@mailwoman/corpus 10.0.0 → 10.2.0

package/out/fr/adapters/finess/postal-lines.js

@@ -0,0 +1,114 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * Readers for the three French postal-line shapes that the FINESS and Annuaire de l'éducation
+ * publications share: the La Poste routing line (`97100 BASSE TERRE`, `97331 CAYENNE CEDEX`), a
+ * line holding a post-office box among other text (`ZA ESPERANCE - BP 26`), and the comparison of
+ * two place names written in different forms (`ST PIERRE` against `Saint-Pierre`).
+ *
+ * Both adapters read French overseas registers, and both publishers write these lines the same
+ * way, so the readers live in one module rather than in each adapter.
+ */
+import { matchCedex } from "@mailwoman/codex/fr/cedex";
+import { foldToken } from "@mailwoman/codex/normalize";
+/**
+ * Five digits, a space, and the routing locality.
+ */
+const ACHEMINEMENT = /^(\d{5}) (.+)$/u;
+/**
+ * Split a La Poste routing line into postcode, locality and an optional trailing `CEDEX`.
+ *
+ * Returns `null` when the line does not open with a five-digit postcode,
+ * holds no locality after it, or holds `CEDEX` anywhere but at its end, because none
+ * of those is a routing line the French layout can print.
+ */
+export function parseAcheminementLine(line) {
+    const collapsed = line.replaceAll(/\s+/gu, " ").trim();
+    const match = ACHEMINEMENT.exec(collapsed);
+    if (!match)
+        return null;
+    const postcode = match[1];
+    const rest = match[2].trim();
+    const cedex = matchCedex(rest);
+    if (!cedex)
+        return { postcode, locality: rest };
+    if (cedex.end !== rest.length)
+        return null;
+    const locality = rest.slice(0, cedex.start).trim();
+    if (!locality)
+        return null;
+    return { postcode, locality, cedex: cedex.matched };
+}
+/**
+ * A post-office box designator and its number.
+ *
+ * `BP` is boîte postale, `CS` is La Poste's course spéciale and `TSA` its tri service
+ * arrivée, all three delivered to a box rather than to the premise.
+ * New Caledonia numbers its boxes with a letter prefix (`BP M2`, `BP MGA 19`, `BP KO243`),
+ * which is why up to three letters may precede the digits.
+ */
+const POST_BOX = /(?<![\p{L}\d])(?:B\.?\s?P\.?|CS|TSA)\s*(?:N°\s*)?(?:[A-Z]{1,3}\s?)?\d+(?![\p{L}\d])/iu;
+/**
+ * Separators a publisher leaves around a box once it is removed: `ZA ESPERANCE - BP 26`.
+ */
+const EDGE_SEPARATORS = /^[\s\-–—/,;:.]+|[\s\-–—/,;:.]+$/gu;
+/**
+ * Take the first post-office box out of a line.
+ *
+ * `SAVANNAH BP 90149` yields `BP 90149` and `SAVANNAH`, and `BP 381-SPRING CONCORDIA -`
+ * yields `BP 381` and `SPRING CONCORDIA`.
+ */
+export function splitPostBox(line) {
+    const collapsed = line.replaceAll(/\s+/gu, " ").trim();
+    const match = POST_BOX.exec(collapsed);
+    if (!match)
+        return { poBox: null, remainder: collapsed.replaceAll(EDGE_SEPARATORS, "") };
+    const before = collapsed.slice(0, match.index).replaceAll(EDGE_SEPARATORS, "");
+    const after = collapsed.slice(match.index + match[0].length).replaceAll(EDGE_SEPARATORS, "");
+    return {
+        poBox: match[0].replaceAll(/\s+/gu, " "),
+        remainder: [before, after].filter((part) => part !== "").join(" "),
+    };
+}
+/**
+ * The opening word of a line that locates a part of a building or a landmark rather than a place:
+ * a building, a floor, a staircase, a door, a corner, a car park, a school's grounds.
+ *
+ * Both publishers put such text on the line a lieu-dit takes (`IMMEUBLE SEMAFA`, `BAT SOUS MARQUE B
+ * - RDC`, `ANGLE DES RUES JEAN JAURES`, `ENCEINTE ECOLE DE TIAPA`), and labeling it a dependent
+ * locality would teach a building name as a place name.
+ * `Lotissement` and `Résidence` are not in the list: in the overseas departments both
+ * name the housing estate an address is in, and TRE-R35 codes both as street types.
+ */
+const PREMISE_DESCRIPTOR = /^(?:imm(?:euble)?|b[aâ]t(?:iment)?|appt?|appartement|étage|etage|rdc|rez[- ]de[- ]chauss[ée]e|esc(?:alier)?|porte|local|angle|centre commercial|ccial|ccal|enceinte|parking|entr[ée]e|lyc[ée]e|coll[èe]ge|[ée]cole|maison|rectorat|vice[- ]rectorat|mairie|face|derri[èe]re|pr[èe]s|[àa] c[ôo]t[ée])(?![\p{L}\d])/iu;
+/**
+ * Whether a line locates part of a building or a landmark rather than a place.
+ */
+export function isPremiseDescriptor(line) {
+    return PREMISE_DESCRIPTOR.test(line.trim());
+}
+/**
+ * The abbreviations La Poste's routing lines use for the hagionymic prefixes.
+ */
+const SAINT_FORMS = { st: "saint", ste: "sainte", sts: "saints", stes: "saintes" };
+/**
+ * A place name folded for comparison: case, diacritics, punctuation and the `ST` abbreviation.
+ */
+function placeKey(value) {
+    return foldToken(value)
+        .replaceAll(/[^\p{L}\d]+/gu, " ")
+        .trim()
+        .split(" ")
+        .map((word) => SAINT_FORMS[word] ?? word)
+        .join(" ");
+}
+/**
+ * Whether two place names write the same place: `ST PIERRE` and `Saint-Pierre`, `MATA'UTU` and `Mata Utu`.
+ */
+export function isSamePlace(left, right) {
+    const leftKey = placeKey(left);
+    return leftKey.length > 0 && leftKey === placeKey(right);
+}
+//# sourceMappingURL=postal-lines.js.map

package/out/fr/adapters/finess/postal-lines.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"postal-lines.js","sourceRoot":"","sources":["../../../../lib/fr/adapters/finess/postal-lines.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AAEH,OAAO,EAAE,UAAU,EAAE,MAAM,2BAA2B,CAAA;AACtD,OAAO,EAAE,SAAS,EAAE,MAAM,4BAA4B,CAAA;AActD;;GAEG;AACH,MAAM,YAAY,GAAG,iBAAiB,CAAA;AAEtC;;;;;;GAMG;AACH,MAAM,UAAU,qBAAqB,CAAC,IAAY;IACjD,MAAM,SAAS,GAAG,IAAI,CAAC,UAAU,CAAC,OAAO,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;IACtD,MAAM,KAAK,GAAG,YAAY,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;IAE1C,IAAI,CAAC,KAAK;QAAE,OAAO,IAAI,CAAA;IAEvB,MAAM,QAAQ,GAAG,KAAK,CAAC,CAAC,CAAE,CAAA;IAC1B,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,CAAA;IAC7B,MAAM,KAAK,GAAG,UAAU,CAAC,IAAI,CAAC,CAAA;IAE9B,IAAI,CAAC,KAAK;QAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAA;IAE/C,IAAI,KAAK,CAAC,GAAG,KAAK,IAAI,CAAC,MAAM;QAAE,OAAO,IAAI,CAAA;IAE1C,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,CAAA;IAElD,IAAI,CAAC,QAAQ;QAAE,OAAO,IAAI,CAAA;IAE1B,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,KAAK,EAAE,KAAK,CAAC,OAAO,EAAE,CAAA;AACpD,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,QAAQ,GAAG,uFAAuF,CAAA;AAExG;;GAEG;AACH,MAAM,eAAe,GAAG,mCAAmC,CAAA;AAgB3D;;;;;GAKG;AACH,MAAM,UAAU,YAAY,CAAC,IAAY;IACxC,MAAM,SAAS,GAAG,IAAI,CAAC,UAAU,CAAC,OAAO,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;IACtD,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;IAEtC,IAAI,CAAC,KAAK;QAAE,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,SAAS,EAAE,SAAS,CAAC,UAAU,CAAC,eAAe,EAAE,EAAE,CAAC,EAAE,CAAA;IAExF,MAAM,MAAM,GAAG,SAAS,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,UAAU,CAAC,eAAe,EAAE,EAAE,CAAC,CAAA;IAC9E,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,UAAU,CAAC,eAAe,EAAE,EAAE,CAAC,CAAA;IAE5F,OAAO;QACN,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,OAAO,EAAE,GAAG,CAAC;QACxC,SAAS,EAAE,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,KAAK,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC;KAClE,CAAA;AACF,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,kBAAkB,GACvB,qTAAqT,CAAA;AAEtT;;GAEG;AACH,MAAM,UAAU,mBAAmB,CAAC,IAAY;IAC/C,OAAO,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAA;AAC5C,CAAC;AAED;;GAEG;AACH,MAAM,WAAW,GAAqC,EAAE,EAAE,EAAE,OAAO,EAAE,GAAG,EAAE,QAAQ,EAAE,GAAG,EAAE,QAAQ,EAAE,IAAI,EAAE,SAAS,EAAE,CAAA;AAEpH;;GAEG;AACH,SAAS,QAAQ,CAAC,KAAa;IAC9B,OAAO,SAAS,CAAC,KAAK,CAAC;SACrB,UAAU,CAAC,eAAe,EAAE,GAAG,CAAC;SAChC,IAAI,EAAE;SACN,KAAK,CAAC,GAAG,CAAC;SACV,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,WAAW,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC;SACxC,IAAI,CAAC,GAAG,CAAC,CAAA;AACZ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,WAAW,CAAC,IAAY,EAAE,KAAa;IACtD,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAA;IAE9B,OAAO,OAAO,CAAC,MAAM,GAAG,CAAC,IAAI,OAAO,KAAK,QAAQ,CAAC,KAAK,CAAC,CAAA;AACzD,CAAC"}

package/out/fr/adapters/finess/voie-type.d.ts

@@ -0,0 +1,77 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * The written street-type word for a FINESS `typvoie` code, and the test for a street name whose
+ * digits belong to the name.
+ *
+ * ## Where the codes come from
+ *
+ * FINESS codes `typvoie` against the Agence du Numérique en Santé's terminology TRE-R35 `TypeVoie`,
+ * published in the Nomenclatures des Objets de Santé at
+ * `https://mos.esante.gouv.fr/NOS/TRE_R35-TypeVoie/FHIR/TRE-R35-TypeVoie` and as FHIR JSON at
+ * `https://ansforge.github.io/IG-terminologie-de-sante/ig/main/CodeSystem-TRE-R35-TypeVoie.json`.
+ * The list holds 170 codes, each one to four capital letters as the FINESS schema's "typeTypeVoie"
+ * pattern `[A-Z]{1,4}` requires. Every code the overseas rows of the 2026-05-04 extract use is in it.
+ *
+ * {@linkcode expandFinessVoieType} reads the codex's `FR_VOIE_TYPES` in reverse first, mapping each
+ * abbreviation and each canonical word back to its canonical word, so `R` reads `rue` and `AV`
+ * reads `avenue`. {@linkcode TRE_R35_VOIE_TYPES} then supplies the codes that table does not hold,
+ * transcribed from TRE-R35, and the one code where the two disagree: TRE-R35 writes `PASS` for
+ * `passe`, where `FR_VOIE_TYPES` lists `pass` as an abbreviation of `passage`. The publisher's list
+ * decides that code, so the local table is consulted before the reverse lookup.
+ */
+/**
+ * TRE-R35 codes that `FR_VOIE_TYPES` does not resolve, or resolves to another word,
+ * with the word an address writes for each.
+ *
+ * Five zone codes keep their acronym, because the acronym is the written form:
+ * an address reads `ZAC DE MOUDONG` rather than `Zone d'aménagement concerté de Moudong`,
+ * and TRE-R35's own display for `HLM` is the acronym.
+ *
+ * `LD` maps to `null`.
+ * A lieu-dit is written by its name only, so the code adds no word to the line.
+ *
+ * `VAL` keeps the code as its word.
+ * TRE-R35 displays it as `Vallée, vallon`, two words the code does not choose between,
+ * and `Val` is itself the written word of many French place names.
+ */
+export declare const TRE_R35_VOIE_TYPES: Readonly<Record<string, string | null>>;
+/**
+ * The code that TRE-R35 gives a care-of line, `Chez`.
+ * Such a line gives the person whose premises the address is.
+ */
+export declare const CARE_OF_VOIE_TYPE = "CHEZ";
+/**
+ * The street-type word for a `typvoie` code, in lower case as the codex writes it.
+ *
+ * Returns `null` for a code that adds no word (`LD`) and `undefined` for a code neither
+ * table resolves, so a caller can refuse a code it cannot read rather than print the code.
+ */
+export declare function expandFinessVoieType(code: string): string | null | undefined;
+/**
+ * A street line split into its leading type word and the name after it.
+ */
+export interface FrenchStreetSplit {
+    street_prefix: string;
+    street: string;
+}
+/**
+ * Split a written street into its leading type word and its name, or return `null`
+ * where the first word is not a street type.
+ *
+ * The BAN reader's {@linkcode decomposeFrStreet} is consulted first, then TRE-R35's words and the
+ * codex's `FR_VOIE_TYPES`, which hold the overseas types it lacks (`quartier`, `lotissement`, `ZAC`).
+ * The name may come back empty, as it does for a line reading only `Rue`.
+ */
+export declare function splitFrenchStreetType(line: string): FrenchStreetSplit | null;
+/**
+ * Whether the digits in a street name belong to the name.
+ *
+ * A name with no digit is not in question and returns `true`.
+ * A name such as `VALLÉE 3 BP 208` or `DIAKA BAT A4 PORTE 2` returns `false`, which is the
+ * case a reader refuses rather than print a box or a door number as part of a street.
+ */
+export declare function isStreetNameDigitPlaced(name: string): boolean;
+//# sourceMappingURL=voie-type.d.ts.map

package/out/fr/adapters/finess/voie-type.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"voie-type.d.ts","sourceRoot":"","sources":["../../../../lib/fr/adapters/finess/voie-type.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AAOH;;;;;;;;;;;;;;GAcG;AACH,eAAO,MAAM,kBAAkB,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC,CAsCtE,CAAA;AAED;;;GAGG;AACH,eAAO,MAAM,iBAAiB,SAAS,CAAA;AAmBvC;;;;;GAKG;AACH,wBAAgB,oBAAoB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,GAAG,SAAS,CAQ5E;AAWD;;GAEG;AACH,MAAM,WAAW,iBAAiB;IACjC,aAAa,EAAE,MAAM,CAAA;IACrB,MAAM,EAAE,MAAM,CAAA;CACd;AAED;;;;;;;GAOG;AACH,wBAAgB,qBAAqB,CAAC,IAAI,EAAE,MAAM,GAAG,iBAAiB,GAAG,IAAI,CAkB5E;AAoBD;;;;;;GAMG;AACH,wBAAgB,uBAAuB,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAM7D"}

package/out/fr/adapters/finess/voie-type.js

@@ -0,0 +1,173 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * The written street-type word for a FINESS `typvoie` code, and the test for a street name whose
+ * digits belong to the name.
+ *
+ * ## Where the codes come from
+ *
+ * FINESS codes `typvoie` against the Agence du Numérique en Santé's terminology TRE-R35 `TypeVoie`,
+ * published in the Nomenclatures des Objets de Santé at
+ * `https://mos.esante.gouv.fr/NOS/TRE_R35-TypeVoie/FHIR/TRE-R35-TypeVoie` and as FHIR JSON at
+ * `https://ansforge.github.io/IG-terminologie-de-sante/ig/main/CodeSystem-TRE-R35-TypeVoie.json`.
+ * The list holds 170 codes, each one to four capital letters as the FINESS schema's "typeTypeVoie"
+ * pattern `[A-Z]{1,4}` requires. Every code the overseas rows of the 2026-05-04 extract use is in it.
+ *
+ * {@linkcode expandFinessVoieType} reads the codex's `FR_VOIE_TYPES` in reverse first, mapping each
+ * abbreviation and each canonical word back to its canonical word, so `R` reads `rue` and `AV`
+ * reads `avenue`. {@linkcode TRE_R35_VOIE_TYPES} then supplies the codes that table does not hold,
+ * transcribed from TRE-R35, and the one code where the two disagree: TRE-R35 writes `PASS` for
+ * `passe`, where `FR_VOIE_TYPES` lists `pass` as an abbreviation of `passage`. The publisher's list
+ * decides that code, so the local table is consulted before the reverse lookup.
+ */
+import { FR_VOIE_TYPES, isFrenchStreetWord } from "@mailwoman/codex/fr/voie";
+import { foldToken } from "@mailwoman/codex/normalize";
+import { decomposeFrStreet } from "#fr/street-decompose";
+/**
+ * TRE-R35 codes that `FR_VOIE_TYPES` does not resolve, or resolves to another word,
+ * with the word an address writes for each.
+ *
+ * Five zone codes keep their acronym, because the acronym is the written form:
+ * an address reads `ZAC DE MOUDONG` rather than `Zone d'aménagement concerté de Moudong`,
+ * and TRE-R35's own display for `HLM` is the acronym.
+ *
+ * `LD` maps to `null`.
+ * A lieu-dit is written by its name only, so the code adds no word to the line.
+ *
+ * `VAL` keeps the code as its word.
+ * TRE-R35 displays it as `Vallée, vallon`, two words the code does not choose between,
+ * and `Val` is itself the written word of many French place names.
+ */
+export const TRE_R35_VOIE_TYPES = {
+    ANSE: "anse",
+    ART: "ancienne route",
+    BRG: "bourg",
+    CAR: "carrefour",
+    CCAL: "centre commercial",
+    CHEM: "cheminement",
+    CHS: "chaussée",
+    CTRE: "centre",
+    EN: "enceinte",
+    ESP: "esplanade",
+    ESPA: "espace",
+    GR: "grande rue",
+    HLM: "HLM",
+    IMM: "immeuble",
+    JARD: "jardin",
+    LD: null,
+    MTE: "montée",
+    PARC: "parc",
+    PASS: "passe",
+    PAT: "patio",
+    PKG: "parking",
+    PLAG: "plage",
+    QU: "quai",
+    QUA: "quartier",
+    RLE: "ruelle",
+    ROC: "rocade",
+    TRA: "traverse",
+    VAL: "val",
+    VGE: "village",
+    VLA: "villa",
+    VLGE: "village",
+    VOI: "voie",
+    ZA: "ZA",
+    ZAC: "ZAC",
+    ZI: "ZI",
+    ZONE: "zone",
+    ZUP: "ZUP",
+};
+/**
+ * The code that TRE-R35 gives a care-of line, `Chez`.
+ * Such a line gives the person whose premises the address is.
+ */
+export const CARE_OF_VOIE_TYPE = "CHEZ";
+/**
+ * Each folded abbreviation and canonical word in `FR_VOIE_TYPES`, mapped to its canonical word.
+ */
+const FR_VOIE_WORD_BY_TOKEN = (() => {
+    const out = new Map();
+    for (const canonical of Object.keys(FR_VOIE_TYPES)) {
+        out.set(foldToken(canonical), canonical);
+        for (const abbreviation of FR_VOIE_TYPES[canonical]) {
+            out.set(foldToken(abbreviation), canonical);
+        }
+    }
+    return out;
+})();
+/**
+ * The street-type word for a `typvoie` code, in lower case as the codex writes it.
+ *
+ * Returns `null` for a code that adds no word (`LD`) and `undefined` for a code neither
+ * table resolves, so a caller can refuse a code it cannot read rather than print the code.
+ */
+export function expandFinessVoieType(code) {
+    const trimmed = code.trim().toUpperCase();
+    if (!trimmed)
+        return null;
+    if (Object.hasOwn(TRE_R35_VOIE_TYPES, trimmed))
+        return TRE_R35_VOIE_TYPES[trimmed];
+    return FR_VOIE_WORD_BY_TOKEN.get(foldToken(trimmed));
+}
+/**
+ * TRE-R35's one-word street types, folded, for the words the BAN reader's dictionary does not hold.
+ */
+const TRE_R35_WORDS = new Set(Object.values(TRE_R35_VOIE_TYPES)
+    .filter((word) => word !== null && !word.includes(" "))
+    .map((word) => foldToken(word)));
+/**
+ * Split a written street into its leading type word and its name, or return `null`
+ * where the first word is not a street type.
+ *
+ * The BAN reader's {@linkcode decomposeFrStreet} is consulted first, then TRE-R35's words and the
+ * codex's `FR_VOIE_TYPES`, which hold the overseas types it lacks (`quartier`, `lotissement`, `ZAC`).
+ * The name may come back empty, as it does for a line reading only `Rue`.
+ */
+export function splitFrenchStreetType(line) {
+    const decomposed = decomposeFrStreet(line);
+    const [first = "", ...rest] = line.split(" ");
+    // The dictionary reads `route nationale` as one type word.
+    // That split leaves the route number by itself as the street name.
+    // Here the type is the first word, and the name keeps `nationale 2`.
+    if (decomposed.prefix && /^\d+[a-z]?$/iu.test(decomposed.street)) {
+        return { street_prefix: first, street: rest.join(" ") };
+    }
+    if (decomposed.prefix)
+        return { street_prefix: decomposed.prefix, street: decomposed.street };
+    if (TRE_R35_WORDS.has(foldToken(first).replace(/\.$/u, "")) || isFrenchStreetWord(first)) {
+        return { street_prefix: first, street: rest.join(" ") };
+    }
+    return null;
+}
+/**
+ * The French month names, folded.
+ * A street name with a date writes the month after the day.
+ */
+const MONTHS = "janvier|fevrier|mars|avril|mai|juin|juillet|aout|septembre|octobre|novembre|decembre";
+/**
+ * Street names whose digits are part of the name rather than a house number or a building.
+ *
+ * A numbered route (`NATIONALE 2`, `DEPARTEMENTAL 5`, `RN1`, `CD 11`, `route Nationale 2`)
+ * and a date (`DU 22 MAI`, `DU 8 MAI 1945`) are both names: a publisher that stores the
+ * house number in its own field leaves these digits in the street because they belong to it.
+ */
+const NUMBERED_NAMES = [
+    /^(?:(?:route|rte|chemin|che|ch) )?(?:(?:de la|du) )?(?:nationale|departementale?|communale|rn|rd|cd|n|d) ?(?:n° ?|n ?)?\d+[a-z]?$/u,
+    new RegExp(`^(?:du |des )?\\d{1,2}(?:er)? (?:${MONTHS})(?: \\d{4})?$`, "u"),
+];
+/**
+ * Whether the digits in a street name belong to the name.
+ *
+ * A name with no digit is not in question and returns `true`.
+ * A name such as `VALLÉE 3 BP 208` or `DIAKA BAT A4 PORTE 2` returns `false`, which is the
+ * case a reader refuses rather than print a box or a door number as part of a street.
+ */
+export function isStreetNameDigitPlaced(name) {
+    if (!/\d/u.test(name))
+        return true;
+    const folded = foldToken(name).replaceAll(/\s+/gu, " ").trim();
+    return NUMBERED_NAMES.some((pattern) => pattern.test(folded));
+}
+//# sourceMappingURL=voie-type.js.map

package/out/fr/adapters/finess/voie-type.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"voie-type.js","sourceRoot":"","sources":["../../../../lib/fr/adapters/finess/voie-type.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AAEH,OAAO,EAAE,aAAa,EAAuB,kBAAkB,EAAE,MAAM,0BAA0B,CAAA;AACjG,OAAO,EAAE,SAAS,EAAE,MAAM,4BAA4B,CAAA;AAEtD,OAAO,EAAE,iBAAiB,EAAE,MAAM,sBAAsB,CAAA;AAExD;;;;;;;;;;;;;;GAcG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAA4C;IAC1E,IAAI,EAAE,MAAM;IACZ,GAAG,EAAE,gBAAgB;IACrB,GAAG,EAAE,OAAO;IACZ,GAAG,EAAE,WAAW;IAChB,IAAI,EAAE,mBAAmB;IACzB,IAAI,EAAE,aAAa;IACnB,GAAG,EAAE,UAAU;IACf,IAAI,EAAE,QAAQ;IACd,EAAE,EAAE,UAAU;IACd,GAAG,EAAE,WAAW;IAChB,IAAI,EAAE,QAAQ;IACd,EAAE,EAAE,YAAY;IAChB,GAAG,EAAE,KAAK;IACV,GAAG,EAAE,UAAU;IACf,IAAI,EAAE,QAAQ;IACd,EAAE,EAAE,IAAI;IACR,GAAG,EAAE,QAAQ;IACb,IAAI,EAAE,MAAM;IACZ,IAAI,EAAE,OAAO;IACb,GAAG,EAAE,OAAO;IACZ,GAAG,EAAE,SAAS;IACd,IAAI,EAAE,OAAO;IACb,EAAE,EAAE,MAAM;IACV,GAAG,EAAE,UAAU;IACf,GAAG,EAAE,QAAQ;IACb,GAAG,EAAE,QAAQ;IACb,GAAG,EAAE,UAAU;IACf,GAAG,EAAE,KAAK;IACV,GAAG,EAAE,SAAS;IACd,GAAG,EAAE,OAAO;IACZ,IAAI,EAAE,SAAS;IACf,GAAG,EAAE,MAAM;IACX,EAAE,EAAE,IAAI;IACR,GAAG,EAAE,KAAK;IACV,EAAE,EAAE,IAAI;IACR,IAAI,EAAE,MAAM;IACZ,GAAG,EAAE,KAAK;CACV,CAAA;AAED;;;GAGG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAAG,MAAM,CAAA;AAEvC;;GAEG;AACH,MAAM,qBAAqB,GAAgC,CAAC,GAAG,EAAE;IAChE,MAAM,GAAG,GAAG,IAAI,GAAG,EAAkB,CAAA;IAErC,KAAK,MAAM,SAAS,IAAI,MAAM,CAAC,IAAI,CAAC,aAAa,CAAqB,EAAE,CAAC;QACxE,GAAG,CAAC,GAAG,CAAC,SAAS,CAAC,SAAS,CAAC,EAAE,SAAS,CAAC,CAAA;QAExC,KAAK,MAAM,YAAY,IAAI,aAAa,CAAC,SAAS,CAAC,EAAE,CAAC;YACrD,GAAG,CAAC,GAAG,CAAC,SAAS,CAAC,YAAY,CAAC,EAAE,SAAS,CAAC,CAAA;QAC5C,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC,CAAC,EAAE,CAAA;AAEJ;;;;;GAKG;AACH,MAAM,UAAU,oBAAoB,CAAC,IAAY;IAChD,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAA;IAEzC,IAAI,CAAC,OAAO;QAAE,OAAO,IAAI,CAAA;IAEzB,IAAI,MAAM,CAAC,MAAM,CAAC,kBAAkB,EAAE,OAAO,CAAC;QAAE,OAAO,kBAAkB,CAAC,OAAO,CAAC,CAAA;IAElF,OAAO,qBAAqB,CAAC,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,CAAA;AACrD,CAAC;AAED;;GAEG;AACH,MAAM,aAAa,GAAwB,IAAI,GAAG,CACjD,MAAM,CAAC,MAAM,CAAC,kBAAkB,CAAC;KAC/B,MAAM,CAAC,CAAC,IAAI,EAAkB,EAAE,CAAC,IAAI,KAAK,IAAI,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC;KACtE,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAChC,CAAA;AAUD;;;;;;;GAOG;AACH,MAAM,UAAU,qBAAqB,CAAC,IAAY;IACjD,MAAM,UAAU,GAAG,iBAAiB,CAAC,IAAI,CAAC,CAAA;IAC1C,MAAM,CAAC,KAAK,GAAG,EAAE,EAAE,GAAG,IAAI,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAA;IAE7C,2DAA2D;IAC3D,mEAAmE;IACnE,qEAAqE;IACrE,IAAI,UAAU,CAAC,MAAM,IAAI,eAAe,CAAC,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,EAAE,CAAC;QAClE,OAAO,EAAE,aAAa,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAA;IACxD,CAAC;IAED,IAAI,UAAU,CAAC,MAAM;QAAE,OAAO,EAAE,aAAa,EAAE,UAAU,CAAC,MAAM,EAAE,MAAM,EAAE,UAAU,CAAC,MAAM,EAAE,CAAA;IAE7F,IAAI,aAAa,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,IAAI,kBAAkB,CAAC,KAAK,CAAC,EAAE,CAAC;QAC1F,OAAO,EAAE,aAAa,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAA;IACxD,CAAC;IAED,OAAO,IAAI,CAAA;AACZ,CAAC;AAED;;;GAGG;AACH,MAAM,MAAM,GAAG,sFAAsF,CAAA;AAErG;;;;;;GAMG;AACH,MAAM,cAAc,GAAsB;IACzC,oIAAoI;IACpI,IAAI,MAAM,CAAC,oCAAoC,MAAM,gBAAgB,EAAE,GAAG,CAAC;CAC3E,CAAA;AAED;;;;;;GAMG;AACH,MAAM,UAAU,uBAAuB,CAAC,IAAY;IACnD,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,IAAI,CAAA;IAElC,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC,UAAU,CAAC,OAAO,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;IAE9D,OAAO,cAAc,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAA;AAC9D,CAAC"}

package/out/fr/insee-country.d.ts

@@ -0,0 +1,45 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The ISO 3166-1 jurisdiction an INSEE commune code belongs to.
+ *
+ * Every French source this repository reads keys its rows by INSEE commune code, and France's
+ * overseas departments and collectivities are separate ISO 3166-1 jurisdictions. A reader that
+ * labels every INSEE-coded row `FR` therefore mislabels the overseas ones. This module is the one
+ * home for that mapping, so the BAN adapter and the French recipes that read the same files agree.
+ */
+/**
+ * The jurisdiction each overseas INSEE department prefix belongs to.
+ *
+ * An INSEE commune code is five characters.
+ * A metropolitan code's department is its first two, including `2A` and `2B` for Corsica,
+ * and an overseas code's department is its first three.
+ *
+ * Every overseas prefix begins `97` or `98`, so three characters separate the
+ * two cases without a length test.
+ *
+ * `984` and `986` are mapped although IGN publishes both files with no header line
+ * and no rows, so a row appearing under either is labeled rather than read as metropolitan.
+ * `989` Clipperton has no ISO 3166-1 alpha-2 code assigned to it, so a row there
+ * has nowhere correct to go and this map leaves it out.
+ */
+export declare const COUNTRY_BY_INSEE_DEPARTMENT: Readonly<Record<string, string>>;
+/**
+ * Every jurisdiction an INSEE-coded French source can produce, so a caller's `--country`
+ * is checked against the set rather than against one value.
+ *
+ * `TF` and `WF` are in the set and currently yield no rows, because the set describes what
+ * the publisher's files can contain rather than what the address-source register has elected.
+ */
+export declare const INSEE_COUNTRIES: readonly string[];
+/**
+ * The country a row belongs to, read from its INSEE commune code.
+ *
+ * A code outside the overseas prefixes is metropolitan France.
+ * An empty or short code also reads `FR`, because a missing commune code does
+ * not move the row to another country.
+ */
+export declare function countryOfInseeCode(codeInsee: string): string;
+//# sourceMappingURL=insee-country.d.ts.map

package/out/fr/insee-country.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"insee-country.d.ts","sourceRoot":"","sources":["../../lib/fr/insee-country.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH;;;;;;;;;;;;;;GAcG;AACH,eAAO,MAAM,2BAA2B,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAaxE,CAAA;AAED;;;;;;GAMG;AACH,eAAO,MAAM,eAAe,EAAE,SAAS,MAAM,EAA0D,CAAA;AAEvG;;;;;;GAMG;AACH,wBAAgB,kBAAkB,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM,CAE5D"}

package/out/fr/insee-country.js

@@ -0,0 +1,60 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The ISO 3166-1 jurisdiction an INSEE commune code belongs to.
+ *
+ * Every French source this repository reads keys its rows by INSEE commune code, and France's
+ * overseas departments and collectivities are separate ISO 3166-1 jurisdictions. A reader that
+ * labels every INSEE-coded row `FR` therefore mislabels the overseas ones. This module is the one
+ * home for that mapping, so the BAN adapter and the French recipes that read the same files agree.
+ */
+/**
+ * The jurisdiction each overseas INSEE department prefix belongs to.
+ *
+ * An INSEE commune code is five characters.
+ * A metropolitan code's department is its first two, including `2A` and `2B` for Corsica,
+ * and an overseas code's department is its first three.
+ *
+ * Every overseas prefix begins `97` or `98`, so three characters separate the
+ * two cases without a length test.
+ *
+ * `984` and `986` are mapped although IGN publishes both files with no header line
+ * and no rows, so a row appearing under either is labeled rather than read as metropolitan.
+ * `989` Clipperton has no ISO 3166-1 alpha-2 code assigned to it, so a row there
+ * has nowhere correct to go and this map leaves it out.
+ */
+export const COUNTRY_BY_INSEE_DEPARTMENT = {
+    "971": "GP",
+    "972": "MQ",
+    "973": "GF",
+    "974": "RE",
+    "975": "PM",
+    "976": "YT",
+    "977": "BL",
+    "978": "MF",
+    "984": "TF",
+    "986": "WF",
+    "987": "PF",
+    "988": "NC",
+};
+/**
+ * Every jurisdiction an INSEE-coded French source can produce, so a caller's `--country`
+ * is checked against the set rather than against one value.
+ *
+ * `TF` and `WF` are in the set and currently yield no rows, because the set describes what
+ * the publisher's files can contain rather than what the address-source register has elected.
+ */
+export const INSEE_COUNTRIES = ["FR", ...Object.values(COUNTRY_BY_INSEE_DEPARTMENT)];
+/**
+ * The country a row belongs to, read from its INSEE commune code.
+ *
+ * A code outside the overseas prefixes is metropolitan France.
+ * An empty or short code also reads `FR`, because a missing commune code does
+ * not move the row to another country.
+ */
+export function countryOfInseeCode(codeInsee) {
+    return COUNTRY_BY_INSEE_DEPARTMENT[codeInsee.trim().slice(0, 3)] ?? "FR";
+}
+//# sourceMappingURL=insee-country.js.map

package/out/fr/insee-country.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"insee-country.js","sourceRoot":"","sources":["../../lib/fr/insee-country.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH;;;;;;;;;;;;;;GAcG;AACH,MAAM,CAAC,MAAM,2BAA2B,GAAqC;IAC5E,KAAK,EAAE,IAAI;IACX,KAAK,EAAE,IAAI;IACX,KAAK,EAAE,IAAI;IACX,KAAK,EAAE,IAAI;IACX,KAAK,EAAE,IAAI;IACX,KAAK,EAAE,IAAI;IACX,KAAK,EAAE,IAAI;IACX,KAAK,EAAE,IAAI;IACX,KAAK,EAAE,IAAI;IACX,KAAK,EAAE,IAAI;IACX,KAAK,EAAE,IAAI;IACX,KAAK,EAAE,IAAI;CACX,CAAA;AAED;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,eAAe,GAAsB,CAAC,IAAI,EAAE,GAAG,MAAM,CAAC,MAAM,CAAC,2BAA2B,CAAC,CAAC,CAAA;AAEvG;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CAAC,SAAiB;IACnD,OAAO,2BAA2B,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,IAAI,CAAA;AACzE,CAAC"}

package/out/fr/recipes/admin-split.d.ts

@@ -3,36 +3,13 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   `fr-admin-split` slice recipe — the FR admin-split coverage slice (night 2026-06-19,
- *   surpass-v1.5.0). Teaches the model to SPLIT the département out of the locality on
- *   bare/space/comma-delimited French place rows — the admin-deciding failure class the pre-GPU
- *   self-validation proved moves the resolved coordinate (collision communes −61%; see
- *   docs/articles/evals/experiments/2026-06-19-fr-admin-split-prevalidation.md). Ported from
- *   scripts/build-fr-admin-split-slice.mjs.
- *
- *   Failure shapes (the model currently mis-handles all of these):
- *
- *   - `Thauron, Creuse` → région dropped to null (the comma+full-name miss)
- *   - `Montredon, Lozère` → région = "ère" (the diacritic subword split, #727)
- *   - (AU analog) `CANBERRA ACT` → the space-delimited admin fuse
- *
- *   The département is the essential admin unit for FR postal geography and maps to the `region`
- *   component tag in our schema. We derive it DETERMINISTICALLY from the real postcode via codex
- *   `departementForCodePostal` (first two digits = département) — salvage-first, no re-derived
- *   table.
- *
- *   Data (`--communes`, opts.communes): REAL BAN (Base Adresse Nationale) commune+postcode+coord
- *   tuples, one per line, TAB-separated `commune <TAB> postcode <TAB> lon <TAB> lat`. Build the
- *   input TSV once from the BAN staging CSV (see the legacy script header). Anchor-ON by
- *   construction: rows carry a REAL postcode token in `raw` + a `postcode` component, so the
- *   training loader paints the anchor feature onto that span automatically. The trailing-postcode
- *   anchor REINFORCES the FR split (FR postcode is trailing, unlike German PLZ-leading — the v0.9.2
- *   scar is positional, not universal).
+ *   Generate French training rows that distinguish a département (`region`) from a commune
+ *   (`locality`). Forms include bare, comma-separated and space-separated names.
+ *   Derive the département from each real BAN postcode. Input rows are tab-separated commune, postcode, longitude and latitude.
  */
 import type { CorpusRecipe } from "#recipes/scaffold";
 /**
- * Slice recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
- * and `description` below for the surface form it generates.
+ * Recipe registered with the corpus builder.
  */
 export declare const frAdminSplitRecipe: CorpusRecipe;
 //# sourceMappingURL=admin-split.d.ts.map

package/out/fr/recipes/admin-split.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"admin-split.d.ts","sourceRoot":"","sources":["../../../lib/fr/recipes/admin-split.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8BG;AASH,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAA;AA8GrD;;;GAGG;AACH,eAAO,MAAM,kBAAkB,EAAE,YAyFhC,CAAA"}
+{"version":3,"file":"admin-split.d.ts","sourceRoot":"","sources":["../../../lib/fr/recipes/admin-split.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAUH,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAA;AAwGrD;;GAEG;AACH,eAAO,MAAM,kBAAkB,EAAE,YA+FhC,CAAA"}