examples : support progress_callback API for addon.node (#2941)

* feat: progress supported

* fix: missing params

* style: Format the code to improve readability

Unified code indentation ensures consistent coding style, enhancing code readability and maintainability.

* feat: support prompt api

---------

Co-authored-by: linxiaodong <calm.lin@wukongsch.com>

examples/addon.node/__test__/whisper.spec.js

@@ -18,6 +18,7 @@ const whisperParamsMock = {
   translate: true,
   no_timestamps: false,
   audio_ctx: 0,
+  max_len: 0,
 };
 
 describe("Run whisper.node", () => {

examples/addon.node/addon.cpp

@@ -128,7 +128,67 @@ void whisper_print_segment_callback(struct whisper_context * ctx, struct whisper
 
 void cb_log_disable(enum ggml_log_level, const char *, void *) {}
 
-int run(whisper_params &params, std::vector<std::vector<std::string>> &result) {
+class ProgressWorker : public Napi::AsyncWorker {
+ public:
+    ProgressWorker(Napi::Function& callback, whisper_params params, Napi::Function progress_callback, Napi::Env env)
+        : Napi::AsyncWorker(callback), params(params), env(env) {
+        // Create thread-safe function
+        if (!progress_callback.IsEmpty()) {
+            tsfn = Napi::ThreadSafeFunction::New(
+                env,
+                progress_callback,
+                "Progress Callback",
+                0,
+                1
+            );
+        }
+    }
+
+    ~ProgressWorker() {
+        if (tsfn) {
+            // Make sure to release the thread-safe function on destruction
+            tsfn.Release();
+        }
+    }
+
+    void Execute() override {
+        // Use custom run function with progress callback support
+        run_with_progress(params, result);
+    }
+
+    void OnOK() override {
+        Napi::HandleScope scope(Env());
+        Napi::Object res = Napi::Array::New(Env(), result.size());
+        for (uint64_t i = 0; i < result.size(); ++i) {
+            Napi::Object tmp = Napi::Array::New(Env(), 3);
+            for (uint64_t j = 0; j < 3; ++j) {
+                tmp[j] = Napi::String::New(Env(), result[i][j]);
+            }
+            res[i] = tmp;
+        }
+        Callback().Call({Env().Null(), res});
+    }
+
+    // Progress callback function - using thread-safe function
+    void OnProgress(int progress) {
+        if (tsfn) {
+            // Use thread-safe function to call JavaScript callback
+            auto callback = [progress](Napi::Env env, Napi::Function jsCallback) {
+                jsCallback.Call({Napi::Number::New(env, progress)});
+            };
+            
+            tsfn.BlockingCall(callback);
+        }
+    }
+
+ private:
+    whisper_params params;
+    std::vector<std::vector<std::string>> result;
+    Napi::Env env;
+    Napi::ThreadSafeFunction tsfn;
+
+    // Custom run function with progress callback support
+    int run_with_progress(whisper_params &params, std::vector<std::vector<std::string>> &result) {
         if (params.no_prints) {
             whisper_log_set(cb_log_disable, NULL);
         }
@@ -144,7 +204,6 @@ int run(whisper_params &params, std::vector<std::vector<std::string>> &result) {
         }
 
         // whisper init
-
         struct whisper_context_params cparams = whisper_context_default_params();
         cparams.use_gpu = params.use_gpu;
         cparams.flash_attn = params.flash_attn;
@@ -155,8 +214,7 @@ int run(whisper_params &params, std::vector<std::vector<std::string>> &result) {
             return 3;
         }
 
-    // if params.pcmf32 is provided, set params.fname_inp to "buffer"
+        // If params.pcmf32 provides, set params.fname_inp as "buffer"
-    // this is simpler than further modifications in the code
         if (!params.pcmf32.empty()) {
             fprintf(stderr, "info: using audio buffer as input\n");
             params.fname_inp.clear();
@@ -170,7 +228,7 @@ int run(whisper_params &params, std::vector<std::vector<std::string>> &result) {
             std::vector<float> pcmf32; // mono-channel F32 PCM
             std::vector<std::vector<float>> pcmf32s; // stereo-channel F32 PCM
 
-        // read the input audio file if params.pcmf32 is not provided
+            // If params.pcmf32 is empty, read input audio file
             if (params.pcmf32.empty()) {
                 if (!::read_audio_data(fname_inp, pcmf32, pcmf32s, params.diarize)) {
                     fprintf(stderr, "error: failed to read audio file '%s'\n", fname_inp.c_str());
@@ -180,14 +238,14 @@ int run(whisper_params &params, std::vector<std::vector<std::string>> &result) {
                 pcmf32 = params.pcmf32;
             }
 
-        // print system information
+            // Print system info
             if (!params.no_prints) {
                 fprintf(stderr, "\n");
                 fprintf(stderr, "system_info: n_threads = %d / %d | %s\n",
                         params.n_threads*params.n_processors, std::thread::hardware_concurrency(), whisper_print_system_info());
             }
 
-        // print some info about the processing
+            // Print processing info
             if (!params.no_prints) {
                 fprintf(stderr, "\n");
                 if (!whisper_is_multilingual(ctx)) {
@@ -208,7 +266,7 @@ int run(whisper_params &params, std::vector<std::vector<std::string>> &result) {
                 fprintf(stderr, "\n");
             }
 
-        // run the inference
+            // Run inference
             {
                 whisper_full_params wparams = whisper_full_default_params(WHISPER_SAMPLING_GREEDY);
 
@@ -241,17 +299,22 @@ int run(whisper_params &params, std::vector<std::vector<std::string>> &result) {
 
                 whisper_print_user_data user_data = { &params, &pcmf32s };
 
-            // this callback is called on each new segment
+                // This callback is called for each new segment
                 if (!wparams.print_realtime) {
                     wparams.new_segment_callback           = whisper_print_segment_callback;
                     wparams.new_segment_callback_user_data = &user_data;
                 }
 
-            // example for abort mechanism
+                // Set progress callback
-            // in this example, we do not abort the processing, but we could if the flag is set to true
+                wparams.progress_callback = [](struct whisper_context * /*ctx*/, struct whisper_state * /*state*/, int progress, void * user_data) {
-            // the callback is called before every encoder run - if it returns false, the processing is aborted
+                    ProgressWorker* worker = static_cast<ProgressWorker*>(user_data);
+                    worker->OnProgress(progress);
+                };
+                wparams.progress_callback_user_data = this;
+
+                // Abort mechanism example
                 {
-                static bool is_aborted = false; // NOTE: this should be atomic to avoid data race
+                    static bool is_aborted = false; // Note: this should be atomic to avoid data races
 
                     wparams.encoder_begin_callback = [](struct whisper_context * /*ctx*/, struct whisper_state * /*state*/, void * user_data) {
                         bool is_aborted = *(bool*)user_data;
@@ -284,36 +347,8 @@ int run(whisper_params &params, std::vector<std::vector<std::string>> &result) {
 
         return 0;
     }
-
-class Worker : public Napi::AsyncWorker {
- public:
-  Worker(Napi::Function& callback, whisper_params params)
-      : Napi::AsyncWorker(callback), params(params) {}
-
-  void Execute() override {
-    run(params, result);
-  }
-
-  void OnOK() override {
-    Napi::HandleScope scope(Env());
-    Napi::Object res = Napi::Array::New(Env(), result.size());
-    for (uint64_t i = 0; i < result.size(); ++i) {
-      Napi::Object tmp = Napi::Array::New(Env(), 3);
-      for (uint64_t j = 0; j < 3; ++j) {
-        tmp[j] = Napi::String::New(Env(), result[i][j]);
-      }
-      res[i] = tmp;
-    }
-    Callback().Call({Env().Null(), res});
-  }
-
- private:
-  whisper_params params;
-  std::vector<std::vector<std::string>> result;
 };
 
-
-
 Napi::Value whisper(const Napi::CallbackInfo& info) {
   Napi::Env env = info.Env();
   if (info.Length() <= 0 || !info[0].IsObject()) {
@@ -333,6 +368,23 @@ Napi::Value whisper(const Napi::CallbackInfo& info) {
   bool comma_in_time = whisper_params.Get("comma_in_time").As<Napi::Boolean>();
   int32_t max_len = whisper_params.Get("max_len").As<Napi::Number>();
   
+  // support prompt
+  std::string prompt = "";
+  if (whisper_params.Has("prompt") && whisper_params.Get("prompt").IsString()) {
+    prompt = whisper_params.Get("prompt").As<Napi::String>();
+  }
+  
+  // Add support for print_progress
+  bool print_progress = false;
+  if (whisper_params.Has("print_progress")) {
+    print_progress = whisper_params.Get("print_progress").As<Napi::Boolean>();
+  }
+  // Add support for progress_callback
+  Napi::Function progress_callback;
+  if (whisper_params.Has("progress_callback") && whisper_params.Get("progress_callback").IsFunction()) {
+    progress_callback = whisper_params.Get("progress_callback").As<Napi::Function>();
+  }
+
   Napi::Value pcmf32Value = whisper_params.Get("pcmf32");
   std::vector<float> pcmf32_vec;
   if (pcmf32Value.IsTypedArray()) {
@@ -355,9 +407,12 @@ Napi::Value whisper(const Napi::CallbackInfo& info) {
   params.pcmf32 = pcmf32_vec;
   params.comma_in_time = comma_in_time;
   params.max_len = max_len;
+  params.print_progress = print_progress;
+  params.prompt = prompt;
 
   Napi::Function callback = info[1].As<Napi::Function>();
-  Worker* worker = new Worker(callback, params);
+  // Create a new Worker class with progress callback support
+  ProgressWorker* worker = new ProgressWorker(callback, params, progress_callback, env);
   worker->Queue();
   return env.Undefined();
 }

examples/addon.node/index.js

@@ -19,6 +19,9 @@ const whisperParams = {
   no_timestamps: false,
   audio_ctx: 0,
   max_len: 0,
+  progress_callback: (progress) => {
+      console.log(`progress: ${progress}%`);
+    }
 };
 
 const arguments = process.argv.slice(2);