forked from cartsnitch/cartsnitch
Squashed 'receiptwitness/' content from commit e8d374a
git-subtree-dir: receiptwitness git-subtree-split: e8d374a89ed8978f429598e02d31b1c5963efe22
This commit is contained in:
Coupon Carl
@@ -0,0 +1,435 @@
|
||||
"""Regression tests: graceful handling of page layout changes.
|
||||
|
||||
Retailers frequently change their API response structures, field names,
|
||||
and nesting. These tests verify that both parsers degrade gracefully when
|
||||
encountering alternative or missing fields — producing valid output
|
||||
instead of crashing.
|
||||
"""
|
||||
|
||||
from decimal import Decimal
|
||||
|
||||
from receiptwitness.parsers.kroger import parse_kroger_receipt
|
||||
from receiptwitness.parsers.meijer import parse_meijer_receipt
|
||||
from receiptwitness.scrapers.base import RawReceipt
|
||||
|
||||
|
||||
class TestKrogerFieldNameVariations:
|
||||
"""Kroger changes field names between app versions and API revisions."""
|
||||
|
||||
def test_alternative_item_key_line_items(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-ALT-1",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"lineItems": [{"description": "MILK", "basePrice": 3.99, "totalPrice": 3.99}],
|
||||
"total": 3.99,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert len(result["items"]) == 1
|
||||
assert result["items"][0]["product_name_raw"] == "MILK"
|
||||
|
||||
def test_alternative_item_key_receipt_items(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-ALT-2",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"receiptItems": [
|
||||
{"description": "EGGS", "basePrice": 5.49, "totalPrice": 5.49}
|
||||
],
|
||||
"total": 5.49,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert len(result["items"]) == 1
|
||||
assert result["items"][0]["product_name_raw"] == "EGGS"
|
||||
|
||||
def test_alternative_description_fields(self):
|
||||
"""Test productName and itemDescription fallbacks."""
|
||||
for field in ("productName", "itemDescription", "name"):
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-DESC",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [{field: "TEST PRODUCT", "basePrice": 1.00, "totalPrice": 1.00}],
|
||||
"total": 1.00,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert result["items"][0]["product_name_raw"] == "TEST PRODUCT"
|
||||
|
||||
def test_alternative_price_fields(self):
|
||||
"""Test unitPrice and price fallbacks for basePrice."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-PRICE-1",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [{"description": "ITEM A", "unitPrice": 2.50, "totalPrice": 2.50}],
|
||||
"total": 2.50,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert result["items"][0]["unit_price"] == Decimal("2.50")
|
||||
|
||||
raw2 = RawReceipt(
|
||||
receipt_id="KR-PRICE-2",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [{"description": "ITEM B", "price": 4.00, "totalPrice": 4.00}],
|
||||
"total": 4.00,
|
||||
}
|
||||
},
|
||||
)
|
||||
result2 = parse_kroger_receipt(raw2)
|
||||
assert result2["items"][0]["unit_price"] == Decimal("4.00")
|
||||
|
||||
def test_alternative_total_fields(self):
|
||||
"""Test orderTotal, grandTotal fallbacks."""
|
||||
for field in ("orderTotal", "grandTotal"):
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-TOT",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={field: 42.50, "detail": {}},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert result["total"] == Decimal("42.50")
|
||||
|
||||
def test_alternative_savings_fields(self):
|
||||
"""Test youSaved and totalDiscount fallbacks."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-SAV-1",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={"youSaved": 5.00, "detail": {}},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert result["savings_total"] == Decimal("5.00")
|
||||
|
||||
def test_alternative_tax_field(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-TAX",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={"salesTax": 3.25, "detail": {}},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert result["tax"] == Decimal("3.25")
|
||||
|
||||
def test_alternative_quantity_field_qty(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-QTY",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [
|
||||
{"description": "APPLES", "qty": 5, "basePrice": 1.00, "totalPrice": 5.00}
|
||||
],
|
||||
"total": 5.00,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert result["items"][0]["quantity"] == Decimal("5")
|
||||
|
||||
def test_alternative_upc_field_kroger_product_id(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-UPC",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [
|
||||
{
|
||||
"description": "ITEM",
|
||||
"krogerProductId": "12345678",
|
||||
"basePrice": 1.00,
|
||||
"totalPrice": 1.00,
|
||||
}
|
||||
],
|
||||
"total": 1.00,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert result["items"][0]["upc"] == "12345678"
|
||||
|
||||
def test_missing_extended_price_computed(self):
|
||||
"""When totalPrice is missing, extended_price = unit_price * quantity."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-CALC",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [{"description": "EGGS", "basePrice": 5.49, "quantity": 2}],
|
||||
"total": 10.98,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert result["items"][0]["extended_price"] == Decimal("5.49") * Decimal("2")
|
||||
|
||||
|
||||
class TestMeijerFieldNameVariations:
|
||||
"""Meijer XHR endpoints may change field names between SPA versions."""
|
||||
|
||||
def test_alternative_item_key_line_items(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-ALT-1",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"lineItems": [{"description": "BANANAS", "price": 0.69, "extendedPrice": 0.69}],
|
||||
"total": 0.69,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert len(result["items"]) == 1
|
||||
assert result["items"][0]["product_name_raw"] == "BANANAS"
|
||||
|
||||
def test_alternative_description_fields(self):
|
||||
for field in ("itemDescription", "name"):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-DESC",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [{field: "TEST ITEM", "price": 1.00, "extendedPrice": 1.00}],
|
||||
"total": 1.00,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["items"][0]["product_name_raw"] == "TEST ITEM"
|
||||
|
||||
def test_alternative_price_field_unit_price(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-PRICE",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [{"description": "MILK", "unitPrice": 3.49, "totalPrice": 3.49}],
|
||||
"total": 3.49,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["items"][0]["unit_price"] == Decimal("3.49")
|
||||
|
||||
def test_alternative_extended_price_field_total_price(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-EXT",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [{"description": "CEREAL", "price": 4.99, "totalPrice": 4.99}],
|
||||
"total": 4.99,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["items"][0]["extended_price"] == Decimal("4.99")
|
||||
|
||||
def test_alternative_total_field_transaction_total(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-TOT",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={"transactionTotal": 55.00, "detail": {}},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["total"] == Decimal("55.00")
|
||||
|
||||
def test_alternative_loyalty_field(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-LOY",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [
|
||||
{
|
||||
"description": "ITEM",
|
||||
"price": 5.00,
|
||||
"extendedPrice": 5.00,
|
||||
"loyaltyDiscount": 0.50,
|
||||
}
|
||||
],
|
||||
"total": 5.00,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["items"][0]["loyalty_discount"] == Decimal("0.50")
|
||||
|
||||
def test_alternative_upc_field_uppercase(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-UPC",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [
|
||||
{
|
||||
"description": "ITEM",
|
||||
"UPC": "0012345678",
|
||||
"price": 1.00,
|
||||
"extendedPrice": 1.00,
|
||||
}
|
||||
],
|
||||
"total": 1.00,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["items"][0]["upc"] == "12345678"
|
||||
|
||||
def test_alternative_category_field(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-CAT",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [
|
||||
{
|
||||
"description": "ITEM",
|
||||
"price": 1.00,
|
||||
"extendedPrice": 1.00,
|
||||
"departmentDescription": "FROZEN",
|
||||
}
|
||||
],
|
||||
"total": 1.00,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["items"][0]["category_raw"] == "FROZEN"
|
||||
|
||||
def test_missing_extended_price_computed(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-CALC",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [{"description": "MILK", "price": 3.49, "quantity": 2}],
|
||||
"total": 6.98,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["items"][0]["extended_price"] == Decimal("3.49") * Decimal("2")
|
||||
|
||||
def test_missing_description_fallback(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-NODESC",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [{"price": 1.00, "extendedPrice": 1.00}],
|
||||
"total": 1.00,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["items"][0]["product_name_raw"] == "UNKNOWN ITEM"
|
||||
|
||||
|
||||
class TestMixedFieldVersions:
|
||||
"""Test receipts that mix field naming conventions (happens during rollouts)."""
|
||||
|
||||
def test_kroger_mixed_item_fields(self):
|
||||
"""Some items use old names, some use new names in same receipt."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-MIX",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [
|
||||
{"description": "OLD STYLE", "basePrice": 2.00, "totalPrice": 2.00},
|
||||
{"productName": "NEW STYLE", "unitPrice": 3.00, "extendedAmount": 3.00},
|
||||
],
|
||||
"total": 5.00,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert len(result["items"]) == 2
|
||||
assert result["items"][0]["product_name_raw"] == "OLD STYLE"
|
||||
assert result["items"][0]["unit_price"] == Decimal("2.00")
|
||||
assert result["items"][1]["product_name_raw"] == "NEW STYLE"
|
||||
assert result["items"][1]["unit_price"] == Decimal("3.00")
|
||||
|
||||
def test_kroger_completely_unknown_structure_no_crash(self):
|
||||
"""Receipt with unrecognized structure should return empty items."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-UNKNOWN",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={"something_unexpected": [1, 2, 3], "detail": {"foo": "bar"}},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert result["receipt_id"] == "KR-UNKNOWN"
|
||||
assert result["items"] == []
|
||||
|
||||
def test_meijer_completely_unknown_structure_no_crash(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-UNKNOWN",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={"something_unexpected": [1, 2, 3], "detail": {"foo": "bar"}},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["receipt_id"] == "MJ-UNKNOWN"
|
||||
assert result["items"] == []
|
||||
|
||||
def test_kroger_null_fields_no_crash(self):
|
||||
"""Fields with None values should be handled gracefully."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-NULL",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [
|
||||
{
|
||||
"description": "ITEM",
|
||||
"basePrice": None,
|
||||
"totalPrice": None,
|
||||
"quantity": None,
|
||||
"upc": None,
|
||||
"department": None,
|
||||
}
|
||||
],
|
||||
"total": None,
|
||||
"subtotal": None,
|
||||
"tax": None,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert result["items"][0]["product_name_raw"] == "ITEM"
|
||||
assert result["items"][0]["unit_price"] == Decimal("0")
|
||||
|
||||
def test_meijer_null_fields_no_crash(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-NULL",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [
|
||||
{
|
||||
"description": "ITEM",
|
||||
"price": None,
|
||||
"extendedPrice": None,
|
||||
"quantity": None,
|
||||
"upc": None,
|
||||
"category": None,
|
||||
}
|
||||
],
|
||||
"total": None,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["items"][0]["product_name_raw"] == "ITEM"
|
||||
assert result["items"][0]["unit_price"] == Decimal("0")
|
||||
@@ -0,0 +1,365 @@
|
||||
"""Regression tests: rate limiting and retry behavior.
|
||||
|
||||
Validates that scrapers enforce human-like delays between requests
|
||||
and handle rate-limit/error responses gracefully without infinite retries.
|
||||
"""
|
||||
|
||||
from datetime import UTC, datetime, timedelta
|
||||
from unittest.mock import AsyncMock, patch
|
||||
|
||||
import pytest
|
||||
|
||||
from receiptwitness.scrapers.base import SessionData
|
||||
from receiptwitness.scrapers.kroger import DEFAULT_USER_AGENT, KrogerScraper
|
||||
from receiptwitness.scrapers.meijer import MeijerScraper
|
||||
|
||||
|
||||
class TestHumanDelayBehavior:
|
||||
"""Verify that human_delay respects configured bounds."""
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_delay_within_bounds(self):
|
||||
"""human_delay should sleep between min_ms/1000 and max_ms/1000 seconds."""
|
||||
scraper = KrogerScraper()
|
||||
sleep_path = "receiptwitness.scrapers.base.asyncio.sleep"
|
||||
with patch(sleep_path, new_callable=AsyncMock) as mock_sleep:
|
||||
await scraper.human_delay(100, 200)
|
||||
mock_sleep.assert_called_once()
|
||||
delay = mock_sleep.call_args[0][0]
|
||||
assert 0.1 <= delay <= 0.2
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_delay_uses_settings_defaults(self):
|
||||
"""Without explicit args, should use settings.min/max_request_delay_ms."""
|
||||
scraper = MeijerScraper()
|
||||
sleep_path = "receiptwitness.scrapers.base.asyncio.sleep"
|
||||
with (
|
||||
patch("receiptwitness.scrapers.base.settings") as mock_settings,
|
||||
patch(sleep_path, new_callable=AsyncMock) as mock_sleep,
|
||||
):
|
||||
mock_settings.min_request_delay_ms = 1000
|
||||
mock_settings.max_request_delay_ms = 5000
|
||||
await scraper.human_delay()
|
||||
mock_sleep.assert_called_once()
|
||||
delay = mock_sleep.call_args[0][0]
|
||||
assert 1.0 <= delay <= 5.0
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_delay_is_randomized(self):
|
||||
"""Multiple calls should produce different delays (probabilistic)."""
|
||||
scraper = KrogerScraper()
|
||||
delays = []
|
||||
sleep_path2 = "receiptwitness.scrapers.base.asyncio.sleep"
|
||||
with patch(sleep_path2, new_callable=AsyncMock) as mock_sleep:
|
||||
for _ in range(20):
|
||||
await scraper.human_delay(100, 5000)
|
||||
delays.append(mock_sleep.call_args[0][0])
|
||||
# With range 100-5000ms, 20 calls should have at least 2 distinct values
|
||||
assert len(set(delays)) >= 2
|
||||
|
||||
|
||||
class TestKrogerRateLimiting:
|
||||
"""Verify Kroger scraper calls human_delay between receipt fetches."""
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_delay_called_between_receipts(self):
|
||||
"""Scraper must call human_delay for each receipt detail fetch."""
|
||||
scraper = KrogerScraper()
|
||||
valid_session = SessionData(
|
||||
cookies=[{"name": "s", "value": "v", "domain": ".kroger.com", "path": "/"}],
|
||||
user_agent=DEFAULT_USER_AGENT,
|
||||
created_at=datetime.now(UTC),
|
||||
expires_at=datetime.now(UTC) + timedelta(hours=2),
|
||||
)
|
||||
|
||||
mock_api_response = AsyncMock()
|
||||
mock_api_response.ok = True
|
||||
mock_api_response.json = AsyncMock(
|
||||
return_value={
|
||||
"orders": [
|
||||
{
|
||||
"orderId": f"KR-{i}",
|
||||
"purchaseDate": "2026-03-10T14:00:00Z",
|
||||
"storeNumber": "357",
|
||||
}
|
||||
for i in range(3)
|
||||
]
|
||||
}
|
||||
)
|
||||
|
||||
mock_detail_response = AsyncMock()
|
||||
mock_detail_response.ok = True
|
||||
mock_detail_response.json = AsyncMock(return_value={})
|
||||
|
||||
mock_request = AsyncMock()
|
||||
mock_request.get = AsyncMock(side_effect=[mock_api_response] + [mock_detail_response] * 3)
|
||||
|
||||
mock_page = AsyncMock()
|
||||
mock_page.goto = AsyncMock()
|
||||
mock_page.request = mock_request
|
||||
|
||||
mock_context = AsyncMock()
|
||||
mock_context.new_page = AsyncMock(return_value=mock_page)
|
||||
mock_context.add_cookies = AsyncMock()
|
||||
mock_context.add_init_script = AsyncMock()
|
||||
mock_browser = AsyncMock()
|
||||
mock_browser.new_context = AsyncMock(return_value=mock_context)
|
||||
mock_context.browser = mock_browser
|
||||
|
||||
mock_pw = AsyncMock()
|
||||
mock_pw.chromium.launch = AsyncMock(return_value=mock_browser)
|
||||
|
||||
with (
|
||||
patch("receiptwitness.scrapers.kroger.async_playwright") as mock_apw,
|
||||
patch.object(scraper, "human_delay", new_callable=AsyncMock) as mock_delay,
|
||||
):
|
||||
mock_cm = AsyncMock()
|
||||
mock_cm.__aenter__ = AsyncMock(return_value=mock_pw)
|
||||
mock_cm.__aexit__ = AsyncMock(return_value=False)
|
||||
mock_apw.return_value = mock_cm
|
||||
|
||||
receipts = await scraper.scrape_receipts(valid_session)
|
||||
|
||||
assert len(receipts) == 3
|
||||
# human_delay called at least once per receipt (after initial page nav)
|
||||
# Plus once for the initial navigation delay
|
||||
assert mock_delay.call_count >= 3
|
||||
|
||||
|
||||
class TestMeijerRateLimiting:
|
||||
"""Verify Meijer scraper calls human_delay between receipt fetches."""
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_delay_called_between_receipts(self):
|
||||
scraper = MeijerScraper()
|
||||
valid_session = SessionData(
|
||||
cookies=[{"name": "s", "value": "v", "domain": ".meijer.com", "path": "/"}],
|
||||
user_agent="test",
|
||||
created_at=datetime.now(UTC),
|
||||
expires_at=datetime.now(UTC) + timedelta(hours=4),
|
||||
)
|
||||
|
||||
mock_api_response = AsyncMock()
|
||||
mock_api_response.ok = True
|
||||
mock_api_response.json = AsyncMock(
|
||||
return_value={
|
||||
"transactions": [
|
||||
{
|
||||
"transactionId": f"TXN-{i}",
|
||||
"transactionDate": "2026-03-10T14:00:00Z",
|
||||
"storeNumber": "42",
|
||||
}
|
||||
for i in range(3)
|
||||
]
|
||||
}
|
||||
)
|
||||
|
||||
mock_detail_response = AsyncMock()
|
||||
mock_detail_response.ok = True
|
||||
mock_detail_response.json = AsyncMock(return_value={})
|
||||
|
||||
mock_request = AsyncMock()
|
||||
mock_request.get = AsyncMock(side_effect=[mock_api_response] + [mock_detail_response] * 3)
|
||||
|
||||
mock_page = AsyncMock()
|
||||
mock_page.goto = AsyncMock()
|
||||
mock_page.request = mock_request
|
||||
|
||||
mock_context = AsyncMock()
|
||||
mock_context.new_page = AsyncMock(return_value=mock_page)
|
||||
mock_context.add_cookies = AsyncMock()
|
||||
mock_context.add_init_script = AsyncMock()
|
||||
mock_browser = AsyncMock()
|
||||
mock_browser.new_context = AsyncMock(return_value=mock_context)
|
||||
mock_context.browser = mock_browser
|
||||
|
||||
mock_pw = AsyncMock()
|
||||
mock_pw.chromium.launch = AsyncMock(return_value=mock_browser)
|
||||
|
||||
with (
|
||||
patch("receiptwitness.scrapers.meijer.async_playwright") as mock_apw,
|
||||
patch.object(scraper, "human_delay", new_callable=AsyncMock) as mock_delay,
|
||||
):
|
||||
mock_cm = AsyncMock()
|
||||
mock_cm.__aenter__ = AsyncMock(return_value=mock_pw)
|
||||
mock_cm.__aexit__ = AsyncMock(return_value=False)
|
||||
mock_apw.return_value = mock_cm
|
||||
|
||||
receipts = await scraper.scrape_receipts(valid_session)
|
||||
|
||||
assert len(receipts) == 3
|
||||
assert mock_delay.call_count >= 3
|
||||
|
||||
|
||||
class TestGracefulErrorRecovery:
|
||||
"""Scrapers should not retry endlessly on errors."""
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_kroger_api_500_returns_empty_not_retry(self):
|
||||
"""500 error should return empty list, not retry."""
|
||||
scraper = KrogerScraper()
|
||||
valid_session = SessionData(
|
||||
cookies=[{"name": "s", "value": "v", "domain": ".kroger.com", "path": "/"}],
|
||||
user_agent=DEFAULT_USER_AGENT,
|
||||
created_at=datetime.now(UTC),
|
||||
expires_at=datetime.now(UTC) + timedelta(hours=2),
|
||||
)
|
||||
|
||||
mock_api_response = AsyncMock()
|
||||
mock_api_response.ok = False
|
||||
mock_api_response.status = 500
|
||||
mock_api_response.status_text = "Internal Server Error"
|
||||
|
||||
mock_request = AsyncMock()
|
||||
mock_request.get = AsyncMock(return_value=mock_api_response)
|
||||
|
||||
mock_page = AsyncMock()
|
||||
mock_page.goto = AsyncMock()
|
||||
mock_page.request = mock_request
|
||||
|
||||
mock_context = AsyncMock()
|
||||
mock_context.new_page = AsyncMock(return_value=mock_page)
|
||||
mock_context.add_cookies = AsyncMock()
|
||||
mock_context.add_init_script = AsyncMock()
|
||||
mock_browser = AsyncMock()
|
||||
mock_browser.new_context = AsyncMock(return_value=mock_context)
|
||||
mock_context.browser = mock_browser
|
||||
|
||||
mock_pw = AsyncMock()
|
||||
mock_pw.chromium.launch = AsyncMock(return_value=mock_browser)
|
||||
|
||||
with (
|
||||
patch("receiptwitness.scrapers.kroger.async_playwright") as mock_apw,
|
||||
patch.object(scraper, "human_delay", new_callable=AsyncMock),
|
||||
):
|
||||
mock_cm = AsyncMock()
|
||||
mock_cm.__aenter__ = AsyncMock(return_value=mock_pw)
|
||||
mock_cm.__aexit__ = AsyncMock(return_value=False)
|
||||
mock_apw.return_value = mock_cm
|
||||
|
||||
receipts = await scraper.scrape_receipts(valid_session)
|
||||
assert receipts == []
|
||||
# Should only call the API once — no retries
|
||||
assert mock_request.get.call_count == 1
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_kroger_429_returns_empty_not_retry(self):
|
||||
"""Rate limit (429) should return empty, not retry."""
|
||||
scraper = KrogerScraper()
|
||||
valid_session = SessionData(
|
||||
cookies=[{"name": "s", "value": "v", "domain": ".kroger.com", "path": "/"}],
|
||||
user_agent=DEFAULT_USER_AGENT,
|
||||
created_at=datetime.now(UTC),
|
||||
expires_at=datetime.now(UTC) + timedelta(hours=2),
|
||||
)
|
||||
|
||||
mock_api_response = AsyncMock()
|
||||
mock_api_response.ok = False
|
||||
mock_api_response.status = 429
|
||||
mock_api_response.status_text = "Too Many Requests"
|
||||
|
||||
mock_request = AsyncMock()
|
||||
mock_request.get = AsyncMock(return_value=mock_api_response)
|
||||
|
||||
mock_page = AsyncMock()
|
||||
mock_page.goto = AsyncMock()
|
||||
mock_page.request = mock_request
|
||||
|
||||
mock_context = AsyncMock()
|
||||
mock_context.new_page = AsyncMock(return_value=mock_page)
|
||||
mock_context.add_cookies = AsyncMock()
|
||||
mock_context.add_init_script = AsyncMock()
|
||||
mock_browser = AsyncMock()
|
||||
mock_browser.new_context = AsyncMock(return_value=mock_context)
|
||||
mock_context.browser = mock_browser
|
||||
|
||||
mock_pw = AsyncMock()
|
||||
mock_pw.chromium.launch = AsyncMock(return_value=mock_browser)
|
||||
|
||||
with (
|
||||
patch("receiptwitness.scrapers.kroger.async_playwright") as mock_apw,
|
||||
patch.object(scraper, "human_delay", new_callable=AsyncMock),
|
||||
):
|
||||
mock_cm = AsyncMock()
|
||||
mock_cm.__aenter__ = AsyncMock(return_value=mock_pw)
|
||||
mock_cm.__aexit__ = AsyncMock(return_value=False)
|
||||
mock_apw.return_value = mock_cm
|
||||
|
||||
receipts = await scraper.scrape_receipts(valid_session)
|
||||
assert receipts == []
|
||||
assert mock_request.get.call_count == 1
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_meijer_detail_exception_continues(self):
|
||||
"""Exception fetching one receipt detail should not abort remaining receipts."""
|
||||
scraper = MeijerScraper()
|
||||
valid_session = SessionData(
|
||||
cookies=[{"name": "s", "value": "v", "domain": ".meijer.com", "path": "/"}],
|
||||
user_agent="test",
|
||||
created_at=datetime.now(UTC),
|
||||
expires_at=datetime.now(UTC) + timedelta(hours=4),
|
||||
)
|
||||
|
||||
mock_api_response = AsyncMock()
|
||||
mock_api_response.ok = True
|
||||
mock_api_response.json = AsyncMock(
|
||||
return_value={
|
||||
"transactions": [
|
||||
{
|
||||
"transactionId": "TXN-1",
|
||||
"transactionDate": "2026-03-10T14:00:00Z",
|
||||
"storeNumber": "42",
|
||||
},
|
||||
{
|
||||
"transactionId": "TXN-2",
|
||||
"transactionDate": "2026-03-11T10:00:00Z",
|
||||
"storeNumber": "42",
|
||||
},
|
||||
]
|
||||
}
|
||||
)
|
||||
|
||||
# First detail call raises exception, second succeeds
|
||||
mock_detail_fail = AsyncMock()
|
||||
mock_detail_fail.ok = False
|
||||
mock_detail_fail.status = 500
|
||||
|
||||
mock_detail_ok = AsyncMock()
|
||||
mock_detail_ok.ok = True
|
||||
mock_detail_ok.json = AsyncMock(return_value={"items": []})
|
||||
|
||||
mock_request = AsyncMock()
|
||||
mock_request.get = AsyncMock(
|
||||
side_effect=[mock_api_response, mock_detail_fail, mock_detail_ok]
|
||||
)
|
||||
|
||||
mock_page = AsyncMock()
|
||||
mock_page.goto = AsyncMock()
|
||||
mock_page.request = mock_request
|
||||
|
||||
mock_context = AsyncMock()
|
||||
mock_context.new_page = AsyncMock(return_value=mock_page)
|
||||
mock_context.add_cookies = AsyncMock()
|
||||
mock_context.add_init_script = AsyncMock()
|
||||
mock_browser = AsyncMock()
|
||||
mock_browser.new_context = AsyncMock(return_value=mock_context)
|
||||
mock_context.browser = mock_browser
|
||||
|
||||
mock_pw = AsyncMock()
|
||||
mock_pw.chromium.launch = AsyncMock(return_value=mock_browser)
|
||||
|
||||
with (
|
||||
patch("receiptwitness.scrapers.meijer.async_playwright") as mock_apw,
|
||||
patch.object(scraper, "human_delay", new_callable=AsyncMock),
|
||||
):
|
||||
mock_cm = AsyncMock()
|
||||
mock_cm.__aenter__ = AsyncMock(return_value=mock_pw)
|
||||
mock_cm.__aexit__ = AsyncMock(return_value=False)
|
||||
mock_apw.return_value = mock_cm
|
||||
|
||||
receipts = await scraper.scrape_receipts(valid_session)
|
||||
|
||||
# Both receipts should be returned — the first with empty detail
|
||||
assert len(receipts) == 2
|
||||
assert receipts[0].raw_data.get("detail") == {}
|
||||
assert receipts[1].receipt_id == "TXN-2"
|
||||
@@ -0,0 +1,364 @@
|
||||
"""Regression tests: scraper output matches expected schema.
|
||||
|
||||
Validates that parsed receipts from both Kroger and Meijer conform to the
|
||||
PurchaseCreate schema contract. Uses recorded fixtures to ensure outputs
|
||||
remain stable across code changes.
|
||||
"""
|
||||
|
||||
from decimal import Decimal
|
||||
|
||||
from receiptwitness.parsers.kroger import parse_kroger_receipt
|
||||
from receiptwitness.parsers.meijer import parse_meijer_receipt
|
||||
from receiptwitness.scrapers.base import RawReceipt
|
||||
|
||||
# Required top-level keys in a parsed receipt
|
||||
RECEIPT_REQUIRED_KEYS = {"receipt_id", "purchase_date", "total", "items", "raw_data"}
|
||||
RECEIPT_OPTIONAL_KEYS = {"subtotal", "tax", "savings_total", "source_url"}
|
||||
|
||||
# Required keys in each parsed item
|
||||
ITEM_REQUIRED_KEYS = {
|
||||
"product_name_raw",
|
||||
"upc",
|
||||
"quantity",
|
||||
"unit_price",
|
||||
"extended_price",
|
||||
}
|
||||
ITEM_OPTIONAL_KEYS = {
|
||||
"regular_price",
|
||||
"sale_price",
|
||||
"coupon_discount",
|
||||
"loyalty_discount",
|
||||
"category_raw",
|
||||
}
|
||||
|
||||
|
||||
def _validate_receipt_schema(result: dict) -> None:
|
||||
"""Assert that a parsed receipt dict conforms to the expected schema."""
|
||||
# All required keys present
|
||||
for key in RECEIPT_REQUIRED_KEYS:
|
||||
assert key in result, f"Missing required key: {key}"
|
||||
|
||||
# Types
|
||||
assert isinstance(result["receipt_id"], str)
|
||||
assert isinstance(result["purchase_date"], str)
|
||||
assert isinstance(result["total"], Decimal)
|
||||
assert isinstance(result["items"], list)
|
||||
assert isinstance(result["raw_data"], dict)
|
||||
|
||||
# Optional keys should be correct types when present
|
||||
if result.get("subtotal") is not None:
|
||||
assert isinstance(result["subtotal"], Decimal)
|
||||
if result.get("tax") is not None:
|
||||
assert isinstance(result["tax"], Decimal)
|
||||
if result.get("savings_total") is not None:
|
||||
assert isinstance(result["savings_total"], Decimal)
|
||||
if result.get("source_url") is not None:
|
||||
assert isinstance(result["source_url"], str)
|
||||
|
||||
# No unexpected keys
|
||||
all_keys = RECEIPT_REQUIRED_KEYS | RECEIPT_OPTIONAL_KEYS
|
||||
for key in result:
|
||||
assert key in all_keys, f"Unexpected key in receipt: {key}"
|
||||
|
||||
|
||||
def _validate_item_schema(item: dict) -> None:
|
||||
"""Assert that a parsed item dict conforms to the expected schema."""
|
||||
for key in ITEM_REQUIRED_KEYS:
|
||||
assert key in item, f"Missing required item key: {key}"
|
||||
|
||||
assert isinstance(item["product_name_raw"], str)
|
||||
assert len(item["product_name_raw"]) > 0
|
||||
assert isinstance(item["quantity"], Decimal)
|
||||
assert isinstance(item["unit_price"], Decimal)
|
||||
assert isinstance(item["extended_price"], Decimal)
|
||||
|
||||
# UPC can be None or str
|
||||
if item["upc"] is not None:
|
||||
assert isinstance(item["upc"], str)
|
||||
# UPC should not have leading zeros (stripped during parsing)
|
||||
assert not item["upc"].startswith("0"), f"UPC has leading zeros: {item['upc']}"
|
||||
|
||||
# Optional Decimal fields
|
||||
for opt_key in ("regular_price", "sale_price", "coupon_discount", "loyalty_discount"):
|
||||
if item.get(opt_key) is not None:
|
||||
assert isinstance(item[opt_key], Decimal), f"{opt_key} should be Decimal"
|
||||
|
||||
if item.get("category_raw") is not None:
|
||||
assert isinstance(item["category_raw"], str)
|
||||
|
||||
# No unexpected keys
|
||||
all_keys = ITEM_REQUIRED_KEYS | ITEM_OPTIONAL_KEYS
|
||||
for key in item:
|
||||
assert key in all_keys, f"Unexpected key in item: {key}"
|
||||
|
||||
|
||||
class TestKrogerSchemaValidation:
|
||||
def test_full_receipt_schema(self, kroger_receipt_data):
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-2026-0312-4471",
|
||||
purchase_date="2026-03-12T16:45:00Z",
|
||||
store_number="00357",
|
||||
raw_data=kroger_receipt_data,
|
||||
source_url="https://www.kroger.com/atlas/v1/receipt/api?orderId=KR-2026-0312-4471",
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
_validate_receipt_schema(result)
|
||||
for item in result["items"]:
|
||||
_validate_item_schema(item)
|
||||
|
||||
def test_item_count_excludes_voided_and_returned(self, kroger_receipt_data):
|
||||
"""Fixture has 10 items, 2 should be excluded (voided + returned)."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-2026-0312-4471",
|
||||
purchase_date="2026-03-12T16:45:00Z",
|
||||
raw_data=kroger_receipt_data,
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert len(result["items"]) == 8
|
||||
|
||||
def test_totals_are_positive_decimals(self, kroger_receipt_data):
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-2026-0312-4471",
|
||||
purchase_date="2026-03-12T16:45:00Z",
|
||||
raw_data=kroger_receipt_data,
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert result["total"] > Decimal("0")
|
||||
assert result["subtotal"] > Decimal("0")
|
||||
assert result["tax"] > Decimal("0")
|
||||
assert result["savings_total"] > Decimal("0")
|
||||
|
||||
def test_receipt_id_preserved(self, kroger_receipt_data):
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-2026-0312-4471",
|
||||
purchase_date="2026-03-12T16:45:00Z",
|
||||
raw_data=kroger_receipt_data,
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
assert result["receipt_id"] == "KR-2026-0312-4471"
|
||||
|
||||
def test_known_product_prices(self, kroger_receipt_data):
|
||||
"""Verify specific products produce correct price extraction."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-2026-0312-4471",
|
||||
purchase_date="2026-03-12T16:45:00Z",
|
||||
raw_data=kroger_receipt_data,
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
items_by_name = {i["product_name_raw"]: i for i in result["items"]}
|
||||
|
||||
# Milk: $3.99, regular $4.29
|
||||
milk = items_by_name["KROGER WHOLE MILK GAL"]
|
||||
assert milk["unit_price"] == Decimal("3.99")
|
||||
assert milk["regular_price"] == Decimal("4.29")
|
||||
assert milk["sale_price"] == Decimal("3.99")
|
||||
|
||||
# Eggs: qty 2, $5.49 each, total $10.98
|
||||
eggs = items_by_name["SIMPLE TRUTH ORG EGGS 12CT"]
|
||||
assert eggs["quantity"] == Decimal("2")
|
||||
assert eggs["unit_price"] == Decimal("5.49")
|
||||
assert eggs["extended_price"] == Decimal("10.98")
|
||||
|
||||
# Deli turkey: weighted item, 0.68 lb
|
||||
turkey = items_by_name["KROGER DELI TURKEY BREAST"]
|
||||
assert turkey["quantity"] == Decimal("0.68")
|
||||
assert turkey["upc"] is None
|
||||
|
||||
def test_multi_quantity_item_correct(self, kroger_receipt_data):
|
||||
"""Pasta is qty=3, unit=$2.49, total=$7.47."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-2026-0312-4471",
|
||||
purchase_date="2026-03-12T16:45:00Z",
|
||||
raw_data=kroger_receipt_data,
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
pasta = [i for i in result["items"] if "PASTA" in i["product_name_raw"]][0]
|
||||
assert pasta["quantity"] == Decimal("3")
|
||||
assert pasta["unit_price"] == Decimal("2.49")
|
||||
assert pasta["extended_price"] == Decimal("7.47")
|
||||
|
||||
def test_coupon_discount_captured(self, kroger_receipt_data):
|
||||
"""Tide Pods has $2.00 coupon."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-2026-0312-4471",
|
||||
purchase_date="2026-03-12T16:45:00Z",
|
||||
raw_data=kroger_receipt_data,
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
tide = [i for i in result["items"] if "TIDE" in i["product_name_raw"]][0]
|
||||
assert tide["coupon_discount"] == Decimal("2.00")
|
||||
|
||||
|
||||
class TestMeijerSchemaValidation:
|
||||
def test_full_receipt_schema(self, meijer_receipt_data):
|
||||
raw = RawReceipt(
|
||||
receipt_id="TXN-2026-0310-001",
|
||||
purchase_date="2026-03-10T14:30:00Z",
|
||||
store_number="42",
|
||||
raw_data=meijer_receipt_data,
|
||||
source_url="https://www.meijer.com/bin/meijer/profile/receipt?receiptId=TXN-2026-0310-001",
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
_validate_receipt_schema(result)
|
||||
for item in result["items"]:
|
||||
_validate_item_schema(item)
|
||||
|
||||
def test_item_count_excludes_voided(self, meijer_receipt_data):
|
||||
"""Fixture has 6 items, 1 should be excluded (voided soda)."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="TXN-2026-0310-001",
|
||||
purchase_date="2026-03-10T14:30:00Z",
|
||||
raw_data=meijer_receipt_data,
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert len(result["items"]) == 5
|
||||
|
||||
def test_totals_are_positive_decimals(self, meijer_receipt_data):
|
||||
raw = RawReceipt(
|
||||
receipt_id="TXN-2026-0310-001",
|
||||
purchase_date="2026-03-10T14:30:00Z",
|
||||
raw_data=meijer_receipt_data,
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["total"] > Decimal("0")
|
||||
assert result["subtotal"] > Decimal("0")
|
||||
assert result["tax"] > Decimal("0")
|
||||
assert result["savings_total"] > Decimal("0")
|
||||
|
||||
def test_receipt_id_preserved(self, meijer_receipt_data):
|
||||
raw = RawReceipt(
|
||||
receipt_id="TXN-2026-0310-001",
|
||||
purchase_date="2026-03-10T14:30:00Z",
|
||||
raw_data=meijer_receipt_data,
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
assert result["receipt_id"] == "TXN-2026-0310-001"
|
||||
|
||||
def test_known_product_prices(self, meijer_receipt_data):
|
||||
"""Verify specific Meijer products produce correct price extraction."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="TXN-2026-0310-001",
|
||||
purchase_date="2026-03-10T14:30:00Z",
|
||||
raw_data=meijer_receipt_data,
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
items_by_name = {i["product_name_raw"]: i for i in result["items"]}
|
||||
|
||||
# Bananas: $0.69
|
||||
bananas = items_by_name["ORGANIC BANANAS"]
|
||||
assert bananas["unit_price"] == Decimal("0.69")
|
||||
assert bananas["mperks_discount"] if "mperks_discount" in bananas else True
|
||||
assert bananas["loyalty_discount"] == Decimal("0.10")
|
||||
|
||||
# Milk: qty 2, $3.49 each, total $6.98
|
||||
milk = items_by_name["MEIJER 2% MILK GAL"]
|
||||
assert milk["quantity"] == Decimal("2")
|
||||
assert milk["unit_price"] == Decimal("3.49")
|
||||
assert milk["extended_price"] == Decimal("6.98")
|
||||
|
||||
# Weighted deli turkey: 0.75 lb at $8.99/lb
|
||||
turkey = items_by_name["WEIGHTED DELI TURKEY"]
|
||||
assert turkey["quantity"] == Decimal("0.75")
|
||||
assert turkey["upc"] is None
|
||||
|
||||
def test_mperks_discount_captured(self, meijer_receipt_data):
|
||||
"""Paper towels has $1.00 mPerks discount."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="TXN-2026-0310-001",
|
||||
purchase_date="2026-03-10T14:30:00Z",
|
||||
raw_data=meijer_receipt_data,
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
towels = [i for i in result["items"] if "PAPER TOWELS" in i["product_name_raw"]][0]
|
||||
assert towels["loyalty_discount"] == Decimal("1.00")
|
||||
assert towels["coupon_discount"] == Decimal("1.00")
|
||||
|
||||
def test_cheerios_coupon_discount(self, meijer_receipt_data):
|
||||
"""Cheerios has $0.50 coupon."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="TXN-2026-0310-001",
|
||||
purchase_date="2026-03-10T14:30:00Z",
|
||||
raw_data=meijer_receipt_data,
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
cheerios = [i for i in result["items"] if "CHEERIOS" in i["product_name_raw"]][0]
|
||||
assert cheerios["coupon_discount"] == Decimal("0.50")
|
||||
|
||||
|
||||
class TestEmptyAndEdgeCaseSchemas:
|
||||
"""Regression tests for edge-case receipts that should not crash."""
|
||||
|
||||
def test_kroger_empty_receipt(self):
|
||||
raw = RawReceipt(receipt_id="KR-EMPTY", purchase_date="2026-03-12", raw_data={})
|
||||
result = parse_kroger_receipt(raw)
|
||||
_validate_receipt_schema(result)
|
||||
assert result["items"] == []
|
||||
assert result["total"] == Decimal("0")
|
||||
|
||||
def test_meijer_empty_receipt(self):
|
||||
raw = RawReceipt(receipt_id="MJ-EMPTY", purchase_date="2026-03-10", raw_data={})
|
||||
result = parse_meijer_receipt(raw)
|
||||
_validate_receipt_schema(result)
|
||||
assert result["items"] == []
|
||||
assert result["total"] == Decimal("0")
|
||||
|
||||
def test_kroger_receipt_no_detail(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-NODET",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={"total": 50.00},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
_validate_receipt_schema(result)
|
||||
assert result["items"] == []
|
||||
assert result["total"] == Decimal("50.00")
|
||||
|
||||
def test_meijer_receipt_no_detail(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-NODET",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={"total": 30.00},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
_validate_receipt_schema(result)
|
||||
assert result["items"] == []
|
||||
assert result["total"] == Decimal("30.00")
|
||||
|
||||
def test_kroger_receipt_all_voided(self):
|
||||
"""A receipt where every item is voided should have 0 items."""
|
||||
raw = RawReceipt(
|
||||
receipt_id="KR-ALLVOID",
|
||||
purchase_date="2026-03-12",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [
|
||||
{"description": "VOIDED A", "basePrice": 5.0, "voided": True},
|
||||
{"description": "VOIDED B", "basePrice": 3.0, "status": "VOIDED"},
|
||||
{"description": "RETURNED C", "basePrice": 7.0, "status": "RETURNED"},
|
||||
{"description": "RETURNED D", "basePrice": 2.0, "returnFlag": True},
|
||||
],
|
||||
"total": 0,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_kroger_receipt(raw)
|
||||
_validate_receipt_schema(result)
|
||||
assert result["items"] == []
|
||||
|
||||
def test_meijer_receipt_all_voided(self):
|
||||
raw = RawReceipt(
|
||||
receipt_id="MJ-ALLVOID",
|
||||
purchase_date="2026-03-10",
|
||||
raw_data={
|
||||
"detail": {
|
||||
"items": [
|
||||
{"description": "VOIDED A", "price": 5.0, "voided": True},
|
||||
{"description": "VOIDED B", "price": 3.0, "status": "VOIDED"},
|
||||
],
|
||||
"total": 0,
|
||||
}
|
||||
},
|
||||
)
|
||||
result = parse_meijer_receipt(raw)
|
||||
_validate_receipt_schema(result)
|
||||
assert result["items"] == []
|
||||
Reference in New Issue
Block a user