fixed bad encoding in international languages

src/parser_interests.py

@@ -9,6 +9,9 @@ def parse_interests(user_data: dict) -> dict:
     user_interests = utils.json_file_converter(interests_path)["topics"]
     interest_categories = utils.json_file_converter(interest_categories_path)
 
+    for interest in user_interests:
+        interest = interest.encode('latin_1').decode('utf8')
+
     interest_category_count = {}
     interest_category_total = {}
 
@@ -58,6 +61,6 @@ def html_interest_list_builder(user_interests: list, interest_categories: dict)
                 corresponding_category = category
             elif interest in politicians:
                 corresponding_category = "Political Orientation"
-        html_interests += start_html + interest + mid_html + corresponding_category + end_html
+        html_interests += start_html + interest.encode('latin_1').decode('utf8') + mid_html + corresponding_category + end_html
 
     return html_interests

src/parser_off_facebook_tracking.py

@@ -6,6 +6,7 @@ def parse_off_facebook_activities(user_data: dict) -> dict:
     off_facebook_activities_path = getcwd() + "/temp/ads_and_businesses/your_off-facebook_activity.json"
     off_facebook_activity_list = utils.json_file_converter(off_facebook_activities_path)["off_facebook_activity"]
 
+
     number_of_websites_tracking_the_user = 0
     number_of_applications_tracking_the_user = 0
 
@@ -16,7 +17,7 @@ def parse_off_facebook_activities(user_data: dict) -> dict:
             number_of_websites_tracking_the_user += 1
         else:
             number_of_applications_tracking_the_user += 1
-        events_per_tracker[tracking_entity["name"]] = len(tracking_entity["events"])
+        events_per_tracker[tracking_entity["name"].encode('latin_1').decode('utf8')] = len(tracking_entity["events"])
 
     # sort dict of events per tracker by descending value order
     events_per_tracker = {k: v for k, v in sorted(events_per_tracker.items(), reverse=True, key=lambda item: item[1])}

src/parser_peer_group.py

@@ -12,6 +12,9 @@ def parse_peer_group(user_data: dict) -> dict:
         return user_data
     peer_group = peer_group_info["friend_peer_group"]
 
+
+    peer_group = peer_group.encode('latin_1').decode('utf8')
+
     user_data["peer_group"] = peer_group
 
     return user_data

src/parser_profile_info.py

@@ -9,7 +9,7 @@ def parse_user_info(user_data: dict) -> dict:
     user_name = profile_data["name"]["full_name"]
     join_year = utils.epoch_to_year(profile_data["registration_timestamp"])
     if "relationship" in profile_data:
-        relationship_status = profile_data["relationship"]["status"]
+        relationship_status = profile_data["relationship"]["status"].encode('latin_1').decode('utf8')
         if "partner" in profile_data["relationship"]:
             relationship_status += " with " + profile_data["relationship"]["partner"]
         relationship_timestamp = utils.epoch_to_year(profile_data["relationship"]["timestamp"])

src/utils.py

@@ -3,7 +3,7 @@
 
 def json_file_converter(file_path: str) -> dict:
     """ Opens a json file and returns a corresponding python list or dictionary object """
-    with open(file_path, "r", encoding="utf8") as json_file:
+    with open(file_path, "r", encoding="latin1") as json_file:
         json_str = json_file.read()
     python_str = json_str.replace("true", "True").replace("false", "False").replace("null", "None")
     python_dict = literal_eval(python_str)
@@ -14,7 +14,6 @@ def epoch_to_year(posix_time: int) -> str:
     # just in case number provided is a string
     posix_time = int(posix_time)
     year = datetime.utcfromtimestamp(posix_time).strftime('%Y')
-    month = datetime.utcfromtimestamp(posix_time).strftime('%m')
     return year
 
 def epoch_to_year_and_month(posix_time: int) -> str: