|
|
|
@ -15,25 +15,28 @@ import re |
|
|
|
def nexus_description_parser(soup): |
|
|
|
|
|
|
|
# Fields to be parsed |
|
|
|
vendor = "-1" # 0 *Vendor_Name |
|
|
|
success = "-1" # 1 Vendor_Successful_Transactions |
|
|
|
rating_vendor = "-1" # 2 Vendor_Rating |
|
|
|
name = "-1" # 3 *Product_Name |
|
|
|
describe = "-1" # 4 Product_Description |
|
|
|
CVE = "-1" # 5 Product_CVE_Classification (Common Vulnerabilities and Exposures) dont worry about that much |
|
|
|
MS = "-1" # 6 Product_MS_Classification (Microsoft Security) dont worry about that much |
|
|
|
category = "-1" # 7 Product_Category |
|
|
|
views = "-1" # 8 Product_Number_Of_Views |
|
|
|
reviews = "-1" # 9 Product_Number_Of_Reviews |
|
|
|
rating_item = "-1" # 10 Product_Rating |
|
|
|
addDate = "-1" # 11 Product_AddedDate |
|
|
|
BTC = "-1" # 12 Product_BTC_SellingPrice |
|
|
|
USD = "-1" # 13 Product_USD_SellingPrice |
|
|
|
EURO = "-1" # 14 Product_EURO_SellingPrice |
|
|
|
sold = "-1" # 15 Product_QuantitySold |
|
|
|
left = "-1" # 16 Product_QuantityLeft |
|
|
|
shipFrom = "-1" # 17 Product_ShippedFrom |
|
|
|
shipTo = "-1" # 18 Product_ShippedTo |
|
|
|
|
|
|
|
vendor = "-1" # 0 *Vendor_Name |
|
|
|
success = "-1" # 1 Vendor_Successful_Transactions |
|
|
|
rating_vendor = "-1" # 2 Vendor_Rating |
|
|
|
name = "-1" # 3 *Product_Name |
|
|
|
describe = "-1" # 4 Product_Description |
|
|
|
CVE = "-1" # 5 Product_CVE_Classification (Common Vulnerabilities and Exposures) |
|
|
|
MS = "-1" # 6 Product_MS_Classification (Microsoft Security) |
|
|
|
category = "-1" # 7 Product_Category |
|
|
|
views = "-1" # 8 Product_Number_Of_Views |
|
|
|
reviews = "-1" # 9 Product_Number_Of_Reviews |
|
|
|
rating_item = "-1" # 10 Product_Rating |
|
|
|
addDate = "-1" # 11 Product_AddedDate |
|
|
|
BTC = "-1" # 12 Product_BTC_SellingPrice |
|
|
|
USD = "-1" # 13 Product_USD_SellingPrice |
|
|
|
EURO = "-1" # 14 Product_EURO_SellingPrice |
|
|
|
sold = "-1" # 15 Product_QuantitySold |
|
|
|
left = "-1" # 16 Product_QuantityLeft |
|
|
|
shipFrom = "-1" # 17 Product_ShippedFrom |
|
|
|
shipTo = "-1" # 18 Product_ShippedTo |
|
|
|
image = "-1" # 19 Product_Image |
|
|
|
vendor_image = "-1" # 20 Vendor_Image |
|
|
|
|
|
|
|
|
|
|
|
#finding the name of the product |
|
|
|
@ -48,6 +51,10 @@ def nexus_description_parser(soup): |
|
|
|
else: |
|
|
|
describe = cleanString(description_div.text.strip()) |
|
|
|
|
|
|
|
# Finding Product Image |
|
|
|
image = soup.find('div', {'class': 'woocommerce-product-gallery__image'}).find('img') |
|
|
|
image = image.get('src') |
|
|
|
image = image.split('base64,')[-1] |
|
|
|
|
|
|
|
#find the category of the product |
|
|
|
name_of_category = soup.find("span", {"class": "posted_in"}).find("a").text |
|
|
|
@ -64,7 +71,7 @@ def nexus_description_parser(soup): |
|
|
|
|
|
|
|
# Populating the final variable (this should be a list with all fields scraped) |
|
|
|
row = (vendor, rating_vendor, success, name, describe, CVE, MS, category, views, reviews, rating_item, addDate, |
|
|
|
BTC, USD, EURO, sold, left, shipFrom, shipTo) |
|
|
|
BTC, USD, EURO, sold, left, shipFrom, shipTo, image, vendor_image) |
|
|
|
|
|
|
|
|
|
|
|
# Sending the results |
|
|
|
@ -78,28 +85,31 @@ def nexus_description_parser(soup): |
|
|
|
def nexus_listing_parser(soup): |
|
|
|
|
|
|
|
# Fields to be parsed |
|
|
|
nm = 0 # *Total_Products (Should be Integer) |
|
|
|
mktName = "Nexus" # 0 *Marketplace_Name |
|
|
|
vendor = [] # 1 *Vendor y |
|
|
|
rating_vendor = [] # 2 Vendor_Rating |
|
|
|
success = [] # 3 Vendor_Successful_Transactions |
|
|
|
name = [] # 4 *Product_Name y |
|
|
|
CVE = [] # 5 Product_CVE_Classification (Common Vulnerabilities and Exposures) |
|
|
|
MS = [] # 6 Product_MS_Classification (Microsoft Security) |
|
|
|
category = [] # 7 Product_Category y |
|
|
|
describe = [] # 8 Product_Description |
|
|
|
views = [] # 9 Product_Number_Of_Views |
|
|
|
reviews = [] # 10 Product_Number_Of_Reviews |
|
|
|
rating_item = [] # 11 Product_Rating |
|
|
|
addDate = [] # 12 Product_AddDate |
|
|
|
BTC = [] # 13 Product_BTC_SellingPrice |
|
|
|
USD = [] # 14 Product_USD_SellingPrice y |
|
|
|
EURO = [] # 15 Product_EURO_SellingPrice |
|
|
|
sold = [] # 16 Product_QuantitySold |
|
|
|
qLeft = [] # 17 Product_QuantityLeft |
|
|
|
shipFrom = [] # 18 Product_ShippedFrom |
|
|
|
shipTo = [] # 19 Product_ShippedTo |
|
|
|
href = [] # 20 Product_Links |
|
|
|
nm = 0 # *Total_Products (Should be Integer) |
|
|
|
mktName = "Nexus" # 0 *Marketplace_Name |
|
|
|
vendor = [] # 1 *Vendor y |
|
|
|
rating_vendor = [] # 2 Vendor_Rating |
|
|
|
success = [] # 3 Vendor_Successful_Transactions |
|
|
|
name = [] # 4 *Product_Name y |
|
|
|
CVE = [] # 5 Product_CVE_Classification (Common Vulnerabilities and Exposures) |
|
|
|
MS = [] # 6 Product_MS_Classification (Microsoft Security) |
|
|
|
category = [] # 7 Product_Category y |
|
|
|
describe = [] # 8 Product_Description |
|
|
|
views = [] # 9 Product_Number_Of_Views |
|
|
|
reviews = [] # 10 Product_Number_Of_Reviews |
|
|
|
rating_item = [] # 11 Product_Rating |
|
|
|
addDate = [] # 12 Product_AddDate |
|
|
|
BTC = [] # 13 Product_BTC_SellingPrice |
|
|
|
USD = [] # 14 Product_USD_SellingPrice y |
|
|
|
EURO = [] # 15 Product_EURO_SellingPrice |
|
|
|
sold = [] # 16 Product_QuantitySold |
|
|
|
qLeft =[] # 17 Product_QuantityLeft |
|
|
|
shipFrom = [] # 18 Product_ShippedFrom |
|
|
|
shipTo = [] # 19 Product_ShippedTo |
|
|
|
image = [] # 20 Product_Image |
|
|
|
image_vendor = [] # 21 Vendor_Image |
|
|
|
href = [] # 22 Product_Links |
|
|
|
|
|
|
|
products_list = soup.find_all('li') |
|
|
|
nm = 0 |
|
|
|
for product in products_list: |
|
|
|
@ -117,10 +127,18 @@ def nexus_listing_parser(soup): |
|
|
|
print("I can't find the link") |
|
|
|
raise e |
|
|
|
|
|
|
|
# Finding Product Image |
|
|
|
product_image = product.find('a', {'class': 'woocommerce-loop-image-link woocommerce-LoopProduct-link woocommerce-loop-product__link'}).find('img') |
|
|
|
product_image = product_image.get('src') |
|
|
|
product_image = product_image.split('base64,')[-1] |
|
|
|
image.append(product_image) |
|
|
|
|
|
|
|
BTC.append("-1") |
|
|
|
|
|
|
|
#everything else appends a -1 |
|
|
|
rating_vendor.append("-1") |
|
|
|
USD.append("-1") |
|
|
|
vendor.append("-1") |
|
|
|
vendor.append(mktName) |
|
|
|
success.append("-1") |
|
|
|
CVE.append("-1") |
|
|
|
MS.append("-1") |
|
|
|
@ -129,12 +147,12 @@ def nexus_listing_parser(soup): |
|
|
|
views.append("-1") |
|
|
|
reviews.append("-1") |
|
|
|
addDate.append("-1") |
|
|
|
BTC.append("-1") |
|
|
|
EURO.append("-1") |
|
|
|
sold.append("-1") |
|
|
|
qLeft.append("-1") |
|
|
|
shipFrom.append("-1") |
|
|
|
shipTo.append("-1") |
|
|
|
image_vendor.append("-1") |
|
|
|
# print("Done! moving onto the next product!") |
|
|
|
# print(len(shipTo)) |
|
|
|
nm += 1 |
|
|
|
@ -145,7 +163,7 @@ def nexus_listing_parser(soup): |
|
|
|
|
|
|
|
# Populate the final variable (this should be a list with all fields scraped) |
|
|
|
return organizeProducts( |
|
|
|
marketplace = "Nexus", |
|
|
|
marketplace = mktName, |
|
|
|
nm = nm, |
|
|
|
vendor = vendor, |
|
|
|
rating_vendor = rating_vendor, |
|
|
|
@ -166,7 +184,9 @@ def nexus_listing_parser(soup): |
|
|
|
qLeft = qLeft, |
|
|
|
shipFrom = shipFrom, |
|
|
|
shipTo = shipTo, |
|
|
|
href = href |
|
|
|
href = href, |
|
|
|
image = image, |
|
|
|
image_vendor = image_vendor |
|
|
|
) |
|
|
|
|
|
|
|
|
|
|
|
|
ericssonmarin-cpp
1 year ago