[fredsimard]

What this changes

Fixes the six bugs found in a full audit of the Swift sources, corrects the documentation where it described behaviour the code did not have, and records the result as 1.0.1. One commit per issue.

Commit	Fix
aef05c8	.bool results from executeAppleScriptViaCommandLine(_:with:) were always false. osascript prints true/false as text, and Int("true") is nil. The string is now compared, with integer coercion kept as a fallback so both execution paths agree.
cf777d5	The same method deadlocked on output larger than the pipe buffer (~64 KB): it waited for the process to exit before draining the pipe, while the child blocked on write. The pipe is now read first.
3c1403a	Variable substitution rescanned text it had already substituted, so a value containing a placeholder name was rewritten by a later variable — a file named $folderPath.txt passed as folderPath had that variable's value spliced into its name. The script is now walked once, matching the longest key at each $.
39af9de	toHFSPath() prepended the startup disk name to every path, so /Volumes/Backup/notes.txt became Macintosh HD:Volumes:Backup:notes.txt instead of Backup:notes.txt. The volume name and mount point now come from the path itself.
e0161b9	An NSNumber bridges to Bool whatever it holds, so NSNumber(value: 1) from JSONSerialization or Objective-C was substituted as true instead of 1. Numbers are unwrapped to the Swift type they actually carry before any type matching.
a5f8452	resetAutomationPermission(for:) returned true whenever tccutil launched, ignoring its exit status.

4ef0079 covers the documentation corrections, and 95b1c9d the changelog. 5956262 adds the local review notes to .gitignore.

Why

Three of these produce silently wrong results rather than errors: a .bool script that answers true is read as false, a filename that happens to contain a variable name is corrupted, and a count arriving as NSNumber is substituted as true. The pipe deadlock hangs the calling thread permanently, and is reachable from a .json listing of a large folder.

How it was tested

• macOS version: 26.6.1, Swift 6.3.3
• Tested against application: none — no change touches a specific app's dictionary.

swift build and swift test pass (22 tests). The two path-conversion tests were updated: they used /Users/roger/Desktop, which does not exist, and toHFSPath() now reads the volume from the file system, so they use NSHomeDirectory().

Four regression tests were added, one per fix that can be pinned deterministically. Each was run against the pre-fix sources and fails there, so they guard the behaviour rather than merely describe it:

Test	Result before the fix
testSubstitutedValueIsNotRescanned	set p to "Roger.txt", the name variable spliced into a filename that contained its placeholder
testNSNumberIsNotRenderedAsBoolean	true true, a count of 1 rendered as a boolean
testJSONNumbersKeepTheirType	true true 1.5, the same through JSONSerialization
testCommandLineBooleanResults	false for a script returning true

The last one executes, unlike the rest of the suite. It runs return true and return false through osascript, which drives no application and needs no Automation permission, so it is safe in CI; the file header notes the exception.

Checklist

• swift build and swift test pass.
• New or changed public API has a /// doc comment describing purpose, parameters and return value.
• Docs updated if behaviour, parameters or return values changed.
• Tests added for anything deterministic (escaping, substitution, parsing).
• No breaking change to public API — no signature changed, but three behaviour changes are worth a reviewer's attention:
  • toHFSPath() now requires the path to exist and returns nil otherwise, where it previously returned a string for any input.
  • Float and UInt variables now render as numeric literals instead of escaped text.
  • resetAutomationPermission(for:) now returns false when tccutil exits non-zero.