Improvements to the SocialMediaAccountExposure module and associated content scripts

[jonathanmayer]

• General

  • [x] Nit: rename domains.js, mediaFbAccounts.js, and mediaYtChannels.js for clarity and consistency (e.g., newsDomains.js, newsFacebookAccounts.js, and newsYouTubeChannels.js) and update references to those files.
  • [x] Remove socialmedia.js and the reference in study.js (seems like this isn't used anymore).
  • [x] Please update the news social media accounts based on Ben's latest data.

• SocialMediaAccountExposure Module

  • [x] Implement support for Twitter social media account exposure (like you've already implemented for Facebook and YouTube).
  • [x] Rename nextPageIdCounter that was copy and pasted from another module, since it's not actually counting pages (e.g., nextSocialMediaAccountExposureIdCounter).
  • [x] Replace the YouTube dynamic content script injection (ytChannelMatchCode and ytChannelMatcher) with a RegExp in extension local storage.
  • [x] Replace the Facebook dynamic content script injection (fbAccountMatchCode and fbAccountMatcher) with a RegExp in extension local storage.
  • [x] Update the two instances of "Warning: unexpected page content update" that were copy and pasted from another module, since that's not the issue that would trigger that debugging output.
  • [x] Define one type of exposure event in JSDoc, and consistently store that object type. The type should include a property that specifies the social media platform.
  • [x] A separate message listener for each social media platform is unnecessary. It would be cleaner to have one message listener and one message format for all the platforms.
  • [x] Update getStudyDataAsObject to reflect the current stored data.

• socialMediaAccountExposure-fb.js Content Script

  • [x] Nit: fix the module JSDoc (see comments on this issue for other content scripts).
  • [x] Condense the private window logic (see comments on this issue for other content scripts).
  • [x] Replace all the instances of x with a more descriptive variable name.
  • [x] Clearly document each step of the DOM parsing logic, with examples. What, exactly, does each of these operations do?
  • [x] Send an array of URLs to the background page rather than a concatenated string.

• socialMediaAccountExposure-youtube.js Content Script

  • [x] Nit: fix the module JSDoc (see comments on this issue for other content scripts).
  • [x] Condense the private window logic (see comments on this issue for other content scripts).
  • [x] The g and m modifiers don't appear necessary for the ytmatcher string.
  • [x] Nit: the checkChannel function would benefit from a more descriptive name (e.g., isVideoPage).
  • [x] The isYoutube function is only used once. That code can be consolidated into the checkChannel function. Also... I don't think we want this check, since a user could be exposed to news videos on pages that aren't a video viewing page. Can also drop the videoTitle logic.
  • [x] The current implementation looks for links in the "channel" format. It should also support links in the "user" format. For example, the CNN channel is both https://www.youtube.com/user/CNN and https://www.youtube.com/channel/UCupvZG-5ko_eiXAupbDfxWw.
  • [x] Add documentation to the checkChannel function clarifying what it's doing. Also rename the x variable.

[jonathanmayer]

Closing this issue, both because @PranayAnchuri worked through all the items and because we paused development on the SocialMediaAccountExposure module half a year ago: https://github.com/citp/news-disinformation-study/commit/0c651ca55375ee6e39a4cbaffae3dd0aacc56ded