Update app.py

app.py

@@ -1,3 +1,34 @@
+
+Hugging Face's logo Hugging Face
+
+Models
+Datasets
+Spaces
+Docs
+Pricing
+
+Spaces:
+cstr
+/
+conceptnet_normalized
+App
+Files
+Community
+Settings
+conceptnet_normalized
+/ app.py
+cstr's picture
+cstr
+Update app.py
+fefc16c
+verified
+3 minutes ago
+raw
+history
+blame
+edit
+delete
+23.4 kB
 import gradio as gr
 import sqlite3
 import pandas as pd
@@ -10,7 +41,7 @@ from collections import defaultdict
 
 # ===== CONFIGURATION =====
 # 1. Point to the NEW normalized database (fixed)
-TARGET_LANGUAGES = ['de', 'en', 'es', 'fr', 'it', 'ja', 'nl', 'pl', 'pt', 'ru', 'zh']
+TARGET_LANGUAGES = ['en', 'fr', 'it', 'de', 'es', 'ar', 'fa', 'grc', 'he', 'la', 'hbo']
 NORMALIZED_REPO_ID = "cstr/conceptnet-normalized-multi"
 NORMALIZED_DB_FILE = "conceptnet_normalized.db"
 
@@ -102,15 +133,15 @@ def get_db_connection():
     return conn
 
 def node_url_to_label(url: str) -> str:
-    """Extract the term from ConceptNet URL: /c/{lang}/{term}/..."""
+    """Extract the term from ConceptNet URL: http://conceptnet.io/c/{lang}/{term}/..."""
     try:
         parts = url.split('/')
-        if len(parts) >= 4:
-            # Index 3 is the term, regardless of whether there's a POS tag
-            return parts[3].replace('_', ' ')
+        # Term is ALWAYS at index 5
+        if len(parts) >= 6 and parts[3] == 'c':
+            return parts[5].replace('_', ' ')
     except:
         pass
-    return url  # fallback
+    return url  # Fallback to full URL if parsing fails
 
 def get_semantic_profile(word: str, lang: str = 'en', selected_relations: List[str] = None, progress=gr.Progress()):
     """
@@ -287,12 +318,11 @@ def get_semantic_profile(word: str, lang: str = 'en', selected_relations: List[s
         traceback.print_exc()
         yield f"**❌ Error:** {e}"
 
-def run_query(start_node, relation, end_node, limit, progress=gr.Progress()):
+def run_query(start_node, start_lang, relation, end_node, end_lang, limit, progress=gr.Progress()):
     """
-    --- REWRITTEN FOR NORMALIZED DB ---
     Query builder using fast integer joins.
     """
-    log_progress(f"Query: start={start_node}, rel={relation}, end={end_node}", "INFO")
+    log_progress(f"Query: start={start_node} ({start_lang}), rel={relation}, end={end_node} ({end_lang})", "INFO")
     progress(0, desc="Building...")
     
     if not DB_PATH:
@@ -318,12 +348,12 @@ def run_query(start_node, relation, end_node, limit, progress=gr.Progress()):
         with get_db_connection() as conn:
             progress(0.3, desc="Adding filters...")
             
-            # Start node
+            # Start node - USE start_lang
             if start_node and start_node.strip():
                 if start_node.startswith('http://'):
                     pattern = f"{start_node}%"
                 else:
-                    pattern = f"{CONCEPTNET_BASE}/c/en/{start_node.strip().lower().replace(' ', '_')}%"
+                    pattern = f"{CONCEPTNET_BASE}/c/{start_lang}/{start_node.strip().lower().replace(' ', '_')}%"
                 where_clauses.append("n_start.node_url LIKE ?")
                 params.append(pattern)
             
@@ -334,12 +364,12 @@ def run_query(start_node, relation, end_node, limit, progress=gr.Progress()):
                     where_clauses.append("r.rel_url = ?")
                     params.append(rel_value)
             
-            # End node
+            # End node - USE end_lang
             if end_node and end_node.strip():
                 if end_node.startswith('http://'):
                     pattern = f"{end_node}%"
                 else:
-                    pattern = f"{CONCEPTNET_BASE}/c/en/{end_node.strip().lower().replace(' ', '_')}%"
+                    pattern = f"{CONCEPTNET_BASE}/c/{end_lang}/{end_node.strip().lower().replace(' ', '_')}%"
                 where_clauses.append("n_end.node_url LIKE ?")
                 params.append(pattern)
             
@@ -472,6 +502,7 @@ with gr.Blocks(title="ConceptNet Explorer", theme=gr.themes.Soft()) as demo:
                 
                 with gr.Row():
                     start_input = gr.Textbox(label="Start Node (word)", placeholder="dog (optional)")
+                    start_lang = gr.Dropdown(choices=TARGET_LANGUAGES, value="en", label="Start Lang", scale=1)
                     rel_input = gr.Dropdown(
                         choices=[""] + list(CONCEPTNET_RELATIONS.keys()), 
                         label="Relation (name)", 
@@ -479,6 +510,7 @@ with gr.Blocks(title="ConceptNet Explorer", theme=gr.themes.Soft()) as demo:
                         info="Leave blank to query all relations"
                     )
                     end_input = gr.Textbox(label="End Node (word)", placeholder="(optional)")
+                    end_lang = gr.Dropdown(choices=TARGET_LANGUAGES, value="en", label="End Lang", scale=1)
                 
                 limit_slider = gr.Slider(label="Limit", minimum=1, maximum=500, value=50, step=1)
                 query_btn = gr.Button("▶️ Run Query", variant="primary", size="lg")
@@ -530,7 +562,7 @@ LIMIT 10
         
         query_btn.click(
             run_query, 
-            inputs=[start_input, rel_input, end_input, limit_slider], 
+            inputs=[start_input, start_lang, rel_input, end_input, end_lang, limit_slider], 
             outputs=[results_output, status_output],
             api_name="run_query"
         )
@@ -560,4 +592,4 @@ if __name__ == "__main__":
         log_progress("APP READY! (Normalized DB)", "SUCCESS")
     else:
         log_progress("APP LAUNCHING WITH ERRORS (DB NOT FOUND)", "ERROR")
-    demo.launch(ssr_mode=False)
+    demo.launch(ssr_mode=False)