From 8e58267bde76586a9755497edc1f14a3cf053366 Mon Sep 17 00:00:00 2001 From: nacim Date: Mon, 28 Jul 2025 16:47:31 +0000 Subject: [PATCH] Actualiser conf/default.yaml --- conf/default.yaml | 123 +++++++++++++++++++++++++--------------------- 1 file changed, 67 insertions(+), 56 deletions(-) diff --git a/conf/default.yaml b/conf/default.yaml index 3317997..c8ad252 100644 --- a/conf/default.yaml +++ b/conf/default.yaml @@ -1,9 +1,9 @@ # ===================================================================== -# CONFIGURATION PRESIDIO - v26 +# CONFIGURATION PRESIDIO - v27 # ===================================================================== supported_languages: [en, fr] -# 1. CONFIGURATION DU MOTEUR NLP +# 1. CONFIGURATION DU MOTEUR NLP (INCHANGÉ) # ===================================================================== nlp_configuration: nlp_engine_name: spacy @@ -27,81 +27,92 @@ nlp_configuration: confidence_thresholds: DEFAULT_CONFIDENCE: 0.65 PERSON: 0.85 - LOCATION: 0.85 + LOCATION: 0.4 ORGANIZATION: 0.55 -# 2. CONFIGURATION DU REGISTRE DES DÉTECTEURS +# 2. CONFIGURATION DU REGISTRE DES DÉTECTEURS (AVEC L'AJOUT POUR LES WARNINGS) # ===================================================================== recognizer_registry: load_predefined_recognizers: true - - # --- DÉTECTEURS --- + recognizers: - - - name: CustomDateRecognizer + # --- On déclare le détecteur NLP de base --- + + - name: FlexibleDateRecognizer supported_language: fr - supported_entity: CUSTOM_DATE + supported_entity: FLEXIBLE_DATE patterns: - - name: Date JJ/MM/AAAA + - name: Date format JJ mois AAAA + 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 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 de naissance", "né le"] + context: ["date", "né le", "signé le", "incident du"] - - name: EmailRecognizer + - name: BelgianAddressRecognizer 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"] - - # --- VOTRE REGEX TÉLÉPHONE --- - - name: CustomBelgianPhoneRecognizer + 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: Belgian Phone Pattern (votre regex adaptée) - - regex: "\\b(((\\+|00)32[ ]?(?:\\(0\\)[ ]?)?)|0)?4\\d{2}(\\s?\\d{2}[\\s\\.]?){3}\\b" - score: 1.0 - context: ["téléphone", "tel", "gsm", "mobile"] - - - name: CustomCreditCardRecognizer - supported_language: fr - supported_entity: CREDIT_CARD_NUMBER - patterns: - - name: Credit Card with spaces (Visa, Mastercard) - regex: "\\b(?:4[0-9]{3}(?:[ -]?[0-9]{4}){3}|5[1-5][0-9]{2}(?:[ -]?[0-9]{4}){3})\\b" + - name: Numéro téléphone Belge (fixe ou mobile) + regex: "\\b0[1-9](?:[./\\s]?\\d{2,3}){3}\\b" score: 0.95 - context: ["carte", "visa", "mastercard"] - - - name: IbanRecognizer - supported_language: fr - supported_entity: IBAN - patterns: [{name: IBAN Pattern, regex: "\\b[A-Z]{2}[0-9]{2}\\s?(?:[A-Z0-9]{4}\\s?){2,7}[A-Z0-9]{1,4}\\b", score: 0.95}] - context: ["iban", "compte"] - + context: ["Tel", "Tél", "téléphone", "gsm", "mobile"] - name: SmartOrganizationRecognizer supported_language: fr supported_entity: ORGANIZATION patterns: - - name: Company Name with Legal Form - regex: "\\b([A-Z][a-zà-ÿ]+(?:\\s[A-Z][a-zà-ÿ]+)*)\\s+(SPRL|SARL|SA|SCS|SNC)\\b" + - name: Nom + Forme légale (DigitalConsult SPRL) + regex: "\\b([A-Z][a-zà-ÿ]+(?:\\s[A-Z][a-zà-ÿ]+)*)\\s+(SPRL|SRL|SA|SCS|SNC)\\b" + score: 0.9 + - name: Forme légale + Nom (SPRL DigitalConsult) + regex: "\\b(SPRL|SRL|SA|SCS|SNC)\\s+([A-Z][a-zà-ÿ]+(?:\\s[A-Z][a-zà-ÿ]+)*)\\b" + score: 0.9 + context: ["société", "entreprise", "gérant de la"] + + - name: ProfessionalIdRecognizer + supported_language: fr + supported_entity: BE_PRO_ID + patterns: + - name: Numéro IEC + regex: "(n°\\sIEC:?|IEC:?)\\s*\\d{6}" score: 1.0 - context: ["société", "entreprise"] + context: ["expert-comptable"] - # --- NUMÉROS D'IDENTIFICATION --- - name: BelgianEnterpriseRecognizer supported_language: fr supported_entity: BE_ENTERPRISE_NUMBER - patterns: [{name: BE Enterprise Number, regex: "\\bBE\\s?0\\d{3}[\\.\\s]?\\d{3}[\\.\\s]?\\d{3}\\b", score: 0.95}] - context: ["numéro d'entreprise", "BCE"] + patterns: + - name: Numéro BCE/TVA Belge (avec ou sans BE) + regex: "\\b(BE)?\\s?0?\\d{3}[\\.\\s]?\\d{3}[\\.\\s]?\\d{3}\\b" + score: 1.0 + context: ["BCE", "TVA", "intracommunautaire"] + - 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"] + - name: IbanRecognizer + supported_language: fr + supported_entity: IBAN + patterns: [{name: IBAN Pattern, regex: "\\b[A-Z]{2}[0-9]{2}\\s?(?:[A-Z0-9]{4}\\s?){2,7}[A-Z0-9]{1,4}\\b", score: 0.95}] + context: ["iban", "compte"] - name: BelgianNRNRecognizer supported_language: fr supported_entity: BE_NATIONAL_REGISTER_NUMBER patterns: [{name: NRN Pattern, regex: "\\b[0-9]{2}\\.[0-9]{2}\\.[0-9]{2}-[0-9]{3}\\.[0-9]{2}\\b", score: 1.0}] context: ["registre national"] - - name: FrenchINSEERecognizer supported_language: fr supported_entity: FR_SOCIAL_SECURITY_NUMBER @@ -111,7 +122,7 @@ recognizer_registry: score: 0.95 context: ["sécurité sociale", "insee", "nir"] -# 3. LISTE D'EXCLUSION +# 3. LISTE D'EXCLUSION # ===================================================================== allow_list: - Adresse @@ -159,40 +170,40 @@ allow_list: - Coordonnées - Témoins + # 4. CONFIGURATION DES TRANSFORMATIONS D'ANONYMISATION # ===================================================================== anonymizer_config: default_anonymizers: - # Entités détectées par l'IA (Spacy) et les règles de base de Presidio PERSON: replace LOCATION: replace ORGANIZATION: replace DATE_TIME: replace MONEY: replace - - # Entités détectées par nos règles personnalisées - CUSTOM_DATE: replace - CREDIT_CARD_NUMBER: replace EMAIL_ADDRESS: replace IBAN: replace BE_ENTERPRISE_NUMBER: replace BE_NATIONAL_REGISTER_NUMBER: replace FR_SOCIAL_SECURITY_NUMBER: replace - SWIFT_CODE: replace BE_PHONE_NUMBER: replace + FLEXIBLE_DATE: replace + BE_ADDRESS: replace + BE_PRO_ID: replace + IP_ADDRESS: replace replacements: PERSON: "" LOCATION: "" ORGANIZATION: "" DATE_TIME: "" - CUSTOM_DATE: "" - CREDIT_CARD_NUMBER: "" MONEY: "" EMAIL_ADDRESS: "" IBAN: "" BE_ENTERPRISE_NUMBER: "" BE_NATIONAL_REGISTER_NUMBER: "" FR_SOCIAL_SECURITY_NUMBER: "" - SWIFT_CODE: "" - BE_PHONE_NUMBER: "" + BE_PHONE_NUMBER: "" + FLEXIBLE_DATE: "" + BE_ADDRESS: "" + BE_PRO_ID: "" + IP_ADDRESS: "" \ No newline at end of file