From c6589f4b822cbaea4aa5551b0d708a381413097d Mon Sep 17 00:00:00 2001 From: ericssonmarin-cpp <85146518+ericssonmarin-cpp@users.noreply.github.com> Date: Fri, 8 Sep 2023 11:51:54 -0700 Subject: [PATCH] Image tracking and version. --- MarketPlaces/AnonMarket/parser.py | 101 +++++++++++--------- MarketPlaces/DB_Connection/db_connection.py | 88 ++++++++++++----- 2 files changed, 122 insertions(+), 67 deletions(-) diff --git a/MarketPlaces/AnonMarket/parser.py b/MarketPlaces/AnonMarket/parser.py index c53283c..5b94e75 100644 --- a/MarketPlaces/AnonMarket/parser.py +++ b/MarketPlaces/AnonMarket/parser.py @@ -14,26 +14,27 @@ import re #return: 'row' that contains a variety of lists that each hold info on the description page def AnonMarket_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 name_of_product = soup.find("div", {"class": "heading"}).text name = cleanString(name_of_product.strip()) @@ -44,6 +45,11 @@ def AnonMarket_description_parser(soup): else: describe = cleanString(description_div.text.strip()) + # Finding Product Image + image = soup.find('div', {'class': 'thumbnails'}).find('img') + image = image.get('src') + image = image.split('base64,')[-1] + info_div = soup.find('div', {'class': 'information'}) table = info_div.find('table') if info_div else None @@ -70,7 +76,7 @@ def AnonMarket_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 @@ -84,28 +90,31 @@ def AnonMarket_description_parser(soup): def AnonMarket_listing_parser(soup): # Fields to be parsed - nm = 0 # *Total_Products (Should be Integer) - mktName = "AnonMarket" # 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 = "AnonMarket" # 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 + base_url = "http://2r7wa5og3ly4umqhmmqqytae6bufl5ql5kz7sorndpqtrkc2ri7tohad.onion" products_list = soup.find_all('div', {'class': 'item'}) @@ -147,6 +156,8 @@ def AnonMarket_listing_parser(soup): qLeft.append("-1") shipFrom.append("-1") shipTo.append("-1") + image.append("-1") + image_vendor.append("-1") nm += 1 @@ -177,7 +188,9 @@ def AnonMarket_listing_parser(soup): qLeft = qLeft, shipFrom = shipFrom, shipTo = shipTo, - href = href + href = href, + image = image, + image_vendor = image_vendor ) diff --git a/MarketPlaces/DB_Connection/db_connection.py b/MarketPlaces/DB_Connection/db_connection.py index eb4d996..48cbc4d 100644 --- a/MarketPlaces/DB_Connection/db_connection.py +++ b/MarketPlaces/DB_Connection/db_connection.py @@ -119,6 +119,24 @@ def getLastVendor(cur): trace = traceback.format_exc() print (trace) +def getLastVendorVersion(cur, vendorId): + + try: + + cur.execute("select version_vendor from vendors_history where vendor_id = %(vendorId)s order by version_vendor desc limit 1", {'vendorId': vendorId}) + + recset = cur.fetchall() + + if recset: + return recset[0][0] + else: + return 0 + + except: + + trace = traceback.format_exc() + print (trace) + def getLastItem(cur): try: @@ -137,6 +155,25 @@ def getLastItem(cur): trace = traceback.format_exc() print (trace) + +def getLastItemVersion(cur, itemId): + try: + + cur.execute("select version_item from items_history where item_id = %(itemId)s order by version_item desc limit 1", + {'itemId': itemId}) + + recset = cur.fetchall() + + if recset: + return recset[0][0] + else: + return 0 + + except: + + trace = traceback.format_exc() + print(trace) + def create_marketPlace(cur, row, url): marketId = verifyMarketPlace(cur, row[0]) @@ -192,10 +229,13 @@ def create_vendor(cur, row, marketId): str(recset[0][4]) != str(row[3] if row[3] != '-1' else None) or str(recset[0][5]) != str(row[21] if row[21] != '-1' else None)): - sql = "Insert into vendors_history (vendor_id, market_id, name_vendor, rating_vendor, successfultransactions_vendor, image_vendor, dateinserted_vendor) " \ - "Values (%s, %s, %s, %s, %s, %s, %s)" + vendorVersionId = int(getLastVendorVersion(cur, vendorId) + 1) + + sql = "Insert into vendors_history (vendor_id, version_vendor, market_id, name_vendor, rating_vendor, " \ + "successfultransactions_vendor, image_vendor, dateinserted_vendor) " \ + "Values (%s, %s, %s, %s, %s, %s, %s, %s)" - recset = [vendorId, marketId, + recset = [vendorId, vendorVersionId, marketId, recset[0][2], recset[0][3], recset[0][4], @@ -277,13 +317,15 @@ def create_items(cur, row, marketId, vendorId): str(recset[0][16]) != str(row[17] if row[17] != '-1' else None) or str(recset[0][17]) != str(row[18] if row[18] != '-1' else None) or str(recset[0][18]) != str(row[19] if row[19] != '-1' else None) or str(recset[0][20]) != str(row[20] if row[20] != '-1' else None)): - sql = "Insert into items_history (item_id, market_id, vendor_id, name_item, description_item, cve_item, ms_item, category_item, " \ + itemVersionId = int(getLastItemVersion(cur, itemId) + 1) + + sql = "Insert into items_history (item_id, version_item, market_id, vendor_id, name_item, description_item, cve_item, ms_item, category_item, " \ "views_item, reviews_item, rating_item, dateadded_item, btc_item, usd_item, euro_item, quantitysold_item, " \ "quantityleft_item, shippedfrom_item, shippedto_item, lastseen_item, image_item, href_item, dateinserted_item, " \ "classification_item) Values (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, " \ - "%s, %s, %s, %s, %s)" + "%s, %s, %s, %s, %s, %s)" - recset = [itemId, marketId, vendorId, + recset = [itemId, itemVersionId, marketId, vendorId, recset[0][3], recset[0][4], recset[0][5], @@ -306,16 +348,16 @@ def create_items(cur, row, marketId, vendorId): recset[0][22], recset[0][23]] - cur.execute(sql, recset) + cur.execute(sql, recset) - sql = "Update items set description_item = %(description_item)s, cve_item = %(cve_item)s, ms_item = %(ms_item)s, " \ - "category_item = %(category_item)s, views_item = %(views_item)s, reviews_item = %(reviews_item)s, " \ - "rating_item = %(rating_item)s, dateadded_item = %(dateadded_item)s, btc_item = %(btc_item)s, " \ - "usd_item = %(usd_item)s, euro_item = %(euro_item)s, quantitysold_item = %(quantitysold_item)s, " \ - "quantityleft_item = %(quantityleft_item)s, shippedfrom_item = %(shippedfrom_item)s, shippedto_item = %(shippedto_item)s, " \ - "lastseen_item = %(lastseen_item)s, image_item = %(image_item)s, dateinserted_item = %(dateinserted_item)s where item_id = %(itemId)s" + sql = "Update items set description_item = %(description_item)s, cve_item = %(cve_item)s, ms_item = %(ms_item)s, " \ + "category_item = %(category_item)s, views_item = %(views_item)s, reviews_item = %(reviews_item)s, " \ + "rating_item = %(rating_item)s, dateadded_item = %(dateadded_item)s, btc_item = %(btc_item)s, " \ + "usd_item = %(usd_item)s, euro_item = %(euro_item)s, quantitysold_item = %(quantitysold_item)s, " \ + "quantityleft_item = %(quantityleft_item)s, shippedfrom_item = %(shippedfrom_item)s, shippedto_item = %(shippedto_item)s, " \ + "lastseen_item = %(lastseen_item)s, image_item = %(image_item)s, dateinserted_item = %(dateinserted_item)s where item_id = %(itemId)s" - cur.execute(sql, {'description_item': row[5] if row[5] != '-1' else None, + cur.execute(sql, {'description_item': row[5] if row[5] != '-1' else None, 'cve_item': row[6] if row[6] != '-1' else None, 'ms_item': row[7] if row[7] != '-1' else None, 'category_item': row[8] if row[8] != '-1' else None, @@ -338,10 +380,10 @@ def create_items(cur, row, marketId, vendorId): else: #updating when was the last time the crawler saw that item - sql = "Update items set lastseen_item = %(lastseen_item)s where item_id = %(itemId)s" + sql = "Update items set lastseen_item = %(lastseen_item)s where item_id = %(itemId)s" - cur.execute(sql, {'lastseen_item': row[23], - 'itemId': itemId}) + cur.execute(sql, {'lastseen_item': row[23], + 'itemId': itemId}) return itemId @@ -367,10 +409,10 @@ def create_database(cur, con): sql = "create unique index unique_vendor ON vendors USING btree (market_id ASC NULLS LAST, name_vendor ASC NULLS LAST)" cur.execute(sql) - sql = "create table vendors_history(vendor_id integer not null, market_id integer not null, name_vendor " \ + sql = "create table vendors_history(vendor_id integer not null, version_vendor integer not null, market_id integer not null, name_vendor " \ "character varying(255) not null, rating_vendor character varying(255), successfultransactions_vendor " \ "integer null, image_vendor character varying(1000000) null, dateinserted_vendor timestamp(6) with time zone not null, " \ - "constraint vendors_history_pk primary key (vendor_id, dateinserted_vendor), constraint vendors_history_vendor_id_fkey foreign key (" \ + "constraint vendors_history_pk primary key (vendor_id, version_vendor), constraint vendors_history_vendor_id_fkey foreign key (" \ "vendor_id) references vendors (vendor_id), constraint vendors_history_market_id_fkey foreign key (" \ "market_id) references marketplaces (market_id))" cur.execute(sql) @@ -392,16 +434,16 @@ def create_database(cur, con): sql = "create unique index unique_item ON items USING btree (market_id ASC NULLS LAST, vendor_id ASC NULLS LAST, name_item ASC NULLS LAST)" cur.execute(sql) - sql = "create table items_history(item_id integer not null, market_id integer not null, vendor_id integer not null, name_item character " \ - "varying(255) not null, description_item character varying(1000000) null, cve_item character varying(" \ - "255) null, ms_item character varying(255) null, category_item character varying(255) null, views_item " \ + sql = "create table items_history(item_id integer not null, version_item integer not null, market_id integer not null, " \ + "vendor_id integer not null, name_item character varying(255) not null, description_item character varying(1000000) null, " \ + "cve_item character varying(255) null, ms_item character varying(255) null, category_item character varying(255) null, views_item " \ "integer null, reviews_item integer null, rating_item character varying(255) null, dateadded_item " \ "character varying(25) null, btc_item character varying(255) null, usd_item character varying(255) " \ "null, euro_item character varying(255) null, quantitysold_item integer null, quantityleft_item " \ "character varying(255) null, shippedfrom_item character varying(255) null, shippedto_item character " \ "varying(255) null, lastseen_item timestamp(6) with time zone not null, image_item character varying(1000000) null, " \ "href_item character varying(255) not null, dateinserted_item timestamp(6) with time zone not null, " \ - "classification_item double precision not null, constraint items_history_pk primary key (item_id, dateinserted_item), " \ + "classification_item double precision not null, constraint items_history_pk primary key (item_id, version_item), " \ "constraint items_history_market_id_fkey foreign key (market_id) references marketplaces (market_id), " \ "constraint items_history_vendor_id_fkey foreign key (vendor_id) references vendors (vendor_id), " \ "constraint items_history_item_id_fkey foreign key (item_id) references items (item_id))"