feat(ai-formatting): wire dictionary_terms through PostProcessOptions to LLM prompt suffix

crates/ai-formatting/src/pipeline.rs

@@ -9,6 +9,9 @@ pub struct PostProcessOptions {
     pub british_english: bool,
     pub anti_hallucination: bool,
     pub format_mode: FormatMode,
+    /// Custom vocabulary terms loaded from the user's dictionary. Injected
+    /// into the LLM cleanup prompt so the model knows how to spell them.
+    pub dictionary_terms: Vec<String>,
 }
 
 /// How aggressively to format the transcript text.
@@ -82,6 +85,19 @@ mod tests {
         ]
     }
 
+    #[test]
+    fn dictionary_terms_stored_on_options() {
+        let options = PostProcessOptions {
+            remove_fillers: false,
+            british_english: false,
+            anti_hallucination: false,
+            format_mode: FormatMode::Raw,
+            dictionary_terms: vec!["Wren".to_string(), "CORBEL".to_string()],
+        };
+        assert_eq!(options.dictionary_terms.len(), 2);
+        assert_eq!(options.dictionary_terms[0], "Wren");
+    }
+
     #[test]
     fn post_process_applies_all_filters() {
         let mut segments = make_segments();
@@ -90,6 +106,7 @@ mod tests {
             british_english: true,
             anti_hallucination: true,
             format_mode: FormatMode::Clean,
+            dictionary_terms: vec![],
         };
 
         post_process_segments(&mut segments, &options);
@@ -110,6 +127,7 @@ mod tests {
             british_english: false,
             anti_hallucination: false,
             format_mode: FormatMode::Smart,
+            dictionary_terms: vec![],
         };
 
         post_process_segments(&mut segments, &options);

src-tauri/src/commands/live.rs

@@ -166,11 +166,20 @@ pub async fn start_live_transcription_session(
     let worker_status = status_channel.clone();
     let worker_results = result_channel.clone();
 
+    let dictionary_terms: Vec<String> =
+        kon_storage::database::list_dictionary(&state.db)
+            .await
+            .unwrap_or_default()
+            .into_iter()
+            .map(|e| e.term)
+            .collect();
+
     let handle = tokio::task::spawn_blocking(move || {
         run_live_session(
             session_id,
             engine,
             config,
+            dictionary_terms,
             worker_results,
             worker_status,
             worker_stop,
@@ -250,6 +259,7 @@ fn run_live_session(
     session_id: u64,
     engine: Arc<LocalEngine>,
     config: StartLiveTranscriptionConfig,
+    dictionary_terms: Vec<String>,
     result_channel: Channel<LiveResultMessage>,
     status_channel: Channel<LiveStatusMessage>,
     stop_flag: Arc<AtomicBool>,
@@ -284,6 +294,7 @@ fn run_live_session(
             &mut inflight,
             session_id,
             &config,
+            &dictionary_terms,
             &result_channel,
             &status_channel,
         )? {}
@@ -386,6 +397,7 @@ fn run_live_session(
             &mut inflight,
             session_id,
             &config,
+            &dictionary_terms,
             &result_channel,
             &status_channel,
         )?;
@@ -512,6 +524,7 @@ fn poll_inference(
     inflight: &mut Option<InferenceTask>,
     session_id: u64,
     config: &StartLiveTranscriptionConfig,
+    dictionary_terms: &[String],
     result_channel: &Channel<LiveResultMessage>,
     status_channel: &Channel<LiveStatusMessage>,
 ) -> Result<Option<bool>, String> {
@@ -531,6 +544,7 @@ fn poll_inference(
                     british_english: config.british_english,
                     anti_hallucination: config.anti_hallucination,
                     format_mode: FormatMode::parse(&config.format_mode),
+                    dictionary_terms: dictionary_terms.to_vec(),
                 },
             );
             let segment_count = segments.len();

src-tauri/src/commands/transcription.rs

@@ -50,6 +50,13 @@ pub async fn transcribe_pcm(
     .await
     .map_err(|e| e.to_string())??;
 
+    let dictionary_terms: Vec<String> = kon_storage::database::list_dictionary(&state.db)
+        .await
+        .unwrap_or_default()
+        .into_iter()
+        .map(|e| e.term)
+        .collect();
+
     let mut segments: Vec<Segment> = timed.transcript.segments().to_vec();
     post_process_segments(
         &mut segments,
@@ -58,6 +65,7 @@ pub async fn transcribe_pcm(
             british_english,
             anti_hallucination,
             format_mode: FormatMode::parse(&format_mode),
+            dictionary_terms,
         },
     );
 
@@ -114,6 +122,13 @@ pub async fn transcribe_file(
     .await
     .map_err(|e| e.to_string())??;
 
+    let dictionary_terms: Vec<String> = kon_storage::database::list_dictionary(&state.db)
+        .await
+        .unwrap_or_default()
+        .into_iter()
+        .map(|e| e.term)
+        .collect();
+
     let mut segments: Vec<Segment> = timed.transcript.segments().to_vec();
     post_process_segments(
         &mut segments,
@@ -122,6 +137,7 @@ pub async fn transcribe_file(
             british_english,
             anti_hallucination,
             format_mode: FormatMode::parse(&format_mode),
+            dictionary_terms,
         },
     );
 
@@ -157,6 +173,13 @@ pub async fn transcribe_pcm_parakeet(
     .await
     .map_err(|e| e.to_string())??;
 
+    let dictionary_terms: Vec<String> = kon_storage::database::list_dictionary(&state.db)
+        .await
+        .unwrap_or_default()
+        .into_iter()
+        .map(|e| e.term)
+        .collect();
+
     let mut segments: Vec<Segment> = timed.transcript.segments().to_vec();
     post_process_segments(
         &mut segments,
@@ -165,6 +188,7 @@ pub async fn transcribe_pcm_parakeet(
             british_english,
             anti_hallucination,
             format_mode: FormatMode::parse(&format_mode),
+            dictionary_terms,
         },
     );