From 8ef21681b54438f4fb74f2be79af91cba4000c5e Mon Sep 17 00:00:00 2001
From: gs-gunjan <Gunjan.Shah@ny.email.gs.com>
Date: Wed, 12 Feb 2025 15:22:28 +0530
Subject: [PATCH] datacube: fixing caching for csv

---
 .changeset/tidy-ads-judge.md                  |  5 ++
 .../src/stores/LegendDataCubeCacheManager.ts  | 67 +++++++++++++------
 2 files changed, 53 insertions(+), 19 deletions(-)
 create mode 100644 .changeset/tidy-ads-judge.md

diff --git a/.changeset/tidy-ads-judge.md b/.changeset/tidy-ads-judge.md
new file mode 100644
index 0000000000..cbb48a8777
--- /dev/null
+++ b/.changeset/tidy-ads-judge.md
@@ -0,0 +1,5 @@
+---
+'@finos/legend-application-data-cube': patch
+---
+
+Fixing Caching for CSV
diff --git a/packages/legend-application-data-cube/src/stores/LegendDataCubeCacheManager.ts b/packages/legend-application-data-cube/src/stores/LegendDataCubeCacheManager.ts
index 03a6dc08a2..95b172780f 100644
--- a/packages/legend-application-data-cube/src/stores/LegendDataCubeCacheManager.ts
+++ b/packages/legend-application-data-cube/src/stores/LegendDataCubeCacheManager.ts
@@ -32,6 +32,7 @@ import {
 } from '@finos/legend-shared';
 import type { CachedDataCubeSource } from '@finos/legend-data-cube';
 import { Type } from 'apache-arrow';
+import CSVParser from 'papaparse';
 
 export class LegendDataCubeDataCubeCacheManager {
   private static readonly DUCKDB_DEFAULT_SCHEMA_NAME = 'main'; // See https://duckdb.org/docs/sql/statements/use.html
@@ -92,34 +93,62 @@ export class LegendDataCubeDataCubeCacheManager {
 
     const connection = await this.database.connect();
 
-    const columnString = result.builder.columns
-      .map((col) => col.name)
-      .join(',');
-
-    const dataString: string[] = [columnString];
-
-    result.result.rows.forEach((row) => {
-      const updatedRows = row.values.map((val) => {
-        if (val !== null && typeof val === 'string') {
-          return `'${val.replaceAll(`'`, `''`)}'`;
-        } else if (val === null) {
-          return `NULL`;
+    const columns: string[] = [];
+    const columnNames: string[] = [];
+    result.builder.columns.forEach((col) => {
+      let colType: string;
+      switch (col.type as string) {
+        case PRIMITIVE_TYPE.BOOLEAN: {
+          colType = 'BOOLEAN';
+          break;
+        }
+        case PRIMITIVE_TYPE.INTEGER: {
+          colType = 'INTEGER';
+          break;
         }
-        return val;
-      });
-      dataString.push(`${updatedRows.join(',')}`);
+        case PRIMITIVE_TYPE.NUMBER:
+        case PRIMITIVE_TYPE.DECIMAL:
+        case PRIMITIVE_TYPE.FLOAT: {
+          colType = 'FLOAT';
+          break;
+        }
+        // We don't use type DATE because DuckDB will automatically convert it to a TIMESTAMP
+        case PRIMITIVE_TYPE.STRICTDATE:
+        case PRIMITIVE_TYPE.DATETIME:
+        case PRIMITIVE_TYPE.DATE: {
+          colType = 'VARCHAR';
+          break;
+        }
+        case PRIMITIVE_TYPE.STRING: {
+          colType = 'VARCHAR';
+          break;
+        }
+        default: {
+          throw new UnsupportedOperationError(
+            `Can't initialize cache: failed to find matching DuckDB type for Pure type '${col.type}'`,
+          );
+        }
+      }
+      columns.push(`"${col.name}" ${colType}`);
+      columnNames.push(col.name);
     });
 
-    const csvString = dataString.join('\n');
+    const CREATE_TABLE_SQL = `CREATE TABLE ${schema}.${table} (${columns.join(',')})`;
+    await connection.query(CREATE_TABLE_SQL);
+
+    const data = result.result.rows.map((row) => row.values);
+
+    const csv = CSVParser.unparse([columnNames, ...data], {
+      escapeChar: `'`,
+      quoteChar: `'`,
+    });
 
-    await this._database?.registerFileText(csvFileName, csvString);
+    await this._database?.registerFileText(csvFileName, csv);
 
     await connection.insertCSVFromPath(csvFileName, {
       schema: schema,
       name: table,
       create: false,
-      header: true,
-      detect: true,
       escape: `'`,
       quote: `'`,
       delimiter: ',',