@doccloak/core 0.8.0

package/rules/es.json

@@ -0,0 +1,158 @@
+{
+  "region": "es",
+  "rules": [
+    {
+      "id": "regex:es:dni",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{8}[\\s-]?[A-Z]\\b",
+      "flags": "gi",
+      "confidence": 0.9,
+      "domains": [
+        "identity"
+      ],
+      "description": "Spanish DNI (8 digits + letter)",
+      "examples": [
+        "12345678Z",
+        "12345678-Z"
+      ],
+      "validate": "dni"
+    },
+    {
+      "id": "regex:es:nie",
+      "entityType": "SSN",
+      "pattern": "\\b[XYZ][\\s-]?\\d{7}[\\s-]?[A-Z]\\b",
+      "flags": "gi",
+      "confidence": 0.9,
+      "domains": [
+        "identity"
+      ],
+      "description": "Spanish NIE for foreigners (X/Y/Z + 7 digits + letter)",
+      "examples": [
+        "X1234567L",
+        "Y-1234567-A"
+      ]
+    },
+    {
+      "id": "regex:es:cif",
+      "entityType": "SSN",
+      "pattern": "\\b[A-H]\\d{7}[0-9A-J]\\b",
+      "flags": "gi",
+      "confidence": 0.8,
+      "domains": [
+        "financial",
+        "legal"
+      ],
+      "description": "Spanish CIF company tax ID",
+      "examples": [
+        "A12345678",
+        "B87654321"
+      ]
+    },
+    {
+      "id": "regex:es:cuenta_bancaria",
+      "entityType": "SSN",
+      "pattern": "\\b(?:ES)?\\d{2}\\s?\\d{4}\\s?\\d{4}\\s?\\d{4}\\s?\\d{4}\\s?\\d{4}\\b",
+      "flags": "gi",
+      "confidence": 0.8,
+      "domains": [
+        "financial"
+      ],
+      "description": "Spanish bank account number (CCC, 20 digits)",
+      "examples": [
+        "ES12 1234 5678 9012 3456 7890"
+      ]
+    },
+    {
+      "id": "regex:es:phone",
+      "entityType": "PHONE",
+      "pattern": "\\b(?:\\+?34[\\s-]?)?[6789]\\d{2}[\\s-]?\\d{3}[\\s-]?\\d{3}\\b",
+      "flags": "g",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "Spanish phone number (9 digits, optionally with +34)",
+      "examples": [
+        "+34 612 345 678",
+        "912 345 678"
+      ]
+    },
+    {
+      "id": "regex:es:postal",
+      "entityType": "ADDRESS",
+      "pattern": "\\b\\d{5}\\s+[\\p{L}][\\p{L}\\s'-]+\\b",
+      "flags": "gu",
+      "confidence": 0.75,
+      "domains": [
+        "contact"
+      ],
+      "description": "Spanish postal code (5 digits) followed by city name",
+      "examples": [
+        "28001 Madrid",
+        "08001 Barcelona"
+      ]
+    },
+    {
+      "id": "regex:es:currency_words",
+      "entityType": "CURRENCY",
+      "pattern": "(?:(?:en\\s+)?(?:letras?|palabras?)\\s*:\\s*|(?:la\\s+)?(?:cantidad|suma)\\s+de\\s+)[\\p{L}\\s-]+(?:euros?|céntimos?|pesos?|centavos?)\\b",
+      "flags": "giu",
+      "confidence": 0.9,
+      "domains": [
+        "financial"
+      ],
+      "description": "Spanish amount written in words (e.g., \"en letras: ocho mil quinientos euros\")",
+      "examples": [
+        "en letras: ocho mil quinientos euros"
+      ]
+    },
+    {
+      "id": "regex:es:street",
+      "entityType": "ADDRESS",
+      "pattern": "(?:(?:C(?:alle)?|Av(?:enida|da)?|Avda|Pza|Plaza|Pl|Paseo|P\\.º|Ronda|Ctra|Carretera|Camino|Travesía|Glorieta)\\.?\\s+)[\\p{L}][\\p{L}\\s'-]+(?:,?\\s*(?:n\\.?º?\\s*)?\\d+)?",
+      "flags": "giu",
+      "confidence": 0.85,
+      "domains": [
+        "contact"
+      ],
+      "description": "Spanish street address (Calle/Avenida/Plaza + name + optional number)",
+      "examples": [
+        "Calle Mayor 15",
+        "Av. de la Constitución 3",
+        "Plaza de España"
+      ]
+    },
+    {
+      "id": "regex:es:company",
+      "entityType": "COMPANY",
+      "pattern": "[\\p{L}][\\p{L}\\s&.']+(?:\\s+)?(?:S\\.L\\.U\\.|S\\.L\\.U|S\\.L\\.|S\\.L|S\\.A\\.U\\.|S\\.A\\.U|S\\.A\\.|S\\.A|S\\.C\\.)\\b",
+      "flags": "gu",
+      "confidence": 0.9,
+      "domains": [
+        "legal",
+        "financial"
+      ],
+      "description": "Spanish company name with legal form (S.L., S.A., S.L.U., S.A.U., S.C.)",
+      "examples": [
+        "Telefónica S.A.",
+        "Construcciones García S.L.",
+        "Inditex S.A."
+      ]
+    },
+    {
+      "id": "regex:es:nuss",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{2}[-/]?\\d{8}[-/]?\\d{2}\\b",
+      "flags": "g",
+      "confidence": 0.75,
+      "domains": [
+        "medical",
+        "hr"
+      ],
+      "description": "Spanish Social Security Number (NUSS/NAF, 12 digits: province + sequential + check)",
+      "examples": [
+        "28-12345678-56"
+      ]
+    }
+  ]
+}

package/rules/fi.json

@@ -0,0 +1,43 @@
+{
+  "region": "fi",
+  "rules": [
+    {
+      "id": "regex:fi:hetu",
+      "entityType": "SSN",
+      "pattern": "\\b(?:0[1-9]|[12]\\d|3[01])(?:0[1-9]|1[0-2])\\d{2}[ABCDEFYXWVU+-]\\d{3}[\\dA-FHJK-NPR-Y]\\b",
+      "flags": "gi",
+      "confidence": 0.9,
+      "domains": [
+        "identity",
+        "hr"
+      ],
+      "description": "Finnish personal identity code (henkilötunnus / HETU)",
+      "examples": [
+        "131052-308T"
+      ],
+      "validate": "hetu"
+    },
+    {
+      "id": "regex:fi:phone",
+      "entityType": "PHONE",
+      "pattern": "\\b(?:\\+?358[\\s-]?|0)[1-9]\\d?[\\s-]?\\d{3,4}[\\s-]?\\d{3,4}\\b",
+      "flags": "g",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "Finnish phone number"
+    },
+    {
+      "id": "regex:fi:postal-city",
+      "entityType": "ADDRESS",
+      "pattern": "\\b\\d{5}\\s+[\\p{L}][\\p{L}\\s-]+\\b",
+      "flags": "gu",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "Finnish postal code followed by city name"
+    }
+  ]
+}

package/rules/fr.json

@@ -0,0 +1,178 @@
+{
+  "region": "fr",
+  "rules": [
+    {
+      "id": "regex:fr:nir",
+      "entityType": "SSN",
+      "pattern": "\\b[12]\\s?\\d{2}\\s?\\d{2}\\s?\\d{2}\\s?\\d{3}\\s?\\d{3}(?:\\s?\\d{2})?\\b",
+      "flags": "g",
+      "confidence": 0.9,
+      "domains": [
+        "identity",
+        "medical",
+        "hr"
+      ],
+      "description": "French NIR / INSEE number (numéro de sécurité sociale, 13+2 digits)",
+      "examples": [
+        "1 85 05 78 049 013 36"
+      ],
+      "validate": "nir"
+    },
+    {
+      "id": "regex:fr:passport",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{2}[A-Z]{2}\\d{5}\\b",
+      "flags": "g",
+      "confidence": 0.75,
+      "domains": [
+        "identity"
+      ],
+      "description": "French passport number (2 digits + 2 letters + 5 digits)",
+      "examples": [
+        "12AB34567"
+      ]
+    },
+    {
+      "id": "regex:fr:cni",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{12}\\b",
+      "flags": "g",
+      "confidence": 0.4,
+      "domains": [
+        "identity"
+      ],
+      "description": "French national ID card number (12 digits, new format)",
+      "examples": [
+        "123456789012"
+      ],
+      "falsePositiveNotes": "Very broad — 12 digits matches many things"
+    },
+    {
+      "id": "regex:fr:siret",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{3}\\s?\\d{3}\\s?\\d{3}(?:\\s?\\d{5})?\\b",
+      "flags": "g",
+      "confidence": 0.65,
+      "domains": [
+        "financial",
+        "legal"
+      ],
+      "description": "French SIRET (14 digits) or SIREN (9 digits) business number",
+      "examples": [
+        "362 521 879 00034",
+        "362 521 879"
+      ],
+      "falsePositiveNotes": "9-digit SIREN overlaps with many formats"
+    },
+    {
+      "id": "regex:fr:vat",
+      "entityType": "SSN",
+      "pattern": "\\b(?:FR)?\\d{2}\\s?\\d{9}\\b",
+      "flags": "gi",
+      "confidence": 0.75,
+      "domains": [
+        "financial",
+        "legal"
+      ],
+      "description": "French VAT number (FR + 2 digits + 9 digits SIREN)",
+      "examples": [
+        "FR12 362521879",
+        "FR 12362521879"
+      ]
+    },
+    {
+      "id": "regex:fr:phone",
+      "entityType": "PHONE",
+      "pattern": "\\b0[1-9](?:\\s?\\d{2}){4}\\b",
+      "flags": "g",
+      "confidence": 0.85,
+      "domains": [
+        "contact"
+      ],
+      "description": "French phone number (0X XX XX XX XX)",
+      "examples": [
+        "01 23 45 67 89",
+        "06 12 34 56 78",
+        "0612345678"
+      ]
+    },
+    {
+      "id": "regex:fr:postal",
+      "entityType": "ADDRESS",
+      "pattern": "\\b\\d{5}\\s+[\\p{L}][\\p{L}\\s'-]+\\b",
+      "flags": "gu",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "French postal code (5 digits) followed by city name",
+      "examples": [
+        "75008 Paris",
+        "69001 Lyon",
+        "13001 Marseille"
+      ]
+    },
+    {
+      "id": "regex:fr:currency_words",
+      "entityType": "CURRENCY",
+      "pattern": "(?:(?:soit|montant)\\s+(?:en\\s+)?(?:lettres?|toutes?\\s+lettres?)\\s*:\\s*|(?:la\\s+)?somme\\s+de\\s+)[\\p{L}\\s'-]+(?:euros?|centimes?)\\b",
+      "flags": "giu",
+      "confidence": 0.9,
+      "domains": [
+        "financial"
+      ],
+      "description": "French amount written in words (e.g., \"soit en lettres: huit mille cinq cents euros\")",
+      "examples": [
+        "soit en lettres: huit mille cinq cents euros"
+      ]
+    },
+    {
+      "id": "regex:fr:street",
+      "entityType": "ADDRESS",
+      "pattern": "\\b\\d{1,5}(?:\\s*(?:bis|ter))?\\s*,?\\s*(?:rue|avenue|av\\.|boulevard|bd\\.?|place|pl\\.|impasse|allée|passage|chemin|route|quai|cours|square|voie|sentier|résidence)\\s+[\\p{L}][\\p{L}\\s'-]+",
+      "flags": "giu",
+      "confidence": 0.9,
+      "domains": [
+        "contact"
+      ],
+      "description": "French street address (number + rue/avenue/boulevard + name)",
+      "examples": [
+        "12 rue de Rivoli",
+        "42 avenue des Champs-Élysées",
+        "8 bis boulevard Haussmann"
+      ]
+    },
+    {
+      "id": "regex:fr:company",
+      "entityType": "COMPANY",
+      "pattern": "[\\p{L}][\\p{L}\\s&.']+(?:\\s+)?(?:SARL|SAS|SA|EURL|SCI|SNC|SASU)\\b",
+      "flags": "gu",
+      "confidence": 0.9,
+      "domains": [
+        "legal",
+        "financial"
+      ],
+      "description": "French company name with legal form (SARL, SAS, SA, EURL, SCI, SNC)",
+      "examples": [
+        "Total SA",
+        "Capgemini SAS",
+        "Dupont et Fils SARL"
+      ]
+    },
+    {
+      "id": "regex:fr:carte_vitale",
+      "entityType": "OTHER",
+      "pattern": "\\b\\d{1}\\s?\\d{2}\\s?\\d{2}\\s?\\d{3}\\s?\\d{3}\\b",
+      "flags": "g",
+      "confidence": 0.7,
+      "domains": [
+        "medical"
+      ],
+      "description": "French Carte Vitale number (same as NIR, 13 digits)",
+      "examples": [
+        "1 85 05 780 490"
+      ],
+      "falsePositiveNotes": "Duplicate of NIR — overlap resolution will keep one"
+    }
+  ]
+}

package/rules/gb.json

@@ -0,0 +1,194 @@
+{
+  "region": "gb",
+  "rules": [
+    {
+      "id": "regex:gb:nino",
+      "entityType": "SSN",
+      "pattern": "\\b[A-Z]{2}\\s?\\d{2}\\s?\\d{2}\\s?\\d{2}\\s?[A-D]\\b",
+      "flags": "gi",
+      "confidence": 0.9,
+      "domains": [
+        "identity",
+        "hr",
+        "financial"
+      ],
+      "description": "UK National Insurance Number (NINO, e.g., AB 12 34 56 C)",
+      "examples": [
+        "AB 12 34 56 C",
+        "CE123456C"
+      ],
+      "validate": "nino"
+    },
+    {
+      "id": "regex:gb:nhs",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{3}\\s?\\d{3}\\s?\\d{4}\\b",
+      "flags": "g",
+      "confidence": 0.6,
+      "domains": [
+        "medical",
+        "identity"
+      ],
+      "description": "UK NHS number (10 digits)",
+      "examples": [
+        "943 476 5919"
+      ],
+      "falsePositiveNotes": "10-digit format overlaps with many phone numbers",
+      "validate": "nhs"
+    },
+    {
+      "id": "regex:gb:passport",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{9}\\b",
+      "flags": "g",
+      "confidence": 0.4,
+      "domains": [
+        "identity"
+      ],
+      "description": "UK passport number (9 digits)",
+      "examples": [
+        "123456789"
+      ],
+      "falsePositiveNotes": "Very broad — 9 digits matches many things. Best paired with context clues."
+    },
+    {
+      "id": "regex:gb:sort_code",
+      "entityType": "OTHER",
+      "pattern": "\\b\\d{2}-\\d{2}-\\d{2}\\b",
+      "flags": "g",
+      "confidence": 0.65,
+      "domains": [
+        "financial"
+      ],
+      "description": "UK bank sort code (XX-XX-XX)",
+      "examples": [
+        "12-34-56"
+      ],
+      "falsePositiveNotes": "Also matches dates in dd-mm-yy format"
+    },
+    {
+      "id": "regex:gb:bank_account",
+      "entityType": "OTHER",
+      "pattern": "\\b\\d{7,8}\\b",
+      "flags": "g",
+      "confidence": 0.3,
+      "domains": [
+        "financial"
+      ],
+      "description": "UK bank account number (7-8 digits)",
+      "examples": [
+        "12345678"
+      ],
+      "falsePositiveNotes": "Extremely broad — disabled by default in low-sensitivity mode"
+    },
+    {
+      "id": "regex:gb:vat",
+      "entityType": "SSN",
+      "pattern": "\\b(?:GB)?\\d{3}\\s?\\d{4}\\s?\\d{2}(?:\\s?\\d{3})?\\b",
+      "flags": "gi",
+      "confidence": 0.75,
+      "domains": [
+        "financial",
+        "legal"
+      ],
+      "description": "UK VAT number (GB + 9 or 12 digits)",
+      "examples": [
+        "GB123 4567 89",
+        "GB123456789"
+      ]
+    },
+    {
+      "id": "regex:gb:phone_landline",
+      "entityType": "PHONE",
+      "pattern": "\\b0\\d{4}\\s?\\d{6}\\b",
+      "flags": "g",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "UK landline phone number (e.g., 020 7946 0958)",
+      "examples": [
+        "01234 567890",
+        "02079 460958"
+      ]
+    },
+    {
+      "id": "regex:gb:phone_mobile",
+      "entityType": "PHONE",
+      "pattern": "\\b07\\d{3}\\s?\\d{6}\\b",
+      "flags": "g",
+      "confidence": 0.85,
+      "domains": [
+        "contact"
+      ],
+      "description": "UK mobile phone number (07XXX XXXXXX)",
+      "examples": [
+        "07911 123456"
+      ]
+    },
+    {
+      "id": "regex:gb:postcode",
+      "entityType": "ADDRESS",
+      "pattern": "\\b[A-Z]{1,2}\\d[A-Z\\d]?\\s?\\d[A-Z]{2}\\b",
+      "flags": "gi",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "UK postcode (e.g., SW1A 1AA, EC2A 1NT, M1 1AA)",
+      "examples": [
+        "SW1A 1AA",
+        "EC2A 1NT",
+        "M1 1AA",
+        "B33 8TH"
+      ]
+    },
+    {
+      "id": "regex:gb:currency_words",
+      "entityType": "CURRENCY",
+      "pattern": "(?:in\\s+(?:the\\s+)?(?:amount|sum)\\s+of|pay(?:able)?(?:\\s+the\\s+(?:amount|sum))?\\s+of)\\s+[\\p{L}\\s-]+(?:pounds?|pence|sterling)\\b",
+      "flags": "giu",
+      "confidence": 0.9,
+      "domains": [
+        "financial"
+      ],
+      "description": "English (UK) amount written in words (e.g., \"in the amount of eight thousand five hundred pounds\")",
+      "examples": [
+        "in the amount of eight thousand five hundred pounds"
+      ]
+    },
+    {
+      "id": "regex:gb:street",
+      "entityType": "ADDRESS",
+      "pattern": "\\b\\d{1,4}[A-Z]?\\s+[A-Z][\\w]+(?:\\s+[A-Z][\\w]+)*\\s+(?:Street|St|Road|Rd|Avenue|Ave|Lane|Ln|Drive|Dr|Close|Cl|Crescent|Cres|Terrace|Terr|Place|Pl|Gardens|Gdns|Grove|Way|Court|Ct|Square|Sq|Mews|Row|Rise|Hill|Park|Gate|Walk|Green|Parade|Wharf)\\.?\\b",
+      "flags": "gi",
+      "confidence": 0.85,
+      "domains": [
+        "contact"
+      ],
+      "description": "UK street address (number + street name + type)",
+      "examples": [
+        "10 Downing Street",
+        "221B Baker Street",
+        "42 Oxford Road"
+      ]
+    },
+    {
+      "id": "regex:gb:company",
+      "entityType": "COMPANY",
+      "pattern": "[\\p{L}][\\p{L}\\s&.']+(?:,?\\s+)?(?:Ltd\\.?|PLC|LLP|L\\.L\\.P\\.)\\b",
+      "flags": "gu",
+      "confidence": 0.85,
+      "domains": [
+        "legal",
+        "financial"
+      ],
+      "description": "UK company name with legal suffix (Ltd, PLC, LLP)",
+      "examples": [
+        "Barclays PLC",
+        "Tesco Ltd",
+        "Deloitte LLP"
+      ]
+    }
+  ]
+}

package/rules/ie.json

@@ -0,0 +1,48 @@
+{
+  "region": "ie",
+  "rules": [
+    {
+      "id": "regex:ie:pps-number",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{7}[A-W][ABWTXZ]?\\b",
+      "flags": "gi",
+      "confidence": 0.9,
+      "domains": [
+        "identity",
+        "hr"
+      ],
+      "description": "Irish PPS (Personal Public Service) number",
+      "examples": [
+        "1234567T",
+        "1234567TW"
+      ],
+      "validate": "pps"
+    },
+    {
+      "id": "regex:ie:phone",
+      "entityType": "PHONE",
+      "pattern": "\\b(?:\\+?353[\\s-]?|0)[1-9]\\d?[\\s-]?\\d{3}[\\s-]?\\d{3,4}\\b",
+      "flags": "g",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "Irish phone number"
+    },
+    {
+      "id": "regex:ie:eircode",
+      "entityType": "ADDRESS",
+      "pattern": "\\b[A-Z]\\d{2}\\s?[A-Z0-9]{4}\\b",
+      "flags": "gi",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "Irish Eircode postal code",
+      "examples": [
+        "A65 F4E2",
+        "D08YX4T"
+      ]
+    }
+  ]
+}