Fix crash

trustgraph-flow/trustgraph/extract/kg/ontology/extract.py

@@ -88,6 +88,7 @@ class Processor(FlowProcessor):
         # Track loaded ontology version
         self.current_ontology_version = None
         self.loaded_ontology_ids = set()
+        self.pending_config = None  # Store config until components initialized
 
     async def initialize_components(self, flow):
         """Initialize OntoRAG components."""
@@ -124,8 +125,12 @@ class Processor(FlowProcessor):
             self.initialized = True
             logger.info("OntoRAG components initialized successfully")
 
-            # NOTE: Ontologies will be loaded via on_ontology_config() handler
-            # when ConfigPush messages arrive (including initial config on startup)
+            # Process pending config if available
+            if self.pending_config:
+                logger.info("Processing pending config from startup")
+                config, version = self.pending_config
+                self.pending_config = None
+                await self.on_ontology_config(config, version)
 
         except Exception as e:
             logger.error(f"Failed to initialize OntoRAG components: {e}", exc_info=True)
@@ -156,6 +161,12 @@ class Processor(FlowProcessor):
                 logger.warning("No 'ontology' section in config")
                 return
 
+            # Check if components are initialized
+            if not self.ontology_loader:
+                logger.debug("Components not yet initialized, storing config for later processing")
+                self.pending_config = (config, version)
+                return
+
             ontology_configs = config["ontology"]
 
             # Parse ontology definitions

trustgraph-flow/trustgraph/extract/kg/ontology/text_processor.py

@@ -14,9 +14,16 @@ logger = logging.getLogger(__name__)
 
 # Ensure required NLTK data is downloaded
 try:
-    nltk.data.find('tokenizers/punkt')
+    nltk.data.find('tokenizers/punkt_tab')
 except LookupError:
-    nltk.download('punkt', quiet=True)
+    try:
+        nltk.download('punkt_tab', quiet=True)
+    except:
+        # Fallback to older punkt if punkt_tab not available
+        try:
+            nltk.download('punkt', quiet=True)
+        except:
+            pass
 
 try:
     nltk.data.find('taggers/averaged_perceptron_tagger')
@@ -44,8 +51,14 @@ class SentenceSplitter:
 
     def __init__(self):
         """Initialize sentence splitter."""
-        self.sent_detector = nltk.data.load('tokenizers/punkt/english.pickle')
-        logger.info("Using NLTK sentence tokenizer")
+        try:
+            # Try newer punkt_tab first
+            self.sent_detector = nltk.data.load('tokenizers/punkt_tab/english/')
+            logger.info("Using NLTK sentence tokenizer (punkt_tab)")
+        except:
+            # Fallback to older punkt
+            self.sent_detector = nltk.data.load('tokenizers/punkt/english.pickle')
+            logger.info("Using NLTK sentence tokenizer (punkt)")
 
     def split(self, text: str) -> List[str]:
         """Split text into sentences.