pii-protector 2.2.1__tar.gz → 2.2.2__tar.gz

{pii_protector-2.2.1 → pii_protector-2.2.2}/PKG-INFO

@@ -1,10 +1,10 @@
 Metadata-Version: 2.4
 Name: pii-protector
-Version: 2.2.1
+Version: 2.2.2
 Summary: Production-grade PII detection with multi-model ensemble
 Author-email: tensoryug <jainsatyam26@gmail.com>
 License-Expression: MIT
-Project-URL: Homepage, https://github.com/sjain26/pii-guard
+Project-URL: Homepage, https://github.com/sjain26/pii-detector
 Keywords: pii,privacy,nlp,detection,presidio,data-protection
 Classifier: Development Status :: 5 - Production/Stable
 Classifier: Intended Audience :: Developers
@@ -22,12 +22,12 @@ Requires-Dist: presidio-analyzer>=2.2; extra == "presidio"
 Requires-Dist: spacy>=3.7; extra == "presidio"
 Provides-Extra: transformers
 Requires-Dist: transformers>=4.40; extra == "transformers"
-Requires-Dist: torch>=2.0; extra == "transformers"
+Requires-Dist: optimum[onnxruntime]>=1.18; extra == "transformers"
 Provides-Extra: full
 Requires-Dist: presidio-analyzer>=2.2; extra == "full"
 Requires-Dist: spacy>=3.7; extra == "full"
 Requires-Dist: transformers>=4.40; extra == "full"
-Requires-Dist: torch>=2.0; extra == "full"
+Requires-Dist: optimum[onnxruntime]>=1.18; extra == "full"
 
 # pii-protector
 

{pii_protector-2.2.1 → pii_protector-2.2.2}/pii_detector/__init__.py

@@ -7,5 +7,5 @@ Multi-model ensemble with pre-compiled patterns and type-aware voting
 from .detector import AdvancedPIIDetector, PIIEntity, PresidioCustomDetector
 from .config.settings import PII_CONFIG
 
-__version__ = "2.2.1"
+__version__ = "2.2.2"
 __all__ = ["AdvancedPIIDetector", "PIIEntity", "PresidioCustomDetector", "PII_CONFIG"]

{pii_protector-2.2.1 → pii_protector-2.2.2}/pii_detector/detector.py

@@ -317,6 +317,25 @@ class AdvancedPIIDetector:
         if self.use_pii_model:
             self._init_pii_transformer()  # PII model structured PII
 
+        if self.use_presidio and not self.presidio:
+            warnings.warn(
+                "Presidio (Layer 2) unavailable — NAME/ORG/LOCATION detection disabled. "
+                "Install: pip install 'pii-protector[presidio]' then: python -m spacy download en_core_web_lg",
+                UserWarning, stacklevel=2,
+            )
+        if self.use_transformers and not self.ner_pipeline:
+            warnings.warn(
+                "NER transformer (Layer 3) unavailable — high-accuracy name detection disabled. "
+                "Install: pip install 'optimum[onnxruntime]>=1.18' 'transformers>=4.40'",
+                UserWarning, stacklevel=2,
+            )
+        if self.use_pii_model and not self.pii_pipeline:
+            warnings.warn(
+                "PII model (Layer 4) unavailable — PHONE/EMAIL/DOB/SSN transformer detection disabled. "
+                "Install: pip install 'optimum[onnxruntime]>=1.18' 'transformers>=4.40'",
+                UserWarning, stacklevel=2,
+            )
+
         logger.info(
             f"PII Detector ready | mode=SEQUENTIAL_LAYERED "
             f"| presidio={'ON' if self.presidio else 'OFF'} "
@@ -363,59 +382,41 @@ class AdvancedPIIDetector:
             logger.warning(f"Presidio unavailable: {e}")
 
     def _init_transformers(self):
-        """NER transformer — high-accuracy NAME/ORG/LOC."""
+        """NER transformer — high-accuracy NAME/ORG/LOC (ONNX Runtime)."""
         try:
-            from transformers import pipeline
-            try:
-                with _suppress_stdout():
-                    self.ner_pipeline = pipeline(
-                        "ner",
-                        model=self.transformer_model_name,
-                        aggregation_strategy="first",
-                        device=0,
-                        model_kwargs={"dtype": "float16"},
-                    )
-                self.ner_pipeline("warmup")  # GPU kernel compile
-                logger.info(f"NER ready on GPU ({self.transformer_model_name})")
-            except Exception:
-                with _suppress_stdout():
-                    self.ner_pipeline = pipeline(
-                        "ner",
-                        model=self.transformer_model_name,
-                        aggregation_strategy="first",
-                        device=-1,
-                    )
-                self.ner_pipeline("warmup")
-                logger.info(f"NER ready on CPU ({self.transformer_model_name})")
+            from optimum.onnxruntime import ORTModelForTokenClassification
+            from transformers import AutoTokenizer, pipeline
+            with _suppress_stdout():
+                model = ORTModelForTokenClassification.from_pretrained(self.transformer_model_name)
+                tokenizer = AutoTokenizer.from_pretrained(self.transformer_model_name)
+            self.ner_pipeline = pipeline(
+                "ner",
+                model=model,
+                tokenizer=tokenizer,
+                aggregation_strategy="first",
+            )
+            self.ner_pipeline("warmup")
             self._ner_sources_count += 1
+            logger.info(f"NER ready ({self.transformer_model_name})")
         except Exception as e:
             logger.warning(f"NER transformer unavailable: {e}")
 
     def _init_pii_transformer(self):
-        """PII model structured PII detection — PHONE, EMAIL, SSN, DOB, ADDRESS etc."""
+        """PII model structured PII detection — PHONE, EMAIL, SSN, DOB, ADDRESS etc. (ONNX Runtime)."""
         try:
-            from transformers import pipeline
-            try:
-                with _suppress_stdout():
-                    self.pii_pipeline = pipeline(
-                        "token-classification",
-                        model=self.pii_model_name,
-                        aggregation_strategy="first",
-                        device=0,
-                        model_kwargs={"dtype": "float16"},
-                    )
-                self.pii_pipeline("warmup")  # GPU kernel compile
-                logger.info(f"PII model ready on GPU ({self.pii_model_name})")
-            except Exception:
-                with _suppress_stdout():
-                    self.pii_pipeline = pipeline(
-                        "token-classification",
-                        model=self.pii_model_name,
-                        aggregation_strategy="first",
-                        device=-1,
-                    )
-                self.pii_pipeline("warmup")
-                logger.info(f"PII model ready on CPU ({self.pii_model_name})")
+            from optimum.onnxruntime import ORTModelForTokenClassification
+            from transformers import AutoTokenizer, pipeline
+            with _suppress_stdout():
+                model = ORTModelForTokenClassification.from_pretrained(self.pii_model_name)
+                tokenizer = AutoTokenizer.from_pretrained(self.pii_model_name)
+            self.pii_pipeline = pipeline(
+                "token-classification",
+                model=model,
+                tokenizer=tokenizer,
+                aggregation_strategy="first",
+            )
+            self.pii_pipeline("warmup")
+            logger.info(f"PII model ready ({self.pii_model_name})")
         except Exception as e:
             logger.warning(f"PII model unavailable: {e}")
 

{pii_protector-2.2.1 → pii_protector-2.2.2}/pii_protector.egg-info/PKG-INFO

@@ -1,10 +1,10 @@
 Metadata-Version: 2.4
 Name: pii-protector
-Version: 2.2.1
+Version: 2.2.2
 Summary: Production-grade PII detection with multi-model ensemble
 Author-email: tensoryug <jainsatyam26@gmail.com>
 License-Expression: MIT
-Project-URL: Homepage, https://github.com/sjain26/pii-guard
+Project-URL: Homepage, https://github.com/sjain26/pii-detector
 Keywords: pii,privacy,nlp,detection,presidio,data-protection
 Classifier: Development Status :: 5 - Production/Stable
 Classifier: Intended Audience :: Developers
@@ -22,12 +22,12 @@ Requires-Dist: presidio-analyzer>=2.2; extra == "presidio"
 Requires-Dist: spacy>=3.7; extra == "presidio"
 Provides-Extra: transformers
 Requires-Dist: transformers>=4.40; extra == "transformers"
-Requires-Dist: torch>=2.0; extra == "transformers"
+Requires-Dist: optimum[onnxruntime]>=1.18; extra == "transformers"
 Provides-Extra: full
 Requires-Dist: presidio-analyzer>=2.2; extra == "full"
 Requires-Dist: spacy>=3.7; extra == "full"
 Requires-Dist: transformers>=4.40; extra == "full"
-Requires-Dist: torch>=2.0; extra == "full"
+Requires-Dist: optimum[onnxruntime]>=1.18; extra == "full"
 
 # pii-protector
 

{pii_protector-2.2.1 → pii_protector-2.2.2}/pii_protector.egg-info/requires.txt

@@ -3,7 +3,7 @@
 presidio-analyzer>=2.2
 spacy>=3.7
 transformers>=4.40
-torch>=2.0
+optimum[onnxruntime]>=1.18
 
 [presidio]
 presidio-analyzer>=2.2
@@ -11,4 +11,4 @@ spacy>=3.7
 
 [transformers]
 transformers>=4.40
-torch>=2.0
+optimum[onnxruntime]>=1.18

{pii_protector-2.2.1 → pii_protector-2.2.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "pii-protector"
-version = "2.2.1"
+version = "2.2.2"
 description = "Production-grade PII detection with multi-model ensemble"
 readme = "README.md"
 license = "MIT"
@@ -35,24 +35,24 @@ presidio = [
     "presidio-analyzer>=2.2",
     "spacy>=3.7",
 ]
-# Layer 3 & 4: NER transformer + PII model
+# Layer 3 & 4: NER transformer + PII model (ONNX Runtime)
 transformers = [
     "transformers>=4.40",
-    "torch>=2.0",
+    "optimum[onnxruntime]>=1.18",
 ]
 # Install everything
 full = [
     "presidio-analyzer>=2.2",
     "spacy>=3.7",
     "transformers>=4.40",
-    "torch>=2.0",
+    "optimum[onnxruntime]>=1.18",
 ]
 
 [project.scripts]
 pii-detect = "pii_detector.cli:main"
 
 [project.urls]
-Homepage = "https://github.com/sjain26/pii-guard"
+Homepage = "https://github.com/sjain26/pii-detector"
 
 [tool.setuptools.packages.find]
 include = ["pii_detector*"]