@flexorch/audit 0.3.1 → 0.5.0

package/dist/index.cjs

@@ -26,6 +26,7 @@ __export(index_exports, {
   detectPii: () => detectPii,
   mask: () => mask,
   noiseMetrics: () => noiseMetrics,
+  noiseRatio: () => noiseRatio,
   qualityMetrics: () => qualityMetrics,
   version: () => version
 });
@@ -33,8 +34,9 @@ module.exports = __toCommonJS(index_exports);
 
 // src/pii.ts
 var EMAIL_RE = /\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,}\b/g;
-var PHONE_INTL_RE = /\+\d{1,3}[\s\-.]?\(?\d{1,4}\)?[\s\-.]?\d{3,4}[\s\-.]?\d{4}\b/g;
+var PHONE_INTL_RE = /(?<![+\d])(\+[1-9][\d\s\-.()]{5,18}\d)(?!\d)/g;
 var IBAN_RE = /\b[A-Z]{2}\d{2}[0-9A-Z]{11,30}\b/g;
+var IBAN_INTL_RE = /\b([A-Z]{2}\d{2}[0-9A-Z]{11,30})\b/g;
 var CC_RE = /\b\d{4}[ \-]\d{4}[ \-]\d{4}[ \-]\d{4}\b/g;
 var IPV4_RE = /\b(?:(?:25[0-5]|2[0-4]\d|[01]?\d\d?)\.){3}(?:25[0-5]|2[0-4]\d|[01]?\d\d?)\b/g;
 var _H = "[0-9a-fA-F]{1,4}";
@@ -45,6 +47,102 @@ var IPV6_RE = new RegExp(
 var PHONE_TR_RE = /\b(?:\+90|0)?\s*5\d{2}\s*\d{3}\s*\d{2}\s*\d{2}\b/g;
 var TCKN_RE = /\b([1-9]\d{10})\b/g;
 var VKN_RE = /\b([1-9]\d{9})\b/g;
+var IBAN_TR_RE = /\bTR\d{2}[0-9A-Z]{22}\b/g;
+var _TR_COMPANY_SUFFIX = "(?:A\\.\u015E\\.|Ltd\\.\\s*\u015Eti\\.|Koll\\.\\s*\u015Eti\\.|Koop\\.|T\\.A\\.\u015E\\.)";
+var _TR_NAME_TOKEN = "(?:ve|ile|[A-Z\xC7\u011E\u0130\xD6\u015E\xDC][A-Za-z\xC7\u011E\u0130\xD6\u015E\xDC\xE7\u011F\u0131\u015F\xF6\u015F\xFC]*\\.?)";
+var COMPANY_NAME_TR_RE = new RegExp(
+  `(?<![A-Za-z\xC7\u011E\u0130\xD6\u015E\xDC\xE7\u011F\u0131\u015F\xF6\u015F\xFC])([A-Z\xC7\u011E\u0130\xD6\u015E\xDC][A-Za-z\xC7\u011E\u0130\xD6\u015E\xDC\xE7\u011F\u0131\u015F\xF6\u015F\xFC]*(?:\\s+${_TR_NAME_TOKEN}){0,6}\\s+${_TR_COMPANY_SUFFIX})`,
+  "gu"
+);
+var MERSIS_RE = /\b([1-9]\d{15})\b/g;
+var POSTAL_CODE_TR_RE = /\b((?:0[1-9]|[1-7]\d|80|81)\d{3})\b/g;
+var _TR_PROVINCES_SORTED = [
+  "Afyonkarahisar",
+  "Kahramanmara\u015F",
+  "K\u0131r\u0131kkale",
+  "K\u0131rklareli",
+  "Diyarbak\u0131r",
+  "Gaziantep",
+  "\u015Eanl\u0131urfa",
+  "Nev\u015Fehir",
+  "Kastamonu",
+  "G\xFCm\xFC\u015Fhane",
+  "Eski\u015Fehir",
+  "Erzincan",
+  "Erzurum",
+  "Denizli",
+  "\xC7anakkale",
+  "Ad\u0131yaman",
+  "Zonguldak",
+  "Tekirda\u011F",
+  "Trabzon",
+  "Tunceli",
+  "Karaman",
+  "Karab\xFCk",
+  "Aksaray",
+  "Antalya",
+  "K\u0131r\u015Fehir",
+  "Osmaniye",
+  "Kocaeli",
+  "Sakarya",
+  "Bart\u0131n",
+  "Bayburt",
+  "Ardahan",
+  "Yozgat",
+  "Ankara",
+  "Amasya",
+  "Artvin",
+  "Bal\u0131kesir",
+  "Bilecik",
+  "Bing\xF6l",
+  "Bitlis",
+  "Burdur",
+  "\xC7ank\u0131r\u0131",
+  "Edirne",
+  "Elaz\u0131\u011F",
+  "Giresun",
+  "Hakkari",
+  "Isparta",
+  "\u0130stanbul",
+  "\u0130zmir",
+  "Kayseri",
+  "K\xFCtahya",
+  "Malatya",
+  "Manisa",
+  "Mardin",
+  "Samsun",
+  "\u015E\u0131rnak",
+  "Sinop",
+  "Tokat",
+  "Hatay",
+  "Konya",
+  "Mu\u011Fla",
+  "Ni\u011Fde",
+  "Rize",
+  "Siirt",
+  "Sivas",
+  "Adana",
+  "Ayd\u0131n",
+  "Bursa",
+  "\xC7orum",
+  "I\u011Fd\u0131r",
+  "Kilis",
+  "Mersin",
+  "Batman",
+  "Yalova",
+  "D\xFCzce",
+  "Ordu",
+  "Kars",
+  "A\u011Fr\u0131",
+  "Bolu",
+  "Van",
+  "U\u015Fak",
+  "Mu\u015F"
+].sort((a, b) => b.length - a.length);
+var PROVINCE_TR_RE = new RegExp(
+  `(?<!\\w)(${_TR_PROVINCES_SORTED.map((p) => p.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")).join("|")})(?!\\w)`,
+  "gu"
+);
 var NAME_PREFIX_TR = "(?:Ad[\u0131i]\\s*(?:Soyad[\u0131i])?|Soyad[\u0131i]|\u0130sim|M\xFC\u015Fteri\\s+Ad[\u0131i]|Yetkili(?:\\s+Ki\u015Fi)?|\xC7al\u0131\u015Fan\\s+Ad[\u0131i]|Personel\\s+Ad[\u0131i]|Ki\u015Fi\\s+Ad[\u0131i]|Sat\u0131c\u0131\\s+Ad[\u0131i]|Al\u0131c\u0131\\s+Ad[\u0131i]|\u0130lgili\\s+Ki\u015Fi|Hesap\\s+Sahibi)";
 var NAME_PREFIX_EN = "(?:Full\\s+Name|Customer\\s+Name|Employee\\s+Name|Contact\\s+Name|Authorized\\s+(?:By|Person)|Account\\s+Holder|(?<!\\bUser\\s)Name)";
 var NAME_VALUE = "([A-Z\xC7\u011E\u0130\xD6\u015E\xDC][a-z\xE7\u011F\u0131\u015F\xF6\u015F\xFC]+(?:\\s+[A-Z\xC7\u011E\u0130\xD6\u015E\xDC][a-z\xE7\u011F\u0131\u015F\xF6\u015F\xFC]+){0,2})";
@@ -52,7 +150,64 @@ var NAME_RE = new RegExp(
   `(?:${NAME_PREFIX_TR}|${NAME_PREFIX_EN})\\s*[:\\-]\\s*${NAME_VALUE}`,
   "gu"
 );
+var _IBAN_INTL_LENGTHS = {
+  AT: 20,
+  BE: 16,
+  BG: 22,
+  HR: 21,
+  CY: 28,
+  CZ: 24,
+  DK: 18,
+  EE: 20,
+  FI: 18,
+  FR: 27,
+  DE: 22,
+  GR: 27,
+  HU: 28,
+  IE: 22,
+  IT: 27,
+  LV: 21,
+  LT: 20,
+  LU: 20,
+  MT: 31,
+  NL: 18,
+  PL: 28,
+  PT: 25,
+  RO: 24,
+  SK: 24,
+  SI: 19,
+  ES: 24,
+  SE: 24,
+  GB: 22,
+  CH: 21,
+  NO: 15
+};
+var _INTL_SUFFIX = "(?:KGaA|GmbH|OHG|GbR|SARL|EURL|S\\.p\\.A\\.|S\\.r\\.l\\.|S\\.n\\.c\\.|S\\.a\\.s\\.|B\\.V\\.|N\\.V\\.|S\\.A\\.|S\\.L\\.|Corp\\.|Inc\\.|Ltd\\.|LLP|LLC|PLC|SpA|Srl|SNC|SAS|BV|NV|SL|SA|Corp|Inc|Ltd|KG|AG|UG)";
+var _UC = "[A-Z\xC0-\u024F]";
+var _WC = "[A-Za-z0-9\xC0-\u024F\\-]";
+var _INTL_NAME_TOKEN = `(?:and|&|${_UC}${_WC}*\\.?)`;
+var COMPANY_NAME_INTL_RE = new RegExp(
+  `(?<![A-Za-z\xC0-\u024F])(${_UC}${_WC}*(?:\\s+${_INTL_NAME_TOKEN}){0,6}\\s+${_INTL_SUFFIX})`,
+  "gu"
+);
 var SSN_RE = /\b(?!000|666|9\d{2})\d{3}-(?!00)\d{2}-(?!0000)\d{4}\b/g;
+var EIN_US_RE = /\b(\d{2}-\d{7})\b/g;
+var ITIN_US_RE = /\b(9\d{2}-(?:7[0-9]|8[0-8]|9[0-24-9])-\d{4})\b/g;
+var STEUER_ID_DE_RE = /\b([1-9]\d{10})\b/g;
+var SVNR_DE_RE = /\b(\d{4}[01]\d[0-3]\d[A-Z]\d{4})\b/g;
+var SIRET_FR_RE = /(?:SIRET|N°\s*SIRET|Num[eé]ro\s+SIRET|RCS)\s*[:#]*\s*(\d{14})\b/gi;
+var SIREN_FR_RE = /(?:SIREN|N°\s*SIREN|Num[eé]ro\s+SIREN)\s*[:#]*\s*(\d{9})\b/gi;
+var INSEE_FR_RE = /\b([12]\d{14})\b/g;
+var CODICE_FISCALE_IT_RE = /\b([A-Z]{6}\d{2}[A-Z]\d{2}[A-Z]\d{3}[A-Z])\b/gi;
+var PARTITA_IVA_IT_RE = /\b(\d{11})\b/g;
+var BSN_NL_RE = /\b(\d{9})\b/g;
+var KVK_NL_RE = /(?:KVK|KvK|Handelsregister(?:nummer)?)\s*[:#]*\s*(\d{8})\b/gi;
+var _DNI_LETTER_TABLE = "TRWAGMYFPDXBNJZSQVHLCKE";
+var DNI_ES_RE = /\b(\d{8}[A-Z])\b/g;
+var NIE_ES_RE = /\b([XYZ]\d{7}[A-Z])\b/g;
+var CIF_ES_RE = /\b([ABCDEFGHJKLMNPQRSUVW]\d{7}[0-9A-J])\b/g;
+var NI_UK_RE = /\b([A-CEGHJ-PR-TW-Z][A-CEGHJ-NPR-TW-Z]\d{6}[ABCD])\b/g;
+var UTR_UK_RE = /(?:UTR|Unique\s+Taxpayer(?:\s+Reference)?)\s*[:#]*\s*(\d{10})\b/gi;
 function validTckn(s) {
   if (s.length !== 11 || s[0] === "0") return false;
   const d = s.split("").map(Number);
@@ -103,19 +258,109 @@ function validIban(s) {
   }
   return remainder === 1;
 }
+function validIbanIntl(s) {
+  const country = s.slice(0, 2);
+  if (country === "TR" || !(country in _IBAN_INTL_LENGTHS)) return false;
+  if (s.length !== _IBAN_INTL_LENGTHS[country]) return false;
+  return validIban(s);
+}
+function validPhoneIntl(raw) {
+  const digits = raw.replace(/\D/g, "");
+  return digits.length >= 7 && digits.length <= 15 && digits.slice(0, 2) !== "90";
+}
+function validSteuerIdDe(s) {
+  if (s.length !== 11 || s[0] === "0") return false;
+  let product = 10;
+  for (let i = 0; i < 10; i++) {
+    let total = (parseInt(s[i]) + product) % 10;
+    if (total === 0) total = 10;
+    product = total * 2 % 11;
+  }
+  let check = 11 - product;
+  if (check === 10) check = 0;
+  return check === parseInt(s[10]);
+}
+function validPartitaIvaIt(s) {
+  if (s.length !== 11 || !/^\d+$/.test(s)) return false;
+  let oddSum = 0;
+  let evenSum = 0;
+  for (let i = 0; i < 10; i += 2) oddSum += parseInt(s[i]);
+  for (let i = 1; i < 10; i += 2) {
+    let v = parseInt(s[i]) * 2;
+    evenSum += v < 10 ? v : v - 9;
+  }
+  return (10 - (oddSum + evenSum) % 10) % 10 === parseInt(s[10]);
+}
+function validBsnNl(s) {
+  if (s.length !== 9 || !/^\d+$/.test(s)) return false;
+  let total = 0;
+  for (let i = 0; i < 8; i++) total += (9 - i) * parseInt(s[i]);
+  total -= parseInt(s[8]);
+  return total > 0 && total % 11 === 0;
+}
+function validDniEs(s) {
+  if (s.length !== 9 || !/^\d{8}/.test(s)) return false;
+  return _DNI_LETTER_TABLE[parseInt(s.slice(0, 8)) % 23] === s[8];
+}
+function validNieEs(s) {
+  if (s.length !== 9 || !"XYZ".includes(s[0])) return false;
+  const prefix = { X: "0", Y: "1", Z: "2" }[s[0]];
+  return _DNI_LETTER_TABLE[parseInt(prefix + s.slice(1, 8)) % 23] === s[8];
+}
+var _NI_UK_FORBIDDEN = /* @__PURE__ */ new Set(["BG", "GB", "KN", "NK", "NT", "TN", "ZZ"]);
+function validNiUk(s) {
+  return !_NI_UK_FORBIDDEN.has(s.slice(0, 2).toUpperCase());
+}
+var _EIN_INVALID_PREFIXES = /* @__PURE__ */ new Set([
+  "00",
+  "07",
+  "08",
+  "09",
+  "17",
+  "18",
+  "19",
+  "28",
+  "29",
+  "49",
+  "69",
+  "70",
+  "78",
+  "79",
+  "89",
+  "96",
+  "97"
+]);
+function validEinUs(s) {
+  return !_EIN_INVALID_PREFIXES.has(s.slice(0, 2));
+}
 var LOCALE_DETECTORS = {
-  tr: /* @__PURE__ */ new Set(["national_id_tr", "tax_id_tr", "phone_tr", "name"]),
-  us: /* @__PURE__ */ new Set(["ssn", "phone"]),
-  eu: /* @__PURE__ */ new Set(["phone"])
+  tr: /* @__PURE__ */ new Set([
+    "national_id_tr",
+    "tax_id_tr",
+    "phone_tr",
+    "name",
+    "iban_tr",
+    "company_name_tr",
+    "mersis_no",
+    "postal_code_tr",
+    "province_tr"
+  ]),
+  us: /* @__PURE__ */ new Set(["ssn", "tax_id_us", "national_id_us", "phone_intl", "company_name_intl"]),
+  eu: /* @__PURE__ */ new Set(["phone_intl", "iban_intl", "company_name_intl"]),
+  de: /* @__PURE__ */ new Set(["tax_id_de", "social_id_de"]),
+  fr: /* @__PURE__ */ new Set(["siret_fr", "company_id_fr", "social_id_fr"]),
+  it: /* @__PURE__ */ new Set(["national_id_it", "tax_id_it"]),
+  nl: /* @__PURE__ */ new Set(["national_id_nl", "company_id_nl"]),
+  es: /* @__PURE__ */ new Set(["national_id_es", "tax_id_es"]),
+  uk: /* @__PURE__ */ new Set(["social_id_uk", "tax_id_uk"])
 };
 var UNIVERSAL = /* @__PURE__ */ new Set(["email", "iban", "credit_card", "ip", "ip_v6"]);
 function activeDetectors(locale) {
-  if (locale === "all") {
+  if (locale === "all" || locale === "und") {
     const active2 = new Set(UNIVERSAL);
     for (const detectors of Object.values(LOCALE_DETECTORS)) {
       detectors.forEach((d) => active2.add(d));
     }
-    if (active2.has("phone_tr")) active2.delete("phone");
     return active2;
   }
   const active = new Set(UNIVERSAL);
@@ -131,18 +376,18 @@ function findAll(re, text, type) {
   }
   return results;
 }
-function detectPii(text, locale = "tr") {
+function detectPii(text, locale = "und") {
   const active = activeDetectors(locale);
   const t = text ?? "";
-  const findings = [];
+  let findings = [];
   if (active.has("email")) findings.push(...findAll(EMAIL_RE, t, "email"));
-  if (active.has("phone")) {
+  if (active.has("phone_intl")) {
     PHONE_INTL_RE.lastIndex = 0;
     let m;
     while ((m = PHONE_INTL_RE.exec(t)) !== null) {
-      const digits = m[0].replace(/\D/g, "").length;
-      if (digits >= 10) {
-        findings.push({ type: "phone", value: m[0], start: m.index, end: m.index + m[0].length });
+      const candidate = m[1];
+      if (validPhoneIntl(candidate)) {
+        findings.push({ type: "phone_intl", value: candidate, start: m.index, end: m.index + candidate.length });
       }
     }
   }
@@ -195,8 +440,180 @@ function detectPii(text, locale = "tr") {
       findings.push({ type: "name", value, start, end: start + value.length });
     }
   }
+  if (active.has("iban_tr")) {
+    IBAN_TR_RE.lastIndex = 0;
+    let m;
+    while ((m = IBAN_TR_RE.exec(t)) !== null) {
+      if (validIban(m[0])) {
+        findings.push({ type: "iban_tr", value: m[0], start: m.index, end: m.index + m[0].length });
+      }
+    }
+  }
+  if (active.has("company_name_tr")) {
+    COMPANY_NAME_TR_RE.lastIndex = 0;
+    let m;
+    while ((m = COMPANY_NAME_TR_RE.exec(t)) !== null) {
+      findings.push({ type: "company_name_tr", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("mersis_no")) {
+    MERSIS_RE.lastIndex = 0;
+    let m;
+    while ((m = MERSIS_RE.exec(t)) !== null) {
+      findings.push({ type: "mersis_no", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("postal_code_tr")) {
+    POSTAL_CODE_TR_RE.lastIndex = 0;
+    let m;
+    while ((m = POSTAL_CODE_TR_RE.exec(t)) !== null) {
+      findings.push({ type: "postal_code_tr", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("province_tr")) {
+    PROVINCE_TR_RE.lastIndex = 0;
+    let m;
+    while ((m = PROVINCE_TR_RE.exec(t)) !== null) {
+      findings.push({ type: "province_tr", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
   if (active.has("ssn")) findings.push(...findAll(SSN_RE, t, "ssn"));
-  return findings.sort((a, b) => a.start - b.start);
+  if (active.has("tax_id_us")) {
+    EIN_US_RE.lastIndex = 0;
+    let m;
+    while ((m = EIN_US_RE.exec(t)) !== null) {
+      if (validEinUs(m[1])) findings.push({ type: "tax_id_us", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("national_id_us")) {
+    ITIN_US_RE.lastIndex = 0;
+    let m;
+    while ((m = ITIN_US_RE.exec(t)) !== null) {
+      findings.push({ type: "national_id_us", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("tax_id_de")) {
+    STEUER_ID_DE_RE.lastIndex = 0;
+    let m;
+    while ((m = STEUER_ID_DE_RE.exec(t)) !== null) {
+      if (validSteuerIdDe(m[1])) findings.push({ type: "tax_id_de", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("social_id_de")) {
+    SVNR_DE_RE.lastIndex = 0;
+    let m;
+    while ((m = SVNR_DE_RE.exec(t)) !== null) {
+      findings.push({ type: "social_id_de", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("siret_fr")) {
+    SIRET_FR_RE.lastIndex = 0;
+    let m;
+    while ((m = SIRET_FR_RE.exec(t)) !== null) {
+      findings.push({ type: "siret_fr", value: m[1], start: m.index + m[0].indexOf(m[1]), end: m.index + m[0].indexOf(m[1]) + m[1].length });
+    }
+  }
+  if (active.has("company_id_fr")) {
+    SIREN_FR_RE.lastIndex = 0;
+    let m;
+    while ((m = SIREN_FR_RE.exec(t)) !== null) {
+      findings.push({ type: "company_id_fr", value: m[1], start: m.index + m[0].indexOf(m[1]), end: m.index + m[0].indexOf(m[1]) + m[1].length });
+    }
+  }
+  if (active.has("social_id_fr")) {
+    INSEE_FR_RE.lastIndex = 0;
+    let m;
+    while ((m = INSEE_FR_RE.exec(t)) !== null) {
+      findings.push({ type: "social_id_fr", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("national_id_it")) {
+    CODICE_FISCALE_IT_RE.lastIndex = 0;
+    let m;
+    while ((m = CODICE_FISCALE_IT_RE.exec(t)) !== null) {
+      findings.push({ type: "national_id_it", value: m[1].toUpperCase(), start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("tax_id_it")) {
+    PARTITA_IVA_IT_RE.lastIndex = 0;
+    let m;
+    while ((m = PARTITA_IVA_IT_RE.exec(t)) !== null) {
+      if (validPartitaIvaIt(m[1])) findings.push({ type: "tax_id_it", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("national_id_nl")) {
+    BSN_NL_RE.lastIndex = 0;
+    let m;
+    while ((m = BSN_NL_RE.exec(t)) !== null) {
+      if (validBsnNl(m[1])) findings.push({ type: "national_id_nl", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("company_id_nl")) {
+    KVK_NL_RE.lastIndex = 0;
+    let m;
+    while ((m = KVK_NL_RE.exec(t)) !== null) {
+      findings.push({ type: "company_id_nl", value: m[1], start: m.index + m[0].indexOf(m[1]), end: m.index + m[0].indexOf(m[1]) + m[1].length });
+    }
+  }
+  if (active.has("national_id_es")) {
+    DNI_ES_RE.lastIndex = 0;
+    let m;
+    while ((m = DNI_ES_RE.exec(t)) !== null) {
+      if (validDniEs(m[1])) findings.push({ type: "national_id_es", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+    NIE_ES_RE.lastIndex = 0;
+    while ((m = NIE_ES_RE.exec(t)) !== null) {
+      if (validNieEs(m[1])) findings.push({ type: "national_id_es", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("tax_id_es")) {
+    CIF_ES_RE.lastIndex = 0;
+    let m;
+    while ((m = CIF_ES_RE.exec(t)) !== null) {
+      findings.push({ type: "tax_id_es", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("social_id_uk")) {
+    NI_UK_RE.lastIndex = 0;
+    let m;
+    while ((m = NI_UK_RE.exec(t)) !== null) {
+      if (validNiUk(m[1])) findings.push({ type: "social_id_uk", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  if (active.has("tax_id_uk")) {
+    UTR_UK_RE.lastIndex = 0;
+    let m;
+    while ((m = UTR_UK_RE.exec(t)) !== null) {
+      findings.push({ type: "tax_id_uk", value: m[1], start: m.index + m[0].indexOf(m[1]), end: m.index + m[0].indexOf(m[1]) + m[1].length });
+    }
+  }
+  if (active.has("iban_intl")) {
+    IBAN_INTL_RE.lastIndex = 0;
+    let m;
+    while ((m = IBAN_INTL_RE.exec(t)) !== null) {
+      const candidate = m[1];
+      if (validIbanIntl(candidate)) {
+        findings.push({ type: "iban_intl", value: candidate, start: m.index, end: m.index + candidate.length });
+      }
+    }
+  }
+  if (active.has("company_name_intl")) {
+    COMPANY_NAME_INTL_RE.lastIndex = 0;
+    let m;
+    while ((m = COMPANY_NAME_INTL_RE.exec(t)) !== null) {
+      findings.push({ type: "company_name_intl", value: m[1], start: m.index, end: m.index + m[1].length });
+    }
+  }
+  findings.sort((a, b) => a.start - b.start);
+  const specificIbanSpans = new Set(
+    findings.filter((f) => f.type === "iban_tr" || f.type === "iban_intl").map((f) => `${f.start}:${f.end}`)
+  );
+  if (specificIbanSpans.size > 0) {
+    findings = findings.filter(
+      (f) => !(f.type === "iban" && specificIbanSpans.has(`${f.start}:${f.end}`))
+    );
+  }
+  return findings;
 }
 
 // src/quality.ts
@@ -218,6 +635,15 @@ function isGarbage(ch) {
   return ch === REPLACEMENT_CHAR || cp <= 31 || cp >= 127 && cp <= 159 || cp >= 57344 && cp <= 63743 || // private use area
   cp >= 55296 && cp <= 57343;
 }
+var LINE_NOISE_RE = /[@#!~*=]{3,}/;
+function noiseRatio(text) {
+  if (!text) return 0;
+  const lines = text.split("\n");
+  const total = lines.length;
+  if (total === 0) return 0;
+  const noisy = lines.filter((line) => !line.trim() || LINE_NOISE_RE.test(line)).length;
+  return Math.round(noisy / total * 1e4) / 1e4;
+}
 function noiseMetrics(text) {
   if (!text) return { garbage_ratio: 0, encoding_ok: true };
   const n = text.length;
@@ -274,7 +700,7 @@ function applyMask(text, findings, strategy = "redact") {
 }
 
 // src/index.ts
-var version = "0.3.1";
+var version = "0.5.0";
 function computeQualityScore(completeness, avgLength, garbageRatio) {
   const lengthScore = Math.min(avgLength / 500, 1);
   const noiseScore = Math.max(0, 1 - garbageRatio * 10);
@@ -287,10 +713,11 @@ function computeQualityGrade(score) {
   return "D";
 }
 function audit(text, options = {}) {
-  const locale = options.locale ?? "tr";
+  const locale = options.locale ?? "und";
   const pii = detectPii(text, locale);
   const quality = qualityMetrics(text);
   const noise = noiseMetrics(text);
+  const noise_ratio = noiseRatio(text);
   const quality_score = computeQualityScore(
     quality.completeness,
     quality.avg_length,
@@ -300,7 +727,7 @@ function audit(text, options = {}) {
   const counts = /* @__PURE__ */ new Map();
   for (const f of pii) counts.set(f.type, (counts.get(f.type) ?? 0) + 1);
   const pii_summary = Array.from(counts.entries()).sort(([a], [b]) => a.localeCompare(b)).map(([type, count]) => ({ type, count }));
-  return { quality_grade, quality_score, pii_summary, pii, quality, noise };
+  return { quality_grade, quality_score, pii_summary, pii, quality, noise, noise_ratio, detected_language: locale };
 }
 function auditBatch(texts, options = {}) {
   if (texts.length === 0) {
@@ -332,6 +759,7 @@ function mask(text, findings, options = {}) {
   detectPii,
   mask,
   noiseMetrics,
+  noiseRatio,
   qualityMetrics,
   version
 });

package/dist/index.d.cts

@@ -17,6 +17,12 @@ interface NoiseMetrics {
     garbage_ratio: number;
     encoding_ok: boolean;
 }
+/**
+ * Fraction of lines that are blank or contain symbol noise (`[@#!~*=]{3+}`).
+ * Mirrors the FlexOrch pipeline quality-step threshold — values above 0.20
+ * indicate a document likely to reduce extraction quality.
+ */
+declare function noiseRatio(text: string): number;
 declare function noiseMetrics(text: string): NoiseMetrics;
 
 type MaskStrategy = "redact" | "replace" | "token" | "hash";
@@ -30,11 +36,14 @@ declare function applyMask(text: string, findings: PiiFinding[], strategy?: Mask
  * import { readFileSync } from "fs"
  *
  * const text = readFileSync("contract.txt", "utf8")
- * const result = audit(text, { locale: "tr" })
+ * const result = audit(text)               // locale defaults to "und" (all detectors)
+ * const result = audit(text, { locale: "tr" })  // Turkish-only detectors
  *
- * result.quality_grade   // "A"
- * result.quality_score   // 0.91
- * result.pii_summary     // [{ type: "national_id_tr", count: 3 }, ...]
+ * result.quality_grade      // "A"
+ * result.quality_score      // 0.91
+ * result.noise_ratio        // 0.03  (line-level noise fraction)
+ * result.detected_language  // "und" (locale passed in — caller controls language)
+ * result.pii_summary        // [{ type: "national_id_tr", count: 3 }, ...]
  *
  * // Raw findings and metrics also available:
  * result.pii             // [{ type, value, start, end }, ...]
@@ -45,7 +54,7 @@ declare function applyMask(text: string, findings: PiiFinding[], strategy?: Mask
  * // "Contact: [REDACTED_EMAIL]"
  */
 
-declare const version = "0.3.1";
+declare const version = "0.5.0";
 type QualityGrade = "A" | "B" | "C" | "D";
 interface PiiSummaryEntry {
     type: string;
@@ -54,10 +63,12 @@ interface PiiSummaryEntry {
 interface AuditOptions {
     /**
      * Active locale-specific detectors.
-     * - "tr"  — Turkish: TCKN, VKN, phone_tr, name  (default)
-     * - "us"  — US: SSN, E.164 phone
-     * - "eu"  — EU: E.164 phone
-     * - "all" — All detectors
+     * - "und" — All detectors combined (default; use when language is unknown)
+     * - "all" — Alias for "und"
+     * - "tr"  — Turkish: TCKN, VKN, phone_tr, name, iban_tr, company_name_tr, mersis_no, postal_code_tr, province_tr
+     * - "us"  — US: SSN, EIN, ITIN, E.164 phone, company_name_intl
+     * - "eu"  — EU: E.164 phone, iban_intl, company_name_intl
+     * - "de" / "fr" / "it" / "nl" / "es" / "uk" — country-specific detectors
      *
      * Universal detectors (email, iban, credit_card, ip, ip_v6) are always active.
      */
@@ -74,6 +85,10 @@ interface AuditResult {
     pii: PiiFinding[];
     quality: QualityMetrics;
     noise: NoiseMetrics;
+    /** Fraction of lines that are blank or contain symbol noise (>0.20 = low quality). */
+    noise_ratio: number;
+    /** The locale value passed to audit() — caller-controlled language selection. */
+    detected_language: string;
 }
 interface MaskOptions {
     /** @default "redact" */
@@ -102,4 +117,4 @@ declare function auditBatch(texts: string[], options?: AuditOptions): BatchAudit
  */
 declare function mask(text: string, findings: PiiFinding[], options?: MaskOptions): string;
 
-export { type AuditOptions, type AuditResult, type BatchAuditResult, type MaskOptions, type MaskStrategy, type NoiseMetrics, type PiiFinding, type PiiSummaryEntry, type QualityGrade, type QualityMetrics, applyMask, audit, auditBatch, detectPii, mask, noiseMetrics, qualityMetrics, version };
+export { type AuditOptions, type AuditResult, type BatchAuditResult, type MaskOptions, type MaskStrategy, type NoiseMetrics, type PiiFinding, type PiiSummaryEntry, type QualityGrade, type QualityMetrics, applyMask, audit, auditBatch, detectPii, mask, noiseMetrics, noiseRatio, qualityMetrics, version };