From d91676db6f0b70ac439d6916aaad8a0617f018a0 Mon Sep 17 00:00:00 2001
From: Mohammad Ahtasham ul Hassan <60315450+aht007@users.noreply.github.com>
Date: Thu, 14 Nov 2024 13:59:28 +0500
Subject: [PATCH] feat: add ct duplicates management command (#295)

* feat: add ct duplicates management command
---
 .../commands/find_ct_duplicate_purchases.py   | 71 +++++++++++++++++++
 1 file changed, 71 insertions(+)
 create mode 100644 commerce_coordinator/apps/commercetools/management/commands/find_ct_duplicate_purchases.py

diff --git a/commerce_coordinator/apps/commercetools/management/commands/find_ct_duplicate_purchases.py b/commerce_coordinator/apps/commercetools/management/commands/find_ct_duplicate_purchases.py
new file mode 100644
index 00000000..2542e8c7
--- /dev/null
+++ b/commerce_coordinator/apps/commercetools/management/commands/find_ct_duplicate_purchases.py
@@ -0,0 +1,71 @@
+import json
+from collections import defaultdict
+
+from commerce_coordinator.apps.commercetools.management.commands._ct_api_client_command import (
+    CommercetoolsAPIClientCommand
+)
+
+""" Those intending to use this script, please take into consideration that currently
+the script doesn't cater for the returns (refunds) and hence
+it will show the duplicates even if the user has returned the product.
+ """
+
+
+class Command(CommercetoolsAPIClientCommand):
+    help = "Find duplicate purchases in commercetools"
+
+    def handle(self, *args, **options):
+
+        order_state = "Complete"
+        # TODO: Change for your use case
+        last_modified_at = "2024-10-23T00:00:00"
+        limit = 500
+        offset = 0
+
+        orders_result = self.ct_api_client.base_client.orders.query(
+            where=[
+                f'orderState="{order_state}"',
+                f'lastModifiedAt>"{last_modified_at}"',
+            ],
+            sort=["completedAt desc", "lastModifiedAt desc"],
+            limit=limit,
+            offset=offset,
+        )
+
+        orders = orders_result.results
+
+        while orders_result.offset + orders_result.limit < orders_result.total:
+            orders_result = self.ct_api_client.base_client.orders.query(
+                where=[
+                    f'orderState="{order_state}"',
+                    'lastModifiedAt>"2024-10-23T00:00:00"',
+                ],
+                sort=["completedAt desc", "lastModifiedAt desc"],
+                limit=orders_result.limit,
+                offset=orders_result.offset + orders_result.limit,
+            )
+            orders.extend(orders_result.results)
+
+        user_orders = defaultdict(lambda: defaultdict(list))
+
+        for order in orders:
+            user_email = order.customer_email
+            line_items = order.line_items
+
+            for item in line_items:
+                sku = item.variant.sku
+                user_orders[user_email][sku].append(order.id)
+
+        duplicate_purchases = {}
+        for user_id, sku_orders in user_orders.items():
+            duplicates = {
+                sku: order_ids
+                for sku, order_ids in sku_orders.items()
+                if len(order_ids) > 1
+            }
+            if duplicates:
+                duplicate_purchases[user_id] = duplicates
+
+        # dump to a json file
+        with open("duplicate_purchases.json", "w") as f:
+            json.dump(duplicate_purchases, f)