@luii/node-tesseract-ocr 2.3.2 → 2.4.0

package/dist/cjs/types.d.ts

@@ -735,6 +735,11 @@ export interface TesseractBeginProcessPagesOptions {
     timeout: number;
     textonly: boolean;
 }
+export interface TesseractAddProcessPageOptions {
+    buffer: Buffer<ArrayBuffer>;
+    filename?: string;
+    progressCallback?: (info: ProgressChangedInfo) => void;
+}
 export interface TesseractProcessPagesStatus {
     active: boolean;
     healthy: boolean;
@@ -844,12 +849,15 @@ export interface TesseractDocumentApi {
     /**
      * Adds one encoded page to the active multipage session.
      * @throws {TesseractRuntimeError} If called before `init(...)`.
-     * @throws {TesseractArgumentError} If `buffer` is not a non-empty Buffer.
-     * @throws {TesseractArgumentError} If `filename` is provided but is not a string.
+     * @param {TesseractAddProcessPageOptions} options Page options.
+     * @throws {TesseractArgumentError} If `options` is missing/invalid.
+     * @throws {TesseractArgumentError} If `options.buffer` is not a non-empty Buffer.
+     * @throws {TesseractArgumentError} If `options.filename` is provided but is not a string.
+     * @throws {TesseractArgumentError} If `options.progressCallback` is provided but is not a function.
      * @throws {TesseractRuntimeError} If no session is active, decode fails, or page processing fails.
      * @throws {TesseractWorkerError} If the worker is closing/stopped.
      */
-    addPage(buffer: Buffer<ArrayBuffer>, filename?: string): Promise<void>;
+    addPage(options: TesseractAddProcessPageOptions): Promise<void>;
     /**
      * Finalizes the active multipage session and returns output PDF path.
      * @throws {TesseractRuntimeError} If called before `init(...)`.
@@ -996,15 +1004,18 @@ export interface TesseractInstance {
      */
     beginProcessPages(options: TesseractBeginProcessPagesOptions): Promise<void>;
     /**
-     * Adds one encoded page to the current multipage session.
+     * Adds one encoded page to the active multipage session.
      * @deprecated use `document.addPage()`
-     * @throws {TesseractArgumentError} If `buffer` is not a non-empty Buffer.
-     * @throws {TesseractArgumentError} If `filename` is provided but is not a string.
+     * @param {TesseractAddProcessPageOptions} options Page options.
+     * @throws {TesseractArgumentError} If `options` is missing/invalid.
+     * @throws {TesseractArgumentError} If `options.buffer` is not a non-empty Buffer.
+     * @throws {TesseractArgumentError} If `options.filename` is provided but is not a string.
+     * @throws {TesseractArgumentError} If `options.progressCallback` is provided but is not a function.
      * @throws {TesseractRuntimeError} If called before `init(...)`.
      * @throws {TesseractRuntimeError} If no session is active, decode fails, or page processing fails.
      * @throws {TesseractWorkerError} If the worker is closing/stopped.
      */
-    addProcessPage(buffer: Buffer<ArrayBuffer>, filename?: string): Promise<void>;
+    addProcessPage(options: TesseractAddProcessPageOptions): Promise<void>;
     /**
      * Finalizes the current multipage session and returns the output PDF path.
      * @deprecated use `document.finish()`

package/dist/esm/types.d.ts

@@ -735,6 +735,11 @@ export interface TesseractBeginProcessPagesOptions {
     timeout: number;
     textonly: boolean;
 }
+export interface TesseractAddProcessPageOptions {
+    buffer: Buffer<ArrayBuffer>;
+    filename?: string;
+    progressCallback?: (info: ProgressChangedInfo) => void;
+}
 export interface TesseractProcessPagesStatus {
     active: boolean;
     healthy: boolean;
@@ -844,12 +849,15 @@ export interface TesseractDocumentApi {
     /**
      * Adds one encoded page to the active multipage session.
      * @throws {TesseractRuntimeError} If called before `init(...)`.
-     * @throws {TesseractArgumentError} If `buffer` is not a non-empty Buffer.
-     * @throws {TesseractArgumentError} If `filename` is provided but is not a string.
+     * @param {TesseractAddProcessPageOptions} options Page options.
+     * @throws {TesseractArgumentError} If `options` is missing/invalid.
+     * @throws {TesseractArgumentError} If `options.buffer` is not a non-empty Buffer.
+     * @throws {TesseractArgumentError} If `options.filename` is provided but is not a string.
+     * @throws {TesseractArgumentError} If `options.progressCallback` is provided but is not a function.
      * @throws {TesseractRuntimeError} If no session is active, decode fails, or page processing fails.
      * @throws {TesseractWorkerError} If the worker is closing/stopped.
      */
-    addPage(buffer: Buffer<ArrayBuffer>, filename?: string): Promise<void>;
+    addPage(options: TesseractAddProcessPageOptions): Promise<void>;
     /**
      * Finalizes the active multipage session and returns output PDF path.
      * @throws {TesseractRuntimeError} If called before `init(...)`.
@@ -996,15 +1004,18 @@ export interface TesseractInstance {
      */
     beginProcessPages(options: TesseractBeginProcessPagesOptions): Promise<void>;
     /**
-     * Adds one encoded page to the current multipage session.
+     * Adds one encoded page to the active multipage session.
      * @deprecated use `document.addPage()`
-     * @throws {TesseractArgumentError} If `buffer` is not a non-empty Buffer.
-     * @throws {TesseractArgumentError} If `filename` is provided but is not a string.
+     * @param {TesseractAddProcessPageOptions} options Page options.
+     * @throws {TesseractArgumentError} If `options` is missing/invalid.
+     * @throws {TesseractArgumentError} If `options.buffer` is not a non-empty Buffer.
+     * @throws {TesseractArgumentError} If `options.filename` is provided but is not a string.
+     * @throws {TesseractArgumentError} If `options.progressCallback` is provided but is not a function.
      * @throws {TesseractRuntimeError} If called before `init(...)`.
      * @throws {TesseractRuntimeError} If no session is active, decode fails, or page processing fails.
      * @throws {TesseractWorkerError} If the worker is closing/stopped.
      */
-    addProcessPage(buffer: Buffer<ArrayBuffer>, filename?: string): Promise<void>;
+    addProcessPage(options: TesseractAddProcessPageOptions): Promise<void>;
     /**
      * Finalizes the current multipage session and returns the output PDF path.
      * @deprecated use `document.finish()`

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@luii/node-tesseract-ocr",
-  "version": "2.3.2",
+  "version": "2.4.0",
   "private": false,
   "binary": {
     "napi_versions": [

package/src/commands.hpp

@@ -458,6 +458,7 @@ struct CommandBeginProcessPages {
 struct CommandAddProcessPage {
   EncodedImageBuffer page;
   std::string filename;
+  std::shared_ptr<MonitorContext> monitor_context;
   Result invoke(tesseract::TessBaseAPI &api,
                 std::optional<ProcessPagesSession> &session,
                 const std::atomic<bool> &initialized) const {
@@ -523,18 +524,48 @@ struct CommandAddProcessPage {
 
     const char *effective_filename =
         filename.empty() ? nullptr : filename.c_str();
+    api.SetInputName(effective_filename);
+    api.SetImage(pix);
 
-    bool success = api.ProcessPage(
-        pix, session->next_page_index, effective_filename, nullptr,
-        session->timeout_millisec, session->renderer.get());
+    bool failed = false;
+    MonitorHandle handle{monitor_context};
+    auto *monitor = monitor_context ? &handle.monitor : nullptr;
+
+    if (session->timeout_millisec > 0) {
+      tesseract::ETEXT_DESC timeout_only_monitor{};
+      if (monitor != nullptr) {
+        monitor->set_deadline_msecs(session->timeout_millisec);
+      } else {
+        timeout_only_monitor.cancel = nullptr;
+        timeout_only_monitor.cancel_this = nullptr;
+        timeout_only_monitor.set_deadline_msecs(session->timeout_millisec);
+        monitor = &timeout_only_monitor;
+      }
+      failed = api.Recognize(monitor) < 0;
+    } else if (api.GetPageSegMode() == tesseract::PSM_OSD_ONLY ||
+               api.GetPageSegMode() == tesseract::PSM_AUTO_ONLY) {
+      tesseract::PageIterator *it = api.AnalyseLayout();
+      if (it == nullptr) {
+        failed = true;
+      } else {
+        delete it;
+      }
+    } else {
+      failed = api.Recognize(monitor) < 0;
+    }
+
+    if (session->renderer && !failed) {
+      failed = !session->renderer->AddImage(&api);
+    }
     pixDestroy(&pix);
 
-    if (!success) {
-      throw_runtime("addProcessPage: ProcessPage failed at page {}",
-                    session->next_page_index);
+    if (!failed) {
+      session->next_page_index++;
+      return ResultVoid{};
     }
 
-    session->next_page_index++;
+    throw_runtime("addProcessPage: ProcessPage failed at page {}",
+                  session->next_page_index);
     return ResultVoid{};
   }
 };

package/src/tesseract_wrapper.cpp

@@ -553,33 +553,55 @@ Napi::Value TesseractWrapper::AddProcessPage(const Napi::CallbackInfo &info) {
   Napi::Env env = info.Env();
   CommandAddProcessPage command{};
 
-  if (info.Length() < 1 || info.Length() > 2) {
-    return RejectTypeError(
-        env, "addProcessPage(buffer, filename?): expected 1 or 2 arguments",
-        "addProcessPage");
+  if (info.Length() != 1 || !info[0].IsObject()) {
+    return RejectTypeError(env,
+                           "addProcessPage(options): options must be an object",
+                           "addProcessPage");
   }
 
-  if (!info[0].IsBuffer()) {
+  Napi::Object options = info[0].As<Napi::Object>();
+
+  Napi::Value buffer_value = options.Get("buffer");
+  if (!buffer_value.IsBuffer()) {
     return RejectTypeError(
-        env, "addProcessPage(buffer, filename?): buffer must be a Buffer",
+        env, "addProcessPage(options): options.buffer must be a Buffer",
         "addProcessPage");
   }
 
-  Napi::Buffer<uint8_t> page_buffer = info[0].As<Napi::Buffer<uint8_t>>();
+  Napi::Buffer<uint8_t> page_buffer = buffer_value.As<Napi::Buffer<uint8_t>>();
   const size_t length = page_buffer.Length();
   if (length == 0) {
     return RejectTypeError(env,
-                           "addProcessPage(buffer, filename?): buffer is empty",
+                           "addProcessPage(options): options.buffer is empty",
                            "addProcessPage");
   }
 
-  if (HasArg(info, 1)) {
-    if (!info[1].IsString()) {
+  Napi::Value filename_value = options.Get("filename");
+  if (!filename_value.IsUndefined() && !filename_value.IsNull()) {
+    if (!filename_value.IsString()) {
       return RejectTypeError(
-          env, "addProcessPage(buffer, filename?): filename must be a string",
+          env, "addProcessPage(options): options.filename must be a string",
           "addProcessPage");
     }
-    command.filename = info[1].As<Napi::String>().Utf8Value();
+    command.filename = filename_value.As<Napi::String>().Utf8Value();
+  }
+
+  Napi::Value progress_callback_value = options.Get("progressCallback");
+  if (!progress_callback_value.IsUndefined() &&
+      !progress_callback_value.IsNull()) {
+    if (!progress_callback_value.IsFunction()) {
+      return RejectTypeError(env,
+                             "addProcessPage(options): "
+                             "options.progressCallback must be a function",
+                             "addProcessPage");
+    }
+
+    Napi::Function progress_callback =
+        progress_callback_value.As<Napi::Function>();
+    Napi::ThreadSafeFunction progress_tsfn = Napi::ThreadSafeFunction::New(
+        env, progress_callback, "tesseract_progress_callback", 0, 1);
+    command.monitor_context =
+        std::make_shared<MonitorContext>(std::move(progress_tsfn));
   }
 
   command.page.bytes.resize(length);