djm158 / help-wanted-bot Goto Github PK

spottedmahn/various-app-feedback#1

We should cache either the search results or the last tweet made so we do not send the same exact tweet twice in a row (#1)

It may be helpful to include a list of the issue's labels to the tweet -- for example, as a newer programmer, it may be nice to see good-first-issue or documentation to help find things to work on.

In the context of Ln 52 of index.js, const buildStatus = async (issue) => {, It looks like a request to issue.labels_url returns the following as the body (I've generalized some fields with <>):

[
  {
    "id": <id>,
    "node_id": "<node_id>",
    "url": "https://api.github.com/repos/<owner-name>/<repo-name>/labels/<label-name>",
    "name": "<LABEL-NAME>",
    "color": "DB9D72",
    "default": false,
    "description": ""
  },
... {<further labels>} ...
]

Reference

This of course would bloat the character count of the tweets, however it looks like most of the tweets being sent currently have plenty of wiggle-room to include labels. (Though I also see a // TODO: check character count that may address this case in the future).

https://twitter.com/azurefunctions/status/995041550043267072?s=21

Twitter flagged us the other day. Need to do more research but it is possibly occurring because we are trying to spam the same tweet when the search results do not change in a ten minute window.

Heroku provides 550 free dyno hours per month, which prevents running the bot continuously. Easiest solution is probably to run it on serverless infrastructure (AWS Lambda has a generous free tier that will likely suit the needs of this app).