Support updateTranscribeAudio.

td/telegram/UpdatesManager.cpp

@@ -53,6 +53,7 @@
 #include "td/telegram/TdDb.h"
 #include "td/telegram/telegram_api.hpp"
 #include "td/telegram/ThemeManager.h"
+#include "td/telegram/VoiceNotesManager.h"
 #include "td/telegram/WebPagesManager.h"
 
 #include "td/actor/MultiPromise.h"
@@ -3431,9 +3432,12 @@ void UpdatesManager::on_update(tl_object_ptr<telegram_api::updateSavedRingtones>
   td_->notification_settings_manager_->reload_saved_ringtones(std::move(promise));
 }
 
-// unsupported updates
-
 void UpdatesManager::on_update(tl_object_ptr<telegram_api::updateTranscribedAudio> update, Promise<Unit> &&promise) {
+  td_->voice_notes_manager_->on_update_transcribed_audio(std::move(update->text_), update->transcription_id_,
+                                                         !update->pending_);
+  promise.set_value(Unit());
 }
 
+// unsupported updates
+
 }  // namespace td

td/telegram/UpdatesManager.h

@@ -526,9 +526,9 @@ class UpdatesManager final : public Actor {
 
   void on_update(tl_object_ptr<telegram_api::updateSavedRingtones> update, Promise<Unit> &&promise);
 
-  // unsupported updates
-
   void on_update(tl_object_ptr<telegram_api::updateTranscribedAudio> update, Promise<Unit> &&promise);
+
+  // unsupported updates
 };
 
 }  // namespace td

td/telegram/VoiceNotesManager.cpp

@@ -241,12 +241,13 @@ void VoiceNotesManager::on_voice_note_transcribed(FileId file_id, string &&text,
   auto voice_note = get_voice_note(file_id);
   CHECK(voice_note != nullptr);
   CHECK(!voice_note->is_transcribed);
-  CHECK(voice_note->transcription_id == 0);
+  CHECK(voice_note->transcription_id == 0 || voice_note->transcription_id == transcription_id);
+  bool is_changed = voice_note->is_transcribed != is_final || voice_note->text != text;
   voice_note->transcription_id = transcription_id;
   voice_note->is_transcribed = is_final;
   voice_note->text = std::move(text);
 
-  if (!voice_note->text.empty() || is_final) {
+  if (is_changed) {
     on_voice_note_transcription_updated(file_id);
   }
 
@@ -258,6 +259,13 @@ void VoiceNotesManager::on_voice_note_transcribed(FileId file_id, string &&text,
     speech_recognition_queries_.erase(it);
 
     set_promises(promises);
+  } else {
+    if (pending_voice_note_transcription_queries_.count(transcription_id) != 0) {
+      on_pending_voice_note_transcription_failed(transcription_id,
+                                                 Status::Error(500, "Receive duplicate recognition identifier"));
+    }
+    bool is_inserted = pending_voice_note_transcription_queries_.emplace(transcription_id, file_id).second;
+    CHECK(is_inserted);
   }
 }
 
@@ -265,7 +273,15 @@ void VoiceNotesManager::on_voice_note_transcription_failed(FileId file_id, Statu
   auto voice_note = get_voice_note(file_id);
   CHECK(voice_note != nullptr);
   CHECK(!voice_note->is_transcribed);
-  CHECK(voice_note->transcription_id == 0);
+
+  if (voice_note->transcription_id != 0) {
+    CHECK(pending_voice_note_transcription_queries_.count(voice_note->transcription_id) == 0);
+    voice_note->transcription_id = 0;
+    if (!voice_note->text.empty()) {
+      voice_note->text.clear();
+      on_voice_note_transcription_updated(file_id);
+    }
+  }
 
   auto it = speech_recognition_queries_.find(file_id);
   CHECK(it != speech_recognition_queries_.end());
@@ -276,6 +292,27 @@ void VoiceNotesManager::on_voice_note_transcription_failed(FileId file_id, Statu
   fail_promises(promises, std::move(error));
 }
 
+void VoiceNotesManager::on_update_transcribed_audio(string &&text, int64 transcription_id, bool is_final) {
+  auto it = pending_voice_note_transcription_queries_.find(transcription_id);
+  if (it == pending_voice_note_transcription_queries_.end()) {
+    return;
+  }
+  auto file_id = it->second;
+  pending_voice_note_transcription_queries_.erase(it);
+
+  on_voice_note_transcribed(file_id, std::move(text), transcription_id, is_final);
+}
+
+void VoiceNotesManager::on_pending_voice_note_transcription_failed(int64 transcription_id, Status &&error) {
+  auto it = pending_voice_note_transcription_queries_.find(transcription_id);
+  if (it == pending_voice_note_transcription_queries_.end()) {
+    return;
+  }
+  auto file_id = it->second;
+  pending_voice_note_transcription_queries_.erase(it);
+  on_voice_note_transcription_failed(file_id, std::move(error));
+}
+
 void VoiceNotesManager::on_voice_note_transcription_updated(FileId file_id) {
   auto it = voice_note_messages_.find(file_id);
   if (it != voice_note_messages_.end()) {

td/telegram/VoiceNotesManager.h

@@ -40,6 +40,8 @@ class VoiceNotesManager {
 
   void recognize_speech(FullMessageId full_message_id, Promise<Unit> &&promise);
 
+  void on_update_transcribed_audio(string &&text, int64 transcription_id, bool is_final);
+
   void on_voice_note_transcribed(FileId file_id, string &&text, int64 transcription_id, bool is_final);
 
   void on_voice_note_transcription_failed(FileId file_id, Status &&error);
@@ -80,12 +82,15 @@ class VoiceNotesManager {
 
   FileId on_get_voice_note(unique_ptr<VoiceNote> new_voice_note, bool replace);
 
+  void on_pending_voice_note_transcription_failed(int64 transcription_id, Status &&error);
+
   void on_voice_note_transcription_updated(FileId file_id);
 
   Td *td_;
   FlatHashMap<FileId, unique_ptr<VoiceNote>, FileIdHash> voice_notes_;
 
   FlatHashMap<FileId, vector<Promise<Unit>>, FileIdHash> speech_recognition_queries_;
+  FlatHashMap<int64, FileId> pending_voice_note_transcription_queries_;
 
   FlatHashMap<FileId, FlatHashSet<FullMessageId, FullMessageIdHash>, FileIdHash> voice_note_messages_;
   FlatHashMap<FullMessageId, FileId, FullMessageIdHash> message_voice_notes_;