amazon-orders 3.2.1__tar.gz → 3.2.3__tar.gz

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/CHANGELOG.md

@@ -4,7 +4,23 @@ All notable changes to this project will be documented in this file.
 
 This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
-## [Unreleased](https://github.com/alexdlaird/amazon-orders/compare/3.2.0...HEAD)
+## [Unreleased](https://github.com/alexdlaird/amazon-orders/compare/3.2.2...HEAD)
+
+## [3.2.3](https://github.com/alexdlaird/amazon-orders/compare/3.2.1...3.2.2) - 2025-02-06
+
+### Added
+
+- `bs4_parser` to the config file, which allows for overriding the parser used by BeautifulSoup, defaulting to the built-in `html.parser`.
+
+## [3.2.2](https://github.com/alexdlaird/amazon-orders/compare/3.2.1...3.2.2) - 2025-01-28
+
+### Added
+
+- Stability improvements.
+
+### Fixed
+
+- Broken parsing of Amazon Fresh orders, these are now skipped.
 
 ## [3.2.1](https://github.com/alexdlaird/amazon-orders/compare/3.2.0...3.2.1) - 2024-11-08
 

{amazon_orders-3.2.1/amazon_orders.egg-info → amazon_orders-3.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.2
 Name: amazon-orders
-Version: 3.2.1
+Version: 3.2.3
 Summary: A CLI and library for interacting with Amazon order history.
 Maintainer-email: Alex Laird <contact@alexlaird.com>
 License: MIT License
@@ -62,6 +62,7 @@ Requires-Dist: responses; extra == "dev"
 Requires-Dist: flask; extra == "dev"
 Requires-Dist: twilio; extra == "dev"
 Requires-Dist: pyngrok; extra == "dev"
+Requires-Dist: lxml; extra == "dev"
 Provides-Extra: docs
 Requires-Dist: Sphinx; extra == "docs"
 Requires-Dist: sphinx-notfound-page; extra == "docs"

{amazon_orders-3.2.1 → amazon_orders-3.2.3/amazon_orders.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.2
 Name: amazon-orders
-Version: 3.2.1
+Version: 3.2.3
 Summary: A CLI and library for interacting with Amazon order history.
 Maintainer-email: Alex Laird <contact@alexlaird.com>
 License: MIT License
@@ -62,6 +62,7 @@ Requires-Dist: responses; extra == "dev"
 Requires-Dist: flask; extra == "dev"
 Requires-Dist: twilio; extra == "dev"
 Requires-Dist: pyngrok; extra == "dev"
+Requires-Dist: lxml; extra == "dev"
 Provides-Extra: docs
 Requires-Dist: Sphinx; extra == "docs"
 Requires-Dist: sphinx-notfound-page; extra == "docs"

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/amazon_orders.egg-info/requires.txt

@@ -16,6 +16,7 @@ responses
 flask
 twilio
 pyngrok
+lxml
 
 [docs]
 Sphinx

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/amazonorders/__init__.py

@@ -1,3 +1,3 @@
 __copyright__ = "Copyright (c) 2024 Alex Laird"
 __license__ = "MIT"
-__version__ = "3.2.1"
+__version__ = "3.2.3"

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/amazonorders/conf.py

@@ -36,6 +36,7 @@ class AmazonOrdersConfig:
             "order_class": "amazonorders.entity.order.Order",
             "shipment_class": "amazonorders.entity.shipment.Shipment",
             "item_class": "amazonorders.entity.item.Item",
+            "bs4_parser": "html.parser",
         }
 
         if os.path.exists(self.config_path):

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/amazonorders/entity/order.py

@@ -146,7 +146,7 @@ class Order(Parsable):
                 data_popover = value.get("data-a-popover", {})  # type: ignore[arg-type, var-annotated]
                 inline_content = data_popover.get("inlineContent")  # type: ignore[union-attr]
                 if inline_content:
-                    value = BeautifulSoup(json.loads(inline_content), "html.parser")
+                    value = BeautifulSoup(json.loads(inline_content), self.config.bs4_parser)
 
         if not value:
             # TODO: there are multiple shipToData tags, we should double check we're picking the right one
@@ -166,7 +166,7 @@ class Order(Parsable):
             )
 
             if parent_tag:
-                value = BeautifulSoup(str(parent_tag.contents[0]).strip(), "html.parser")
+                value = BeautifulSoup(str(parent_tag.contents[0]).strip(), self.config.bs4_parser)
 
         if not value:
             return None

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/amazonorders/orders.py

@@ -73,6 +73,12 @@ class AmazonOrders:
             response_parsed = self.amazon_session.last_response_parsed
 
             for order_tag in util.select(response_parsed, self.config.selectors.ORDER_HISTORY_ENTITY_SELECTOR):
+                # If we find a brand logo (for instance, Amazon Fresh), we don't know how to parse this. If we know how
+                # to do this in the future, we can implement it, but right now we have no reliable way, so skipping
+                # these orders.
+                if util.select(order_tag, self.config.selectors.ORDER_HISTORY_BRAND_SELECTOR):
+                    continue
+
                 order: Order = self.config.order_cls(order_tag, self.config)
 
                 if full_details:

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/amazonorders/selectors.py

@@ -55,6 +55,7 @@ class Selectors:
     SHIPMENT_ENTITY_SELECTOR = ["[data-component='orderCard'] [data-component='shipments'] .a-box",
                                 "div.shipment",
                                 "div.delivery-box"]
+    ORDER_HISTORY_BRAND_SELECTOR = ".brand-info-box .brand-logo img"
 
     #####################################
     # CSS selectors for Item fields

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/amazonorders/session.py

@@ -141,7 +141,7 @@ class AmazonSession:
 
         self.last_response = self.session.request(method, url, **kwargs)
         self.last_response_parsed = BeautifulSoup(self.last_response.text,
-                                                  "html.parser")
+                                                  self.config.bs4_parser)
 
         cookies = dict_from_cookiejar(self.session.cookies)
         if os.path.exists(self.config.cookie_jar_path):

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/amazonorders/transactions.py

@@ -17,19 +17,15 @@ from amazonorders.session import AmazonSession
 logger = logging.getLogger(__name__)
 
 
-def _get_today() -> datetime.date:
-    return datetime.date.today()
-
-
-def _parse_transaction_form_tag(
-        form_tag: Tag, config: AmazonOrdersConfig
-) -> Tuple[List[Transaction], Optional[str], Optional[Dict[str, str]]]:
+def _parse_transaction_form_tag(form_tag: Tag,
+                                config: AmazonOrdersConfig) \
+        -> Tuple[List[Transaction], Optional[str], Optional[Dict[str, str]]]:
     transactions = []
     date_container_tags = util.select(form_tag, config.selectors.TRANSACTION_DATE_CONTAINERS_SELECTOR)
     for date_container_tag in date_container_tags:
         date_tag = util.select_one(date_container_tag, config.selectors.FIELD_TRANSACTION_COMPLETED_DATE_SELECTOR)
         if not date_tag:
-            logger.warning("Could not find date tag in transaction form.")
+            logger.warning("Could not find date tag in Transaction form.")
             continue
 
         date_str = date_tag.text
@@ -38,9 +34,7 @@ def _parse_transaction_form_tag(
         transactions_container_tag = date_container_tag.find_next_sibling(
             config.selectors.TRANSACTIONS_CONTAINER_SELECTOR)
         if not isinstance(transactions_container_tag, Tag):
-            logger.warning(
-                "Could not find transactions container tag in transaction form."
-            )
+            logger.warning("Could not find transactions container tag in Transaction form.")
             continue
 
         transaction_tags = util.select(transactions_container_tag, config.selectors.TRANSACTIONS_SELECTOR)
@@ -52,7 +46,7 @@ def _parse_transaction_form_tag(
     form_ie_input = util.select_one(form_tag, config.selectors.TRANSACTIONS_NEXT_PAGE_INPUT_IE_SELECTOR)
     next_page_input = util.select_one(form_tag, config.selectors.TRANSACTIONS_NEXT_PAGE_INPUT_SELECTOR)
     if not next_page_input or not form_state_input or not form_ie_input:
-        return (transactions, None, None)
+        return transactions, None, None
 
     next_page_post_url = str(form_tag["action"])
     next_page_post_data = {
@@ -61,7 +55,7 @@ def _parse_transaction_form_tag(
         str(next_page_input["name"]): "",
     }
 
-    return (transactions, next_page_post_url, next_page_post_data)
+    return transactions, next_page_post_url, next_page_post_data
 
 
 class AmazonTransactions:
@@ -100,7 +94,7 @@ class AmazonTransactions:
         if not self.amazon_session.is_authenticated:
             raise AmazonOrdersError("Call AmazonSession.login() to authenticate first.")
 
-        min_date = _get_today() - datetime.timedelta(days=days)
+        min_date = datetime.date.today() - datetime.timedelta(days=days)
 
         self.amazon_session.get(self.config.constants.TRANSACTION_HISTORY_LANDING_URL)
         if not self.amazon_session.last_response_parsed:

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/pyproject.toml

@@ -41,7 +41,8 @@ dev = [
     "responses",
     "flask",
     "twilio",
-    "pyngrok"
+    "pyngrok",
+    "lxml"
 ]
 docs = [
     "Sphinx",

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/tests/test_cli.py

@@ -86,10 +86,10 @@ class TestCli(UnitTestCase):
         self.assertIn("Order #112-2961628-4757846", response.output)
 
     @responses.activate
-    @patch("amazonorders.transactions._get_today")
+    @patch("amazonorders.transactions.datetime", wraps=datetime)
     def test_transactions_command(self, mock_get_today: Mock):
         # GIVEN
-        mock_get_today.return_value = datetime.date(2024, 10, 11)
+        mock_get_today.date.today.return_value = datetime.date(2024, 10, 11)
         days = 1
         self.given_login_responses_success()
         with open(os.path.join(self.RESOURCES_DIR, "get-transactions.html"), "r", encoding="utf-8") as f:

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/tests/test_conf.py

@@ -41,6 +41,7 @@ class TestConf(TestCase):
         self.assertEqual(10, config.max_auth_attempts)
         self.assertEqual(self.test_output_dir, config.output_dir)
         self.assertEqual(self.test_cookie_jar_path, config.cookie_jar_path)
+        self.assertEqual("html.parser", config.bs4_parser)
 
         # GIVEN
         config.save()
@@ -48,7 +49,8 @@ class TestConf(TestCase):
         # THEN
         self.assertTrue(os.path.exists(config_path))
         with open(config.config_path, "r") as f:
-            self.assertEqual("""constants_class: amazonorders.constants.Constants
+            self.assertEqual("""bs4_parser: html.parser
+constants_class: amazonorders.constants.Constants
 cookie_jar_path: {}
 item_class: amazonorders.entity.item.Item
 max_auth_attempts: 10

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/tests/test_orders.py

@@ -103,6 +103,30 @@ class TestOrders(UnitTestCase):
         self.assertEqual(1, resp2.call_count)
         self.assertEqual(1, resp3.call_count)
 
+    @responses.activate
+    def test_get_order_history_skip_fresh(self):
+        # GIVEN
+        self.amazon_session.is_authenticated = True
+        year = 2024
+        start_index = 0
+        resp1 = self.given_order_history_landing_exists()
+        with open(os.path.join(self.RESOURCES_DIR, "order-history-fresh.html"), "r",
+                  encoding="utf-8") as f:
+            resp2 = responses.add(
+                responses.GET,
+                self.test_config.constants.ORDER_HISTORY_URL,
+                body=f.read(),
+                status=200,
+            )
+
+        # WHEN
+        orders = self.amazon_orders.get_order_history(year=year, start_index=start_index)
+
+        # THEN
+        self.assertEqual(9, len(orders))
+        self.assertEqual(1, resp1.call_count)
+        self.assertEqual(1, resp2.call_count)
+
     @responses.activate
     def test_get_order_history_full_details(self):
         # GIVEN

{amazon_orders-3.2.1 → amazon_orders-3.2.3}/tests/test_transactions.py

@@ -31,16 +31,16 @@ class TestOrders(UnitTestCase):
         self.amazon_transactions = AmazonTransactions(self.amazon_session)
 
     @responses.activate
-    @patch("amazonorders.transactions._get_today")
+    @patch("amazonorders.transactions.datetime", wraps=datetime)
     def test_transactions_command(self, mock_get_today: Mock):
         # GIVEN
-        mock_get_today.return_value = datetime.date(2024, 10, 11)
+        mock_get_today.date.today.return_value = datetime.date(2024, 10, 11)
         days = 1
         self.amazon_session.is_authenticated = True
         with open(
-            os.path.join(self.RESOURCES_DIR, "get-transactions.html"),
-            "r",
-            encoding="utf-8",
+                os.path.join(self.RESOURCES_DIR, "get-transactions.html"),
+                "r",
+                encoding="utf-8",
         ) as f:
             responses.add(
                 responses.GET,
@@ -57,7 +57,7 @@ class TestOrders(UnitTestCase):
 
     def test_parse_transaction_form_tag(self):
         # GIVEN
-        parsed = BeautifulSoup(TEST_PARSE_TRANSACTION_FORM_TAG_HTML, "html.parser")
+        parsed = BeautifulSoup(TEST_PARSE_TRANSACTION_FORM_TAG_HTML, self.test_config.bs4_parser)
         form_tag = parsed.select_one("form")
 
         # WHEN