Merge pull request #258 from Crinibus/fix/amazon-domain

Remove characters other than digits, commas (,) and periods (.) when getting price for Amazon

scraper/domains.py

@@ -254,9 +254,8 @@ def _get_product_name(self) -> str:
         return self.request_data.find("span", id="productTitle").text.strip()
 
     def _get_product_price(self) -> float:
-        return float(
-            self.request_data.find("span", class_="a-price").span.text.replace("$", "").replace(",", "").replace(" ", "")
-        )
+        raw_price = self.request_data.find("span", class_="a-price").span.text.replace(",", "").replace(" ", "")
+        return float(get_number_string(raw_price))
 
     def _get_product_currency(self) -> str:
         regex_pattern = "%22currencyCode%22%3A%22(.{3})%22"
@@ -548,6 +547,13 @@ def get_website_handler(url: str) -> BaseWebsiteHandler:
     return website_handler(url)
 
 
+def get_number_string(value: str) -> str:
+    """Return string with only digits, commas (,) and periods (.)"""
+    text_pattern = re.compile(r"[^\d.,]+")
+    result = text_pattern.sub("", value)
+    return result
+
+
 SUPPORTED_DOMAINS: dict[str, BaseWebsiteHandler] = {
     "komplett": KomplettHandler,
     "proshop": ProshopHandler,

tests/test_domains.py

@@ -1,7 +1,7 @@
 from dataclasses import dataclass
 import pytest
 
-from scraper.domains import get_website_name
+from scraper.domains import get_website_name, get_number_string
 
 
 @dataclass
@@ -42,3 +42,24 @@ def test_get_website_name(url: str, setting: UrlSetting, expected: str) -> None:
         keep_subdomain=setting.keep_subdomain,
     )
     assert result == expected
+
+
+test_price_values = [
+    ("USD 12.40", "12.40"),
+    ("$234.00", "234.00"),
+    ("£345.37", "345.37"),
+    ("486,89 kr", "486,89"),
+    ("$345.37", "345.37"),
+    ("£1345.37", "1345.37"),
+    ("1345,37 DKK", "1345,37"),
+    ("1345.37 DKK", "1345.37"),
+    ("USD 1345.37", "1345.37"),
+    ("USD 10345.37", "10345.37"),
+]
+
+
+@pytest.mark.parametrize("value,expected", test_price_values)
+def test_get_number_string(value: str, expected: str) -> None:
+    result = get_number_string(value)
+
+    assert result == expected

tests/test_objects.json

@@ -37,10 +37,10 @@
       "expected_currency": "DKK"
     },
     "amazon": {
-      "link": "https://www.amazon.com/Sony-WH-1000XM5-Canceling-Headphones-Hands-Free/dp/B09XS7JWHH",
-      "expected_title": "Sony WH-1000XM5 The Best Wireless Noise Canceling Headphones with Auto Noise Canceling Optimizer, Crystal Clear Hands-Free Calling, and Alexa Voice Control, Black",
-      "expected_id": "B09XS7JWHH",
-      "expected_currency": "USD"
+      "link": "https://www.amazon.de/-/en/Google-Pixel-Pro-Smartphone-Obsidian/dp/B0DG9DD9VN",
+      "expected_title": "Google Pixel 9 Pro (512GB, Obsi, EU / UK) + Pixel 9/9 Pro Case, Obsidian",
+      "expected_id": "B0DG9DD9VN",
+      "expected_currency": "EUR"
     },
     "ebay_with_itm": {
       "link": "https://www.ebay.com/itm/265771092654",