Add a simple notebook that actually runs some diffs.

devtools/sqlite-table-diff.ipynb

@@ -0,0 +1,91 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "\"\"\"Example of diffing tables across multiple different SQLite DBs.\n",
+    "\n",
+    "The tables must have the same name/schema. This is intended for use in\n",
+    "investigating validation test errors.\n",
+    "\"\"\"\n",
+    "import sqlite3\n",
+    "from pathlib import Path\n",
+    "from typing import Iterable\n",
+    "\n",
+    "import pandas as pd\n",
+    "\n",
+    "from pudl.helpers import diff_wide_tables, TableDiff\n",
+    "from pudl.metadata.classes import Resource\n",
+    "from pudl.metadata.fields import apply_pudl_dtypes\n",
+    "\n",
+    "\n",
+    "def table_diff(\n",
+    "        table_name: str,\n",
+    "        old_conn,\n",
+    "        new_conn,\n",
+    "        ignore_cols: Iterable[str] = (\"plant_id_ferc1\",),\n",
+    "        addl_key_cols: Iterable[str] = (),\n",
+    "    ) -> TableDiff:\n",
+    "\n",
+    "    \"\"\"Diff two versions of the same table that live in SQL databases.\n",
+    "\n",
+    "    The table has to have the same name + columns in both DBs.\n",
+    "    \"\"\"\n",
+    "    query = f\"SELECT * FROM {table_name}\"\n",
+    "    old_table = apply_pudl_dtypes(pd.read_sql(query, old_conn))\n",
+    "    new_table = apply_pudl_dtypes(pd.read_sql(query, new_conn))\n",
+    "\n",
+    "    cols = list(set(old_table.columns) - set(ignore_cols))\n",
+    "\n",
+    "    primary_key = list(set(Resource.from_id(table_name).schema.primary_key).union(set(addl_key_cols)))\n",
+    "    return diff_wide_tables(primary_key, old_table[cols], new_table[cols])\n",
+    "\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "new_db = sqlite3.connect(Path(\"~/Downloads/pudl.sqlite\").expanduser())\n",
+    "old_db = sqlite3.connect(Path(\"~/Downloads/pudl (2).sqlite\").expanduser())\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "table_name = \"denorm_plants_steam_ferc1\"\n",
+    "diff = table_diff(table_name, old_db, new_db, ignore_cols=(\"plant_id_ferc1\", \"plant_id_pudl\"), addl_key_cols=(\"report_year\", \"utility_id_pudl\"))\n",
+    "diff.changed"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "pudl-dev",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

src/pudl/helpers.py

@@ -1889,6 +1889,15 @@ def diff_wide_tables(
     )
     old_aligned, new_aligned = old_melted.align(new_melted)
     comparison = old_aligned.compare(new_aligned, result_names=("old", "new"))
+    if comparison.empty:
+        return TableDiff(
+            deleted=pd.DataFrame(),
+            added=pd.DataFrame(),
+            changed=pd.DataFrame(),
+            old_df=old,
+            new_df=new,
+        )
+
     old_values = comparison[("value", "old")]
     new_values = comparison[("value", "new")]
     added = comparison[old_values.isna() & new_values.notna()]