fix: align with presidio API

.pre-commit-config.yaml

@@ -24,20 +24,20 @@ repos:
         args: [ --pytest-test-first ]
     # -   id: check-added-large-files
 -   repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.16.3
+    rev: v0.16.8
     hooks:
       - id: ruff
       # Run the formatter
       - id: ruff-format
         exclude: docs/|sphinx/
 -   repo: https://github.com/allganize/ty-pre-commit
     # Ty version.
-    rev: "v0.0.72"
+    rev: "v0.0.82"
     hooks:
       # Run the type checker.
       - id: ty-check
 -   repo: https://github.com/abravalheri/validate-pyproject
-    rev: 'v0.25'
+    rev: '0.26'
     hooks:
       - id: validate-pyproject
         # Optional extra validations from SchemaStore:

pyproject.toml

@@ -37,7 +37,6 @@ dependencies = [
     "pyarrow==24.0.0",
     "python-dotenv==1.2.2",
     "pytz==2024.1",
-    # "safetensors==0.4.4",
     "scipy>=1.17.1",
     "spacy==3.8.14",
     "SPARQLWrapper==2.0.0",
@@ -58,6 +57,9 @@ rest = [
 # flair = [
 #   "flair>=0.15.1",
 # ]
+presidio = [
+  "presidio-analyzer==2.2.364",
+]
 dev = [
   "bandit",
   "faker",
@@ -178,6 +180,6 @@ omit = [
 [tool.coverage.report]
 # Regress guard: overall coverage must not drop below this.
 # Update this value when coverage genuinely improves.
-fail_under = 33
+fail_under = 80
 show_missing = true
 skip_covered = false

src/risk_assessment/classification/unstructured/presidio.py

@@ -20,16 +20,20 @@ def __init__(
             default_score_threshold: float = 0,
             supported_languages: list[str] | None = None,
         ):
+            super().__init__(type_mapping)
             if supported_languages is None:
                 supported_languages = ["en"]
             self.type_mapping = type_mapping
 
-            self.analyzer = AnalyzerEngine(
-                registry=registry,
-                nlp_engine=nlp_engine,
-                default_score_threshold=default_score_threshold,
-                supported_languages=supported_languages,
-            )
+            kwargs: dict = {
+                "default_score_threshold": default_score_threshold,
+                "supported_languages": supported_languages,
+            }
+            if registry is not None:
+                kwargs["registry"] = registry
+            if nlp_engine is not None:
+                kwargs["nlp_engine"] = nlp_engine
+            self.analyzer = AnalyzerEngine(**kwargs)
 
         def extract(self, text: str) -> list[Entity]:
             return [

src/risk_assessment/readi/text_tokenizer.py

@@ -17,10 +17,11 @@
 
 import warnings
 from abc import ABC, abstractmethod
+from typing import cast
 
 import MeCab
 from nltk.tokenize import WordPunctTokenizer
-from transformers import AutoTokenizer
+from transformers import AutoTokenizer, PreTrainedTokenizer, PreTrainedTokenizerFast
 
 from risk_assessment.readi.sentence_tokenizer import SentenceTokenizer
 
@@ -151,7 +152,10 @@ def __init__(self, model_name: str = "FacebookAI/roberta-base", device: str = "c
             device: Device string for downstream model inference. Defaults to ``"cpu"``.
         """
         self.device = device
-        self.tokenizer = AutoTokenizer.from_pretrained(model_name)  # nosec
+        self.tokenizer: PreTrainedTokenizer | PreTrainedTokenizerFast = cast(
+            PreTrainedTokenizer | PreTrainedTokenizerFast,
+            AutoTokenizer.from_pretrained(model_name),  # nosec
+        )
 
     def span_tokenize(self, text: str) -> list[tuple[int, int]]:
         tokenized_text = self.tokenizer(text)