@mailwoman/corpus 10.0.0 → 10.1.0

package/out/fr/adapters/annuaire-education/adapter.js

@@ -0,0 +1,475 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `fr-annuaire-education-overseas`: the schools, education offices and guidance centers of France's
+ * overseas jurisdictions, read from the education ministry's Annuaire de l'éducation.
+ *
+ * Input is JSON Lines, one record per line, as the OpenDataSoft export of the dataset
+ * `fr-en-annuaire-education` on `data.education.gouv.fr` serves it at
+ * `/api/explore/v2.1/catalog/datasets/fr-en-annuaire-education/exports/jsonl`. The export keys each
+ * record by the dataset's field names (`nom_etablissement`, `adresse_1`, `adresse_2`, `adresse_3`,
+ * `code_postal`, `nom_commune`, `code_departement`, `etat`), and {@linkcode AnnuaireRecord} declares
+ * the ones this adapter reads. `fetchAnnuaireEducationOverseas` asks the export for the overseas departments
+ * alone, and the adapter filters again, so a whole-country export reads the same.
+ *
+ * ## Which rows are read
+ *
+ * `code_departement` is the three-digit INSEE department: `971` to `976`, `977` Saint-Barthélemy,
+ * `978` Saint-Martin, `986` Wallis-et-Futuna, `987` French Polynesia and `988` New Caledonia. The
+ * jurisdiction is read through `COUNTRY_BY_INSEE_DEPARTMENT`, the table the BAN adapter uses. A
+ * metropolitan row is refused under {@linkcode AnnuaireRefusal.NotOverseas}, and a row whose `etat`
+ * is not `OUVERT` under {@linkcode AnnuaireRefusal.NotOpen}.
+ *
+ * ## Organizations only
+ *
+ * Every record is an education establishment or office. One private school among the overseas
+ * records writes its address care of a named person (`Chez …`), which places it at that person's
+ * premises, so a line opening with `Chez` refuses the row under {@linkcode AnnuaireRefusal.CareOfPerson}.
+ *
+ * ## The three address lines
+ *
+ * `adresse_1` is the street line, `adresse_2` a complement, and `adresse_3` La Poste's routing line
+ * when the publisher filled it. The publisher writes them as people typed them, so each is read by
+ * what it holds rather than by its position:
+ *
+ * - **A street.** `5 rue Maréchal Leclerc` opens with a number and a street-type word, and
+ *   `Rue Marcel Bonin` with the word alone. The word becomes `street_prefix` through
+ *   {@linkcode splitFrenchStreetType}, which reads the BAN reader's dictionary and then TRE-R35's
+ *   street types for the words it lacks (`quartier`, `lotissement`, `ZAC`). A number of `0` is the
+ *   publisher's placeholder for none. `Bourg` or `Village` written by itself refers to the
+ *   settlement's center and is read as a place.
+ * - **A district under a street.** A complement after a street line whose type word is not a
+ *   thoroughfare (`Quartier d'Orléans`, `BOIS DE NEFLES`, `Plaine Adam`) refers to the district the
+ *   street is in, and becomes `dependent_locality`.
+ * - **A place.** `MATA UTU`, `Gustavia`, `ATOLL DE TAKAROA` hold no number and no street-type word.
+ *   In these jurisdictions such a line gives the village, quarter or lieu-dit the school stands in,
+ *   so it becomes `dependent_locality`. A leading `Lieu-dit` is dropped from it.
+ * - **A building or a landmark.** `Lycée de Poindimié`, `ENCEINTE ECOLE DE TIAPA`, `PARKING MEGA`
+ *   locate the school by another site and are left out of the row, counted under
+ *   {@linkcode AnnuaireTrim.LinePremiseDescriptor}.
+ * - **A box.** `BP 4238`, `FAAA / BP 509`, `BP M2` becomes `po_box`, and the rest of the line is
+ *   read again as one of the shapes above.
+ *
+ * French Polynesia's complement line opens with the publisher's archipelago code (`IDV TAHITI`,
+ * `ISLV RAIATEA`, `TG RANGIROA`, `MARQ UA POU`, `AUSTRALES RIMATARA`), which is an administrative
+ * grouping rather than part of the address, so a remainder opening with one is left out and counted
+ * under {@linkcode AnnuaireTrim.ArchipelagoCode}.
+ *
+ * A digit the reading above cannot place refuses the row: `PK 4 APOOITI UTUROA RAIATEA` is a
+ * kilometer point, `TSOUNDZOU 2` and `ZUP 1` are numbered quarters this reader cannot tell from a
+ * building number. A complement that writes the same place as the street line or the locality is
+ * dropped, and a row left with two different place lines is refused under
+ * {@linkcode AnnuaireRefusal.PlaceLinesConflict} rather than guessing which one is the lieu-dit.
+ *
+ * The postcode, locality and `cedex` come from `adresse_3` where it parses as a routing line, and
+ * from `code_postal` and `nom_commune` otherwise. The codex layout of every jurisdiction here except
+ * French Polynesia prints `cedex`, so a Polynesian `CEDEX` row is refused as unrenderable.
+ *
+ * ## Identity and license
+ *
+ * The row id is content-addressed over the aligned components, and a second record rendering the
+ * same row is refused under {@linkcode AnnuaireRefusal.Duplicate}.
+ *
+ * The dataset's catalog record states `license` `Licence Ouverte v2.0 (Etalab)` with `license_url`
+ * `http://www.etalab.gouv.fr/wp-content/uploads/2017/04/ETALAB-Licence-Ouverte-v2.0.pdf`, and gives
+ * the publisher as `DNE - Ministère de l'Education Nationale`. That license states « Le « Réutilisateur
+ * » est libre de réutiliser l'« Information » : […] de l'adapter, la modifier, l'extraire et la
+ * transformer, pour créer des « Informations dérivées », des produits ou des services », « Sous
+ * réserve de : mentionner la paternité de l'« Information » : sa source (au moins le nom du «
+ * Concédant ») et la date de dernière mise à jour de l'« Information » réutilisée. » Every row
+ * records the license, and {@linkcode FR_ANNUAIRE_EDUCATION_ATTRIBUTION} is the credit, which the
+ * model card completes with the update date the fetch manifest records.
+ *
+ * The adapter honors `opts.limit`, `opts.signal` and `opts.country`, and counts every refusal
+ * under its reason in `opts.dropped`.
+ */
+import { formatAddressRow } from "@mailwoman/codex/address/format";
+import { foldToken } from "@mailwoman/codex/normalize";
+import { JSONSpliterator } from "spliterator";
+import { UnsupportedCountryError } from "#adapters/errors";
+import { stableSourceID } from "#adapters/source-id";
+import { isPremiseDescriptor, isSamePlace, parseAcheminementLine, splitPostBox } from "#fr/adapters/finess/postal-lines";
+import { isStreetNameDigitPlaced, splitFrenchStreetType } from "#fr/adapters/finess/voie-type";
+import { COUNTRY_BY_INSEE_DEPARTMENT } from "#fr/insee-country";
+import { SourceRegister } from "#registers";
+import { AddressRole, countDropped, SurfaceOrigin, } from "#types";
+/**
+ * Registry id for this adapter.
+ *
+ * Stamped into every row it emits, so a corpus record can be traced back to the dataset it came from.
+ */
+export const FR_ANNUAIRE_EDUCATION_ADAPTER_ID = "fr-annuaire-education-overseas";
+/**
+ * The license the dataset's catalog record states.
+ */
+export const FR_ANNUAIRE_EDUCATION_LICENSE = "Licence Ouverte 2.0";
+/**
+ * The credit Licence Ouverte 2.0 requires: the licensor's name, beside the date of the data's last update.
+ */
+export const FR_ANNUAIRE_EDUCATION_ATTRIBUTION = "Annuaire de l'éducation, DNE - Ministère de l'Education Nationale, via data.education.gouv.fr (fr-en-annuaire-education)";
+/**
+ * The overseas INSEE departments this adapter reads: every prefix in `COUNTRY_BY_INSEE_DEPARTMENT`
+ * except `984`, the French Southern Territories, which has no schools.
+ */
+export const FR_ANNUAIRE_EDUCATION_DEPARTEMENTS = [
+    "971",
+    "972",
+    "973",
+    "974",
+    "975",
+    "976",
+    "977",
+    "978",
+    "986",
+    "987",
+    "988",
+];
+/**
+ * Every jurisdiction this adapter can emit.
+ */
+export const FR_ANNUAIRE_EDUCATION_COUNTRIES = FR_ANNUAIRE_EDUCATION_DEPARTEMENTS.map((code) => COUNTRY_BY_INSEE_DEPARTMENT[code]);
+/**
+ * Why a record did not become a row.
+ */
+export const AnnuaireRefusal = {
+    /**
+     * The record's department is metropolitan or not one this adapter reads.
+     */
+    NotOverseas: "row:not-overseas",
+    /**
+     * The establishment's `etat` is not `OUVERT`.
+     */
+    NotOpen: "row:not-open",
+    /**
+     * An address line is written care of a person.
+     */
+    CareOfPerson: "row:care-of-person",
+    /**
+     * Neither `adresse_3` nor `code_postal` gives a five-digit postcode.
+     */
+    PostcodeAbsent: "row:postcode-absent",
+    /**
+     * Neither `adresse_3` nor `nom_commune` gives a locality.
+     */
+    LocalityAbsent: "row:locality-absent",
+    /**
+     * A line holds a digit the reading cannot place as a house number, a box or a numbered street name.
+     */
+    DigitUnplaced: "row:digit-unplaced",
+    /**
+     * A street-type word with no name after it, as `0 route` is.
+     */
+    StreetNameAbsent: "row:street-name-absent",
+    /**
+     * The two lines give two different places, and the record does not mark either one as the lieu-dit.
+     */
+    PlaceLinesConflict: "row:place-lines-conflict",
+    /**
+     * The record gives a locality and a venue without a street, a place or a box.
+     */
+    DeliveryLineAbsent: "row:delivery-line-absent",
+    /**
+     * The jurisdiction's layout does not print every component the row carries.
+     */
+    Unrenderable: "row:unrenderable",
+    /**
+     * Another record already produced this exact row.
+     */
+    Duplicate: "row:duplicate",
+};
+/**
+ * Why an admitted row left out a line the record holds.
+ */
+export const AnnuaireTrim = {
+    /**
+     * A line locates a building or a landmark (`Lycée de Poindimié`) rather than a street or a place.
+     */
+    LinePremiseDescriptor: "component:line:premise-descriptor",
+    /**
+     * A complement opens with French Polynesia's archipelago code, an administrative grouping.
+     */
+    ArchipelagoCode: "component:dependent_locality:archipelago-code",
+};
+/**
+ * The archipelago codes French Polynesia's complement lines open with: Îles du Vent,
+ * Îles sous le Vent, Tuamotu-Gambier, Marquises and Australes.
+ */
+const ARCHIPELAGO_CODE = /^(?:idv|islv|tg|marq|australes)(?![\p{L}\d])/iu;
+/**
+ * A leading house number, with a repetition index or letter suffix: `5`, `12 bis`, `5A`.
+ */
+const LEADING_NUMBER = /^(\d+)(?:\s*(bis|ter|quater)(?![\p{L}])|\s?([A-Za-z])(?![\p{L}'’]))?[\s,]+(.+)$/iu;
+/**
+ * A leading `Lieu-dit` marker, which a place line drops.
+ */
+const LIEU_DIT_MARKER = /^lieu[- ]dit\s+/iu;
+/**
+ * A line opening with `Chez`, the care-of form.
+ */
+const CARE_OF_LINE = /^chez(?![\p{L}\d])/iu;
+/**
+ * Street-type words that name a way people travel along, as opposed to a district, an estate, a zone
+ * or a landform (`quartier`, `ZAC`, `bois`, `plaine`) that the libpostal dictionary also lists.
+ */
+const THOROUGHFARE_WORDS = new Set([
+    "rue",
+    "r",
+    "avenue",
+    "av",
+    "boulevard",
+    "bd",
+    "chemin",
+    "ch",
+    "che",
+    "route",
+    "rte",
+    "route nationale",
+    "allee",
+    "impasse",
+    "imp",
+    "place",
+    "pl",
+    "voie",
+    "sentier",
+    "ruelle",
+    "passage",
+    "quai",
+    "cours",
+    "square",
+    "rocade",
+    "traverse",
+    "montee",
+    "promenade",
+    "esplanade",
+    "rond-point",
+]);
+/**
+ * Street-type words that, written alone, name a settlement's center: `Bourg`, `Village`.
+ */
+const LONE_PLACE_WORDS = new Set(["bourg", "village", "le bourg"]);
+function isThoroughfareWord(word) {
+    return THOROUGHFARE_WORDS.has(foldToken(word).replace(/\.$/u, ""));
+}
+function field(value) {
+    return (value ?? "").replaceAll(/\s+/gu, " ").trim();
+}
+/**
+ * Read one address line, after any post-office box has been taken out of it.
+ */
+export function readAnnuaireLine(line) {
+    const text = line.replace(LIEU_DIT_MARKER, "").trim();
+    if (ARCHIPELAGO_CODE.test(text))
+        return { kind: "archipelago" };
+    if (isPremiseDescriptor(text))
+        return { kind: "premise" };
+    const numbered = LEADING_NUMBER.exec(text);
+    const body = numbered ? numbered[4].trim() : text;
+    // The number as written, suffix included (`22 bis`, `5A`), and none for the publisher's `0`.
+    const house = numbered && !/^0+$/u.test(numbered[1])
+        ? text
+            .slice(0, text.length - numbered[4].length)
+            .replace(/[\s,]+$/u, "")
+            .replace(/^0+(?=\d)/u, "")
+        : "";
+    const split = splitFrenchStreetType(body);
+    if (split) {
+        if (!split.street) {
+            // `Bourg` and `Village` alone name the settlement's center, which is a place.
+            if (!house && split.street_prefix && LONE_PLACE_WORDS.has(foldToken(split.street_prefix))) {
+                return { kind: "place", place: split.street_prefix };
+            }
+            return { kind: "refused", reason: AnnuaireRefusal.StreetNameAbsent };
+        }
+        if (!isStreetNameDigitPlaced(split.street) && !isStreetNameDigitPlaced(body)) {
+            return { kind: "refused", reason: AnnuaireRefusal.DigitUnplaced };
+        }
+        return { kind: "street", ...split, ...(house ? { house_number: house } : {}) };
+    }
+    // A number before a word that is not a street type still opens a street line.
+    if (numbered) {
+        if (/\d/u.test(body))
+            return { kind: "refused", reason: AnnuaireRefusal.DigitUnplaced };
+        return { kind: "street", street: body, ...(house ? { house_number: house } : {}) };
+    }
+    if (/\d/u.test(text))
+        return { kind: "refused", reason: AnnuaireRefusal.DigitUnplaced };
+    return { kind: "place", place: text };
+}
+/**
+ * Read one export record into a row, or report why it yielded none.
+ *
+ * Exported because the refusal reasons are what this adapter is measured by.
+ */
+export function readAnnuaireRecord(record) {
+    const departement = field(record.code_departement);
+    const country = FR_ANNUAIRE_EDUCATION_DEPARTEMENTS.includes(departement)
+        ? COUNTRY_BY_INSEE_DEPARTMENT[departement]
+        : undefined;
+    if (!country)
+        return { refused: AnnuaireRefusal.NotOverseas };
+    if (field(record.etat).toUpperCase() !== "OUVERT")
+        return { refused: AnnuaireRefusal.NotOpen };
+    const line1 = field(record.adresse_1);
+    const line2 = field(record.adresse_2);
+    if (CARE_OF_LINE.test(line1) || CARE_OF_LINE.test(line2))
+        return { refused: AnnuaireRefusal.CareOfPerson };
+    const routing = parseAcheminementLine(field(record.adresse_3));
+    const postcode = routing?.postcode ?? field(record.code_postal);
+    const locality = routing?.locality ?? field(record.nom_commune);
+    if (!/^\d{5}$/u.test(postcode))
+        return { refused: AnnuaireRefusal.PostcodeAbsent };
+    if (!locality)
+        return { refused: AnnuaireRefusal.LocalityAbsent };
+    const components = {};
+    const trimmed = [];
+    const places = [];
+    const venue = field(record.nom_etablissement);
+    if (venue) {
+        components.venue = venue;
+    }
+    for (const line of [line1, line2]) {
+        if (!line)
+            continue;
+        const { poBox, remainder } = splitPostBox(line);
+        if (poBox && !components.po_box) {
+            components.po_box = poBox;
+        }
+        if (!remainder)
+            continue;
+        const reading = readAnnuaireLine(remainder);
+        switch (reading.kind) {
+            case "refused":
+                return { refused: reading.reason };
+            case "premise":
+                trimmed.push(AnnuaireTrim.LinePremiseDescriptor);
+                break;
+            case "archipelago":
+                trimmed.push(AnnuaireTrim.ArchipelagoCode);
+                break;
+            case "street":
+                if (components.street) {
+                    const written = `${reading.street_prefix ?? ""} ${reading.street}`.trim();
+                    // A complement under a street line that is not itself a thoroughfare gives the district the
+                    // street is in: `Quartier d'Orléans`, `ZAC de Moudong`, `BOIS DE NEFLES`, `Plaine Adam`.
+                    if (!reading.house_number && !isThoroughfareWord(reading.street_prefix ?? "")) {
+                        places.push(written);
+                        break;
+                    }
+                    // The same holds the other way round: a district line followed by a thoroughfare.
+                    if (!components.house_number && !isThoroughfareWord(components.street_prefix ?? "")) {
+                        places.push(`${components.street_prefix ?? ""} ${components.street}`.trim());
+                        delete components.street_prefix;
+                        components.street = reading.street;
+                        if (reading.street_prefix) {
+                            components.street_prefix = reading.street_prefix;
+                        }
+                        if (reading.house_number) {
+                            components.house_number = reading.house_number;
+                        }
+                        break;
+                    }
+                    if (!isSamePlace(written, `${components.street_prefix ?? ""} ${components.street}`)) {
+                        return { refused: AnnuaireRefusal.PlaceLinesConflict };
+                    }
+                    break;
+                }
+                if (reading.house_number) {
+                    components.house_number = reading.house_number;
+                }
+                if (reading.street_prefix) {
+                    components.street_prefix = reading.street_prefix;
+                }
+                components.street = reading.street;
+                break;
+            case "place":
+                if (isSamePlace(reading.place, locality) || isSamePlace(reading.place, field(record.nom_commune)))
+                    break;
+                if (places.some((place) => isSamePlace(place, reading.place)))
+                    break;
+                places.push(reading.place);
+                break;
+        }
+    }
+    // A complement that repeats the street's name (`1 rue aloe vera` over `ALOE VERA`) adds no place.
+    const distinctPlaces = places.filter((place) => !components.street ||
+        (!isSamePlace(place, components.street) &&
+            !isSamePlace(place, `${components.street_prefix ?? ""} ${components.street}`)));
+    if (distinctPlaces.length > 1)
+        return { refused: AnnuaireRefusal.PlaceLinesConflict };
+    if (distinctPlaces[0]) {
+        components.dependent_locality = distinctPlaces[0];
+    }
+    if (!components.street && !components.po_box && !components.dependent_locality) {
+        return { refused: AnnuaireRefusal.DeliveryLineAbsent };
+    }
+    components.postcode = postcode;
+    components.locality = locality;
+    if (routing?.cedex) {
+        components.cedex = routing.cedex;
+    }
+    const rendered = formatAddressRow(components, country, { singleLine: true });
+    if (!rendered || rendered.unplaced.length)
+        return { refused: AnnuaireRefusal.Unrenderable };
+    return {
+        admitted: {
+            raw: rendered.raw,
+            components: rendered.components,
+            country,
+            locale: `fr-${country}`,
+            source: FR_ANNUAIRE_EDUCATION_ADAPTER_ID,
+            source_id: stableSourceID(FR_ANNUAIRE_EDUCATION_ADAPTER_ID, rendered.components),
+            corpus_version: "",
+            license: FR_ANNUAIRE_EDUCATION_LICENSE,
+        },
+        trimmed,
+    };
+}
+export function createAnnuaireEducationAdapter() {
+    return {
+        id: FR_ANNUAIRE_EDUCATION_ADAPTER_ID,
+        defaultLicense: FR_ANNUAIRE_EDUCATION_LICENSE,
+        addressRole: AddressRole.Facility,
+        register: SourceRegister.AnnuaireEducation,
+        surface: SurfaceOrigin.Rendered,
+        description: "Annuaire de l'éducation: the addresses of schools and education offices in France's overseas jurisdictions.",
+        async *rows(opts) {
+            if (opts.country && !FR_ANNUAIRE_EDUCATION_COUNTRIES.includes(opts.country)) {
+                throw new UnsupportedCountryError(FR_ANNUAIRE_EDUCATION_ADAPTER_ID, FR_ANNUAIRE_EDUCATION_COUNTRIES, opts.country);
+            }
+            const seen = new Set();
+            let emitted = 0;
+            for await (const record of JSONSpliterator.fromAsync(opts.inputPath)) {
+                if (opts.signal?.aborted)
+                    break;
+                if (opts.limit !== undefined && emitted >= opts.limit)
+                    break;
+                const reading = readAnnuaireRecord(record);
+                if (!("admitted" in reading)) {
+                    countDropped(opts, reading.refused);
+                    continue;
+                }
+                // A caller naming one jurisdiction reads that jurisdiction's rows,
+                // and the others are not refusals of the source.
+                if (opts.country && reading.admitted.country !== opts.country)
+                    continue;
+                if (seen.has(reading.admitted.source_id)) {
+                    countDropped(opts, AnnuaireRefusal.Duplicate);
+                    continue;
+                }
+                for (const reason of reading.trimmed) {
+                    countDropped(opts, reason);
+                }
+                seen.add(reading.admitted.source_id);
+                yield reading.admitted;
+                emitted++;
+            }
+        },
+    };
+}
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export const annuaireEducationAdapter = createAnnuaireEducationAdapter();
+//# sourceMappingURL=adapter.js.map

package/out/fr/adapters/annuaire-education/adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../lib/fr/adapters/annuaire-education/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAsFG;AAEH,OAAO,EAAE,gBAAgB,EAAE,MAAM,iCAAiC,CAAA;AAClE,OAAO,EAAE,SAAS,EAAE,MAAM,4BAA4B,CAAA;AACtD,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAE7C,OAAO,EAAE,uBAAuB,EAAE,MAAM,kBAAkB,CAAA;AAC1D,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAA;AACpD,OAAO,EAAE,mBAAmB,EAAE,WAAW,EAAE,qBAAqB,EAAE,YAAY,EAAE,MAAM,kCAAkC,CAAA;AACxH,OAAO,EAAE,uBAAuB,EAAE,qBAAqB,EAAE,MAAM,+BAA+B,CAAA;AAC9F,OAAO,EAAE,2BAA2B,EAAE,MAAM,mBAAmB,CAAA;AAC/D,OAAO,EAAE,cAAc,EAAE,MAAM,YAAY,CAAA;AAC3C,OAAO,EACN,WAAW,EAGX,YAAY,EAEZ,aAAa,GACb,MAAM,QAAQ,CAAA;AAEf;;;;GAIG;AACH,MAAM,CAAC,MAAM,gCAAgC,GAAG,gCAAgC,CAAA;AAEhF;;GAEG;AACH,MAAM,CAAC,MAAM,6BAA6B,GAAG,qBAAqB,CAAA;AAElE;;GAEG;AACH,MAAM,CAAC,MAAM,iCAAiC,GAC7C,0HAA0H,CAAA;AAE3H;;;GAGG;AACH,MAAM,CAAC,MAAM,kCAAkC,GAAsB;IACpE,KAAK;IACL,KAAK;IACL,KAAK;IACL,KAAK;IACL,KAAK;IACL,KAAK;IACL,KAAK;IACL,KAAK;IACL,KAAK;IACL,KAAK;IACL,KAAK;CACL,CAAA;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,+BAA+B,GAAsB,kCAAkC,CAAC,GAAG,CACvG,CAAC,IAAI,EAAE,EAAE,CAAC,2BAA2B,CAAC,IAAI,CAAE,CAC5C,CAAA;AAmBD;;GAEG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG;IAC9B;;OAEG;IACH,WAAW,EAAE,kBAAkB;IAC/B;;OAEG;IACH,OAAO,EAAE,cAAc;IACvB;;OAEG;IACH,YAAY,EAAE,oBAAoB;IAClC;;OAEG;IACH,cAAc,EAAE,qBAAqB;IACrC;;OAEG;IACH,cAAc,EAAE,qBAAqB;IACrC;;OAEG;IACH,aAAa,EAAE,oBAAoB;IACnC;;OAEG;IACH,gBAAgB,EAAE,wBAAwB;IAC1C;;OAEG;IACH,kBAAkB,EAAE,0BAA0B;IAC9C;;OAEG;IACH,kBAAkB,EAAE,0BAA0B;IAC9C;;OAEG;IACH,YAAY,EAAE,kBAAkB;IAChC;;OAEG;IACH,SAAS,EAAE,eAAe;CACjB,CAAA;AAIV;;GAEG;AACH,MAAM,CAAC,MAAM,YAAY,GAAG;IAC3B;;OAEG;IACH,qBAAqB,EAAE,mCAAmC;IAC1D;;OAEG;IACH,eAAe,EAAE,+CAA+C;CACvD,CAAA;AAWV;;;GAGG;AACH,MAAM,gBAAgB,GAAG,gDAAgD,CAAA;AAEzE;;GAEG;AACH,MAAM,cAAc,GAAG,mFAAmF,CAAA;AAE1G;;GAEG;AACH,MAAM,eAAe,GAAG,mBAAmB,CAAA;AAE3C;;GAEG;AACH,MAAM,YAAY,GAAG,sBAAsB,CAAA;AAE3C;;;GAGG;AACH,MAAM,kBAAkB,GAAwB,IAAI,GAAG,CAAC;IACvD,KAAK;IACL,GAAG;IACH,QAAQ;IACR,IAAI;IACJ,WAAW;IACX,IAAI;IACJ,QAAQ;IACR,IAAI;IACJ,KAAK;IACL,OAAO;IACP,KAAK;IACL,iBAAiB;IACjB,OAAO;IACP,SAAS;IACT,KAAK;IACL,OAAO;IACP,IAAI;IACJ,MAAM;IACN,SAAS;IACT,QAAQ;IACR,SAAS;IACT,MAAM;IACN,OAAO;IACP,QAAQ;IACR,QAAQ;IACR,UAAU;IACV,QAAQ;IACR,WAAW;IACX,WAAW;IACX,YAAY;CACZ,CAAC,CAAA;AAEF;;GAEG;AACH,MAAM,gBAAgB,GAAwB,IAAI,GAAG,CAAC,CAAC,OAAO,EAAE,SAAS,EAAE,UAAU,CAAC,CAAC,CAAA;AAEvF,SAAS,kBAAkB,CAAC,IAAY;IACvC,OAAO,kBAAkB,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,CAAA;AACnE,CAAC;AAED,SAAS,KAAK,CAAC,KAAgC;IAC9C,OAAO,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC,UAAU,CAAC,OAAO,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;AACrD,CAAC;AAYD;;GAEG;AACH,MAAM,UAAU,gBAAgB,CAAC,IAAY;IAC5C,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,eAAe,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;IAErD,IAAI,gBAAgB,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,aAAa,EAAE,CAAA;IAE/D,IAAI,mBAAmB,CAAC,IAAI,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,CAAA;IAEzD,MAAM,QAAQ,GAAG,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IAC1C,MAAM,IAAI,GAAG,QAAQ,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,IAAI,CAAA;IAElD,6FAA6F;IAC7F,MAAM,KAAK,GACV,QAAQ,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAE,CAAC;QACtC,CAAC,CAAC,IAAI;aACH,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,QAAQ,CAAC,CAAC,CAAE,CAAC,MAAM,CAAC;aAC3C,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC;aACvB,OAAO,CAAC,YAAY,EAAE,EAAE,CAAC;QAC5B,CAAC,CAAC,EAAE,CAAA;IAEN,MAAM,KAAK,GAAG,qBAAqB,CAAC,IAAI,CAAC,CAAA;IAEzC,IAAI,KAAK,EAAE,CAAC;QACX,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC;YACnB,8EAA8E;YAC9E,IAAI,CAAC,KAAK,IAAI,KAAK,CAAC,aAAa,IAAI,gBAAgB,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,CAAC,aAAa,CAAC,CAAC,EAAE,CAAC;gBAC3F,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,CAAC,aAAa,EAAE,CAAA;YACrD,CAAC;YAED,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,MAAM,EAAE,eAAe,CAAC,gBAAgB,EAAE,CAAA;QACrE,CAAC;QAED,IAAI,CAAC,uBAAuB,CAAC,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC,uBAAuB,CAAC,IAAI,CAAC,EAAE,CAAC;YAC9E,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,MAAM,EAAE,eAAe,CAAC,aAAa,EAAE,CAAA;QAClE,CAAC;QAED,OAAO,EAAE,IAAI,EAAE,QAAQ,EAAE,GAAG,KAAK,EAAE,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAA;IAC/E,CAAC;IAED,8EAA8E;IAC9E,IAAI,QAAQ,EAAE,CAAC;QACd,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;YAAE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,MAAM,EAAE,eAAe,CAAC,aAAa,EAAE,CAAA;QAEvF,OAAO,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAA;IACnF,CAAC;IAED,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,MAAM,EAAE,eAAe,CAAC,aAAa,EAAE,CAAA;IAEvF,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,CAAA;AACtC,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,kBAAkB,CAAC,MAAsB;IACxD,MAAM,WAAW,GAAG,KAAK,CAAC,MAAM,CAAC,gBAAgB,CAAC,CAAA;IAElD,MAAM,OAAO,GAAG,kCAAkC,CAAC,QAAQ,CAAC,WAAW,CAAC;QACvE,CAAC,CAAC,2BAA2B,CAAC,WAAW,CAAC;QAC1C,CAAC,CAAC,SAAS,CAAA;IAEZ,IAAI,CAAC,OAAO;QAAE,OAAO,EAAE,OAAO,EAAE,eAAe,CAAC,WAAW,EAAE,CAAA;IAE7D,IAAI,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,WAAW,EAAE,KAAK,QAAQ;QAAE,OAAO,EAAE,OAAO,EAAE,eAAe,CAAC,OAAO,EAAE,CAAA;IAE9F,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC,SAAS,CAAC,CAAA;IACrC,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC,SAAS,CAAC,CAAA;IAErC,IAAI,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,OAAO,EAAE,OAAO,EAAE,eAAe,CAAC,YAAY,EAAE,CAAA;IAE1G,MAAM,OAAO,GAAG,qBAAqB,CAAC,KAAK,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,CAAA;IAC9D,MAAM,QAAQ,GAAG,OAAO,EAAE,QAAQ,IAAI,KAAK,CAAC,MAAM,CAAC,WAAW,CAAC,CAAA;IAC/D,MAAM,QAAQ,GAAG,OAAO,EAAE,QAAQ,IAAI,KAAK,CAAC,MAAM,CAAC,WAAW,CAAC,CAAA;IAE/D,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,QAAQ,CAAC;QAAE,OAAO,EAAE,OAAO,EAAE,eAAe,CAAC,cAAc,EAAE,CAAA;IAElF,IAAI,CAAC,QAAQ;QAAE,OAAO,EAAE,OAAO,EAAE,eAAe,CAAC,cAAc,EAAE,CAAA;IAEjE,MAAM,UAAU,GAA+B,EAAE,CAAA;IACjD,MAAM,OAAO,GAAmB,EAAE,CAAA;IAClC,MAAM,MAAM,GAAa,EAAE,CAAA;IAC3B,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC,iBAAiB,CAAC,CAAA;IAE7C,IAAI,KAAK,EAAE,CAAC;QACX,UAAU,CAAC,KAAK,GAAG,KAAK,CAAA;IACzB,CAAC;IAED,KAAK,MAAM,IAAI,IAAI,CAAC,KAAK,EAAE,KAAK,CAAC,EAAE,CAAC;QACnC,IAAI,CAAC,IAAI;YAAE,SAAQ;QAEnB,MAAM,EAAE,KAAK,EAAE,SAAS,EAAE,GAAG,YAAY,CAAC,IAAI,CAAC,CAAA;QAE/C,IAAI,KAAK,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE,CAAC;YACjC,UAAU,CAAC,MAAM,GAAG,KAAK,CAAA;QAC1B,CAAC;QAED,IAAI,CAAC,SAAS;YAAE,SAAQ;QAExB,MAAM,OAAO,GAAG,gBAAgB,CAAC,SAAS,CAAC,CAAA;QAE3C,QAAQ,OAAO,CAAC,IAAI,EAAE,CAAC;YACtB,KAAK,SAAS;gBACb,OAAO,EAAE,OAAO,EAAE,OAAO,CAAC,MAAM,EAAE,CAAA;YACnC,KAAK,SAAS;gBACb,OAAO,CAAC,IAAI,CAAC,YAAY,CAAC,qBAAqB,CAAC,CAAA;gBAChD,MAAK;YACN,KAAK,aAAa;gBACjB,OAAO,CAAC,IAAI,CAAC,YAAY,CAAC,eAAe,CAAC,CAAA;gBAC1C,MAAK;YACN,KAAK,QAAQ;gBACZ,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC;oBACvB,MAAM,OAAO,GAAG,GAAG,OAAO,CAAC,aAAa,IAAI,EAAE,IAAI,OAAO,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,CAAA;oBAEzE,4FAA4F;oBAC5F,yFAAyF;oBACzF,IAAI,CAAC,OAAO,CAAC,YAAY,IAAI,CAAC,kBAAkB,CAAC,OAAO,CAAC,aAAa,IAAI,EAAE,CAAC,EAAE,CAAC;wBAC/E,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;wBAEpB,MAAK;oBACN,CAAC;oBAED,kFAAkF;oBAClF,IAAI,CAAC,UAAU,CAAC,YAAY,IAAI,CAAC,kBAAkB,CAAC,UAAU,CAAC,aAAa,IAAI,EAAE,CAAC,EAAE,CAAC;wBACrF,MAAM,CAAC,IAAI,CAAC,GAAG,UAAU,CAAC,aAAa,IAAI,EAAE,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,CAAC,CAAA;wBAC5E,OAAO,UAAU,CAAC,aAAa,CAAA;wBAC/B,UAAU,CAAC,MAAM,GAAG,OAAO,CAAC,MAAM,CAAA;wBAElC,IAAI,OAAO,CAAC,aAAa,EAAE,CAAC;4BAC3B,UAAU,CAAC,aAAa,GAAG,OAAO,CAAC,aAAa,CAAA;wBACjD,CAAC;wBAED,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;4BAC1B,UAAU,CAAC,YAAY,GAAG,OAAO,CAAC,YAAY,CAAA;wBAC/C,CAAC;wBAED,MAAK;oBACN,CAAC;oBAED,IAAI,CAAC,WAAW,CAAC,OAAO,EAAE,GAAG,UAAU,CAAC,aAAa,IAAI,EAAE,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC,EAAE,CAAC;wBACrF,OAAO,EAAE,OAAO,EAAE,eAAe,CAAC,kBAAkB,EAAE,CAAA;oBACvD,CAAC;oBAED,MAAK;gBACN,CAAC;gBAED,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;oBAC1B,UAAU,CAAC,YAAY,GAAG,OAAO,CAAC,YAAY,CAAA;gBAC/C,CAAC;gBAED,IAAI,OAAO,CAAC,aAAa,EAAE,CAAC;oBAC3B,UAAU,CAAC,aAAa,GAAG,OAAO,CAAC,aAAa,CAAA;gBACjD,CAAC;gBAED,UAAU,CAAC,MAAM,GAAG,OAAO,CAAC,MAAM,CAAA;gBAClC,MAAK;YACN,KAAK,OAAO;gBACX,IAAI,WAAW,CAAC,OAAO,CAAC,KAAK,EAAE,QAAQ,CAAC,IAAI,WAAW,CAAC,OAAO,CAAC,KAAK,EAAE,KAAK,CAAC,MAAM,CAAC,WAAW,CAAC,CAAC;oBAAE,MAAK;gBAExG,IAAI,MAAM,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,WAAW,CAAC,KAAK,EAAE,OAAO,CAAC,KAAK,CAAC,CAAC;oBAAE,MAAK;gBAEpE,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,CAAA;gBAC1B,MAAK;QACP,CAAC;IACF,CAAC;IAED,kGAAkG;IAClG,MAAM,cAAc,GAAG,MAAM,CAAC,MAAM,CACnC,CAAC,KAAK,EAAE,EAAE,CACT,CAAC,UAAU,CAAC,MAAM;QAClB,CAAC,CAAC,WAAW,CAAC,KAAK,EAAE,UAAU,CAAC,MAAM,CAAC;YACtC,CAAC,WAAW,CAAC,KAAK,EAAE,GAAG,UAAU,CAAC,aAAa,IAAI,EAAE,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC,CAAC,CAChF,CAAA;IAED,IAAI,cAAc,CAAC,MAAM,GAAG,CAAC;QAAE,OAAO,EAAE,OAAO,EAAE,eAAe,CAAC,kBAAkB,EAAE,CAAA;IAErF,IAAI,cAAc,CAAC,CAAC,CAAC,EAAE,CAAC;QACvB,UAAU,CAAC,kBAAkB,GAAG,cAAc,CAAC,CAAC,CAAC,CAAA;IAClD,CAAC;IAED,IAAI,CAAC,UAAU,CAAC,MAAM,IAAI,CAAC,UAAU,CAAC,MAAM,IAAI,CAAC,UAAU,CAAC,kBAAkB,EAAE,CAAC;QAChF,OAAO,EAAE,OAAO,EAAE,eAAe,CAAC,kBAAkB,EAAE,CAAA;IACvD,CAAC;IAED,UAAU,CAAC,QAAQ,GAAG,QAAQ,CAAA;IAC9B,UAAU,CAAC,QAAQ,GAAG,QAAQ,CAAA;IAE9B,IAAI,OAAO,EAAE,KAAK,EAAE,CAAC;QACpB,UAAU,CAAC,KAAK,GAAG,OAAO,CAAC,KAAK,CAAA;IACjC,CAAC;IAED,MAAM,QAAQ,GAAG,gBAAgB,CAAC,UAAU,EAAE,OAAO,EAAE,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC,CAAA;IAE5E,IAAI,CAAC,QAAQ,IAAI,QAAQ,CAAC,QAAQ,CAAC,MAAM;QAAE,OAAO,EAAE,OAAO,EAAE,eAAe,CAAC,YAAY,EAAE,CAAA;IAE3F,OAAO;QACN,QAAQ,EAAE;YACT,GAAG,EAAE,QAAQ,CAAC,GAAG;YACjB,UAAU,EAAE,QAAQ,CAAC,UAAU;YAC/B,OAAO;YACP,MAAM,EAAE,MAAM,OAAO,EAAE;YACvB,MAAM,EAAE,gCAAgC;YACxC,SAAS,EAAE,cAAc,CAAC,gCAAgC,EAAE,QAAQ,CAAC,UAAU,CAAC;YAChF,cAAc,EAAE,EAAE;YAClB,OAAO,EAAE,6BAA6B;SACtC;QACD,OAAO;KACP,CAAA;AACF,CAAC;AAED,MAAM,UAAU,8BAA8B;IAC7C,OAAO;QACN,EAAE,EAAE,gCAAgC;QACpC,cAAc,EAAE,6BAA6B;QAC7C,WAAW,EAAE,WAAW,CAAC,QAAQ;QACjC,QAAQ,EAAE,cAAc,CAAC,iBAAiB;QAC1C,OAAO,EAAE,aAAa,CAAC,QAAQ;QAC/B,WAAW,EACV,6GAA6G;QAE9G,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,IAAI,IAAI,CAAC,OAAO,IAAI,CAAC,+BAA+B,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;gBAC7E,MAAM,IAAI,uBAAuB,CAChC,gCAAgC,EAChC,+BAA+B,EAC/B,IAAI,CAAC,OAAO,CACZ,CAAA;YACF,CAAC;YAED,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;YAC9B,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,KAAK,EAAE,MAAM,MAAM,IAAI,eAAe,CAAC,SAAS,CAAiB,IAAI,CAAC,SAAS,CAAC,EAAE,CAAC;gBACtF,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;oBAAE,MAAK;gBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;oBAAE,MAAK;gBAE5D,MAAM,OAAO,GAAG,kBAAkB,CAAC,MAAM,CAAC,CAAA;gBAE1C,IAAI,CAAC,CAAC,UAAU,IAAI,OAAO,CAAC,EAAE,CAAC;oBAC9B,YAAY,CAAC,IAAI,EAAE,OAAO,CAAC,OAAO,CAAC,CAAA;oBAEnC,SAAQ;gBACT,CAAC;gBAED,mEAAmE;gBACnE,iDAAiD;gBACjD,IAAI,IAAI,CAAC,OAAO,IAAI,OAAO,CAAC,QAAQ,CAAC,OAAO,KAAK,IAAI,CAAC,OAAO;oBAAE,SAAQ;gBAEvE,IAAI,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;oBAC1C,YAAY,CAAC,IAAI,EAAE,eAAe,CAAC,SAAS,CAAC,CAAA;oBAE7C,SAAQ;gBACT,CAAC;gBAED,KAAK,MAAM,MAAM,IAAI,OAAO,CAAC,OAAO,EAAE,CAAC;oBACtC,YAAY,CAAC,IAAI,EAAE,MAAM,CAAC,CAAA;gBAC3B,CAAC;gBAED,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAA;gBACpC,MAAM,OAAO,CAAC,QAAQ,CAAA;gBAEtB,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,8BAA8B,EAAE,CAAA"}

package/out/fr/adapters/ban/adapter.d.ts

@@ -3,37 +3,31 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   `ban`: Base Adresse Nationale CSV adapter (FR street-level).
+ * `ban`: Base Adresse Nationale CSV adapter, street-level over eleven jurisdictions.
  *
- *   Input: a CSV dump from `adresse.data.gouv.fr` (semicolon-separated, ~25M rows nationally). The
- *   adapter only reads the small set of columns needed for the corpus:
+ * Input is a semicolon-separated CSV dump from `adresse.data.gouv.fr`. The adapter reads `numero`
+ * as `house_number`, `rep` as a repetition index appended to it, `nom_voie` as `street`,
+ * `code_postal` as `postcode`, and `nom_commune` as `locality`. BAN carries no region column, so
+ * region is left to the wof-postalcode and wof-admin cross-reference at corpus build time.
  *
- *   - `numero` → `house_number`
- *   - `rep` → repetition index ("bis", "ter") appended to house_number
- *   - `nom_voie` → `street` (full road name; includes the prefix "Rue", "Avenue", etc.)
- *   - `code_postal` → `postcode`
- *   - `nom_commune` → `locality`
+ * IGN publishes one file per INSEE department beside the whole-country extract, and the overseas
+ * departments carry the same 23-column header as the metropolitan rows. A row's country is therefore
+ * derived from `code_insee` rather than asserted by the caller, which reads the same whether the
+ * input is one department's file or an extract spanning several.
  *
- *   `region` and `country` are not in BAN. The adapter stamps `country: "FR"` on every row; region is
- *   left for the wof-postalcode + wof-admin cross-reference at corpus build time (a future pass;
- *   for Phase 1 the row's region is simply absent).
+ * The official BAN is dual-licensed under Licence Ouverte 2.0 and ODbL.
+ * This adapter elects Licence Ouverte 2.0. Its terms permit training with attribution.
+ * It records that license on every row. The model card must include BAN attribution.
  *
- *   License: the official BAN (adresse.data.gouv.fr) is DUAL-licensed — Licence Ouverte 2.0 (Etalab,
- *   attribution-only) OR ODbL (share-alike). We ELECT Licence Ouverte 2.0 (issue #26 Tier B:
- *   allowed for training with attribution; the ODbL option's share-alike obligation would defeat
- *   the proprietary-weights goal). Stamped onto every row as `Licence Ouverte 2.0` — NOT the older
- *   conservative `ODbL-1.0` label, which wrongly read as Tier-C-denied in the corpus license audit.
- *   The model card MUST carry the BAN attribution (Tier B obligation).
- *
- *   The adapter is streaming-aware: `CSVSpliterator.fromAsync` reads the `;`-delimited dump row by
- *   row, so a 25M-row file never sits in memory. Honors `opts.limit` for fixture / smoke runs,
- *   `opts.signal` for cancellation, and `opts.country` for a self-consistency check (errors if
- *   country !== FR).
+ * The adapter streams with `CSVSpliterator.fromAsync`, so a 25M-row file never sits in memory. It
+ * honors `opts.limit` and `opts.signal`. `opts.country` is optional: when set it both rejects a
+ * jurisdiction BAN does not publish and keeps only the rows of the one named.
  */
-import type { CorpusAdapter } from "#types";
+import { type CorpusAdapter } from "#types";
 /**
- * Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset
- * it came from.
+ * Registry id for this adapter.
+ *
+ * Stamped into every row it emits, so a corpus record can be traced back to the dataset it came from.
  */
 export declare const BAN_ADAPTER_ID = "ban";
 export declare function createBanAdapter(): CorpusAdapter;

package/out/fr/adapters/ban/adapter.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/fr/adapters/ban/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA+BG;AAOH,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEzE;;;GAGG;AACH,eAAO,MAAM,cAAc,QAAQ,CAAA;AA4BnC,wBAAgB,gBAAgB,IAAI,aAAa,CAiFhD;AAED;;GAEG;AACH,eAAO,MAAM,UAAU,eAAqB,CAAA"}
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/fr/adapters/ban/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AAWH,OAAO,EAIN,KAAK,aAAa,EAGlB,MAAM,QAAQ,CAAA;AAEf;;;;GAIG;AACH,eAAO,MAAM,cAAc,QAAQ,CAAA;AAkBnC,wBAAgB,gBAAgB,IAAI,aAAa,CAkHhD;AAED;;GAEG;AACH,eAAO,MAAM,UAAU,eAAqB,CAAA"}