update sounds fetch

src/app/vinyl/sounds/fetch.py

@@ -24,28 +24,54 @@ def get_page_count(html_content):
 
 
 def parse_page(html_content):
+    entries = []
     soup = BeautifulSoup(html_content, "html.parser")
-
+    for product in soup.find_all("div", {"class": "search-product"}):
-    # Extract the name (artist - album) from the h5 tag
+        item_id = product.find("a", rel=True)["rel"][0]
-    names = list(map(lambda x: x.get_text(strip=True), soup.find_all("h5")))
+        name = product.find("h5").text.strip()
-
+        artist_title = name.split("-")
-    # Remove 'Telefoon', 'E-mail', 'Facebook'
+        artist = artist_title[0].strip()
-    names = list(filter(lambda x: " -" in x, names))
+        title = artist_title[1].strip()
-
+        price = (
-    # Extract the numerical id from the a tag
+            product.find("span", class_="product-price")
-    ids = list(map(lambda x: x["rel"][0], soup.find_all("a", rel=True)))
+            .text.strip()
-
+            .replace("€", "")
-    # Extract the price
+            .strip()
-    prices = list(
-        map(
-            lambda x: float(x.get_text(strip=True).split()[1]),
-            soup.find_all("span", class_="product-price"),
-        )
         )
 
-    df = pd.DataFrame({"id": ids, "name": names, "price": prices})
+        entry = {
+            "id": item_id,
+            "name": name,
+            "artist": artist,
+            "title": title,
+            "price": price,
+        }
+        if detail := product.find("h6", {"class": "hide-for-small"}):
+            entry["detail"] = detail.text
+        if supply := product.find("div", {"class": "product-voorraad"}):
+            entry["supply"] = supply.text
 
-    return df
+        for info in product.find_all("div", {"class": "product-info"}):
+            info = info.text.split(":")
+            if "Genre" in info[0]:
+                entry["genre"] = info[1].strip()
+            if "Releasedatum" in info[0]:
+                entry["release"] = info[1].strip()
+            entries.append(entry)
+
+    return pd.DataFrame(entries).reindex(
+        columns=[
+            "id",
+            "name",
+            "artist",
+            "title",
+            "price",
+            "supply",
+            "release",
+            "genre",
+            "detail",
+        ]
+    )
 
 
 def fetch_deals():