dsh-doc-assistant 0.1.14 → 0.1.18

package/src/host/ocr.js

@@ -1,253 +1,163 @@
-/**
- * Recognize text on a page PNG with macOS Vision.
- *
- * The reader draws only the PNG. These boxes are hit targets: the recognized
- * string is returned to the selection menu and is not painted again.
- * Coordinates are fractions of the image, origin at the top left, matching
- * `pdfText.js`. Other platforms, or a machine without the Swift compiler,
- * return an empty line list and the reader keeps the PDF text layer.
- *
- * @module dsh-doc-assistant/host/ocr
+/** Local page OCR, using macOS Vision or Windows' built-in WinRT OCR.
+ * Both implementations return top-left, normalized hit boxes. No text is
+ * painted over the page and no document contents leave the machine here.
  */
-
-import { spawn, execFile } from 'node:child_process';
-import { createHash } from 'node:crypto';
-import { access, chmod, mkdir, readFile, rename, stat, writeFile } from 'node:fs/promises';
-import { tmpdir } from 'node:os';
-import { join } from 'node:path';
-import { promisify } from 'node:util';
+import { spawn } from 'node:child_process';
+import { readFile, rename, stat, writeFile } from 'node:fs/promises';
+import { win32 } from 'node:path';
 import { fileURLToPath } from 'node:url';
 
-const execFileAsync = promisify(execFile);
-const SCRIPT = fileURLToPath(new URL('./ocr_vision.swift', import.meta.url));
-const MIN_CONFIDENCE = 0.2;
+const MAC_SCRIPT = fileURLToPath(new URL('./ocr_vision.js', import.meta.url));
+const WINDOWS_SCRIPT = fileURLToPath(new URL('./ocr_windows.ps1', import.meta.url));
+const CACHE_VERSION = 2;
 const MAX_LINES = 500;
-
-/** @type {Promise<string|null> | null} */
-let binaryPromise = null;
-
-/** @type {Map<string, Promise<{ aspect: number, lines: object[] }>>} */
+const MAX_OUTPUT = 2_000_000;
 const inflight = new Map();
 
-function clamp01(value) {
-  const number = Number(value);
-  if (!Number.isFinite(number)) return 0;
-  return Math.max(0, Math.min(1, number));
-}
-
-function round4(value) {
-  return Math.round(value * 10000) / 10000;
-}
-
-function compact(text) {
-  return String(text || '').replace(/\s+/g, '').toLowerCase();
-}
-
-/** Same words recognized twice, once as a line and once as a tighter box. */
+function clamp01(value) { return Math.max(0, Math.min(1, Number(value) || 0)); }
+function round4(value) { return Math.round(value * 10000) / 10000; }
+function compact(value) { return String(value || '').replace(/\s+/g, '').toLowerCase(); }
 function sameSpot(a, b) {
   const ix = Math.min(a.x + a.w, b.x + b.w) - Math.max(a.x, b.x);
   const iy = Math.min(a.y + a.h, b.y + b.h) - Math.max(a.y, b.y);
-  if (ix <= 0 || iy <= 0) return false;
   const smaller = Math.min(a.w * a.h, b.w * b.h);
-  return smaller > 0 && (ix * iy) / smaller > 0.6;
+  return ix > 0 && iy > 0 && smaller > 0 && ix * iy / smaller > .6;
 }
 
-/**
- * Drop empty, low-confidence, and duplicate boxes. `y` is already the top edge.
- *
- * @param {unknown} raw
- * @returns {Array<{ text: string, x: number, y: number, w: number, h: number, ocr: true }>}
- */
+/** Drop empty, invalid, low-confidence and duplicate hit boxes. */
 export function normalizeOcrLines(raw) {
   if (!Array.isArray(raw)) return [];
-  /** @type {Array<{ text: string, x: number, y: number, w: number, h: number, confidence: number }>} */
   const rows = [];
   for (const row of raw) {
     if (!row || typeof row !== 'object') continue;
     const text = String(row.text || '').replace(/\s+/g, ' ').trim();
     if (!text || text.length > 400) continue;
+    if (![row.x, row.y, row.w, row.h].every(value => Number.isFinite(Number(value)))) continue;
     const confidence = Number(row.confidence);
-    if (Number.isFinite(confidence) && confidence < MIN_CONFIDENCE) continue;
-    const w = clamp01(row.w);
-    const h = clamp01(row.h);
-    if (w < 0.003 || h < 0.003) continue;
-    rows.push({
-      text,
-      x: round4(clamp01(row.x)),
-      y: round4(clamp01(row.y)),
-      w: round4(w),
-      h: round4(h),
-      confidence: Number.isFinite(confidence) ? confidence : 1,
-    });
+    if (Number.isFinite(confidence) && confidence < .2) continue;
+    const x = clamp01(row.x), y = clamp01(row.y);
+    const w = Math.min(clamp01(row.w), 1 - x), h = Math.min(clamp01(row.h), 1 - y);
+    if (w < .003 || h < .003) continue;
+    rows.push({ text, x: round4(x), y: round4(y), w: round4(w), h: round4(h),
+      confidence: Number.isFinite(confidence) ? confidence : 1 });
   }
-  rows.sort((a, b) => b.confidence - a.confidence || (b.w * b.h) - (a.w * a.h));
-  /** @type {typeof rows} */
+  rows.sort((a, b) => b.confidence - a.confidence || b.w * b.h - a.w * a.h);
   const kept = [];
   for (const row of rows) {
-    const key = compact(row.text);
-    const duplicate = kept.some((other) => compact(other.text) === key && sameSpot(row, other));
-    if (!duplicate) kept.push(row);
+    if (!kept.some(other => compact(other.text) === compact(row.text) && sameSpot(row, other))) kept.push(row);
   }
   kept.sort((a, b) => a.y - b.y || a.x - b.x);
-  if (kept.length > MAX_LINES) kept.length = MAX_LINES;
-  return kept.map((row) => ({
-    text: row.text,
-    x: row.x,
-    y: row.y,
-    w: row.w,
-    h: row.h,
-    ocr: true,
-  }));
-}
-
-async function scriptHash() {
-  const source = await readFile(SCRIPT);
-  return createHash('sha256').update(source).digest('hex').slice(0, 16);
-}
-
-async function compileBinary(home) {
-  if (process.platform !== 'darwin') return null;
-  const hash = await scriptHash();
-  const dir = join(home, 'bin');
-  const dest = join(dir, `ocr-vision-${hash}`);
-  try {
-    await access(dest);
-    return dest;
-  } catch {
-    // Compile once, then reuse the binary.
-  }
-  await mkdir(dir, { recursive: true });
-  const tmp = join(tmpdir(), `dsh-ocr-vision-${hash}`);
-  try {
-    await execFileAsync('xcrun', ['swiftc', '-O', '-o', tmp, SCRIPT], {
-      timeout: 180_000,
-      maxBuffer: 4_000_000,
-    });
-  } catch (error) {
-    const message = error instanceof Error ? error.message : String(error);
-    console.error(`[dsh-doc-assistant] 页面文字识别不可用：${message.split('\n')[0]}`);
-    return null;
-  }
-  await chmod(tmp, 0o755);
-  await rename(tmp, dest);
-  return dest;
+  return kept.slice(0, MAX_LINES).map(({ confidence, ...row }) => ({ ...row, ocr: true }));
 }
 
-function ensureBinary(home) {
-  if (process.platform !== 'darwin') return Promise.resolve(null);
-  if (!binaryPromise) {
-    binaryPromise = compileBinary(home).catch((error) => {
-      const message = error instanceof Error ? error.message : String(error);
-      console.error(`[dsh-doc-assistant] 页面文字识别不可用：${message.split('\n')[0]}`);
-      return null;
-    });
-  }
-  return binaryPromise;
+/** Exported so platform launch/argument handling can be checked on any OS. */
+export function nativeOcrCommand(pngPath, platform = process.platform) {
+  if (platform === 'darwin') return { command: '/usr/bin/osascript',
+    args: ['-l', 'JavaScript', MAC_SCRIPT, pngPath], engine: 'macos-vision' };
+  if (platform === 'win32') return {
+    command: win32.join(process.env.SystemRoot || 'C:\\Windows', 'System32', 'WindowsPowerShell', 'v1.0', 'powershell.exe'),
+    args: ['-NoLogo', '-NoProfile', '-NonInteractive', '-ExecutionPolicy', 'Bypass', '-File', WINDOWS_SCRIPT, '-ImagePath', pngPath],
+    engine: 'windows-ocr' };
+  return null;
 }
 
-function runBinary(bin, pngPath) {
-  return new Promise((resolve) => {
-    let child;
-    try {
-      child = spawn(bin, [pngPath], { stdio: ['ignore', 'pipe', 'pipe'] });
-    } catch {
-      resolve(null);
-      return;
-    }
+function runNative(command, args) {
+  return new Promise(resolve => {
+    let child, done = false, size = 0;
     const out = [];
-    const timer = setTimeout(() => {
-      child.kill('SIGKILL');
-    }, 60_000);
-    child.stdout.on('data', (chunk) => out.push(chunk));
-    child.on('error', () => {
-      clearTimeout(timer);
-      resolve(null);
+    const finish = value => {
+      if (done) return;
+      done = true; clearTimeout(timer); resolve(value);
+    };
+    const timer = setTimeout(() => { child?.kill(); finish({ ok: false, code: 'timeout' }); }, 60_000);
+    try { child = spawn(command, args, { stdio: ['ignore', 'pipe', 'pipe'], windowsHide: true }); }
+    catch { finish({ ok: false, code: 'unavailable' }); return; }
+    child.stdout.on('data', chunk => {
+      size += chunk.length;
+      if (size > MAX_OUTPUT) { child.kill(); finish({ ok: false, code: 'output' }); }
+      else out.push(chunk);
     });
+    child.stderr.on('data', () => {});
+    child.on('error', () => finish({ ok: false, code: 'unavailable' }));
     child.on('close', () => {
-      clearTimeout(timer);
-      const text = Buffer.concat(out).toString('utf8').trim();
-      try {
-        const json = JSON.parse(text);
-        resolve(json && json.ok === true ? json : null);
-      } catch {
-        resolve(null);
-      }
+      try { finish(JSON.parse(Buffer.concat(out).toString('utf8').replace(/^\uFEFF/, '').trim())); }
+      catch { finish({ ok: false, code: 'unavailable' }); }
     });
   });
 }
 
-async function readCache(cachePath, pngStat) {
+function errorMessage(code, engine) {
+  if (code === 'language') return 'Windows 未安装 OCR 语言：请在「设置 → 时间和语言 → 语言和区域」添加中文（简体）的文字识别组件。';
+  if (code === 'timeout') return '这一页文字识别超时，请重试或使用支持图片的模型提问。';
+  if (code === 'image') return '无法读取这一页的图片。';
+  if (code === 'recognition') return '这一页文字识别失败，请重试或使用支持图片的模型提问。';
+  if (engine === 'windows-ocr') return 'Windows 系统文字识别不可用，请检查 OCR 语言组件，或使用支持图片的模型提问。';
+  if (engine === 'macos-vision') return 'macOS 系统文字识别不可用，请重试或使用支持图片的模型提问。';
+  return '当前系统没有可用的本地文字识别，请使用支持图片的模型提问。';
+}
+
+async function cachedResult(cachePath, info) {
   try {
-    const cached = JSON.parse(await readFile(cachePath, 'utf8'));
-    if (!cached || cached.ok !== true || !Array.isArray(cached.lines)) return null;
-    if (cached.pngSize !== pngStat.size) return null;
-    if (Number(cached.pngMtime) !== Math.round(pngStat.mtimeMs)) return null;
-    return cached;
-  } catch {
-    return null;
-  }
+    const result = JSON.parse(await readFile(cachePath, 'utf8'));
+    if (result?.ok !== true || result.version !== CACHE_VERSION || !Array.isArray(result.lines) ||
+        result.pngSize !== info.size || result.pngMtime !== Math.round(info.mtimeMs)) return null;
+    return result;
+  } catch { return null; }
 }
 
-/**
- * OCR one cached page PNG. Failures are an empty list, never a throw.
- *
- * @param {string} pngPath
- * @param {string} home plugin data directory
- * @returns {Promise<{ aspect: number, lines: object[] }>}
- */
-export async function recognizePage(pngPath, home) {
-  const empty = { aspect: 0, lines: [] };
-  if (process.platform !== 'darwin') return empty;
-  const pending = inflight.get(pngPath);
-  if (pending) return pending;
-  const job = recognizeUncached(pngPath, home).finally(() => {
-    if (inflight.get(pngPath) === job) inflight.delete(pngPath);
-  });
-  inflight.set(pngPath, job);
-  return job;
+function pageResult(raw, engine, cached = false) {
+  const lines = normalizeOcrLines(raw.lines);
+  const width = Number(raw.width) || 0, height = Number(raw.height) || 0;
+  const language = typeof raw.language === 'string' ? raw.language.slice(0, 80) : '';
+  const warning = engine === 'windows-ocr' && language && !/^zh\b/i.test(language)
+    ? '系统没有中文 OCR 语言，中文识别可能不完整；可添加中文文字识别组件或使用支持图片的模型。' : '';
+  return { aspect: width > 0 && height > 0 ? width / height : 0, lines,
+    ocr: { status: lines.length ? 'ready' : 'empty', engine, language, cached, ...(warning ? { warning, code: 'missing-chinese' } : {}) } };
 }
 
-async function recognizeUncached(pngPath, home) {
-  const empty = { aspect: 0, lines: [] };
-  let pngStat;
+/** Cache-only access for chat indexing. Never starts native OCR or compilation. */
+export async function cachedPageRecognition(imagePath) {
   try {
-    pngStat = await stat(pngPath);
-  } catch {
-    return empty;
-  }
-  if (!pngStat.isFile() || pngStat.size < 32) return empty;
-  const cachePath = `${pngPath}.ocr.json`;
-  const cached = await readCache(cachePath, pngStat);
-  if (cached) {
-    const width = Number(cached.width) || 0;
-    const height = Number(cached.height) || 0;
-    return {
-      aspect: width > 0 && height > 0 ? width / height : 0,
-      lines: normalizeOcrLines(cached.lines),
-    };
-  }
-  const bin = await ensureBinary(home);
-  if (!bin) return empty;
-  const result = await runBinary(bin, pngPath);
-  if (!result) return empty;
-  const lines = normalizeOcrLines(result.lines);
-  const width = Number(result.width) || 0;
-  const height = Number(result.height) || 0;
-  const payload = {
-    ok: true,
-    pngSize: pngStat.size,
-    pngMtime: Math.round(pngStat.mtimeMs),
-    width,
-    height,
-    lines,
-  };
-  try {
-    await writeFile(cachePath, JSON.stringify(payload));
-  } catch {
-    // The page can still be selected; the next open recognizes it again.
-  }
-  return {
-    aspect: width > 0 && height > 0 ? width / height : 0,
-    lines,
-  };
+    const info = await stat(imagePath);
+    const cached = await cachedResult(`${imagePath}.ocr.json`, info);
+    return cached ? pageResult(cached, cached.engine || 'local-ocr', true) : null;
+  } catch { return null; }
+}
+
+/** Recognize one cached raster. Failed runs are reported and never cached.
+ * `home` stays in the signature for callers of previous releases. A test or
+ * platform adapter may provide `platform`/`run`; production uses native APIs.
+ */
+export async function recognizePage(pngPath, _home, options = {}) {
+  const descriptor = nativeOcrCommand(pngPath, options.platform);
+  const engine = descriptor?.engine || 'unavailable';
+  const empty = (code = 'unavailable') => ({ aspect: 0, lines: [],
+    ocr: { status: code === 'unavailable' || code === 'language' ? 'unavailable' : 'error', engine,
+      code, error: errorMessage(code, engine) } });
+  if (!descriptor) return empty();
+  const key = `${engine}\0${pngPath}`;
+  if (inflight.has(key)) return inflight.get(key);
+  const job = (async () => {
+    let info;
+    try { info = await stat(pngPath); } catch { return empty('image'); }
+    if (!info.isFile() || info.size < 32) return empty('image');
+    const cachePath = `${pngPath}.ocr.json`;
+    const cached = await cachedResult(cachePath, info);
+    if (cached) return pageResult(cached, cached.engine || engine, true);
+    let raw;
+    try { raw = await (options.run || runNative)(descriptor.command, descriptor.args); }
+    catch { return empty(); }
+    if (!raw || raw.ok !== true) return empty(raw?.code || 'unavailable');
+    if (!(Number(raw.width) > 0 && Number(raw.height) > 0)) return empty('image');
+    const result = pageResult(raw, engine);
+    const payload = { ok: true, version: CACHE_VERSION, pngSize: info.size,
+      pngMtime: Math.round(info.mtimeMs), engine, width: Number(raw.width) || 0,
+      height: Number(raw.height) || 0, language: result.ocr.language, lines: result.lines };
+    const temp = `${cachePath}.${process.pid}.tmp`;
+    try { await writeFile(temp, JSON.stringify(payload)); await rename(temp, cachePath); }
+    catch { /* A read-only cache directory must not prevent selecting a page. */ }
+    return result;
+  })().finally(() => { if (inflight.get(key) === job) inflight.delete(key); });
+  inflight.set(key, job);
+  return job;
 }

package/src/host/ocr_vision.js

@@ -0,0 +1,39 @@
+// Run by macOS /usr/bin/osascript -l JavaScript. Vision is provided by macOS;
+// unlike the Swift helper this path does not require Xcode command-line tools.
+ObjC.import('AppKit');
+ObjC.import('Foundation');
+ObjC.import('Vision');
+
+function run(argv) {
+  try {
+    var data = $.NSData.dataWithContentsOfFile(argv[0]);
+    var bitmap = $.NSBitmapImageRep.imageRepWithData(data);
+    if (!bitmap) return JSON.stringify({ ok: false, code: 'image', lines: [] });
+    var request = $.VNRecognizeTextRequest.alloc.init;
+    request.recognitionLevel = 0; // VNRequestTextRecognitionLevelAccurate
+    request.usesLanguageCorrection = true;
+    request.minimumTextHeight = 0.008;
+    request.recognitionLanguages = $(['zh-Hans', 'en-US']);
+    var handler = $.VNImageRequestHandler.alloc.initWithCGImageOptions(bitmap.CGImage, $({}));
+    var error = Ref();
+    if (!handler.performRequestsError($([request]), error)) {
+      return JSON.stringify({ ok: false, code: 'recognition', lines: [] });
+    }
+    var results = request.results;
+    var lines = [];
+    for (var i = 0; i < Number(results.count); i += 1) {
+      var observation = results.objectAtIndex(i);
+      var candidates = observation.topCandidates(1);
+      if (Number(candidates.count) === 0) continue;
+      var best = candidates.objectAtIndex(0);
+      var box = observation.boundingBox;
+      lines.push({ text: ObjC.unwrap(best.string), x: Number(box.origin.x),
+        y: 1 - Number(box.origin.y) - Number(box.size.height),
+        w: Number(box.size.width), h: Number(box.size.height), confidence: Number(best.confidence) });
+    }
+    return JSON.stringify({ ok: true, width: Number(bitmap.pixelsWide),
+      height: Number(bitmap.pixelsHigh), language: 'zh-Hans,en-US', lines: lines });
+  } catch (error) {
+    return JSON.stringify({ ok: false, code: 'unavailable', lines: [] });
+  }
+}

package/src/host/ocr_windows.ps1

@@ -0,0 +1,89 @@
+# Windows PowerShell 5.1 provides WinRT interop without a separately installed
+# OCR application. Keep the source ASCII so PowerShell's legacy parser reads it
+# identically on Windows machines with different system code pages.
+param([Parameter(Mandatory=$true)][string]$ImagePath)
+$ErrorActionPreference = 'Stop'
+[Console]::OutputEncoding = New-Object System.Text.UTF8Encoding($false)
+$ProgressPreference = 'SilentlyContinue'
+
+try {
+  Add-Type -AssemblyName System.Runtime.WindowsRuntime
+  $null = [Windows.Storage.StorageFile, Windows.Storage, ContentType=WindowsRuntime]
+  $null = [Windows.Storage.Streams.IRandomAccessStream, Windows.Storage.Streams, ContentType=WindowsRuntime]
+  $null = [Windows.Storage.FileAccessMode, Windows.Storage, ContentType=WindowsRuntime]
+  $null = [Windows.Graphics.Imaging.BitmapDecoder, Windows.Graphics.Imaging, ContentType=WindowsRuntime]
+  $null = [Windows.Graphics.Imaging.BitmapTransform, Windows.Graphics.Imaging, ContentType=WindowsRuntime]
+  $null = [Windows.Graphics.Imaging.SoftwareBitmap, Windows.Graphics.Imaging, ContentType=WindowsRuntime]
+  $null = [Windows.Graphics.Imaging.BitmapPixelFormat, Windows.Graphics.Imaging, ContentType=WindowsRuntime]
+  $null = [Windows.Graphics.Imaging.BitmapAlphaMode, Windows.Graphics.Imaging, ContentType=WindowsRuntime]
+  $null = [Windows.Graphics.Imaging.ExifOrientationMode, Windows.Graphics.Imaging, ContentType=WindowsRuntime]
+  $null = [Windows.Graphics.Imaging.ColorManagementMode, Windows.Graphics.Imaging, ContentType=WindowsRuntime]
+  $null = [Windows.Media.Ocr.OcrEngine, Windows.Foundation, ContentType=WindowsRuntime]
+  $null = [Windows.Media.Ocr.OcrResult, Windows.Foundation, ContentType=WindowsRuntime]
+  $null = [Windows.Globalization.Language, Windows.Globalization, ContentType=WindowsRuntime]
+
+  $asTask = [System.WindowsRuntimeSystemExtensions].GetMethods() | Where-Object {
+    $_.Name -eq 'AsTask' -and $_.IsGenericMethodDefinition -and
+    $_.GetParameters().Count -eq 1 -and $_.GetParameters()[0].ParameterType.Name -eq 'IAsyncOperation`1'
+  } | Select-Object -First 1
+  function Await($operation, [Type]$type) {
+    $task = $asTask.MakeGenericMethod($type).Invoke($null, @($operation))
+    $task.GetAwaiter().GetResult()
+  }
+
+  $engines = @()
+  foreach ($tag in @('zh-Hans', 'en-US')) {
+    $language = New-Object Windows.Globalization.Language($tag)
+    $engine = [Windows.Media.Ocr.OcrEngine]::TryCreateFromLanguage($language)
+    if ($null -ne $engine) { $engines += $engine }
+  }
+  if ($engines.Count -eq 0) {
+    $engine = [Windows.Media.Ocr.OcrEngine]::TryCreateFromUserProfileLanguages()
+    if ($null -ne $engine) { $engines += $engine }
+  }
+  if ($engines.Count -eq 0) {
+    @{ ok=$false; code='language'; lines=@() } | ConvertTo-Json -Depth 8 -Compress
+    exit 0
+  }
+
+  $file = Await ([Windows.Storage.StorageFile]::GetFileFromPathAsync($ImagePath)) ([Windows.Storage.StorageFile])
+  $stream = Await ($file.OpenAsync([Windows.Storage.FileAccessMode]::Read)) ([Windows.Storage.Streams.IRandomAccessStream])
+  try {
+    $decoder = Await ([Windows.Graphics.Imaging.BitmapDecoder]::CreateAsync($stream)) ([Windows.Graphics.Imaging.BitmapDecoder])
+    $transform = New-Object Windows.Graphics.Imaging.BitmapTransform
+    $max = [Windows.Media.Ocr.OcrEngine]::MaxImageDimension
+    $scale = [Math]::Min(1.0, $max / [double][Math]::Max($decoder.PixelWidth, $decoder.PixelHeight))
+    $transform.ScaledWidth = [uint32][Math]::Max(1, [Math]::Floor($decoder.PixelWidth * $scale))
+    $transform.ScaledHeight = [uint32][Math]::Max(1, [Math]::Floor($decoder.PixelHeight * $scale))
+    $bitmap = Await ($decoder.GetSoftwareBitmapAsync(
+      [Windows.Graphics.Imaging.BitmapPixelFormat]::Bgra8,
+      [Windows.Graphics.Imaging.BitmapAlphaMode]::Premultiplied, $transform,
+      [Windows.Graphics.Imaging.ExifOrientationMode]::IgnoreExifOrientation,
+      [Windows.Graphics.Imaging.ColorManagementMode]::DoNotColorManage)) ([Windows.Graphics.Imaging.SoftwareBitmap])
+    try {
+      $lines = @()
+      # Chinese recognizers also handle Latin slide labels. Use the first
+      # available engine to avoid overlaying two different transcriptions.
+      $engine = $engines[0]
+      $result = Await ($engine.RecognizeAsync($bitmap)) ([Windows.Media.Ocr.OcrResult])
+      foreach ($line in $result.Lines) {
+        $minX = [double]::MaxValue; $minY = [double]::MaxValue
+        $maxX = 0.0; $maxY = 0.0
+        foreach ($word in $line.Words) {
+          $rect = $word.BoundingRect
+          $minX = [Math]::Min($minX, $rect.X); $minY = [Math]::Min($minY, $rect.Y)
+          $maxX = [Math]::Max($maxX, $rect.X + $rect.Width)
+          $maxY = [Math]::Max($maxY, $rect.Y + $rect.Height)
+        }
+        if ($maxX -gt $minX -and $maxY -gt $minY) {
+          $lines += @{ text=$line.Text; x=$minX / $bitmap.PixelWidth; y=$minY / $bitmap.PixelHeight;
+            w=($maxX - $minX) / $bitmap.PixelWidth; h=($maxY - $minY) / $bitmap.PixelHeight; confidence=1 }
+        }
+      }
+      @{ ok=$true; width=$bitmap.PixelWidth; height=$bitmap.PixelHeight;
+        language=$engine.RecognizerLanguage.LanguageTag; lines=@($lines) } | ConvertTo-Json -Depth 8 -Compress
+    } finally { if ($null -ne $bitmap) { $bitmap.Dispose() } }
+  } finally { if ($null -ne $stream) { $stream.Dispose() } }
+} catch {
+  @{ ok=$false; code='unavailable'; lines=@() } | ConvertTo-Json -Depth 8 -Compress
+}

package/src/host/office.js

@@ -114,7 +114,9 @@ function runTool(args, payload, timeoutMs) {
         const message = json && typeof json.error === 'string' && json.error
           ? json.error
           : '文档处理失败';
-        finish(new Error(message));
+        const failure = new Error(message);
+        if (typeof json?.code === 'string' && /^WORD_LAYOUT_/.test(json.code)) failure.code = json.code;
+        finish(failure);
         return;
       }
       finish(null);