parse_csv: fix parsing youtube comment for newer takeout exports

[karlicoss]

noticed some errors in promnesia now that new comments started being extracted after my previous fix

of course, comments are comma-separated jsons with one like per each json 🤡 If you have any good ideas how to parse it nicer, let me know! Sadly, stdlib json parser doesn't support some sort of streaming mode :(

[karlicoss]

Thought I had a bug for a second, but turned out to be fine, so reopened!

[purarue]

of course, comments are comma-separated jsons with one like per each json

uh, this fix is probably fine, I just cant even visualize what that means...? Could you post/link me to example data just so I understand

[karlicoss]

Ah, you can see the example here! https://github.com/seanbreckenridge/google_takeout_parser/pull/79/files#diff-8789098ed2a3a26eb03a72c01b6432558382e227eb4cc9bc3cbd50f570f9857bR80

Something like this: contentJSON='{"text":"> I am about to get buried in the concrete"},{"text":"\n"},{"text":"the most normal Veritasium video!"}'

[purarue]

bleh... I don't why they just cant stick to some standard format...

I guess they have to innovate™

thanks for the fix 👍

[purarue]

feel free to ping me if you want me to do a new release (though I dont think this impacts other CIs)

[karlicoss]

Yeah, it's dynamically called anyway, so no need for new release. Thanks!