@bicharts/shape-core 0.5.7 → 0.5.8

package/dist/index.mjs

@@ -1,2 +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{a as dt,b as yt}from"./chunk-RIM62XDD.mjs";import{a as st,b as ct,c as at,d as ut,e as lt,f as ft}from"./chunk-WIPW5J2O.mjs";import{a as tt}from"./chunk-QEH7WTLB.mjs";import{a as nt,b as et}from"./chunk-2LR73DNS.mjs";import{a as ot,b as it,c as rt}from"./chunk-NWDPA7DG.mjs";import{a as j,b as J,c as Q,d as Y}from"./chunk-JY46MGFW.mjs";import{a as x,f as P,g as K,h as V,i as f,j as q,k as F,l as b,m as U,n as W,o as X,p as v,q as Z}from"./chunk-RPZZSD7D.mjs";import"./chunk-JH3W3DH7.mjs";import{a as H,b as E,c as B}from"./chunk-WBEROMBG.mjs";var A=85,T=2,pt=100,N=400,mt=97;function L(r){return!!P(r)}function p(r){let n=new Set;for(let e of r){if(e==null)continue;let i=x(String(e));if(i&&!n.has(i)&&(n.add(i),n.size>=N))break}return Array.from(n)}function g(r){let n=p(r);if(n.length===0||ht(r)<mt)return!1;let e=n.filter(c=>L(c)&&!f(c)).length,i=n.filter(c=>!L(c)&&!!f(c)).length;return e>=1&&e>=i}var gt=new Set(["USA","CAN","MEX"]);function z(r,n){if(!f(r))return!1;let e=P(r);return e?n==="world"?!0:gt.has(e):!1}function ht(r){let n=p(r);if(n.length===0)return 0;let e=0;for(let i of n)L(i)&&e++;return Math.round(e/n.length*1e3)/10}function M(r){let n=p(r);if(n.length===0)return 0;let e=0;for(let i of n)f(i)&&e++;return Math.round(e/n.length*1e3)/10}function _(r){let n=new Set;for(let e of r){if(e==null)continue;let i=String(e).trim();if(i&&!n.has(i)&&(n.add(i),n.size>=N))break}return Array.from(n)}function D(r){let n=_(r);if(n.length===0)return 0;let e=0;for(let i of n)b(i).length>0&&e++;return Math.round(e/n.length*1e3)/10}function Ct(r,n){let e=p(r);if(e.length===0)return 0;let i=0;for(let c of e)v(c,n)&&i++;return Math.round(i/e.length*1e3)/10}function k(r,n){let e=0;for(let i of p(r))n(i)&&e++;return e}function Pt(r,n){let e=0;for(let i of _(r))n(i)&&e++;return e}function bt(r,n,e,i){let c=[],m=[],o={},w=new Set(r),y=t=>n.map(s=>s[t]);for(let t of["city","zip","lat","lon","country"]){let s=e?.[t];s&&w.has(s)&&(o[t]=s)}let l=e?.state;if(l&&w.has(l)){let t=y(l),s=p(t);s.length>0&&s.every(a=>z(a,i))?m.push(`state=${l} (every value is "CA", which is both Canada and California)`):g(t)?(m.push(`state=${l} (every value is a country, not a state)`),o.country||(o.country=l)):M(t)<A?m.push(`state=${l} (only ${M(t)}% of values are states/provinces)`):o.state=l}let d=new Set(Object.values(o).filter(Boolean)),h=r.filter(t=>!d.has(t)&&!t.startsWith("__"));if(!o.state){let t=null;for(let s of h){let a=y(s);if(g(a))continue;let u=M(a);if(u<A)continue;let S=k(a,C=>!!f(C)),O=k(a,C=>!!f(C)&&!z(C,i));S<T&&O<1||(!t||u>t.pct)&&(t={name:s,pct:u})}t&&(o.state=t.name,d.add(t.name),c.push(`state=${t.name} (${t.pct}% states/provinces)`))}if(!o.zip)for(let t of h){if(d.has(t))continue;let s=y(t),a=D(s);if(!(a<pt)&&!(Pt(s,u=>b(u).length>0)<T)){o.zip=t,d.add(t),c.push(`zip=${t} (${a}% ZIP codes)`);break}}if(!o.country){for(let t of h)if(!d.has(t)&&g(y(t))){o.country=t,d.add(t),c.push(`country=${t}`);break}}if(!o.city){let t=null;for(let s of h){if(d.has(s))continue;let a=y(s),u=Ct(a,i);u<A||k(a,S=>v(S,i))<T||(!t||u>t.pct)&&(t={name:s,pct:u})}t&&(o.city=t.name,c.push(`city=${t.name} (${t.pct}% known cities)`))}let G=o.country?y(o.country):[],R=!!o.country&&g(G),$=R?new Set(G.map(t=>P(t==null?null:String(t))).filter(Boolean)).size:0,I=R&&$>=2;return o.country&&!I&&!(o.city||o.state||o.zip||o.lat&&o.lon)&&m.push(R?`country=${o.country} (every row resolves to the same country - nothing to place apart)`:`country=${o.country} (values are not country identifiers)`),{bind:!!(o.city||o.state||o.zip||I||o.lat&&o.lon)?o:null,backfilled:c,refused:m}}export{rt as GET_RANDOM,ft as IndexedText,it as SIMPLE_STRING_HASH,ot as STR,M as admin1MatchPct,Y as buildGeoIsoColumn,X as buildGeoPointColumns,Z as cityMatchPct,ut as classifyAdditivity,lt as classifyNumericValueNature,ct as classifyTemporal,tt as detectFormatSignature,j as detectGeo,B as detectOrdinalDomain,dt as engineTypeForSqlType,at as hostAggHint,yt as ingest,V as isCityTableLoaded,U as isGeoPointAmbiguity,st as isIdentifierName,Q as isJoinGeoKind,H as isOrdinalFriendlyName,g as looksLikeCountryColumn,et as monthLookupFor,q as normalizeCountry,nt as normalizeMonthKey,x as normalizePlaceName,K as registerCityTable,f as resolveAdmin1,W as resolveGeoPoint,bt as resolvePointRoles,E as safeDistinctValuesToShip,J as toGeoIso,D as zipMatchPct,b as zipPrefixCandidates,F as zipToPrefix3};
+import{a as yt,b as mt}from"./chunk-RIM62XDD.mjs";import{a as ct,b as ut,c as lt,d as ft,e as pt,f as dt}from"./chunk-WIPW5J2O.mjs";import{a as et}from"./chunk-QEH7WTLB.mjs";import{a as ot,b as it}from"./chunk-2LR73DNS.mjs";import{a as rt,b as st,c as at}from"./chunk-NWDPA7DG.mjs";import{a as Q,b as Y,c as tt,d as nt}from"./chunk-JY46MGFW.mjs";import{a as P,f as b,g as q,h as F,i as f,j as Z,k as U,l as A,m as W,n as X,o as j,p as S,q as J}from"./chunk-RPZZSD7D.mjs";import"./chunk-JH3W3DH7.mjs";import{a as O,b as K,c as V}from"./chunk-WBEROMBG.mjs";var k=97,gt=new Set(["n a","none","null","nil","unknown","not applicable","not available","not specified","unspecified","missing","blank","empty","tbd","to be determined"]);function M(i){return i?gt.has(i):!0}var v=85,L=2,ht=100,N=400;function w(i){return!!b(i)}function y(i){let n=new Set;for(let e of i){if(e==null)continue;let r=P(String(e));if(!M(r)&&!n.has(r)&&(n.add(r),n.size>=N))break}return Array.from(n)}function g(i){let n=y(i);if(n.length===0||Pt(i)<97)return!1;let e=n.filter(a=>w(a)&&!f(a)).length,r=n.filter(a=>!w(a)&&!!f(a)).length;return e>=1&&e>=r}var Ct=new Set(["USA","CAN","MEX"]);function E(i,n){if(!f(i))return!1;let e=b(i);return e?n==="world"?!0:Ct.has(e):!1}function Pt(i){let n=y(i);if(n.length===0)return 0;let e=0;for(let r of n)w(r)&&e++;return Math.round(e/n.length*1e3)/10}function T(i){let n=y(i);if(n.length===0)return 0;let e=0;for(let r of n)f(r)&&e++;return Math.round(e/n.length*1e3)/10}function $(i){let n=new Set;for(let e of i){if(e==null)continue;let r=String(e).trim();if(!M(P(r))&&!n.has(r)&&(n.add(r),n.size>=N))break}return Array.from(n)}function B(i){let n=$(i);if(n.length===0)return 0;let e=0;for(let r of n)A(r).length>0&&e++;return Math.round(e/n.length*1e3)/10}function bt(i,n){let e=y(i);if(e.length===0)return 0;let r=0;for(let a of e)S(a,n)&&r++;return Math.round(r/e.length*1e3)/10}function G(i,n){let e=0;for(let r of y(i))n(r)&&e++;return e}function At(i,n){let e=0;for(let r of $(i))n(r)&&e++;return e}function Mt(i,n,e,r){let a=[],m=[],o={},I=new Set(i),d=t=>n.map(s=>s[t]);for(let t of["city","zip","lat","lon","country"]){let s=e?.[t];s&&I.has(s)&&(o[t]=s)}let l=e?.state;if(l&&I.has(l)){let t=d(l),s=y(t);s.length>0&&s.every(c=>E(c,r))?m.push(`state=${l} (every value is "CA", which is both Canada and California)`):g(t)?(m.push(`state=${l} (every value is a country, not a state)`),o.country||(o.country=l)):T(t)<v?m.push(`state=${l} (only ${T(t)}% of values are states/provinces)`):o.state=l}let p=new Set(Object.values(o).filter(Boolean)),h=i.filter(t=>!p.has(t)&&!t.startsWith("__"));if(!o.state){let t=null;for(let s of h){let c=d(s);if(g(c))continue;let u=T(c);if(u<v)continue;let R=G(c,C=>!!f(C)),H=G(c,C=>!!f(C)&&!E(C,r));R<L&&H<1||(!t||u>t.pct)&&(t={name:s,pct:u})}t&&(o.state=t.name,p.add(t.name),a.push(`state=${t.name} (${t.pct}% states/provinces)`))}if(!o.zip)for(let t of h){if(p.has(t))continue;let s=d(t),c=B(s);if(!(c<ht)&&!(At(s,u=>A(u).length>0)<L)){o.zip=t,p.add(t),a.push(`zip=${t} (${c}% ZIP codes)`);break}}if(!o.country){for(let t of h)if(!p.has(t)&&g(d(t))){o.country=t,p.add(t),a.push(`country=${t}`);break}}if(!o.city){let t=null;for(let s of h){if(p.has(s))continue;let c=d(s),u=bt(c,r);u<v||G(c,R=>S(R,r))<L||(!t||u>t.pct)&&(t={name:s,pct:u})}t&&(o.city=t.name,a.push(`city=${t.name} (${t.pct}% known cities)`))}let _=o.country?d(o.country):[],x=!!o.country&&g(_),D=x?new Set(_.map(t=>b(t==null?null:String(t))).filter(Boolean)).size:0,z=x&&D>=2;return o.country&&!z&&!(o.city||o.state||o.zip||o.lat&&o.lon)&&m.push(x?`country=${o.country} (every row resolves to the same country - nothing to place apart)`:`country=${o.country} (values are not country identifiers)`),{bind:!!(o.city||o.state||o.zip||z||o.lat&&o.lon)?o:null,backfilled:a,refused:m}}export{k as GAZETTEER_MATCH_PCT,at as GET_RANDOM,dt as IndexedText,st as SIMPLE_STRING_HASH,rt as STR,T as admin1MatchPct,nt as buildGeoIsoColumn,j as buildGeoPointColumns,J as cityMatchPct,ft as classifyAdditivity,pt as classifyNumericValueNature,ut as classifyTemporal,et as detectFormatSignature,Q as detectGeo,V as detectOrdinalDomain,yt as engineTypeForSqlType,lt as hostAggHint,mt as ingest,M as isBlankLike,F as isCityTableLoaded,W as isGeoPointAmbiguity,ct as isIdentifierName,tt as isJoinGeoKind,O as isOrdinalFriendlyName,g as looksLikeCountryColumn,it as monthLookupFor,Z as normalizeCountry,ot as normalizeMonthKey,P as normalizePlaceName,q as registerCityTable,f as resolveAdmin1,X as resolveGeoPoint,Mt as resolvePointRoles,K as safeDistinctValuesToShip,Y as toGeoIso,B as zipMatchPct,A as zipPrefixCandidates,U as zipToPrefix3};

package/dist/types/index.d.ts

@@ -3,6 +3,7 @@ export type { IValueCollection } from "./indexedText";
 export { classifyTemporal, classifyAdditivity, classifyNumericValueNature, hostAggHint, isIdentifierName, } from "./indexedText";
 export type { LLMColumnWithValue } from "./models";
 export type { LLMRequestModifier, SimpleColumn, VegaRendererPayload, LLMClientHints, LLMRequestCode, LLMQualifyResult, LLMRequestCodeResult, GetLicenseStatusResult, } from "./models";
+export { GAZETTEER_MATCH_PCT, isBlankLike } from "./matchQuality";
 export { detectOrdinalDomain, safeDistinctValuesToShip, isOrdinalFriendlyName } from "./ordinalDetector";
 export { detectFormatSignature } from "./formatDetector";
 export { detectGeo, toGeoIso, buildGeoIsoColumn, isJoinGeoKind } from "./geoDetector";

package/dist/types/matchQuality.d.ts

@@ -0,0 +1,23 @@
+/**
+ * THE KNOB. Minimum share of a column's non-blank DISTINCT values that must resolve against
+ * the gazetteer before the column can claim a geographic role.
+ *
+ * Tune here and every classifier that opts in moves together. Higher is stricter about
+ * claiming a role at all; lower risks a column of something else claiming one. 97 leaves
+ * room for the odd typo in an otherwise clean column while staying far out of reach of a
+ * column that is not really places.
+ *
+ * Deliberately NOT applied to the looser name roles (city/state at 85%), which tolerate
+ * messier real-world text by design — raising those is a separate decision with its own
+ * consequences, and belongs to Joel, not to this constant quietly widening its reach.
+ */
+export declare const GAZETTEER_MATCH_PCT = 97;
+/**
+ * Is this ALREADY-NORMALIZED token a blank in disguise?
+ *
+ * Callers pass the output of normalizePlaceName, so genuine blanks are "" by then; this
+ * catches the ones a person typed. Excluded from the denominator, not counted as a failed
+ * match — a column of 40 countries and one "N/A" is a clean country column with a hole in
+ * it, not a 97.6%-quality one, and the difference decides whether the map draws.
+ */
+export declare function isBlankLike(normalized: string): boolean;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bicharts/shape-core",
-  "version": "0.5.7",
+  "version": "0.5.8",
   "description": "Host-agnostic data-shape profiler: ingest a dataset's rows through IValueCollection.addRow and read back the measured shape as an LLMColumnWithValue[] payload \u2014 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",