From 063ac3a4fd9a74ef296b91552d4d5c923f9cd886 Mon Sep 17 00:00:00 2001
From: polyfloyd <floyd@polyfloyd.net>
Date: Sun, 22 Dec 2024 18:03:41 +0100
Subject: [PATCH] Fix AH scraping

---
 inflatinator/revbank.py       |  4 +--
 inflatinator/scrapers.py      | 47 ++++++++++++++++++-----------------
 inflatinator/scrapers_test.py | 10 ++++++++
 requirements.txt              |  1 +
 4 files changed, 37 insertions(+), 25 deletions(-)

diff --git a/inflatinator/revbank.py b/inflatinator/revbank.py
index 46bb3e3..5ddb534 100644
--- a/inflatinator/revbank.py
+++ b/inflatinator/revbank.py
@@ -67,8 +67,8 @@ class NoAutoUpdate(Exception):
 
 
 def find_product_details(product: Product):
-    if (ah_sku := product.metadata.get('ah', None)):
-        return scrapers.ah_get_by_sku(ah_sku, int(product.metadata['qty']))
+    if 'ah' in product.metadata:
+        return scrapers.ah_get_by_gtin(product.aliases[0])
     if 'sligro' in product.metadata:
         return scrapers.sligro_get_by_gtin(product.aliases[0])
     raise NoAutoUpdate()
diff --git a/inflatinator/scrapers.py b/inflatinator/scrapers.py
index 18609d3..e2b6652 100644
--- a/inflatinator/scrapers.py
+++ b/inflatinator/scrapers.py
@@ -1,3 +1,4 @@
+from dataclasses import dataclass
 from decimal import Decimal
 from functools import reduce
 from pyquery import PyQuery as pq
@@ -7,47 +8,47 @@ import os
 import requests
 import subprocess
 import logging
+from supermarktconnector.ah import AHConnector
+from typing import List
 
 
 vat = Decimal('1.09')
 
 
+@dataclass
 class Product:
-    def __init__(self, *, name, price, gtin, units, aliases=[]):
-        self.name = name
-        self.price = price
-        self.gtin = gtin
-        self.units = units
-        self.aliases = aliases
+    name: str
+    price: Decimal
+    gtin: str
+    units: int
+    aliases: List[str]
 
     def __str__(self):
         return self.name
 
 
-def links_get(url):
-    compl = subprocess.run(['links', '-source', url], capture_output=True)
-    return compl.stdout
+_ah = None
 
+def ah_get_by_gtin(gtin13):
+    assert re.match(r'^\d{13}$', gtin13)
 
-def ah_get_by_sku(ah_sku, units):
-    assert re.match(r'^wi\d+$', ah_sku)
+    global _ah
+    if not _ah:
+        _ah = AHConnector()
 
-    html_src = links_get(f'https://www.ah.nl/producten/product/{ah_sku}')
-    doc = pq(html_src)
+    ah_prod = _ah.get_product_by_barcode(gtin13)
 
-    ld_jsons = doc('script[type="application/ld+json"]')
-    for j in ld_jsons:
-        schema = json.loads(j.text)
-        if schema['@type'] == 'Product' and schema['sku'] == ah_sku:
-            break
-    else:
-        raise Exception(f'ah.nl returned no JSON metadata for SKU {ah_sku}')
+    units_description = ah_prod['salesUnitSize']
+    units = 1
+    if (m := re.search(r'^(\d+)', units_description)):
+        units = int(m[1])
 
     return Product(
-        name=schema['name'],
-        price=Decimal(schema['offers']['price']),
-        gtin=schema['gtin13'],
+        name=ah_prod['title'],
+        price=Decimal(ah_prod['priceBeforeBonus']),
+        gtin=gtin13,
         units=units,
+        aliases=[],
     )
 
 
diff --git a/inflatinator/scrapers_test.py b/inflatinator/scrapers_test.py
index 0f293f2..02fd1bc 100644
--- a/inflatinator/scrapers_test.py
+++ b/inflatinator/scrapers_test.py
@@ -1,6 +1,16 @@
 from scrapers import *
 
 
+def test_scrape_ah():
+    # Ola Liuk
+    prod = ah_get_by_gtin('8711327538481')
+    assert type(prod) is Product
+    assert prod.name == 'Ola Liuk'
+    assert prod.gtin == '8711327538481'
+    assert prod.units == 8
+    assert prod.aliases == []
+
+
 def test_scrape_sligro():
     # Cola zero sugar
     prod = sligro_get_by_gtin('5000112659184')
diff --git a/requirements.txt b/requirements.txt
index a8df116..68df88e 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,3 +1,4 @@
 pyquery
 pytest
 requests
+supermarktconnector