update intro, singleton conceptnet client

app.py

@@ -1,25 +1,62 @@
 # ============================================================================
-# GERMAN LINGUISTICS HUB (CONSOLIDATED APP V3)
+# GERMAN LINGUISTICS HUB (CONSOLIDATED APP V22)
 #
 # This script combines multiple NLP tools into a single Gradio interface.
 #
+# ============================================================================
 # TABS & FUNCTIONALITY:
-# 1. Comprehensive Analyzer (DE):
-#    - CONTEXTUAL analysis of full sentences.
-#    - Ranks all semantics by relevance to the sentence.
-# 2. Word Encyclopedia (DE): (NEW!)
+# ============================================================================
+#
+# --- PRIMARY TABS ---
+#
+# 1. Word Encyclopedia (DE):
 #    - NON-CONTEXTUAL analysis of single words.
-#    - Finds ALL grammatical (Pattern) and semantic (OdeNet, ConceptNet)
-#      possibilities, cross-validated and grouped by Part-of-Speech.
-#    - Ideal for enriching word lists.
-# 3. spaCy Analyzer (Multi-lingual): Direct spaCy output.
-# 4. Grammar Check (DE): LanguageTool.
-# 5. Inflections (DE): Direct Pattern.de output.
-# 6. Thesaurus (DE): Direct OdeNet output.
-# 7. ConceptNet (Direct): Direct ConceptNet API output.
+#    - Multi-engine dispatcher with user selection and automatic fallback:
+#      (Wiktionary -> DWDSmor -> HanTa -> IWNLP)
+#    - Aggregates all grammatical (Wiktionary, Pattern) and semantic 
+#      (Wiktionary, OdeNet, ConceptNet) possibilities, grouped by Part-of-Speech.
+#    - Validates and filters artifacts (e.g., "abgeschnitten", "lauf").
+#
+# 2. Comprehensive Analyzer (DE):
+#    - CONTEXTUAL analysis of full sentences.
+#    - Uses the Word Encyclopedia's dispatcher for robust lemma analysis.
+#    - Ranks all semantic senses (Wiktionary, OdeNet) by relevance to the sentence.
+#
+# --- STANDALONE TOOL TABS ---
+#
+# 3. spaCy Analyzer (Multi-lingual): 
+#    - Direct, raw spaCy output (NER, POS, dependencies) for multiple languages.
+#
+# 4. Grammar Check (DE): 
+#    - Direct LanguageTool output.
+#
+# --- RAW ENGINE TABS (for debugging & comparison) ---
+#
+# 5. Engine: Wiktionary (DE): 
+#    - Standalone access to the Wiktionary DB (Primary) engine.
+#
+# 6. Engine: DWDSmor (DE):
+#    - Standalone access to the DWDSmor (Fallback 1) engine.
+#
+# 7. Engine: HanTa (DE):
+#    - Standalone access to the HanTa (Fallback 2) engine.
+#
+# 8. Engine: IWNLP-spaCy (DE):
+#    - Standalone access to the IWNLP-spaCy (Fallback 3) engine.
+#
+# --- RAW COMPONENT TABS (for debugging & comparison) ---
+#
+# 9. Component: Inflections (DE): 
+#    - Direct access to the `pattern.de` library.
+#
+# 10. Component: Thesaurus (DE): 
+#     - Direct access to the `OdeNet` library.
+#
+# 11. Component: ConceptNet (Direct): 
+#     - Direct access to the ConceptNet API.
+#
 # ============================================================================
 
-
 # ============================================================================
 # 1. CONSOLIDATED IMPORTS
 # ============================================================================
@@ -188,6 +225,9 @@ WIKTIONARY_AVAILABLE = False
 CONCEPTNET_CACHE: Dict[Tuple[str, str], Any] = {}
 CONCEPTNET_LOCK = threading.Lock()
 
+CONCEPTNET_CLIENT: Optional[Client] = None 
+CONCEPTNET_CLIENT_LOCK = threading.Lock() 
+
 # --- HanTa Tagger Cache & Lock ---
 HANTA_TAGGER_INSTANCE: Optional[HanoverTagger] = None
 HANTA_TAGGER_LOCK = threading.Lock()
@@ -1320,6 +1360,29 @@ def word_appears_in_inflections(word: str, inflections: Dict[str, Any], pos_type
 # ============================================================================
 # 6b. CONCEPTNET HELPER LOGIC (V2 - ROBUST PARSER)
 # ============================================================================
+def get_conceptnet_client() -> Optional[Client]:
+    """ Thread-safe function to get a single instance of the Gradio Client. """
+    global CONCEPTNET_CLIENT
+    if not GRADIO_CLIENT_AVAILABLE:
+        return None
+    
+    if CONCEPTNET_CLIENT:
+        return CONCEPTNET_CLIENT
+    
+    with CONCEPTNET_CLIENT_LOCK:
+        if CONCEPTNET_CLIENT:
+            return CONCEPTNET_CLIENT
+        try:
+            print("Initializing Gradio Client for ConceptNet...")
+            client = Client("cstr/conceptnet_normalized")
+            print("✓ Gradio Client for ConceptNet initialized.")
+            CONCEPTNET_CLIENT = client
+            return CONCEPTNET_CLIENT
+        except Exception as e:
+            print(f"✗ CRITICAL: Failed to initialize ConceptNet Gradio Client: {e}")
+            traceback.print_exc()
+            return None
+
 def conceptnet_get_relations(word: str, language: str = 'de') -> Dict[str, Any]:
     """
     Fetches relations from the cstr/conceptnet_normalized Gradio API.
@@ -1346,7 +1409,9 @@ def conceptnet_get_relations(word: str, language: str = 'de') -> Dict[str, Any]:
     
     try:
         # --- 2. Call Gradio API ---
-        client = Client("cstr/conceptnet_normalized")
+        client = get_conceptnet_client() # <-- USE HELPER
+        if not client:
+            return {"error": "ConceptNet Gradio Client is not available."}
         
         selected_relations = [
             "RelatedTo", "IsA", "PartOf", "HasA", "UsedFor",
@@ -3521,10 +3586,21 @@ if __name__ == "__main__":
         print("✓ Requests library is available.")
     print("--- Requests Done ---\n")
 
+    # --- 8. Initialize ConceptNet Client ---
+    print("--- Initializing ConceptNet Client ---")
+    if GRADIO_CLIENT_AVAILABLE:
+        try:
+            get_conceptnet_client() # Call the function to load the client
+        except Exception as e:
+            print(f"✗ FAILED to start ConceptNet Client: {e}")
+    else:
+        print("INFO: gradio_client not available, skipping ConceptNet client.")
+    print("--- ConceptNet Client Done ---\n")
+
     print("="*70)
     print("All services initialized. Launching Gradio Hub...")
     print("="*70 + "\n")
     
-    # --- 8. Launch Gradio ---
+    # --- 9. Launch Gradio ---
     demo = create_consolidated_interface()
     demo.launch(server_name="0.0.0.0", server_port=7860, show_error=True)