add hf to discord mapping

app.py

@@ -86,7 +86,8 @@ def get_blend_es_data():
                                     user_mapping[user_id] = f"User-{user_id[:8]}"
 
             for user_id, count in dataset_contributions.items():
-                username = user_mapping.get(user_id, f"User-{user_id[:8]}")
+                hf_username = user_mapping.get(user_id, f"User-{user_id[:8]}")
+                username = get_discord_username(hf_username)
                 data.append(
                     {"source": "blend-es", "username": username, "count": count}
                 )
@@ -126,26 +127,63 @@ def get_include_data():
     return data
 
 
-def get_mail_to_username_mapping():
-    mail_to_discord = {}
+def get_user_mapping():
+    """
+    Get cached mapping of emails and hf_usernames to discord usernames.
+    Returns a tuple of (email_to_discord, hf_username_to_discord) mappings.
+    """
+    email_to_discord = {}
+    hf_username_to_discord = {}
+
     try:
         if os.path.exists(PARTICIPANTS_CSV):
             mapping_df = pd.read_csv(PARTICIPANTS_CSV)
+
+            # Map emails to discord usernames
             if "gmail" in mapping_df.columns and "discord" in mapping_df.columns:
                 for _, row in mapping_df.iterrows():
                     mail = row["gmail"]
                     discord = row["discord"]
-                    if pd.notna(mail) and pd.notna(discord):
-                        mail_to_discord[mail.lower()] = discord.lower()
+                    if pd.notna(mail) and pd.notna(discord) and discord != "NA":
+                        email_to_discord[mail.lower()] = discord.lower()
+
+            # Map hf_usernames to discord usernames
+            if "hf_username" in mapping_df.columns and "discord" in mapping_df.columns:
+                for _, row in mapping_df.iterrows():
+                    hf_username = row["hf_username"]
+                    discord = row["discord"]
+                    if pd.notna(hf_username) and pd.notna(discord) and discord != "NA":
+                        hf_username_to_discord[hf_username.lower()] = discord.lower()
+
     except Exception as e:
         print(f"Error loading {PARTICIPANTS_CSV}: {e}")
 
-    return mail_to_discord
+    return email_to_discord, hf_username_to_discord
+
+
+def get_discord_username(identifier):
+    """
+    Get discord username from either email or hf_username. Returns the discord username if found, otherwise returns the identifier.
+    """
+    email_to_discord, hf_username_to_discord = get_user_mapping()
+
+    # Try to find discord username by email first
+    if "@" in identifier:
+        discord_username = email_to_discord.get(identifier.lower())
+        if discord_username:
+            return discord_username
+
+    # Try to find discord username by hf_username
+    discord_username = hf_username_to_discord.get(identifier.lower())
+    if discord_username:
+        return discord_username
+
+    # Fallback: use identifier as username
+    return identifier.split("@")[0] if "@" in identifier else identifier
 
 
 def get_estereotipos_data():
     data = []
-    mail_to_discord = get_mail_to_username_mapping()
 
     try:
         if os.path.exists(STEREOTYPES_CSV):
@@ -159,10 +197,7 @@ def get_estereotipos_data():
                         mail_counts[mail.lower()] += int(count)
 
                 for mail, count in mail_counts.items():
-                    username = mail_to_discord.get(mail.lower(), "")
-                    if not username:
-                        username = mail.split("@")[0] if "@" in mail else mail
-
+                    username = get_discord_username(mail)
                     data.append(
                         {"source": "estereotipos", "username": username, "count": count}
                     )
@@ -174,7 +209,6 @@ def get_estereotipos_data():
 
 def get_arena_data():
     data = []
-    mail_to_discord = get_mail_to_username_mapping()
 
     try:
         if os.path.exists(ARENA_JSON):
@@ -193,10 +227,7 @@ def get_arena_data():
                             mail_counts[mail.lower()] += 1
 
             for mail, count in mail_counts.items():
-                username = mail_to_discord.get(mail.lower(), "")
-                if not username:
-                    username = mail.split("@")[0] if "@" in mail else mail
-
+                username = get_discord_username(mail)
                 data.append({"source": "arena", "username": username, "count": count})
     except Exception as e:
         print(f"Error loading {ARENA_JSON}: {e}")