Files
Presidio/conf/default.yaml
2025-07-28 15:43:18 +00:00

150 lines
4.8 KiB
YAML

# =====================================================================
# CONFIGURATION PRESIDIO - v28 (FINALE ET OPTIMISÉE)
# =====================================================================
supported_languages: [fr, en]
# 1. CONFIGURATION DU MOTEUR NLP
# =====================================================================
nlp_configuration:
nlp_engine_name: spacy
models:
- lang_code: en
model_name: en_core_web_lg
- lang_code: fr
model_name: fr_core_news_lg
ner_model_configuration:
labels_to_ignore:
- MISC
- CARDINAL
- EVENT
- LANGUAGE
- LAW
- ORDINAL
- PERCENT
- PRODUCT
- QUANTITY
- WORK_OF_ART
confidence_thresholds:
PERSON: 0.85
LOCATION: 0.8
ORGANIZATION: 0.7
# 2. CONFIGURATION DU REGISTRE DES DÉTECTEURS
# =====================================================================
recognizer_registry:
load_predefined_recognizers: true
recognizers:
- name: SpacyRecognizer
supported_language: fr
- name: SpacyRecognizer
supported_language: en
- name: FlexibleDateRecognizer
supported_language: fr
supported_entity: FLEXIBLE_DATE
patterns:
- name: Date format JJ mois AAAA (ex: 15 mars 2025)
regex: "\\b(0?[1-9]|[12][0-9]|3[01])\\s+(janvier|février|mars|avril|mai|juin|juillet|août|septembre|octobre|novembre|décembre)\\s+(19|20)\\d{2}\\b"
score: 1.0
- name: Date format JJ/MM/AAAA (ex: 12/08/1978)
regex: "\\b(0[1-9]|[12][0-9]|3[01])[-/.](0[1-9]|1[012])[-/.](19|20)\\d{2}\\b"
score: 1.0
context: ["date", "né le", "signé le", "incident du"]
- name: BelgianAddressRecognizer
supported_language: fr
supported_entity: BE_ADDRESS
patterns:
- name: Adresse Belge complète
regex: "\\b(?:\\d{1,4}[A-Za-z]?(?:\\s*,)?\\s+)?(?:Avenue|Rue|Boulevard|Chaussée|Place|Quai|Impasse|Drève)(?:\\s+(?:de|la|le|d'|des))?(?:\\s+[A-Z][a-zà-ÿ'-]+)+,?(?:\\s+\\d{1,4}[A-Za-z]?)?,\\s*\\d{4}\\s+[A-Za-zà-ÿ'-]+"
score: 1.0
context: ["demeurant", "adresse", "siège social", "bureaux situés"]
- name: BelgianPhoneRecognizer
supported_language: fr
supported_entity: BE_PHONE_NUMBER
patterns:
- name: Numéro téléphone Belge avec séparateurs
regex: "\\b0[1-9](?:[./\\s]?\\d{2,3}){3}\\b"
score: 0.95
context: ["Tel", "Tél", "téléphone"]
- name: BelgianBankAccountRecognizer
supported_language: fr
supported_entity: BE_BANK_ACCOUNT
patterns:
- name: IBAN Belge
regex: "\\bBE\\d{2}[\\s]?\\d{4}[\\s]?\\d{4}[\\s]?\\d{4}\\b"
score: 1.0
context: ["IBAN", "compte", "bancaires"]
- name: BankOrganizationRecognizer
supported_language: fr
supported_entity: ORGANIZATION
deny_list: ["Banque Nationale de Belgique"]
supported_tokens: ["BNP Paribas Fortis", "CBC Banque", "KBC", "ING", "Belfius"]
score: 0.9
- name: BelgianEnterpriseVatRecognizer
supported_language: fr
supported_entity: BE_ENTERPRISE_VAT_NUMBER
patterns:
- name: Numéro BCE/TVA Belge
regex: "\\b(BE)?\\s?0?\\d{3}[\\.\\s]?\\d{3}[\\.\\s]?\\d{3}\\b"
score: 1.0
context: ["BCE", "TVA", "intracommunautaire"]
- name: SpecificIdRecognizer
supported_language: fr
supported_entity: SPECIFIC_ID
patterns:
- name: Numéro IEC
regex: "(n°\\sIEC:?|IEC:?)\\s*\\d{6}"
score: 1.0
context: ["expert-comptable"]
- name: EmailRecognizer
supported_language: fr
supported_entity: EMAIL_ADDRESS
patterns: [{name: Email Pattern, regex: "\\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\\.[A-Z|a-z]{2,}\\b", score: 1.0}]
context: ["email", "courriel", "mail"]
# 3. LISTE D'EXCLUSION (ALLOW LIST)
# =====================================================================
allow_list:
- Contrat
- SPRL
- Loi
- Euro
# 4. CONFIGURATION DES TRANSFORMATIONS D'ANONYMISATION
# =====================================================================
anonymizer_config:
default_anonymizers:
PERSON: replace
LOCATION: replace
ORGANIZATION: replace
IP_ADDRESS: replace
MONEY: replace
FLEXIBLE_DATE: replace
BE_ADDRESS: replace
BE_PHONE_NUMBER: replace
BE_BANK_ACCOUNT: replace
BE_ENTERPRISE_VAT_NUMBER: replace
SPECIFIC_ID: replace
EMAIL_ADDRESS: replace
replacements:
PERSON: "<PERSONNE>"
LOCATION: "<LIEU>"
ORGANIZATION: "<ORGANISATION>"
IP_ADDRESS: "<ADRESSE_IP>"
MONEY: "<MONTANT>"
FLEXIBLE_DATE: "<DATE>"
BE_ADDRESS: "<ADRESSE_BELGE>"
BE_PHONE_NUMBER: "<TELEPHONE_BE>"
BE_BANK_ACCOUNT: "<COMPTE_BANCAIRE_BE>"
BE_ENTERPRISE_VAT_NUMBER: "<NUM_BCE_TVA_BE>"
SPECIFIC_ID: "<ID_SPECIFIQUE>"
EMAIL_ADDRESS: "<EMAIL>"