Log API responses for comments #55
Description
The YouTube commentCount is apparently pretty complicated (which is a nice way of saying inaccurate).
As I captured in a gist, it does not seem to be reliable. The count can be high due to comment filtering/moderation, or low than due to channel author comments not be counted, etc. Recently, I ran into this example:
I expected to retrieve one additional comment from the video, which was determined by comparing the local database commentCount to the API response commentCount. Attempting to pull the one comment, I apparently downloaded the pinned comment by the channel author, in addition to 5 of its replies, all of which had been posted 10/13/2024, so hardly the newest comments like my logic expects.
I'll need to gather my own data to determine what exactly is going on here. The work proposed is as follows:
- Add argument for logging json API responses (maybe
--log-json |-j`)- json should be logged to a file
- A separate utility python script should be written to parse and analyze the json