Merge pull request #56260 from mihir-kandoi/pg-wo-stock-report-rowcount

fix(manufacturing): keep Work Order Stock report one row per item (MariaDB parity)

erpnext/manufacturing/report/work_order_stock_report/test_work_order_stock_report.py

@@ -8,9 +8,9 @@ from erpnext.tests.utils import ERPNextTestSuite
 
 class TestWorkOrderStockReport(ERPNextTestSuite):
 	def test_report_executes_and_lists_work_order(self):
-		# get_item_list computes build_qty by multiplying bin/bom/bom_item columns that are not
-		# functionally dependent on the grouped item_code; they must be in the GROUP BY for the
-		# report to run on Postgres. This exercises that query on both engines.
+		# get_item_list aggregates build_qty with Max() and groups by item_code, so it returns one
+		# row per item_code and runs on Postgres (which rejects non-grouped selected columns).
+		# This exercises that query on both engines.
 		from erpnext.manufacturing.doctype.work_order.test_work_order import make_wo_order_test_record
 		from erpnext.manufacturing.report.work_order_stock_report.work_order_stock_report import execute
 

erpnext/manufacturing/report/work_order_stock_report/work_order_stock_report.py

@@ -4,7 +4,7 @@
 
 import frappe
 from frappe import _
-from frappe.query_builder.functions import IfNull
+from frappe.query_builder.functions import IfNull, Max, Sum
 from frappe.utils import cint
 
 
@@ -40,17 +40,23 @@ def get_item_list(wo_list, filters):
 					)
 					.select(
 						bom_item.item_code.as_("item_code"),
-						IfNull(bin.actual_qty * bom.quantity / bom_item.stock_qty, 0).as_("build_qty"),
+						# Aggregate so the query stays one row per item_code and is Postgres
+						# GROUP-BY-valid. A BOM may list the same item on several lines; adding the
+						# qty columns to GROUP BY would split the row and change the row count on
+						# MariaDB. actual_qty (single warehouse) and bom.quantity (single BOM) are
+						# pinned to one value, so Max() returns that value; the per-unit requirement
+						# is the TOTAL of this item across its lines, so stock_qty is summed. For the
+						# common single-line item this equals the prior expression exactly.
+						IfNull(Max(bin.actual_qty) * Max(bom.quantity) / Sum(bom_item.stock_qty), 0).as_(
+							"build_qty"
+						),
 					)
 					.where(
 						(bom.name == bom_item.parent)
 						& (bom_item.item_code == wo_item_details.item_code)
 						& (bom.name == wo_details.bom_no)
 					)
-					# build_qty multiplies columns from bin/bom/bom_item that aren't functionally
-					# dependent on the grouped item_code, so postgres requires them in the GROUP BY.
-					# The WHERE pins bom, item and warehouse to single rows, so this stays one row.
-					.groupby(bom_item.item_code, bom.quantity, bom_item.stock_qty, bin.actual_qty)
+					.groupby(bom_item.item_code)
 				).run(as_dict=1)
 
 				stock_qty = 0