custom anonymizer as class

.gitignore

@@ -1,3 +1,2 @@
-__pycache__/*
 .ipynb_checkpoints
 __pycache__

app.py

@@ -9,8 +9,8 @@ import pandas as pd
 import streamlit as st
 from annotated_text import annotated_text
 
-from piilo.engines.analyzer import prepare_analyzer
-from piilo.engines.anonymizer import surrogate_anonymizer
+from piilo.engines.analyzer import CustomAnalyzer
+from piilo.engines.anonymizer import SurrogateAnonymizer
 
 os.environ['TOKENIZERS_PARALLELISM'] = 'false'
 warnings.filterwarnings('ignore')
@@ -26,14 +26,12 @@ def analyzer_engine():
             {'lang_code': 'en', 'model_name': 'en_student_name_detector'}],
     }
 
-    analyzer = prepare_analyzer(configuration)
-
-    return analyzer
+    return CustomAnalyzer(configuration=configuration)
 
 @st.cache(allow_output_mutation=True)
 def anonymizer_engine():
     '''Return generate surrogate anonymizer.'''
-    return surrogate_anonymizer()
+    return SurrogateAnonymizer()
 
 def annotate(text, st_analyze_results, st_entities):
     tokens = []

piilo/engines/analyzer.py

@@ -117,25 +117,27 @@ class CustomSpacyRecognizer(LocalRecognizer):
             [entity in egrp and label in lgrp for egrp, lgrp in check_label_groups]
         )
 
-def prepare_analyzer(configuration):
-    '''Handle Preparation of Analyzer Engine for Presidio.'''
-
-    spacy_recognizer = CustomSpacyRecognizer()
-
-    # Create NLP engine based on configuration
-    provider = NlpEngineProvider(nlp_configuration=configuration)
-    nlp_engine = provider.create_engine()
-
-    # add rule-based recognizers
-    registry = RecognizerRegistry()
-    registry.load_predefined_recognizers(nlp_engine=nlp_engine)
-    registry.add_recognizer(spacy_recognizer)
+class CustomAnalyzer(AnalyzerEngine):
+    '''Custom Analyzer Engine for Presidio.'''
+    
+    def __init__(self, configuration):
+        
+        spacy_recognizer = CustomSpacyRecognizer()
+        
+        # Create NLP engine based on configuration
+        provider = NlpEngineProvider(nlp_configuration=configuration)
+        nlp_engine = provider.create_engine()
 
-    # remove the nlp engine we passed, to use custom label mappings
-    registry.remove_recognizer('SpacyRecognizer')
+        # add rule-based recognizers
+        registry = RecognizerRegistry()
+        registry.load_predefined_recognizers(nlp_engine=nlp_engine)
+        registry.add_recognizer(spacy_recognizer)
 
-    analyzer = AnalyzerEngine(nlp_engine=nlp_engine,
-                              registry=registry,
-                              supported_languages=['en'])
+        # remove the nlp engine we passed, to use custom label mappings
+        registry.remove_recognizer('SpacyRecognizer')
 
-    return analyzer
+        super().__init__(
+            nlp_engine=nlp_engine,
+            registry=registry,
+            supported_languages=['en']
+            )

piilo/engines/anonymizer.py

@@ -14,6 +14,7 @@ data = Path(__file__).parent.parent.parent / 'data'
 name_table = data / 'ascii_names.parquet'
 
 logger = logging.getLogger('anonymizer')
+
 class NameDatabase(NameDataset):
     '''A wrapper around the names_dataset.NameDataset class.
     '''
@@ -45,7 +46,7 @@ class NameDatabase(NameDataset):
         country = NameWrapper(self.search(last_names)).country
         return country if country else None
     
-class surrogate_anonymizer(AnonymizerEngine):
+class SurrogateAnonymizer(AnonymizerEngine):
     '''A wrapper around the presidio_anonymizer.AnonymizerEngine class.
     '''
     

piilo/main.py

@@ -5,8 +5,8 @@ import logging
 from fastapi import FastAPI
 from fastapi.middleware.cors import CORSMiddleware
 
-from engines.analyzer import prepare_analyzer
-from engines.anonymizer import surrogate_anonymizer
+from engines.analyzer import CustomAnalyzer
+from engines.anonymizer import SurrogateAnonymizer
 from models.anonymize import AnonymizeRequest, AnonymizeResponse
 
 configuration = {
@@ -19,8 +19,8 @@ logger = logging.getLogger('api')
 logging.basicConfig(level=logging.INFO)
 
 logger.info("Loading Custom Presidio Analyzer and Anonymizer...")
-analyzer = prepare_analyzer(configuration)
-anonymizer = surrogate_anonymizer()
+analyzer = CustomAnalyzer(configuration)
+anonymizer = SurrogateAnonymizer()
 logger.info("Loading Successful!")
 
 app = FastAPI()