Account for missing contentDescription in Tjendrawasih Kip...

inflatinator/scrapers.py

@@ -103,17 +103,10 @@ def sligro_get_by_gtin(gtin13):
     product = body['products'][0]
     sku = product["code"]
 
-    # Query the product page itself, there is more info that we need on there. The 'url' field in
-    # the product object gives a 404, but the actual product page URL can be created from the search
-    # results.
-    url_slug = '-'.join([product['brandName'], product['name'], product['contentDescription']])\
-        .replace(' ', '-')\
-        .replace('\'', '-')\
-        .replace('&', '-')\
-        .replace(',', '')\
-        .replace('%', '')\
-        .lower()
-    prod_resp = requests.get(f'https://www.sligro.nl/p.{sku}.html/{url_slug}.html')
+    # Query the product page itself, there is more info that we need on there. In the website, the
+    # final path element is a derivation of the contentDescription field. It must be present, but
+    # matches anything.
+    prod_resp = requests.get(f'https://www.sligro.nl/p.{sku}.html/product.html')
     prod_resp.raise_for_status()
 
     product_page = pq(prod_resp.text)
@@ -125,8 +118,7 @@ def sligro_get_by_gtin(gtin13):
     if sub_gtin:
         sub_gtin = sub_gtin.lstrip('0')
 
-    # The contentDescription field holds the number of individual packages per box sold.
-    units, volume = parse_content_description(product['contentDescription'])
+    units, volume = get_packaging_info(product)
 
     # Pricing requires logging in and is on a separate endpoint...
     pricing_resp = sligro_client().get(f'https://www.sligro.nl/api/cart/sligro-nl/customerorganizationdatas?productCodes={sku}')
@@ -149,6 +141,13 @@ def sligro_get_by_gtin(gtin13):
     )
 
 
+def get_packaging_info(product):
+    if product['gtin'] == '08712641001903':  # Tjendrawasih Bapao kip: no contentDescription field?
+        return 12, '120 gram'
+    # The contentDescription field holds the number of individual packages per box sold.
+    return parse_content_description(product['contentDescription'])
+
+
 # The contentDescription seems to have a formatting consistent enough for regex matching. Some
 # products have multiple levels of packaging, but the last or only component is always the
 # volume or weight.