pi-mega-compact 0.4.2 → 0.4.3

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-mega-compact",
-  "version": "0.4.2",
+  "version": "0.4.3",
   "description": "Layered, local, vector-backed context compressor for pi — supersede/collapse/cluster compaction with deduped inline recall.",
   "type": "module",
   "license": "BSD-2-Clause",

package/src/store/sprint10.test.ts

@@ -174,6 +174,38 @@ test("checkAllIntegrity covers every session", () => {
   assert.ok(reports.every((r: { ok: boolean }) => r.ok));
 });
 
+// --- schema migration (v0.4.2: original_token_estimate) -------------------
+
+test("Sprint 10 migration: pre-0.4.2 db gains original_token_estimate and repoStats works", () => {
+  const dir = join(baseTmp, `run-${counter++}`);
+  // Simulate a v0.4.1-era store: openStore() builds the full current schema,
+  // then drop the original_token_estimate column that 0.4.2 added. CREATE TABLE
+  // IF NOT EXISTS is a no-op on an existing table, which is exactly why the
+  // shipped v0.4.2 crashed at runtime on old repos ("no such column").
+  const db = openStore(dir);
+  db.exec("ALTER TABLE context_chunks DROP COLUMN original_token_estimate;");
+  closeStore(dir);
+
+  // Re-open through the real code path — must ALTER the column in, not crash.
+  const vs = new VectorStore({ dedupSim: 0.9, stateDir: dir });
+  vs.add({
+    sessionId: "sess_mig",
+    summary: "legacy region",
+    regionText: "some region text to compact",
+    tokenEstimate: 100,
+    originalTokenEstimate: 500,
+    timestamp: 1,
+  });
+  const repo = vs.repoStats();
+  // No "no such column" crash = migration succeeded; totals reflect the new col.
+  assert.equal(repo.checkpointCount, 1);
+  assert.equal(repo.originalTokens, 500, "originalTokens read from migrated column");
+  // Re-open a second time to prove idempotency (column already exists).
+  closeStore(dir);
+  const vs2 = new VectorStore({ dedupSim: 0.9, stateDir: dir });
+  assert.equal(vs2.repoStats().originalTokens, 500, "stable across re-open");
+});
+
 // --- cleanup ---------------------------------------------------------------
 
 test("Sprint 10 cleanup", () => {

package/src/store/sqlite.ts

@@ -190,6 +190,12 @@ function initSchema(db: Database.Database): void {
       tokenize='trigram'
     );
   `);
+  // Idempotent column migrations. `CREATE TABLE IF NOT EXISTS` is a no-op on a
+  // pre-existing table, so new columns added to context_chunks after a store was
+  // first created (e.g. original_token_estimate in v0.4.2) must be ALTERed in for
+  // databases created by an older version — otherwise repoStats()/upsert crash
+  // with "no such column" and the extension fails to load. Additive only.
+  ensureColumn(db, "context_chunks", "original_token_estimate", "INTEGER");
   const v = db.prepare("SELECT value FROM meta WHERE key='schema_version'").get() as
     | { value: string }
     | undefined;
@@ -198,6 +204,19 @@ function initSchema(db: Database.Database): void {
   }
 }
 
+/**
+ * Add `column` (with `decl`, e.g. "INTEGER") to `table` if it does not already
+ * exist. Idempotent: checks PRAGMA table_info first, so it is safe to run on
+ * every open. Table/column/decl are code-controlled constants (never user
+ * input), so the unavoidable identifier interpolation here does not violate
+ * PREVENT-002 (no external data reaches this SQL).
+ */
+function ensureColumn(db: Database.Database, table: string, column: string, decl: string): void {
+  const cols = db.prepare(`PRAGMA table_info(${table})`).all() as Array<{ name: string }>;
+  if (cols.some((c) => c.name === column)) return;
+  db.exec(`ALTER TABLE ${table} ADD COLUMN ${column} ${decl}`);
+}
+
 /** Read a string-valued meta key (or undefined). Used for cumulative counters. */
 export function getMeta(key: string, stateDir: string = getStateDir()): string | undefined {
   const db = openStore(stateDir);