# Recognizer pour SIRET/SIREN français
recognizer_registry:
  recognizers:
    - name: FrenchSIRETRecognizer
      supported_language: fr
      supported_entity: FR_SIRET
      patterns:
        - name: SIRET complet
          regex: "\\b[0-9]{3}\\s?[0-9]{3}\\s?[0-9]{3}\\s?[0-9]{5}\\b"
          score: 0.9
        - name: SIREN
          regex: "\\b[0-9]{3}\\s?[0-9]{3}\\s?[0-9]{3}\\b"
          score: 0.85
        - name: SIRET avec espaces
          regex: "\\bSIRET\\s*:?\\s*[0-9]{3}\\s[0-9]{3}\\s[0-9]{3}\\s[0-9]{5}\\b"
          score: 0.95
      context: ["SIRET", "SIREN", "établissement", "entreprise", "société"]