@finbheara/domain 0.5.0 → 0.7.0

package/dist/place.js

@@ -15,7 +15,7 @@ import {
   placesRelation2,
   probablySamePlace2,
   sameReading2
-} from "./chunks/place-0j6k50wf.js";
+} from "./chunks/place-h7tvdfyp.js";
 export {
   DEFAULT_PLACE_CAPACITY2 as DEFAULT_PLACE_CAPACITY,
   DEFAULT_PLACE_READER2 as DEFAULT_PLACE_READER,

package/dist/promotion.js

@@ -46,7 +46,7 @@ function samePromotedRow(was, now) {
 }
 var STOOD_FOR = /^(?:'|(?![A-Za-z])\p{L})$/u;
 function questionMarksRead(was, now) {
-  const a = [...collapse(was)];
+  const a = [...collapse(mojibakeRead(was))];
   if (!a.includes("?"))
     return a.join("");
   const b = [...collapse(now)];
@@ -67,6 +67,19 @@ function questionMarksRead(was, now) {
   }
   return out.join("");
 }
+function mojibakeRead(print) {
+  let out = print;
+  const codes = [...out].map((c) => c.codePointAt(0) ?? 0);
+  if (codes.some((c) => c >= 128) && codes.every((c) => c <= 255)) {
+    const bytes = Uint8Array.from(codes);
+    try {
+      const decoded = new TextDecoder("utf-8", { fatal: true }).decode(bytes);
+      if (decoded !== out)
+        out = decoded;
+    } catch {}
+  }
+  return out.replace(/\uFFFD/gu, "?");
+}
 function samePromotedName(was, now) {
   const read = questionMarksRead(was, now);
   if (read === null)
@@ -86,20 +99,27 @@ function samePromotedName(was, now) {
   return glued !== "" && (glued === nameFold(`${given}${surname}`) || glued === nameFold(`${surname}${given}`));
 }
 function samePromotedHeir(was, now) {
-  if (was.name !== "" && was.name === now.name)
+  if (collapse(was.name) !== "" && collapse(was.name) === collapse(now.name))
     return true;
   const school = questionMarksRead(was.school, now.school);
   if (school === null || !samePromotedSchool(school, now.school))
     return false;
+  if (collapse(was.name) === "" && collapse(now.name) !== "")
+    return true;
   return samePromotedName(was.name, now.name);
 }
+function samePromotedPair(was, now) {
+  return samePromotedRow(was, now) || samePromotedHeir(was, now);
+}
 export {
   CHAR_SYNONYM_CLASSES,
   collapse,
   foldCharSynonyms,
+  mojibakeRead,
   questionMarksRead,
   samePromotedHeir,
   samePromotedName,
+  samePromotedPair,
   samePromotedRow,
   samePromotedSchool
 };

package/dist/school.js

@@ -1,11 +1,14 @@
 import {
+  NEVER_NAME,
+  printZones,
+  opensLocation,
   place2,
   isPlaceToken2,
   countryCodeOf2,
   areAllPlaceTokens2,
   placeRelation2,
   sameReading2
-} from "./chunks/place-0j6k50wf.js";
+} from "./chunks/place-h7tvdfyp.js";
 import {
   validity
 } from "./chunks/place-q9kwsmzd.js";
@@ -870,130 +873,6 @@ function truncatedInstitution(tail) {
   return found;
 }
 
-var QUALIFIER_DELIMITER = /[,;:()[\]/–—-]/;
-var NEVER_PLACE = new Map([
-  ["de", "name particle: `De <Surname>`, and Spanish `de <Place>`"],
-  ["ni", "Irish patronymic particle: `Ní <Surname>`"],
-  ["mo", "Irish possessive: `Mo <Word>`"],
-  ["na", "Irish article: `Rince na <Word>`"],
-  ["in", "Irish preposition: `<Word> In <Word>`; and `Inc` truncated to `In`"],
-  ["co", "Company: `<Surname> & Co.`. Bracketed `(US-CO)` is still Colorado"],
-  ["id", "Irish Dance: `<Surname> School of ID`"],
-  ["sd", "School of Dance: `<Surname> SD`"]
-]);
-var NEVER_NAME = new Map([
-  ["de", "name particle: `De <Surname>`; and Spanish `de <Place>`"],
-  ["ni", "Irish patronymic particle: `Ní <Surname>`"],
-  ["ui", "Irish patronymic particle: `Uí <Surname>`"],
-  ["mo", "Irish possessive: `Mo <Word>`"],
-  ["na", "Irish article: `Rince na <Word>`"],
-  ["an", "Irish article: `<Word> an Rince`"],
-  ["in", "Irish preposition: `<Word> In <Word>`; and `Inc` truncated to `In`"],
-  ["co", "Company: `<Surname> & Co`"],
-  ["id", "Irish Dance: `ID de <Place>` keys the place, as `Irish Dancing de <Place>` does"],
-  ["sd", "School of Dance: `<Surname> SD`"],
-  ["da", "`Dance` truncated: `<Name> Academy of Irish Da` keys the name"]
-]);
-var GENERIC_TLD = "com|net|org|edu|gov|mil|int|info|biz";
-var WEB_SPAN = new RegExp([
-  "[A-Za-z][A-Za-z0-9+.-]*://\\S+",
-  "(?:^|[\\s([])www\\.\\S+",
-  "[^\\s@]+@[^\\s@]+\\.[A-Za-z]{2,}",
-  `(?<![A-Za-z0-9.@-])(?:[A-Za-z0-9][A-Za-z0-9-]*\\.)*[A-Za-z0-9][A-Za-z0-9-]+\\.(?:${GENERIC_TLD})(?![A-Za-z0-9-])`
-].join("|"), "giu");
-function printZones(raw) {
-  const qualifier = new Set;
-  const bracketed = new Set;
-  const qualifierAny = new Set;
-  const runs = [...raw.matchAll(/[A-Za-z0-9]+/g)];
-  const web = [];
-  for (const m of raw.matchAll(WEB_SPAN))
-    web.push([m.index ?? 0, (m.index ?? 0) + m[0].length]);
-  const inside = new Set;
-  const outside = new Set;
-  const trailing = new Set;
-  for (let i = runs.length - 1;i > 0; i--) {
-    const here = runs[i];
-    const prev = runs[i - 1];
-    if (here === undefined || prev === undefined)
-      break;
-    const gap = raw.slice((prev.index ?? 0) + prev[0].length, here.index ?? 0);
-    if (!QUALIFIER_DELIMITER.test(gap))
-      break;
-    if (i - 1 === 0)
-      break;
-    trailing.add(i - 1);
-  }
-  let depth = 0;
-  let cursor = 0;
-  for (let i = 0;i < runs.length; i++) {
-    const run = runs[i];
-    const text = run?.[0];
-    if (run === undefined || text === undefined)
-      continue;
-    const at = run.index ?? cursor;
-    for (const ch of raw.slice(cursor, at)) {
-      if (ch === "(" || ch === "[")
-        depth++;
-      else if ((ch === ")" || ch === "]") && depth > 0)
-        depth--;
-    }
-    cursor = at + text.length;
-    const folded = text.toLowerCase().normalize("NFD").replace(/[̀-ͯ]/g, "");
-    (web.some(([s, e]) => at >= s && at < e) ? inside : outside).add(folded);
-    const before = raw.slice(0, at).trimEnd().slice(-1);
-    const after = raw.slice(cursor).trimStart().slice(0, 1);
-    const atQualifier = i === runs.length - 1 || before !== "" && QUALIFIER_DELIMITER.test(before) || after !== "" && QUALIFIER_DELIMITER.test(after) && trailing.has(i);
-    if (atQualifier)
-      qualifierAny.add(folded);
-    if (folded.length > 2)
-      continue;
-    if (atQualifier)
-      qualifier.add(folded);
-    if (depth > 0)
-      bracketed.add(folded);
-  }
-  const webOnly = new Set([...inside].filter((f) => !outside.has(f)));
-  return { qualifier, bracketed, qualifierAny, web, webOnly, tailCode: tailCode(raw, runs) };
-}
-var GOVERNS_THE_NEXT_WORD = new Set(["de", "mo"]);
-function tailCode(raw, runs) {
-  const last = runs[runs.length - 1];
-  const prev = runs[runs.length - 2];
-  if (last === undefined || prev === undefined)
-    return null;
-  const text = last[0];
-  if (text.length > 2 || text !== text.toUpperCase() || !/^[A-Z]+$/.test(text))
-    return null;
-  const at = last.index ?? 0;
-  const folded = text.toLowerCase();
-  const gap = raw.slice((prev.index ?? 0) + prev[0].length, at);
-  const marked = /[,;:/\\\u2013\u2014]/u.test(gap) || /\s-|-\s/u.test(gap);
-  const bare = GOVERNS_THE_NEXT_WORD.has(folded) && /^\s+$/u.test(gap) && /\p{Ll}/u.test(raw);
-  if (!marked && !bare)
-    return null;
-  const only = runs.filter((r) => r[0].toLowerCase() === folded).length === 1;
-  return { folded, at, only };
-}
-function opensLocation(folded, at, index) {
-  if (index === undefined ? at.webOnly.has(folded) : inWebSpan(at, index))
-    return false;
-  if (folded.length > 2)
-    return true;
-  if (NEVER_PLACE.has(folded))
-    return at.bracketed.has(folded) || isTailCode(folded, at, index);
-  return at.qualifier.has(folded);
-}
-function isTailCode(folded, at, index) {
-  const tail = at.tailCode;
-  if (tail === null || tail.folded !== folded)
-    return false;
-  return index === undefined ? tail.only : index === tail.at;
-}
-function inWebSpan(at, index) {
-  return at.web.some(([start, end]) => index >= start && index < end);
-}
-
 var SCHOOL_FLUFF_WORDS = new Set([
   "academy",
   "irish",
@@ -1140,6 +1019,13 @@ function createSchoolNameReader(options = {}) {
   const CP1252_HIGH = "€‚ƒ„…†‡ˆ‰Š‹Œ" + "Ž‘’“”•–—˜™š›" + "œžŸ";
   const BYTE_OF_CP1252 = new Map([...CP1252_HIGH].map((ch, i) => [ch.codePointAt(0), 128 + i]));
   const MOJIBAKE = /[Â-ßà-ï][€-¿‚-›ŒŠŽ]/u;
+  const MOJIBAKE_ALL = new RegExp(MOJIBAKE.source, "gu");
+  const NON_LATIN_LETTER = /(?!\p{Script=Latin})\p{L}/u;
+  function isRepair(before, after) {
+    if (!NON_LATIN_LETTER.test(after))
+      return true;
+    return (before.match(MOJIBAKE_ALL)?.length ?? 0) >= 2;
+  }
   function undoMojibake(s) {
     const bytes = new Uint8Array(s.length);
     for (let i = 0;i < s.length; i++) {
@@ -1159,14 +1045,19 @@ function createSchoolNameReader(options = {}) {
     let s = print;
     for (let round = 0;round < 3 && MOJIBAKE.test(s); round++) {
       const undone = undoMojibake(s);
-      if (undone === undefined || undone === s)
+      if (undone === undefined || undone === s || !isRepair(s, undone))
         return s;
       s = undone;
     }
     return s;
   }
+  const SPACING_ACUTE_AFTER_VOWEL = /((?<=\p{L})[AEIOUaeiou]|[AEIUaeiu])[´ˊ](?![sS]\b)/gu;
+  const SPACING_ACUTE = /[´ˊ]/gu;
+  function readSpacingAcute(print) {
+    return print.replace(SPACING_ACUTE_AFTER_VOWEL, "$1́").replace(SPACING_ACUTE, "'");
+  }
   function renderSn1(print) {
-    return normalizeApostrophes(repairMojibake(print).normalize("NFKC")).replace(/\p{Cf}+/gu, "").replace(/\p{Cc}+/gu, " ").replace(/\s+/gu, " ").trim();
+    return normalizeApostrophes(readSpacingAcute(repairMojibake(print)).normalize("NFKC")).replace(/\p{Cf}+/gu, "").replace(/\p{Cc}+/gu, " ").replace(/\s+/gu, " ").trim();
   }
   const COMPOUND_OPENERS = new Set([
     "hyphen",
@@ -2035,8 +1926,8 @@ function isColumnCutOf(a, b) {
   if (!comparable(a) || !comparable(b))
     return false;
   const cutFrom = (cut, whole) => {
-    const c = foldPrint(cut.sn0);
-    const w = foldPrint(whole.sn0);
+    const c = foldPrint(cut.sn1);
+    const w = foldPrint(whole.sn1);
     if (c.length === 0 || c.length >= w.length || !w.startsWith(c))
       return false;
     if (w[c.length] === " ")

package/dist/types/age-title.d.ts

@@ -0,0 +1,12 @@
+export declare const TITLE_RANGE_MIN = 3;
+export declare const TITLE_RANGE_MAX = 40;
+export interface AgeTitleSpan {
+    readonly start: number;
+    readonly end: number;
+    readonly text: string;
+}
+export interface AgeTitleToken {
+    readonly token: string;
+    readonly span: AgeTitleSpan | null;
+}
+export declare function ageTitleToken(title: string): AgeTitleToken;

package/dist/types/age-types.d.ts

@@ -9,8 +9,10 @@ export interface BirthYears {
 export interface AgeLadder {
     readonly caps: readonly number[];
     readonly overs: readonly number[];
-    readonly ranges: readonly number[];
+    readonly ranges: readonly AgeSpan[];
+    readonly adult?: boolean;
 }
+export type AgeSpan = readonly [number, number];
 export interface YOBContext {
     readonly year?: number | null;
 }
@@ -21,9 +23,9 @@ export interface AgeContext {
 }
 export type AgeArm = "band" | "yob" | "open" | "empty";
 export type AgeKind = "under" | "over" | "range" | "adult" | "senior" | "yob";
-export type AgeRefusal = "not-a-band" | "backwards-range" | "senior-without-ladder" | "below-youngest-band" | "year-out-of-range";
+export type AgeRefusal = "not-a-band" | "backwards-range" | "overlapping-ranges" | "senior-without-ladder" | "below-youngest-band" | "year-out-of-range";
 export type AgeSource = "sibling-narrowed" | "printed-band" | "printed-range" | "printed-yob" | "open";
-export type AgeClause = "under-floored-on-rung" | "youngest-absorbs" | "over-ceiling-from-next-over" | "over-ceiling-refused-gap" | "and-over-no-ceiling" | "range-overlapping" | "range-inclusive" | "range-floor-on-rung" | "range-floor-widened" | "range-adult-floor";
+export type AgeClause = "under-floored-on-rung" | "youngest-absorbs" | "over-ceiling-from-next-over" | "over-ceiling-refused-gap" | "and-over-no-ceiling" | "adult-under-ceiling" | "range-overlapping" | "range-as-under" | "range-top-meets-over" | "range-adult-floor";
 export type AgeRelation = "same" | "contains" | "overlaps" | "disjoint" | "silent";
 export type AgeRelationReason = "equal" | "a-contains-b" | "b-contains-a" | "partial" | "a-older" | "a-younger" | "a-not-a-band" | "b-not-a-band";
 export interface AgeComparison {

package/dist/types/age.d.ts

@@ -1,6 +1,8 @@
-import type { AgeArm, AgeClause, AgeComparison, AgeContext, AgeInterval, AgeKind, AgeLadder, AgeMisuse, AgeRefusal, AgeRelation, AgeRelationReason, AgeSource, BirthYears, YOBContext, YoungFloorVerdict } from "./age-types.ts";
+import { type AgeTitleSpan } from "./age-title.ts";
+import type { AgeArm, AgeClause, AgeComparison, AgeContext, AgeInterval, AgeKind, AgeLadder, AgeMisuse, AgeRefusal, AgeRelation, AgeRelationReason, AgeSource, AgeSpan, BirthYears, YOBContext, YoungFloorVerdict } from "./age-types.ts";
 import { type EmptyReason, type Validated } from "./validated.ts";
-export type { AgeArm, AgeClause, AgeComparison, AgeContext, AgeInterval, AgeKind, AgeLadder, AgeMisuse, AgeRefusal, AgeRelation, AgeRelationReason, AgeSource, BirthYears, EmptyReason, Validated, YOBContext, YoungFloorVerdict, };
+export { type AgeTitleSpan, type AgeTitleToken, TITLE_RANGE_MAX, TITLE_RANGE_MIN, ageTitleToken, } from "./age-title.ts";
+export type { AgeArm, AgeClause, AgeComparison, AgeContext, AgeInterval, AgeKind, AgeLadder, AgeMisuse, AgeRefusal, AgeRelation, AgeRelationReason, AgeSource, AgeSpan, BirthYears, EmptyReason, Validated, YOBContext, YoungFloorVerdict, };
 export declare const YOUNGEST_FLOOR = 0;
 export declare const ADULT_FLOOR = 18;
 export declare const MAJORS_FLOOR_YEARS = 3;
@@ -54,9 +56,16 @@ export declare class AgeMisuseError extends Error {
 }
 declare function fromPrint(print: string, context?: AgeContext): Age;
 declare function fromYOB(from: number | null, to?: number | null, context?: YOBContext): Age;
+export interface AgeInTitle {
+    readonly age: Age;
+    readonly token: string;
+    readonly span: AgeTitleSpan | null;
+}
+declare function fromTitle(title: string, context?: AgeContext): AgeInTitle;
 export declare const Age: Readonly<{
     from: typeof fromPrint;
     fromYOB: typeof fromYOB;
+    fromTitle: typeof fromTitle;
 }>;
 export declare function birthYearsOfInterval(iv: AgeInterval | null, year: number): BirthYears;
 export declare function ageRelation(a: Age, b: Age): AgeComparison;

package/dist/types/place-types.d.ts

@@ -8,13 +8,16 @@ export interface PlaceNode {
     readonly prints: readonly string[];
     readonly codes: readonly string[];
     readonly truncations?: readonly string[];
+    readonly headings?: readonly string[];
+    readonly runPrints?: readonly string[];
+    readonly listedCutsOnly?: true;
     readonly iso2?: string;
     readonly iso3?: string;
     readonly isoNumeric?: string;
     readonly oireachtas?: string;
 }
-export type PlaceMatch = "id" | "code" | "print" | "phrase" | "iso" | "county" | "truncation";
-export type PlaceRefusal = "school" | "organization" | "placeholder" | "damaged";
+export type PlaceMatch = "id" | "code" | "print" | "phrase" | "iso" | "county" | "truncation" | "desmushed";
+export type PlaceRefusal = "school" | "organization" | "placeholder" | "damaged" | "web" | "online" | "notice";
 export type PlaceRelation = "same" | "contains" | "overlaps" | "conflict" | "silent";
 export interface PlaceContext {
     readonly region?: string;

package/dist/types/promotion.d.ts

@@ -12,5 +12,7 @@ export declare const collapse: (s: string) => string;
 export declare function samePromotedSchool(was: string, now: string): boolean;
 export declare function samePromotedRow(was: PromotedPrint, now: PromotedPrint): boolean;
 export declare function questionMarksRead(was: string, now: string): string | null;
+export declare function mojibakeRead(print: string): string;
 export declare function samePromotedName(was: string, now: string): boolean;
 export declare function samePromotedHeir(was: PromotedPrint, now: PromotedPrint): boolean;
+export declare function samePromotedPair(was: PromotedPrint, now: PromotedPrint): boolean;

package/dist/types/school-location.d.ts

@@ -12,5 +12,6 @@ export interface PrintZones {
         readonly only: boolean;
     } | null;
 }
+export declare function printsWeb(raw: string): boolean;
 export declare function printZones(raw: string): PrintZones;
 export declare function opensLocation(folded: string, at: PrintZones, index?: number): boolean;

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@finbheara/domain",
-	"version": "0.5.0",
+	"version": "0.7.0",
 	"description": "Places, CLRG regions, school names, ages and dances as they are printed: read, compared and told apart",
 	"keywords": [
 		"places",
@@ -73,6 +73,6 @@
 		"prepublishOnly": "bun scripts/build.ts && bun scripts/guard-tarball.ts"
 	},
 	"dependencies": {
-		"@finbheara/names": "0.17.0"
+		"@finbheara/names": "0.18.0"
 	}
 }