"in_reply_to_user_id" is coming as float, but should be a string.

Question

"in_reply_to_user_id" is coming as float, but should be a string.

patrick-lee-warren opened this issue 3 years ago · 7 comments

patrick-lee-warren commented 3 years ago

The other ids seem to be correctly specified as str, but this one is coming in with a decimal.

Answer 1 · 2021-10-12T00:18:50.000Z

Same deal for pinned_tweet_id, for the user pull.

Answer 2 · 2021-10-12T14:59:55.000Z

I see this behavior as well. twitter alchemy correctly parses both of these as integers, so the issue must be on the pandas/saving side of things. We should see if things go wrong at this line, for example.

Answer 3 · 2021-10-13T14:48:17.000Z

Carl's theory is that the missing values are causing the pandas series to be coerced to float.

Answer 4 · 2021-10-17T23:52:04.000Z

We aren't ever going to be any mathematical operations on any of these ids. There is no reason to treat the as anything but strings, ever.

Answer 5 · 2021-10-18T00:12:56.000Z

This has now been resolved in Pull Request #22. The float data-typed id columns specified above have been converted to remove the decimal and can now be interpreted as strings correctly.

Answer 6 · 2021-10-21T15:12:46.000Z

I made the necessary change in twitter alchemy: dhudsmith/twitter-alchemy@fa00f93#diff-ea93147253fd85a5596145f5e805a803cdad3416b82c4d75b81a99006f345784

I also changed the requirements file to match: d679581

@NickDeas, I tested and it looks ok. I think we just need to rip out the fix_floats logic to restore compatibility with python >= 3.6

Answer 7 · 2021-10-21T16:58:13.000Z

fix_floats and references removed, and ReadMe reverted to old compatibility. In my testing it looked to work as well.