projects:instagram_crawler [wiki]

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
===== Technical Overview =====
  * multiple 7%%"%% or 10%%"%% displays and/or projectors
  * each display/projector runs a slideshow of images/videos found on instagram with its own unique hashtag
  * 1 audio speaker associated with each display
  * each audio speaker plays audio from video when video is shown, plays chooses randomly from a set of audio files when showing images
  * images are shown between X and Y seconds (?)
  * entire video is shown or random clips of video are shown for between X and Y seconds (?)
  * crossfade between images and between video
  * projectors and 7%%"%%/10%%"%% displays are never mirroring each other (?, hopefully?)

===== Already-purchased Hardware =====
  * We already have rPI Picos, 7%%"%% displays for rPI, projector
  * Picos don't run Linux and therefore don't run Python or Java. Probably C++based like the ESP32
  * Picos don't support the 7%%"%% displays
  * Projector interface unknown but probably VGA/HDMI

===== Possible Technical Solutions =====
==== Preexisting Software ====
  * [[https://instalooter.readthedocs.io/en/latest/|Instalooter]] is a Python-based Instagram scraper that grabs images and video from hashtags and saves them in a  directory
    * [[https://github.com/althonos/InstaLooter|github]]
    * [[https://instalooter.readthedocs.io/en/latest/install.html|Installation]]
    * tested and seems to work perfectly with both images and video with sound for this application
    * opens browser on first run (only)
    * update: does not seem to work anymore?
  * [[https://github.com/yogeshwaran01/instagramy|instagramy]] is another Python-based scraper
  * [[https://apify.com/about|Apify]] looks to be a private API for Instagram that can do the job with API calls. There's a free tier but haven't dug in further
  * There's a couple of Java-based [[https://github.com/topics/instagram-java-scraper|Instagram scrapers]] perhaps worth considering
  * [[https://github.com/instaloader/instaloader|Instaloader]] is yet another Python-based scraper but this one seems to be actively developed
  * there's this mysterious [[https://github.com/89z/mech|Mech]] software that doesn't seem well documented
  * It could be the best route tho is [[https://developers.facebook.com/docs/instagram-api/guides/hashtag-search/|the official api]]
==== Hardware ====
=== rPI ===
Purchase rPIs that will run Instalooter and connect with the 7%%"%% displays we already have. Write/find/modify slideshow app to show the images/video and play the audio

This is a great solution except for that rPIs that work with the 7%%"%% displays will probably cost about $100 each in today's market after power supply and memory are considered. We also would need to figure out amplification and speakers for each Pi. In spite of this, it is likely the fastest way to get results. 
=== Android Tablets ===
Purchase Android tablets. Rip tablets out of their case as appropriate to hide electronics etc. Port Instalooter, roll our own scraper, use an existing Java scraper, or use Kivy to move Instalooter to Android. Write/find/modify slideshow app to show the images/video and play the audio. Make some simple circuitry such that tablets turn on automatically when power is applied (push the "on" switch)

This is the least expensive option as used Android tablets are cheap. It will take some work to get a scraper working in Java/Kotlin. Not sure how much work and could be a bigger job than it's worth.

=== Android Tablets and Central Computer ===
Purchase Android tablets. Rip out of case as appropriate to hide electronics etc. Have a central computer (probably an old Chromebook Laptop hidden somewhere with lid closed) do the scraping. Android tablets get the images/video wirelessly from central computer and show images. Write/find/modify slideshow app to show the images/video and play the audio. Make some simple circuitry such that tablets turn on automatically when power is applied (push the "on" switch)

This shouldn't be too much more work than just using the rPIs since we can use Instalooter without modification. It's at least a lot less work than the previous option.