Microsoft Presidioでセンシティブな文字列を検出・マスクする

# テキスト・画像共通の解析用パッケージっぽい
!pip install presidio_analyzer

# テキスト秘匿化用パッケージ
!pip install presidio_anonymizer

# 画像秘匿化用パッケージ
!pip install presidio_image_redactor

!sudo apt install tesseract-ocr
!sudo apt install libtesseract-dev

!python -m spacy download en_core_web_lg

from presidio_analyzer import AnalyzerEngine

text="My phone number is 212-555-5555"

# エンジンを初期化してNLPモジュールをロードする
analyzer = AnalyzerEngine()

# analyzerに解析させる
results = analyzer.analyze(
    text=text,
    entities=["PHONE_NUMBER"],
    language='en'
)

print(results)

WARNING:presidio-analyzer:configuration file /usr/local/lib/python3.10/dist-packages/conf/default.yaml not found.  Using default config: {'nlp_engine_name': 'spacy', 'models': [{'lang_code': 'en', 'model_name': 'en_core_web_lg'}]}.
WARNING:presidio-analyzer:configuration file is missing 'ner_model_configuration'. Using default
WARNING:presidio-analyzer:model_to_presidio_entity_mapping is missing from configuration, using default
WARNING:presidio-analyzer:low_score_entity_names is missing from configuration, using default
WARNING:presidio-analyzer:labels_to_ignore is missing from configuration, using default
[type: PHONE_NUMBER, start: 19, end: 31, score: 0.75]

from presidio_anonymizer import AnonymizerEngine

# anonymizerを初期化
anonymizer = AnonymizerEngine()

# テキストとanalyzerの結果をanonymizerに渡す
anonymized_text = anonymizer.anonymize(text=text,analyzer_results=results)

print(anonymized_text)

text: My phone number is <PHONE_NUMBER>
items:
[
    {'start': 19, 'end': 33, 'entity_type': 'PHONE_NUMBER', 'text': '<PHONE_NUMBER>', 'operator': 'replace'}
]

from presidio_image_redactor import ImageRedactorEngine
from PIL import Image

image = Image.open("original_image.png")

redactor = ImageRedactorEngine()
redactor.redact(image=image)

from presidio_analyzer import PatternRecognizer

titles_list = [
    "Sir",
    "Ma'am",
    "Madam",
    "Mr.",
    "Mrs.",
    "Ms.",
    "Miss",
    "Dr.",
    "Professor",
]

# PatternRecognizerを定義する
titles_recognizer = PatternRecognizer(supported_entity="TITLE", deny_list=titles_list)

text1 = "I suspect Professor Plum, in the Dining Room, with the candlestick"

# PatternRecognizerで直接解析する
result = titles_recognizer.analyze(text1, entities=["TITLE"])

print(f"Result:\n {result}")

Result:
 [type: TITLE, start: 10, end: 19, score: 1.0]

from presidio_analyzer import PatternRecognizer
from presidio_analyzer import AnalyzerEngine

titles_list = [
    "Sir",
    "Ma'am",
    "Madam",
    "Mr.",
    "Mrs.",
    "Ms.",
    "Miss",
    "Dr.",
    "Professor",
]

# PatternRecognizerを定義
titles_recognizer = PatternRecognizer(supported_entity="TITLE", deny_list=titles_list)

# AnalyzerEngineの初期化
analyzer = AnalyzerEngine()

# AnalyzerEngineに、定義したPatternRecognizerを追加
analyzer.registry.add_recognizer(titles_recognizer)

text1 = "I suspect Professor Plum, in the Dining Room, with the candlestick."

# AnalyzerEngineで解析
result = analyzer.analyze(text=text1, language="en")

print(f"Result:\n {result}")

Result:
 [type: TITLE, start: 10, end: 19, score: 1.0, type: PERSON, start: 20, end: 24, score: 0.85]

print("Identified these PII entities:")
for r in result:
    print(f"- {text1[r.start:r.end]} as {r.entity_type}")

Identified these PII entities:
- Professor as TITLE
- Plum as PERSON

from presidio_analyzer import Pattern, PatternRecognizer

# `Pattern`オブジェクトを使って正規表現パターンを定義
numbers_pattern = Pattern(name="numbers_pattern", regex="\d+", score=0.5)

# 定義した正規表現パターンからPatternRecognizerを定義。正規表現パターンは複数定義可能。
number_recognizer = PatternRecognizer(
    supported_entity="NUMBER", patterns=[numbers_pattern]
)

text = "I live in 510 Broad st."

numbers_result = number_recognizer.analyze(text=text, entities=["NUMBER"])

print("Result:", numbers_result)
print("Identified these PII entities:")
for r in numbers_result:
    print(f"- {text[r.start:r.end]} as {r.entity_type}")

Result: [type: NUMBER, start: 10, end: 13, score: 0.5]
Identified these PII entities:
- 510 Broad as TITLE
- st. as PERSON

from typing import List
from presidio_analyzer import EntityRecognizer, RecognizerResult
from presidio_analyzer.nlp_engine import NlpArtifacts


class MyRecognizer(EntityRecognizer):
    def load(self) -> None:
        """ロードは不要"""
        pass

    def analyze(
        self, text: str, entities: List[str], nlp_artifacts: NlpArtifacts
    ) -> List[RecognizerResult]:
        """
        特定のPIIを検出するロジックをここに書く
        """
        pass

from typing import List
from presidio_analyzer import EntityRecognizer, RecognizerResult
from presidio_analyzer.nlp_engine import NlpArtifacts


class NumbersRecognizer(EntityRecognizer):

    expected_confidence_level = 0.7  # recognizerが期待する信頼レベル

    def load(self) -> None:
        """ロードは不要"""
        pass

    def analyze(
        self, text: str, entities: List[str], nlp_artifacts: NlpArtifacts
    ) -> List[RecognizerResult]:
        """
        数値表現（123 とか One Two Threeとか)を示すトークンが含まれているかを解析する
        """
        results = []

        # spaCyで分割されたトークンに対し順次 `token.like_num` を呼ぶ
        for token in nlp_artifacts.tokens:
            if token.like_num:
                result = RecognizerResult(
                    entity_type="NUMBER",
                    start=token.idx,
                    end=token.idx + len(token),
                    score=self.expected_confidence_level,
                )
                results.append(result)
        return results


# NumbersRecognizerからrecognizerインスタンスを初期化
new_numbers_recognizer = NumbersRecognizer(supported_entities=["NUMBER"])

from presidio_analyzer import AnalyzerEngine

text = "Roberto lives in Five 10 Broad st."
analyzer = AnalyzerEngine()
analyzer.registry.add_recognizer(new_numbers_recognizer)

numbers_result = analyzer.analyze(text=text, language="en")
print("Results:")
for res in numbers_result:
    print(f"- {str(res)}")
print("Identified these PII entities:")
for r in numbers_result:
    print(f"- {text[r.start:r.end]} as {r.entity_type}")

Results:
- type: PERSON, start: 0, end: 7, score: 0.85
- type: NUMBER, start: 17, end: 21, score: 0.7
- type: NUMBER, start: 22, end: 24, score: 0.7
Identified these PII entities:
- Roberto as PERSON
- Five as NUMBER
- 10 as NUMBER

!python -m spacy download es_core_news_md

import spacy
spacy.cli.download("es_core_news_md")

from presidio_analyzer import AnalyzerEngine
from presidio_analyzer.nlp_engine import NlpEngineProvider


# NLPエンジン名と、言語コードとモデルのマッピング、を設定する
configuration = {
    "nlp_engine_name": "spacy",
    "models": [
        {"lang_code": "es", "model_name": "es_core_news_md"},
        {"lang_code": "en", "model_name": "en_core_web_lg"},
    ],
}

# マッピング設定からNLPエンジンを作成
provider = NlpEngineProvider(nlp_configuration=configuration)
nlp_engine_with_spanish = provider.create_engine()

# 作成したNLPエンジンとサポートする言語をAnalyzerEngineに渡して初期化
analyzer = AnalyzerEngine(
    nlp_engine=nlp_engine_with_spanish, supported_languages=["en", "es"]
)

# スペイン語で解析
text_spanish = "Mi nombre es Morris"
results_spanish = analyzer.analyze(text=text_spanish, language="es")
print("Results from Spanish request:")
print(results_spanish)
print("Identified these PII entities:")
for r in results_spanish:
    print(f"- {text_spanish[r.start:r.end]} as {r.entity_type}")

print()

# 英語で解析
text_english = "My name is Morris"
results_english = analyzer.analyze(text=text_english, language="en")
print("Results from English request:")
print(results_english)
print("Identified these PII entities:")
for r in results_english:
    print(f"- {text_english[r.start:r.end]} as {r.entity_type}")

Results from Spanish request:
[type: PERSON, start: 13, end: 19, score: 0.85]
Identified these PII entities:
- Morris as PERSON

Results from English request:
[type: PERSON, start: 11, end: 17, score: 0.85]
Identified these PII entities:
- Morris as PERSON

from presidio_analyzer import (
    Pattern,
    PatternRecognizer,
    RecognizerRegistry,
    AnalyzerEngine,
)

# 正規表現で郵便番号を検出するPatternRecognizerを作成
regex = r"(\b\d{5}(?:\-\d{4})?\b)"
zipcode_pattern = Pattern(name="zip code (weak)", regex=regex, score=0.01)
zipcode_recognizer = PatternRecognizer(
    supported_entity="US_ZIP_CODE", patterns=[zipcode_pattern]
)

# AnalyzerEngineのレジストリに追加
registry = RecognizerRegistry()
registry.add_recognizer(zipcode_recognizer)
analyzer = AnalyzerEngine(registry=registry)

# テキストを解析
text = "My zip code is 90210"
result = analyzer.analyze(text=text, language="en")
print(f"Result:\n {result}")
print("Identified these PII entities:")
for r in result:
    print(f"- {text[r.start:r.end]} as {r.entity_type}")
print()

Result:
 [type: US_ZIP_CODE, start: 15, end: 20, score: 0.01]
Identified these PII entities:
- 90210 as US_ZIP_CODE

from presidio_analyzer import (
    Pattern,
    PatternRecognizer,
    RecognizerRegistry,
    AnalyzerEngine,
)
from presidio_analyzer import AnalyzerEngine, RecognizerRegistry
from presidio_analyzer.context_aware_enhancers import LemmaContextAwareEnhancer

# 正規表現パターンとコンテキストワードを設定したPatternRecognizerを作成
regex = r"(\b\d{5}(?:\-\d{4})?\b)"
zipcode_pattern = Pattern(name="zip code (weak)", regex=regex, score=0.01)
zipcode_recognizer_w_context = PatternRecognizer(
    supported_entity="US_ZIP_CODE",
    patterns=[zipcode_pattern],
    context=["zip", "zipcode"],     # コンテキストワードの設定
)

# AnalyzerEngineのレジストリに追加
registry = RecognizerRegistry()
registry.add_recognizer(zipcode_recognizer_w_context)
analyzer = AnalyzerEngine(registry=registry)

# テキストを解析
text = "My zip code is 90210"
result = analyzer.analyze(text=text, language="en")
print(f"Result:\n {result}")
print("Identified these PII entities:")
for r in result:
    print(f"- {text[r.start:r.end]} as {r.entity_type}")
print()

Result:
 [type: US_ZIP_CODE, start: 15, end: 20, score: 0.4]
Identified these PII entities:
- 90210 as US_ZIP_CODE

from presidio_analyzer import (
    Pattern,
    PatternRecognizer,
    RecognizerRegistry,
    AnalyzerEngine,
)
from presidio_analyzer import AnalyzerEngine, RecognizerRegistry
from presidio_analyzer.context_aware_enhancers import LemmaContextAwareEnhancer

# 正規表現パターンとコンテキストワードを設定したPatternRecognizerを作成
regex = r"(\b\d{5}(?:\-\d{4})?\b)"
zipcode_pattern = Pattern(name="zip code (weak)", regex=regex, score=0.01)
zipcode_recognizer_w_context = PatternRecognizer(
    supported_entity="US_ZIP_CODE",
    patterns=[zipcode_pattern],
    context=["zip", "zipcode"],     # コンテキストワードの設定
)

# LemmaContextAwareEnhancerオブジェクトを作成して、コンテキスト合致時の加算スコアと最小スコアを設定
context_aware_enhancer = LemmaContextAwareEnhancer(
    context_similarity_factor=0.45,
    min_score_with_context_similarity=0.4
)

# AnalyzerEngineのレジストリに追加
registry = RecognizerRegistry()
registry.add_recognizer(zipcode_recognizer_w_context)
analyzer = AnalyzerEngine(
    registry=registry,
    context_aware_enhancer=context_aware_enhancer,     # スコア設定を行ったLemmaContextAwareEnhancerを指定
)

# テキストを解析
text = "My zip code is 90210"
result = analyzer.analyze(text=text, language="en")
print(f"Result:\n {result}")
print("Identified these PII entities:")
for r in result:
    print(f"- {text[r.start:r.end]} as {r.entity_type}")
print()

Result:
 [type: US_ZIP_CODE, start: 15, end: 20, score: 0.46]
Identified these PII entities:
- 90210 as US_ZIP_CODE

record = {"column_name": "zip", "text": "My code is 90210"}
result = analyzer.analyze(text=record["text"], language="en", context=[record["column_name"]])
print(f"Result:\n {result}")
print("Identified these PII entities:")
for r in result:
    print(f"- {text[r.start:r.end]} as {r.entity_type}")
print()

Result:
 [type: US_ZIP_CODE, start: 11, end: 16, score: 0.46]
Identified these PII entities:
- 90210 as US_ZIP_CODE

from presidio_analyzer import AnalyzerEngine
import pprint

analyzer = AnalyzerEngine()

text="My phone number is 212-555-5555"

results = analyzer.analyze(
    text=text, language="en", return_decision_process=True
)

# 通常の結果出力の場合
# print(results)
# で [type: PHONE_NUMBER, start: 19, end: 31, score: 0.75] が返ってくる。

decision_process = results[0].analysis_explanation

pp = pprint.PrettyPrinter()
print("Decision process output:\n")
pp.pprint(decision_process.__dict__)

Decision process output:

{'original_score': 0.4,
 'pattern': None,
 'pattern_name': None,
 'recognizer': 'ABCMeta',
 'regex_flags': None,
 'score': 0.75,
 'score_context_improvement': 0.35,
 'supportive_context_word': 'phone',
 'textual_explanation': 'Recognized as US region phone number, using '
                        'PhoneRecognizer',
 'validation_result': None}

recognizers:
  -
    name: "Zip code Recognizer"
    supported_language: "de"
    patterns:
      -
         name: "zip code (weak)"
         regex: "(\\b\\d{5}(?:\\-\\d{4})?\\b)"
         score: 0.01
    context:
     - zip
     - code
    supported_entity: "ZIP"
  -
    name: "Titles recognizer"
    supported_language: "en"
    supported_entity: "TITLE"
    deny_list:
      - Mr.
      - Mrs.
      - Ms.
      - Miss
      - Dr.
      - Prof.

from presidio_analyzer import AnalyzerEngine, RecognizerRegistry

yaml_file = "recognizers.yaml"

registry = RecognizerRegistry()
registry.add_recognizers_from_yaml(yaml_file)

analyzer = AnalyzerEngine(registry=registry)

text = "Mr. Plum wrote a book"
result = analyzer.analyze(text=text, language="en")

print("Result:")
for res in result:
    print(f"- {str(res)}")
print()
print("Identified these PII entities:")
for r in result:
    print(f"- {text[r.start:r.end]} as {r.entity_type}")
print()

Result:
- type: TITLE, start: 0, end: 3, score: 1.0

Identified these PII entities:
- Mr. as TITLE

from presidio_analyzer import AnalyzerEngine, RecognizerRegistry

yaml_file = "recognizers.yaml"

registry = RecognizerRegistry()
registry.load_predefined_recognizers()  # すべての事前定義されたrecognizersを読み込む (クレジットカード, 電話番号等)
registry.add_recognizers_from_yaml(yaml_file)

analyzer = AnalyzerEngine(registry=registry)

text = "Mr. Plum wrote a book"
result = analyzer.analyze(text=text, language="en")

print("Result:")
for res in result:
    print(f"- {str(res)}")
print()
print("Identified these PII entities:")
for r in result:
    print(f"- {text[r.start:r.end]} as {r.entity_type}")
print()

Result:
- type: TITLE, start: 0, end: 3, score: 1.0
- type: PERSON, start: 4, end: 8, score: 0.85

Identified these PII entities:
- Mr. as TITLE
- Plum as PERSON

loaded_recognizers = registry.get_recognizers(language="en", all_fields=True)
for r in loaded_recognizers:
    print("{}:\n  {},\n  {}".format(r.name, r.supported_entities, r.context))        

UsBankRecognizer:
  ['US_BANK_NUMBER'],
  ['bankcheck', 'account', 'account#', 'acct', 'save', 'debit']
UsLicenseRecognizer:
  ['US_DRIVER_LICENSE'],
  ['driver', 'license', 'permit', 'lic', 'identification', 'dls', 'cdls', 'lic#', 'driving']
UsItinRecognizer:
  ['US_ITIN'],
  ['individual', 'taxpayer', 'itin', 'tax', 'payer', 'taxid', 'tin']
UsPassportRecognizer:
  ['US_PASSPORT'],
  ['us', 'united', 'states', 'passport', 'passport#', 'travel', 'document']
UsSsnRecognizer:
  ['US_SSN'],
  ['social', 'security', 'ssn', 'ssns', 'ssn#', 'ss#', 'ssid']
NhsRecognizer:
  ['UK_NHS'],
  ['national health service', 'nhs', 'health services authority', 'health authority']
SgFinRecognizer:
  ['SG_NRIC_FIN'],
  ['fin', 'fin#', 'nric', 'nric#']
AuAbnRecognizer:
  ['AU_ABN'],
  ['australian business number', 'abn']
AuAcnRecognizer:
  ['AU_ACN'],
  ['australian company number', 'acn']
AuTfnRecognizer:
  ['AU_TFN'],
  ['tax file number', 'tfn']
AuMedicareRecognizer:
  ['AU_MEDICARE'],
  ['medicare']
InPanRecognizer:
  ['IN_PAN'],
  ['permanent account number', 'pan']
CreditCardRecognizer:
  ['CREDIT_CARD'],
  ['credit', 'card', 'visa', 'mastercard', 'cc ', 'amex', 'discover', 'jcb', 'diners', 'maestro', 'instapayment']
CryptoRecognizer:
  ['CRYPTO'],
  ['wallet', 'btc', 'bitcoin', 'crypto']
DateRecognizer:
  ['DATE_TIME'],
  ['date', 'birthday']
EmailRecognizer:
  ['EMAIL_ADDRESS'],
  ['email']
IbanRecognizer:
  ['IBAN_CODE'],
  ['iban', 'bank', 'transaction']
IpRecognizer:
  ['IP_ADDRESS'],
  ['ip', 'ipv4', 'ipv6']
MedicalLicenseRecognizer:
  ['MEDICAL_LICENSE'],
  ['medical', 'certificate', 'DEA']
PhoneRecognizer:
  ['PHONE_NUMBER'],
  ['phone', 'number', 'telephone', 'cell', 'cellphone', 'mobile', 'call']
UrlRecognizer:
  ['URL'],
  ['url', 'website', 'link']
SpacyRecognizer:
  ['DATE_TIME', 'NRP', 'LOCATION', 'PERSON', 'ORGANIZATION'],
  []
Titles recognizer:
  ['TITLE'],
  None

from presidio_anonymizer import AnonymizerEngine
from presidio_anonymizer.entities import RecognizerResult

text = (
    "Hello, Mr. Williams. My name is John Doe. Call me John. \n"
    "My phone number is 212-555-5555 and email adderss is example@example.com. \n"
    "My credit card number is 1111-2222-3333-4444."
)

analyzer = AnalyzerEngine()

analzer_results = analyzer.analyze(
    text=text,
    language='en',
)

print("Analyzer Results:")
for ar in analzer_results:
    print(f"- {str(ar)}")

print()

# AnonymizerEngineを初期化
engine = AnonymizerEngine()

# テキストとそのテキストのanalyzerの結果をanonymizeメソッドに渡す
result = engine.anonymize(
    text=text, analyzer_results=analzer_results
)

print("De-identified text")
print(result.text)

Analyzer Results:
- type: EMAIL_ADDRESS, start: 110, end: 129, score: 1.0
- type: CREDIT_CARD, start: 157, end: 176, score: 1.0
- type: PERSON, start: 11, end: 19, score: 0.85
- type: PERSON, start: 32, end: 40, score: 0.85
- type: PERSON, start: 50, end: 54, score: 0.85
- type: PHONE_NUMBER, start: 76, end: 88, score: 0.75
- type: URL, start: 118, end: 129, score: 0.5

De-identified text
Hello, Mr. <PERSON>. My name is <PERSON>. Call me <PERSON>. 
My phone number is <PHONE_NUMBER> and email adderss is <EMAIL_ADDRESS>. 
My credit card number is <CREDIT_CARD>.

from presidio_analyzer import AnalyzerEngine, RecognizerRegistry
from presidio_anonymizer import AnonymizerEngine
from presidio_anonymizer.entities import RecognizerResult, OperatorConfig

text = (
    "Hello, Mr. Williams. My name is John Doe. Call me John. \n"
    "My phone number is 212-555-5555 and email adderss is example@example.com. \n"
    "My credit card number is 1111-2222-3333-4444."
)

analyzer = AnalyzerEngine()

analyzer_results = analyzer.analyze(
    text=text,
    language='en',
)

print("Analyzer Results:")
for ar in analyzer_results:
    print(f"- {str(ar)}")

print()

# AnonymizerEngineを初期化
engine = AnonymizerEngine()

# anonymizeの処理内容を定義
operators = {
    # デフォルトは文字列"<ANONYMIZED>"に置き換え
    "DEFAULT": OperatorConfig(
        "replace", {"new_value": "<ANONYMIZED>"}
    ),
    # PERSONの場合は文字列"<SOMEONE>"に置き換え
    "PERSON": OperatorConfig(
        "replace", {"new_value": "<SOMEONE>"}
    ),
    # EMAIL_ADDRESSの場合はmd5ハッシュ化
    "EMAIL_ADDRESS": OperatorConfig(
        "hash", {"hash_type": "md5"}
    ),
    # PHONE_NUMBERの場合は後ろから9文字文を"*"でマスク
    "PHONE_NUMBER": OperatorConfig(
        "mask",
        {
            "type": "mask",
            "masking_char": "*",
            "chars_to_mask": 9,
            "from_end": True,
        },
    ),
    # CREDIT_CARDの場合は先頭から16文字文を"X"でマスク
    "CREDIT_CARD": OperatorConfig(
        "mask",
        {
            "type": "mask",
            "masking_char": "X",
            "chars_to_mask":15,
            "from_end": False,
        },
    ),
}

# テキストとそのテキストのanalyzerの結果をanonymizeメソッドに渡す。
# operatorsでanonymizeの処理定義を設定できる
result = engine.anonymize(
    text=text, analyzer_results=analyzer_results, operators=operators,
)

print("De-identified text")
print(result.text)

Analyzer Results:
- type: EMAIL_ADDRESS, start: 110, end: 129, score: 1.0
- type: CREDIT_CARD, start: 157, end: 176, score: 1.0
- type: PERSON, start: 11, end: 19, score: 0.85
- type: PERSON, start: 32, end: 40, score: 0.85
- type: PERSON, start: 50, end: 54, score: 0.85
- type: PHONE_NUMBER, start: 76, end: 88, score: 0.75
- type: URL, start: 118, end: 129, score: 0.5

De-identified text
Hello, Mr. <SOMEONE>. My name is <SOMEONE>. Call me <SOMEONE>. 
My phone number is 212********* and email adderss is 23463b99b62a72f26ed677cc556c44e8. 
My credit card number is XXXXXXXXXXXXXXX4444.

from presidio_anonymizer import AnonymizerEngine
from presidio_anonymizer.entities import OperatorConfig, EngineResult, RecognizerResult
from faker import Faker

text = "Hello, Mr. Williams. My name is John Doe. Call me John."

analyzer = AnalyzerEngine()

analyzer_results = analyzer.analyze(
    text=text,
    language='en',
)

print("Analyzer Results:")
for ar in analyzer_results:
    print(f"- {str(ar)}")

print()

# fakerを使って架空の名前を返す関数を用意する（引数は使わなくても値を受ける必要がある）
fake = Faker()

def fake_name(x):
    return fake.name()


# PERSONエンティティに対してlambdaを使ってカスタムな処理を定義
operators = {"PERSON": OperatorConfig("custom", {"lambda": fake_name})}

anonymizer = AnonymizerEngine()

anonymized_results = anonymizer.anonymize(
    text=text, analyzer_results=analyzer_results, operators=operators
)

print(anonymized_results.text)

Analyzer Results:
- type: PERSON, start: 11, end: 19, score: 0.85
- type: PERSON, start: 32, end: 40, score: 0.85
- type: PERSON, start: 50, end: 54, score: 0.85

Hello, Mr. Stephanie Gutierrez. My name is Robert Griffin. Call me Marc White.

from presidio_analyzer import AnalyzerEngine

text="My name is James Bond"

analyzer = AnalyzerEngine()

analyzer_results = analyzer.analyze(
    text=text,
    language='en',
)

print("Analyzer Results:")
for ar in analyzer_results:
    print(f"- {str(ar)}")
print("Identified these PII entities:")
for r in analyzer_results:
    print(f"- {text[r.start:r.end]} as {r.entity_type}")
print()

Analyzer Results:
- type: PERSON, start: 11, end: 21, score: 0.85
Identified these PII entities:
- James Bond as PERSON

from presidio_anonymizer import AnonymizerEngine, DeanonymizeEngine
from presidio_anonymizer.entities import (
    RecognizerResult,
    OperatorResult,
    OperatorConfig,
)

# 暗号化・復号化の鍵を設定
crypto_key = "WmZq4t7w!z%C&F)J"

anonymizer = AnonymizerEngine()

# テキストとanalyzerの結果を渡してanonymizeを実行
# OperatorConfigでPERSONを"encrypt"するよう設定
anonymize_result = engine.anonymize(
    text=text,
    analyzer_results=analyzer_results,
    operators={"PERSON": OperatorConfig("encrypt", {"key": crypto_key})},
)

anonymize_result

text: My name is eW/Ql/BgmJCpW36wOHHlYGnxiyAGovKLf6r6UCEhfII=
items:
[
    {'start': 11, 'end': 55, 'entity_type': 'PERSON', 'text': 'eW/Ql/BgmJCpW36wOHHlYGnxiyAGovKLf6r6UCEhfII=', 'operator': 'encrypt'}
]

# DeanonymizerEngineを初期化
engine = DeanonymizeEngine()

# anonymizerdされたテキストとエンティティを渡してdeanonymizeを実行
# OperatorConfigでPERSONを"dencrypt"するよう設定
deanonymized_result = engine.deanonymize(
    text=anonymize_result.text,
    entities=anonymize_result.items, 
    operators={"DEFAULT": OperatorConfig("decrypt", {"key": crypto_key})},
)

deanonymized_result

text: My name is James Bond
items:
[
    {'start': 11, 'end': 21, 'entity_type': 'PERSON', 'text': 'James Bond', 'operator': 'decrypt'}
]

from presidio_anonymizer.operators import Decrypt

# anonymizerdされたエンティティのテキストを取得
encrypted_entity_value = anonymize_result.items[0].text

# Decryptで直接復号化
Decrypt().operate(text=encrypted_entity_value, params={"key": crypto_key})

from presidio_analyzer import AnalyzerEngine

websites_list = [
    "bing.com",
]

text = "My favorite website is bing.com, his is microsoft.com"

analyzer = AnalyzerEngine()

result = analyzer.analyze(
    text = text,
    language = 'en',
    allow_list = websites_list     # allow_listで指定する
)

print("Analyzer Results:")
for ar in result:
    print(f"- {str(ar)}")
print("Identified these PII entities:")
for r in result:
    print(f"- {text[r.start:r.end]} as {r.entity_type}")
print()

Analyzer Results:
- type: URL, start: 40, end: 53, score: 0.85
Identified these PII entities:
- microsoft.com as URL

!pip install presidio_analyzer
!pip install presidio_anonymizer
!pip install presidio_image_redactor

!python -m spacy download ja_core_news_sm
!python -m spacy download ja_core_news_md
!python -m spacy download ja_core_news_lg
!python -m spacy download ja_core_news_trf

!sudo apt install tesseract-ocr
!sudo apt install libtesseract-dev

from presidio_analyzer import AnalyzerEngine
from presidio_anonymizer import AnonymizerEngine
from presidio_analyzer.nlp_engine import NlpEngineProvider

configuration = {
    "nlp_engine_name": "spacy",
    "models": [
        {"lang_code": "ja", "model_name": "ja_core_news_lg"},
    ],
}

provider = NlpEngineProvider(nlp_configuration=configuration)
nlp_engine = provider.create_engine()

analyzer = AnalyzerEngine(
    nlp_engine=nlp_engine, supported_languages=["ja"]
)

anonymizer = AnonymizerEngine()

text = """\
ドウデュース（欧字名:Do Deuce、2019年5月7日 - ）は、日本の競走馬。主な勝ち鞍は2021年の朝日杯フューチュリティステークス、2022年の東京優駿、2023年の有馬記念。

馬名の意味は「する＋テニス用語（勝利目前の意味）」。2021年のJRA賞最優秀2歳牡馬である。

戦績
デビュー前
2019年5月7日、北海道安平町のノーザンファームで誕生。松島正昭が代表を務める株式会社キーファーズの所有馬となり、ノーザンファーム空港牧場で育成の後、栗東トレーニングセンターの友道康夫厩舎に入厩した。

2歳（2021年）
9月5日に小倉競馬場で行われた2歳新馬戦（芝1800メートル）に武豊鞍上で出走。1番人気に推されると、レースは直線でガイアフォースとの追い比べをクビ差制してデビュー勝ちを果たした。

次走はリステッド競走のアイビーステークスを選択。2番人気に推され、レースでは追い比べから抜け出すと、最後は追い込んできたグランシエロをクビ差凌いで優勝、デビュー2連勝とした。

続いて朝日杯フューチュリティステークスに出走。重賞勝ち馬セリフォスやジオグリフをはじめとした自身と同じ無敗馬が多く顔を揃える中、3番人気に支持される。レースでは直線で外に出すと、先に抜け出していたセリフォスを半馬身差で差し切り優勝、無傷3連勝でGI初制覇を果たした。鞍上の武豊はこの競走22回目の挑戦で初制覇となり、日本の中央競馬 (JRA) の平地GI完全制覇までホープフルステークスを残すのみとした。また馬主である松島及びキーファーズにとっては初の単独所有馬によるGI勝利、並びに国内GI初制覇となった。\
"""

analyzer_result = analyzer.analyze(
    text=text,
    language="ja"
)

print("Results:")
for res in analyzer_result:
    print(f"- {str(res)}")
print("Identified these PII entities:")
for r in analyzer_result:
    print(f"- {text[r.start:r.end]} as {r.entity_type}")

Results:
- type: DATE_TIME, start: 20, end: 29, score: 0.85
- type: LOCATION, start: 35, end: 37, score: 0.85
- type: DATE_TIME, start: 48, end: 53, score: 0.85
- type: DATE_TIME, start: 71, end: 76, score: 0.85
- type: DATE_TIME, start: 82, end: 87, score: 0.85
- type: DATE_TIME, start: 121, end: 126, score: 0.85
- type: DATE_TIME, start: 153, end: 162, score: 0.85
- type: LOCATION, start: 163, end: 166, score: 0.85
- type: PERSON, start: 182, end: 186, score: 0.85
- type: PERSON, start: 242, end: 246, score: 0.85
- type: DATE_TIME, start: 259, end: 264, score: 0.85
- type: DATE_TIME, start: 266, end: 270, score: 0.85
- type: PERSON, start: 324, end: 327, score: 0.85
- type: LOCATION, start: 418, end: 424, score: 0.85
- type: LOCATION, start: 475, end: 480, score: 0.85
- type: PERSON, start: 481, end: 486, score: 0.85
- type: LOCATION, start: 545, end: 550, score: 0.85
- type: PERSON, start: 583, end: 585, score: 0.85
- type: LOCATION, start: 605, end: 607, score: 0.85
- type: LOCATION, start: 656, end: 658, score: 0.85
Identified these PII entities:
- 2019年5月7日 as DATE_TIME
- 日本 as LOCATION
- 2021年 as DATE_TIME
- 2022年 as DATE_TIME
- 2023年 as DATE_TIME
- 2021年 as DATE_TIME
- 2019年5月7日 as DATE_TIME
- 北海道 as LOCATION
- 松島正昭 as PERSON
- 友道康夫 as PERSON
- 2021年 as DATE_TIME
- 9月5日 as DATE_TIME
- ガイア as PERSON
- グランシエロ as LOCATION
- セリフォス as LOCATION
- ジオグリフ as PERSON
- セリフォス as LOCATION
- 武豊 as PERSON
- 日本 as LOCATION
- 松島 as LOCATION

result = anonymizer.anonymize(
    text=text,
    analyzer_results=analyzer_result
)

print(result.text)

- 2019年5月7日 as DATE_TIME
- 日本 as LOCATION
- 2021年 as DATE_TIME
- 2022年 as DATE_TIME
- 2023年 as DATE_TIME
- 2021年 as DATE_TIME
- 2019年5月7日 as DATE_TIME
- 北海道 as LOCATION
- 安平町 as LOCATION
- 松島正昭 as PERSON
- 友道康夫 as PERSON
- 2021年 as DATE_TIME
- 9月5日 as DATE_TIME
- 武豊鞍上 as PERSON
- ガイア as PERSON
- グランシエロ as LOCATION
- 鞍上 as PERSON
- 武豊 as PERSON
- 日本 as LOCATION
- JRA as LOCATION
- 松島 as PERSON

- 2019年5月7日 as DATE_TIME
- 日本 as LOCATION
- 2021年 as DATE_TIME
- 2022年 as DATE_TIME
- 2023年 as DATE_TIME
- 2021年 as DATE_TIME
- 2019年5月7日 as DATE_TIME
- 北海道安平町 as LOCATION
- 松島正昭 as PERSON
- 友道康夫厩舎 as PERSON
- 2021年 as DATE_TIME
- 9月5日 as DATE_TIME
- ガイア as PERSON
- ジオグリフ as PERSON
- 鞍上 as PERSON
- 武豊 as PERSON
- 日本 as LOCATION

Identified these PII entities:
- ドウデュース as PERSON
- 2019年5月7日 as DATE_TIME
- 日本 as LOCATION
- 2021年 as DATE_TIME
- 2022年 as DATE_TIME
- 2023年 as DATE_TIME
- 2021年 as DATE_TIME
- 2019年5月7日 as DATE_TIME
- 北海道安平町 as LOCATION
- ノーザンファーム as LOCATION
- 松島正昭 as PERSON
- 友道康夫 as PERSON
- 2021年 as DATE_TIME
- 9月5日 as DATE_TIME
- 武豊 as PERSON
- ガイアフォース as PERSON
- セリフォス as PERSON
- ジオグリフ as PERSON
- セリフォス as PERSON
- 武豊 as PERSON
- 日本 as LOCATION
- 松島 as PERSON
- キーファーズ as PERSON