chg: [screenshot + har directory] add option to change screenshots directory

bin/lib/ConfigLoader.py

@@ -41,6 +41,15 @@ class ConfigLoader(object):
                                   db=self.cfg.getint(redis_name, "db"),
                                   decode_responses=decode_responses )
 
+    def get_files_directory(self, key_name):
+        directory_path = self.cfg.get('Directories', key_name)
+        # full path
+        if directory_path[0] == '/':
+            return directory_path
+        else:
+            directory_path = os.path.join(os.environ['AIL_HOME'], directory_path)
+            return directory_path
+
     def get_config_str(self, section, key_name):
         return self.cfg.get(section, key_name)
 

bin/lib/Screenshot.py

@@ -20,7 +20,7 @@ import ConfigLoader
 config_loader = ConfigLoader.ConfigLoader()
 r_serv_onion = config_loader.get_redis_conn("ARDB_Onion")
 r_serv_metadata = config_loader.get_redis_conn("ARDB_Metadata")
-SCREENSHOT_FOLDER = os.path.join(os.environ['AIL_HOME'], config_loader.get_config_str("Directories", "crawled_screenshot"), 'screenshot')
+SCREENSHOT_FOLDER = config_loader.get_files_directory('screenshot')
 config_loader = None
 
 # get screenshot relative path

bin/packages/HiddenServices.py

@@ -60,11 +60,10 @@ class HiddenServices(object):
             self.paste_directory = os.path.join(os.environ['AIL_HOME'], config_loader.get_config_str("Directories", "pastes"))
             self.paste_crawled_directory = os.path.join(self.paste_directory, config_loader.get_config_str("Directories", "crawled"))
             self.paste_crawled_directory_name = config_loader.get_config_str("Directories", "crawled")
-            self.screenshot_directory = os.path.join(os.environ['AIL_HOME'], config_loader.get_config_str("Directories", "crawled_screenshot"))
+            self.screenshot_directory = config_loader.get_files_directory('screenshot')
-            self.screenshot_directory_screenshot = os.path.join(self.screenshot_directory, 'screenshot')
         elif type == 'i2p':
             self.paste_directory = os.path.join(os.environ['AIL_HOME'], config_loader.get_config_str("Directories", "crawled_screenshot"))
-            self.screenshot_directory = os.path.join(os.environ['AIL_HOME'], config_loader.get_config_str("Directories", "crawled_screenshot"))
+            self.screenshot_directory = config_loader.get_files_directory('screenshot')
         else:
             ## TODO: # FIXME: add error
             pass

bin/packages/Item.py

@@ -34,7 +34,8 @@ PASTES_FOLDER = os.path.join(os.path.realpath(PASTES_FOLDER), '')
 
 r_cache = config_loader.get_redis_conn("Redis_Cache")
 r_serv_metadata = config_loader.get_redis_conn("ARDB_Metadata")
-screenshot_directory = os.path.join(os.environ['AIL_HOME'], config_loader.get_config_str("Directories", "crawled_screenshot"))
+screenshot_directory = config_loader.get_files_directory('screenshot')
+har_directory = config_loader.get_files_directory('har')
 
 config_loader = None
 
@@ -388,7 +389,7 @@ def get_item_screenshot(item_id):
     return ''
 
 def get_item_har_name(item_id):
-    har_path = os.path.join(screenshot_directory, item_id) + '.json'
+    har_path = os.path.join(har_directory, item_id) + '.json'
     if os.path.isfile(har_path):
         return har_path
     else:

bin/torcrawler/TorSplashCrawler.py

@@ -26,7 +26,7 @@ sys.path.append(os.environ['AIL_BIN'])
 from Helper import Process
 
 sys.path.append(os.path.join(os.environ['AIL_BIN'], 'lib'))
-#import ConfigLoader
+import ConfigLoader
 import Screenshot
 import crawlers
 
@@ -133,7 +133,11 @@ class TorSplashCrawler():
             config_section = 'Crawler'
             self.p = Process(config_section)
             self.item_dir = os.path.join(self.p.config.get("Directories", "crawled"), date_str )
-            self.har_dir = os.path.join(os.environ['AIL_HOME'], self.p.config.get("Directories", "crawled_screenshot"), date_str )
+
+            config_loader = ConfigLoader.ConfigLoader()
+            self.har_dir = os.path.join(config_loader.get_files_directory('har') , date_str )
+            config_loader = None
+
             self.r_serv_log_submit = redis.StrictRedis(
                 host=self.p.config.get("Redis_Log_submit", "host"),
                 port=self.p.config.getint("Redis_Log_submit", "port"),

configs/core.cfg.sample

@@ -4,7 +4,8 @@ dicofilters = Dicos
 pastes = PASTES
 hash = HASHS
 crawled = crawled
-crawled_screenshot = CRAWLED_SCREENSHOT
+har = CRAWLED_SCREENSHOT
+screenshot = CRAWLED_SCREENSHOT/screenshot
 
 wordtrending_csv = var/www/static/csv/wordstrendingdata
 wordsfile = files/wordfile
@@ -221,6 +222,11 @@ host = localhost
 port = 6382
 db = 10
 
+[Kvrocks_Meta]
+host = localhost
+port = 6383
+db = 0
+
 [Url]
 cc_critical = DE
 
@@ -278,6 +284,7 @@ default_crawler_closespider_pagecount = 50
 default_crawler_user_agent = Mozilla/5.0 (Windows NT 6.1; rv:60.0) Gecko/20100101 Firefox/60.0
 splash_url = http://127.0.0.1
 splash_port = 8050-8052
+domain_proxy = onion.foundation
 
 [IP]
 # list of comma-separated CIDR that you wish to be alerted for. e.g:

update/v1.5/Update-ARDB_Onions_screenshots.py

@@ -33,7 +33,7 @@ if __name__ == '__main__':
     config_loader = ConfigLoader.ConfigLoader()
 
     SCREENSHOT_FOLDER = os.path.join(os.environ['AIL_HOME'], config_loader.get_config_str("Directories", "crawled_screenshot"))
-    NEW_SCREENSHOT_FOLDER = os.path.join(os.environ['AIL_HOME'], config_loader.get_config_str("Directories", "crawled_screenshot"), 'screenshot')
+    NEW_SCREENSHOT_FOLDER = config_loader.get_files_directory('screenshot')
 
     PASTES_FOLDER = os.path.join(os.environ['AIL_HOME'], config_loader.get_config_str("Directories", "pastes")) + '/'
 

var/www/modules/Flask_config.py

@@ -100,7 +100,7 @@ dict_update_description = {'v1.5':{'nb_background_update': 5, 'update_warning_me
 UPLOAD_FOLDER = os.path.join(os.environ['AIL_FLASK'], 'submitted')
 
 PASTES_FOLDER = os.path.join(os.environ['AIL_HOME'], config_loader.get_config_str("Directories", "pastes")) + '/'
-SCREENSHOT_FOLDER = os.path.join(os.environ['AIL_HOME'], config_loader.get_config_str("Directories", "crawled_screenshot"), 'screenshot')
+SCREENSHOT_FOLDER = config_loader.get_files_directory('screenshot')
 
 REPO_ORIGIN = 'https://github.com/ail-project/ail-framework.git'