This reverts commit 0911477.

urllib3 is needed to postprocess/parse/URL for sanitization and privacy purpose (bibanon#192)

We fix this by carefully redacting the IP address in the JSON fields known to contain it.

* Revert "Remove urllib3 dependency"

This reverts commit 0911477.

urllib3 is needed to postprocess/parse/URL for sanitization and privacy purpose (bibanon#192)

* IA currently leaks the IP address of the submitter. This is bad.

We fix this by carefully redacting the IP address in the JSON fields known to contain it.

* added tests

* flake8 linting

- Use latest urllib3 instead of that specific version

…oaded the way

  yt-dlp keep track of downloaded items.

- Amend bibanon#19 by adding optional idempotency between runs:
  While concurrent-instances can still rely on a "clean" download directory (bibanon#19)
  a single upload node can use --use-upload-archive can avoid source files removal.
  Subsequent runs of `tubeup` will omit files already uploaded.

- Fixes bibanon#23 (and part of bibanon#233)
- NB: Ability for upload_ia() to return None, paves the way to fix bibanon#36 or bibanon#109

…dia files.

If `tubeup` is passed the path of info.json files instead of URL,
then it will parse them and, if `--output` was set correctly (same value as previous `yt-dlp`)
then recover the basename of the local files associated with this video.

If used intelligently, especially in conjunction with:
--ignore-existing-item # needed, for now, to avoid the download-archive codepath
--yt=--format=...      # use the same download format
--yt=--no-overwrites
... then existing files may be uploaded without having been re-downloaded/overwritten.

Usage isn't foolproof. Use at your own risks.

…video basename)

…e uploaded