Update the display for language and country codes.

requirements.txt

@@ -3,3 +3,5 @@ pandas
 pyarrow
 huggingface-hub
 datasets
+pycountry
+babel

src/streamlit_app.py

@@ -5,6 +5,8 @@ import hashlib
 import json
 import tempfile
 from huggingface_hub import HfApi, login, hf_hub_download
+import pycountry
+from babel import Locale
 
 # Hugging Face Dataset configuration
 # In HF Spaces, variables and secrets are available as environment variables
@@ -84,6 +86,72 @@ def discover_available_jurisdictions():
         # Return empty list as fallback
         return []
 
+def get_jurisdiction_display_name(jurisdiction_code):
+    """
+    Convert jurisdiction code (e.g., 'hr-hr') to display name (e.g., 'Croatian-Croatia').
+    
+    Uses ISO 639-1 (language) and ISO 3166-1 (country) codes to generate human-readable names.
+    
+    Args:
+        jurisdiction_code: String in format 'language-country' (e.g., 'hr-hr', 'en-us')
+    
+    Returns:
+        Display name string (e.g., 'Croatian-Croatia') or original code if conversion fails
+    """
+    try:
+        # Parse jurisdiction code (e.g., "hr-hr" -> language="hr", country="HR")
+        parts = jurisdiction_code.lower().split('-')
+        if len(parts) != 2:
+            return jurisdiction_code  # Fallback to original if format is wrong
+        
+        language_code, country_code = parts[0], parts[1].upper()
+        
+        # Get language name using babel
+        language_name = None
+        try:
+            # Try to parse locale (e.g., "hr_HR" or "en_US")
+            locale_str = f"{language_code}_{country_code}"
+            locale = Locale.parse(locale_str)
+            language_name = locale.get_language_name('en')
+            if language_name:
+                language_name = language_name.title()
+        except Exception:
+            pass
+        
+        # Fallback: try pycountry for language
+        if not language_name:
+            try:
+                lang = pycountry.languages.get(alpha_2=language_code)
+                language_name = lang.name
+            except Exception:
+                language_name = language_code.upper()
+        
+        # Get country name using pycountry
+        country_name = None
+        try:
+            country = pycountry.countries.get(alpha_2=country_code)
+            country_name = country.name
+        except Exception:
+            country_name = country_code
+        
+        return f"{language_name}-{country_name}"
+    except Exception:
+        # Fallback to original code if anything goes wrong
+        return jurisdiction_code
+
+@st.cache_data
+def get_jurisdiction_display_mapping(jurisdiction_codes):
+    """
+    Create a mapping from jurisdiction codes to display names.
+    
+    Args:
+        jurisdiction_codes: List of jurisdiction code strings
+    
+    Returns:
+        Dictionary mapping codes to display names
+    """
+    return {code: get_jurisdiction_display_name(code) for code in jurisdiction_codes}
+
 @st.cache_data
 def load_grading_template(jurisdiction):
     """Load grading template from Hugging Face Dataset for the specified jurisdiction"""
@@ -695,12 +763,29 @@ if not st.session_state.logged_in:
         st.error("❌ **Error**: No jurisdictions found in the repository. Please ensure the repository structure is correct.")
         st.stop()
     
-    # Determine default index for selectbox
+    # Get display name mapping
+    display_mapping = get_jurisdiction_display_mapping(available_jurisdictions)
+    
+    # Determine default index for selectbox - prioritize hr-hr
     default_index = 0
-    if st.session_state.jurisdiction and st.session_state.jurisdiction in available_jurisdictions:
+    if "hr-hr" in available_jurisdictions:
+        default_index = available_jurisdictions.index("hr-hr")
+        # Set hr-hr as default in session state if not already set
+        if not st.session_state.jurisdiction:
+            st.session_state.jurisdiction = "hr-hr"
+    elif st.session_state.jurisdiction and st.session_state.jurisdiction in available_jurisdictions:
         default_index = available_jurisdictions.index(st.session_state.jurisdiction)
     
-    jurisdiction = st.selectbox("Jurisdiction", options=available_jurisdictions, index=default_index)
+    # Create format function to show display names
+    def format_jurisdiction(code):
+        return display_mapping.get(code, code)
+    
+    jurisdiction = st.selectbox(
+        "Jurisdiction", 
+        options=available_jurisdictions, 
+        index=default_index,
+        format_func=format_jurisdiction
+    )
     st.session_state.jurisdiction = jurisdiction
     
     username = st.text_input("Username")
@@ -788,7 +873,8 @@ elif st.session_state.logged_in:
             st.write(f"HF Token configured: {HF_TOKEN is not None}")
             st.write(f"HF API initialized: {hf_api is not None}")
             if username:
-                st.write(f"Jurisdiction: `{jurisdiction}`")
+                jurisdiction_display = get_jurisdiction_display_name(jurisdiction)
+                st.write(f"Jurisdiction: {jurisdiction_display} (`{jurisdiction}`)")
                 st.write(f"User parquet file: `{jurisdiction}/users/{username}_answers.parquet`")
                 st.write(f"Users file: `{jurisdiction}/users/users.json`")