Skip images with certain urls (#213)

* Add flag to ignore certain urls

* Updated README and parser arguments

* Fix error on missing -iu option

Resolves issues related to pull request #213.
Removed duplicated validation

Author: crainbows

google_images_download/google_images_download.py

@@ -40,7 +40,7 @@
              "output_directory", "image_directory", "no_directory", "proxy", "similar_images", "specific_site",
              "print_urls", "print_size", "print_paths", "metadata", "extract_metadata", "socket_timeout",
              "thumbnail", "thumbnail_only", "language", "prefix", "chromedriver", "related_images", "safe_search", "no_numbering",
-             "offset", "no_download","save_source","silent_mode"]
+             "offset", "no_download","save_source","silent_mode","ignore_urls"]
 
 
 def user_input():
@@ -112,6 +112,7 @@ def user_input():
         parser.add_argument('-nn', '--no_numbering', default=False, help="Allows you to exclude the default numbering of images", action="store_true")
         parser.add_argument('-of', '--offset', help="Where to start in the fetched links", type=str, required=False)
         parser.add_argument('-nd', '--no_download', default=False, help="Prints the URLs of the images and/or thumbnails without downloading them", action="store_true")
+        parser.add_argument('-iu', '--ignore_urls', default=False, help="delimited list input of image urls/keywords to ignore", type=str)
         parser.add_argument('-sil', '--silent_mode', default=False, help="Remains silent. Does not print notification messages on the terminal", action="store_true")
         parser.add_argument('-is', '--save_source', help="creates a text file containing a list of downloaded images along with source page url", type=str, required=False)
 
@@ -507,7 +508,7 @@ def create_directories(self,main_directory, dir_name,thumbnail,thumbnail_only):
 
 
     # Download Images
-    def download_image_thumbnail(self,image_url,main_directory,dir_name,return_image_name,print_urls,socket_timeout,print_size,no_download,save_source,img_src):
+    def download_image_thumbnail(self,image_url,main_directory,dir_name,return_image_name,print_urls,socket_timeout,print_size,no_download,save_source,img_src,ignore_urls):
         if print_urls or no_download:
             print("Image URL: " + image_url)
         if no_download:
@@ -574,7 +575,10 @@ def download_image_thumbnail(self,image_url,main_directory,dir_name,return_image
 
 
     # Download Images
-    def download_image(self,image_url,image_format,main_directory,dir_name,count,print_urls,socket_timeout,prefix,print_size,no_numbering,no_download,save_source,img_src,silent_mode,thumbnail_only,format):
+    def download_image(self,image_url,image_format,main_directory,dir_name,count,print_urls,socket_timeout,prefix,print_size,no_numbering,no_download,save_source,img_src,silent_mode,thumbnail_only,format,ignore_urls):
+        if ignore_urls:
+            if any(url in image_url for url in ignore_urls.split(',')):
+                return "fail","Image ignored",None,None
         if thumbnail_only:
             return "success", "Skipping image download...", str(image_url[(image_url.rfind('/')) + 1:]), image_url
         if not silent_mode:
@@ -759,14 +763,14 @@ def _get_all_items(self,page,main_directory,dir_name,limit,arguments):
                         print("\nImage Metadata: " + str(object))
 
                 #download the images
-                download_status,download_message,return_image_name,absolute_path = self.download_image(object['image_link'],object['image_format'],main_directory,dir_name,count,arguments['print_urls'],arguments['socket_timeout'],arguments['prefix'],arguments['print_size'],arguments['no_numbering'],arguments['no_download'],arguments['save_source'],object['image_source'],arguments["silent_mode"],arguments["thumbnail_only"],arguments['format'])
+                download_status,download_message,return_image_name,absolute_path = self.download_image(object['image_link'],object['image_format'],main_directory,dir_name,count,arguments['print_urls'],arguments['socket_timeout'],arguments['prefix'],arguments['print_size'],arguments['no_numbering'],arguments['no_download'],arguments['save_source'],object['image_source'],arguments["silent_mode"],arguments["thumbnail_only"],arguments['format'],arguments['ignore_urls'])
                 if not arguments["silent_mode"]:
                     print(download_message)
                 if download_status == "success":
 
                     # download image_thumbnails
                     if arguments['thumbnail'] or arguments["thumbnail_only"]:
-                        download_status, download_message_thumbnail = self.download_image_thumbnail(object['image_thumbnail_url'],main_directory,dir_name,return_image_name,arguments['print_urls'],arguments['socket_timeout'],arguments['print_size'],arguments['no_download'],arguments['save_source'],object['image_source'])
+                        download_status, download_message_thumbnail = self.download_image_thumbnail(object['image_thumbnail_url'],main_directory,dir_name,return_image_name,arguments['print_urls'],arguments['socket_timeout'],arguments['print_size'],arguments['no_download'],arguments['save_source'],object['image_source'],arguments['ignore_urls'])
                         if not arguments["silent_mode"]:
                             print(download_message_thumbnail)