@doccloak/core 0.8.0

package/package.json

@@ -0,0 +1,75 @@
+{
+  "name": "@doccloak/core",
+  "version": "0.8.0",
+  "description": "Environment-agnostic document anonymization engine for the DocCloak ecosystem.",
+  "license": "Apache-2.0",
+  "author": "Witold Lojek",
+  "type": "module",
+  "main": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "default": "./dist/index.js"
+    },
+    "./dom": {
+      "types": "./dist/dom/index.d.ts",
+      "default": "./dist/dom/index.js"
+    },
+    "./worker-protocol": {
+      "types": "./dist/worker-protocol.d.ts",
+      "default": "./dist/worker-protocol.js"
+    }
+  },
+  "files": [
+    "dist",
+    "rules",
+    "LICENSE",
+    "NOTICE",
+    "README.md",
+    "CHANGELOG.md"
+  ],
+  "keywords": [
+    "anonymization",
+    "pii",
+    "ner",
+    "redaction",
+    "privacy",
+    "gdpr",
+    "onnx",
+    "docx",
+    "ocr"
+  ],
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/WLojek/DocCloak.Core.git"
+  },
+  "homepage": "https://github.com/WLojek/DocCloak.Core#readme",
+  "bugs": {
+    "url": "https://github.com/WLojek/DocCloak.Core/issues"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "scripts": {
+    "build": "rm -rf dist && tsc -p tsconfig.build.json && node scripts/fix-dts-extensions.mjs",
+    "prepack": "npm run build && node scripts/package-exports.mjs dist",
+    "postpack": "node scripts/package-exports.mjs src",
+    "rules:gen": "node scripts/generate-rules-module.mjs",
+    "rules:check": "node scripts/generate-rules-module.mjs --check",
+    "test": "vitest run",
+    "test:watch": "vitest",
+    "typecheck": "tsc --noEmit"
+  },
+  "devDependencies": {
+    "jsdom": "24.1.3",
+    "typescript": "5.8.3",
+    "vitest": "4.1.0"
+  },
+  "dependencies": {
+    "cfb": "1.2.2",
+    "jszip": "3.10.1",
+    "onnxruntime-web": "1.19.2",
+    "tesseract.js": "7.0.0"
+  }
+}

package/rules/at.json

@@ -0,0 +1,70 @@
+{
+  "region": "at",
+  "rules": [
+    {
+      "id": "regex:at:svnr",
+      "entityType": "SSN",
+      "pattern": "\\b[1-9]\\d{3}\\s?(?:0[1-9]|[12]\\d|3[01])(?:0[1-9]|1[0-2])\\d{2}\\b",
+      "flags": "g",
+      "confidence": 0.85,
+      "domains": [
+        "identity",
+        "hr",
+        "medical"
+      ],
+      "description": "Austrian social insurance number (SVNR / Sozialversicherungsnummer, 10 digits)",
+      "examples": [
+        "1237 010180"
+      ],
+      "validate": "svnr"
+    },
+    {
+      "id": "regex:at:phone",
+      "entityType": "PHONE",
+      "pattern": "\\b(?:\\+?43[\\s-]?|0)\\d{1,4}[\\s-]?\\d{4,10}\\b",
+      "flags": "g",
+      "confidence": 0.75,
+      "domains": [
+        "contact"
+      ],
+      "description": "Austrian phone number (variable-length area codes)",
+      "examples": [
+        "+43 1 12345678",
+        "0664 1234567"
+      ]
+    },
+    {
+      "id": "regex:at:postal",
+      "entityType": "ADDRESS",
+      "pattern": "\\b\\d{4}\\s+[\\p{L}][\\p{L}\\s-]+\\b",
+      "flags": "gu",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "Austrian postal code (4 digits) followed by city name",
+      "examples": [
+        "1010 Wien",
+        "5020 Salzburg",
+        "8010 Graz"
+      ]
+    },
+    {
+      "id": "regex:at:street",
+      "entityType": "ADDRESS",
+      "pattern": "[\\p{L}][\\p{L}\\s-]*(?:straße|strasse|str\\.|gasse|weg|platz|ring|allee)\\s+\\d+[\\p{L}]?(?:\\s*[/\\\\]\\s*\\d+)?",
+      "flags": "giu",
+      "confidence": 0.85,
+      "domains": [
+        "contact"
+      ],
+      "description": "Austrian street address (name + Straße/Gasse/Weg/Platz + number)",
+      "examples": [
+        "Mariahilfer Straße 45",
+        "Währinger Str. 12",
+        "Hauptplatz 1",
+        "Neubaugasse 7/3"
+      ]
+    }
+  ]
+}

package/rules/be.json

@@ -0,0 +1,70 @@
+{
+  "region": "be",
+  "rules": [
+    {
+      "id": "regex:be:national_register",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{2}[.\\s]?\\d{2}[.\\s]?\\d{2}[-.\\s]?\\d{3}[.\\s]?\\d{2}\\b",
+      "flags": "g",
+      "confidence": 0.9,
+      "domains": [
+        "identity",
+        "hr"
+      ],
+      "description": "Belgian National Register Number (Rijksregisternummer / Numéro de registre national)",
+      "examples": [
+        "85.07.15-123.97",
+        "85071512397"
+      ],
+      "validate": "belgianNrn"
+    },
+    {
+      "id": "regex:be:phone",
+      "entityType": "PHONE",
+      "pattern": "\\b(?:\\+?32[\\s-]?|0)(?:4\\d{2}[\\s-]?\\d{2}[\\s-]?\\d{2}[\\s-]?\\d{2}|[1-9]\\d?[\\s-]?\\d{3}[\\s-]?\\d{2}[\\s-]?\\d{2})\\b",
+      "flags": "g",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "Belgian phone number (mobile 04xx or landline)",
+      "examples": [
+        "+32 475 12 34 56",
+        "02 123 45 67"
+      ]
+    },
+    {
+      "id": "regex:be:postal",
+      "entityType": "ADDRESS",
+      "pattern": "\\b[1-9]\\d{3}\\s+[\\p{L}][\\p{L}\\s'-]+\\b",
+      "flags": "gu",
+      "confidence": 0.75,
+      "domains": [
+        "contact"
+      ],
+      "description": "Belgian postal code (4 digits, 1000-9999) followed by city name",
+      "examples": [
+        "1000 Bruxelles",
+        "2000 Antwerpen",
+        "9000 Gent"
+      ],
+      "falsePositiveNotes": "4-digit numbers are common"
+    },
+    {
+      "id": "regex:be:street",
+      "entityType": "ADDRESS",
+      "pattern": "(?:straat|laan|weg|plein|steenweg|lei|dreef|boulevard|avenue|rue|place|chaussée)[\\p{L}\\s'-]+\\d{1,5}[\\p{L}]?|[\\p{L}][\\p{L}\\s'-]*(?:straat|laan|weg|plein|steenweg|lei|dreef)\\s+\\d{1,5}[\\p{L}]?",
+      "flags": "giu",
+      "confidence": 0.85,
+      "domains": [
+        "contact"
+      ],
+      "description": "Belgian street address (Dutch/French street types + number)",
+      "examples": [
+        "Kerkstraat 12",
+        "Avenue Louise 54",
+        "Rue de la Loi 16"
+      ]
+    }
+  ]
+}

package/rules/ch.json

@@ -0,0 +1,66 @@
+{
+  "region": "ch",
+  "rules": [
+    {
+      "id": "regex:ch:ahv-avs-number",
+      "entityType": "SSN",
+      "pattern": "\\b756[.\\s]?\\d{4}[.\\s]?\\d{4}[.\\s]?\\d{2}\\b",
+      "flags": "g",
+      "confidence": 0.95,
+      "domains": [
+        "identity",
+        "hr",
+        "medical"
+      ],
+      "description": "Swiss AHV/AVS social security number (new format, EAN-13 based)",
+      "examples": [
+        "756.1234.5678.97"
+      ],
+      "validate": "ahv"
+    },
+    {
+      "id": "regex:ch:phone",
+      "entityType": "PHONE",
+      "pattern": "\\b(?:\\+?41[\\s-]?|0)[1-9]\\d[\\s-]?\\d{3}[\\s-]?\\d{2}[\\s-]?\\d{2}\\b",
+      "flags": "g",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "Swiss phone number"
+    },
+    {
+      "id": "regex:ch:postal-city",
+      "entityType": "ADDRESS",
+      "pattern": "\\b\\d{4}\\s+[\\p{L}][\\p{L}\\s-]+\\b",
+      "flags": "gu",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "Swiss postal code followed by city name"
+    },
+    {
+      "id": "regex:ch:street-german",
+      "entityType": "ADDRESS",
+      "pattern": "[\\p{L}][\\p{L}\\s-]*(?:strasse|str\\.|weg|gasse|platz|allee|rain|matte)\\s+\\d+[\\p{L}]?",
+      "flags": "giu",
+      "confidence": 0.85,
+      "domains": [
+        "contact"
+      ],
+      "description": "Swiss street address in German (e.g. Bahnhofstrasse 10)"
+    },
+    {
+      "id": "regex:ch:street-french",
+      "entityType": "ADDRESS",
+      "pattern": "(?:rue|avenue|chemin|route|place|boulevard)\\s+[\\p{L}][\\p{L}\\s'-]+\\s*\\d+",
+      "flags": "giu",
+      "confidence": 0.85,
+      "domains": [
+        "contact"
+      ],
+      "description": "Swiss street address in French (e.g. rue de Lausanne 12)"
+    }
+  ]
+}

package/rules/cn.json

@@ -0,0 +1,155 @@
+{
+  "region": "cn",
+  "rules": [
+    {
+      "id": "regex:cn:resident_id",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{6}(?:19|20)\\d{2}(?:0[1-9]|1[0-2])(?:0[1-9]|[12]\\d|3[01])\\d{3}[\\dXx]\\b",
+      "flags": "g",
+      "confidence": 0.9,
+      "domains": [
+        "identity"
+      ],
+      "description": "Chinese Resident Identity Card number (18 digits)",
+      "examples": [
+        "110101199003070003"
+      ]
+    },
+    {
+      "id": "regex:cn:passport",
+      "entityType": "SSN",
+      "pattern": "\\b[GDE]\\d{8}\\b",
+      "flags": "g",
+      "confidence": 0.75,
+      "domains": [
+        "identity"
+      ],
+      "description": "Chinese passport number (G/D/E + 8 digits)",
+      "examples": [
+        "G12345678",
+        "E00000001"
+      ]
+    },
+    {
+      "id": "regex:cn:bank_card",
+      "entityType": "OTHER",
+      "pattern": "\\b\\d{15,19}\\b",
+      "flags": "g",
+      "confidence": 0.4,
+      "domains": [
+        "financial"
+      ],
+      "description": "Chinese bank card number (15-19 digits)",
+      "examples": [
+        "6222021234567890123"
+      ],
+      "falsePositiveNotes": "Any 15-19 digit run matches; low confidence by design"
+    },
+    {
+      "id": "regex:cn:uscc",
+      "entityType": "SSN",
+      "pattern": "\\b[0-9A-Z]{18}\\b",
+      "flags": "g",
+      "confidence": 0.6,
+      "domains": [
+        "identity",
+        "financial"
+      ],
+      "description": "Chinese Unified Social Credit Code (18 alphanumeric)",
+      "examples": [
+        "91110000710921000K"
+      ]
+    },
+    {
+      "id": "regex:cn:phone_mobile",
+      "entityType": "PHONE",
+      "pattern": "\\b(?:\\+?86[\\s-]?)?1[3-9]\\d[\\s-]?\\d{4}[\\s-]?\\d{4}\\b",
+      "flags": "g",
+      "confidence": 0.85,
+      "domains": [
+        "contact"
+      ],
+      "description": "Chinese mobile phone number (1XX-XXXX-XXXX, optionally with +86)",
+      "examples": [
+        "+86 138 1234 5678",
+        "13812345678"
+      ]
+    },
+    {
+      "id": "regex:cn:phone_landline",
+      "entityType": "PHONE",
+      "pattern": "\\b0\\d{2,3}[\\s-]?\\d{7,8}\\b",
+      "flags": "g",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "Chinese landline phone number (0XX-XXXXXXXX)",
+      "examples": [
+        "010-12345678",
+        "021-87654321"
+      ]
+    },
+    {
+      "id": "regex:cn:postal",
+      "entityType": "ADDRESS",
+      "pattern": "\\b\\d{6}\\b",
+      "flags": "g",
+      "confidence": 0.35,
+      "domains": [
+        "contact"
+      ],
+      "description": "Chinese postal code (6 digits)",
+      "examples": [
+        "100000",
+        "200000"
+      ],
+      "falsePositiveNotes": "Any standalone 6-digit number matches; low confidence by design"
+    },
+    {
+      "id": "regex:cn:currency_hanzi",
+      "entityType": "CURRENCY",
+      "pattern": "[零壹贰叁肆伍陆柒捌玖拾佰仟萬億一二三四五六七八九十百千万亿]+\\s*(?:元|圆|块)",
+      "flags": "g",
+      "confidence": 0.9,
+      "domains": [
+        "financial"
+      ],
+      "description": "Chinese amount in hanzi numerals + yuan sign (e.g., 捌仟伍佰元)",
+      "examples": [
+        "捌仟伍佰元",
+        "八千五百元"
+      ]
+    },
+    {
+      "id": "regex:cn:address",
+      "entityType": "ADDRESS",
+      "pattern": "[\\u4E00-\\u9FFF]+[省市][\\u4E00-\\u9FFF]+[市区县][\\u4E00-\\u9FFF\\d-]+",
+      "flags": "g",
+      "confidence": 0.85,
+      "domains": [
+        "contact"
+      ],
+      "description": "Chinese address (province/city + district + street)",
+      "examples": [
+        "北京市朝阳区建国路1号"
+      ]
+    },
+    {
+      "id": "regex:cn:company",
+      "entityType": "COMPANY",
+      "pattern": "[\\u4E00-\\u9FFF][\\u4E00-\\u9FFF\\s]+(?:有限公司|股份有限公司|集团|有限责任公司)",
+      "flags": "g",
+      "confidence": 0.9,
+      "domains": [
+        "general"
+      ],
+      "description": "Chinese company name with legal form",
+      "examples": [
+        "阿里巴巴集团",
+        "华为技术有限公司",
+        "腾讯科技股份有限公司"
+      ]
+    }
+  ]
+}

package/rules/de.json

@@ -0,0 +1,192 @@
+{
+  "region": "de",
+  "rules": [
+    {
+      "id": "regex:de:personalausweis",
+      "entityType": "SSN",
+      "pattern": "\\b[CFGHJKLMNPRTVWXYZ]\\d{2}[CFGHJKLMNPRTVWXYZ0-9]{6}\\d\\b",
+      "flags": "g",
+      "confidence": 0.75,
+      "domains": [
+        "identity"
+      ],
+      "description": "German ID card number (Personalausweisnummer)",
+      "examples": [
+        "T220001297"
+      ]
+    },
+    {
+      "id": "regex:de:passport",
+      "entityType": "SSN",
+      "pattern": "\\b[CFGHJKLMNPRTVWXYZ]\\d{2}[CFGHJKLMNPRTVWXYZ0-9]{5}\\d[A-Z]\\d{7}\\b",
+      "flags": "g",
+      "confidence": 0.75,
+      "domains": [
+        "identity"
+      ],
+      "description": "German passport number",
+      "examples": [
+        "C01X00T47D1234567"
+      ]
+    },
+    {
+      "id": "regex:de:steuernummer",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{2,3}\\/?\\d{3}\\/?\\d{4,5}\\b",
+      "flags": "g",
+      "confidence": 0.6,
+      "domains": [
+        "financial"
+      ],
+      "description": "German tax number (Steuernummer, 10-11 digits, format varies by state)",
+      "examples": [
+        "93/815/08152",
+        "2181508152"
+      ],
+      "falsePositiveNotes": "Format varies by Bundesland — hard to validate precisely"
+    },
+    {
+      "id": "regex:de:steuerid",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{2}\\s?\\d{3}\\s?\\d{3}\\s?\\d{3}\\b",
+      "flags": "g",
+      "confidence": 0.8,
+      "domains": [
+        "financial",
+        "identity"
+      ],
+      "description": "German tax ID (Steuerliche Identifikationsnummer, 11 digits)",
+      "examples": [
+        "12 345 678 903"
+      ],
+      "validate": "steuerId"
+    },
+    {
+      "id": "regex:de:vat",
+      "entityType": "SSN",
+      "pattern": "\\b(?:DE)?\\d{9}\\b",
+      "flags": "gi",
+      "confidence": 0.65,
+      "domains": [
+        "financial",
+        "legal"
+      ],
+      "description": "German VAT number (USt-IdNr., DE + 9 digits)",
+      "examples": [
+        "DE123456789"
+      ],
+      "falsePositiveNotes": "Without DE prefix, 9 digits alone has many false positives"
+    },
+    {
+      "id": "regex:de:phone",
+      "entityType": "PHONE",
+      "pattern": "\\b(?:\\+?49[\\s-]?)?\\(?\\d{2,5}\\)?[\\s-]?\\d{3,10}\\b",
+      "flags": "g",
+      "confidence": 0.7,
+      "domains": [
+        "contact"
+      ],
+      "description": "German phone number (variable-length area codes)",
+      "examples": [
+        "+49 30 12345678",
+        "089/12345678",
+        "(030) 12345678"
+      ],
+      "falsePositiveNotes": "Variable-length area codes make this pattern broad"
+    },
+    {
+      "id": "regex:de:postal",
+      "entityType": "ADDRESS",
+      "pattern": "\\b\\d{5}\\s+[\\p{L}][\\p{L}\\s-]+\\b",
+      "flags": "gu",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "German postal code (5 digits) followed by city name",
+      "examples": [
+        "10115 Berlin",
+        "80331 München"
+      ]
+    },
+    {
+      "id": "regex:de:currency_words",
+      "entityType": "CURRENCY",
+      "pattern": "(?:in\\s+(?:Höhe|Worten?)\\s+(?:von\\s+)?|(?:Betrag|Summe)\\s+von\\s+)[\\p{L}\\s-]+(?:Euro|Cent)\\b",
+      "flags": "giu",
+      "confidence": 0.9,
+      "domains": [
+        "financial"
+      ],
+      "description": "German amount written in words (e.g., \"in Höhe von achttausendfünfhundert Euro\")",
+      "examples": [
+        "in Höhe von achttausendfünfhundert Euro"
+      ]
+    },
+    {
+      "id": "regex:de:street",
+      "entityType": "ADDRESS",
+      "pattern": "[\\p{L}][\\p{L}\\s-]*(?:straße|strasse|str\\.|weg|gasse|platz|allee|ring|damm|ufer|chaussee)\\s+\\d+[\\p{L}]?(?:\\s*[/\\\\]\\s*\\d+)?",
+      "flags": "giu",
+      "confidence": 0.9,
+      "domains": [
+        "contact"
+      ],
+      "description": "German street address (name + straße/str./weg/gasse/platz + number)",
+      "examples": [
+        "Friedrichstraße 43",
+        "Berliner Str. 12",
+        "Hauptplatz 1",
+        "Am Damm 7a"
+      ]
+    },
+    {
+      "id": "regex:de:company",
+      "entityType": "COMPANY",
+      "pattern": "[\\p{L}][\\p{L}\\s&.]+(?:\\s+)?(?:GmbH|AG|KG|OHG|e\\.?\\s?V\\.?|GbR|UG|mbH|KGaA)\\b",
+      "flags": "gu",
+      "confidence": 0.9,
+      "domains": [
+        "legal",
+        "financial"
+      ],
+      "description": "German company name with legal form (GmbH, AG, KG, OHG, e.V., GbR, UG)",
+      "examples": [
+        "Deutsche Bank AG",
+        "Siemens GmbH",
+        "Bosch KG"
+      ]
+    },
+    {
+      "id": "regex:de:icd10gm",
+      "entityType": "OTHER",
+      "pattern": "\\b[A-Z]\\d{2}(?:\\.\\d{1,2})?(?:\\s?[GLRAZ])?\\b",
+      "flags": "g",
+      "confidence": 0.55,
+      "domains": [
+        "medical"
+      ],
+      "description": "ICD-10-GM diagnosis code (German modification)",
+      "examples": [
+        "J18.9 G",
+        "E11.65"
+      ],
+      "falsePositiveNotes": "Very short codes — high false positive risk"
+    },
+    {
+      "id": "regex:de:sozialversicherung",
+      "entityType": "SSN",
+      "pattern": "\\b\\d{2}(?:0[1-9]|[12]\\d|3[01])(?:0[1-9]|1[0-2])\\d{2}[A-Z]\\d{2}\\d\\b",
+      "flags": "gi",
+      "confidence": 0.75,
+      "domains": [
+        "hr",
+        "identity"
+      ],
+      "description": "German social insurance number (Sozialversicherungsnummer, 12 chars: area + DDMMYY birth date + letter + serial + check digit)",
+      "examples": [
+        "12010182A056"
+      ]
+    }
+  ]
+}

package/rules/dk.json

@@ -0,0 +1,43 @@
+{
+  "region": "dk",
+  "rules": [
+    {
+      "id": "regex:dk:cpr-number",
+      "entityType": "SSN",
+      "pattern": "\\b(?:0[1-9]|[12]\\d|3[01])(?:0[1-9]|1[0-2])\\d{2}[-\\s]?\\d{4}\\b",
+      "flags": "g",
+      "confidence": 0.8,
+      "domains": [
+        "identity",
+        "hr"
+      ],
+      "description": "Danish CPR (Central Person Register) number",
+      "examples": [
+        "010190-1234"
+      ],
+      "falsePositiveNotes": "Post-2007 CPR numbers may not satisfy modulo-11 check"
+    },
+    {
+      "id": "regex:dk:phone",
+      "entityType": "PHONE",
+      "pattern": "\\b(?:\\+?45[\\s-]?)?\\d{2}[\\s-]?\\d{2}[\\s-]?\\d{2}[\\s-]?\\d{2}\\b",
+      "flags": "g",
+      "confidence": 0.75,
+      "domains": [
+        "contact"
+      ],
+      "description": "Danish phone number"
+    },
+    {
+      "id": "regex:dk:postal-city",
+      "entityType": "ADDRESS",
+      "pattern": "\\b\\d{4}\\s+[\\p{L}][\\p{L}\\s-]+\\b",
+      "flags": "gu",
+      "confidence": 0.8,
+      "domains": [
+        "contact"
+      ],
+      "description": "Danish postal code followed by city name"
+    }
+  ]
+}