@taloswork/ose-shared 0.1.10 → 0.1.12

package/skills/document/bin/ocr-worker.cjs

@@ -1,4 +1,5 @@
 #!/usr/bin/env node
+process.noDeprecation = true;
 var __getOwnPropNames = Object.getOwnPropertyNames;
 var __commonJS = (cb, mod) => function __require() {
   return mod || (0, cb[__getOwnPropNames(cb)[0]])((mod = { exports: {} }).exports, mod), mod.exports;

package/skills/document/bin/ocr.cjs

@@ -132,7 +132,7 @@ var require_runtime = __commonJS({
         return { __await: arg };
       };
       function AsyncIterator(generator, PromiseImpl) {
-        function invoke(method, arg, resolve, reject) {
+        function invoke(method, arg, resolve2, reject) {
           var record = tryCatch(generator[method], generator, arg);
           if (record.type === "throw") {
             reject(record.arg);
@@ -141,24 +141,24 @@ var require_runtime = __commonJS({
             var value = result.value;
             if (value && typeof value === "object" && hasOwn.call(value, "__await")) {
               return PromiseImpl.resolve(value.__await).then(function(value2) {
-                invoke("next", value2, resolve, reject);
+                invoke("next", value2, resolve2, reject);
               }, function(err) {
-                invoke("throw", err, resolve, reject);
+                invoke("throw", err, resolve2, reject);
               });
             }
             return PromiseImpl.resolve(value).then(function(unwrapped) {
               result.value = unwrapped;
-              resolve(result);
+              resolve2(result);
             }, function(error) {
-              return invoke("throw", error, resolve, reject);
+              return invoke("throw", error, resolve2, reject);
             });
           }
         }
         var previousPromise;
         function enqueue(method, arg) {
           function callInvokeWithMethodAndArg() {
-            return new PromiseImpl(function(resolve, reject) {
-              invoke(method, arg, resolve, reject);
+            return new PromiseImpl(function(resolve2, reject) {
+              invoke(method, arg, resolve2, reject);
             });
           }
           return previousPromise = // If enqueue has been called before, then we want to wait until
@@ -612,13 +612,13 @@ var require_createScheduler = __commonJS({
           }
         }
       };
-      const queue = (action, payload) => new Promise((resolve, reject) => {
+      const queue = (action, payload) => new Promise((resolve2, reject) => {
         const job = createJob({ action, payload });
         jobQueue.push(async (w) => {
           jobQueue.shift();
           runningWorkers[w.id] = job;
           try {
-            resolve(await w[action].apply(exports2, [...payload, job.id]));
+            resolve2(await w[action].apply(exports2, [...payload, job.id]));
           } catch (err) {
             reject(err);
           } finally {
@@ -6600,7 +6600,7 @@ var require_lib3 = __commonJS({
       let accum = [];
       let accumBytes = 0;
       let abort = false;
-      return new Body.Promise(function(resolve, reject) {
+      return new Body.Promise(function(resolve2, reject) {
         let resTimeout;
         if (_this4.timeout) {
           resTimeout = setTimeout(function() {
@@ -6634,7 +6634,7 @@ var require_lib3 = __commonJS({
           }
           clearTimeout(resTimeout);
           try {
-            resolve(Buffer.concat(accum, accumBytes));
+            resolve2(Buffer.concat(accum, accumBytes));
           } catch (err) {
             reject(new FetchError(`Could not create Buffer from response body for ${_this4.url}: ${err.message}`, "system", err));
           }
@@ -7309,7 +7309,7 @@ var require_lib3 = __commonJS({
         throw new Error("native promise missing, set fetch.Promise to your favorite alternative");
       }
       Body.Promise = fetch.Promise;
-      return new fetch.Promise(function(resolve, reject) {
+      return new fetch.Promise(function(resolve2, reject) {
         const request = new Request(url, opts);
         const options = getNodeRequestOptions(request);
         const send = (options.protocol === "https:" ? https : http).request;
@@ -7442,7 +7442,7 @@ var require_lib3 = __commonJS({
                   requestOpts.body = void 0;
                   requestOpts.headers.delete("content-length");
                 }
-                resolve(fetch(new Request(locationURL, requestOpts)));
+                resolve2(fetch(new Request(locationURL, requestOpts)));
                 finalize();
                 return;
             }
@@ -7463,7 +7463,7 @@ var require_lib3 = __commonJS({
           const codings = headers.get("Content-Encoding");
           if (!request.compress || request.method === "HEAD" || codings === null || res.statusCode === 204 || res.statusCode === 304) {
             response = new Response(body, response_options);
-            resolve(response);
+            resolve2(response);
             return;
           }
           const zlibOptions = {
@@ -7473,7 +7473,7 @@ var require_lib3 = __commonJS({
           if (codings == "gzip" || codings == "x-gzip") {
             body = body.pipe(zlib.createGunzip(zlibOptions));
             response = new Response(body, response_options);
-            resolve(response);
+            resolve2(response);
             return;
           }
           if (codings == "deflate" || codings == "x-deflate") {
@@ -7485,12 +7485,12 @@ var require_lib3 = __commonJS({
                 body = body.pipe(zlib.createInflateRaw());
               }
               response = new Response(body, response_options);
-              resolve(response);
+              resolve2(response);
             });
             raw.on("end", function() {
               if (!response) {
                 response = new Response(body, response_options);
-                resolve(response);
+                resolve2(response);
               }
             });
             return;
@@ -7498,11 +7498,11 @@ var require_lib3 = __commonJS({
           if (codings == "br" && typeof zlib.createBrotliDecompress === "function") {
             body = body.pipe(zlib.createBrotliDecompress());
             response = new Response(body, response_options);
-            resolve(response);
+            resolve2(response);
             return;
           }
           response = new Response(body, response_options);
-          resolve(response);
+          resolve2(response);
         });
         writeToStream(req, request);
       });
@@ -7662,8 +7662,8 @@ var require_createWorker = __commonJS({
       const lstmOnlyCore = [OEM.DEFAULT, OEM.LSTM_ONLY].includes(oem) && !options.legacyCore;
       let workerResReject;
       let workerResResolve;
-      const workerRes = new Promise((resolve, reject) => {
-        workerResResolve = resolve;
+      const workerRes = new Promise((resolve2, reject) => {
+        workerResResolve = resolve2;
         workerResReject = reject;
       });
       const workerError = (event) => {
@@ -7678,10 +7678,10 @@ var require_createWorker = __commonJS({
       const setReject = (promiseId, rej) => {
         rejects[promiseId] = rej;
       };
-      const startJob = ({ id: jobId, action, payload }) => new Promise((resolve, reject) => {
+      const startJob = ({ id: jobId, action, payload }) => new Promise((resolve2, reject) => {
         log(`[${id}]: Start ${jobId}, action=${action}`);
         const promiseId = `${action}-${jobId}`;
-        setResolve(promiseId, resolve);
+        setResolve(promiseId, resolve2);
         setReject(promiseId, reject);
         send(worker, {
           workerId: id,
@@ -8025,34 +8025,124 @@ var require_src = __commonJS({
 
 // apps/admin/data/skills.lib/document/bin-src/ocr.mjs
 var import_promises = require("node:fs/promises");
+var import_node_path3 = require("node:path");
+
+// packages/core/dist/document/ocr.js
+var import_node_fs = require("node:fs");
+var import_node_path2 = require("node:path");
+
+// packages/core/dist/fs/paths.js
+var import_node_os = require("node:os");
 var import_node_path = require("node:path");
+var TALOS_RESERVED_DIR_NAMES = {
+  /** `<profilesRoot>/.bootstrap/` — pre-login installation configuration. */
+  bootstrap: ".bootstrap",
+  /** `bundled/.tmp/` — install staging and vacated installs awaiting cleanup. */
+  installScratch: ".tmp",
+  /** `bundled/.quarantine/` — integrity-failed bundles kept for post-mortem. */
+  quarantine: ".quarantine",
+  /** `data/ose/<id>/sessions/.locks/` — cross-process session ownership locks. */
+  sessionLocks: ".locks"
+};
+var TALOS_BOOTSTRAP_DIR_NAME = TALOS_RESERVED_DIR_NAMES.bootstrap;
+function getTalosOsCacheDir() {
+  if (process.platform === "darwin")
+    return (0, import_node_path.join)((0, import_node_os.homedir)(), "Library", "Caches", "Talos");
+  if (process.platform === "win32") {
+    const local = process.env.LOCALAPPDATA?.trim();
+    return (0, import_node_path.join)(local ? local : (0, import_node_path.join)((0, import_node_os.homedir)(), "AppData", "Local"), "Talos", "Cache");
+  }
+  const xdg = process.env.XDG_CACHE_HOME?.trim();
+  return (0, import_node_path.join)(xdg ? xdg : (0, import_node_path.join)((0, import_node_os.homedir)(), ".cache"), "talos");
+}
 
 // packages/core/dist/document/ocr.js
-var LANG_RE = /^[a-z]{3}(\+[a-z]{3})*$/;
+var LANG_RE = /^[a-z]{3}(_[a-z]+)*(\+[a-z]{3}(_[a-z]+)*)*$/;
+function isValidOcrLang(lang) {
+  return LANG_RE.test(lang);
+}
+function defaultOcrCachePath() {
+  return (0, import_node_path2.join)(getTalosOsCacheDir(), "tessdata");
+}
+var CJK = "[\\u2e80-\\u2eff\\u3000-\\u303f\\u3040-\\u30ff\\u3100-\\u312f\\u31a0-\\u31ff\\u3200-\\u33ff\\u3400-\\u4dbf\\u4e00-\\u9fff\\uf900-\\ufaff\\ufe30-\\ufe4f\\uff00-\\uffef]";
+var CJK_GAP = new RegExp(`(${CJK})[ \\t]+(?=${CJK})`, "g");
+function tidyCjkSpacing(text) {
+  return text.replace(CJK_GAP, "$1");
+}
+var UNSPACED_LANG = /^(chi_|jpn)/;
+var oneLine = (e) => String(e instanceof Error ? e.message : e).replace(/^Error:\s*/, "").replace(/\s+/g, " ").trim();
 async function ocrImage(image, opts = {}) {
   if (!image || image.length === 0)
     throw new Error("empty image");
   const lang = opts.lang ?? "eng";
-  if (!LANG_RE.test(lang))
-    throw new Error(`invalid lang "${lang}" (expected e.g. "eng" or "eng+deu")`);
+  if (!isValidOcrLang(lang))
+    throw new Error(`invalid lang "${lang}" (expected e.g. "eng", "chi_sim" or "chi_sim+eng")`);
+  const langs = lang.split("+");
+  const cachePath = opts.cachePath || defaultOcrCachePath();
+  let langPath = opts.langPath;
+  let cacheMethod;
+  let notBundled = langs;
+  if (!langPath && opts.bundledLangPath) {
+    const bundledDir = opts.bundledLangPath;
+    const packOf = (l) => (0, import_node_path2.join)(bundledDir, `${l}.traineddata.gz`);
+    notBundled = langs.filter((l) => !(0, import_node_fs.existsSync)(packOf(l)));
+    if (notBundled.length === 0) {
+      langPath = bundledDir;
+      cacheMethod = "none";
+    } else {
+      try {
+        (0, import_node_fs.mkdirSync)(cachePath, { recursive: true });
+        for (const l of langs) {
+          const cached = (0, import_node_path2.join)(cachePath, `${l}.traineddata`);
+          if (!notBundled.includes(l) && !(0, import_node_fs.existsSync)(cached))
+            (0, import_node_fs.copyFileSync)(packOf(l), cached);
+        }
+      } catch {
+      }
+    }
+  }
+  if (cacheMethod !== "none") {
+    try {
+      (0, import_node_fs.mkdirSync)(cachePath, { recursive: true });
+    } catch {
+    }
+  }
   const mod = await Promise.resolve().then(() => __toESM(require_src(), 1));
-  const recognize = mod.recognize ?? mod.default?.recognize;
-  if (typeof recognize !== "function")
-    throw new Error("tesseract.js is unavailable (recognize not found)");
-  let result;
+  const createWorker = mod.createWorker ?? mod.default?.createWorker;
+  if (typeof createWorker !== "function")
+    throw new Error("tesseract.js is unavailable (createWorker not found)");
+  let fail;
+  const failed = new Promise((_, reject) => {
+    fail = reject;
+  });
+  failed.catch(() => void 0);
+  let worker;
   try {
-    result = await recognize(Buffer.from(image), lang, {
-      ...opts.langPath ? { langPath: opts.langPath } : {},
-      ...opts.cachePath ? { cachePath: opts.cachePath } : {},
-      ...opts.workerPath ? { workerPath: opts.workerPath } : {}
-    });
+    worker = await Promise.race([
+      createWorker(lang, 1, {
+        cachePath,
+        ...langPath ? { langPath } : {},
+        ...cacheMethod ? { cacheMethod } : {},
+        ...opts.workerPath ? { workerPath: opts.workerPath } : {},
+        errorHandler: (e) => fail(new Error(oneLine(e)))
+      }),
+      failed
+    ]);
+    if (langs.some((l) => UNSPACED_LANG.test(l)))
+      await worker.setParameters({ preserve_interword_spaces: "1" });
+    const result = await Promise.race([worker.recognize(Buffer.from(image)), failed]);
+    return { text: tidyCjkSpacing((result.data.text ?? "").trim()), confidence: result.data.confidence ?? 0 };
   } catch (e) {
-    throw new Error(`OCR failed (language data unavailable offline?): ${e.message}`);
+    const msg = oneLine(e);
+    const which = notBundled.length && !langPath ? ` for ${notBundled.join("+")} (not bundled; downloaded on first use, needs network)` : "";
+    throw new Error(worker ? `OCR failed: ${msg}` : `OCR failed: could not load language data${which}: ${msg}`);
+  } finally {
+    await worker?.terminate().catch(() => void 0);
   }
-  return { text: (result.data.text ?? "").trim(), confidence: result.data.confidence ?? 0 };
 }
 
 // apps/admin/data/skills.lib/document/bin-src/ocr.mjs
+process.noDeprecation = true;
 function die(msg, code = 1) {
   process.stderr.write(`ocr: ${msg}
 `);
@@ -8062,15 +8152,16 @@ var IMG = /* @__PURE__ */ new Set([".png", ".jpg", ".jpeg", ".bmp", ".pbm", ".we
 async function main() {
   const [file, lang] = process.argv.slice(2);
   if (!file) die("usage: ocr <image.png|jpg> [lang]", 64);
-  const ext = (0, import_node_path.extname)(file).toLowerCase();
+  const ext = (0, import_node_path3.extname)(file).toLowerCase();
   if (!IMG.has(ext)) die(`expected an image (${[...IMG].join("/")}); for PDFs render a page first, then OCR`, 64);
   const bytes = new Uint8Array(await (0, import_promises.readFile)(file).catch((e) => die(`cannot read ${file}: ${e.message}`)));
   const cachePath = process.env.TALOS_TESSDATA || void 0;
   const langPath = process.env.TALOS_TESSDATA_LANGPATH || void 0;
-  const workerPath = (0, import_node_path.join)(__dirname, "ocr-worker.cjs");
+  const workerPath = (0, import_node_path3.join)(__dirname, "ocr-worker.cjs");
+  const bundledLangPath = (0, import_node_path3.join)(__dirname, "..", "assets", "tessdata");
   let result;
   try {
-    result = await ocrImage(bytes, { lang: lang || "eng", cachePath, langPath, workerPath });
+    result = await ocrImage(bytes, { lang: lang || "eng", cachePath, langPath, bundledLangPath, workerPath });
   } catch (e) {
     die(e.message);
   }