2024-10-29-13-19-50 - wip-data-pipeline

2-Python/Python.ipynb

@@ -10,10 +10,21 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 8,
    "id": "34b5f60b-2da0-41d9-be58-f410b365c194",
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Cleaning up (if any existing) tpch db file tpch.db\n",
+      "Creating TPCH input data at tpch.db\n",
+      "Cleaning up (if any existing) sqlite3 db file example.db\n",
+      "Creating sqlite database file at example.db\n"
+     ]
+    }
+   ],
    "source": [
     "! python ../setup.py"
    ]
@@ -367,7 +378,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 58,
+   "execution_count": 9,
    "id": "f193c1a7-7597-42be-9497-87926ef79a27",
    "metadata": {},
    "outputs": [],
@@ -379,21 +390,21 @@
     "cursor = conn.cursor()\n",
     "\n",
     "# Connect to DuckDB and load TPC-H tables into Pandas DataFrames\n",
-    "customer_df = con.sql(\"SELECT * FROM customer\").df()\n",
-    "orders_df = con.sql(\"SELECT * FROM orders\").df()\n",
-    "lineitem_df = con.sql(\"SELECT * FROM lineitem\").df()\n",
-    "nation_df = con.sql(\"SELECT * FROM nation\").df()\n",
-    "region_df = con.sql(\"SELECT * FROM region\").df()\n",
-    "supplier_df = con.sql(\"SELECT * FROM supplier\").df()\n",
-    "part_df = con.sql(\"SELECT * FROM part\").df()\n",
-    "partsupp_df = con.sql(\"SELECT * FROM partsupp\").df()\n",
+    "customer_df = cursor.sql(\"SELECT * FROM customer\").df()\n",
+    "orders_df = cursor.sql(\"SELECT * FROM orders\").df()\n",
+    "lineitem_df = cursor.sql(\"SELECT * FROM lineitem\").df()\n",
+    "nation_df = cursor.sql(\"SELECT * FROM nation\").df()\n",
+    "region_df = cursor.sql(\"SELECT * FROM region\").df()\n",
+    "supplier_df = cursor.sql(\"SELECT * FROM supplier\").df()\n",
+    "part_df = cursor.sql(\"SELECT * FROM part\").df()\n",
+    "partsupp_df = cursor.sql(\"SELECT * FROM partsupp\").df()\n",
     "\n",
     "conn.close()"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 60,
+   "execution_count": 10,
    "id": "a9080b3a-d11b-4cca-9754-a2c0b35dbe9d",
    "metadata": {},
    "outputs": [],
@@ -403,7 +414,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 61,
+   "execution_count": 11,
    "id": "6a85008e-166e-4136-9611-ee9dc0fc7d61",
    "metadata": {},
    "outputs": [
@@ -591,7 +602,7 @@
        "320  ng the final, bold requests. furiously regular...  "
       ]
      },
-     "execution_count": 61,
+     "execution_count": 11,
      "metadata": {},
      "output_type": "execute_result"
     }