Add regex mode for word selection definition

[eyupcanakman]

Summary

• Add two modes for word selection definition in General preferences: List (default) and Regex.
• Keep default behavior in List mode and allow advanced users to opt into regex-based word selection.
• In Regex mode, select the regex match containing the click location, and fall back to classic list-based selection when there is no valid containing match.

Changes

• sources/iTermPreferences.h

  • Add iTermWordSelectionDefinitionMode.
  • Add keys for word selection mode and regex value.

• sources/iTermPreferences.m

  • Add defaults/computed values for mode and regex.
  • Add regex validation helper.

• sources/GeneralPreferencesViewController.m

  • Add a List / Regex popup next to "Characters considered part of a word".
  • Persist and restore per-mode values when switching modes.

• sources/iTermTextExtractor.m

  • Add regex-mode selection path for the containing match.
  • Keep list-mode selection behavior as fallback.

• iTerm2XCTests/iTermTextExtractorTest.m

  • Add regression coverage for regex-mode matching, invalid/empty regex fallback, and list-mode behavior.

Test

• xcodebuild build -project iTerm2.xcodeproj -scheme iTerm2 -destination 'platform=macOS' CODE_SIGNING_ALLOWED=NO CODE_SIGNING_REQUIRED=NO CODE_SIGN_IDENTITY= ENABLE_DEBUG_DYLIB=NO
  • Result: BUILD SUCCEEDED
• xcodebuild test -project iTerm2.xcodeproj -scheme iTerm2Tests -destination 'platform=macOS' CODE_SIGNING_ALLOWED=NO CODE_SIGNING_REQUIRED=NO CODE_SIGN_IDENTITY= ENABLE_DEBUG_DYLIB=NO -only-testing:iTerm2XCTests/iTermTextExtractorTest
  • Local result: build stopped before test execution by existing compile errors in iTerm2XCTests/iTermRuleTest.m on this checkout.

[gnachman]

This is the problem smart selection is meant to address. I know smart selection is not a great replacement for word selection, and quad-click is inconvenient, so there is definitely room for improvement but I think as written this would break too many workflows. Can you provide more details on your thinking and how you would use this?

[eyupcanakman]

@gnachman I understand the concern. I frequently need to copy full URLs from logs to a browser, and smart selection or quad-click feels too slow for this. To avoid breaking existing workflows, would you accept this as an opt-in toggle in Preferences > Advanced? This keeps the default unchanged but solves the issue for users like me who need it.

Let me know and I will update the PR.

[gnachman]

I think a better solution would be to extend "characters considered part of a word for selection" to be a regular expression.

Then you could put (https?://) in there, solving the problem pretty simply.

There would need to be two modes to switch between "list of characters" and "regex", although in the implementation they could both be implemented as regular expressions.

[eyupcanakman]

That makes a lot of sense. I will work on implementing this approach and update the PR.

[eyupcanakman]

@gnachman Implemented your suggestion. Word selection now has List (default) and Regex modes.

[gnachman]

This ended up being pretty complex. Committed as 54bd855

[eyupcanakman]

Got it, thanks. I'll check master to see how you implemented it.