Retry OCR on upscaled images when nothing is read

Tiny UI text (around 7pt) reads as nothing at native size — verified
locally. When a pass returns fewer than three characters, OCR retries at
2x and 3x and keeps the fullest result. The smoke test covers a tiny-text
case so the fallback stays working.

💘 Generated with Crush

Assisted-by: Crush:deepseek-v4-flash-0731

Sources/Kapture/App.swift

@@ -123,8 +123,8 @@ enum LoginItem {
     }
 }
 
-func renderTestImage(_ text: String, fontSize: CGFloat = 56) -> CGImage? {
+func renderTestImage(_ text: String, fontSize: CGFloat = 56, width: CGFloat = 1900, height: CGFloat = 220) -> CGImage? {
-    let size = NSSize(width: 1900, height: 220)
+    let size = NSSize(width: width, height: height)
     guard let rep = NSBitmapImageRep(
         bitmapDataPlanes: nil, pixelsWide: Int(size.width), pixelsHigh: Int(size.height),
         bitsPerSample: 8, samplesPerPixel: 4, hasAlpha: true, isPlanar: false,
@@ -138,15 +138,20 @@ func renderTestImage(_ text: String, fontSize: CGFloat = 56) -> CGImage? {
         .font: NSFont.systemFont(ofSize: fontSize),
         .foregroundColor: NSColor.black,
     ]
-    text.draw(at: NSPoint(x: 60, y: 75), withAttributes: attrs)
+    // Centre the text vertically so tiny canvases still contain it.
+    let textHeight = ("Ag" as NSString).size(withAttributes: attrs).height
+    let y = max(4, (size.height - textHeight) / 2)
+    text.draw(at: NSPoint(x: 20, y: y), withAttributes: attrs)
     NSGraphicsContext.restoreGraphicsState()
     return rep.cgImage
 }
 
 func runOCRSelfTest() {
-    let cases: [(expected: String, contains: String)] = [
+    let cases: [(expected: String, contains: String, fontSize: CGFloat, width: CGFloat, height: CGFloat)] = [
-        ("Kapture OCR test 1234, hello world", "Kapture OCR test 1234"),
+        ("Kapture OCR test 1234, hello world", "Kapture OCR test 1234", 56, 1900, 220),
-        ("你好世界，这是一个测试", "你好世界"),
+        ("你好世界，这是一个测试", "你好世界", 56, 1900, 220),
+        // Tiny text: needs the upscale retry to read anything at all.
+        ("会话导出并发冲突报错 export fail", "导出", 7, 320, 40),
     ]
     var index = 0
 
@@ -154,9 +159,9 @@ func runOCRSelfTest() {
         guard index < cases.count else { exit(0) }
         let current = cases[index]
         index += 1
-        guard let image = renderTestImage(current.expected) else { exit(1) }
+        guard let image = renderTestImage(current.expected, fontSize: current.fontSize, width: current.width, height: current.height) else { exit(1) }
         OCRService.shared.recognize(image) { recognized in
-            print("case:     \(current.expected)")
+            print("case:     \(current.expected) @\(Int(current.fontSize))pt")
             print("got:      \(recognized)")
             let ok = recognized.contains(current.contains)
             print(ok ? "PASS" : "FAIL")

Sources/Kapture/OCRService.swift

@@ -8,34 +8,66 @@ final class OCRService {
 
     func recognize(_ cgImage: CGImage, completion: @escaping (String) -> Void) {
         queue.async {
-            let request = VNRecognizeTextRequest()
+            let text = Self.perform(cgImage)
-            request.recognitionLevel = .accurate
+            // Tiny text (small UI labels) often reads as nothing at native
-            request.usesLanguageCorrection = true
+            // size; retry upscaled and keep whichever pass read more.
-            request.recognitionLanguages = ["en-US", "zh-Hans", "zh-Hant", "ja-JP", "ko-KR"]
+            guard text.count < 3 else {
-            // Without this, Vision defaults to the first listed language and
+                completion(text)
-            // renders CJK as Latin garbage (verified: 你好世界 -> "177WR").
-            if #available(macOS 13.0, *) {
-                request.automaticallyDetectsLanguage = true
-            }
-
-            let handler = VNImageRequestHandler(cgImage: cgImage, options: [:])
-            do {
-                try handler.perform([request])
-            } catch {
-                completion("")
                 return
             }
-
+            var best = text
-            let observations = request.results ?? []
+            for factor in [CGFloat(2), CGFloat(3)] {
-            let lines = observations
+                guard let upscaled = Self.upscaled(cgImage, factor: factor) else { continue }
-                .sorted { a, b in
+                let candidate = Self.perform(upscaled)
-                    let ay = a.boundingBox.midY
+                if candidate.count > best.count { best = candidate }
-                    let by = b.boundingBox.midY
+                if best.count >= 3 { break }
-                    if abs(ay - by) > 0.02 { return ay > by }
+            }
-                    return a.boundingBox.minX < b.boundingBox.minX
+            completion(best)
-                }
-                .compactMap { $0.topCandidates(1).first?.string }
-            completion(lines.joined(separator: "\n"))
         }
     }
+
+    private static func perform(_ cgImage: CGImage) -> String {
+        let request = VNRecognizeTextRequest()
+        request.recognitionLevel = .accurate
+        request.usesLanguageCorrection = true
+        request.recognitionLanguages = ["en-US", "zh-Hans", "zh-Hant", "ja-JP", "ko-KR"]
+        // Without this, Vision defaults to the first listed language and
+        // renders CJK as Latin garbage (verified: 你好世界 -> "177WR").
+        if #available(macOS 13.0, *) {
+            request.automaticallyDetectsLanguage = true
+        }
+
+        let handler = VNImageRequestHandler(cgImage: cgImage, options: [:])
+        do {
+            try handler.perform([request])
+        } catch {
+            return ""
+        }
+
+        let observations = request.results ?? []
+        let lines = observations
+            .sorted { a, b in
+                let ay = a.boundingBox.midY
+                let by = b.boundingBox.midY
+                if abs(ay - by) > 0.02 { return ay > by }
+                return a.boundingBox.minX < b.boundingBox.minX
+            }
+            .compactMap { $0.topCandidates(1).first?.string }
+        return lines.joined(separator: "\n")
+    }
+
+    private static func upscaled(_ cgImage: CGImage, factor: CGFloat) -> CGImage? {
+        let width = Int(CGFloat(cgImage.width) * factor)
+        let height = Int(CGFloat(cgImage.height) * factor)
+        guard width > 0, height > 0, width <= 20000, height <= 20000 else { return nil }
+        guard let ctx = CGContext(
+            data: nil, width: width, height: height,
+            bitsPerComponent: 8, bytesPerRow: 0,
+            space: CGColorSpaceCreateDeviceRGB(),
+            bitmapInfo: CGImageAlphaInfo.premultipliedLast.rawValue
+        ) else { return nil }
+        ctx.interpolationQuality = .high
+        ctx.draw(cgImage, in: CGRect(x: 0, y: 0, width: CGFloat(width), height: CGFloat(height)))
+        return ctx.makeImage()
+    }
 }

Sources/Kapture/OverlayView.swift

@@ -97,7 +97,9 @@ final class OverlayView: NSView {
         }
 
         // Esc hint — the only way out once clicks are consumed by the tap.
-        let hint = "Drag to select · hover + click a window to capture it · Esc to cancel"
+        let hint = frozenImage != nil
+            ? "Frozen screen · drag to select · Esc to cancel"
+            : "Drag to select · hover + click a window to capture it · Esc to cancel"
         let hintAttrs: [NSAttributedString.Key: Any] = [
             .font: NSFont.monospacedSystemFont(ofSize: 12, weight: .medium),
             .foregroundColor: NSColor.white.withAlphaComponent(0.85),
@@ -115,22 +117,17 @@ final class OverlayView: NSView {
 
         guard let controller else { return }
 
-        // Frozen mode: no window highlight; the marquee is cut straight out
+        // Frozen mode: same look as live mode — a dark veil over everything,
-        // of the still image so it shows the frozen pixels.
+        // with the selection showing the frozen pixels undimmed.
-        if frozenImage != nil {
+        if let frozenImage {
+            NSColor.black.withAlphaComponent(0.35).setFill()
+            NSBezierPath(rect: bounds).fill()
+
             if let sel = controller.marqueeRect, sel.width > 0, sel.height > 0 {
                 let vf = viewRect(sel)
-                // Dim everything except the selection (even-odd fill).
-                let maskPath = CGMutablePath()
-                maskPath.addRect(bounds)
-                maskPath.addRect(vf)
                 NSGraphicsContext.current?.saveGraphicsState()
-                let ctx = NSGraphicsContext.current?.cgContext
+                NSBezierPath(rect: vf).addClip()
-                ctx?.addPath(maskPath)
+                drawFrozen(frozenImage)
-                ctx?.fillPath(using: .evenOdd)
-                NSColor.black.withAlphaComponent(0.35).setFill()
-                ctx?.addPath(maskPath)
-                ctx?.fillPath(using: .evenOdd)
                 NSGraphicsContext.current?.restoreGraphicsState()
 
                 let border = NSBezierPath(rect: vf)

build-app.sh

@@ -108,9 +108,9 @@ cat > "$APP/Contents/Info.plist" <<'PLIST'
     <key>CFBundleIconFile</key>
     <string>Icon</string>
     <key>CFBundleShortVersionString</key>
-    <string>1.10.0</string>
+    <string>1.10.2</string>
     <key>CFBundleVersion</key>
-    <string>100</string>
+    <string>102</string>
     <key>LSUIElement</key>
     <true/>
     <key>NSHighResolutionCapable</key>