perf: use execa to run swift directly

assets/recognizeText.swift

@@ -0,0 +1,82 @@
+// The Swift Programming Language
+// https://docs.swift.org/swift-book
+// swift build -c release
+
+import Cocoa
+import Vision
+
+let recognitionLevel = VNRequestTextRecognitionLevel.accurate
+let usesLanguageCorrection = true
+
+func captureScreen() -> CGImage? {
+    let screenRect = NSScreen.main?.frame ?? .zero
+    let imageRef = CGWindowListCreateImage(screenRect, .optionOnScreenOnly, kCGNullWindowID, .bestResolution)
+    return imageRef
+}
+
+func captureSelectedArea() -> CGImage? {
+    let task = Process()
+    task.launchPath = "/usr/sbin/screencapture"
+    task.arguments = ["-i", "-c"]
+    task.launch()
+    task.waitUntilExit()
+
+    guard let pasteboard = NSPasteboard.general.pasteboardItems?.first,
+          let fileType = pasteboard.types.first,
+          let data = pasteboard.data(forType: fileType),
+          let image = NSImage(data: data) else {
+        fputs("Error: failed to capture selected area\n", stderr)
+        return nil
+    }
+
+    var proposedRect = NSRect.zero
+    guard let imgRef = image.cgImage(forProposedRect: &proposedRect, context: nil, hints: nil) else {
+        fputs("Error: failed to convert NSImage to CGImage for captured area\n", stderr)
+        return nil
+    }
+
+    return imgRef
+}
+
+func main() -> Int32 {
+    let imgRef: CGImage? = captureSelectedArea()
+
+    guard let capturedImage = imgRef else {
+        fputs("Error: failed to capture image\n", stderr)
+        return 1
+    }
+
+    let request = VNRecognizeTextRequest()
+    request.recognitionLevel = recognitionLevel
+    request.usesLanguageCorrection = usesLanguageCorrection
+    if #available(macOS 13.0, *) {
+        request.automaticallyDetectsLanguage = true
+    }
+
+// "en-US", "fr-FR", "it-IT", "de-DE", "es-ES", "pt-BR", "zh-Hans", "zh-Hant", "yue-Hans", "yue-Hant", "ko-KR", "ja-JP", "ru-RU", "uk-UA"
+    request.recognitionLanguages = ["zh-Hans", "zh-Hant", "en-US", "ja-JP", "fr-FR", "de-DE", "es-ES", "pt-BR", "it-IT", "ko-KR", "ru-RU", "uk-UA"]
+
+    do {
+        try VNImageRequestHandler(cgImage: capturedImage, options: [:]).perform([request])
+    } catch {
+        fputs("Error: \(error.localizedDescription)", stderr)
+        return 1
+    }
+
+    guard let observations = request.results else {
+        fputs("Error: could not get text recognition results", stderr)
+        return 1
+    }
+
+    // Currently, we just join all recognized text by '\n'
+    // TODO: use the bounding box frame and the end character of each line to determine how to join the text.
+    let obs = observations
+        .map { $0.topCandidates(1).first?.string ?? "" }
+        .joined(separator: "\n")
+
+    print(obs)
+
+    return 0
+}
+
+exit(main())

src/ocr.tsx

@@ -1,3 +1,13 @@
+/*
+ * @author: tisfeng
+ * @createTime: 2023-05-15 23:31
+ * @lastEditor: tisfeng
+ * @lastEditTime: 2023-05-17 18:42
+ * @fileName: ocr.tsx
+ *
+ * Copyright (c) 2023 by ${git_name}, All Rights Reserved.
+ */
+
 import { closeMainWindow, open, showHUD } from "@raycast/api";
 import { recognizeText } from "./recognizeText";
 
@@ -6,11 +16,10 @@ export default async function command() {
 
   try {
     const recognizedText = await recognizeText();
-
     if (!recognizedText) {
       return await showHUD("❌ No text detected!");
     }
-    console.log(recognizedText);
+    console.log(`Recognized text: ${recognizedText}`);
 
     const encodedQueryText = encodeURIComponent(recognizedText);
     const easyDictUrl = `raycast://extensions/isfeng/easydict/easydict?fallbackText=${encodedQueryText}`;

src/recognizeText.ts

@@ -1,14 +1,21 @@
+/*
+ * @author: tisfeng
+ * @createTime: 2023-05-15 23:31
+ * @lastEditor: tisfeng
+ * @lastEditTime: 2023-05-17 18:40
+ * @fileName: recognizeText.ts
+ *
+ * Copyright (c) 2023 by ${git_name}, All Rights Reserved.
+ */
+
 import { environment } from "@raycast/api";
-import { join } from "path";
 import { execa, ExecaError } from "execa";
-import { chmod } from "fs/promises";
+import { join } from "path";
 
 const recognizeText = async () => {
-  const command = join(environment.assetsPath, "RecognizeText");
-  await chmod(command, "755");
-
   try {
-    const { stdout } = await execa(command);
+    const filePath = join(environment.assetsPath, "recognizeText.swift");
+    const { stdout } = await execa("swift", [filePath]);
     return stdout;
   } catch (error) {
     if ((error as ExecaError).stdout === "No text selected") {