Support nested parens in license expressions by Liam-DeVoe · Pull Request #931 · pypa/packaging

Liam-DeVoe · 2025-09-08T02:39:08Z

Closes #928.

I've tested this with the following hypothesis test:

Details

import lark
from hypothesis import given, settings
from hypothesis.extra.lark import from_lark
from packaging.metadata import licenses

# naming and order follows
# https://spdx.github.io/spdx-spec/v2.2.2/SPDX-license-expressions/
grammar = """
start: license_expression

idstring: (ALPHA | DIGIT | "-" | ".")+
# a few hardcoded test licenses and exceptions
license_id: "MIT"
          | "Apache-2.0"
          | "GPL-3.0"
          | "BSD-3-Clause"
          | "ISC"
license_exception_id: "LLVM-exception"
                    | "Classpath-exception-2.0"
                    | "GCC-exception-3.1"

# DocumentRef is a valid prefix according to
# https://spdx.github.io/spdx-spec/v2.2.2/SPDX-license-expressions/, but
# we don't implement it.
# license_ref: ("DocumentRef-" idstring ":")? "LicenseRef-" idstring
license_ref: "LicenseRef-" idstring

simple_expression: license_id
                 | license_id "+"
                 | license_ref
compound_expression: simple_expression
                   | simple_expression " WITH " license_exception_id
                   | compound_expression " AND " compound_expression
                   | compound_expression " OR " compound_expression
                   | "(" compound_expression ")"
license_expression: simple_expression | compound_expression

%import common.LETTER -> ALPHA
%import common.DIGIT
"""

parser = lark.Lark(grammar)

@given(from_lark(parser))
@settings(deadline=None)
def test_canonicalize_license_expression(expression):
    licenses.canonicalize_license_expression(expression)

I'm very happy to add this test to the suite if you think it's worth the hypothesis dependency.

Liam-DeVoe · 2025-09-08T02:40:45Z

+            # we don't canonicalize redundant parens, instead leaving them as-is
+            # in the license expression.
+            ("(MIT)", "(MIT)"),
+            ("((MIT))", "((MIT))"),
+            ("(( MIT ))", "((MIT))"),
+            ("((MIT AND (MIT)))", "((MIT AND (MIT)))"),


we could make it so redundant parens are canonicalized, but we'd have to modify the parser to be a proper parser instead of deferring to python's compile.

henryiii

I think this makes sense as a first step; longer term I assume we should canonicalize parens?

Liam-DeVoe · 2025-10-08T19:13:08Z

I thought so at the time, but I'm actually not sure! Here's what PEP 639 has to say:

If tools choose to validate the SPDX expression, they also SHOULD store a case-normalized version of the License-Expression field using the reference case for each SPDX license identifier and uppercase for the AND, OR and WITH keywords.

which reads to me that parentheses should not be normalized. However, this might simply be an oversight.

If we do want to normalize parens, we'll have to decide what to do about technically-unnecessary parens for explicit operator precedence, for example a AND b OR c <-> (a AND b) OR c. Some might complain if this were normalized to the less-readable no-parens version, though it would certainly make the normalizing logic simpler.

This PR contains the following updates: | Package | Type | Update | Change | |---|---|---|---| | [packaging](https://github.com/pypa/packaging) | packages | major | `==25.0` → `==26.0` | --- ### Release Notes <details> <summary>pypa/packaging (packaging)</summary> ### [`v26.0`](https://github.com/pypa/packaging/releases/tag/26.0) [Compare Source](pypa/packaging@25.0...26.0) Read about the performance improvements here: <https://iscinumpy.dev/post/packaging-faster>. #### What's Changed Features: - PEP 751: support pylock by [@sbidoul](https://github.com/sbidoul) in [#900](pypa/packaging#900) - PEP 794: import name metadata by [@brettcannon](https://github.com/brettcannon) in [#948](pypa/packaging#948) - Support writing metadata by [@henryiii](https://github.com/henryiii) in [#846](pypa/packaging#846) - Support `__replace__` for `Version` by [@henryiii](https://github.com/henryiii) in [#1003](pypa/packaging#1003) - Support positional pattern matching for `Version` and `Specifier` by [@henryiii](https://github.com/henryiii) in [#1004](pypa/packaging#1004) Behavior adaptations: - PEP 440 handling of prereleases for `Specifier.contains`, `SpecifierSet.contains`, and `SpecifierSet.filter` by [@notatallshaw](https://github.com/notatallshaw) in [#897](pypa/packaging#897) - Handle PEP 440 edge case in `SpecifierSet.filter` by [@notatallshaw](https://github.com/notatallshaw) in [#942](pypa/packaging#942) - Adjust arbitrary equality intersection preservation in `SpecifierSet` by [@notatallshaw](https://github.com/notatallshaw) in [#951](pypa/packaging#951) - Return `False` instead of raising for `.contains` with invalid version by [@Liam-DeVoe](https://github.com/Liam-DeVoe) in [#932](pypa/packaging#932) - Support arbitrary equality on arbitrary strings for `Specifier` and `SpecifierSet`'s `filter` and `contains` method. by [@notatallshaw](https://github.com/notatallshaw) in [#954](pypa/packaging#954) - Only try to parse as `Version` on certain marker keys, return `False` on unequal ordered comparsions by [@JP-Ellis](https://github.com/JP-Ellis) in [#939](pypa/packaging#939) Fixes: - Update `_hash` when unpickling `Tag()` by [@dholth](https://github.com/dholth) in [#860](pypa/packaging#860) - Correct comment and simplify implicit prerelease handling in `Specifier.prereleases` by [@notatallshaw](https://github.com/notatallshaw) in [#896](pypa/packaging#896) - Use explicit `_GLibCVersion` `NamedTuple` in `_manylinux` by [@cthoyt](https://github.com/cthoyt) in [#868](pypa/packaging#868) - Detect invalid license expressions containing `()` by [@bwoodsend](https://github.com/bwoodsend) in [#879](pypa/packaging#879) - Correct regex for metadata `'name'` format by [@di](https://github.com/di) in [#925](pypa/packaging#925) - Improve the message around expecting a semicolon by [@pradyunsg](https://github.com/pradyunsg) in [#833](pypa/packaging#833) - Support nested parens in license expressions by [@Liam-DeVoe](https://github.com/Liam-DeVoe) in [#931](pypa/packaging#931) - Add space before at symbol in `Requirements` string by [@henryiii](https://github.com/henryiii) in [#953](pypa/packaging#953) - A root logger use found by ruff LOG, use `packaging` logger instead by [@henryiii](https://github.com/henryiii) in [#965](pypa/packaging#965) - Better support for subclassing `Marker` and `Requirement` by [@henryiii](https://github.com/henryiii) in [#1022](pypa/packaging#1022) - Normalize all extras, not just if it comes first by [@henryiii](https://github.com/henryiii) in [#1024](pypa/packaging#1024) - Don't produce a broken repr if `Marker` fails to construct by [@henryiii](https://github.com/henryiii) in [#1033](pypa/packaging#1033) Performance: - Avoid recompiling regexes in the tokenizer for a 3x speedup by [@hauntsaninja](https://github.com/hauntsaninja) in [#1019](pypa/packaging#1019) - Improve performance in `_manylinux.py` by [@cthoyt](https://github.com/cthoyt) in [#869](pypa/packaging#869) - Minor cleanups to `Version` by [@bearomorphism](https://github.com/bearomorphism) in [#913](pypa/packaging#913) - Skip redundant creation of `Version`s in specifier comparison by [@notatallshaw](https://github.com/notatallshaw) in [#986](pypa/packaging#986) - Cache `Specifier`'s Version by [@notatallshaw](https://github.com/notatallshaw) in [#985](pypa/packaging#985) - Make `Version` a little faster by [@henryiii](https://github.com/henryiii) in [#987](pypa/packaging#987) - Minor `Version` regex cleanup by [@henryiii](https://github.com/henryiii) in [#990](pypa/packaging#990) - Faster regex on Python 3.11.5+ by [@henryiii](https://github.com/henryiii) in [#988](pypa/packaging#988) and [#1055](pypa/packaging#1055) - Lazily calculate `_key` in `Version` by [@notatallshaw](https://github.com/notatallshaw) in [#989](pypa/packaging#989) and regression for `packaging_legacy` fixed by [@henryiii](https://github.com/henryiii) in [#1048](pypa/packaging#1048) - Faster `canonicalize_version` by [@henryiii](https://github.com/henryiii) in [#993](pypa/packaging#993) - Use `fullmatch` in a couple more places by [@henryiii](https://github.com/henryiii) in [#992](pypa/packaging#992) - Use `fullmatch` for markers too by [@henryiii](https://github.com/henryiii) in [#1029](pypa/packaging#1029) - Use `map` instead of generator by [@henryiii](https://github.com/henryiii) in [#996](pypa/packaging#996) - Deprecate `._version` (`_Version`, a `NamedTuple`) by [@henryiii](https://github.com/henryiii) in [#995](pypa/packaging#995) and [#1062](pypa/packaging#1062) - Avoid duplicate `Version` creation in `canonicalize_version` by [@henryiii](https://github.com/henryiii) in [#994](pypa/packaging#994) - Add `__slots__` to `Version` by [@henryiii](https://github.com/henryiii) in [#1001](pypa/packaging#1001) - Add `__slots__` to `Specifier`s by [@henryiii](https://github.com/henryiii) in [#1002](pypa/packaging#1002) - Add `__slots__` to `Node`s by [@henryiii](https://github.com/henryiii) in [#1032](pypa/packaging#1032) - Use `version.__replace__` in specifier comparison by [@notatallshaw](https://github.com/notatallshaw) in [#999](pypa/packaging#999) - Use `_get_spec_version` in more places in `Specifier` by [@notatallshaw](https://github.com/notatallshaw) in [#1005](pypa/packaging#1005) - Pull `set` construction out of function by [@henryiii](https://github.com/henryiii) in [#1012](pypa/packaging#1012) - Letter normalization dict for prereleases and the like by [@henryiii](https://github.com/henryiii) in [#1014](pypa/packaging#1014) - Avoid normalizing extras again when comparing by [@henryiii](https://github.com/henryiii) in [#1028](pypa/packaging#1028) - Speed up `Version.__str__` by about 10% by [@henryiii](https://github.com/henryiii) in [#997](pypa/packaging#997) - Increase the performance of `canonicalize_name` by avoiding a regex by [@henryiii](https://github.com/henryiii) in [#1030](pypa/packaging#1030), [#1047](pypa/packaging#1047), and [#1064](pypa/packaging#1064) - Faster zero stripping by [@henryiii](https://github.com/henryiii) in [#1058](pypa/packaging#1058) Type annotations: - Fix a type annotation by [@brettcannon](https://github.com/brettcannon) in [#907](pypa/packaging#907) - Fix tags return type in `parse_wheel_filename` docs by [@ncoghlan](https://github.com/ncoghlan) in [#973](pypa/packaging#973) - Add type hint for `_version` in `.version.Version` by [@brettcannon](https://github.com/brettcannon) in [#927](pypa/packaging#927) - Changed static type annotations in prereleases setter method in `specifier.py` by [@subhajitsaha01](https://github.com/subhajitsaha01) in [#930](pypa/packaging#930) - Statically type the tests by [@henryiii](https://github.com/henryiii) in [#982](pypa/packaging#982) Internal: - Test and declare support Python 3.14 by [@henryiii](https://github.com/henryiii) in [#901](pypa/packaging#901) - Modernize and speed up tests on Python 3.14 by [@henryiii](https://github.com/henryiii) in [#903](pypa/packaging#903) - Change our license metadata to use an SPDX license expression by [@cdce8p](https://github.com/cdce8p) in [#881](pypa/packaging#881) - No need for `license-files` by [@DimitriPapadopoulos](https://github.com/DimitriPapadopoulos) in [#924](pypa/packaging#924) - Update mypy by [@hauntsaninja](https://github.com/hauntsaninja) in [#891](pypa/packaging#891) - Some config updates by [@henryiii](https://github.com/henryiii) in [#902](pypa/packaging#902) - Add spell check and rst check by [@henryiii](https://github.com/henryiii) in [#904](pypa/packaging#904) - Clean up ruff ignores by [@henryiii](https://github.com/henryiii) in [#905](pypa/packaging#905) - Update example for env marker `python_version` by [@trim21](https://github.com/trim21) in [#908](pypa/packaging#908) - Move codespell configuration into pyproject.toml by [@yarikoptic](https://github.com/yarikoptic) in [#910](pypa/packaging#910) - Check warning a little more precisely by [@henryiii](https://github.com/henryiii) in [#837](pypa/packaging#837) - Speed up mypy a little by [@henryiii](https://github.com/henryiii) in [#836](pypa/packaging#836) - Apply ruff/flake8-pyi rules (PYI) by [@DimitriPapadopoulos](https://github.com/DimitriPapadopoulos) in [#835](pypa/packaging#835) - Better local runs for codespell by [@henryiii](https://github.com/henryiii) in [#911](pypa/packaging#911) - Remove outdated/confusing Gist link by [@stefan6419846](https://github.com/stefan6419846) in [#921](pypa/packaging#921) - Fix docs and docs ci after [#897](pypa/packaging#897) landed by [@notatallshaw](https://github.com/notatallshaw) in [#926](pypa/packaging#926) - Run twine-check on push in CI by [@EpicWink](https://github.com/EpicWink) in [#922](pypa/packaging#922) - `ruff` was renamed `ruff-check` in pre-commit by [@henryiii](https://github.com/henryiii) in [#933](pypa/packaging#933) - Fix incorrectly implicitly concatenated string in specifiers test by [@notatallshaw](https://github.com/notatallshaw) in [#946](pypa/packaging#946) - Simplify conditional by [@ofek](https://github.com/ofek) in [#949](pypa/packaging#949) - Modernize nox, use dependency-groups for tests by [@henryiii](https://github.com/henryiii) in [#952](pypa/packaging#952) - Add more checks that don't affect anything by [@henryiii](https://github.com/henryiii) in [#957](pypa/packaging#957) - Enable Ruff ISC rule by [@henryiii](https://github.com/henryiii) in [#959](pypa/packaging#959) - Ruff code FLY by [@henryiii](https://github.com/henryiii) in [#963](pypa/packaging#963) - pytest `log_level` is better than `log_cli_level` by [@henryiii](https://github.com/henryiii) in [#956](pypa/packaging#956) - Ruff code TRY by [@henryiii](https://github.com/henryiii) in [#961](pypa/packaging#961) - Add the ruff PL checks by [@henryiii](https://github.com/henryiii) in [#964](pypa/packaging#964) - Enable Ruff ARG rules by [@henryiii](https://github.com/henryiii) in [#958](pypa/packaging#958) - Ruff PT code (pytest) by [@henryiii](https://github.com/henryiii) in [#960](pypa/packaging#960) - Add ruff DTZ by [@henryiii](https://github.com/henryiii) in [#968](pypa/packaging#968) - Add ruff BLE by [@henryiii](https://github.com/henryiii) in [#967](pypa/packaging#967) - Add the ruff SIM checks by [@henryiii](https://github.com/henryiii) in [#966](pypa/packaging#966) - Adding ruff PERF by [@henryiii](https://github.com/henryiii) in [#969](pypa/packaging#969) - Move some config into coverage config by [@henryiii](https://github.com/henryiii) in [#971](pypa/packaging#971) - Check ruff C4 by [@henryiii](https://github.com/henryiii) in [#962](pypa/packaging#962) - Adding ruff T20 by [@henryiii](https://github.com/henryiii) in [#972](pypa/packaging#972) - Add a tests pass job by [@henryiii](https://github.com/henryiii) in [#977](pypa/packaging#977) - Add ruff TC by [@henryiii](https://github.com/henryiii) in [#980](pypa/packaging#980) - Adding part of ruff RET by [@henryiii](https://github.com/henryiii) in [#979](pypa/packaging#979) - Reorder mypy check by [@henryiii](https://github.com/henryiii) in [#983](pypa/packaging#983) - Enable ruff ALL by [@henryiii](https://github.com/henryiii) in [#984](pypa/packaging#984) - Link back to repo/source in furo by [@henryiii](https://github.com/henryiii) in [#991](pypa/packaging#991) - Add case insensitivity tests for arbitrary equality by [@notatallshaw](https://github.com/notatallshaw) in [#975](pypa/packaging#975) - Synchronize documentation and code for markers by [@zahlman](https://github.com/zahlman) in [#1008](pypa/packaging#1008) - Use `partition` in `_parse_project_urls` by [@henryiii](https://github.com/henryiii) in [#1013](pypa/packaging#1013) - auto-skip the dependabot PRs in the release changelog generation by [@henryiii](https://github.com/henryiii) in [#1016](pypa/packaging#1016) - Update unreleased section in changelog by [@henryiii](https://github.com/henryiii) in [#1017](pypa/packaging#1017) - Fix PR role to match extlinks by [@hugovk](https://github.com/hugovk) in [#1020](pypa/packaging#1020) - Mention new parts in README by [@henryiii](https://github.com/henryiii) in [#1023](pypa/packaging#1023) - Replace a couple of asserts with else by [@henryiii](https://github.com/henryiii) in [#1027](pypa/packaging#1027) - Simplify and/or check a little more by [@henryiii](https://github.com/henryiii) in [#1031](pypa/packaging#1031) - Use slim runner for all check by [@henryiii](https://github.com/henryiii) in [#1021](pypa/packaging#1021) - Use typos instead of codespell by [@henryiii](https://github.com/henryiii) in [#1015](pypa/packaging#1015) - Update changelog with recent additions by [@henryiii](https://github.com/henryiii) in [#1034](pypa/packaging#1034) - Publish to PyPI via GitHub CI by [@EpicWink](https://github.com/EpicWink) in [#893](pypa/packaging#893) - Use prek for faster pre-commit lint step by [@henryiii](https://github.com/henryiii) in [#1037](pypa/packaging#1037) - Add help text to noxfile by [@henryiii](https://github.com/henryiii) in [#1038](pypa/packaging#1038) - Update licenses to 3.27 by [@henryiii](https://github.com/henryiii) in [#1036](pypa/packaging#1036) - Use relative import in `packaging.licenses` by [@notatallshaw](https://github.com/notatallshaw) in [#1039](pypa/packaging#1039) - Add zizmor and tighten up CI by [@henryiii](https://github.com/henryiii) in [#1035](pypa/packaging#1035) - Fix release script by [@henryiii](https://github.com/henryiii) in [#1040](pypa/packaging#1040) - Fix using a dev version (again) by [@henryiii](https://github.com/henryiii) in [#1041](pypa/packaging#1041) - Fix type hint of function used with `contextlib.contextmanager` by [@SpecLad](https://github.com/SpecLad) in [#1046](pypa/packaging#1046) - Always run tests by [@henryiii](https://github.com/henryiii) in [#1044](pypa/packaging#1044) - Fix a changelog number by [@henryiii](https://github.com/henryiii) in [#1042](pypa/packaging#1042) - Fix the publish job by [@henryiii](https://github.com/henryiii) in [#1043](pypa/packaging#1043) - Get the correct tag on publish by [@henryiii](https://github.com/henryiii) in [#1045](pypa/packaging#1045) - Test on first public release of CPython 3.11 and newer by [@henryiii](https://github.com/henryiii) in [#1056](pypa/packaging#1056) - Fix publication job (again) by [@henryiii](https://github.com/henryiii) in [#1051](pypa/packaging#1051) - Use `actionlint` to check CI workflows by [@miketheman](https://github.com/miketheman) in [#1052](pypa/packaging#1052) - Fix formatting of distribution types in metadata.rst by [@brettcannon](https://github.com/brettcannon) in [#1053](pypa/packaging#1053) #### New Contributors - [@cdce8p](https://github.com/cdce8p) made their first contribution in [#881](pypa/packaging#881) - [@dholth](https://github.com/dholth) made their first contribution in [#860](pypa/packaging#860) - [@trim21](https://github.com/trim21) made their first contribution in [#908](pypa/packaging#908) - [@yarikoptic](https://github.com/yarikoptic) made their first contribution in [#910](pypa/packaging#910) - [@cthoyt](https://github.com/cthoyt) made their first contribution in [#868](pypa/packaging#868) - [@bwoodsend](https://github.com/bwoodsend) made their first contribution in [#879](pypa/packaging#879) - [@stefan6419846](https://github.com/stefan6419846) made their first contribution in [#921](pypa/packaging#921) - [@bearomorphism](https://github.com/bearomorphism) made their first contribution in [#913](pypa/packaging#913) - [@EpicWink](https://github.com/EpicWink) made their first contribution in [#922](pypa/packaging#922) - [@Liam-DeVoe](https://github.com/Liam-DeVoe) made their first contribution in [#932](pypa/packaging#932) - [@subhajitsaha01](https://github.com/subhajitsaha01) made their first contribution in [#930](pypa/packaging#930) - [@ncoghlan](https://github.com/ncoghlan) made their first contribution in [#973](pypa/packaging#973) - [@zahlman](https://github.com/zahlman) made their first contribution in [#1008](pypa/packaging#1008) - [@JP-Ellis](https://github.com/JP-Ellis) made their first contribution in [#939](pypa/packaging#939) #### Since last RC Fixes: - Restore `._version` as a compat shim by [@henryiii](https://github.com/henryiii) in [#1062](pypa/packaging#1062) Performance: - Dual replace by [@henryiii](https://github.com/henryiii) in [#1064](pypa/packaging#1064) Documentaiton: - Prepare for 26.0 final by [@henryiii](https://github.com/henryiii) in [#1063](pypa/packaging#1063) **Full Changelog**: <pypa/packaging@26.0rc3...26.0> </details> --- ### Configuration 📅 **Schedule**: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined). 🚦 **Automerge**: Disabled by config. Please merge this manually once you are satisfied. ♻ **Rebasing**: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox. 🔕 **Ignore**: Close this PR and you won't be reminded about this update again. --- - [ ] If you want to rebase/retry this PR, check this box --- This PR has been generated by [Renovate Bot](https://github.com/renovatebot/renovate).  Reviewed-on: https://kubara.git.onstackit.cloud/STACKIT/kubara/pulls/280

support nested parens

2e06626

Liam-DeVoe commented Sep 8, 2025

View reviewed changes

format

75157b6

Liam-DeVoe mentioned this pull request Oct 4, 2025

Return False instead of raising for .contains with invalid version #932

Merged

henryiii approved these changes Oct 8, 2025

View reviewed changes

Merge branch 'main' into nested-parens

d17255a

henryiii merged commit a1f7056 into pypa:main Oct 29, 2025
39 checks passed

Liam-DeVoe deleted the nested-parens branch October 29, 2025 22:46

github-actions Bot mentioned this pull request Feb 12, 2026

Bump packaging from 25.0 to 26.0 senzing-factory/dockerhub-util#176

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Support nested parens in license expressions#931

Support nested parens in license expressions#931
henryiii merged 3 commits intopypa:mainfrom
Liam-DeVoe:nested-parens

Liam-DeVoe commented Sep 8, 2025

Uh oh!

Liam-DeVoe Sep 8, 2025

Uh oh!

henryiii left a comment

Uh oh!

Liam-DeVoe commented Oct 8, 2025 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

Liam-DeVoe commented Sep 8, 2025

Uh oh!

Liam-DeVoe Sep 8, 2025

Choose a reason for hiding this comment

Uh oh!

henryiii left a comment

Choose a reason for hiding this comment

Uh oh!

Liam-DeVoe commented Oct 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Liam-DeVoe commented Oct 8, 2025 •

edited

Loading