amazon-orders 3.2.2__tar.gz → 3.2.4__tar.gz

{amazon_orders-3.2.2 → amazon_orders-3.2.4}/CHANGELOG.md

@@ -4,7 +4,24 @@ 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/3.2.2...HEAD)
+## [Unreleased](https://github.com/alexdlaird/amazon-orders/compare/3.2.4...HEAD)
+
+## [3.2.4](https://github.com/alexdlaird/amazon-orders/compare/3.2.3...3.2.4) - 2025-02-11
+
+### Added
+
+- `Order.promotion_applied` field, which is now parsed alongside other subtotals.
+
+### Fixed
+
+- Broken parsing of Whole Foods Market orders, these are now skipped.
+- Parsing issue on `Order.order_number` and `Order.order_placed` due to changes in Amazon.com DOM.
+
+## [3.2.3](https://github.com/alexdlaird/amazon-orders/compare/3.2.2...3.2.3) - 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
 
@@ -257,7 +274,7 @@ Build and stability improvements.
 - [AuthForm](https://amazon-orders.readthedocs.io/api.html#amazonorders.forms.AuthForm) abstract class, and
   migrated all auth flow items to subclasses of this class.
 - [Parsable.simple_parse()](https://amazon-orders.readthedocs.io/api.html#amazonorders.entities.parsable.Parsable.simple_parse),
-  which can handle most basic fields when parised with CSS selectors.
+  which can handle most basic fields when parsed with CSS selectors.
 - Stability improvements.
 - Test improvements.
 

{amazon_orders-3.2.2/amazon_orders.egg-info → amazon_orders-3.2.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: amazon-orders
-Version: 3.2.2
+Version: 3.2.4
 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"
@@ -87,7 +88,7 @@ be used to interact with Amazon.com's consumer-facing website.
 
 This works by parsing website data from Amazon.com. A periodic build validates functionality to ensure its
 stability, but as Amazon provides no official API to use, this package may break at any time. This
-package only officially supports the English version of the website.
+package only officially supports the English, `.com` version of Amazon.
 
 ## Installation
 

{amazon_orders-3.2.2 → amazon_orders-3.2.4}/README.md

@@ -12,7 +12,7 @@ be used to interact with Amazon.com's consumer-facing website.
 
 This works by parsing website data from Amazon.com. A periodic build validates functionality to ensure its
 stability, but as Amazon provides no official API to use, this package may break at any time. This
-package only officially supports the English version of the website.
+package only officially supports the English, `.com` version of Amazon.
 
 ## Installation
 

{amazon_orders-3.2.2 → amazon_orders-3.2.4/amazon_orders.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: amazon-orders
-Version: 3.2.2
+Version: 3.2.4
 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"
@@ -87,7 +88,7 @@ be used to interact with Amazon.com's consumer-facing website.
 
 This works by parsing website data from Amazon.com. A periodic build validates functionality to ensure its
 stability, but as Amazon provides no official API to use, this package may break at any time. This
-package only officially supports the English version of the website.
+package only officially supports the English, `.com` version of Amazon.
 
 ## Installation
 

{amazon_orders-3.2.2 → amazon_orders-3.2.4}/amazon_orders.egg-info/requires.txt

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

{amazon_orders-3.2.2 → amazon_orders-3.2.4}/amazonorders/__init__.py

@@ -1,3 +1,3 @@
 __copyright__ = "Copyright (c) 2024 Alex Laird"
 __license__ = "MIT"
-__version__ = "3.2.2"
+__version__ = "3.2.4"

{amazon_orders-3.2.2 → amazon_orders-3.2.4}/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-3.2.2 → amazon_orders-3.2.4}/amazonorders/entity/order.py

@@ -43,7 +43,9 @@ class Order(Parsable):
         #: The Order number.
         self.order_number: str = clone.order_number if clone else self.safe_simple_parse(
             selector=self.config.selectors.FIELD_ORDER_NUMBER_SELECTOR,
-            required=True)
+            required=True,
+            prefix_split="#",
+            prefix_split_fuzzy=True)
         #: The Order details link.
         self.order_details_link: Optional[str] = clone.order_details_link if clone else self.safe_parse(
             self._parse_order_details_link)
@@ -51,7 +53,10 @@ class Order(Parsable):
         self.grand_total: float = clone.grand_total if clone else self.safe_parse(self._parse_grand_total)
         #: The Order placed date.
         self.order_placed_date: date = clone.order_placed_date if clone else self.safe_simple_parse(
-            selector=self.config.selectors.FIELD_ORDER_PLACED_DATE_SELECTOR, parse_date=True)
+            selector=self.config.selectors.FIELD_ORDER_PLACED_DATE_SELECTOR,
+            suffix_split="Order #",
+            suffix_split_fuzzy=True,
+            parse_date=True)
         #: The Order Recipients.
         self.recipient: Recipient = clone.recipient if clone else self.safe_parse(self._parse_recipient)
 
@@ -69,6 +74,8 @@ class Order(Parsable):
         self.subtotal: Optional[float] = self._if_full_details(self._parse_currency("subtotal"))
         #: The Order shipping total. Only populated when ``full_details`` is ``True``.
         self.shipping_total: Optional[float] = self._if_full_details(self._parse_currency("shipping"))
+        #: The Order promotion applied. Only populated when ``full_details`` is ``True``.
+        self.promotion_applied: Optional[float] = self._if_full_details(self._parse_currency("promotion"))
         #: The Order Subscribe & Save discount. Only populated when ``full_details`` is ``True``.
         self.subscription_discount: Optional[float] = self._if_full_details(self._parse_currency("subscribe"))
         #: The Order total before tax. Only populated when ``full_details`` is ``True``.
@@ -146,7 +153,7 @@ 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 +173,7 @@ 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-3.2.2 → amazon_orders-3.2.4}/amazonorders/entity/parsable.py

@@ -67,7 +67,10 @@ class Parsable:
                      required: bool = False,
                      prefix_split: Optional[str] = None,
                      wrap_tag: Optional[Type] = None,
-                     parse_date: bool = False) -> Any:
+                     parse_date: bool = False,
+                     prefix_split_fuzzy: bool = False,
+                     suffix_split: Optional[str] = None,
+                     suffix_split_fuzzy: bool = False) -> Any:
         """
         Will attempt to extract the text value of the given CSS selector(s) for a field, and
         is suitable for most basic functionality on a well-formed page.
@@ -86,6 +89,9 @@ class Parsable:
         :param wrap_tag: Wrap the selected tag in this class before returning.
         :param parse_date: ``True`` if the resulting value should be fuzzy parsed in to a date (returning ``None`` if
             parsing fails).
+        :param prefix_split_fuzzy: ``True`` if the value should still be used even if ``prefix_split`` is not found.
+        :param suffix_split: Only select the field with the given suffix, returning the left side of the split if so.
+        :param suffix_split_fuzzy: ``True`` if the value should still be used even if ``suffix_split`` is not found.
         :return: The cleaned up return value from the parsed ``selector``.
         """
         if isinstance(selector, str):
@@ -109,12 +115,24 @@ class Parsable:
 
                         if prefix_split:
                             if prefix_split not in tag.text:
-                                continue
+                                if prefix_split_fuzzy:
+                                    value = tag.text.strip()
+                                else:
+                                    continue
                             else:
                                 value = tag.text.strip().split(prefix_split)[1]
                         else:
                             value = tag.text
 
+                        if suffix_split:
+                            if suffix_split not in value:
+                                if suffix_split_fuzzy:
+                                    value = value.strip()
+                                else:
+                                    continue
+                            else:
+                                value = value.strip().split(suffix_split)[0]
+
                         if wrap_tag:
                             value = wrap_tag(tag, self.config)
                         else:
@@ -142,7 +160,7 @@ class Parsable:
         """
         A helper function that uses :func:`simple_parse` as the ``parse_function()`` passed to :func:`safe_parse`.
 
-        :param selector: The selector to pass to :func:`simple_parse`.
+        :param selector: The CSS selector to pass to :func:`simple_parse`.
         :param kwargs: The ``kwargs`` will be passed to ``parse_function``.
         :return: The return value from :func:`simple_parse`.
         """

{amazon_orders-3.2.2 → amazon_orders-3.2.4}/amazonorders/orders.py

@@ -73,9 +73,8 @@ 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.
+                # First check if this Order is known to be of a type that we do not currently have a way to support
+                # parsing, meaning it should be skipped
                 if util.select(order_tag, self.config.selectors.ORDER_HISTORY_BRAND_SELECTOR):
                     continue
 

{amazon_orders-3.2.2 → amazon_orders-3.2.4}/amazonorders/selectors.py

@@ -55,7 +55,14 @@ 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"
+    # Selectors defined here, if found in an Order, will cause the Order to be skipped, since it means we currently
+    # do not have a way to support fully parsing its details
+    ORDER_HISTORY_BRAND_SELECTOR = [
+        # Amazon Fresh is not supported
+        ".brand-info-box .brand-logo img",
+        # Whole Foods Market is not supported
+        "a.yohtmlc-order-details-link[href^='/wholefoodsmarket']"
+    ]
 
     #####################################
     # CSS selectors for Item fields
@@ -81,12 +88,14 @@ class Selectors:
     #####################################
 
     FIELD_ORDER_DETAILS_LINK_SELECTOR = "a.yohtmlc-order-details-link"
-    FIELD_ORDER_NUMBER_SELECTOR = [".order-date-invoice-item bdi[dir='ltr']",
+    FIELD_ORDER_NUMBER_SELECTOR = ["[data-component='briefOrderInfo'] div.a-column",
+                                   ".order-date-invoice-item bdi[dir='ltr']",
                                    "bdi[dir='ltr']",
                                    "span[dir='ltr']"]
     FIELD_ORDER_GRAND_TOTAL_SELECTOR = ["div.yohtmlc-order-total span.value",
                                         "div.order-header div.a-column.a-span2"]
-    FIELD_ORDER_PLACED_DATE_SELECTOR = ["span.order-date-invoice-item",
+    FIELD_ORDER_PLACED_DATE_SELECTOR = ["[data-component='briefOrderInfo'] div.a-column",
+                                        "span.order-date-invoice-item",
                                         "div.a-span3"]
     FIELD_ORDER_PAYMENT_METHOD_SELECTOR = "img.pmts-payment-credit-card-instrument-logo"
     FIELD_ORDER_PAYMENT_METHOD_LAST_4_SELECTOR = "span:has(img.pmts-payment-credit-card-instrument-logo):last-child"

{amazon_orders-3.2.2 → amazon_orders-3.2.4}/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-3.2.2 → amazon_orders-3.2.4}/pyproject.toml

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

{amazon_orders-3.2.2 → amazon_orders-3.2.4}/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-3.2.2 → amazon_orders-3.2.4}/tests/test_orders.py

@@ -127,6 +127,30 @@ class TestOrders(UnitTestCase):
         self.assertEqual(1, resp1.call_count)
         self.assertEqual(1, resp2.call_count)
 
+    @responses.activate
+    def test_get_order_history_skip_wholefoods(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-wholefoods.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-3.2.2 → amazon_orders-3.2.4}/tests/test_transactions.py

@@ -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