Update Ozon-parser.py

This commit is contained in:
Igor Volochay
2024-07-16 14:27:56 +03:00
committed by GitHub
parent ad2663bd3c
commit 6a5ffd22e1
+9 -3
View File
@@ -41,9 +41,11 @@ def get_product_info(product_url):
image_url = json.loads(json_data["seo"]["script"][0]["innerHTML"])["image"] image_url = json.loads(json_data["seo"]["script"][0]["innerHTML"])["image"]
price = json.loads(json_data["seo"]["script"][0]["innerHTML"])["offers"]["price"] + " " +\ price = json.loads(json_data["seo"]["script"][0]["innerHTML"])["offers"]["price"] + " " +\
json.loads(json_data["seo"]["script"][0]["innerHTML"])["offers"]["priceCurrency"] json.loads(json_data["seo"]["script"][0]["innerHTML"])["offers"]["priceCurrency"]
rating = json.loads(json_data["seo"]["script"][0]["innerHTML"]["ratingValue"])
rating_counter = json.loads(json_data["seo"]["script"][0]["innerHTML"]["reviewCount"])
product_id = json.loads(json_data["seo"]["script"][0]["innerHTML"])["sku"] product_id = json.loads(json_data["seo"]["script"][0]["innerHTML"])["sku"]
return (product_id, full_name, description, price, image_url) return (product_id, full_name, description, price, rating, rating_counter, image_url)
def get_mainpage_cards(driver, url): def get_mainpage_cards(driver, url):
driver.get(url) driver.get(url)
@@ -70,11 +72,13 @@ def get_mainpage_cards(driver, url):
card_url = card[2].find("a", href=True)["href"] card_url = card[2].find("a", href=True)["href"]
product_url = "https://ozon.ru/" + card_url product_url = "https://ozon.ru/" + card_url
product_id, full_name, description, price, image_url = get_product_info(card_url) product_id, full_name, description, price, rating, rating_counter, image_url = get_product_info(card_url)
card_info = {product_id: {"short_name": card_name, card_info = {product_id: {"short_name": card_name,
"full_name": full_name, "full_name": full_name,
"description": description, "description": description,
"url": product_url, "url": product_url,
"rating": rating,
"rating_counter": rating_counter,
"price": price, "price": price,
"image_url": image_url "image_url": image_url
} }
@@ -103,11 +107,13 @@ def get_searchpage_cards(driver, url, all_cards = []):
product_url = "https://ozon.ru/" + card_url product_url = "https://ozon.ru/" + card_url
product_id, full_name, description, price, image_url = get_product_info(card_url) product_id, full_name, description, price, rating, rating_counter, image_url = get_product_info(card_url)
card_info = {product_id: {"short_name": card_name, card_info = {product_id: {"short_name": card_name,
"full_name": full_name, "full_name": full_name,
"description": description, "description": description,
"url": product_url, "url": product_url,
"rating": rating,
"rating_counter": rating_counter,
"price": price, "price": price,
"image_url": image_url "image_url": image_url
} }