@bicharts/shape-core 0.1.13 → 0.1.14

package/dist/index.mjs

@@ -1,251 +1,2 @@
 /*! @bicharts/shape-core — Apache-2.0. Bundled reference data: GeoNames (https://www.geonames.org/) CC BY 4.0; US Census/TIGER (public domain). Full text: NOTICE in this package. */
-import {
-  IndexedText,
-  classifyAdditivity,
-  classifyNumericValueNature,
-  classifyTemporal,
-  hostAggHint,
-  isIdentifierName
-} from "./chunk-NGV56PGT.mjs";
-import {
-  GET_RANDOM,
-  SIMPLE_STRING_HASH,
-  STR
-} from "./chunk-VLMGGCN4.mjs";
-import {
-  buildGeoIsoColumn,
-  detectGeo,
-  isJoinGeoKind,
-  toGeoIso
-} from "./chunk-YJVU7G63.mjs";
-import {
-  buildGeoPointColumns,
-  cityMatchPct,
-  isKnownCity,
-  normalizeCountry,
-  normalizePlaceName,
-  resolveAdmin1,
-  resolveGeoPoint,
-  zipPrefixCandidates,
-  zipToPrefix3
-} from "./chunk-QTEE53OQ.mjs";
-import {
-  detectOrdinalDomain,
-  isOrdinalFriendlyName,
-  safeDistinctValuesToShip
-} from "./chunk-VHWC3THS.mjs";
-import {
-  detectFormatSignature
-} from "./chunk-ICDJLWEM.mjs";
-import {
-  monthLookupFor,
-  normalizeMonthKey
-} from "./chunk-RZAZ7PZK.mjs";
-
-// src/geoPointRoles.ts
-var THRESHOLD_PCT = 85;
-var MIN_DISTINCT_BACKFILL = 2;
-var ZIP_THRESHOLD_PCT = 100;
-var MAX_DISTINCT_SCAN = 400;
-var COUNTRY_IDS = /* @__PURE__ */ new Set([
-  "us",
-  "usa",
-  "u s",
-  "u s a",
-  "united states",
-  "united states of america",
-  "america",
-  "ca",
-  "can",
-  "canada",
-  "mx",
-  "mex",
-  "mexico",
-  "estados unidos mexicanos"
-]);
-function distinctNormalized(values) {
-  const seen = /* @__PURE__ */ new Set();
-  for (const v of values) {
-    if (v === null || v === void 0) continue;
-    const k = normalizePlaceName(String(v));
-    if (!k) continue;
-    if (!seen.has(k)) {
-      seen.add(k);
-      if (seen.size >= MAX_DISTINCT_SCAN) break;
-    }
-  }
-  return Array.from(seen);
-}
-function looksLikeCountryColumn(values) {
-  const d = distinctNormalized(values);
-  if (d.length === 0) return false;
-  if (!d.every((v) => COUNTRY_IDS.has(v))) return false;
-  return d.some((v) => !resolveAdmin1(v));
-}
-function isAmbiguousCountryState(normalized) {
-  return COUNTRY_IDS.has(normalized) && !!resolveAdmin1(normalized);
-}
-function admin1MatchPct(values) {
-  const d = distinctNormalized(values);
-  if (d.length === 0) return 0;
-  let n = 0;
-  for (const v of d) if (resolveAdmin1(v)) n++;
-  return Math.round(n / d.length * 1e3) / 10;
-}
-function distinctRaw(values) {
-  const seen = /* @__PURE__ */ new Set();
-  for (const v of values) {
-    if (v === null || v === void 0) continue;
-    const s = String(v).trim();
-    if (!s) continue;
-    if (!seen.has(s)) {
-      seen.add(s);
-      if (seen.size >= MAX_DISTINCT_SCAN) break;
-    }
-  }
-  return Array.from(seen);
-}
-function zipMatchPct(values) {
-  const d = distinctRaw(values);
-  if (d.length === 0) return 0;
-  let n = 0;
-  for (const v of d) if (zipPrefixCandidates(v).length > 0) n++;
-  return Math.round(n / d.length * 1e3) / 10;
-}
-function cityPct(values) {
-  const d = distinctNormalized(values);
-  if (d.length === 0) return 0;
-  let n = 0;
-  for (const v of d) if (isKnownCity(v)) n++;
-  return Math.round(n / d.length * 1e3) / 10;
-}
-function distinctMatches(values, pred) {
-  let n = 0;
-  for (const v of distinctNormalized(values)) if (pred(v)) n++;
-  return n;
-}
-function distinctRawMatches(values, pred) {
-  let n = 0;
-  for (const v of distinctRaw(values)) if (pred(v)) n++;
-  return n;
-}
-function resolvePointRoles(columns, rows, hint) {
-  const backfilled = [];
-  const refused = [];
-  const out = {};
-  const colSet = new Set(columns);
-  const valuesOf = (name) => rows.map((r) => r[name]);
-  for (const role of ["city", "zip", "lat", "lon", "country"]) {
-    const name = hint?.[role];
-    if (name && colSet.has(name)) out[role] = name;
-  }
-  const hintedState = hint?.state;
-  if (hintedState && colSet.has(hintedState)) {
-    const vals = valuesOf(hintedState);
-    const dvals = distinctNormalized(vals);
-    if (dvals.length > 0 && dvals.every(isAmbiguousCountryState)) {
-      refused.push(`state=${hintedState} (every value is "CA", which is both Canada and California)`);
-    } else if (looksLikeCountryColumn(vals)) {
-      refused.push(`state=${hintedState} (every value is a country, not a state)`);
-      if (!out.country) out.country = hintedState;
-    } else if (admin1MatchPct(vals) < THRESHOLD_PCT) {
-      refused.push(`state=${hintedState} (only ${admin1MatchPct(vals)}% of values are states/provinces)`);
-    } else {
-      out.state = hintedState;
-    }
-  }
-  const taken = new Set(Object.values(out).filter(Boolean));
-  const candidates = columns.filter((c) => !taken.has(c) && !c.startsWith("__"));
-  if (!out.state) {
-    let best = null;
-    for (const c of candidates) {
-      const vals = valuesOf(c);
-      if (looksLikeCountryColumn(vals)) continue;
-      const pct = admin1MatchPct(vals);
-      if (pct < THRESHOLD_PCT) continue;
-      const hits = distinctMatches(vals, (v) => !!resolveAdmin1(v));
-      const unambiguous = distinctMatches(vals, (v) => !!resolveAdmin1(v) && !isAmbiguousCountryState(v));
-      if (hits < MIN_DISTINCT_BACKFILL && unambiguous < 1) continue;
-      if (!best || pct > best.pct) best = { name: c, pct };
-    }
-    if (best) {
-      out.state = best.name;
-      taken.add(best.name);
-      backfilled.push(`state=${best.name} (${best.pct}% states/provinces)`);
-    }
-  }
-  if (!out.zip) {
-    for (const c of candidates) {
-      if (taken.has(c)) continue;
-      const vals = valuesOf(c);
-      const pct = zipMatchPct(vals);
-      if (pct < ZIP_THRESHOLD_PCT) continue;
-      if (distinctRawMatches(vals, (v) => zipPrefixCandidates(v).length > 0) < MIN_DISTINCT_BACKFILL) continue;
-      out.zip = c;
-      taken.add(c);
-      backfilled.push(`zip=${c} (${pct}% ZIP codes)`);
-      break;
-    }
-  }
-  if (!out.country) {
-    for (const c of candidates) {
-      if (taken.has(c)) continue;
-      if (!looksLikeCountryColumn(valuesOf(c))) continue;
-      out.country = c;
-      taken.add(c);
-      backfilled.push(`country=${c}`);
-      break;
-    }
-  }
-  if (!out.city) {
-    let best = null;
-    for (const c of candidates) {
-      if (taken.has(c)) continue;
-      const vals = valuesOf(c);
-      const pct = cityPct(vals);
-      if (pct < THRESHOLD_PCT) continue;
-      if (distinctMatches(vals, (v) => isKnownCity(v)) < MIN_DISTINCT_BACKFILL) continue;
-      if (!best || pct > best.pct) best = { name: c, pct };
-    }
-    if (best) {
-      out.city = best.name;
-      backfilled.push(`city=${best.name} (${best.pct}% known cities)`);
-    }
-  }
-  const any = !!(out.city || out.state || out.zip || out.lat && out.lon);
-  return { bind: any ? out : null, backfilled, refused };
-}
-export {
-  GET_RANDOM,
-  IndexedText,
-  SIMPLE_STRING_HASH,
-  STR,
-  admin1MatchPct,
-  buildGeoIsoColumn,
-  buildGeoPointColumns,
-  cityMatchPct,
-  classifyAdditivity,
-  classifyNumericValueNature,
-  classifyTemporal,
-  detectFormatSignature,
-  detectGeo,
-  detectOrdinalDomain,
-  hostAggHint,
-  isIdentifierName,
-  isJoinGeoKind,
-  isOrdinalFriendlyName,
-  looksLikeCountryColumn,
-  monthLookupFor,
-  normalizeCountry,
-  normalizeMonthKey,
-  normalizePlaceName,
-  resolveAdmin1,
-  resolveGeoPoint,
-  resolvePointRoles,
-  safeDistinctValuesToShip,
-  toGeoIso,
-  zipMatchPct,
-  zipPrefixCandidates,
-  zipToPrefix3
-};
-//# sourceMappingURL=index.mjs.map
+import{a as U,b as X,c as Y,d as tt,e as nt,f as et}from"./chunk-XWCQ7BJB.mjs";import{a as j,b as J,c as Q}from"./chunk-NWDPA7DG.mjs";import{a as K,b as V,c as E,d as F}from"./chunk-GY54KQLP.mjs";import{a as C,b as f,c as $,d as D,e as g,f as H,g as O,h as b,i as B}from"./chunk-3UOCZ2D6.mjs";import{a as G,b as N,c as _}from"./chunk-WBEROMBG.mjs";import{a as q}from"./chunk-QEH7WTLB.mjs";import{a as W,b as Z}from"./chunk-2LR73DNS.mjs";var R=85,S=2,ot=100,L=400,w=new Set(["us","usa","u s","u s a","united states","united states of america","america","ca","can","canada","mx","mex","mexico","estados unidos mexicanos"]);function m(i){let n=new Set;for(let e of i){if(e==null)continue;let o=C(String(e));if(o&&!n.has(o)&&(n.add(o),n.size>=L))break}return Array.from(n)}function P(i){let n=m(i);return n.length===0||!n.every(e=>w.has(e))?!1:n.some(e=>!f(e))}function M(i){return w.has(i)&&!!f(i)}function x(i){let n=m(i);if(n.length===0)return 0;let e=0;for(let o of n)f(o)&&e++;return Math.round(e/n.length*1e3)/10}function z(i){let n=new Set;for(let e of i){if(e==null)continue;let o=String(e).trim();if(o&&!n.has(o)&&(n.add(o),n.size>=L))break}return Array.from(n)}function I(i){let n=z(i);if(n.length===0)return 0;let e=0;for(let o of n)g(o).length>0&&e++;return Math.round(e/n.length*1e3)/10}function it(i){let n=m(i);if(n.length===0)return 0;let e=0;for(let o of n)b(o)&&e++;return Math.round(e/n.length*1e3)/10}function k(i,n){let e=0;for(let o of m(i))n(o)&&e++;return e}function st(i,n){let e=0;for(let o of z(i))n(o)&&e++;return e}function rt(i,n,e){let o=[],p=[],s={},A=new Set(i),d=t=>n.map(r=>r[t]);for(let t of["city","zip","lat","lon","country"]){let r=e?.[t];r&&A.has(r)&&(s[t]=r)}let u=e?.state;if(u&&A.has(u)){let t=d(u),r=m(t);r.length>0&&r.every(M)?p.push(`state=${u} (every value is "CA", which is both Canada and California)`):P(t)?(p.push(`state=${u} (every value is a country, not a state)`),s.country||(s.country=u)):x(t)<R?p.push(`state=${u} (only ${x(t)}% of values are states/provinces)`):s.state=u}let l=new Set(Object.values(s).filter(Boolean)),y=i.filter(t=>!l.has(t)&&!t.startsWith("__"));if(!s.state){let t=null;for(let r of y){let a=d(r);if(P(a))continue;let c=x(a);if(c<R)continue;let v=k(a,h=>!!f(h)),T=k(a,h=>!!f(h)&&!M(h));v<S&&T<1||(!t||c>t.pct)&&(t={name:r,pct:c})}t&&(s.state=t.name,l.add(t.name),o.push(`state=${t.name} (${t.pct}% states/provinces)`))}if(!s.zip)for(let t of y){if(l.has(t))continue;let r=d(t),a=I(r);if(!(a<ot)&&!(st(r,c=>g(c).length>0)<S)){s.zip=t,l.add(t),o.push(`zip=${t} (${a}% ZIP codes)`);break}}if(!s.country){for(let t of y)if(!l.has(t)&&P(d(t))){s.country=t,l.add(t),o.push(`country=${t}`);break}}if(!s.city){let t=null;for(let r of y){if(l.has(r))continue;let a=d(r),c=it(a);c<R||k(a,v=>b(v))<S||(!t||c>t.pct)&&(t={name:r,pct:c})}t&&(s.city=t.name,o.push(`city=${t.name} (${t.pct}% known cities)`))}return{bind:!!(s.city||s.state||s.zip||s.lat&&s.lon)?s:null,backfilled:o,refused:p}}export{Q as GET_RANDOM,et as IndexedText,J as SIMPLE_STRING_HASH,j as STR,x as admin1MatchPct,F as buildGeoIsoColumn,O as buildGeoPointColumns,B as cityMatchPct,tt as classifyAdditivity,nt as classifyNumericValueNature,X as classifyTemporal,q as detectFormatSignature,K as detectGeo,_ as detectOrdinalDomain,Y as hostAggHint,U as isIdentifierName,E as isJoinGeoKind,G as isOrdinalFriendlyName,P as looksLikeCountryColumn,Z as monthLookupFor,$ as normalizeCountry,W as normalizeMonthKey,C as normalizePlaceName,f as resolveAdmin1,H as resolveGeoPoint,rt as resolvePointRoles,N as safeDistinctValuesToShip,V as toGeoIso,I as zipMatchPct,g as zipPrefixCandidates,D as zipToPrefix3};

package/dist/indexedText.mjs

@@ -1,24 +1,2 @@
 /*! @bicharts/shape-core — Apache-2.0. Bundled reference data: GeoNames (https://www.geonames.org/) CC BY 4.0; US Census/TIGER (public domain). Full text: NOTICE in this package. */
-import {
-  IndexedText,
-  classifyAdditivity,
-  classifyNumericValueNature,
-  classifyTemporal,
-  hostAggHint,
-  isIdentifierName
-} from "./chunk-NGV56PGT.mjs";
-import "./chunk-VLMGGCN4.mjs";
-import "./chunk-YJVU7G63.mjs";
-import "./chunk-QTEE53OQ.mjs";
-import "./chunk-VHWC3THS.mjs";
-import "./chunk-ICDJLWEM.mjs";
-import "./chunk-RZAZ7PZK.mjs";
-export {
-  IndexedText,
-  classifyAdditivity,
-  classifyNumericValueNature,
-  classifyTemporal,
-  hostAggHint,
-  isIdentifierName
-};
-//# sourceMappingURL=indexedText.mjs.map
+import{a,b,c,d,e,f}from"./chunk-XWCQ7BJB.mjs";import"./chunk-NWDPA7DG.mjs";import"./chunk-GY54KQLP.mjs";import"./chunk-3UOCZ2D6.mjs";import"./chunk-WBEROMBG.mjs";import"./chunk-QEH7WTLB.mjs";import"./chunk-2LR73DNS.mjs";export{f as IndexedText,d as classifyAdditivity,e as classifyNumericValueNature,b as classifyTemporal,c as hostAggHint,a as isIdentifierName};

package/dist/models.mjs

@@ -1,2 +1 @@
 /*! @bicharts/shape-core — Apache-2.0. Bundled reference data: GeoNames (https://www.geonames.org/) CC BY 4.0; US Census/TIGER (public domain). Full text: NOTICE in this package. */
-//# sourceMappingURL=models.mjs.map

package/dist/monthNames.mjs

@@ -1,10 +1,2 @@
 /*! @bicharts/shape-core — Apache-2.0. Bundled reference data: GeoNames (https://www.geonames.org/) CC BY 4.0; US Census/TIGER (public domain). Full text: NOTICE in this package. */
-import {
-  monthLookupFor,
-  normalizeMonthKey
-} from "./chunk-RZAZ7PZK.mjs";
-export {
-  monthLookupFor,
-  normalizeMonthKey
-};
-//# sourceMappingURL=monthNames.mjs.map
+import{a,b}from"./chunk-2LR73DNS.mjs";export{b as monthLookupFor,a as normalizeMonthKey};

package/dist/ordinalDetector.mjs

@@ -1,12 +1,2 @@
 /*! @bicharts/shape-core — Apache-2.0. Bundled reference data: GeoNames (https://www.geonames.org/) CC BY 4.0; US Census/TIGER (public domain). Full text: NOTICE in this package. */
-import {
-  detectOrdinalDomain,
-  isOrdinalFriendlyName,
-  safeDistinctValuesToShip
-} from "./chunk-VHWC3THS.mjs";
-export {
-  detectOrdinalDomain,
-  isOrdinalFriendlyName,
-  safeDistinctValuesToShip
-};
-//# sourceMappingURL=ordinalDetector.mjs.map
+import{a,b,c}from"./chunk-WBEROMBG.mjs";export{c as detectOrdinalDomain,a as isOrdinalFriendlyName,b as safeDistinctValuesToShip};

package/dist/util.mjs

@@ -1,12 +1,2 @@
 /*! @bicharts/shape-core — Apache-2.0. Bundled reference data: GeoNames (https://www.geonames.org/) CC BY 4.0; US Census/TIGER (public domain). Full text: NOTICE in this package. */
-import {
-  GET_RANDOM,
-  SIMPLE_STRING_HASH,
-  STR
-} from "./chunk-VLMGGCN4.mjs";
-export {
-  GET_RANDOM,
-  SIMPLE_STRING_HASH,
-  STR
-};
-//# sourceMappingURL=util.mjs.map
+import{a,b,c}from"./chunk-NWDPA7DG.mjs";export{c as GET_RANDOM,b as SIMPLE_STRING_HASH,a as STR};

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bicharts/shape-core",
-  "version": "0.1.13",
+  "version": "0.1.14",
   "description": "Host-agnostic data-shape profiler: ingest a dataset's rows through IValueCollection.addRow and read back the measured shape as an LLMColumnWithValue[] payload — types, cardinality, temporal/ordinal detection, geographic region and point resolution, cross-column signals. MEASUREMENT only: the policy that decides what a measured shape means for chart selection lives server-side.",
   "license": "Apache-2.0",
   "type": "module",
@@ -78,7 +78,7 @@
   },
   "scripts": {
     "build": "node build.mjs",
-    "prepack": "node build.mjs",
+    "prepack": "node build.mjs --minify",
     "typecheck": "tsc -p tsconfig.json --noEmit"
   }
 }

package/dist/chunk-ICDJLWEM.mjs

@@ -1,83 +0,0 @@
-/*! @bicharts/shape-core — Apache-2.0. Bundled reference data: GeoNames (https://www.geonames.org/) CC BY 4.0; US Census/TIGER (public domain). Full text: NOTICE in this package. */
-
-// src/formatDetector.ts
-var SIGNATURE_FLOOR = 0.7;
-function classifyOne(v) {
-  if (v === null || v === void 0) return null;
-  const s = String(v);
-  if (s.length === 0) return null;
-  const len = s.length;
-  if (len === 36 && /^[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$/i.test(s)) {
-    return "UUID";
-  }
-  if (/^[^\s@]+@[^\s@]+\.[A-Za-z]{2,}$/.test(s)) return "EMAIL_LIKE";
-  if (/^https?:\/\//i.test(s) || /^www\./i.test(s)) return "URL_LIKE";
-  if (/^[+\d][\d\s().\-]{5,17}\d$/.test(s)) {
-    const digitCount = (s.match(/\d/g) || []).length;
-    if (digitCount >= 7 && digitCount <= 15) return "PHONE_LIKE";
-  }
-  if (/^\d{4}-\d{2}-\d{2}([T ]\d{2}:\d{2}(:\d{2})?(\.\d+)?Z?)?$/.test(s)) return "DATE_STR";
-  if (/^\d{1,2}[\/\-]\d{1,2}[\/\-]\d{2,4}$/.test(s)) return "DATE_STR";
-  if (/^(true|false|yes|no|t|f|y|n)$/i.test(s)) return "BOOLEAN";
-  if (/^\d+$/.test(s)) {
-    if (len >= 6) return "NUMERIC_ID";
-    return null;
-  }
-  if (len >= 8 && /^[0-9a-f]+$/i.test(s)) return "HEX_ID";
-  if (/^[A-Z][A-Z0-9]*$/.test(s) && !/^\d+$/.test(s)) {
-    return "ALL_UPPER";
-  }
-  if (len >= 6 && /^[A-Za-z0-9_-]+$/.test(s) && /\d/.test(s) && /[A-Za-z]/.test(s)) {
-    return "OPAQUE_ID_ALPHANUMERIC";
-  }
-  const wordCount = s.trim().split(/\s+/).length;
-  const hasInternalPunct = /[.!?,;:]/.test(s);
-  if (wordCount >= 2) {
-    if (hasInternalPunct || /[a-z][A-Z]/.test(s) || s.length > 40) return "SENTENCE_TEXT";
-    if (/^[A-Z]/.test(s) && /\b[A-Z][a-z]+/.test(s)) return "TITLE_CASE_WORDS";
-    if (/^[a-z]/.test(s) && !/[A-Z]/.test(s)) return "LOWER_WORDS";
-    return "SENTENCE_TEXT";
-  }
-  if (/^[A-Z][a-z]+$/.test(s)) return "TITLE_CASE_WORDS";
-  if (/^[a-z]+$/.test(s)) return "LOWER_WORDS";
-  return null;
-}
-function detectFormatSignature(values) {
-  if (!values || values.length === 0) return "OTHER";
-  const counts = /* @__PURE__ */ new Map();
-  const lengthBySig = /* @__PURE__ */ new Map();
-  let classified = 0;
-  for (const v of values) {
-    const sig = classifyOne(v);
-    if (!sig) continue;
-    classified++;
-    counts.set(sig, (counts.get(sig) || 0) + 1);
-    if (!lengthBySig.has(sig)) lengthBySig.set(sig, /* @__PURE__ */ new Set());
-    lengthBySig.get(sig).add(String(v).length);
-  }
-  if (classified === 0) return "OTHER";
-  let topSig = null;
-  let topCount = 0;
-  for (const [sig, count] of counts.entries()) {
-    if (count > topCount) {
-      topCount = count;
-      topSig = sig;
-    }
-  }
-  if (!topSig) return "OTHER";
-  const dominance = topCount / values.length;
-  if (dominance < SIGNATURE_FLOOR) return "MIXED";
-  if (topSig === "ALL_UPPER") {
-    const lens = lengthBySig.get("ALL_UPPER");
-    if (lens.size === 1) {
-      const n = lens.values().next().value;
-      return `ALL_UPPER_FIXED_${n}`;
-    }
-  }
-  return topSig;
-}
-
-export {
-  detectFormatSignature
-};
-//# sourceMappingURL=chunk-ICDJLWEM.mjs.map

package/dist/chunk-ICDJLWEM.mjs.map

@@ -1,7 +0,0 @@
-{
-  "version": 3,
-  "sources": ["../src/formatDetector.ts"],
-  "sourcesContent": ["// \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n// Client-side format-signature detector\n// \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n//\n// Replaces `topCategoryValues` (obfuscated string samples) for String-dataType\n// columns. Instead of shipping 5 class-preserving random tokens for the LLM\n// to infer format-class from, we classify ONCE on the client (where the raw\n// values are visible) and ship a single enum-like signature.\n//\n// Genesis (2026-05-27): the server's DataShapeAdvisor.IsLikelyIdColumn was\n// already walking each character of the obfuscated tokens to infer \"id-like\"\n// \u2014 that's a format-classifier in the wrong place. Obfuscation preserves\n// character classes by design, so the server-side inference accidentally\n// works, but the logic belongs on the client where it sees real data. Side\n// benefit: no obfuscated values ship for Strings \u2192 the LLM can't hardcode\n// one as a runtime filter (image-bug 2026-05-26).\n//\n// Signature catalog (returned values):\n//   ALL_UPPER_FIXED_<n>   \u2014 every value is uppercase letters, length exactly n\n//   ALL_UPPER             \u2014 uppercase-only, mixed length\n//   TITLE_CASE_WORDS      \u2014 proper-name shape (e.g. \"Delta Air Lines\")\n//   LOWER_WORDS           \u2014 lowercase-only words, often slugs / tags\n//   SENTENCE_TEXT         \u2014 multi-word strings with mixed casing / punctuation\n//   EMAIL_LIKE            \u2014 contains @ + a domain suffix\n//   URL_LIKE              \u2014 starts with http(s):// or contains www.\n//   PHONE_LIKE            \u2014 digits + parens/dashes/spaces, length 7-15\n//   DATE_STR              \u2014 ISO-ish or slash-formatted dates\n//   NUMERIC_ID            \u2014 all-digit strings of length \u2265 6\n//   HEX_ID                \u2014 hex characters [0-9a-f]+ of length \u2265 8\n//   UUID                  \u2014 8-4-4-4-12 hex with dashes\n//   OPAQUE_ID_ALPHANUMERIC \u2014 alphanumeric + dashes/underscores, no whitespace,\n//                            mixed letters+digits, length \u2265 6\n//   BOOLEAN               \u2014 Boolean-rendered text (True/False, Yes/No)\n//   MIXED                 \u2014 distinct values don't agree on a single signature\n//   OTHER                 \u2014 no signature matched (short categorical labels)\n//\n// Threshold: a signature is returned if \u2265 70% of the sampled distinct values\n// match it. The 70% floor tolerates a few stragglers (e.g. one \"TBD\" in an\n// otherwise ID-like column) without dropping to OTHER unnecessarily.\n\nconst SIGNATURE_FLOOR = 0.70;\n\n// Per-value classifier. Returns the highest-specificity signature that fits\n// a single value, or null if none.\nfunction classifyOne(v: string): string | null {\n    if (v === null || v === undefined) return null;\n    const s = String(v);\n    if (s.length === 0) return null;\n    const len = s.length;\n\n    // UUID \u2014 most specific, check first.\n    if (len === 36 && /^[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$/i.test(s)) {\n        return \"UUID\";\n    }\n\n    // EMAIL \u2014 @ and a TLD-shaped suffix.\n    if (/^[^\\s@]+@[^\\s@]+\\.[A-Za-z]{2,}$/.test(s)) return \"EMAIL_LIKE\";\n\n    // URL \u2014 http(s):// or www.\n    if (/^https?:\\/\\//i.test(s) || /^www\\./i.test(s)) return \"URL_LIKE\";\n\n    // PHONE \u2014 digits with separators, total digit count 7-15.\n    if (/^[+\\d][\\d\\s().\\-]{5,17}\\d$/.test(s)) {\n        const digitCount = (s.match(/\\d/g) || []).length;\n        if (digitCount >= 7 && digitCount <= 15) return \"PHONE_LIKE\";\n    }\n\n    // DATE \u2014 ISO-ish or common slash/dash formats.\n    if (/^\\d{4}-\\d{2}-\\d{2}([T ]\\d{2}:\\d{2}(:\\d{2})?(\\.\\d+)?Z?)?$/.test(s)) return \"DATE_STR\";\n    if (/^\\d{1,2}[\\/\\-]\\d{1,2}[\\/\\-]\\d{2,4}$/.test(s)) return \"DATE_STR\";\n\n    // BOOLEAN \u2014 common text renderings.\n    if (/^(true|false|yes|no|t|f|y|n)$/i.test(s)) return \"BOOLEAN\";\n\n    // NUMERIC_ID \u2014 long all-digit (short ones are codes / years / ordinals).\n    if (/^\\d+$/.test(s)) {\n        if (len >= 6) return \"NUMERIC_ID\";\n        return null; // short numeric \u2192 falls through to OTHER\n    }\n\n    // HEX_ID \u2014 hex characters only, length \u2265 8.\n    if (len >= 8 && /^[0-9a-f]+$/i.test(s)) return \"HEX_ID\";\n\n    // ALL_UPPER \u2014 uppercase letters and digits, no spaces (codes / abbreviations).\n    if (/^[A-Z][A-Z0-9]*$/.test(s) && !/^\\d+$/.test(s)) {\n        // Caller aggregates by length to decide ALL_UPPER_FIXED_n vs ALL_UPPER.\n        return \"ALL_UPPER\";\n    }\n\n    // OPAQUE_ID \u2014 alphanumeric + dashes/underscores, no whitespace, has BOTH\n    // letters and digits. Length \u2265 6 to skip short tokens like \"A1\" / \"Q2\".\n    if (len >= 6 && /^[A-Za-z0-9_-]+$/.test(s) && /\\d/.test(s) && /[A-Za-z]/.test(s)) {\n        return \"OPAQUE_ID_ALPHANUMERIC\";\n    }\n\n    // Words-and-text classifiers (whitespace OK).\n    // SENTENCE_TEXT \u2014 multi-word with mixed casing OR punctuation.\n    const wordCount = s.trim().split(/\\s+/).length;\n    const hasInternalPunct = /[.!?,;:]/.test(s);\n    if (wordCount >= 2) {\n        if (hasInternalPunct || /[a-z][A-Z]/.test(s) || s.length > 40) return \"SENTENCE_TEXT\";\n        // Multi-word, no internal punct, \u2264 40 chars \u2014 check casing.\n        if (/^[A-Z]/.test(s) && /\\b[A-Z][a-z]+/.test(s)) return \"TITLE_CASE_WORDS\";\n        if (/^[a-z]/.test(s) && !/[A-Z]/.test(s)) return \"LOWER_WORDS\";\n        return \"SENTENCE_TEXT\";\n    }\n\n    // Single word.\n    if (/^[A-Z][a-z]+$/.test(s)) return \"TITLE_CASE_WORDS\";\n    if (/^[a-z]+$/.test(s)) return \"LOWER_WORDS\";\n\n    return null;\n}\n\nexport function detectFormatSignature(values: string[]): string {\n    if (!values || values.length === 0) return \"OTHER\";\n\n    const counts = new Map<string, number>();\n    const lengthBySig = new Map<string, Set<number>>();\n    let classified = 0;\n\n    for (const v of values) {\n        const sig = classifyOne(v);\n        if (!sig) continue;\n        classified++;\n        counts.set(sig, (counts.get(sig) || 0) + 1);\n        if (!lengthBySig.has(sig)) lengthBySig.set(sig, new Set());\n        lengthBySig.get(sig)!.add(String(v).length);\n    }\n\n    if (classified === 0) return \"OTHER\";\n\n    // Pick the dominant signature.\n    let topSig: string | null = null;\n    let topCount = 0;\n    for (const [sig, count] of counts.entries()) {\n        if (count > topCount) {\n            topCount = count;\n            topSig = sig;\n        }\n    }\n    if (!topSig) return \"OTHER\";\n\n    const dominance = topCount / values.length;\n    if (dominance < SIGNATURE_FLOOR) return \"MIXED\";\n\n    // ALL_UPPER_FIXED_n specialization. If every classified-as-ALL_UPPER value\n    // shares the same length, emit ALL_UPPER_FIXED_<n>. Otherwise plain\n    // ALL_UPPER. The product spec (2026-05-27): \"use n instead of hardcoding 2/3\".\n    if (topSig === \"ALL_UPPER\") {\n        const lens = lengthBySig.get(\"ALL_UPPER\")!;\n        if (lens.size === 1) {\n            const n = lens.values().next().value;\n            return `ALL_UPPER_FIXED_${n}`;\n        }\n    }\n\n    return topSig;\n}\n"],
-  "mappings": ";;;AAwCA,IAAM,kBAAkB;AAIxB,SAAS,YAAY,GAA0B;AAC3C,MAAI,MAAM,QAAQ,MAAM,OAAW,QAAO;AAC1C,QAAM,IAAI,OAAO,CAAC;AAClB,MAAI,EAAE,WAAW,EAAG,QAAO;AAC3B,QAAM,MAAM,EAAE;AAGd,MAAI,QAAQ,MAAM,kEAAkE,KAAK,CAAC,GAAG;AACzF,WAAO;AAAA,EACX;AAGA,MAAI,kCAAkC,KAAK,CAAC,EAAG,QAAO;AAGtD,MAAI,gBAAgB,KAAK,CAAC,KAAK,UAAU,KAAK,CAAC,EAAG,QAAO;AAGzD,MAAI,6BAA6B,KAAK,CAAC,GAAG;AACtC,UAAM,cAAc,EAAE,MAAM,KAAK,KAAK,CAAC,GAAG;AAC1C,QAAI,cAAc,KAAK,cAAc,GAAI,QAAO;AAAA,EACpD;AAGA,MAAI,2DAA2D,KAAK,CAAC,EAAG,QAAO;AAC/E,MAAI,sCAAsC,KAAK,CAAC,EAAG,QAAO;AAG1D,MAAI,iCAAiC,KAAK,CAAC,EAAG,QAAO;AAGrD,MAAI,QAAQ,KAAK,CAAC,GAAG;AACjB,QAAI,OAAO,EAAG,QAAO;AACrB,WAAO;AAAA,EACX;AAGA,MAAI,OAAO,KAAK,eAAe,KAAK,CAAC,EAAG,QAAO;AAG/C,MAAI,mBAAmB,KAAK,CAAC,KAAK,CAAC,QAAQ,KAAK,CAAC,GAAG;AAEhD,WAAO;AAAA,EACX;AAIA,MAAI,OAAO,KAAK,mBAAmB,KAAK,CAAC,KAAK,KAAK,KAAK,CAAC,KAAK,WAAW,KAAK,CAAC,GAAG;AAC9E,WAAO;AAAA,EACX;AAIA,QAAM,YAAY,EAAE,KAAK,EAAE,MAAM,KAAK,EAAE;AACxC,QAAM,mBAAmB,WAAW,KAAK,CAAC;AAC1C,MAAI,aAAa,GAAG;AAChB,QAAI,oBAAoB,aAAa,KAAK,CAAC,KAAK,EAAE,SAAS,GAAI,QAAO;AAEtE,QAAI,SAAS,KAAK,CAAC,KAAK,gBAAgB,KAAK,CAAC,EAAG,QAAO;AACxD,QAAI,SAAS,KAAK,CAAC,KAAK,CAAC,QAAQ,KAAK,CAAC,EAAG,QAAO;AACjD,WAAO;AAAA,EACX;AAGA,MAAI,gBAAgB,KAAK,CAAC,EAAG,QAAO;AACpC,MAAI,WAAW,KAAK,CAAC,EAAG,QAAO;AAE/B,SAAO;AACX;AAEO,SAAS,sBAAsB,QAA0B;AAC5D,MAAI,CAAC,UAAU,OAAO,WAAW,EAAG,QAAO;AAE3C,QAAM,SAAS,oBAAI,IAAoB;AACvC,QAAM,cAAc,oBAAI,IAAyB;AACjD,MAAI,aAAa;AAEjB,aAAW,KAAK,QAAQ;AACpB,UAAM,MAAM,YAAY,CAAC;AACzB,QAAI,CAAC,IAAK;AACV;AACA,WAAO,IAAI,MAAM,OAAO,IAAI,GAAG,KAAK,KAAK,CAAC;AAC1C,QAAI,CAAC,YAAY,IAAI,GAAG,EAAG,aAAY,IAAI,KAAK,oBAAI,IAAI,CAAC;AACzD,gBAAY,IAAI,GAAG,EAAG,IAAI,OAAO,CAAC,EAAE,MAAM;AAAA,EAC9C;AAEA,MAAI,eAAe,EAAG,QAAO;AAG7B,MAAI,SAAwB;AAC5B,MAAI,WAAW;AACf,aAAW,CAAC,KAAK,KAAK,KAAK,OAAO,QAAQ,GAAG;AACzC,QAAI,QAAQ,UAAU;AAClB,iBAAW;AACX,eAAS;AAAA,IACb;AAAA,EACJ;AACA,MAAI,CAAC,OAAQ,QAAO;AAEpB,QAAM,YAAY,WAAW,OAAO;AACpC,MAAI,YAAY,gBAAiB,QAAO;AAKxC,MAAI,WAAW,aAAa;AACxB,UAAM,OAAO,YAAY,IAAI,WAAW;AACxC,QAAI,KAAK,SAAS,GAAG;AACjB,YAAM,IAAI,KAAK,OAAO,EAAE,KAAK,EAAE;AAC/B,aAAO,mBAAmB,CAAC;AAAA,IAC/B;AAAA,EACJ;AAEA,SAAO;AACX;",
-  "names": []
-}