Arxiv.py Versions Save

Deprecating change.

Deprecates submodule-style imports for the base package: import arxiv.arxiv and from arxiv import arxiv. Use import arxiv.

What's Changed

• Move package logic into __init__.py by @lukasschwab in https://github.com/lukasschwab/arxiv.py/pull/141
• Add an unencoded query example to README.md by @lukasschwab in https://github.com/lukasschwab/arxiv.py/pull/143
• Restructure README.md: focus on generated docs by @lukasschwab in https://github.com/lukasschwab/arxiv.py/pull/144

Full Changelog: https://github.com/lukasschwab/arxiv.py/compare/2.0.0...2.1.0

What's Changed

Breaking change.

• Primary goal is to improve the client's resilience against API instability. See ongoing investigation on #129.
• Basic usage — i.e. everything covered in the README — should be unaffected.

Breaking

• Previously deprecated Client.get/Search.get removed in #137
• Misleading type hints/defaults for max_results/delay_seconds replaced in #138
  • Constructors are expected to be backwards-compatible, but this is untested behavior
• UnexpectedEmptyPageError and HTTPError constructor signatures changed in #136
  • Sets a User-Agent header on the API request
  • Introduces dependency on requests
  • Introduces Client-encapsulated session reuse
  • feedparser no longer swallows connection errors; instead, retry per Client behavior

Deprecating

• Search.results deprecated: prefer Client.results with client reuse.

Other

• Format with black in https://github.com/lukasschwab/arxiv.py/pull/120
• Standardize on HTTPS over HTTP in https://github.com/lukasschwab/arxiv.py/pull/131
• Improve logging: standard formatting, verbose tests in https://github.com/lukasschwab/arxiv.py/pull/132
• Use setup-python@v4, cache pip dependencies in https://github.com/lukasschwab/arxiv.py/pull/133
• Pin feedparser==6.0.6 in https://github.com/lukasschwab/arxiv.py/pull/135

Full Changelog: https://github.com/lukasschwab/arxiv.py/compare/1.4.8...2.0.0

What's Changed

Nonbreaking change.

• Adapt pdoc docs.yml docs-deploy workflow by @lukasschwab in https://github.com/lukasschwab/arxiv.py/pull/113
• Remove ./docs directory from git by @lukasschwab in https://github.com/lukasschwab/arxiv.py/pull/114
• Simplify pdoc build, eliminate nav badges by @lukasschwab in https://github.com/lukasschwab/arxiv.py/pull/115
• Replace / with _ in legacy-format IDs in download filenames by @lukasschwab in https://github.com/lukasschwab/arxiv.py/pull/118

Full Changelog: https://github.com/lukasschwab/arxiv.py/compare/1.4.7...1.4.8

Nonbreaking change. No logical changes from 1.4.6.

• Indicate Python version in trove classifiers by @lukasschwab in https://github.com/lukasschwab/arxiv.py/pull/112

Nonbreaking change.

• Move linting from flake8 to ruff by @lukasschwab in https://github.com/lukasschwab/arxiv.py/pull/110
• Set python_requires >=3.7 by @lukasschwab in https://github.com/lukasschwab/arxiv.py/pull/111

Full Changelog: https://github.com/lukasschwab/arxiv.py/compare/1.4.5...1.4.6

Nonbreaking change.

• Disambiguate generic list type hints by @lukasschwab in https://github.com/lukasschwab/arxiv.py/pull/109

Nonbreaking change.

• Fix docstring: updated and published are datetimes (#106, #107)
• Try to make docs-generation simpler for contributors. See #108.
  • Pin pyenv Python version to v3.10.4.
  • Bump pdoc to v13.1.0.

NOTE: this tag should really point to 67b02aa. The tagged commit has the wrong version number in setup.py.

Nonbreaking change.

• Add docstring warning about Search.query encoding (#91)
• Fiz "abstrace" typo in Result docstrings (#96)
• Parameterize initial search offset for error recovery (#103)

Nonbreaking change.

• Run retries recursively to sleep between erroneous responses (#53)
• Skip invalid entries from nonexistent id_list IDs (#81)

Nonbreaking change.

• Extract result.comment from entry.get('arxiv_comment)' rather than entry.get('comment') (#79)