@finbheara/domain 0.11.0 → 0.13.0

package/README.md

@@ -43,7 +43,7 @@ ESM, Node ≥ 22 or Bun ≥ 1.4.2. MIT. Not affiliated with or endorsed by any d
 | `@finbheara/domain/noun-index` | `createIndex`, `indexSchema`, `personEncoder`, and the `SqlHandle`/`NounEncoder` types: a database-backed name map (below) |
 | `@finbheara/domain/dance-types` | `danceTypes()`, `DANCE_TYPES`, `rhythmOfTime`, and the shared `DanceMatch`/`DanceRelation`/`SetRhythm`/`TimeSignature` types |
 | `@finbheara/domain/promotion` | `samePromotedRow`, `samePromotedSchool`, `samePromotedHeir`, `samePromotedName`, `questionMarksRead`, `foldCharSynonyms`: is a row the same row after a promotion (below) |
-| `@finbheara/domain/person` | `seatName`, `dancerName`, `teacherName`, `roleName`, `roleSentinelOf`, the role lists and `SHARED_ROW_SENTINELS`: a person's print read in its role (below) |
+| `@finbheara/domain/person` | `seatName`, `dancerName`, `teacherName`, `roleName`, `roleSentinelOf`, `rolePrintsNoName`, `dancerPrintsNoName`, the role lists and `SHARED_ROW_SENTINELS`: a person's print read in its role (below) |
 
 The library does no I/O and installs nothing process-wide. Every reader has a default and a
 factory: build your own with options when the default is not what you want.
@@ -105,7 +105,7 @@ A `Place` is one of four arms: `resolved`, `plural` (a print naming more than on
 says they cannot be one place; `same`, `contains` and `overlaps` do not.
 
 A print that is clearly not a place is refused, with its reason: a placeholder
-(`[address withheld]`, `Other`), a URL anywhere in the print (`web`), an online event in the
+(`[address withheld]`, `TBA`), a URL anywhere in the print (`web`), an online event in the
 venue field (`online`), a change-of-venue notice (`notice`), or a replacement character where
 the source lost bytes (`damaged`). These are asked only where the store reads nothing, so a
 refusal never takes a reading away. Mojibake that can be decoded back is not refused: repair it
@@ -171,7 +171,8 @@ compareSchoolNames("Ardvarna Academy", "Ardvarna Academy (Ontario)", reader);
 ### Your own garbage
 
 The shipped refusals are generic: a card number, keyboard filler, a run with no letter in it,
-the shared placeholder list of `@finbheara/names` (`N/A`, `No School`, ...). What YOUR data
+what a school cell prints where no school was entered (`N/A`, `No School`, `-- Choose One --`,
+...). What YOUR data
 prints in a school field and is not a school, such as an entry system's own "school not
 listed" instruction, a test entry, or a note typed into the wrong field, is yours to declare.
 Pass it as `garbage`, the same idea as `phraseClassifier(lexicon, { garbageVocab })` in
@@ -482,6 +483,9 @@ seatName("Aoife Mhaolvarna").nn2;                    // "Mhaolvarna<<Aoife": nam
   and the school cell's (`SCHOOL_CELL_SENTINELS` on `/school`, asked by the school reader before
   its own register), so a sentinel a whole row prints has one definition.
 - Each row is the folded print (`foldNameKey`), what it stands for, and where it was seen.
+- **The print alone** has the same answer: `rolePrintsNoName(role, print)` (and
+  `dancerPrintsNoName(print)`) is `roleName(role, print).printsNoName()` without the parse, for
+  a walk that parses no name. Names' `isNoNamePrint` alone misses a role's own sentinel.
 
 ## Known gaps
 

package/dist/chunks/{place-tv86hmps.js → place-17e26rga.js}

@@ -1,9 +1,11 @@
+import {
+  SHARED_ROW_SENTINELS2,
+  sentinelIn2
+} from "./place-tcgfvemw.js";
 import {
   validity
 } from "./place-q9kwsmzd.js";
 
-import { isPlaceholderPrint } from "@finbheara/names/normalize";
-
 var PLACE_NODES = [
   {
     id: "world",
@@ -205,6 +207,8 @@ var PLACE_NODES = [
       "mid america region",
       "mid america region usa",
       "mid american region",
+      "middle america",
+      "middle america us",
       "reigiun lar meiricea",
       "usa mid america region"
     ],
@@ -234,7 +238,13 @@ var PLACE_NODES = [
     name: "New England Region USA",
     geo: "USA",
     admin: "IDTANA",
-    prints: ["new england", "new england region", "new england region usa", "reigiun sasana nua"],
+    prints: [
+      "new england",
+      "new england region",
+      "new england region usa",
+      "new england us",
+      "reigiun sasana nua"
+    ],
     codes: ["NERO"],
     oireachtas: "NERO"
   },
@@ -251,6 +261,7 @@ var PLACE_NODES = [
       "so usa",
       "southern region",
       "southern region usa",
+      "southern us",
       "southern usa",
       "usa southern region"
     ],
@@ -3965,7 +3976,7 @@ var PLACE_NODES = [
     name: "District of Columbia",
     geo: "USA",
     admin: "IDTANA-SRO",
-    prints: ["district of columbia"],
+    prints: ["d c", "district of columbia", "washington d c", "washington dc"],
     codes: ["DC", "DC-USA", "US-DC", "USA-DC"]
   },
   {
@@ -6478,6 +6489,15 @@ var PLACE_NODES = [
     prints: ["salt lake city"],
     codes: []
   },
+  {
+    id: "USA-VA/dulles",
+    kind: "city",
+    name: "Dulles",
+    geo: "USA-VA",
+    admin: "IDTANA-SRO",
+    prints: ["dulles", "washington dulles"],
+    codes: []
+  },
   {
     id: "USA-VA/leesburg",
     kind: "city",
@@ -6763,6 +6783,8 @@ function inWebSpan(at, index) {
 }
 
 var DEFAULT_PLACE_CAPACITY2 = 8192;
+var PLACE_CELL_SENTINELS = [...SHARED_ROW_SENTINELS2];
+var placeCellSentinelOf = sentinelIn2(PLACE_CELL_SENTINELS);
 function createPlaceReader2(options = {}) {
   const nodes = options.tree ?? PLACE_NODES;
   let index;
@@ -6804,7 +6826,7 @@ function createPlaceReader2(options = {}) {
       return "online";
     if (folded === NOTICE || folded.startsWith(`${NOTICE} `))
       return "notice";
-    if (isPlaceholderPrint(raw))
+    if (placeCellSentinelOf(raw) !== undefined)
       return "placeholder";
     return;
   }
@@ -7397,6 +7419,24 @@ function createPlaceReader2(options = {}) {
     }
     return;
   }
+  function taggedHead(core, declared) {
+    const m = /^(.*\S)\s*(?:\(([^()[\]{}]*)\)|\[([^()[\]{}]*)\])$/su.exec(core);
+    if (m === null)
+      return;
+    const head = m[1];
+    const tag = (m[2] ?? m[3] ?? "").trim();
+    if (fold(head) === "" || fold(tag) === "")
+      return;
+    const h = read(head, declared);
+    const t = read(tag, declared);
+    if (h.refusal !== null || t.refusal !== null || h.heads.length === 0 || t.heads.length === 0)
+      return;
+    const held = (n) => t.heads.some((x) => isAncestor(x, n));
+    if (h.members !== undefined)
+      return h.members.every(held) ? h : undefined;
+    const kept = h.heads.filter(held);
+    return kept.length === 0 ? undefined : { ...h, heads: kept };
+  }
   function pieced(left, country) {
     if (namedInside(left, country).length > 0)
       return [left];
@@ -7551,7 +7591,7 @@ function createPlaceReader2(options = {}) {
     }
     const first = sane[0];
     if (first === undefined)
-      return (smushed ? desmush(core, declared) : undefined) ?? NOTHING;
+      return taggedHead(core, declared) ?? (smushed ? desmush(core, declared) : undefined) ?? NOTHING;
     if (sane.length === 1)
       return first;
     const heads = [];
@@ -7847,10 +7887,89 @@ function createPlaceReader2(options = {}) {
     return made;
   }
   function placeFromId(id) {
-    const n = idx().byId.get(id.trim());
-    if (n === undefined)
+    const key = id.trim();
+    if (key.includes("+"))
+      return pluralFromId(id, key);
+    const n = idx().byId.get(key);
+    if (n !== undefined)
+      return build(id, [chainOf(n)], { heads: [n], match: "id", truncated: false, refusal: null }, []);
+    for (let at = key.lastIndexOf("/");at > 0; at = key.lastIndexOf("/", at - 1)) {
+      const above = idx().byId.get(key.slice(0, at));
+      if (above !== undefined)
+        return build(id, [chainOf(above)], { heads: [above], match: "id", truncated: true, refusal: null }, []);
+    }
+    return build(id, [], NOTHING, []);
+  }
+  function pluralFromId(id, key) {
+    const members = [];
+    for (const part of key.split("+")) {
+      const n = idx().byId.get(part);
+      if (n === undefined || members.some((m) => m.id === n.id))
+        return build(id, [], NOTHING, []);
+      members.push(n);
+    }
+    if (members.length < 2)
       return build(id, [], NOTHING, []);
-    return build(id, [chainOf(n)], { heads: [n], match: "id", truncated: false, refusal: null }, []);
+    for (let a = 0;a < members.length; a++)
+      for (let b = a + 1;b < members.length; b++)
+        if (!twoPlaces(members[a], members[b]))
+          return build(id, [], NOTHING, []);
+    members.sort((x, y) => x.id.localeCompare(y.id));
+    return build(id, chainsFor(members), {
+      heads: members,
+      match: "id",
+      truncated: false,
+      refusal: null,
+      members
+    }, []);
+  }
+  const MAX_WINDOW = 6;
+  function segmentsOf(text) {
+    return text.split(/[;:()[\]{}|]+/).map((seg) => seg.split(/[\s,\u2013\u2014]+|\s-\s|-(?=\p{Lu})/u).filter((t) => t !== "" && t !== "-")).filter((seg) => seg.length > 0);
+  }
+  function placeWindows(text, ctx) {
+    const found = [];
+    for (const tokens of segmentsOf(text)) {
+      const taken = new Array(tokens.length).fill(false);
+      const here = [];
+      for (let w = Math.min(MAX_WINDOW, tokens.length);w >= 1; w--)
+        for (let i = 0;i + w <= tokens.length; i++) {
+          if (taken.slice(i, i + w).some((x) => x))
+            continue;
+          const span = tokens.slice(i, i + w);
+          const p = place(span.join(" "), ctx);
+          if (!p.isPlace() || p.truncated)
+            continue;
+          for (let k = i;k < i + w; k++)
+            taken[k] = true;
+          here.push({ at: i, r: { p, tokens: span } });
+        }
+      here.sort((a, b) => a.at - b.at);
+      for (const [k, h] of here.entries()) {
+        const n = here[k + 1];
+        if (n !== undefined && n.at === h.at + h.r.tokens.length)
+          h.r.next = n.r;
+      }
+      found.push(...here.map((f) => f.r));
+    }
+    return found;
+  }
+  function placeParts(print) {
+    const core = unbracket(repair(print));
+    const { tokens, seps } = cut(core);
+    const pieces = [];
+    let run = tokens[0] ?? "";
+    for (let i = 0;i < seps.length; i++) {
+      const next = tokens[i + 1] ?? "";
+      if (seps[i] === "space")
+        run = `${run} ${next}`;
+      else {
+        pieces.push(run);
+        run = next;
+      }
+    }
+    pieces.push(run);
+    return pieces.filter((p) => p.trim() !== "").map((p) => place(p));
   }
   function oireachtasOfRegion(regionId) {
     if (regionId === undefined)
@@ -7945,6 +8064,8 @@ function createPlaceReader2(options = {}) {
     countryCodeOf,
     areAllPlaceTokens,
     placeFromId,
+    placeParts,
+    placeWindows,
     oireachtasOfRegion,
     regionOfOireachtas,
     placeStats,
@@ -7971,6 +8092,12 @@ function areAllPlaceTokens2(tokens, off = 0, declared = false) {
 function placeFromId2(id) {
   return DEFAULT_PLACE_READER2.placeFromId(id);
 }
+function placeWindows2(text, ctx) {
+  return DEFAULT_PLACE_READER2.placeWindows(text, ctx);
+}
+function placeParts2(print) {
+  return DEFAULT_PLACE_READER2.placeParts(print);
+}
 function oireachtasOfRegion2(regionId) {
   return DEFAULT_PLACE_READER2.oireachtasOfRegion(regionId);
 }
@@ -7996,4 +8123,4 @@ function sameReading2(a, b) {
   return DEFAULT_PLACE_READER2.sameReading(a, b);
 }
 
-export { NEVER_NAME, printZones, opensLocation, DEFAULT_PLACE_CAPACITY2, createPlaceReader2, DEFAULT_PLACE_READER2, place2, isPlaceToken2, countryCodeOf2, areAllPlaceTokens2, placeFromId2, oireachtasOfRegion2, regionOfOireachtas2, placeStats2, resetPlaceCache2, placeRelation2, placesRelation2, probablySamePlace2, sameReading2 };
+export { NEVER_NAME, printZones, opensLocation, DEFAULT_PLACE_CAPACITY2, createPlaceReader2, DEFAULT_PLACE_READER2, place2, isPlaceToken2, countryCodeOf2, areAllPlaceTokens2, placeFromId2, placeWindows2, placeParts2, oireachtasOfRegion2, regionOfOireachtas2, placeStats2, resetPlaceCache2, placeRelation2, placesRelation2, probablySamePlace2, sameReading2 };

package/dist/person.js

@@ -5,6 +5,7 @@ import {
 
 import {
   NAME_REFUSED,
+  isNoNamePrint,
   personName
 } from "@finbheara/names/normalize";
 var ADJUDICATOR_SENTINELS = [
@@ -110,6 +111,10 @@ function roleName(role, print, order) {
   const sentinel = MATCHERS[role](print);
   return sentinel === undefined ? read : new RoleRefusedName(read, role, sentinel);
 }
+function rolePrintsNoName(role, print) {
+  return MATCHERS[role](print) !== undefined || isNoNamePrint(print);
+}
+var dancerPrintsNoName = (print) => rolePrintsNoName("dancer", print);
 var seatName = (print) => roleName("adjudicator", print, "given-first");
 var dancerName = (print, order) => roleName("dancer", print, order);
 var teacherName = (print, order) => roleName("teacher", print, order);
@@ -120,7 +125,9 @@ export {
   SHARED_ROW_SENTINELS2 as SHARED_ROW_SENTINELS,
   TEACHER_SENTINELS,
   dancerName,
+  dancerPrintsNoName,
   roleName,
+  rolePrintsNoName,
   roleSentinelOf,
   seatName,
   sentinelIn2 as sentinelIn,

package/dist/place.js

@@ -7,6 +7,8 @@ import {
   countryCodeOf2,
   areAllPlaceTokens2,
   placeFromId2,
+  placeWindows2,
+  placeParts2,
   oireachtasOfRegion2,
   regionOfOireachtas2,
   placeStats2,
@@ -15,7 +17,8 @@ import {
   placesRelation2,
   probablySamePlace2,
   sameReading2
-} from "./chunks/place-tv86hmps.js";
+} from "./chunks/place-17e26rga.js";
+import"./chunks/place-tcgfvemw.js";
 export {
   DEFAULT_PLACE_CAPACITY2 as DEFAULT_PLACE_CAPACITY,
   DEFAULT_PLACE_READER2 as DEFAULT_PLACE_READER,
@@ -26,8 +29,10 @@ export {
   oireachtasOfRegion2 as oireachtasOfRegion,
   place2 as place,
   placeFromId2 as placeFromId,
+  placeParts2 as placeParts,
   placeRelation2 as placeRelation,
   placeStats2 as placeStats,
+  placeWindows2 as placeWindows,
   placesRelation2 as placesRelation,
   probablySamePlace2 as probablySamePlace,
   regionOfOireachtas2 as regionOfOireachtas,

package/dist/school.js

@@ -1,3 +1,7 @@
+import {
+  SHARED_ROW_SENTINELS2,
+  sentinelIn2
+} from "./chunks/place-tcgfvemw.js";
 import {
   NEVER_NAME,
   printZones,
@@ -8,11 +12,7 @@ import {
   areAllPlaceTokens2,
   placeRelation2,
   sameReading2
-} from "./chunks/place-tv86hmps.js";
-import {
-  SHARED_ROW_SENTINELS2,
-  sentinelIn2
-} from "./chunks/place-tcgfvemw.js";
+} from "./chunks/place-17e26rga.js";
 import {
   validity
 } from "./chunks/place-q9kwsmzd.js";
@@ -21,8 +21,6 @@ import { normalizeApostrophes as normalizeApostrophes2 } from "@finbheara/names/
 import { SURNAME_SEPARATOR, lexiconKey, titleCaseNamePart } from "@finbheara/names/normalize";
 import { personName } from "@finbheara/names/normalize";
 
-import { placeholderOf } from "@finbheara/names/normalize";
-
 var SCHOOL_CELL_OWN_SENTINELS = [
   {
     print: "no school",
@@ -35,6 +33,69 @@ var SCHOOL_CELL_OWN_SENTINELS = [
   {
     print: "independent dancer",
     authority: "`Independent Dancer`, `Independent dancer`, `Independent Dancer?`: 8 school cells"
+  },
+  {
+    print: "non dancing",
+    authority: "`Non-dancing`: 107 school cells, an entry form's option for a non-dancer"
+  },
+  {
+    print: "school not selected",
+    match: "prefix",
+    authority: "`SCHOOL NOT SELECTED - update your ... profile`: 45 school cells, an entry form's instruction"
+  },
+  {
+    print: "choose one",
+    authority: "`-- Choose One --`: 41 school cells, an entry form's unchosen option"
+  },
+  {
+    print: "no school selected",
+    match: "prefix",
+    authority: "`NO SCHOOL SELECTED - Please change your profile`: 16 school cells, an entry form's instruction"
+  },
+  {
+    print: "no school associations",
+    authority: "`No School Associations`: 13 school cells"
+  },
+  { print: "no lessons yet", authority: "`No lessons yet`: 5 school cells" },
+  { print: "non dancer", authority: "`Non-dancer`: 4 school cells" },
+  { print: "non dance competitions", authority: "`Non-Dance Competitions`: 3 school cells" },
+  {
+    print: "not listed",
+    match: "prefix",
+    authority: "`--Not Listed--`: 2 school cells, an entry form's option"
+  },
+  { print: "other", authority: "`Other`: 2 school cells, an entry form's option" },
+  { print: "not certified", authority: "`Not Certified`: 1 school cell" },
+  { print: "n e e d", authority: "`(n e e d ) ____`: 1 school cell, keyboard filler" },
+  { print: "none", authority: "`None`: 1 school cell" },
+  {
+    print: "unknown irish dancing school",
+    authority: "`Unknown Irish Dancing School`: 1 school cell"
+  },
+  { print: "none inactive dancer", authority: "`None - Inactive Dancer`: 1 school cell" },
+  { print: "not applicable", authority: "`Not applicable`: a school print, no school entered" },
+  {
+    print: "no school judge",
+    authority: "`No School - Judge`: an entry form's option for an adjudicator"
+  },
+  { print: "no dance school", authority: "`No Dance School`: an entry form's option" },
+  { print: "unknown music school", authority: "`Unknown Music School`: an entry form's option" },
+  { print: "school deleted", authority: "`---SCHOOL DELETED`: an entry form's option" },
+  { print: "select school", authority: "`--Select School--`: an entry form's unchosen option" },
+  { print: "school other", authority: "`School Other`: an entry form's option" },
+  { print: "non", authority: "`Non`: an entry form's option, the truncation of `Non-dancer`" },
+  {
+    print: "x closed select new school",
+    authority: "`X - CLOSED - Select New School`: an entry form's option for a closed school"
+  },
+  {
+    print: "x school closed please select another school",
+    authority: "`X -SCHOOL CLOSED - Please select another school`: an entry form's option"
+  },
+  { print: "default school", authority: "`Default School`: an entry form's default" },
+  {
+    print: "my school of irish dance",
+    authority: "`My School of Irish Dance`: an entry form's example text"
   }
 ];
 var SCHOOL_CELL_SENTINELS = [
@@ -282,7 +343,7 @@ function notASchoolIn(rows) {
     const text = normalizeForRefusal(print);
     if (text === "")
       return;
-    const placeholder = schoolCellSentinelOf(text) ?? placeholderOf(text);
+    const placeholder = schoolCellSentinelOf(text);
     if (placeholder !== undefined)
       return placeholderRow(placeholder);
     for (const row of rows) {

package/dist/types/person.d.ts

@@ -9,6 +9,8 @@ export declare const TEACHER_SENTINELS: readonly RowSentinel[];
 export declare const ROLE_SENTINELS: Readonly<Record<PersonRole, readonly RowSentinel[]>>;
 export declare function roleSentinelOf(role: PersonRole, print: string): RowSentinel | undefined;
 export declare function roleName(role: PersonRole, print: string, order?: ParseOrder): PersonName;
+export declare function rolePrintsNoName(role: PersonRole, print: string): boolean;
+export declare const dancerPrintsNoName: (print: string) => boolean;
 export declare const seatName: (print: string) => PersonName;
 export declare const dancerName: (print: string, order?: ParseOrder) => PersonName;
 export declare const teacherName: (print: string, order?: ParseOrder) => PersonName;

package/dist/types/place.d.ts

@@ -51,6 +51,11 @@ export interface UnreadPlace extends AbstractPlace {
     readonly arm: "unread";
 }
 export type Place = ResolvedPlace | PluralPlaces | AmbiguousPlace | UnreadPlace;
+export interface PlaceWindow {
+    readonly p: Place;
+    readonly tokens: readonly string[];
+    next?: PlaceWindow;
+}
 export type PlaceRefusals = (print: string) => PlaceRefusal | undefined;
 export interface PlaceReaderOptions {
     readonly tree?: readonly PlaceNode[];
@@ -64,6 +69,8 @@ export declare function createPlaceReader(options?: PlaceReaderOptions): {
     countryCodeOf: (token: string) => string | undefined;
     areAllPlaceTokens: (tokens: readonly string[], off?: number, declared?: boolean) => boolean;
     placeFromId: (id: string) => Place;
+    placeParts: (print: string) => Place[];
+    placeWindows: (text: string, ctx?: string | PlaceContext) => PlaceWindow[];
     oireachtasOfRegion: (regionId: string | undefined) => string | undefined;
     regionOfOireachtas: (code: string) => Place;
     placeStats: () => {
@@ -84,6 +91,8 @@ export declare function isPlaceToken(token: string, cut?: boolean, declared?: bo
 export declare function countryCodeOf(token: string): string | undefined;
 export declare function areAllPlaceTokens(tokens: readonly string[], off?: number, declared?: boolean): boolean;
 export declare function placeFromId(id: string): Place;
+export declare function placeWindows(text: string, ctx?: string | PlaceContext): PlaceWindow[];
+export declare function placeParts(print: string): Place[];
 export declare function oireachtasOfRegion(regionId: string | undefined): string | undefined;
 export declare function regionOfOireachtas(code: string): Place;
 export declare function placeStats(): {

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@finbheara/domain",
-	"version": "0.11.0",
+	"version": "0.13.0",
 	"description": "Places, CLRG regions, school names, ages, dances and a person's print in its role, as they are printed: read, compared and told apart",
 	"keywords": [
 		"places",