From a3a911171ce7836108e598082ccfbcb1653d31ae Mon Sep 17 00:00:00 2001
From: Eren Akbiyik <eren@moose.education>
Date: Mon, 21 Nov 2022 17:22:14 +0000
Subject: [PATCH] convert vector to use with C api

---
 examples/stream/stream.cpp |  3 ++-
 whisper.cpp                | 11 ++++++++---
 whisper.h                  |  5 ++++-
 3 files changed, 14 insertions(+), 5 deletions(-)

diff --git a/examples/stream/stream.cpp b/examples/stream/stream.cpp
index 2256e69..a5bb692 100644
--- a/examples/stream/stream.cpp
+++ b/examples/stream/stream.cpp
@@ -345,7 +345,8 @@ int main(int argc, char ** argv) {
             wparams.audio_ctx            = params.audio_ctx;
             wparams.speed_up             = params.speed_up;
 
-            wparams.prompt_tokens        = &prompt_tokens;
+            wparams.prompt_tokens        = &prompt_tokens[0];
+            wparams.prompt_n_tokens      = prompt_tokens.size();
             
             if (whisper_full(ctx, wparams, pcmf32.data(), pcmf32.size()) != 0) {
                 fprintf(stderr, "%s: failed to process audio\n", argv[0]);
diff --git a/whisper.cpp b/whisper.cpp
index 48bdb31..f8c1ada 100644
--- a/whisper.cpp
+++ b/whisper.cpp
@@ -2413,6 +2413,8 @@ struct whisper_full_params whisper_full_default_params(enum whisper_sampling_str
                     /*.audio_ctx            =*/ 0,
 
                     /*.prompt_tokens        =*/ nullptr,
+                    /*.prompt_n_tokens      =*/ 0,
+
                     /*.language             =*/ "en",
 
                     /*.greedy               =*/ {
@@ -2457,6 +2459,8 @@ struct whisper_full_params whisper_full_default_params(enum whisper_sampling_str
                     /*.audio_ctx            =*/ 0,
 
                     /*.prompt_tokens        =*/ nullptr,
+                    /*.prompt_n_tokens      =*/ 0,
+
                     /*.language             =*/ "en",
 
                     /*.greedy               =*/ {
@@ -2588,10 +2592,11 @@ int whisper_full(
 
     // Prepend the prompt tokens to the prompt_past
     if (params.prompt_tokens) {
-        for (int i = 0; i < (int) params.prompt_tokens->size(); i++) {
-            prompt_past.push_back((*params.prompt_tokens)[i]);
+        // Parse tokens from the pointer (it points to an std::vector)
+        for (int i = 0; i < params.prompt_n_tokens; i++) {
+            prompt_past.push_back(params.prompt_tokens[i]);
         }
-        std::rotate(prompt_past.begin(), prompt_past.end() - params.prompt_tokens->size(), prompt_past.end());
+        std::rotate(prompt_past.begin(), prompt_past.end() - params.prompt_n_tokens, prompt_past.end());
     }
 
     // overwrite audio_ctx
diff --git a/whisper.h b/whisper.h
index 917b459..ad989e9 100644
--- a/whisper.h
+++ b/whisper.h
@@ -209,7 +209,10 @@ extern "C" {
         bool speed_up;  // speed-up the audio by 2x using Phase Vocoder
         int  audio_ctx; // overwrite the audio context size (0 = use default)
 
-        const std::vector<whisper_token> * prompt_tokens;
+        // std::vector<whisper_token>: tokents to provide the whisper model as initial prompt
+        const whisper_token * prompt_tokens;
+        int prompt_n_tokens;
+
         const char * language;
 
         struct {