Attempting to give more information on the table columns

app.py

@@ -18,6 +18,7 @@ Generation:
 
 import gradio as gr
 import icij_utils
+import sqlalchemy
 import smolagents
 import os
 import pathlib
@@ -65,8 +66,8 @@ for table, doc in metadata.TABLE_DOCS.items():
     if table in schema:
         for col_name, col_type in schema[table].items():
             col_doc = metadata.COLUMN_DOCS.get(table, {}).get(col_name, "No documentation available")
-            #tool_description += f"  - {col_name}: {col_type}: {col_doc}\n"
-            tool_description += f"  - {col_name}: {col_type}\n"
+            tool_description += f"  - {col_name}: {col_type}: {col_doc}\n"
+            #tool_description += f"  - {col_name}: {col_type}\n"
         
 # Add source documentation
 #tool_description += "\n\nSource IDs:\n"
@@ -150,8 +151,12 @@ with gr.Blocks() as demo:
     with gr.Accordion("Sample questions", open=False):
         gr.Examples(
             [
-                ["",],
-                ["",],
+                [
+                    (
+                        "Can you list the entities with an address in Canada? "
+                        "Please give the name of the entity an its  address."
+                    ),
+                ],
             ],
             inputs=[question,],
             outputs=[response,],

icij_utils.py

@@ -229,7 +229,7 @@ class ICIJDatabaseMetadata:
             'name': "Legal name of the offshore entity",
             'original_name': "Name in original language/character set",
             'former_name': "Previous names of the entity",
-            'jurisdiction': "Country/region where the entity is registered",
+            #'jurisdiction': "Country/region where the entity is registered",
             'jurisdiction_description': "Detailed description of the jurisdiction",
             'company_type': "Legal structure of the entity (e.g., corporation, trust)",
             'address': "Primary registered address",
@@ -240,29 +240,31 @@ class ICIJDatabaseMetadata:
             'dorm_date': "Date when entity became dormant",
             'status': "Current status of the entity",
             'service_provider': "Firm that provided offshore services",
-            'source_id': "Identifier for the leak source"
+            'country_codes': '3 letter abbreviations of country names',
+            'countries': 'name of country',
+            'sourceID': "Identifier for the leak source"
         },
         
         'others': {
             'name': "Name of the miscellaneous party or item",
             'type': "Type of the other party (e.g., vessel, legal case)",
             'incorporation_date': "Date of incorporation or creation if applicable",
-            'jurisdiction': "Jurisdiction associated with the party",
+            'jurisdiction': "2 letter code of the Jurisdiction associated with the party",
+            'jurisdiction-description': 'full name of the jurisdiction',
             'countries': "Countries associated with the party",
             'status': "Current status",
             'internal_id': "Unique identifier within the leak data",
             'address': "Associated address if available",
-            'source_id': "Identifier for the leak source",
+            'sourceID': "Identifier for the leak source",
             'valid_until': "Date until which the information is valid"
         },
         
         'officers': {
             'name': "Name of the individual or organization",
-            'country_codes': "Countries connected to the officer",
-            'source_id': "Identifier for the leak source",
+            'countries': 'full name of the country connected to the officer',
+            'country_codes': "3 letter code of the countries connected to the officer",
+            'sourceID': "Identifier for the leak source",
             'valid_until': "Date until which the information is valid",
-            'status': "Current status of the officer",
-            'internal_id': "Unique identifier within the leak data"
         },
         
         'intermediaries': {
@@ -270,28 +272,28 @@ class ICIJDatabaseMetadata:
             'internal_id': "Unique identifier within the leak data",
             'address': "Business address",
             'status': "Current status",
-            'country_codes': "Countries where intermediary operates",
-            'source_id': "Identifier for the leak source"
+            'countries': "Countries where intermediary operates",
+            'country_codes': "3 letter abbreviations of the countries where intermediary operates",
+            'sourceID': "Identifier for the leak source"
         },
         
         'addresses': {
             'address': "Full address text",
             'name': "Name associated with address",
-            'country_codes': "Country codes for the address",
+            'country_codes': "3 letter country codes for the address",
             'countries': "Full country names",
-            'source_id': "Identifier for the leak source",
+            'sourceID': "Identifier for the leak source",
             'valid_until': "Date until which address is valid",
-            'internal_id': "Unique identifier within the leak data"
         },
         
         'relationships': {
-            'from_id': "Internal ID of the source node",
-            'to_id': "Internal ID of the target node",
+            'node_id_start': "Internal ID of the source node",
+            'node_id_end': "Internal ID of the target node",
             'rel_type': "Type of relationship (e.g., shareholder, director)",
             'link': "Additional details about the relationship",
             'start_date': "When the relationship began",
             'end_date': "When the relationship ended",
-            'source_id': "Identifier for the leak source",
+            'sourceID': "Identifier for the leak source",
             'status': "Current status of the relationship"
         }
     }