@noy-db/hub 0.2.0-pre.7 → 0.2.0-pre.8

package/dist/chunk-6STK5TQP.js

@@ -0,0 +1,139 @@
+import {
+  ScriptViolationError
+} from "./chunk-B6PB7JLN.js";
+
+// src/i18n/script.ts
+var LATIN_BASE = /* @__PURE__ */ new Set([
+  "en",
+  "fr",
+  "de",
+  "es",
+  "it",
+  "pt",
+  "nl",
+  "sv",
+  "no",
+  "da",
+  "fi",
+  "is",
+  "pl",
+  "cs",
+  "sk",
+  "hu",
+  "ro",
+  "hr",
+  "sl",
+  "et",
+  "lv",
+  "lt",
+  "tr",
+  "vi",
+  "id",
+  "ms",
+  "tl",
+  "sw",
+  "af",
+  "ca",
+  "gl",
+  "eu",
+  "cy",
+  "ga"
+]);
+var SCRIPT_TABLE = {
+  th: ["Thai"],
+  ko: ["Hangul", "Han"],
+  ja: ["Han", "Hiragana", "Katakana"],
+  zh: ["Han"],
+  ar: ["Arabic"],
+  fa: ["Arabic"],
+  ur: ["Arabic"],
+  ru: ["Cyrillic"],
+  uk: ["Cyrillic"],
+  bg: ["Cyrillic"],
+  sr: ["Cyrillic"],
+  he: ["Hebrew"],
+  el: ["Greek"],
+  hi: ["Devanagari"],
+  ta: ["Tamil"],
+  km: ["Khmer"],
+  lo: ["Lao"],
+  my: ["Myanmar"]
+};
+var SUBTAG_SCRIPTS = {
+  Latn: ["Latin"],
+  Cyrl: ["Cyrillic", "Latin"],
+  Hans: ["Han", "Latin"],
+  Hant: ["Han", "Latin"],
+  Thai: ["Thai", "Latin"],
+  Arab: ["Arabic", "Latin"]
+};
+function inferScripts(locale) {
+  const parts = locale.split("-");
+  const subtag = parts.find((t) => /^[A-Z][a-z]{3}$/.test(t));
+  if (subtag && SUBTAG_SCRIPTS[subtag]) return SUBTAG_SCRIPTS[subtag];
+  const base = (parts[0] ?? "").toLowerCase();
+  if (LATIN_BASE.has(base)) return ["Latin"];
+  const primary = SCRIPT_TABLE[base];
+  if (primary) return [...primary, "Latin"];
+  return ["Latin"];
+}
+function allowedFor(descriptor, locale) {
+  const script = descriptor.options.script;
+  if (script && script !== "auto") {
+    const explicit = script[locale];
+    if (explicit) return explicit;
+  }
+  return inferScripts(locale);
+}
+var BASELINE = String.raw`\p{White_Space}\p{Script=Common}\p{Script=Inherited}\p{Mark}`;
+function fullMatcher(scripts) {
+  const cls = scripts.map((s) => `\\p{Script=${s}}`).join("");
+  return new RegExp(`^[${BASELINE}${cls}]*$`, "u");
+}
+function charMatcher(scripts) {
+  const cls = scripts.map((s) => `\\p{Script=${s}}`).join("");
+  return new RegExp(`[${BASELINE}${cls}]`, "u");
+}
+function offendingSample(str, scripts) {
+  const ok = charMatcher(scripts);
+  const bad = [];
+  for (const ch of str) {
+    if (!ok.test(ch)) bad.push(ch);
+    if (bad.length >= 8) break;
+  }
+  return bad.join("");
+}
+function stripDisallowed(str, scripts) {
+  const ok = charMatcher(scripts);
+  let out = "";
+  for (const ch of str) if (ok.test(ch)) out += ch;
+  return out;
+}
+function enforceScript(value, field, descriptor) {
+  const opt = descriptor.options;
+  if (!opt.script) return { value, warnings: [] };
+  const mode = opt.onScriptViolation ?? "reject";
+  const warnings = [];
+  let out = value;
+  for (const [locale, raw] of Object.entries(value)) {
+    if (typeof raw !== "string") continue;
+    const allowed = allowedFor(descriptor, locale);
+    if (fullMatcher(allowed).test(raw)) continue;
+    const sample = offendingSample(raw, allowed);
+    if (mode === "reject") {
+      throw new ScriptViolationError(field, locale, allowed, sample);
+    }
+    warnings.push({ field, locale, expected: allowed, sample });
+    if (mode === "filter") {
+      if (out === value) out = { ...value };
+      out[locale] = stripDisallowed(raw, allowed);
+    }
+  }
+  return { value: out, warnings };
+}
+
+export {
+  inferScripts,
+  enforceScript
+};
+//# sourceMappingURL=chunk-6STK5TQP.js.map

package/dist/chunk-6STK5TQP.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/i18n/script.ts"],"sourcesContent":["/**\n * Per-locale script enforcement for `i18nText` fields (write-time).\n *\n * Each locale slot's string is validated against an allowed set of\n * Unicode scripts. `'auto'` infers the set from the locale code with\n * **asymmetric Latin tolerance** (#283): every non-Latin-script locale\n * also allows `Latin`, because proper names and addresses in those\n * locales routinely embed Latin brand/building/technical names — while\n * Latin-script locales do NOT allow other scripts, so the common error\n * (e.g. Thai text dumped into an `en` slot) is still caught.\n *\n * The always-on baseline is `Common` (digits, punctuation), `Inherited`\n * and `Mark` (combining diacritics, joiners, harakat, tone marks), and\n * whitespace — so Latin digits and in-script combining marks never\n * false-reject.\n *\n * @public\n */\nimport { ScriptViolationError } from '../errors.js'\nimport type { I18nTextDescriptor } from './core.js'\n\n/** Locales whose base language is written in the Latin script. */\nconst LATIN_BASE = new Set([\n  'en', 'fr', 'de', 'es', 'it', 'pt', 'nl', 'sv', 'no', 'da', 'fi', 'is',\n  'pl', 'cs', 'sk', 'hu', 'ro', 'hr', 'sl', 'et', 'lv', 'lt', 'tr', 'vi',\n  'id', 'ms', 'tl', 'sw', 'af', 'ca', 'gl', 'eu', 'cy', 'ga',\n])\n\n/** Base-language → primary (non-Latin) scripts. Latin is appended by inferScripts. */\nconst SCRIPT_TABLE: Record<string, readonly string[]> = {\n  th: ['Thai'],\n  ko: ['Hangul', 'Han'],\n  ja: ['Han', 'Hiragana', 'Katakana'],\n  zh: ['Han'],\n  ar: ['Arabic'],\n  fa: ['Arabic'],\n  ur: ['Arabic'],\n  ru: ['Cyrillic'],\n  uk: ['Cyrillic'],\n  bg: ['Cyrillic'],\n  sr: ['Cyrillic'],\n  he: ['Hebrew'],\n  el: ['Greek'],\n  hi: ['Devanagari'],\n  ta: ['Tamil'],\n  km: ['Khmer'],\n  lo: ['Lao'],\n  my: ['Myanmar'],\n}\n\n/** Map a BCP-47 script subtag (e.g. `Latn`, `Cyrl`) to allowed scripts. */\nconst SUBTAG_SCRIPTS: Record<string, readonly string[]> = {\n  Latn: ['Latin'],\n  Cyrl: ['Cyrillic', 'Latin'],\n  Hans: ['Han', 'Latin'],\n  Hant: ['Han', 'Latin'],\n  Thai: ['Thai', 'Latin'],\n  Arab: ['Arabic', 'Latin'],\n}\n\n/**\n * Infer the allowed Unicode scripts for a BCP-47 locale, with asymmetric\n * Latin tolerance. A script subtag (`th-Latn`) wins over the base\n * language. Unknown locales default to `['Latin']`.\n */\nexport function inferScripts(locale: string): readonly string[] {\n  const parts = locale.split('-')\n  const subtag = parts.find((t) => /^[A-Z][a-z]{3}$/.test(t))\n  if (subtag && SUBTAG_SCRIPTS[subtag]) return SUBTAG_SCRIPTS[subtag]\n\n  const base = (parts[0] ?? '').toLowerCase()\n  if (LATIN_BASE.has(base)) return ['Latin']\n  const primary = SCRIPT_TABLE[base]\n  if (primary) return [...primary, 'Latin'] // asymmetric Latin tolerance (#283)\n  return ['Latin']\n}\n\n/** Resolve the allowed scripts for a field's locale slot. */\nfunction allowedFor(descriptor: I18nTextDescriptor, locale: string): readonly string[] {\n  const script = descriptor.options.script\n  if (script && script !== 'auto') {\n    const explicit = script[locale]\n    if (explicit) return explicit\n  }\n  return inferScripts(locale)\n}\n\n/** Always-allowed baseline character classes (besides the named scripts). */\nconst BASELINE = String.raw`\\p{White_Space}\\p{Script=Common}\\p{Script=Inherited}\\p{Mark}`\n\n/** Build a whole-string matcher for the allowed scripts. */\nfunction fullMatcher(scripts: readonly string[]): RegExp {\n  const cls = scripts.map((s) => `\\\\p{Script=${s}}`).join('')\n  return new RegExp(`^[${BASELINE}${cls}]*$`, 'u')\n}\n\n/** Build a single-character matcher (for sampling / stripping). */\nfunction charMatcher(scripts: readonly string[]): RegExp {\n  const cls = scripts.map((s) => `\\\\p{Script=${s}}`).join('')\n  return new RegExp(`[${BASELINE}${cls}]`, 'u')\n}\n\n/** Collect a short sample of characters that violate the allowed scripts. */\nfunction offendingSample(str: string, scripts: readonly string[]): string {\n  const ok = charMatcher(scripts)\n  const bad: string[] = []\n  for (const ch of str) {\n    if (!ok.test(ch)) bad.push(ch)\n    if (bad.length >= 8) break\n  }\n  return bad.join('')\n}\n\n/** Remove characters that violate the allowed scripts. */\nfunction stripDisallowed(str: string, scripts: readonly string[]): string {\n  const ok = charMatcher(scripts)\n  let out = ''\n  for (const ch of str) if (ok.test(ch)) out += ch\n  return out\n}\n\n/** A non-fatal script violation recorded under `'filter'`/`'warn'` modes. */\nexport interface ScriptWarning {\n  readonly field: string\n  readonly locale: string\n  readonly expected: readonly string[]\n  readonly sample: string\n}\n\n/**\n * Enforce a field's script constraint over an i18nText value map.\n *\n * - No `script` option ⇒ returns the value unchanged.\n * - `onScriptViolation: 'reject'` (default) ⇒ throws {@link ScriptViolationError}.\n * - `'filter'` ⇒ returns a copy with disallowed characters stripped + warnings.\n * - `'warn'` ⇒ returns the value unchanged + warnings.\n */\nexport function enforceScript(\n  value: Record<string, unknown>,\n  field: string,\n  descriptor: I18nTextDescriptor,\n): { value: Record<string, unknown>; warnings: ScriptWarning[] } {\n  const opt = descriptor.options\n  if (!opt.script) return { value, warnings: [] }\n\n  const mode = opt.onScriptViolation ?? 'reject'\n  const warnings: ScriptWarning[] = []\n  let out = value\n\n  for (const [locale, raw] of Object.entries(value)) {\n    if (typeof raw !== 'string') continue\n    const allowed = allowedFor(descriptor, locale)\n    if (fullMatcher(allowed).test(raw)) continue\n\n    const sample = offendingSample(raw, allowed)\n    if (mode === 'reject') {\n      throw new ScriptViolationError(field, locale, allowed, sample)\n    }\n    warnings.push({ field, locale, expected: allowed, sample })\n    if (mode === 'filter') {\n      if (out === value) out = { ...value }\n      out[locale] = stripDisallowed(raw, allowed)\n    }\n  }\n\n  return { value: out, warnings }\n}\n"],"mappings":";;;;;AAsBA,IAAM,aAAa,oBAAI,IAAI;AAAA,EACzB;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAClE;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAClE;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AAAA,EAAM;AACxD,CAAC;AAGD,IAAM,eAAkD;AAAA,EACtD,IAAI,CAAC,MAAM;AAAA,EACX,IAAI,CAAC,UAAU,KAAK;AAAA,EACpB,IAAI,CAAC,OAAO,YAAY,UAAU;AAAA,EAClC,IAAI,CAAC,KAAK;AAAA,EACV,IAAI,CAAC,QAAQ;AAAA,EACb,IAAI,CAAC,QAAQ;AAAA,EACb,IAAI,CAAC,QAAQ;AAAA,EACb,IAAI,CAAC,UAAU;AAAA,EACf,IAAI,CAAC,UAAU;AAAA,EACf,IAAI,CAAC,UAAU;AAAA,EACf,IAAI,CAAC,UAAU;AAAA,EACf,IAAI,CAAC,QAAQ;AAAA,EACb,IAAI,CAAC,OAAO;AAAA,EACZ,IAAI,CAAC,YAAY;AAAA,EACjB,IAAI,CAAC,OAAO;AAAA,EACZ,IAAI,CAAC,OAAO;AAAA,EACZ,IAAI,CAAC,KAAK;AAAA,EACV,IAAI,CAAC,SAAS;AAChB;AAGA,IAAM,iBAAoD;AAAA,EACxD,MAAM,CAAC,OAAO;AAAA,EACd,MAAM,CAAC,YAAY,OAAO;AAAA,EAC1B,MAAM,CAAC,OAAO,OAAO;AAAA,EACrB,MAAM,CAAC,OAAO,OAAO;AAAA,EACrB,MAAM,CAAC,QAAQ,OAAO;AAAA,EACtB,MAAM,CAAC,UAAU,OAAO;AAC1B;AAOO,SAAS,aAAa,QAAmC;AAC9D,QAAM,QAAQ,OAAO,MAAM,GAAG;AAC9B,QAAM,SAAS,MAAM,KAAK,CAAC,MAAM,kBAAkB,KAAK,CAAC,CAAC;AAC1D,MAAI,UAAU,eAAe,MAAM,EAAG,QAAO,eAAe,MAAM;AAElE,QAAM,QAAQ,MAAM,CAAC,KAAK,IAAI,YAAY;AAC1C,MAAI,WAAW,IAAI,IAAI,EAAG,QAAO,CAAC,OAAO;AACzC,QAAM,UAAU,aAAa,IAAI;AACjC,MAAI,QAAS,QAAO,CAAC,GAAG,SAAS,OAAO;AACxC,SAAO,CAAC,OAAO;AACjB;AAGA,SAAS,WAAW,YAAgC,QAAmC;AACrF,QAAM,SAAS,WAAW,QAAQ;AAClC,MAAI,UAAU,WAAW,QAAQ;AAC/B,UAAM,WAAW,OAAO,MAAM;AAC9B,QAAI,SAAU,QAAO;AAAA,EACvB;AACA,SAAO,aAAa,MAAM;AAC5B;AAGA,IAAM,WAAW,OAAO;AAGxB,SAAS,YAAY,SAAoC;AACvD,QAAM,MAAM,QAAQ,IAAI,CAAC,MAAM,cAAc,CAAC,GAAG,EAAE,KAAK,EAAE;AAC1D,SAAO,IAAI,OAAO,KAAK,QAAQ,GAAG,GAAG,OAAO,GAAG;AACjD;AAGA,SAAS,YAAY,SAAoC;AACvD,QAAM,MAAM,QAAQ,IAAI,CAAC,MAAM,cAAc,CAAC,GAAG,EAAE,KAAK,EAAE;AAC1D,SAAO,IAAI,OAAO,IAAI,QAAQ,GAAG,GAAG,KAAK,GAAG;AAC9C;AAGA,SAAS,gBAAgB,KAAa,SAAoC;AACxE,QAAM,KAAK,YAAY,OAAO;AAC9B,QAAM,MAAgB,CAAC;AACvB,aAAW,MAAM,KAAK;AACpB,QAAI,CAAC,GAAG,KAAK,EAAE,EAAG,KAAI,KAAK,EAAE;AAC7B,QAAI,IAAI,UAAU,EAAG;AAAA,EACvB;AACA,SAAO,IAAI,KAAK,EAAE;AACpB;AAGA,SAAS,gBAAgB,KAAa,SAAoC;AACxE,QAAM,KAAK,YAAY,OAAO;AAC9B,MAAI,MAAM;AACV,aAAW,MAAM,IAAK,KAAI,GAAG,KAAK,EAAE,EAAG,QAAO;AAC9C,SAAO;AACT;AAkBO,SAAS,cACd,OACA,OACA,YAC+D;AAC/D,QAAM,MAAM,WAAW;AACvB,MAAI,CAAC,IAAI,OAAQ,QAAO,EAAE,OAAO,UAAU,CAAC,EAAE;AAE9C,QAAM,OAAO,IAAI,qBAAqB;AACtC,QAAM,WAA4B,CAAC;AACnC,MAAI,MAAM;AAEV,aAAW,CAAC,QAAQ,GAAG,KAAK,OAAO,QAAQ,KAAK,GAAG;AACjD,QAAI,OAAO,QAAQ,SAAU;AAC7B,UAAM,UAAU,WAAW,YAAY,MAAM;AAC7C,QAAI,YAAY,OAAO,EAAE,KAAK,GAAG,EAAG;AAEpC,UAAM,SAAS,gBAAgB,KAAK,OAAO;AAC3C,QAAI,SAAS,UAAU;AACrB,YAAM,IAAI,qBAAqB,OAAO,QAAQ,SAAS,MAAM;AAAA,IAC/D;AACA,aAAS,KAAK,EAAE,OAAO,QAAQ,UAAU,SAAS,OAAO,CAAC;AAC1D,QAAI,SAAS,UAAU;AACrB,UAAI,QAAQ,MAAO,OAAM,EAAE,GAAG,MAAM;AACpC,UAAI,MAAM,IAAI,gBAAgB,KAAK,OAAO;AAAA,IAC5C;AAAA,EACF;AAEA,SAAO,EAAE,OAAO,KAAK,SAAS;AAChC;","names":[]}

package/dist/{chunk-REQJJIPP.js → chunk-B6PB7JLN.js}

@@ -513,6 +513,27 @@ var LocaleNotSpecifiedError = class extends NoydbError {
     this.field = field;
   }
 };
+var ScriptViolationError = class extends NoydbError {
+  /** The field whose value violated its script constraint. */
+  field;
+  /** The locale slot (e.g. `'en'`) that was checked. */
+  locale;
+  /** The Unicode scripts allowed for this slot. */
+  expected;
+  /** A short sample of the offending characters, for diagnostics. */
+  sample;
+  constructor(field, locale, expected, sample, message) {
+    super(
+      "SCRIPT_VIOLATION",
+      message ?? `Field "${field}" slot "${locale}" expects script(s) [${expected.join(", ")}] but contains disallowed character(s): "${sample}".`
+    );
+    this.name = "ScriptViolationError";
+    this.field = field;
+    this.locale = locale;
+    this.expected = expected;
+    this.sample = sample;
+  }
+};
 var TranslatorNotConfiguredError = class extends NoydbError {
   /** The field that requested auto-translation. */
   field;
@@ -893,6 +914,7 @@ export {
   DictKeyInUseError,
   MissingTranslationError,
   LocaleNotSpecifiedError,
+  ScriptViolationError,
   TranslatorNotConfiguredError,
   BackupLedgerError,
   BackupCorruptedError,
@@ -924,4 +946,4 @@ export {
   OverlayIdMismatchError,
   SnapshotNotFoundError
 };
-//# sourceMappingURL=chunk-REQJJIPP.js.map
+//# sourceMappingURL=chunk-B6PB7JLN.js.map