rubocop: Avoid duplicate arch-specific versions (and more) in casks

[issyl0]

• Have you followed the guidelines in our Contributing document?
• Have you checked to ensure there aren't other open Pull Requests for the same change?
• Have you added an explanation of what your changes do and why you'd like us to include them?
• Have you written new tests for your changes? Here's an example.
• Have you successfully run brew lgtm (style, typechecking and tests) with your changes locally?

---

• AI was used to generate or assist with generating this PR. Please specify below how you used AI to help you, and what steps you have taken to manually verify the changes.

• Out of curiosity I gave the task to Copilot with context from the PR with the manual fixes, plus the Slack thread that requested the RuboCop rule for this.

• Immensely unsatisfying compared to doing it myself (@issyl0), to be honest, although much more efficient for getting started!

• Supersedes rubocop: Avoid duplicate arch-specific versions in casks #21723 which got into a weird state.

---

• Casks sometimes end up with on_arm/on_intel blocks whose version values are identical, such as if at some point they released an ARM-only version which had to have its own stanza.
• Making this an autocorrecting RuboCop means that it'll save maintainer time spent manually editing these like in Combine duplicate arch-specific versions homebrew-cask#252975.

[Copilot]

Pull request overview

This PR enhances the RuboCop::Cop::Cask::OnSystemConditionals cop to autocorrect casks that redundantly nest arch-specific blocks when the version is identical (and to improve/extend existing sha256-nesting simplifications), reducing manual maintainer edits in homebrew-cask.

Changes:

• Add a new simplification pass to collapse on_intel/on_arm blocks when both contain identical version (and consolidate sha256 appropriately).
• Tighten sha256-nesting autocorrect by requiring the arch blocks to be siblings and removing whole lines on correction.
• Extend specs and Sorbet RBI signatures to cover the new pattern matching.

Reviewed changes

Copilot reviewed 2 out of 3 changed files in this pull request and generated 4 comments.

File	Description
Library/Homebrew/rubocops/cask/on_system_conditionals.rb	Adds simplify_arch_version_stanzas, adjusts sha256 simplification, and introduces an AST search for version+sha pairs.
Library/Homebrew/rubocops/cask/on_system_conditionals.rbi	Adds Sorbet signature for version_and_sha256_on_arch_stanzas.
Library/Homebrew/test/rubocops/cask/on_system_conditionals_spec.rb	Updates offense messaging expectations and adds new test cases for identical arch-version scenarios (including nested on_os).

Files not reviewed (1)

• Library/Homebrew/rubocops/cask/on_system_conditionals.rbi: Language not supported

Comments suppressed due to low confidence (1)

Library/Homebrew/rubocops/cask/on_system_conditionals.rb:92

• simplify_arch_version_stanzas has the same “last one wins” behavior as simplify_sha256_stanzas: it overwrites nodes[:arm]/nodes[:intel] for every match and then processes at most one pair. This will miss additional offending pairs (e.g., multiple on_os blocks each containing an on_intel/on_arm pair) and can also miss the only offending pair if a later non-matching block overwrites one side. Suggest grouping by block_node.parent (or walking sibling blocks) and applying the simplification per parent scope.

        def simplify_arch_version_stanzas
          nodes = {}

          version_and_sha256_on_arch_stanzas(cask_body) do |block_node, arch_method, version_value, sha256_value|
            arch = arch_method.to_s.delete_prefix("on_").to_sym
            nodes[arch] = { node: block_node, version_value:, sha256_value: }
          end

          return if !nodes.key?(:arm) || !nodes.key?(:intel)
          return if nodes[:arm][:node].parent != nodes[:intel][:node].parent

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[MikeMcQuaid]

Looks good when 🟢 and you're happy, thanks @issyl0!

[issyl0]

For d433677 I'm not sure if there actually are any, but it seemed legit as an idea to check for.