feat(transcripts): migration v5 meta — starred, tags, template, language, segments persisted

crates/storage/src/database.rs

@@ -88,7 +88,7 @@ pub async fn insert_transcript(
 
 pub async fn get_transcript(pool: &SqlitePool, id: &str) -> Result<Option<TranscriptRow>> {
     let row = sqlx::query(
-        "SELECT id, text, source, title, audio_path, duration, engine, model_id, inference_ms, sample_rate, audio_channels, format_mode, remove_fillers, british_english, anti_hallucination, created_at FROM transcripts WHERE id = ?",
+        "SELECT id, text, source, title, audio_path, duration, engine, model_id, inference_ms, sample_rate, audio_channels, format_mode, remove_fillers, british_english, anti_hallucination, created_at, starred, manual_tags, template, language, segments_json FROM transcripts WHERE id = ?",
     )
     .bind(id)
     .fetch_optional(pool)
@@ -110,7 +110,7 @@ pub async fn list_transcripts_paged(
     offset: i64,
 ) -> Result<Vec<TranscriptRow>> {
     let rows = sqlx::query(
-        "SELECT id, text, source, title, audio_path, duration, engine, model_id, inference_ms, sample_rate, audio_channels, format_mode, remove_fillers, british_english, anti_hallucination, created_at FROM transcripts ORDER BY created_at DESC LIMIT ? OFFSET ?",
+        "SELECT id, text, source, title, audio_path, duration, engine, model_id, inference_ms, sample_rate, audio_channels, format_mode, remove_fillers, british_english, anti_hallucination, created_at, starred, manual_tags, template, language, segments_json FROM transcripts ORDER BY created_at DESC LIMIT ? OFFSET ?",
     )
     .bind(limit)
     .bind(offset)
@@ -179,6 +179,49 @@ pub async fn update_transcript(
     }
 }
 
+/// Patch-style update for the Task 2.5 transcripts_meta columns (starred,
+/// manual_tags, template, language, segments_json). Follows the same
+/// COALESCE pattern as `update_task`: each `Option::Some` overwrites,
+/// each `None` preserves the existing column value. Returns the refreshed
+/// row or errors if `id` does not exist after the UPDATE.
+///
+/// Keeping this separate from `update_transcript` (text / title) means the
+/// existing command surface stays stable — the viewer and history store
+/// keep calling `update_transcript` for content edits, and call this for
+/// UI-metadata persistence.
+pub async fn update_transcript_meta(
+    pool: &SqlitePool,
+    id: &str,
+    starred: Option<bool>,
+    manual_tags: Option<&str>,
+    template: Option<&str>,
+    language: Option<&str>,
+    segments_json: Option<&str>,
+) -> Result<TranscriptRow> {
+    sqlx::query(
+        "UPDATE transcripts SET \
+         starred = COALESCE(?, starred), \
+         manual_tags = COALESCE(?, manual_tags), \
+         template = COALESCE(?, template), \
+         language = COALESCE(?, language), \
+         segments_json = COALESCE(?, segments_json) \
+         WHERE id = ?",
+    )
+    .bind(starred.map(|b| b as i64))
+    .bind(manual_tags)
+    .bind(template)
+    .bind(language)
+    .bind(segments_json)
+    .bind(id)
+    .execute(pool)
+    .await
+    .map_err(|e| KonError::StorageError(format!("update_transcript_meta: {e}")))?;
+
+    get_transcript(pool, id)
+        .await?
+        .ok_or_else(|| KonError::StorageError(format!("update_transcript_meta: transcript {id} not found")))
+}
+
 pub async fn delete_transcript(pool: &SqlitePool, id: &str) -> Result<()> {
     sqlx::query("DELETE FROM transcripts WHERE id = ?")
         .bind(id)
@@ -198,7 +241,7 @@ pub async fn search_transcripts(
     limit: i64,
 ) -> Result<Vec<TranscriptRow>> {
     let rows = sqlx::query(
-        "SELECT t.id, t.text, t.source, t.title, t.audio_path, t.duration, t.engine, t.model_id, t.inference_ms, t.sample_rate, t.audio_channels, t.format_mode, t.remove_fillers, t.british_english, t.anti_hallucination, t.created_at \
+        "SELECT t.id, t.text, t.source, t.title, t.audio_path, t.duration, t.engine, t.model_id, t.inference_ms, t.sample_rate, t.audio_channels, t.format_mode, t.remove_fillers, t.british_english, t.anti_hallucination, t.created_at, t.starred, t.manual_tags, t.template, t.language, t.segments_json \
          FROM transcripts t \
          JOIN transcripts_fts fts ON fts.rowid = t.rowid \
          WHERE transcripts_fts MATCH ? \
@@ -512,6 +555,13 @@ pub struct TranscriptRow {
     pub british_english: bool,
     pub anti_hallucination: bool,
     pub created_at: String,
+    // Task 2.5 — transcripts_meta (migration v5). Persists the UI metadata
+    // that previously lived in the removed localStorage `kon_history` cache.
+    pub starred: bool,
+    pub manual_tags: String,
+    pub template: String,
+    pub language: String,
+    pub segments_json: String,
 }
 
 #[derive(Debug, Clone)]
@@ -547,6 +597,12 @@ fn transcript_row_from(r: &sqlx::sqlite::SqliteRow) -> TranscriptRow {
         british_english: r.get("british_english"),
         anti_hallucination: r.get("anti_hallucination"),
         created_at: r.get("created_at"),
+        // Task 2.5 (migration v5). INTEGER 0/1 → bool via `!= 0`.
+        starred: r.get::<i64, _>("starred") != 0,
+        manual_tags: r.get("manual_tags"),
+        template: r.get("template"),
+        language: r.get("language"),
+        segments_json: r.get("segments_json"),
     }
 }
 
@@ -743,6 +799,107 @@ mod tests {
         assert!(parent.done, "parent should auto-complete when all children done");
     }
 
+    #[tokio::test]
+    async fn update_transcript_meta_happy_path() {
+        // Task 2.5 — insert a transcript, update starred=true, read it back.
+        let pool = test_pool().await;
+        insert_transcript(
+            &pool,
+            &InsertTranscriptParams {
+                id: "tm1",
+                text: "body",
+                source: "microphone",
+                title: Some("Meta happy"),
+                audio_path: None,
+                duration: 0.0,
+                engine: None,
+                model_id: None,
+                inference_ms: None,
+                sample_rate: None,
+                audio_channels: None,
+                format_mode: None,
+                remove_fillers: false,
+                british_english: false,
+                anti_hallucination: false,
+            },
+        )
+        .await
+        .unwrap();
+
+        let row = update_transcript_meta(&pool, "tm1", Some(true), None, None, None, None)
+            .await
+            .unwrap();
+        assert!(row.starred, "starred must flip true after update");
+
+        let fetched = get_transcript(&pool, "tm1").await.unwrap().unwrap();
+        assert!(fetched.starred, "starred must persist via SELECT");
+        assert_eq!(fetched.manual_tags, "");
+        assert_eq!(fetched.template, "");
+        assert_eq!(fetched.language, "");
+        assert_eq!(fetched.segments_json, "");
+    }
+
+    #[tokio::test]
+    async fn update_transcript_meta_partial_leaves_others_unchanged() {
+        // Task 2.5 — partial update: only segments_json changes; the other
+        // meta columns must be preserved by COALESCE.
+        let pool = test_pool().await;
+        insert_transcript(
+            &pool,
+            &InsertTranscriptParams {
+                id: "tm2",
+                text: "body",
+                source: "microphone",
+                title: None,
+                audio_path: None,
+                duration: 0.0,
+                engine: None,
+                model_id: None,
+                inference_ms: None,
+                sample_rate: None,
+                audio_channels: None,
+                format_mode: None,
+                remove_fillers: false,
+                british_english: false,
+                anti_hallucination: false,
+            },
+        )
+        .await
+        .unwrap();
+
+        // Seed starred + tags + template + language first.
+        update_transcript_meta(
+            &pool,
+            "tm2",
+            Some(true),
+            Some("urgent,review"),
+            Some("Meeting"),
+            Some("en-GB"),
+            None,
+        )
+        .await
+        .unwrap();
+
+        // Now update only segments_json.
+        let row = update_transcript_meta(
+            &pool,
+            "tm2",
+            None,
+            None,
+            None,
+            None,
+            Some(r#"[{"start":0.0,"end":1.0,"text":"hi"}]"#),
+        )
+        .await
+        .unwrap();
+
+        assert!(row.starred, "starred must survive partial update");
+        assert_eq!(row.manual_tags, "urgent,review");
+        assert_eq!(row.template, "Meeting");
+        assert_eq!(row.language, "en-GB");
+        assert_eq!(row.segments_json, r#"[{"start":0.0,"end":1.0,"text":"hi"}]"#);
+    }
+
     #[tokio::test]
     async fn update_task_overwrites_provided_fields() {
         // Task 2.6 — happy path: insert, update bucket + effort, read back.

crates/storage/src/lib.rs

@@ -7,7 +7,8 @@ pub use database::{
     delete_dictionary_entry, delete_task, delete_transcript, get_setting, get_task_by_id, uncomplete_task,
     get_transcript, init, insert_subtask, insert_task, insert_transcript, list_dictionary,
     list_recent_errors, list_subtasks, list_tasks, list_transcripts, list_transcripts_paged,
-    log_error, search_transcripts, set_setting, update_task, update_transcript, DictionaryEntry,
+    log_error, search_transcripts, set_setting, update_task, update_transcript,
-    ErrorLogRow, InsertTranscriptParams, TaskRow, TranscriptRow,
+    update_transcript_meta, DictionaryEntry, ErrorLogRow, InsertTranscriptParams, TaskRow,
+    TranscriptRow,
 };
 pub use file_storage::{app_data_dir, crashes_dir, database_path, logs_dir, recordings_dir};

crates/storage/src/migrations.rs

@@ -115,6 +115,17 @@ const MIGRATIONS: &[(i64, &str, &str)] = &[
     (4, "tasks_meta: notes column for persisted free-text", r#"
         ALTER TABLE tasks ADD COLUMN notes TEXT NOT NULL DEFAULT ''
     "#),
+    (
+        5,
+        "transcripts_meta",
+        r#"
+        ALTER TABLE transcripts ADD COLUMN starred INTEGER NOT NULL DEFAULT 0;
+        ALTER TABLE transcripts ADD COLUMN manual_tags TEXT NOT NULL DEFAULT '';
+        ALTER TABLE transcripts ADD COLUMN template TEXT NOT NULL DEFAULT '';
+        ALTER TABLE transcripts ADD COLUMN language TEXT NOT NULL DEFAULT '';
+        ALTER TABLE transcripts ADD COLUMN segments_json TEXT NOT NULL DEFAULT '';
+        "#,
+    ),
 ];
 
 /// Split SQL into individual statements, respecting BEGIN...END trigger blocks.
@@ -215,7 +226,7 @@ mod tests {
             .fetch_one(&pool)
             .await
             .unwrap();
-        assert_eq!(count, 4);
+        assert_eq!(count, 5);
 
         sqlx::query("INSERT INTO settings (key, value) VALUES ('test', 'value')")
             .execute(&pool)
@@ -237,7 +248,7 @@ mod tests {
             .fetch_one(&pool)
             .await
             .unwrap();
-        assert_eq!(count, 4);
+        assert_eq!(count, 5);
     }
 
     #[tokio::test]
@@ -268,6 +279,34 @@ mod tests {
         }
     }
 
+    #[tokio::test]
+    async fn migration_transcripts_meta_adds_columns() {
+        // Task 2.5 — verify starred / manual_tags / template / language /
+        // segments_json are all present on the transcripts table after
+        // migrations run. All five are added by v5.
+        let pool = SqlitePoolOptions::new()
+            .max_connections(1)
+            .connect("sqlite::memory:")
+            .await
+            .expect("pool");
+        run_migrations(&pool).await.expect("migrate");
+
+        let info = sqlx::query("PRAGMA table_info(transcripts)")
+            .fetch_all(&pool)
+            .await
+            .unwrap();
+        let names: Vec<String> = info
+            .iter()
+            .map(|r| r.get::<String, _>("name"))
+            .collect();
+        for col in ["starred", "manual_tags", "template", "language", "segments_json"] {
+            assert!(
+                names.contains(&col.to_string()),
+                "transcripts must have {col}; got {names:?}"
+            );
+        }
+    }
+
     #[tokio::test]
     async fn test_parent_task_id_cascade_delete() {
         let pool = SqlitePoolOptions::new()

src-tauri/src/commands/transcripts.rs

@@ -13,7 +13,8 @@ use kon_storage::{
     delete_transcript as db_delete_transcript, get_transcript as db_get_transcript,
     insert_transcript as db_insert_transcript, list_dictionary,
     list_transcripts_paged, search_transcripts as db_search_transcripts,
-    update_transcript as db_update_transcript, DictionaryEntry, InsertTranscriptParams,
+    update_transcript as db_update_transcript,
+    update_transcript_meta as db_update_transcript_meta, DictionaryEntry, InsertTranscriptParams,
     TranscriptRow,
 };
 
@@ -22,6 +23,10 @@ use crate::AppState;
 /// Frontend-facing shape of a stored transcript. Mirrors `TranscriptRow`
 /// from the storage crate but drops fields the UI does not need yet
 /// (sample_rate, audio_channels, format flags) and renames to camelCase.
+///
+/// Task 2.5 — `starred`, `manualTags`, `template`, `language`, `segmentsJson`
+/// were added to back the viewer metadata that previously lived only in the
+/// removed `kon_history` localStorage cache.
 #[derive(Debug, Clone, Serialize)]
 #[serde(rename_all = "camelCase")]
 pub struct TranscriptDto {
@@ -34,6 +39,11 @@ pub struct TranscriptDto {
     pub engine: Option<String>,
     pub model_id: Option<String>,
     pub created_at: String,
+    pub starred: bool,
+    pub manual_tags: String,
+    pub template: String,
+    pub language: String,
+    pub segments_json: String,
 }
 
 impl From<TranscriptRow> for TranscriptDto {
@@ -48,6 +58,11 @@ impl From<TranscriptRow> for TranscriptDto {
             engine: r.engine,
             model_id: r.model_id,
             created_at: r.created_at,
+            starred: r.starred,
+            manual_tags: r.manual_tags,
+            template: r.template,
+            language: r.language,
+            segments_json: r.segments_json,
         }
     }
 }
@@ -179,6 +194,45 @@ pub async fn search_transcripts(
         .map_err(|e| e.to_string())
 }
 
+/// Task 2.5 — patch-style update for the transcripts_meta columns. Each
+/// field is optional: `Some` overwrites, omitted / `None` preserves via
+/// COALESCE server-side. Separate from `update_transcript` (text / title)
+/// so the existing command surface stays untouched.
+#[derive(Debug, Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub struct UpdateTranscriptMetaRequest {
+    #[serde(default)]
+    pub starred: Option<bool>,
+    #[serde(default)]
+    pub manual_tags: Option<String>,
+    #[serde(default)]
+    pub template: Option<String>,
+    #[serde(default)]
+    pub language: Option<String>,
+    #[serde(default)]
+    pub segments_json: Option<String>,
+}
+
+#[tauri::command]
+pub async fn update_transcript_meta_cmd(
+    state: tauri::State<'_, AppState>,
+    id: String,
+    patch: UpdateTranscriptMetaRequest,
+) -> Result<TranscriptDto, String> {
+    let row = db_update_transcript_meta(
+        &state.db,
+        &id,
+        patch.starred,
+        patch.manual_tags.as_deref(),
+        patch.template.as_deref(),
+        patch.language.as_deref(),
+        patch.segments_json.as_deref(),
+    )
+    .await
+    .map_err(|e| e.to_string())?;
+    Ok(TranscriptDto::from(row))
+}
+
 // --- Dictionary commands (custom vocabulary) ---
 
 #[derive(Debug, Clone, Serialize)]

src-tauri/src/lib.rs

@@ -271,6 +271,7 @@ pub fn run() {
             commands::transcripts::count_transcripts_command,
             commands::transcripts::get_transcript,
             commands::transcripts::update_transcript,
+            commands::transcripts::update_transcript_meta_cmd,
             commands::transcripts::delete_transcript,
             commands::transcripts::search_transcripts,
             commands::transcripts::list_dictionary_command,

src/lib/stores/page.svelte.js

@@ -94,10 +94,30 @@ import { toasts } from "$lib/stores/toasts.svelte.js";
 export const history = $state([]);
 
 /** Shape a TranscriptDto row from `list_transcripts` into the UI entry shape
- *  that HistoryPage expects. UI-only fields that were never persisted to
+ *  that HistoryPage expects.
- *  SQLite (segments, starred, manualTags, language, template) are absent. */
+ *
+ *  Task 2.5 — the meta columns (starred, manualTags, template, language,
+ *  segments_json) are now carried by SQLite. `manualTags` is stored comma-
+ *  joined in a single TEXT column (empty string = no tags); `segmentsJson`
+ *  is a serialised JSON array of segment objects (empty string = no
+ *  segments).
+ */
 function mapTranscriptRow(row) {
   const text = row.text ?? "";
+  // manual_tags is a single TEXT column. Empty string ⇒ empty array.
+  const rawTags = row.manualTags ?? "";
+  const manualTags = rawTags ? rawTags.split(",").filter(Boolean) : [];
+  // segments_json is a serialised array. Empty string ⇒ empty array.
+  const rawSegments = row.segmentsJson ?? "";
+  let segments = [];
+  if (rawSegments) {
+    try {
+      const parsed = JSON.parse(rawSegments);
+      if (Array.isArray(parsed)) segments = parsed;
+    } catch (err) {
+      console.warn("mapTranscriptRow: segmentsJson parse failed", err);
+    }
+  }
   return {
     id: row.id,
     text,
@@ -110,6 +130,11 @@ function mapTranscriptRow(row) {
     createdAt: row.createdAt ?? null,
     date: row.createdAt ? new Date(row.createdAt).toLocaleString("en-GB") : "",
     preview: text.slice(0, 120),
+    starred: !!row.starred,
+    manualTags,
+    template: row.template ?? "",
+    language: row.language ?? "",
+    segments,
   };
 }
 
@@ -138,9 +163,9 @@ if (typeof window !== "undefined" && hasTauriRuntime()) {
 // `saveHistory` is retained as a no-op for callers in HistoryPage that use
 // it as a mutation hook (tag add/remove, clear-all). SQLite writes happen
 // via the specific `add_transcript` / `update_transcript` /
-// `delete_transcript` commands in the functions below; ad-hoc UI-only
+// `update_transcript_meta_cmd` / `delete_transcript` commands in the
-// mutations (e.g. manualTags) are session-scoped because they are not
+// functions below. HistoryPage's manualTags mutations will need to call
-// backed by a SQLite column.
+// `saveTranscriptMeta` directly (Task 2.5) to persist past reload.
 export function saveHistory() {}
 
 /**
@@ -211,6 +236,47 @@ export async function renameHistoryEntry(id, updates) {
   }
 }
 
+/**
+ * Task 2.5 — persist viewer metadata to SQLite via update_transcript_meta_cmd.
+ * `patch` may contain any of:
+ *   - `starred: boolean`
+ *   - `manualTags: string[] | string`    (array joined on ',' server-side)
+ *   - `template: string`
+ *   - `language: string`
+ *   - `segments: Array<object>`          (JSON.stringify'd before the wire hop)
+ * Omitted fields are preserved by COALESCE server-side. The in-memory
+ * history row (if present) is refreshed from the returned canonical row
+ * so the main-window list stays consistent.
+ *
+ * Returns nothing — callers needing the new row should read `history`.
+ *
+ * @param {string} id
+ * @param {Record<string, any>} patch
+ */
+export async function saveTranscriptMeta(id, patch) {
+  if (!hasTauriRuntime()) return;
+  try {
+    const payload = {
+      starred: patch.starred,
+      manualTags: patch.manualTags == null
+        ? undefined
+        : (Array.isArray(patch.manualTags)
+            ? patch.manualTags.join(",")
+            : String(patch.manualTags)),
+      template: patch.template,
+      language: patch.language,
+      segmentsJson: patch.segments ? JSON.stringify(patch.segments) : undefined,
+    };
+    const row = await invoke("update_transcript_meta_cmd", { id: String(id), patch: payload });
+    const idx = history.findIndex((h) => String(h.id) === String(id));
+    if (idx !== -1) {
+      history[idx] = mapTranscriptRow(row);
+    }
+  } catch (err) {
+    toasts.error("Failed to save transcript metadata", String(err));
+  }
+}
+
 export function deleteFromHistory(index) {
   const entry = history[index];
   history.splice(index, 1);

src/routes/viewer/+page.svelte

@@ -4,6 +4,7 @@
   import { convertFileSrc, invoke } from "@tauri-apps/api/core";
   import { formatTime, formatDuration } from "$lib/utils/time.js";
   import { PLAYBACK_SPEEDS } from "$lib/utils/constants.js";
+  import { saveTranscriptMeta } from "$lib/stores/page.svelte.js";
 
   let item = $state(null);
   let audioEl = $state(null);
@@ -184,6 +185,10 @@
       // Update full text
       item.text = item.segments.map((s) => s.text).join(" ").trim();
       saveItemToHistory();
+      // Task 2.5 — persist segment boundaries (text + starred flags) to
+      // SQLite via segments_json. update_transcript above only covers text
+      // and title; the segment array needs the dedicated meta command.
+      persistSegments();
     }
     editingIdx = -1;
     editingText = "";
@@ -194,20 +199,35 @@
       item.segments.splice(idx, 1);
       item.text = item.segments.map((s) => s.text).join(" ").trim();
       saveItemToHistory();
+      // Task 2.5 — persist the shrunk segment array.
+      persistSegments();
       if (editingIdx === idx) { editingIdx = -1; }
     }
   }
 
   function toggleStar(idx) {
     if (item?.segments) {
-      // TODO: Phase 2 remaining Task 1.5 — starred is not a SQLite column
-      // on the transcripts table yet, so this mutation is in-memory only
-      // and will not survive a reload. Task 1.5 will extend update_transcript
-      // (and the transcripts schema) to carry starred segment state.
       item.segments[idx].starred = !item.segments[idx].starred;
+      // Task 2.5 — persist the updated segments array (incl. starred flags)
+      // via the new update_transcript_meta_cmd. Before this, starring a
+      // segment was in-memory only and vanished on reload.
+      persistSegments();
     }
   }
 
+  // Task 2.5 — single entry point for segments_json persistence. Callers
+  // already guard with `if (item?.segments)` so the snapshot here is safe;
+  // the @ts-ignore matches the file-wide pattern where `item` inferred as
+  // `null` from `$state(null)` narrows to `never` after guards. Replacing
+  // the annotation needs a full viewer re-typing pass — out of scope for
+  // Task 2.5, tracked by the file's 80-odd pre-existing TS errors.
+  function persistSegments() {
+    if (!item) return;
+    /** @type {any} */
+    const snap = item;
+    saveTranscriptMeta(snap.id, { segments: snap.segments });
+  }
+
   function copySegment(idx) {
     if (item?.segments?.[idx]) {
       const text = item.segments[idx].text.trim();
@@ -217,13 +237,11 @@
     }
   }
 
-  // Persist edits back to SQLite via update_transcript (the canonical store).
+  // Persist edits back to SQLite. `update_transcript` covers `text` / `title`;
-  // Only `text` and `title` are accepted by update_transcript today. Segment
+  // segment structure, per-segment starred flags, manualTags, template and
-  // structure, segment-level starred flags, and manualTags are not SQLite
+  // language go through `update_transcript_meta_cmd` (Task 2.5) via the
-  // columns yet — Phase 2 remaining Task 1.5 will extend the command and the
+  // store's `saveTranscriptMeta` helper — callers in this file wire those
-  // transcripts schema to cover those. Until then, segment-level edits that
+  // up alongside the plain-text update below.
-  // change `item.text` (the joined full transcript) survive reload; the
-  // segment array itself does not.
   //
   // The cross-window `kon_viewer_item` localStorage payload stays in sync so
   // a second viewer window hydrating from it reflects the latest edit.