Tablib

Breaking changes

- Dropped Python 2 support
- Dependencies are now all optional. To install `tablib` as before with all
possible supported formats, run `pip install tablib[all]`

Improvements

- Formats can now be dynamically registered through the
`tablib.formats.registry.refister` API (256).
- Tablib methods expecting data input (`detect_format`, `import_set`,
`Dataset.load`, `Databook.load`) now accepts file-like objects in addition
to raw strings and bytestrings (440).

Bugfixes

- Fixed a crash when exporting an empty string with the ReST format (368)
- Error cells from imported .xls files contain now the error string (202)

Deprecations

- The 0.14.x series will be the last to support Python 2

Breaking changes

- Dropped Python 3.4 support

Improvements

- Added Python 3.7 and 3.8 support
- The project is now maintained by the Jazzband team, https://jazzband.co
- Improved format autodetection and added autodetection for the odf format.
- Added search to all documentation pages
- Open xlsx workbooks in read-only mode (316)
- Unpin requirements
- Only install backports.csv on Python 2

Bugfixes

- Fixed `DataBook().load` parameter ordering (first stream, then format).
- Fixed a regression for xlsx exports where non-string values were forced to
strings (314)
- Fixed xlsx format detection (which was often detected as `xls` format)

- Added reStructuredText output capability (336)
- Added Jira output capability
- Stopped calling openpyxl deprecated methods (accessing cells, removing sheets)
(openpyxl minimal version is now 2.4.0)
- Fixed a circular dependency issue in JSON output (332)
- Fixed Unicode error for the CSV export on Python 2 (215)
- Removed usage of optional `ujson` (311)
- Dropped Python 3.3 support

- Favor `Dataset.export(<format>)` over `Dataset.<format>` syntax in docs
- Make Panda dependency optional

- Add initial Panda DataFrame support
- Dropped Python 2.6 support

- Use `yaml.safe_load` for importing yaml.