Update utils/non_melting_point.py

utils/non_melting_point.py

@@ -8,6 +8,8 @@ import re
 from tqdm import tqdm
 from spacy.cli import download
 
+# Add imports for the curated transformer support
+from spacy.util import registry
 
 # Logging setup
 logging.basicConfig(level=logging.WARNING, format="%(asctime)s - %(levelname)s - %(message)s")
@@ -25,6 +27,18 @@ class NgramProcessor:
         if models is None:
             models = ["en_core_web_trf"]
         
+        # Check and register curated transformer if needed
+        if "curated_transformer" not in registry.factories.get_names("pipe", "en"):
+            try:
+                # Try to import and register the curated transformer
+                import spacy_curated_transformers
+                from spacy_curated_transformers import CuratedTransformer
+                from spacy.language import Language
+                Language.factory("curated_transformer", func=CuratedTransformer)
+                tqdm.write("[NgramProcessor] Registered curated_transformer factory")
+            except ImportError:
+                tqdm.write("[NgramProcessor] Warning: spacy_curated_transformers not found. Some models may not load correctly.")
+        
         # Load specified model
         self.models = {}
         for model_name in models: