the terminal for a key-code, which will be available in a browser window.

Description of available narrator arguments

• sys_tts: Automatically uses the OS built-in speech command
  • MacOSX: NS Speech Synthesizer
  • Linux: espeak
  • Windows: not implemented
• py_tts: Uses pyttsx3
• google_tts: Uses gTTS
• dc_tts: Deep Convolutional TTS

Python

import wiki_movie
movie = wiki_movie.WikiMovie("Computer Science")
movie.make_movie()

Standalone ImageDownloader, for scraping google image searches

from wiki_movie.image import ImageDownloader
downloader = ImageDownloader('Python', num_requested=100)
downloader.find_and_download()

Optionally add list of search terms to combine with the main keyword.

from wiki_movie.image import ImageDownloader
downloader = ImageDownloader('Python', ['big', 'small', 'green'])
# A search will be run for each: 'Python big', 'Python small', 'Python green'
downloader.find_and_download()

Possible ImageMagick/MoviePy issue:

On some systems, an issue arises with MoviePy's use of ImageMagick, and you may need to alter the policy.xlm file of ImageMagick (see this issue).

Steps to fix:

1. Find the location of the ImageMagick policy file:

   • convert -list policy

   • The top line of the output should look like Path: /etc/ImageMagick-6/policy.xml

   • If instead you see Path: [built-in], then no changes may be necessary.

2. Edit the line of the policy file which refers to the "path" domain, to allow rights for read and write.

   • sudo nano /etc/ImageMagick-6/policy.xml

   • Original:

     <policy domain="path" rights="none" pattern="@*"/>

   • Edited:

     <policy domain="path" rights="read | write" pattern="@*"/>

References:

• ImageMagick
• Ghostscript
• MoviePy
• Wikipedia-API (python)
• pyttsx3