我完全不会 Lua，但我用嘴做了一个 OCR

zh-Hans
zh-Hant
en-US

local hs = hs
local hotkey = hs.hotkey
local task = hs.task
local alert = hs.alert
local pasteboard = hs.pasteboard
local timer = hs.timer
local screenRecordingState = hs.screenRecordingState
local menubar = hs.menubar
local image = hs.image
local base64 = hs.base64

local ocr = {}

local screenshotCommand = "/usr/sbin/screencapture"
local osascriptCommand = "/usr/bin/osascript"
local menuBarIconSvg = [[
<svg xmlns="http://www.w3.org/2000/svg" width="18" height="18" viewBox="0 0 18 18">
  <g fill="none" stroke="black" stroke-width="1.05" stroke-linecap="round" stroke-linejoin="round" vector-effect="non-scaling-stroke">
    <path d="M2.4 5.1V3.6C2.4 2.9 2.9 2.4 3.6 2.4H5.1"/>
    <path d="M12.9 2.4H14.4C15.1 2.4 15.6 2.9 15.6 3.6V5.1"/>
    <path d="M15.6 12.9V14.4C15.6 15.1 15.1 15.6 14.4 15.6H12.9"/>
    <path d="M5.1 15.6H3.6C2.9 15.6 2.4 15.1 2.4 14.4V12.9"/>
    <circle cx="5.1" cy="9" r="1.9"/>
    <path d="M10.65 7H10.25C9.1 7 8.25 7.8 8.25 9C8.25 10.2 9.1 11 10.25 11H10.65"/>
    <path d="M12.1 11V7H13.55C14.35 7 14.9 7.5 14.9 8.2C14.9 8.95 14.35 9.45 13.55 9.45H12.1"/>
    <path d="M13.45 9.45L14.9 11"/>
  </g>
</svg>
]]

-- Cleanup stale temp files on load
for f in io.popen('ls /tmp/ocr_* 2>/dev/null'):lines() do
  os.remove(f)
end

local function trim(s) return (s or ""):gsub("^%s+", ""):gsub("%s+$", "") end

local function tmp(suffix)
  return "/tmp/ocr_" .. tostring(os.time()) .. "_" .. tostring(math.random(100000)) .. (suffix or "")
end

local function q(s) return '"' .. tostring(s):gsub("\\", "\\\\"):gsub('"', '\\"') .. '"' end

local function buildAppleScript(imagePath)
  local qp = q(imagePath)
  return table.concat({
    'use framework "Foundation"',
    'use framework "Vision"',
    'use scripting additions',
    "",
    "set theFile to current application's |NSURL|'s fileURLWithPath:" .. qp,
    "set requestHandler to current application's VNImageRequestHandler's alloc()'s initWithURL:theFile options:(missing value)",
    "set theRequest to current application's VNRecognizeTextRequest's alloc()'s init()",
    "theRequest's setRecognitionLevel:(current application's VNRequestTextRecognitionLevelAccurate)",
    'theRequest\'s setRecognitionLanguages:{"zh-Hans", "zh-Hant", "en-US"}',
    "theRequest's setUsesLanguageCorrection:false",
    "requestHandler's performRequests:(current application's NSArray's arrayWithObject:(theRequest)) |error|:(missing value)",
    "set theResults to theRequest's results()",
    'if theResults is missing value then return ""',
    "set theArray to current application's NSMutableArray's new()",
    "repeat with aResult in theResults",
    "    set theCandidates to aResult's topCandidates:1",
    "    if (theCandidates's |count|()) > 0 then",
    "        (theArray's addObject:(((theCandidates's objectAtIndex:0)'s |string|())))",
    "    end if",
    "end repeat",
    "return (theArray's componentsJoinedByString:linefeed) as text",
  }, "\n")
end

local function runOCR(imagePath, callback)
  local scriptPath = tmp(".applescript")
  local f, err = io.open(scriptPath, "w")
  if not f then
    os.remove(scriptPath)
    os.remove(imagePath)
    callback(nil, "OCR failed")
    return
  end
  f:write(buildAppleScript(imagePath))
  f:close()

  local t = task.new(osascriptCommand, function(exitCode, stdout, stderr)
    os.remove(scriptPath)

    if exitCode ~= 0 then
      os.remove(imagePath)
      callback(nil, "OCR failed")
      return
    end

    callback(stdout:gsub("\r?\n$", ""), nil)
  end, {scriptPath})

  if not t then
    os.remove(scriptPath)
    os.remove(imagePath)
    callback(nil, "OCR failed")
    return
  end

  t:start()
end

function ocr.runOnImage(imagePath)
  runOCR(imagePath, function(text, err)
    os.remove(imagePath)

    if not text then
      alert.show(err, 3)
      return
    end

    if text == "" then
      alert.show("No text found", 2)
      return
    end

    pasteboard.setContents(text)
  end)
end

function ocr.captureSelectionAndRecognize()
  if not screenRecordingState(false) then
    screenRecordingState(true)
    alert.show("Screen capture denied", 3)
    return
  end

  local imagePath = tmp(".png")
  local t = task.new(screenshotCommand, function(exitCode, _, stderr)
    if exitCode ~= 0 then
      os.remove(imagePath)
      alert.show("Screen capture failed", 1.8)
      return
    end

    timer.doAfter(0.05, function()
      ocr.runOnImage(imagePath)
    end)
  end, {"-i", "-x", imagePath})

  if not t then
    os.remove(imagePath)
    alert.show("Screen capture failed", 2)
    return
  end

  t:start()
end

local modifiers = {"ctrl"}
local triggerKey = "d"

if hotkey.assignable(modifiers, triggerKey) then
  hotkey.bind(modifiers, triggerKey, nil, function()
    ocr.captureSelectionAndRecognize()
  end)
else
  alert.show("Ctrl+" .. triggerKey .. " is reserved", 3)
end

local function createMenuBarItem()
  local item = menubar.new(true, "ocr_menu_bar")
  if not item then
    alert.show("OCR menu bar item failed", 3)
    return nil
  end

  local icon = image.imageFromURL("data:image/svg+xml;base64," .. base64.encode(menuBarIconSvg))
  if icon then
    item:setIcon(icon, true)
  else
    item:setTitle("OCR")
  end

  item:setTooltip("OCR selection")
  item:setClickCallback(function()
    ocr.captureSelectionAndRecognize()
  end)

  return item
end

ocr.menuBar = createMenuBarItem()

return ocr

require("ocr")