Merge pull request #1477 from sawanmsis/exclude_nsfw

Excluded NSFW sites by default.

sherlock/resources/data.json

@@ -294,7 +294,8 @@
     "url": "https://pt.bongacams.com/profile/{}",
     "urlMain": "https://pt.bongacams.com",
     "username_claimed": "asuna-black",
-    "username_unclaimed": "noonewouldeverusethis77777"
+    "username_unclaimed": "noonewouldeverusethis77777",
+    "isNSFW": true
   },
   "Bookcrossing": {
     "errorType": "status_code",
@@ -369,7 +370,8 @@
     "url": "https://chaturbate.com/{}",
     "urlMain": "https://chaturbate.com",
     "username_claimed": "cute18cute",
-    "username_unclaimed": "noonewouldeverusethis77777"
+    "username_unclaimed": "noonewouldeverusethis77777",
+    "isNSFW": true
   },
   "Chess": {
     "errorMsg": "Username is valid",
@@ -1385,7 +1387,8 @@
     "urlMain": "https://onlyfans.com/",
     "urlProbe": "https://onlyfans.com/api2/v2/users/{}",
     "username_claimed": "theemilylynne",
-    "username_unclaimed": "noonewouldeverusethis7"
+    "username_unclaimed": "noonewouldeverusethis7",
+    "isNSFW": true
   },
   "OpenStreetMap": {
     "errorType": "status_code",
@@ -1511,7 +1514,8 @@
     "url": "https://pornhub.com/users/{}",
     "urlMain": "https://pornhub.com/",
     "username_claimed": "blue",
-    "username_unclaimed": "noonewouldeverusethis7"
+    "username_unclaimed": "noonewouldeverusethis7",
+    "isNSFW": true
   },
   "ProductHunt": {
     "errorMsg": "We seem to have lost this page",
@@ -2117,8 +2121,8 @@
     "url": "https://en.wikipedia.org/wiki/Special:CentralAuth/{}?uselang=qqx",
     "urlMain": "https://www.wikipedia.org/",
     "username_claimed": "Hoadlck",
-    "username_unclaimed": "noonewouldeverusethis8"  
-   },
+    "username_unclaimed": "noonewouldeverusethis8"
+  },
   "Windy": {
     "errorType": "status_code",
     "url": "https://community.windy.com/user/{}",
@@ -2178,7 +2182,8 @@
     "url": "https://xvideos.com/profiles/{}",
     "urlMain": "https://xvideos.com/",
     "username_claimed": "blue",
-    "username_unclaimed": "noonewouldeverusethis7"
+    "username_unclaimed": "noonewouldeverusethis7",
+    "isNSFW": true
   },
   "YouNow": {
     "errorMsg": "No users found",
@@ -2201,7 +2206,8 @@
     "url": "https://youporn.com/uservids/{}",
     "urlMain": "https://youporn.com",
     "username_claimed": "blue",
-    "username_unclaimed": "noonewouldeverusethis77777"
+    "username_unclaimed": "noonewouldeverusethis77777",
+    "isNSFW": true
   },
   "Zhihu": {
     "errorMsg": "404",
@@ -2704,4 +2710,4 @@
     "username_claimed": "kossher",
     "username_unclaimed": "noonewouldeverusethis7"
   }
-}
+}

sherlock/sherlock.py

@@ -561,6 +561,10 @@ def main():
                         action="store_true", default=False,
                         help="Force the use of the local data.json file.")
 
+    parser.add_argument("--nsfw",
+                        action="store_true", default=False,
+                        help="Include checking of NSFW sites from default list. Default False")
+
     args = parser.parse_args()
     
     # If the user presses CTRL-C, exit gracefully without throwing errors
@@ -624,6 +628,9 @@ def main():
         print(f"ERROR:  {error}")
         sys.exit(1)
 
+    if not args.nsfw:
+        sites.remove_nsfw_sites()
+
     # Create original dictionary from SitesInformation() object.
     # Eventually, the rest of the code will be updated to use the new object
     # directly, but this will glue the two pieces together.

sherlock/sites.py

@@ -8,7 +8,7 @@ import requests
 
 class SiteInformation:
     def __init__(self, name, url_home, url_username_format, username_claimed,
-                 username_unclaimed, information):
+                 username_unclaimed, information, is_nsfw):
         """Create Site Information Object.
 
         Contains information about a specific website.
@@ -52,6 +52,7 @@ class SiteInformation:
         self.username_claimed = username_claimed
         self.username_unclaimed = username_unclaimed
         self.information = information
+        self.is_nsfw  = is_nsfw
 
         return
 
@@ -162,7 +163,9 @@ class SitesInformation:
                                     site_data[site_name]["url"],
                                     site_data[site_name]["username_claimed"],
                                     site_data[site_name]["username_unclaimed"],
-                                    site_data[site_name]
+                                    site_data[site_name],
+                                    site_data[site_name].get("isNSFW",False)
+
                                     )
             except KeyError as error:
                 raise ValueError(
@@ -171,6 +174,23 @@ class SitesInformation:
 
         return
 
+    def remove_nsfw_sites(self):
+        """
+        Remove NSFW sites from the sites, if isNSFW flag is true for site
+
+        Keyword Arguments:
+        self                   -- This object.
+
+        Return Value:
+        None
+        """
+        sites = {}
+        for site in self.sites:
+            if self.sites[site].is_nsfw:
+                continue
+            sites[site] = self.sites[site]  
+        self.sites =  sites
+
     def site_name_list(self):
         """Get Site Name List.