jake/vOOice
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #429 from klaudworks/fix/ocr-window-detection

Browse Source 
Fix OCR capturing VoiceInk status overlay instead of frontmost app window
This commit is contained in:
Prakash Joshi Pax 2025-12-07 12:27:31 +05:45 committed by GitHub
commit 45dbd57722
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 42 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

52
VoiceInk/Services/ScreenCaptureService.swift
View File

@ -14,20 +14,52 @@ class ScreenCaptureService: ObservableObject {
category: "aienhancement" category: "aienhancement"
) )
private func getActiveWindowInfo() -> (title: String, ownerName: String, windowID: CGWindowID)? { private struct WindowCandidate {
let windowListInfo = CGWindowListCopyWindowInfo([.optionOnScreenOnly], kCGNullWindowID) as? [[String: Any]] ?? [] let title: String
let ownerName: String
let windowID: CGWindowID
let ownerPID: pid_t
let layer: Int32
}
if let frontWindow = windowListInfo.first(where: { info in private func getActiveWindowInfo() -> (title: String, ownerName: String, windowID: CGWindowID)? {
let layer = info[kCGWindowLayer as String] as? Int32 ?? 0 let currentPID = ProcessInfo.processInfo.processIdentifier
return layer == 0 let frontmostPID = NSWorkspace.shared.frontmostApplication?.processIdentifier
}) { let windowListInfo = CGWindowListCopyWindowInfo([.optionOnScreenOnly], kCGNullWindowID) as? [[String: Any]] ?? []
guard let windowID = frontWindow[kCGWindowNumber as String] as? CGWindowID,
let ownerName = frontWindow[kCGWindowOwnerName as String] as? String, let candidates = windowListInfo.compactMap { info -> WindowCandidate? in
let title = frontWindow[kCGWindowName as String] as? String else { guard let windowID = info[kCGWindowNumber as String] as? CGWindowID,
let ownerName = info[kCGWindowOwnerName as String] as? String,
let ownerPIDNumber = info[kCGWindowOwnerPID as String] as? NSNumber,
let layer = info[kCGWindowLayer as String] as? Int32 else {
return nil return nil
} }
return (title: title, ownerName: ownerName, windowID: windowID) let rawTitle = (info[kCGWindowName as String] as? String)?.trimmingCharacters(in: .whitespacesAndNewlines)
let resolvedTitle = rawTitle?.isEmpty == false ? rawTitle! : ownerName
return WindowCandidate(
title: resolvedTitle,
ownerName: ownerName,
windowID: windowID,
ownerPID: ownerPIDNumber.int32Value,
layer: layer
)
}
func isEligible(_ candidate: WindowCandidate) -> Bool {
guard candidate.layer == 0 else { return false }
guard candidate.ownerPID != currentPID else { return false }
return true
}
if let frontmostPID = frontmostPID,
let focusedWindow = candidates.first(where: { isEligible($0) && $0.ownerPID == frontmostPID }) {
return (title: focusedWindow.title, ownerName: focusedWindow.ownerName, windowID: focusedWindow.windowID)
}
if let fallbackWindow = candidates.first(where: isEligible) {
return (title: fallbackWindow.title, ownerName: fallbackWindow.ownerName, windowID: fallbackWindow.windowID)
} }
return nil return nil