@mailwoman/corpus 9.4.0 → 10.1.0

package/out/adapters/wof/postalcode-json/adapter.js

@@ -0,0 +1,165 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Emits corpus rows from Who's On First postalcode GeoJSON repositories.
+ *
+ *   The input directory must hold the `whosonfirst-data-postalcode-<cc>` repositories and the matching
+ *   `whosonfirst-data-admin-<cc>` repositories. Postcode records point to their ancestors by `wof:parent_id`, so the
+ *   admin records must be in the same walk.
+ *
+ *   Each postcode record produces one row per pair of name slot and hierarchy variant. The name slots are the canonical
+ *   `wof:name` (slot `default`) plus any `name:*` variants. The hierarchy variants are the postcode by itself, then with
+ *   locality, region and country added in turn. Ancestor names always use the canonical `wof:name`.
+ *
+ *   `source_id` has the form `wof-postalcode-<wof_id>-<name-slot>-<hierarchy-variant>`.
+ */
+import { COUNTRY_DISPLAY_NAME, emitWOFJSONRows, LOCALE_BY_COUNTRY, nameSlotsFor as wofNameSlotsFor, } from "#adapters/wof/json-rows";
+import { SourceRegister } from "#registers";
+import { AddressRole, SurfaceOrigin } from "#types";
+import { US_STATE_BY_ABBREVIATION } from "#us/fips-state";
+import { buildAncestorNameIndex, walkFeatures } from "#utils";
+/**
+ * Maps a lowercase US state name to its USPS code.
+ *
+ * WOF stores full state names such as `Oregon`, and the address layout prints components unchanged.
+ * The adapter substitutes the USPS code so the row matches how a US postal address is written.
+ */
+const US_STATE_ABBREVIATION_BY_NAME = new Map(Object.values(US_STATE_BY_ABBREVIATION).map((state) => [state.name.toLowerCase(), state.abbreviation]));
+/**
+ * Returns the region text to print.
+ *
+ * The adapter converts US state names to USPS codes.
+ * Other names pass through.
+ */
+function regionSurface(country, name) {
+    if (country !== "US")
+        return name;
+    return US_STATE_ABBREVIATION_BY_NAME.get(name.trim().toLowerCase()) ?? name;
+}
+/**
+ * Maps a WOF placetype to a component tag, or returns `undefined` for a placetype this adapter skips.
+ *
+ * The mapping also filters records, so it covers only `postalcode`
+ * and the ancestor placetypes the variants use.
+ * The admin adapter keeps its own mapping.
+ */
+function placetypeToTag(placetype) {
+    switch (placetype) {
+        case "country":
+        case "nation":
+            return "country";
+        case "macroregion":
+        case "region":
+            return "region";
+        case "locality":
+            return "locality";
+        case "postalcode":
+            return "postcode";
+        default:
+            return undefined;
+    }
+}
+/**
+ * Builds the hierarchy variants for a postcode record, or returns an empty list for any other placetype.
+ *
+ * `selfName` is the postcode text for the current name slot.
+ */
+export function postcodeVariantsFor(row, ancestry, selfName) {
+    if (placetypeToTag(row.placetype) !== "postcode")
+        return [];
+    const locality = ancestry.locality;
+    const region = ancestry.region;
+    const country = ancestry.country;
+    const countryDisplay = COUNTRY_DISPLAY_NAME[row.country] ?? country ?? row.country;
+    const variants = [{ suffix: "self", components: { postcode: selfName } }];
+    if (locality) {
+        variants.push({
+            suffix: "with-locality",
+            components: { postcode: selfName, locality },
+        });
+    }
+    if (locality && region) {
+        variants.push({
+            suffix: "with-locality-region",
+            components: { postcode: selfName, locality, region },
+        });
+    }
+    if (locality && region && country) {
+        variants.push({
+            suffix: "with-locality-region-country",
+            components: {
+                postcode: selfName,
+                locality,
+                region: regionSurface(row.country, region),
+                country: countryDisplay,
+            },
+        });
+    }
+    return variants;
+}
+/**
+ * Returns a record's name slots.
+ *
+ * The `default` slot holds `wof:name` unchanged.
+ * The other slots hold `name:*` variants that differ from it.
+ */
+export function nameSlotsFor(rec) {
+    return wofNameSlotsFor(rec);
+}
+/**
+ * The source ID on rows from this adapter.
+ */
+export const WOF_POSTALCODE_ADAPTER_ID = "wof-postalcode";
+/**
+ * Creates the WOF postalcode adapter.
+ */
+export function createWOFPostalcodeAdapter() {
+    return {
+        id: WOF_POSTALCODE_ADAPTER_ID,
+        defaultLicense: "CC0-1.0",
+        addressRole: AddressRole.Premise,
+        register: SourceRegister.WhosOnFirst,
+        surface: SurfaceOrigin.Rendered,
+        description: "Who's On First postalcode GeoJSON bundles (postcode → locality/region pairs). Ancestor names from sibling admin repos.",
+        async *rows(opts) {
+            // The index keeps admin records as well as postcodes so that postcode ancestry resolves.
+            const byID = new Map();
+            for await (const rec of walkFeatures(opts.inputPath, { signal: opts.signal })) {
+                if (opts.signal?.aborted)
+                    return;
+                if (opts.country && rec.country !== opts.country)
+                    continue;
+                if (!placetypeToTag(rec.placetype))
+                    continue;
+                byID.set(rec.id, rec);
+            }
+            const ancestry = buildAncestorNameIndex(byID, placetypeToTag);
+            // The adapter read admin records so postcode ancestry resolves.
+            // The index above now holds the three ancestor names read by each postcode variant.
+            // Their removal here releases every admin record's `nameVariants` map before the emit pass.
+            // `shouldEmit` skipped those records anyway.
+            for (const [id, record] of byID) {
+                if (placetypeToTag(record.placetype) !== "postcode") {
+                    byID.delete(id);
+                }
+            }
+            yield* emitWOFJSONRows({
+                records: byID,
+                ancestry,
+                adapterOptions: opts,
+                adapterID: WOF_POSTALCODE_ADAPTER_ID,
+                localeByCountry: LOCALE_BY_COUNTRY,
+                shouldEmit: (record) => placetypeToTag(record.placetype) === "postcode",
+                nameSlotsFor,
+                variantsFor: postcodeVariantsFor,
+            });
+        },
+    };
+}
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export const wofPostalcodeAdapter = createWOFPostalcodeAdapter();
+//# sourceMappingURL=adapter.js.map

package/out/adapters/wof/postalcode-json/adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../lib/adapters/wof/postalcode-json/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AAKH,OAAO,EACN,oBAAoB,EACpB,eAAe,EACf,iBAAiB,EACjB,YAAY,IAAI,eAAe,GAE/B,MAAM,yBAAyB,CAAA;AAChC,OAAO,EAAE,cAAc,EAAE,MAAM,YAAY,CAAA;AAC3C,OAAO,EAAE,WAAW,EAA8D,aAAa,EAAE,MAAM,QAAQ,CAAA;AAC/G,OAAO,EAAE,wBAAwB,EAAE,MAAM,gBAAgB,CAAA;AACzD,OAAO,EAAE,sBAAsB,EAAE,YAAY,EAAsC,MAAM,QAAQ,CAAA;AAEjG;;;;;GAKG;AACH,MAAM,6BAA6B,GAAgC,IAAI,GAAG,CACzE,MAAM,CAAC,MAAM,CAAC,wBAAwB,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,WAAW,EAAE,EAAE,KAAK,CAAC,YAAY,CAAC,CAAC,CACtG,CAAA;AAED;;;;;GAKG;AACH,SAAS,aAAa,CAAC,OAAe,EAAE,IAAY;IACnD,IAAI,OAAO,KAAK,IAAI;QAAE,OAAO,IAAI,CAAA;IAEjC,OAAO,6BAA6B,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC,IAAI,IAAI,CAAA;AAC5E,CAAC;AAED;;;;;;GAMG;AACH,SAAS,cAAc,CAAC,SAAwC;IAC/D,QAAQ,SAAS,EAAE,CAAC;QACnB,KAAK,SAAS,CAAC;QACf,KAAK,QAAQ;YACZ,OAAO,SAAS,CAAA;QACjB,KAAK,aAAa,CAAC;QACnB,KAAK,QAAQ;YACZ,OAAO,QAAQ,CAAA;QAChB,KAAK,UAAU;YACd,OAAO,UAAU,CAAA;QAClB,KAAK,YAAY;YAChB,OAAO,UAAU,CAAA;QAClB;YACC,OAAO,SAAS,CAAA;IAClB,CAAC;AACF,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,mBAAmB,CAAC,GAAc,EAAE,QAAuB,EAAE,QAAgB;IAC5F,IAAI,cAAc,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,UAAU;QAAE,OAAO,EAAE,CAAA;IAE3D,MAAM,QAAQ,GAAG,QAAQ,CAAC,QAAQ,CAAA;IAClC,MAAM,MAAM,GAAG,QAAQ,CAAC,MAAM,CAAA;IAC9B,MAAM,OAAO,GAAG,QAAQ,CAAC,OAAO,CAAA;IAChC,MAAM,cAAc,GAAG,oBAAoB,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,OAAO,IAAI,GAAG,CAAC,OAAO,CAAA;IAElF,MAAM,QAAQ,GAAqB,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,EAAE,QAAQ,EAAE,QAAQ,EAAE,EAAE,CAAC,CAAA;IAE3F,IAAI,QAAQ,EAAE,CAAC;QACd,QAAQ,CAAC,IAAI,CAAC;YACb,MAAM,EAAE,eAAe;YACvB,UAAU,EAAE,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE;SAC5C,CAAC,CAAA;IACH,CAAC;IAED,IAAI,QAAQ,IAAI,MAAM,EAAE,CAAC;QACxB,QAAQ,CAAC,IAAI,CAAC;YACb,MAAM,EAAE,sBAAsB;YAC9B,UAAU,EAAE,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE;SACpD,CAAC,CAAA;IACH,CAAC;IAED,IAAI,QAAQ,IAAI,MAAM,IAAI,OAAO,EAAE,CAAC;QACnC,QAAQ,CAAC,IAAI,CAAC;YACb,MAAM,EAAE,8BAA8B;YACtC,UAAU,EAAE;gBACX,QAAQ,EAAE,QAAQ;gBAClB,QAAQ;gBACR,MAAM,EAAE,aAAa,CAAC,GAAG,CAAC,OAAO,EAAE,MAAM,CAAC;gBAC1C,OAAO,EAAE,cAAc;aACvB;SACD,CAAC,CAAA;IACH,CAAC;IAED,OAAO,QAAQ,CAAA;AAChB,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,YAAY,CAAC,GAAc;IAC1C,OAAO,eAAe,CAAC,GAAG,CAAC,CAAA;AAC5B,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,yBAAyB,GAAG,gBAAgB,CAAA;AAEzD;;GAEG;AACH,MAAM,UAAU,0BAA0B;IACzC,OAAO;QACN,EAAE,EAAE,yBAAyB;QAC7B,cAAc,EAAE,SAAS;QACzB,WAAW,EAAE,WAAW,CAAC,OAAO;QAChC,QAAQ,EAAE,cAAc,CAAC,WAAW;QACpC,OAAO,EAAE,aAAa,CAAC,QAAQ;QAC/B,WAAW,EACV,wHAAwH;QAEzH,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,yFAAyF;YACzF,MAAM,IAAI,GAAG,IAAI,GAAG,EAAqB,CAAA;YAEzC,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,YAAY,CAAC,IAAI,CAAC,SAAS,EAAE,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,CAAC;gBAC/E,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;oBAAE,OAAM;gBAEhC,IAAI,IAAI,CAAC,OAAO,IAAI,GAAG,CAAC,OAAO,KAAK,IAAI,CAAC,OAAO;oBAAE,SAAQ;gBAE1D,IAAI,CAAC,cAAc,CAAC,GAAG,CAAC,SAAS,CAAC;oBAAE,SAAQ;gBAC5C,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,EAAE,GAAG,CAAC,CAAA;YACtB,CAAC;YAED,MAAM,QAAQ,GAAG,sBAAsB,CAAC,IAAI,EAAE,cAAc,CAAC,CAAA;YAE7D,gEAAgE;YAChE,oFAAoF;YACpF,4FAA4F;YAC5F,6CAA6C;YAC7C,KAAK,MAAM,CAAC,EAAE,EAAE,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC;gBACjC,IAAI,cAAc,CAAC,MAAM,CAAC,SAAS,CAAC,KAAK,UAAU,EAAE,CAAC;oBACrD,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,CAAA;gBAChB,CAAC;YACF,CAAC;YAED,KAAK,CAAC,CAAC,eAAe,CAAC;gBACtB,OAAO,EAAE,IAAI;gBACb,QAAQ;gBACR,cAAc,EAAE,IAAI;gBACpB,SAAS,EAAE,yBAAyB;gBACpC,eAAe,EAAE,iBAAiB;gBAClC,UAAU,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,cAAc,CAAC,MAAM,CAAC,SAAS,CAAC,KAAK,UAAU;gBACvE,YAAY;gBACZ,WAAW,EAAE,mBAAmB;aAChC,CAAC,CAAA;QACH,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAG,0BAA0B,EAAE,CAAA"}

package/out/adapters.d.ts

@@ -0,0 +1,75 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Adapter registry bootstrap.
+ *
+ *   The module registers every built-in adapter with `defaultAdapterRegistry` when loaded. The CLI
+ *   (`commands/corpus/list.tsx`, `commands/corpus/run.tsx`) imports it once at startup.
+ *
+ *   Each adapter has its own directory, and the publisher decides which parent it sits under. An
+ *   adapter reading one country's publisher lives under that country, at `lib/<iso2>/adapters/<id>/`:
+ *   G-NAF under `au`, Digitaal Vlaanderen under `be`, ČÚZK under `cz`, Klimadatastyrelsen under
+ *   `dk`, Maa- ja Ruumiamet under `ee`, Ryhti under `fi`, BAN, FINESS and the Annuaire de l'éducation under `fr`, ANAC under `it`, the
+ *   Administration du
+ *   cadastre et de la topographie under `lu`, the State Address Register under `lv`, Kartverket's Matrikkelen
+ *   under `no`, EMUiA under `pl`, Romania's school network under `ro`, and the FCC and USGov readers under `us`. A national publisher
+ *   stays there even when its own files cover that state's dependencies, because the publisher is
+ *   what the adapter reads: BAN carries France and ten overseas jurisdictions, Matrikkelen carries
+ *   Norway and Svalbard, and Ryhti's one national file carries Åland beside the Finnish mainland.
+ *
+ *   Flanders is the counter-case worth stating. Its adapter sits under `be` and covers the Flemish
+ *   Region alone, because Brussels and Wallonia publish the INSPIRE theme through their own
+ *   services, so Belgium needs three adapters rather than one.
+ *
+ *   Spain needs four, for the same reason. `es-catastro` reads the Dirección General del Catastro's
+ *   52 provinces, and `es-bizkaia`, `es-gipuzkoa` and `es-navarra` read the foral cadastres, which
+ *   state their own terms and which the national feed links rather than serves.
+ *
+ *   An adapter reading a publisher that is not any one country's lives here, at `lib/adapters/<id>/`.
+ *   That is the aggregators, `./openaddresses/`, `./osm/`, `./overture/`, `./geonames/` and `./wof/`,
+ *   plus `./po-box/`.
+ *
+ *   A new adapter joins the `BUILTIN_ADAPTERS` list below. A test that needs a pristine registry
+ *   constructs its own `InMemoryAdapterRegistry` rather than mutating the default.
+ *
+ *   The WOF adapters keep their canonical ids, `wof-admin` and `wof-postalcode`, so existing
+ *   `mailwoman corpus build` callsites continue to work. `./wof/admin/json/` and
+ *   `./wof/postalcode-json/` hold the implementations.
+ */
+import type { CorpusAdapter } from "#types";
+/**
+ * Built-in adapters.
+ *
+ * Order is significant, because `corpus build` iterates this list to drive every adapter in turn.
+ * The order runs coarse-first (admin, then postcode), then street-level
+ * (BAN FR, tiger US, OpenAddresses global), then adversarial-source (FCC BDC US, HRSA fqhc US).
+ *
+ * The `usgov-samhsa-treatment-locator` adapter is intentionally absent from this list.
+ * The samhsa Open Data Foundry bulk CSV it was written against is no longer publicly distributed.
+ *
+ * Its factory and export remain available so an operator who obtains a compatible CSV can hand-register it.
+ * Re-add it here once a stable public source returns.
+ */
+export declare const BUILTIN_ADAPTERS: readonly CorpusAdapter[];
+export { BAN_ADAPTER_ID, banAdapter } from "#fr/adapters/ban/adapter";
+export { MATRIKKELEN_ADAPTER_ID, matrikkelenAdapter } from "#no/adapters/matrikkelen/adapter";
+export { FCC_BDC_ADAPTER_ID, FCC_BDC_DEFAULT_LICENSE, fccBdcAdapter } from "#us/adapters/fcc-bdc/adapter";
+export { GEONAMES_POSTAL_ADAPTER_ID, GEONAMES_POSTAL_DEFAULT_LICENSE, geonamesPostalAdapter, } from "#adapters/geonames/postal/adapter";
+export { GEONAMES_ADAPTER_ID, GEONAMES_DEFAULT_LICENSE, geonamesAdapter } from "#adapters/geonames/adapter";
+export { OPENADDRESSES_ADAPTER_ID, OPENADDRESSES_DEFAULT_LICENSE, openaddressesAdapter, } from "#adapters/openaddresses/adapter";
+export { STATE_HI_SCHOOLS_ADAPTER_ID, STATE_HI_SCHOOLS_DEFAULT_LICENSE, stateHiSchoolsAdapter, } from "#us/adapters/state/hi-schools/adapter";
+export { STATE_IA_CONTRACTORS_ADAPTER_ID, STATE_IA_CONTRACTORS_DEFAULT_LICENSE, stateIaContractorsAdapter, } from "#us/adapters/state/ia-contractors/adapter";
+export { STATE_NY_NOTARIES_ADAPTER_ID, STATE_NY_NOTARIES_DEFAULT_LICENSE, stateNyNotariesAdapter, } from "#us/adapters/state/ny-notaries/adapter";
+export { STATE_TX_NOTARIES_ADAPTER_ID, STATE_TX_NOTARIES_DEFAULT_LICENSE, stateTxNotariesAdapter, } from "#us/adapters/state/tx-notaries/adapter";
+export { TIGER_ADAPTER_ID, TIGER_DEFAULT_LICENSE, tigerAdapter } from "#us/adapters/tiger/adapter";
+export { USGOV_HRSA_FQHC_ADAPTER_ID, USGOV_HRSA_FQHC_DEFAULT_LICENSE, usgovHrsaFqhcAdapter, } from "#us/adapters/usgov/hrsa-fqhc/adapter";
+export { USGOV_IMLS_PLS_ADAPTER_ID, USGOV_IMLS_PLS_DEFAULT_LICENSE, usgovImlsPlsAdapter, } from "#us/adapters/usgov/imls-pls/adapter";
+export { USGOV_IRS_BMF_ADAPTER_ID, USGOV_IRS_BMF_DEFAULT_LICENSE, USGovIRSBMFAdapter, } from "#us/adapters/usgov/irs-bmf/adapter";
+export { USGOV_NAD_ADAPTER_ID, USGOV_NAD_DEFAULT_LICENSE, usgovNADAdapter } from "#us/adapters/usgov/nad/adapter";
+export { USGOV_NPPES_ADAPTER_ID, USGOV_NPPES_DEFAULT_LICENSE, usgovNPPESAdapter, } from "#us/adapters/usgov/nppes/adapter";
+export { USGOV_SAMHSA_ADAPTER_ID, USGOV_SAMHSA_DEFAULT_LICENSE, usgovSamhsaTreatmentLocatorAdapter, } from "#us/adapters/usgov/samhsa-treatment-locator/adapter";
+export { WOF_ADMIN_ADAPTER_ID, wofAdminAdapter } from "#adapters/wof/admin/json/adapter";
+export { WOF_POSTALCODE_ADAPTER_ID, wofPostalcodeAdapter } from "#adapters/wof/postalcode-json/adapter";
+//# sourceMappingURL=adapters.d.ts.map

package/out/adapters.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapters.d.ts","sourceRoot":"","sources":["../lib/adapters.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAuCG;AAmCH,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,QAAQ,CAAA;AAa3C;;;;;;;;;;;;GAYG;AACH,eAAO,MAAM,gBAAgB,EAAE,SAAS,aAAa,EA4CpD,CAAA;AAQD,OAAO,EAAE,cAAc,EAAE,UAAU,EAAE,MAAM,0BAA0B,CAAA;AACrE,OAAO,EAAE,sBAAsB,EAAE,kBAAkB,EAAE,MAAM,kCAAkC,CAAA;AAC7F,OAAO,EAAE,kBAAkB,EAAE,uBAAuB,EAAE,aAAa,EAAE,MAAM,8BAA8B,CAAA;AAEzG,OAAO,EACN,0BAA0B,EAC1B,+BAA+B,EAC/B,qBAAqB,GACrB,MAAM,mCAAmC,CAAA;AAE1C,OAAO,EAAE,mBAAmB,EAAE,wBAAwB,EAAE,eAAe,EAAE,MAAM,4BAA4B,CAAA;AAE3G,OAAO,EACN,wBAAwB,EACxB,6BAA6B,EAC7B,oBAAoB,GACpB,MAAM,iCAAiC,CAAA;AAExC,OAAO,EACN,2BAA2B,EAC3B,gCAAgC,EAChC,qBAAqB,GACrB,MAAM,uCAAuC,CAAA;AAE9C,OAAO,EACN,+BAA+B,EAC/B,oCAAoC,EACpC,yBAAyB,GACzB,MAAM,2CAA2C,CAAA;AAElD,OAAO,EACN,4BAA4B,EAC5B,iCAAiC,EACjC,sBAAsB,GACtB,MAAM,wCAAwC,CAAA;AAE/C,OAAO,EACN,4BAA4B,EAC5B,iCAAiC,EACjC,sBAAsB,GACtB,MAAM,wCAAwC,CAAA;AAE/C,OAAO,EAAE,gBAAgB,EAAE,qBAAqB,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAA;AAElG,OAAO,EACN,0BAA0B,EAC1B,+BAA+B,EAC/B,oBAAoB,GACpB,MAAM,sCAAsC,CAAA;AAE7C,OAAO,EACN,yBAAyB,EACzB,8BAA8B,EAC9B,mBAAmB,GACnB,MAAM,qCAAqC,CAAA;AAE5C,OAAO,EACN,wBAAwB,EACxB,6BAA6B,EAC7B,kBAAkB,GAClB,MAAM,oCAAoC,CAAA;AAE3C,OAAO,EAAE,oBAAoB,EAAE,yBAAyB,EAAE,eAAe,EAAE,MAAM,gCAAgC,CAAA;AAEjH,OAAO,EACN,sBAAsB,EACtB,2BAA2B,EAC3B,iBAAiB,GACjB,MAAM,kCAAkC,CAAA;AAEzC,OAAO,EACN,uBAAuB,EACvB,4BAA4B,EAC5B,kCAAkC,GAClC,MAAM,qDAAqD,CAAA;AAE5D,OAAO,EAAE,oBAAoB,EAAE,eAAe,EAAE,MAAM,kCAAkC,CAAA;AACxF,OAAO,EAAE,yBAAyB,EAAE,oBAAoB,EAAE,MAAM,uCAAuC,CAAA"}

package/out/adapters.js

@@ -0,0 +1,167 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Adapter registry bootstrap.
+ *
+ *   The module registers every built-in adapter with `defaultAdapterRegistry` when loaded. The CLI
+ *   (`commands/corpus/list.tsx`, `commands/corpus/run.tsx`) imports it once at startup.
+ *
+ *   Each adapter has its own directory, and the publisher decides which parent it sits under. An
+ *   adapter reading one country's publisher lives under that country, at `lib/<iso2>/adapters/<id>/`:
+ *   G-NAF under `au`, Digitaal Vlaanderen under `be`, ČÚZK under `cz`, Klimadatastyrelsen under
+ *   `dk`, Maa- ja Ruumiamet under `ee`, Ryhti under `fi`, BAN, FINESS and the Annuaire de l'éducation under `fr`, ANAC under `it`, the
+ *   Administration du
+ *   cadastre et de la topographie under `lu`, the State Address Register under `lv`, Kartverket's Matrikkelen
+ *   under `no`, EMUiA under `pl`, Romania's school network under `ro`, and the FCC and USGov readers under `us`. A national publisher
+ *   stays there even when its own files cover that state's dependencies, because the publisher is
+ *   what the adapter reads: BAN carries France and ten overseas jurisdictions, Matrikkelen carries
+ *   Norway and Svalbard, and Ryhti's one national file carries Åland beside the Finnish mainland.
+ *
+ *   Flanders is the counter-case worth stating. Its adapter sits under `be` and covers the Flemish
+ *   Region alone, because Brussels and Wallonia publish the INSPIRE theme through their own
+ *   services, so Belgium needs three adapters rather than one.
+ *
+ *   Spain needs four, for the same reason. `es-catastro` reads the Dirección General del Catastro's
+ *   52 provinces, and `es-bizkaia`, `es-gipuzkoa` and `es-navarra` read the foral cadastres, which
+ *   state their own terms and which the national feed links rather than serves.
+ *
+ *   An adapter reading a publisher that is not any one country's lives here, at `lib/adapters/<id>/`.
+ *   That is the aggregators, `./openaddresses/`, `./osm/`, `./overture/`, `./geonames/` and `./wof/`,
+ *   plus `./po-box/`.
+ *
+ *   A new adapter joins the `BUILTIN_ADAPTERS` list below. A test that needs a pristine registry
+ *   constructs its own `InMemoryAdapterRegistry` rather than mutating the default.
+ *
+ *   The WOF adapters keep their canonical ids, `wof-admin` and `wof-postalcode`, so existing
+ *   `mailwoman corpus build` callsites continue to work. `./wof/admin/json/` and
+ *   `./wof/postalcode-json/` hold the implementations.
+ */
+import { cordisAdapter } from "#adapters/cordis/adapter";
+import { geonamesAdapter } from "#adapters/geonames/adapter";
+import { geonamesPostalAdapter } from "#adapters/geonames/postal/adapter";
+import { gleifAdapter } from "#adapters/gleif/adapter";
+import { openaddressesAdapter } from "#adapters/openaddresses/adapter";
+import { osmAdapter } from "#adapters/osm/adapter";
+import { overtureAdapter } from "#adapters/overture/adapter";
+import { defaultAdapterRegistry } from "#adapters/registry";
+import { wofAdminAdapter } from "#adapters/wof/admin/json/adapter";
+import { wofPostalcodeAdapter } from "#adapters/wof/postalcode-json/adapter";
+import { gnafAdapter } from "#au/adapters/gnaf/adapter";
+import { brusselsAdapter } from "#be/adapters/brussels/adapter";
+import { vlaanderenAdapter } from "#be/adapters/vlaanderen/adapter";
+import { wallonieAdapter } from "#be/adapters/wallonie/adapter";
+import { czCuzkAdapter } from "#cz/adapters/cuzk/adapter";
+import { dkInspireAdapter } from "#dk/adapters/inspire/adapter";
+import { adsAdapter } from "#ee/adapters/ads/adapter";
+import { esBizkaiaAdapter } from "#es/adapters/bizkaia/adapter";
+import { esCatastroAdapter } from "#es/adapters/catastro/adapter";
+import { esGipuzkoaAdapter } from "#es/adapters/gipuzkoa/adapter";
+import { esNavarraAdapter } from "#es/adapters/navarra/adapter";
+import { ryhtiAdapter } from "#fi/adapters/ryhti/adapter";
+import { annuaireEducationAdapter } from "#fr/adapters/annuaire-education/adapter";
+import { banAdapter } from "#fr/adapters/ban/adapter";
+import { finessAdapter } from "#fr/adapters/finess/adapter";
+import { itANACAdapter } from "#it/adapters/anac/adapter";
+import { bdAdressesAdapter } from "#lu/adapters/bd-adresses/adapter";
+import { varisAdapter } from "#lv/adapters/varis/adapter";
+import { nlKadasterAdapter } from "#nl/adapters/kadaster/adapter";
+import { matrikkelenAdapter } from "#no/adapters/matrikkelen/adapter";
+import { emuiaAdapter } from "#pl/adapters/emuia/adapter";
+import { reteaScolaraAdapter } from "#ro/adapters/retea-scolara/adapter";
+import { skInspireAdapter } from "#sk/adapters/inspire/adapter";
+import { fccBdcAdapter } from "#us/adapters/fcc-bdc/adapter";
+import { stateHiSchoolsAdapter } from "#us/adapters/state/hi-schools/adapter";
+import { stateIaContractorsAdapter } from "#us/adapters/state/ia-contractors/adapter";
+import { stateNyNotariesAdapter } from "#us/adapters/state/ny-notaries/adapter";
+import { stateTxNotariesAdapter } from "#us/adapters/state/tx-notaries/adapter";
+import { tigerAdapter } from "#us/adapters/tiger/adapter";
+import { usgovHrsaFqhcAdapter } from "#us/adapters/usgov/hrsa-fqhc/adapter";
+import { usgovImlsPlsAdapter } from "#us/adapters/usgov/imls-pls/adapter";
+import { USGovIRSBMFAdapter } from "#us/adapters/usgov/irs-bmf/adapter";
+import { usgovNADAdapter } from "#us/adapters/usgov/nad/adapter";
+import { usgovNPPESAdapter } from "#us/adapters/usgov/nppes/adapter";
+/**
+ * Built-in adapters.
+ *
+ * Order is significant, because `corpus build` iterates this list to drive every adapter in turn.
+ * The order runs coarse-first (admin, then postcode), then street-level
+ * (BAN FR, tiger US, OpenAddresses global), then adversarial-source (FCC BDC US, HRSA fqhc US).
+ *
+ * The `usgov-samhsa-treatment-locator` adapter is intentionally absent from this list.
+ * The samhsa Open Data Foundry bulk CSV it was written against is no longer publicly distributed.
+ *
+ * Its factory and export remain available so an operator who obtains a compatible CSV can hand-register it.
+ * Re-add it here once a stable public source returns.
+ */
+export const BUILTIN_ADAPTERS = [
+    wofAdminAdapter,
+    wofPostalcodeAdapter,
+    geonamesAdapter,
+    geonamesPostalAdapter,
+    banAdapter,
+    matrikkelenAdapter,
+    ryhtiAdapter,
+    czCuzkAdapter,
+    dkInspireAdapter,
+    adsAdapter,
+    emuiaAdapter,
+    vlaanderenAdapter,
+    wallonieAdapter,
+    brusselsAdapter,
+    nlKadasterAdapter,
+    bdAdressesAdapter,
+    skInspireAdapter,
+    esCatastroAdapter,
+    esBizkaiaAdapter,
+    esGipuzkoaAdapter,
+    esNavarraAdapter,
+    tigerAdapter,
+    openaddressesAdapter,
+    overtureAdapter,
+    osmAdapter,
+    gnafAdapter,
+    fccBdcAdapter,
+    usgovHrsaFqhcAdapter,
+    usgovNPPESAdapter,
+    usgovNADAdapter,
+    usgovImlsPlsAdapter,
+    USGovIRSBMFAdapter,
+    stateIaContractorsAdapter,
+    stateTxNotariesAdapter,
+    stateNyNotariesAdapter,
+    stateHiSchoolsAdapter,
+    itANACAdapter,
+    reteaScolaraAdapter,
+    cordisAdapter,
+    varisAdapter,
+    finessAdapter,
+    annuaireEducationAdapter,
+    gleifAdapter,
+];
+for (const adapter of BUILTIN_ADAPTERS) {
+    if (!defaultAdapterRegistry.get(adapter.id)) {
+        defaultAdapterRegistry.register(adapter);
+    }
+}
+export { BAN_ADAPTER_ID, banAdapter } from "#fr/adapters/ban/adapter";
+export { MATRIKKELEN_ADAPTER_ID, matrikkelenAdapter } from "#no/adapters/matrikkelen/adapter";
+export { FCC_BDC_ADAPTER_ID, FCC_BDC_DEFAULT_LICENSE, fccBdcAdapter } from "#us/adapters/fcc-bdc/adapter";
+export { GEONAMES_POSTAL_ADAPTER_ID, GEONAMES_POSTAL_DEFAULT_LICENSE, geonamesPostalAdapter, } from "#adapters/geonames/postal/adapter";
+export { GEONAMES_ADAPTER_ID, GEONAMES_DEFAULT_LICENSE, geonamesAdapter } from "#adapters/geonames/adapter";
+export { OPENADDRESSES_ADAPTER_ID, OPENADDRESSES_DEFAULT_LICENSE, openaddressesAdapter, } from "#adapters/openaddresses/adapter";
+export { STATE_HI_SCHOOLS_ADAPTER_ID, STATE_HI_SCHOOLS_DEFAULT_LICENSE, stateHiSchoolsAdapter, } from "#us/adapters/state/hi-schools/adapter";
+export { STATE_IA_CONTRACTORS_ADAPTER_ID, STATE_IA_CONTRACTORS_DEFAULT_LICENSE, stateIaContractorsAdapter, } from "#us/adapters/state/ia-contractors/adapter";
+export { STATE_NY_NOTARIES_ADAPTER_ID, STATE_NY_NOTARIES_DEFAULT_LICENSE, stateNyNotariesAdapter, } from "#us/adapters/state/ny-notaries/adapter";
+export { STATE_TX_NOTARIES_ADAPTER_ID, STATE_TX_NOTARIES_DEFAULT_LICENSE, stateTxNotariesAdapter, } from "#us/adapters/state/tx-notaries/adapter";
+export { TIGER_ADAPTER_ID, TIGER_DEFAULT_LICENSE, tigerAdapter } from "#us/adapters/tiger/adapter";
+export { USGOV_HRSA_FQHC_ADAPTER_ID, USGOV_HRSA_FQHC_DEFAULT_LICENSE, usgovHrsaFqhcAdapter, } from "#us/adapters/usgov/hrsa-fqhc/adapter";
+export { USGOV_IMLS_PLS_ADAPTER_ID, USGOV_IMLS_PLS_DEFAULT_LICENSE, usgovImlsPlsAdapter, } from "#us/adapters/usgov/imls-pls/adapter";
+export { USGOV_IRS_BMF_ADAPTER_ID, USGOV_IRS_BMF_DEFAULT_LICENSE, USGovIRSBMFAdapter, } from "#us/adapters/usgov/irs-bmf/adapter";
+export { USGOV_NAD_ADAPTER_ID, USGOV_NAD_DEFAULT_LICENSE, usgovNADAdapter } from "#us/adapters/usgov/nad/adapter";
+export { USGOV_NPPES_ADAPTER_ID, USGOV_NPPES_DEFAULT_LICENSE, usgovNPPESAdapter, } from "#us/adapters/usgov/nppes/adapter";
+export { USGOV_SAMHSA_ADAPTER_ID, USGOV_SAMHSA_DEFAULT_LICENSE, usgovSamhsaTreatmentLocatorAdapter, } from "#us/adapters/usgov/samhsa-treatment-locator/adapter";
+export { WOF_ADMIN_ADAPTER_ID, wofAdminAdapter } from "#adapters/wof/admin/json/adapter";
+export { WOF_POSTALCODE_ADAPTER_ID, wofPostalcodeAdapter } from "#adapters/wof/postalcode-json/adapter";
+//# sourceMappingURL=adapters.js.map

package/out/adapters.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapters.js","sourceRoot":"","sources":["../lib/adapters.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAuCG;AAEH,OAAO,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAA;AACxD,OAAO,EAAE,eAAe,EAAE,MAAM,4BAA4B,CAAA;AAC5D,OAAO,EAAE,qBAAqB,EAAE,MAAM,mCAAmC,CAAA;AACzE,OAAO,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAA;AACtD,OAAO,EAAE,oBAAoB,EAAE,MAAM,iCAAiC,CAAA;AACtE,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAA;AAClD,OAAO,EAAE,eAAe,EAAE,MAAM,4BAA4B,CAAA;AAC5D,OAAO,EAAE,sBAAsB,EAAE,MAAM,oBAAoB,CAAA;AAC3D,OAAO,EAAE,eAAe,EAAE,MAAM,kCAAkC,CAAA;AAClE,OAAO,EAAE,oBAAoB,EAAE,MAAM,uCAAuC,CAAA;AAC5E,OAAO,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAA;AACvD,OAAO,EAAE,eAAe,EAAE,MAAM,+BAA+B,CAAA;AAC/D,OAAO,EAAE,iBAAiB,EAAE,MAAM,iCAAiC,CAAA;AACnE,OAAO,EAAE,eAAe,EAAE,MAAM,+BAA+B,CAAA;AAC/D,OAAO,EAAE,aAAa,EAAE,MAAM,2BAA2B,CAAA;AACzD,OAAO,EAAE,gBAAgB,EAAE,MAAM,8BAA8B,CAAA;AAC/D,OAAO,EAAE,UAAU,EAAE,MAAM,0BAA0B,CAAA;AACrD,OAAO,EAAE,gBAAgB,EAAE,MAAM,8BAA8B,CAAA;AAC/D,OAAO,EAAE,iBAAiB,EAAE,MAAM,+BAA+B,CAAA;AACjE,OAAO,EAAE,iBAAiB,EAAE,MAAM,+BAA+B,CAAA;AACjE,OAAO,EAAE,gBAAgB,EAAE,MAAM,8BAA8B,CAAA;AAC/D,OAAO,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAA;AACzD,OAAO,EAAE,wBAAwB,EAAE,MAAM,yCAAyC,CAAA;AAClF,OAAO,EAAE,UAAU,EAAE,MAAM,0BAA0B,CAAA;AACrD,OAAO,EAAE,aAAa,EAAE,MAAM,6BAA6B,CAAA;AAC3D,OAAO,EAAE,aAAa,EAAE,MAAM,2BAA2B,CAAA;AACzD,OAAO,EAAE,iBAAiB,EAAE,MAAM,kCAAkC,CAAA;AACpE,OAAO,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAA;AACzD,OAAO,EAAE,iBAAiB,EAAE,MAAM,+BAA+B,CAAA;AACjE,OAAO,EAAE,kBAAkB,EAAE,MAAM,kCAAkC,CAAA;AACrE,OAAO,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAA;AACzD,OAAO,EAAE,mBAAmB,EAAE,MAAM,oCAAoC,CAAA;AACxE,OAAO,EAAE,gBAAgB,EAAE,MAAM,8BAA8B,CAAA;AAE/D,OAAO,EAAE,aAAa,EAAE,MAAM,8BAA8B,CAAA;AAC5D,OAAO,EAAE,qBAAqB,EAAE,MAAM,uCAAuC,CAAA;AAC7E,OAAO,EAAE,yBAAyB,EAAE,MAAM,2CAA2C,CAAA;AACrF,OAAO,EAAE,sBAAsB,EAAE,MAAM,wCAAwC,CAAA;AAC/E,OAAO,EAAE,sBAAsB,EAAE,MAAM,wCAAwC,CAAA;AAC/E,OAAO,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAA;AACzD,OAAO,EAAE,oBAAoB,EAAE,MAAM,sCAAsC,CAAA;AAC3E,OAAO,EAAE,mBAAmB,EAAE,MAAM,qCAAqC,CAAA;AACzE,OAAO,EAAE,kBAAkB,EAAE,MAAM,oCAAoC,CAAA;AACvE,OAAO,EAAE,eAAe,EAAE,MAAM,gCAAgC,CAAA;AAChE,OAAO,EAAE,iBAAiB,EAAE,MAAM,kCAAkC,CAAA;AAEpE;;;;;;;;;;;;GAYG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAA6B;IACzD,eAAe;IACf,oBAAoB;IACpB,eAAe;IACf,qBAAqB;IACrB,UAAU;IACV,kBAAkB;IAClB,YAAY;IACZ,aAAa;IACb,gBAAgB;IAChB,UAAU;IACV,YAAY;IACZ,iBAAiB;IACjB,eAAe;IACf,eAAe;IACf,iBAAiB;IACjB,iBAAiB;IACjB,gBAAgB;IAChB,iBAAiB;IACjB,gBAAgB;IAChB,iBAAiB;IACjB,gBAAgB;IAChB,YAAY;IACZ,oBAAoB;IACpB,eAAe;IACf,UAAU;IACV,WAAW;IACX,aAAa;IACb,oBAAoB;IACpB,iBAAiB;IACjB,eAAe;IACf,mBAAmB;IACnB,kBAAkB;IAClB,yBAAyB;IACzB,sBAAsB;IACtB,sBAAsB;IACtB,qBAAqB;IACrB,aAAa;IACb,mBAAmB;IACnB,aAAa;IACb,YAAY;IACZ,aAAa;IACb,wBAAwB;IACxB,YAAY;CACZ,CAAA;AAED,KAAK,MAAM,OAAO,IAAI,gBAAgB,EAAE,CAAC;IACxC,IAAI,CAAC,sBAAsB,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC,EAAE,CAAC;QAC7C,sBAAsB,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAA;IACzC,CAAC;AACF,CAAC;AAED,OAAO,EAAE,cAAc,EAAE,UAAU,EAAE,MAAM,0BAA0B,CAAA;AACrE,OAAO,EAAE,sBAAsB,EAAE,kBAAkB,EAAE,MAAM,kCAAkC,CAAA;AAC7F,OAAO,EAAE,kBAAkB,EAAE,uBAAuB,EAAE,aAAa,EAAE,MAAM,8BAA8B,CAAA;AAEzG,OAAO,EACN,0BAA0B,EAC1B,+BAA+B,EAC/B,qBAAqB,GACrB,MAAM,mCAAmC,CAAA;AAE1C,OAAO,EAAE,mBAAmB,EAAE,wBAAwB,EAAE,eAAe,EAAE,MAAM,4BAA4B,CAAA;AAE3G,OAAO,EACN,wBAAwB,EACxB,6BAA6B,EAC7B,oBAAoB,GACpB,MAAM,iCAAiC,CAAA;AAExC,OAAO,EACN,2BAA2B,EAC3B,gCAAgC,EAChC,qBAAqB,GACrB,MAAM,uCAAuC,CAAA;AAE9C,OAAO,EACN,+BAA+B,EAC/B,oCAAoC,EACpC,yBAAyB,GACzB,MAAM,2CAA2C,CAAA;AAElD,OAAO,EACN,4BAA4B,EAC5B,iCAAiC,EACjC,sBAAsB,GACtB,MAAM,wCAAwC,CAAA;AAE/C,OAAO,EACN,4BAA4B,EAC5B,iCAAiC,EACjC,sBAAsB,GACtB,MAAM,wCAAwC,CAAA;AAE/C,OAAO,EAAE,gBAAgB,EAAE,qBAAqB,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAA;AAElG,OAAO,EACN,0BAA0B,EAC1B,+BAA+B,EAC/B,oBAAoB,GACpB,MAAM,sCAAsC,CAAA;AAE7C,OAAO,EACN,yBAAyB,EACzB,8BAA8B,EAC9B,mBAAmB,GACnB,MAAM,qCAAqC,CAAA;AAE5C,OAAO,EACN,wBAAwB,EACxB,6BAA6B,EAC7B,kBAAkB,GAClB,MAAM,oCAAoC,CAAA;AAE3C,OAAO,EAAE,oBAAoB,EAAE,yBAAyB,EAAE,eAAe,EAAE,MAAM,gCAAgC,CAAA;AAEjH,OAAO,EACN,sBAAsB,EACtB,2BAA2B,EAC3B,iBAAiB,GACjB,MAAM,kCAAkC,CAAA;AAEzC,OAAO,EACN,uBAAuB,EACvB,4BAA4B,EAC5B,kCAAkC,GAClC,MAAM,qDAAqD,CAAA;AAE5D,OAAO,EAAE,oBAAoB,EAAE,eAAe,EAAE,MAAM,kCAAkC,CAAA;AACxF,OAAO,EAAE,yBAAyB,EAAE,oBAAoB,EAAE,MAAM,uCAAuC,CAAA"}

package/out/au/adapters/gnaf/adapter.d.ts

@@ -0,0 +1,37 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * G-NAF (Australia) corpus adapter.
+ *
+ * The model mis-parses Australian addresses in their native postcode-first and house-number-last
+ * order, where it tags a leading four-digit postcode as a house number and swaps street against
+ * locality. This adapter renders each assembled G-NAF tuple (see {@link ./assemble}) in three real
+ * AU layouts, rotated by row index, so the locality and postcode each land in every position across
+ * the corpus. Input is the assembled component jsonl, one tuple per line. Open G-NAF license
+ * requires attribution to "Geoscape Australia".
+ */
+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.
+ */
+export declare const GNAF_ADAPTER_ID = "gnaf";
+/**
+ * Open G-NAF is freely redistributable with attribution to Geoscape Australia (CC-BY-style).
+ */
+export declare const GNAF_DEFAULT_LICENSE = "CC-BY-4.0";
+/**
+ * Build the G-NAF adapter.
+ *
+ * `inputPath` is the assembled component jsonl (see {@link ./assemble}); it is
+ * country-pinned to AU regardless of `opts.country` (G-NAF is Australia-only).
+ */
+export declare function createGNAFAdapter(): CorpusAdapter;
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export declare const gnafAdapter: CorpusAdapter;
+//# sourceMappingURL=adapter.d.ts.map

package/out/au/adapters/gnaf/adapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/au/adapters/gnaf/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAYH,OAAO,EAAuD,KAAK,aAAa,EAAiB,MAAM,QAAQ,CAAA;AAE/G;;;;GAIG;AACH,eAAO,MAAM,eAAe,SAAS,CAAA;AACrC;;GAEG;AACH,eAAO,MAAM,oBAAoB,cAAc,CAAA;AA8B/C;;;;;GAKG;AACH,wBAAgB,iBAAiB,IAAI,aAAa,CAkEjD;AAED;;GAEG;AACH,eAAO,MAAM,WAAW,eAAsB,CAAA"}

package/out/au/adapters/gnaf/adapter.js

@@ -0,0 +1,119 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * G-NAF (Australia) corpus adapter.
+ *
+ * The model mis-parses Australian addresses in their native postcode-first and house-number-last
+ * order, where it tags a leading four-digit postcode as a house number and swaps street against
+ * locality. This adapter renders each assembled G-NAF tuple (see {@link ./assemble}) in three real
+ * AU layouts, rotated by row index, so the locality and postcode each land in every position across
+ * the corpus. Input is the assembled component jsonl, one tuple per line. Open G-NAF license
+ * requires attribution to "Geoscape Australia".
+ */
+/* oxlint-disable mailwoman/prefer-home -- this adapter reads Australia's national register and writes AU surfaces
+   only. One of the forms below is a deliberate malformation. The postcode-first order is the dominant failure this
+   source exists to teach against — and a renderer that produces well-formed addresses cannot express it. */
+import { componentsPresentIn } from "@mailwoman/codex/address/format";
+import { tryParsingJSON } from "@mailwoman/core/json";
+import { TextSpliterator } from "spliterator";
+import { stableSourceID } from "#adapters/source-id";
+import { SourceRegister } from "#registers";
+import { AddressRole, 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 GNAF_ADAPTER_ID = "gnaf";
+/**
+ * Open G-NAF is freely redistributable with attribution to Geoscape Australia (CC-BY-style).
+ */
+export const GNAF_DEFAULT_LICENSE = "CC-BY-4.0";
+/**
+ * The address layouts an AU address actually arrives in.
+ *
+ * The model already handles postcode-trailing (canonical); the two postcode-leading
+ * forms are the ones it fails, so they show the change.
+ * We keep the canonical form too so the retrain doesn't forget it.
+ */
+function renderOrders(c) {
+    const region = c.region ? ` ${c.region}` : "";
+    return [
+        // real-AU canonical: number-first, street, suburb [state] postcode — "50 Barry Street, Carlton NSW 2000"
+        `${c.house_number} ${c.street}, ${c.locality}${region} ${c.postcode}`,
+        // postcode-first (the dominant failure): "2000 Carlton, Barry Street 50"
+        `${c.postcode} ${c.locality}, ${c.street} ${c.house_number}`,
+        // locality-first: "Carlton, 2000, Barry Street 50"
+        `${c.locality}, ${c.postcode}, ${c.street} ${c.house_number}`,
+    ];
+}
+/**
+ * Build the G-NAF adapter.
+ *
+ * `inputPath` is the assembled component jsonl (see {@link ./assemble}); it is
+ * country-pinned to AU regardless of `opts.country` (G-NAF is Australia-only).
+ */
+export function createGNAFAdapter() {
+    return {
+        id: GNAF_ADAPTER_ID,
+        defaultLicense: GNAF_DEFAULT_LICENSE,
+        addressRole: AddressRole.Premise,
+        register: SourceRegister.GNAF,
+        surface: SurfaceOrigin.Rendered,
+        description: "G-NAF (Australia): assembled address tuples rendered in multiple word orders (canonical / postcode-first / locality-first) — teaches the model AU's postcode-first layout.",
+        async *rows(opts) {
+            let emitted = 0;
+            let idx = 0;
+            // TextSpliterator auto-disposes on loop completion and on an early `break` (abort / limit).
+            for await (const line of TextSpliterator.fromAsync(opts.inputPath)) {
+                if (opts.signal?.aborted)
+                    break;
+                if (opts.limit !== undefined && emitted >= opts.limit)
+                    break;
+                const t = tryParsingJSON(line);
+                if (t === null)
+                    continue;
+                if (!t.house_number || !t.street || !t.locality || !t.postcode)
+                    continue;
+                const orders = renderOrders(t);
+                const order = idx % orders.length;
+                idx++;
+                const raw = orders[order];
+                const components = {
+                    house_number: t.house_number,
+                    street: t.street,
+                    locality: t.locality,
+                    postcode: t.postcode,
+                };
+                // `region` appears only in the canonical render.
+                // The postcode-leading layouts omit it so verbatim alignment never breaks.
+                if (order === 0 && t.region) {
+                    components.region = t.region;
+                }
+                // `raw` is one of three deliberate word orders, two of which no layout prints.
+                // Containment is therefore checked against the string this adapter built.
+                const aligned = componentsPresentIn(components, raw);
+                if (!Object.keys(aligned).length)
+                    continue;
+                yield {
+                    raw,
+                    components: aligned,
+                    country: "AU",
+                    locale: "en-AU",
+                    source: GNAF_ADAPTER_ID,
+                    source_id: `${stableSourceID(GNAF_ADAPTER_ID, aligned)}-o${order}`,
+                    corpus_version: "",
+                    license: GNAF_DEFAULT_LICENSE,
+                };
+                emitted++;
+            }
+        },
+    };
+}
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export const gnafAdapter = createGNAFAdapter();
+//# sourceMappingURL=adapter.js.map

package/out/au/adapters/gnaf/adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../lib/au/adapters/gnaf/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH;;4GAE4G;AAE5G,OAAO,EAAE,mBAAmB,EAAE,MAAM,iCAAiC,CAAA;AACrE,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAA;AACrD,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAE7C,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAA;AACpD,OAAO,EAAE,cAAc,EAAE,MAAM,YAAY,CAAA;AAC3C,OAAO,EAAE,WAAW,EAA8D,aAAa,EAAE,MAAM,QAAQ,CAAA;AAE/G;;;;GAIG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,MAAM,CAAA;AACrC;;GAEG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAG,WAAW,CAAA;AAU/C;;;;;;GAMG;AACH,SAAS,YAAY,CAAC,CAAY;IACjC,MAAM,MAAM,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,EAAE,CAAA;IAE7C,OAAO;QACN,yGAAyG;QACzG,GAAG,CAAC,CAAC,YAAY,IAAI,CAAC,CAAC,MAAM,KAAK,CAAC,CAAC,QAAQ,GAAG,MAAM,IAAI,CAAC,CAAC,QAAQ,EAAE;QACrE,yEAAyE;QACzE,GAAG,CAAC,CAAC,QAAQ,IAAI,CAAC,CAAC,QAAQ,KAAK,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,YAAY,EAAE;QAC5D,mDAAmD;QACnD,GAAG,CAAC,CAAC,QAAQ,KAAK,CAAC,CAAC,QAAQ,KAAK,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,YAAY,EAAE;KAC7D,CAAA;AACF,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,iBAAiB;IAChC,OAAO;QACN,EAAE,EAAE,eAAe;QACnB,cAAc,EAAE,oBAAoB;QACpC,WAAW,EAAE,WAAW,CAAC,OAAO;QAChC,QAAQ,EAAE,cAAc,CAAC,IAAI;QAC7B,OAAO,EAAE,aAAa,CAAC,QAAQ;QAC/B,WAAW,EACV,4KAA4K;QAE7K,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,IAAI,OAAO,GAAG,CAAC,CAAA;YACf,IAAI,GAAG,GAAG,CAAC,CAAA;YAEX,4FAA4F;YAC5F,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,eAAe,CAAC,SAAS,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,CAAC;gBACpE,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,CAAC,GAAG,cAAc,CAAY,IAAI,CAAC,CAAA;gBAEzC,IAAI,CAAC,KAAK,IAAI;oBAAE,SAAQ;gBAExB,IAAI,CAAC,CAAC,CAAC,YAAY,IAAI,CAAC,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,CAAC,QAAQ,IAAI,CAAC,CAAC,CAAC,QAAQ;oBAAE,SAAQ;gBAExE,MAAM,MAAM,GAAG,YAAY,CAAC,CAAC,CAAC,CAAA;gBAC9B,MAAM,KAAK,GAAG,GAAG,GAAG,MAAM,CAAC,MAAM,CAAA;gBAEjC,GAAG,EAAE,CAAA;gBACL,MAAM,GAAG,GAAG,MAAM,CAAC,KAAK,CAAE,CAAA;gBAE1B,MAAM,UAAU,GAA+B;oBAC9C,YAAY,EAAE,CAAC,CAAC,YAAY;oBAC5B,MAAM,EAAE,CAAC,CAAC,MAAM;oBAChB,QAAQ,EAAE,CAAC,CAAC,QAAQ;oBACpB,QAAQ,EAAE,CAAC,CAAC,QAAQ;iBACpB,CAAA;gBAED,iDAAiD;gBACjD,2EAA2E;gBAC3E,IAAI,KAAK,KAAK,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,CAAC;oBAC7B,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,MAAM,CAAA;gBAC7B,CAAC;gBAED,+EAA+E;gBAC/E,0EAA0E;gBAC1E,MAAM,OAAO,GAAG,mBAAmB,CAAC,UAAU,EAAE,GAAG,CAAC,CAAA;gBAEpD,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM;oBAAE,SAAQ;gBAE1C,MAAM;oBACL,GAAG;oBACH,UAAU,EAAE,OAAO;oBACnB,OAAO,EAAE,IAAI;oBACb,MAAM,EAAE,OAAO;oBACf,MAAM,EAAE,eAAe;oBACvB,SAAS,EAAE,GAAG,cAAc,CAAC,eAAe,EAAE,OAAO,CAAC,KAAK,KAAK,EAAE;oBAClE,cAAc,EAAE,EAAE;oBAClB,OAAO,EAAE,oBAAoB;iBAC7B,CAAA;gBAED,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,iBAAiB,EAAE,CAAA"}