Accept Voice Memos drags and add batch file transcription

Fluid.xcodeproj/project.pbxproj

@@ -20,6 +20,8 @@
 		A62300000000000000000002 /* AudioBufferConverterTests.swift in Sources */ = {isa = PBXBuildFile; fileRef = A62300000000000000000001 /* AudioBufferConverterTests.swift */; };
 		C0DE63600000000000000002 /* AudioEngineRetirementDrainTests.swift in Sources */ = {isa = PBXBuildFile; fileRef = C0DE63600000000000000001 /* AudioEngineRetirementDrainTests.swift */; };
 		DA7100020000000000000002 /* DirectAudioReliabilityTests.swift in Sources */ = {isa = PBXBuildFile; fileRef = DA7100010000000000000001 /* DirectAudioReliabilityTests.swift */; };
+		A62300000000000000000022 /* BatchTranscriptionCoordinatorTests.swift in Sources */ = {isa = PBXBuildFile; fileRef = A62300000000000000000021 /* BatchTranscriptionCoordinatorTests.swift */; };
+		A62300000000000000000032 /* PromiseDropSupportTests.swift in Sources */ = {isa = PBXBuildFile; fileRef = A62300000000000000000031 /* PromiseDropSupportTests.swift */; };
 		7CDB0A2F2F3C4D5600FB7CAD /* dictation_fixture.wav in Resources */ = {isa = PBXBuildFile; fileRef = 7CDB0A2B2F3C4D5600FB7CAD /* dictation_fixture.wav */; };
 		7CDB0A302F3C4D5600FB7CAD /* XCTest.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 7CDB0A2C2F3C4D5600FB7CAD /* XCTest.framework */; };
 		7CE006BD2E80EBE600DDCCD6 /* AppUpdater in Frameworks */ = {isa = PBXBuildFile; productRef = 7CE006BC2E80EBE600DDCCD6 /* AppUpdater */; };
@@ -56,6 +58,8 @@
 		A62300000000000000000001 /* AudioBufferConverterTests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = AudioBufferConverterTests.swift; sourceTree = "<group>"; };
 		C0DE63600000000000000001 /* AudioEngineRetirementDrainTests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = AudioEngineRetirementDrainTests.swift; sourceTree = "<group>"; };
 		DA7100010000000000000001 /* DirectAudioReliabilityTests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = DirectAudioReliabilityTests.swift; sourceTree = "<group>"; };
+		A62300000000000000000021 /* BatchTranscriptionCoordinatorTests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = BatchTranscriptionCoordinatorTests.swift; sourceTree = "<group>"; };
+		A62300000000000000000031 /* PromiseDropSupportTests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = PromiseDropSupportTests.swift; sourceTree = "<group>"; };
 		7C078D8F2E3B339200FB7CAC /* FluidVoice Debug.app */ = {isa = PBXFileReference; explicitFileType = wrapper.application; includeInIndex = 0; path = "FluidVoice Debug.app"; sourceTree = BUILT_PRODUCTS_DIR; };
 		7C91B0022F42AA0100C0DEF0 /* HotkeyShortcutTests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = HotkeyShortcutTests.swift; sourceTree = "<group>"; };
 		7CDB0A202F3C4D5600FB7CAD /* FluidDictationIntegrationTests.xctest */ = {isa = PBXFileReference; explicitFileType = wrapper.cfbundle; includeInIndex = 0; path = FluidDictationIntegrationTests.xctest; sourceTree = BUILT_PRODUCTS_DIR; };
@@ -135,6 +139,8 @@
 				A62300000000000000000001 /* AudioBufferConverterTests.swift */,
 				C0DE63600000000000000001 /* AudioEngineRetirementDrainTests.swift */,
 				DA7100010000000000000001 /* DirectAudioReliabilityTests.swift */,
+				A62300000000000000000021 /* BatchTranscriptionCoordinatorTests.swift */,
+				A62300000000000000000031 /* PromiseDropSupportTests.swift */,
 			);
 			path = FluidDictationIntegrationTests;
 			sourceTree = "<group>";
@@ -295,6 +301,8 @@
 				A62300000000000000000002 /* AudioBufferConverterTests.swift in Sources */,
 				C0DE63600000000000000002 /* AudioEngineRetirementDrainTests.swift in Sources */,
 				DA7100020000000000000002 /* DirectAudioReliabilityTests.swift in Sources */,
+				A62300000000000000000022 /* BatchTranscriptionCoordinatorTests.swift in Sources */,
+				A62300000000000000000032 /* PromiseDropSupportTests.swift in Sources */,
 			);
 			runOnlyForDeploymentPostprocessing = 0;
 		};

Sources/Fluid/ContentView.swift

@@ -2012,6 +2012,8 @@ struct ContentView: View {
     // MARK: - Stop and Process Transcription
 
     private func stopAndProcessTranscription(route: DictationOutputRoute = .normal) async {
+        defer { FileTranscriptionSession.shared.endDictationIntent() } // released after the final transcription pass
+
         DebugLogger.shared.debug("stopAndProcessTranscription called", source: "ContentView")
         DebugLogger.shared.info("Output route selected: \(route.rawValue)", source: "ContentView")
         self.appBench("stop_path_enter route=\(route.rawValue)")
@@ -3074,6 +3076,12 @@ struct ContentView: View {
 
     /// Capture app context at start to avoid mismatches if the user switches apps mid-session
     private func startRecording() {
+        // avoid concurrent inference on the shared ASR model
+        guard !FileTranscriptionSession.isBatchTranscribing else {
+            self.notifyDictationBlockedByBatch()
+            return
+        }
+
         let model = SettingsStore.shared.selectedSpeechModel
         DebugLogger.shared.info(
             "ContentView: startRecording() for model=\(model.displayName), supportsStreaming=\(model.supportsStreaming)",
@@ -3084,6 +3092,8 @@ struct ContentView: View {
             return
         }
 
+        FileTranscriptionSession.shared.beginDictationIntent() // synchronous, before any await, so an ignored start can't latch it
+
         self.advanceOverlayLifecycle()
         self.setActiveRecordingMode(.dictate)
         let shouldShowDictationOverlay = !self.isRecordingForCommand
@@ -3119,6 +3129,7 @@ struct ContentView: View {
             })
             if startOutcome == .failed {
                 self.menuBarManager.hideRecordingOverlayImmediately(reason: "asr_start_failed")
+                FileTranscriptionSession.shared.endDictationIntent()
             }
         }
 
@@ -3344,15 +3355,19 @@ struct ContentView: View {
             },
             commandModeCallback: {
                 DebugLogger.shared.info("Command mode triggered", source: "ContentView")
-                self.captureRecordingContext()
-
-                // Set flag so stopAndProcessTranscription knows to process as command
-                self.setActiveRecordingMode(.command)
+                // Bypasses beginDictationRecording, so it needs its own batch guard — and before the mode
+                // flip, which would otherwise leave the overlay in command mode with nothing recording.
+                guard !FileTranscriptionSession.isBatchTranscribing else {
+                    self.notifyDictationBlockedByBatch()
+                    return
+                }
 
-                // Set overlay mode to command
+                self.captureRecordingContext()
+                self.setActiveRecordingMode(.command) // tells stopAndProcessTranscription to process as command
                 self.menuBarManager.setOverlayMode(.command)
 
-                guard !self.asr.isRunningOrStarting else { return }
+                guard !self.asr.isRunningOrStarting else { return } // mid-recording mode switch ends here: no restart
+                FileTranscriptionSession.shared.beginDictationIntent()
 
                 self.advanceOverlayLifecycle()
 
@@ -3370,12 +3385,19 @@ struct ContentView: View {
                         self.menuBarManager.hideRecordingOverlayImmediately(
                             reason: "command_asr_start_failed"
                         )
+                        FileTranscriptionSession.shared.endDictationIntent()
                     }
                 }
             },
             rewriteModeCallback: {
                 guard !self.showPrivateAIEditModeUnavailableIfNeeded() else { return }
 
+                // Ahead of the capture and mode flip, which would otherwise strand the overlay in edit mode.
+                guard !FileTranscriptionSession.isBatchTranscribing else {
+                    self.notifyDictationBlockedByBatch()
+                    return
+                }
+
                 self.captureRecordingContext()
 
                 // Try to capture text first while still in the other app
@@ -3401,7 +3423,8 @@ struct ContentView: View {
                 // Set flag so stopAndProcessTranscription knows to process as rewrite
                 self.setActiveRecordingMode(.edit)
 
-                guard !self.asr.isRunningOrStarting else { return }
+                guard !self.asr.isRunningOrStarting else { return } // mid-recording mode switch ends here: no restart
+                FileTranscriptionSession.shared.beginDictationIntent()
 
                 self.advanceOverlayLifecycle()
 
@@ -3416,6 +3439,7 @@ struct ContentView: View {
                         self.menuBarManager.hideRecordingOverlayImmediately(
                             reason: "edit_asr_start_failed"
                         )
+                        FileTranscriptionSession.shared.endDictationIntent()
                     }
                 }
             },
@@ -3727,6 +3751,10 @@ extension ContentView {
     }
 
     private func beginDictationRecording(for slot: SettingsStore.DictationShortcutSlot, mode: ActiveRecordingMode) {
+        guard !FileTranscriptionSession.isBatchTranscribing else {
+            self.notifyDictationBlockedByBatch()
+            return
+        }
         DebugLogger.shared.debug("Begin dictation recording for slot \(slot.rawValue)", source: "ContentView")
         self.appBench("begin_recording slot=\(slot.rawValue) mode=\(mode.rawValue)")
         if self.isOnboardingVoicePlaygroundStepActive {
@@ -3744,6 +3772,7 @@ extension ContentView {
             self.appBench("asr_start_skipped reason=already_running_or_starting")
             return
         }
+        FileTranscriptionSession.shared.beginDictationIntent() // claimed before the async start, since isRunning hasn't flipped yet
         self.advanceOverlayLifecycle()
         if self.asr.micStatus == .authorized {
             self.appBench("overlay_mode_request mode=Dictation")
@@ -3765,6 +3794,7 @@ extension ContentView {
             })
             if startOutcome == .failed {
                 self.menuBarManager.hideRecordingOverlayImmediately(reason: "asr_start_failed")
+                FileTranscriptionSession.shared.endDictationIntent()
             }
             DebugLogger.shared.benchmark(
                 "APP_BENCH",
@@ -3780,6 +3810,15 @@ extension ContentView {
         self.beginDictationRecording(for: .secondary, mode: mode)
     }
 
+    /// reuses the stop cue as feedback when blocked by a batch
+    private func notifyDictationBlockedByBatch() {
+        DebugLogger.shared.warning(
+            "Dictation blocked: batch file transcription in progress",
+            source: "ContentView"
+        )
+        TranscriptionSoundPlayer.shared.playStopSound()
+    }
+
     private func appBench(_ message: String) {
         DebugLogger.shared.benchmark("APP_BENCH", message: message, source: "AppBenchmark")
     }

Sources/Fluid/Fluid-Bridging-Header.h

@@ -2,5 +2,6 @@
 #define FLUID_BRIDGING_HEADER_H
 
 #include "CoreAudioCaptureSupportBridge.h"
+#include "ObjCExceptionCatcher.h"
 
 #endif

Sources/Fluid/ObjCExceptionCatcher.h

@@ -0,0 +1,15 @@
+#ifndef OBJC_EXCEPTION_CATCHER_H
+#define OBJC_EXCEPTION_CATCHER_H
+
+#import <Foundation/Foundation.h>
+
+NS_ASSUME_NONNULL_BEGIN
+
+/// Runs the block, returning a raised exception's description or nil if it completed.
+/// Pasteboard and file-promise calls raise, and a C++ terminate handler in-process
+/// turns any uncaught NSException into abort().
+NSString *_Nullable FluidCatchObjCException(void (NS_NOESCAPE ^block)(void));
+
+NS_ASSUME_NONNULL_END
+
+#endif

Sources/Fluid/ObjCExceptionCatcher.m

@@ -0,0 +1,10 @@
+#import "ObjCExceptionCatcher.h"
+
+NSString *FluidCatchObjCException(void (NS_NOESCAPE ^block)(void)) {
+    @try {
+        block();
+        return nil;
+    } @catch (NSException *exception) {
+        return exception.description ?: @"unknown Objective-C exception";
+    }
+}

Sources/Fluid/Services/ASRService.swift

@@ -2161,6 +2161,9 @@ final class ASRService: ObservableObject {
     }
 
     func stopWithoutTranscription() async {
+        // Cancel paths skip stop-and-process; above the guard to cover the starting phase.
+        defer { FileTranscriptionSession.endDictationIntentIfCreated() }
+
         if self.isStarting, self.isRunning == false {
             await self.cancelPendingAudioCaptureStart(reason: "stop_without_transcription")
         }