@gscdump/engine 0.28.3 → 0.30.0

package/dist/sql-fragments.d.mts

@@ -18,4 +18,27 @@ declare const METRIC_EXPR: Record<Metric, string>;
  * on the resolved column expression so drizzle can pass a column ref.
  */
 declare function topLevelPagePredicateSql(pathExpr: string): string;
-export { METRIC_EXPR, escapeLike, topLevelPagePredicateSql };
+/**
+ * How a canonicalized date column is emitted by {@link dateReplaceClause}:
+ * - `'date'` keeps a real `DATE` value (`CAST(col AS DATE)`). Right for views
+ *   and `.duckdb` exports the app re-queries, where the column type matters.
+ * - `'string'` emits an ISO `YYYY-MM-DD` string (`strftime(CAST(col AS DATE)…)`).
+ *   Right for row materialisation to JSON/CSV/NDJSON, where a `DATE` would
+ *   serialize as an opaque object / epoch.
+ */
+type DateCanonicalForm = 'date' | 'string';
+/**
+ * Build a `read_parquet` `REPLACE (…)` clause that canonicalizes legacy `date`
+ * columns. `date` lands as VARCHAR in older parquets (BYTE_ARRAY/UTF8, written
+ * before the schema enforced DATE); DuckDB infers the column type from the file,
+ * so without this every read path would expose VARCHAR despite SCHEMAS declaring
+ * DATE. The `CAST(col AS DATE)` is a no-op for already-DATE columns and
+ * vectorized parsing for VARCHAR ones, so output stays canonical either way.
+ *
+ * Pure: the caller passes the table's DATE column names (derived from `SCHEMAS`)
+ * so this fragment carries no schema/drizzle dependency. Returns `''` when the
+ * table has no DATE columns, so callers can interpolate it unconditionally:
+ *   `SELECT * ${dateReplaceClause(cols)} FROM read_parquet(…)`.
+ */
+declare function dateReplaceClause(dateColumns: readonly string[], form?: DateCanonicalForm): string;
+export { DateCanonicalForm, METRIC_EXPR, dateReplaceClause, escapeLike, topLevelPagePredicateSql };

package/dist/sql-fragments.mjs

@@ -10,4 +10,9 @@ const METRIC_EXPR = {
 function topLevelPagePredicateSql(pathExpr) {
 	return `LENGTH(${pathExpr}) - LENGTH(REPLACE(${pathExpr}, '/', '')) <= 1`;
 }
-export { METRIC_EXPR, escapeLike, topLevelPagePredicateSql };
+function dateReplaceClause(dateColumns, form = "string") {
+	if (dateColumns.length === 0) return "";
+	const cast = (n) => form === "date" ? `CAST(${n} AS DATE) AS ${n}` : `strftime(CAST(${n} AS DATE), '%Y-%m-%d') AS ${n}`;
+	return `REPLACE (${dateColumns.map(cast).join(", ")})`;
+}
+export { METRIC_EXPR, dateReplaceClause, escapeLike, topLevelPagePredicateSql };

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@gscdump/engine",
   "type": "module",
-  "version": "0.28.3",
+  "version": "0.30.0",
   "description": "Append-only Parquet/DuckDB storage engine + planner + adapters for the gscdump pipeline. Node + edge runtimes; opt-in heavy peers.",
   "author": {
     "name": "Harlan Wilton",
@@ -172,8 +172,8 @@
   },
   "peerDependencies": {
     "@duckdb/duckdb-wasm": "^1.32.0",
-    "hyparquet": "^1.26.0",
-    "hyparquet-writer": "^0.15.6"
+    "hyparquet": "^1.26.1",
+    "hyparquet-writer": "^0.16.1"
   },
   "peerDependenciesMeta": {
     "@duckdb/duckdb-wasm": {
@@ -188,11 +188,11 @@
   },
   "dependencies": {
     "drizzle-orm": "1.0.0-rc.3",
-    "hyparquet": "^1.26.0",
-    "hyparquet-writer": "^0.15.6",
+    "hyparquet": "^1.26.1",
+    "hyparquet-writer": "^0.16.1",
     "proper-lockfile": "^4.1.2",
-    "@gscdump/contracts": "0.28.3",
-    "gscdump": "0.28.3"
+    "@gscdump/contracts": "0.30.0",
+    "gscdump": "0.30.0"
   },
   "devDependencies": {
     "@duckdb/duckdb-wasm": "^1.32.0",