@mailwoman/corpus 9.4.0 → 10.1.0

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

@@ -1,197 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `wof-admin`: Who's On First admin GeoJSON-bundle adapter.
- *
- *   **Phase 1.5.1 pivot.** The original Phase 1.5 SQLite adapter (formerly at
- *   `packages/corpus/lib/adapters/wof-admin/`, removed in this same change) was replaced by this
- *   one because the SQLite distribution path was unworkable for the real corpus build:
- *
- *   1. `dist.whosonfirst.org/sqlite/` is dead (NXDOMAIN); the Geocode-Earth mirror is the only one.
- *   2. The Geocode-Earth-hosted postalcode DB tags every row `mz:is_current = -1` ("unknown but treated
- *        as active"); the SQLite adapter's `is_current = 1` predicate emitted zero rows.
- *   3. The `names` table in the SQLite distribution is empty — localized `name:*` variants live in a
- *        separate distribution. The St. Petersburg / Mt. Vernon / Ft. Lauderdale alternation cases
- *        (the original Phase 1.5.1 motivator) cannot be solved on the SQLite path even with a
- *        patched `is_current` predicate.
- *
- *   Input: a directory containing one or more cloned `whosonfirst-data-admin-<cc>` GitHub repos. Each
- *   repo has `data/XXX/YYY/ZZZ/<wof-id>.geojson` files; `**\/*.geojson` walks the tree recursively.
- *   Alternate-geometry siblings (`-alt-*`) are skipped — they're separate exports of the same
- *   record, not new records.
- *
- *   Per record, the adapter emits one row per `(name-variant, hierarchy-variant)` pair:
- *
- *   - **Name variants**: the canonical `wof:name` (slot key `default`) plus every `name:*` localized
- *       variant present on the feature (`name:eng_x_preferred`, `name:eng_x_colloquial`,
- *       `name:rus_x_preferred`, ...). This is the Phase 1.5.1 fix for the St. Petersburg case:
- *       `"Saint Petersburg"` (canonical) and `"St. Petersburg"` (eng_x_colloquial) both become
- *       training rows for the same WOF id.
- *   - **Hierarchy variants** (unchanged from the SQLite adapter): locality → 3 variants, region → 2,
- *       country → 1, county → 1.
- *
- *   `source_id` is `wof-admin-<wof_id>-<name-slot>-<hierarchy-variant>`. The previous SQLite adapter
- *   used `wof-admin-<wof_id>-<hierarchy-variant>` (no name slot); the new format adds a name-slot
- *   segment so the colloquial / preferred / per-locale variants survive dedup independently.
- *
- *   License: CC0. The adapter stamps every row with `CC0-1.0`.
- */
-import { COUNTRY_DISPLAY_NAME, emitWOFJSONRows, LOCALE_BY_COUNTRY, nameSlotsFor as wofNameSlotsFor, } from "#adapters/wof-json-rows";
-import { buildAncestryIndex, walkFeatures } from "#utils";
-/**
- * Map a WOF placetype to a Mailwoman `ComponentTag`, or `undefined` to skip.
- *
- * Per-adapter deliberately (the postalcode adapter carries its own): each table is a record FILTER for its adapter's
- * emission set, not a shared vocabulary.
- */
-function placetypeToTag(placetype) {
-    switch (placetype) {
-        case "country":
-        case "nation":
-            return "country";
-        case "macroregion":
-        case "region":
-            return "region";
-        case "macrocounty":
-        case "county":
-        case "localadmin":
-            return "subregion";
-        case "locality":
-            return "locality";
-        case "borough":
-        case "macrohood":
-        case "neighbourhood":
-        case "microhood":
-            return "dependent_locality";
-        default:
-            return undefined;
-    }
-}
-/**
- * Compute the hierarchy variants for a record given its ancestry chain and the chosen `selfName`.
- *
- * `selfName` is the surface form to use for the record's own component (locality / region / country / subregion).
- * Callers pass the canonical `wof:name` for the `"default"` slot and a `name:*` localized value for variant slots;
- * ancestor names always come from the ancestor's canonical `wof:name`.
- *
- * Country variants substitute `COUNTRY_DISPLAY_NAME` for the default slot so the OpenCage template produces the
- * canonicalized form (`"United States of America"`), matching the legacy SQLite adapter's behavior.
- */
-export function variantsFor(row, ancestry, selfName) {
-    const selfTag = placetypeToTag(row.placetype);
-    if (!selfTag)
-        return [];
-    const region = ancestry.find((a) => placetypeToTag(a.placetype) === "region");
-    const country = ancestry.find((a) => placetypeToTag(a.placetype) === "country");
-    const countryDisplay = COUNTRY_DISPLAY_NAME[row.country] ?? country?.name ?? row.country;
-    const variants = [];
-    switch (selfTag) {
-        case "locality":
-        case "dependent_locality": {
-            variants.push({ suffix: "self", components: { [selfTag]: selfName } });
-            if (region) {
-                variants.push({
-                    suffix: "with-region",
-                    components: { [selfTag]: selfName, region: region.name },
-                });
-            }
-            if (region && country) {
-                variants.push({
-                    suffix: "with-region-country",
-                    components: { [selfTag]: selfName, region: region.name, country: countryDisplay },
-                });
-            }
-            else if (!region && country) {
-                variants.push({
-                    suffix: "with-country",
-                    components: { [selfTag]: selfName, country: countryDisplay },
-                });
-            }
-            return variants;
-        }
-        case "region": {
-            variants.push({ suffix: "self", components: { region: selfName } });
-            if (country) {
-                variants.push({
-                    suffix: "with-country",
-                    components: { region: selfName, country: countryDisplay },
-                });
-            }
-            return variants;
-        }
-        case "country": {
-            variants.push({ suffix: "self", components: { country: selfName } });
-            return variants;
-        }
-        case "subregion": {
-            variants.push({ suffix: "self", components: { subregion: selfName } });
-            return variants;
-        }
-        default:
-            return [];
-    }
-}
-/**
- * Build the per-record name-slot list. The canonical `"default"` slot uses the OpenCage-canonical country form when the
- * record is itself a country (matches SQLite-adapter behavior); every other placetype's default slot uses `wof:name`
- * verbatim.
- */
-export function nameSlotsFor(rec) {
-    return wofNameSlotsFor(rec, {
-        canonicalName: (record) => placetypeToTag(record.placetype) === "country"
-            ? (COUNTRY_DISPLAY_NAME[record.country] ?? record.name)
-            : record.name,
-    });
-}
-/**
- * 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 WOF_ADMIN_ADAPTER_ID = "wof-admin";
-/**
- * Construct the wof-admin JSON-bundle adapter. The adapter is stateless across runs; calling this twice with the same
- * input directory produces byte-identical `canonical.jsonl` (records are emitted in sorted `wof:id` order to be
- * insensitive to filesystem walk ordering).
- */
-export function createWOFAdminAdapter() {
-    return {
-        id: WOF_ADMIN_ADAPTER_ID,
-        defaultLicense: "CC0-1.0",
-        description: "Who's On First admin GeoJSON bundles (countries, regions, counties, localities) — multi-name variants per record.",
-        async *rows(opts) {
-            // Pass 1: scan every GeoJSON file once, build the in-memory record index.
-            // We keep only records whose placetype maps to a ComponentTag — irrelevant placetypes
-            // (campus, county-region hybrids on which Mailwoman has no opinion) are dropped here so
-            // they don't inflate the ancestry index. Country-filtered runs prune to the matching
-            // country code too; the ancestors of a same-country record live in the same admin repo.
-            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 = buildAncestryIndex(byID);
-            // Pass 2: emit rows in sorted-id order for deterministic JSONL.
-            yield* emitWOFJSONRows({
-                records: byID,
-                ancestry,
-                adapterOptions: opts,
-                adapterID: WOF_ADMIN_ADAPTER_ID,
-                localeByCountry: LOCALE_BY_COUNTRY,
-                nameSlotsFor,
-                variantsFor,
-            });
-        },
-    };
-}
-/**
- * Single shared instance, suitable for `defaultAdapterRegistry`.
- */
-export const wofAdminAdapter = createWOFAdminAdapter();
-//# sourceMappingURL=adapter.js.map

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

@@ -1 +0,0 @@
-{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../lib/adapters/wof-admin-json/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAuCG;AAKH,OAAO,EACN,oBAAoB,EACpB,eAAe,EACf,iBAAiB,EACjB,YAAY,IAAI,eAAe,GAE/B,MAAM,yBAAyB,CAAA;AAEhC,OAAO,EAAE,kBAAkB,EAAE,YAAY,EAAkB,MAAM,QAAQ,CAAA;AAEzE;;;;;GAKG;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,aAAa,CAAC;QACnB,KAAK,QAAQ,CAAC;QACd,KAAK,YAAY;YAChB,OAAO,WAAW,CAAA;QACnB,KAAK,UAAU;YACd,OAAO,UAAU,CAAA;QAClB,KAAK,SAAS,CAAC;QACf,KAAK,WAAW,CAAC;QACjB,KAAK,eAAe,CAAC;QACrB,KAAK,WAAW;YACf,OAAO,oBAAoB,CAAA;QAC5B;YACC,OAAO,SAAS,CAAA;IAClB,CAAC;AACF,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,UAAU,WAAW,CAAC,GAAc,EAAE,QAAqB,EAAE,QAAgB;IAClF,MAAM,OAAO,GAAG,cAAc,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;IAE7C,IAAI,CAAC,OAAO;QAAE,OAAO,EAAE,CAAA;IAEvB,MAAM,MAAM,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,cAAc,CAAC,CAAC,CAAC,SAAS,CAAC,KAAK,QAAQ,CAAC,CAAA;IAC7E,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,cAAc,CAAC,CAAC,CAAC,SAAS,CAAC,KAAK,SAAS,CAAC,CAAA;IAC/E,MAAM,cAAc,GAAG,oBAAoB,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,OAAO,EAAE,IAAI,IAAI,GAAG,CAAC,OAAO,CAAA;IAExF,MAAM,QAAQ,GAAqB,EAAE,CAAA;IAErC,QAAQ,OAAO,EAAE,CAAC;QACjB,KAAK,UAAU,CAAC;QAChB,KAAK,oBAAoB,CAAC,CAAC,CAAC;YAC3B,QAAQ,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,EAAE,CAAC,OAAO,CAAC,EAAE,QAAQ,EAAE,EAAE,CAAC,CAAA;YAEtE,IAAI,MAAM,EAAE,CAAC;gBACZ,QAAQ,CAAC,IAAI,CAAC;oBACb,MAAM,EAAE,aAAa;oBACrB,UAAU,EAAE,EAAE,CAAC,OAAO,CAAC,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,CAAC,IAAI,EAAE;iBACxD,CAAC,CAAA;YACH,CAAC;YAED,IAAI,MAAM,IAAI,OAAO,EAAE,CAAC;gBACvB,QAAQ,CAAC,IAAI,CAAC;oBACb,MAAM,EAAE,qBAAqB;oBAC7B,UAAU,EAAE,EAAE,CAAC,OAAO,CAAC,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,CAAC,IAAI,EAAE,OAAO,EAAE,cAAc,EAAE;iBACjF,CAAC,CAAA;YACH,CAAC;iBAAM,IAAI,CAAC,MAAM,IAAI,OAAO,EAAE,CAAC;gBAC/B,QAAQ,CAAC,IAAI,CAAC;oBACb,MAAM,EAAE,cAAc;oBACtB,UAAU,EAAE,EAAE,CAAC,OAAO,CAAC,EAAE,QAAQ,EAAE,OAAO,EAAE,cAAc,EAAE;iBAC5D,CAAC,CAAA;YACH,CAAC;YAED,OAAO,QAAQ,CAAA;QAChB,CAAC;QAED,KAAK,QAAQ,CAAC,CAAC,CAAC;YACf,QAAQ,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,EAAE,MAAM,EAAE,QAAQ,EAAE,EAAE,CAAC,CAAA;YAEnE,IAAI,OAAO,EAAE,CAAC;gBACb,QAAQ,CAAC,IAAI,CAAC;oBACb,MAAM,EAAE,cAAc;oBACtB,UAAU,EAAE,EAAE,MAAM,EAAE,QAAQ,EAAE,OAAO,EAAE,cAAc,EAAE;iBACzD,CAAC,CAAA;YACH,CAAC;YAED,OAAO,QAAQ,CAAA;QAChB,CAAC;QAED,KAAK,SAAS,CAAC,CAAC,CAAC;YAChB,QAAQ,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,EAAE,OAAO,EAAE,QAAQ,EAAE,EAAE,CAAC,CAAA;YAEpE,OAAO,QAAQ,CAAA;QAChB,CAAC;QAED,KAAK,WAAW,CAAC,CAAC,CAAC;YAClB,QAAQ,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,EAAE,SAAS,EAAE,QAAQ,EAAE,EAAE,CAAC,CAAA;YAEtE,OAAO,QAAQ,CAAA;QAChB,CAAC;QAED;YACC,OAAO,EAAE,CAAA;IACX,CAAC;AACF,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,YAAY,CAAC,GAAc;IAC1C,OAAO,eAAe,CAAC,GAAG,EAAE;QAC3B,aAAa,EAAE,CAAC,MAAM,EAAE,EAAE,CACzB,cAAc,CAAC,MAAM,CAAC,SAAS,CAAC,KAAK,SAAS;YAC7C,CAAC,CAAC,CAAC,oBAAoB,CAAC,MAAM,CAAC,OAAO,CAAC,IAAI,MAAM,CAAC,IAAI,CAAC;YACvD,CAAC,CAAC,MAAM,CAAC,IAAI;KACf,CAAC,CAAA;AACH,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAG,WAAW,CAAA;AAE/C;;;;GAIG;AACH,MAAM,UAAU,qBAAqB;IACpC,OAAO;QACN,EAAE,EAAE,oBAAoB;QACxB,cAAc,EAAE,SAAS;QACzB,WAAW,EACV,mHAAmH;QAEpH,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,0EAA0E;YAC1E,sFAAsF;YACtF,wFAAwF;YACxF,qFAAqF;YACrF,wFAAwF;YACxF,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,kBAAkB,CAAC,IAAI,CAAC,CAAA;YAEzC,gEAAgE;YAChE,KAAK,CAAC,CAAC,eAAe,CAAC;gBACtB,OAAO,EAAE,IAAI;gBACb,QAAQ;gBACR,cAAc,EAAE,IAAI;gBACpB,SAAS,EAAE,oBAAoB;gBAC/B,eAAe,EAAE,iBAAiB;gBAClC,YAAY;gBACZ,WAAW;aACX,CAAC,CAAA;QACH,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,qBAAqB,EAAE,CAAA"}

package/out/adapters/wof-json-rows.d.ts

@@ -1,62 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- * @file Deterministic canonical-row emission shared by WOF GeoJSON adapters.
- */
-import type { ComponentTag } from "@mailwoman/core/types";
-import type { AdapterOptions, CanonicalRow } from "#types";
-import { type WOFRecord } from "#utils";
-/**
- * Display name for the country, keyed by ISO 3166-1 alpha-2.
- *
- * Must be the **OpenCage-canonical** surface form: the `address-formatter` library expands some country names en route
- * to its output (e.g. `"United States"` → `"United States of America"`). If `components.country` and the formatted
- * `raw` disagree, alignment will fail downstream. Keying off the canonical form keeps the two in lockstep.
- *
- * Phase 1 US + FR only; extend as new locales come online. Missing countries fall back to the country row's `wof:name`,
- * accepting the alignment risk for non-canonicalized names.
- */
-export declare const COUNTRY_DISPLAY_NAME: Record<string, string>;
-/**
- * BCP-47 locale defaulting for the corpus row's `locale` field. Defaulted by country.
- */
-export declare const LOCALE_BY_COUNTRY: Record<string, string>;
-export interface WOFVariantSpec {
-    suffix: string;
-    components: Partial<Record<ComponentTag, string>>;
-}
-export interface NameSlotOptions {
-    /**
-     * Canonical surface for the record's own `"default"` slot. Default `rec.name` verbatim; the admin adapter substitutes
-     * the OpenCage-canonical {@link COUNTRY_DISPLAY_NAME} for country records.
-     */
-    canonicalName?: (rec: WOFRecord) => string;
-}
-/**
- * Build the per-record name-slot list: the canonical `"default"` slot, then every `name:*` variant deduplicated against
- * it so a redundant `"default"`-equivalent row is not emitted under a localized key.
- */
-export declare function nameSlotsFor(rec: WOFRecord, options?: NameSlotOptions): Array<{
-    key: string;
-    value: string;
-}>;
-interface EmitWOFJSONRowsOptions {
-    records: ReadonlyMap<number, WOFRecord>;
-    ancestry: ReadonlyMap<number, WOFRecord[]>;
-    adapterOptions: AdapterOptions;
-    adapterID: string;
-    localeByCountry: Readonly<Record<string, string>>;
-    shouldEmit?: (record: WOFRecord) => boolean;
-    nameSlotsFor: (record: WOFRecord) => Array<{
-        key: string;
-        value: string;
-    }>;
-    variantsFor: (record: WOFRecord, ancestry: WOFRecord[], selfName: string) => WOFVariantSpec[];
-}
-/**
- * Emit aligned rows in WOF-id order, enforcing the adapter limit across all name and hierarchy variants.
- */
-export declare function emitWOFJSONRows(options: EmitWOFJSONRowsOptions): Generator<CanonicalRow>;
-export {};
-//# sourceMappingURL=wof-json-rows.d.ts.map

package/out/adapters/wof-json-rows.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"wof-json-rows.d.ts","sourceRoot":"","sources":["../../lib/adapters/wof-json-rows.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,uBAAuB,CAAA;AAGzD,OAAO,KAAK,EAAE,cAAc,EAAE,YAAY,EAAE,MAAM,QAAQ,CAAA;AAC1D,OAAO,EAAoB,KAAK,SAAS,EAAE,MAAM,QAAQ,CAAA;AAEzD;;;;;;;;;GASG;AACH,eAAO,MAAM,oBAAoB,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAGvD,CAAA;AAED;;GAEG;AACH,eAAO,MAAM,iBAAiB,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAGpD,CAAA;AAED,MAAM,WAAW,cAAc;IAC9B,MAAM,EAAE,MAAM,CAAA;IACd,UAAU,EAAE,OAAO,CAAC,MAAM,CAAC,YAAY,EAAE,MAAM,CAAC,CAAC,CAAA;CACjD;AAED,MAAM,WAAW,eAAe;IAC/B;;;OAGG;IACH,aAAa,CAAC,EAAE,CAAC,GAAG,EAAE,SAAS,KAAK,MAAM,CAAA;CAC1C;AAED;;;GAGG;AACH,wBAAgB,YAAY,CAAC,GAAG,EAAE,SAAS,EAAE,OAAO,GAAE,eAAoB,GAAG,KAAK,CAAC;IAAE,GAAG,EAAE,MAAM,CAAC;IAAC,KAAK,EAAE,MAAM,CAAA;CAAE,CAAC,CAajH;AAED,UAAU,sBAAsB;IAC/B,OAAO,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,CAAC,CAAA;IACvC,QAAQ,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,EAAE,CAAC,CAAA;IAC1C,cAAc,EAAE,cAAc,CAAA;IAC9B,SAAS,EAAE,MAAM,CAAA;IACjB,eAAe,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;IACjD,UAAU,CAAC,EAAE,CAAC,MAAM,EAAE,SAAS,KAAK,OAAO,CAAA;IAC3C,YAAY,EAAE,CAAC,MAAM,EAAE,SAAS,KAAK,KAAK,CAAC;QAAE,GAAG,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAA;KAAE,CAAC,CAAA;IAC1E,WAAW,EAAE,CAAC,MAAM,EAAE,SAAS,EAAE,QAAQ,EAAE,SAAS,EAAE,EAAE,QAAQ,EAAE,MAAM,KAAK,cAAc,EAAE,CAAA;CAC7F;AAED;;GAEG;AACH,wBAAiB,eAAe,CAAC,OAAO,EAAE,sBAAsB,GAAG,SAAS,CAAC,YAAY,CAAC,CA8CzF"}

package/out/adapters/wof-json-rows.js

@@ -1,83 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- * @file Deterministic canonical-row emission shared by WOF GeoJSON adapters.
- */
-import { formatAddress, reconcileComponents } from "@mailwoman/formatter";
-import { normalizeNameKey } from "#utils";
-/**
- * Display name for the country, keyed by ISO 3166-1 alpha-2.
- *
- * Must be the **OpenCage-canonical** surface form: the `address-formatter` library expands some country names en route
- * to its output (e.g. `"United States"` → `"United States of America"`). If `components.country` and the formatted
- * `raw` disagree, alignment will fail downstream. Keying off the canonical form keeps the two in lockstep.
- *
- * Phase 1 US + FR only; extend as new locales come online. Missing countries fall back to the country row's `wof:name`,
- * accepting the alignment risk for non-canonicalized names.
- */
-export const COUNTRY_DISPLAY_NAME = {
-    US: "United States of America",
-    FR: "France",
-};
-/**
- * BCP-47 locale defaulting for the corpus row's `locale` field. Defaulted by country.
- */
-export const LOCALE_BY_COUNTRY = {
-    US: "en-US",
-    FR: "fr-FR",
-};
-/**
- * Build the per-record name-slot list: the canonical `"default"` slot, then every `name:*` variant deduplicated against
- * it so a redundant `"default"`-equivalent row is not emitted under a localized key.
- */
-export function nameSlotsFor(rec, options = {}) {
-    const canonicalSelfName = options.canonicalName?.(rec) ?? rec.name;
-    const seen = new Set([canonicalSelfName]);
-    const slots = [{ key: "default", value: canonicalSelfName }];
-    for (const [rawKey, value] of rec.nameVariants) {
-        if (seen.has(value))
-            continue;
-        seen.add(value);
-        slots.push({ key: normalizeNameKey(rawKey), value });
-    }
-    return slots;
-}
-/**
- * Emit aligned rows in WOF-id order, enforcing the adapter limit across all name and hierarchy variants.
- */
-export function* emitWOFJSONRows(options) {
-    const { records, ancestry, adapterOptions, adapterID, localeByCountry, shouldEmit = () => true, nameSlotsFor: slotsForRecord, variantsFor, } = options;
-    let emitted = 0;
-    for (const id of [...records.keys()].toSorted((a, b) => a - b)) {
-        if (adapterOptions.signal?.aborted)
-            return;
-        const record = records.get(id);
-        if (!shouldEmit(record))
-            continue;
-        for (const slot of slotsForRecord(record)) {
-            for (const variant of variantsFor(record, ancestry.get(id) ?? [], slot.value)) {
-                if (adapterOptions.limit !== undefined && emitted >= adapterOptions.limit)
-                    return;
-                const raw = formatAddress(variant.components, record.country, { separator: ", " });
-                if (!raw)
-                    continue;
-                const components = reconcileComponents(variant.components, raw);
-                if (!Object.keys(components).length)
-                    continue;
-                yield {
-                    raw,
-                    components,
-                    country: record.country,
-                    locale: localeByCountry[record.country],
-                    source: adapterID,
-                    source_id: `${adapterID}-${record.id}-${slot.key}-${variant.suffix}`,
-                    corpus_version: "",
-                    license: "CC0-1.0",
-                };
-                emitted++;
-            }
-        }
-    }
-}
-//# sourceMappingURL=wof-json-rows.js.map

package/out/adapters/wof-json-rows.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"wof-json-rows.js","sourceRoot":"","sources":["../../lib/adapters/wof-json-rows.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAGH,OAAO,EAAE,aAAa,EAAE,mBAAmB,EAAE,MAAM,sBAAsB,CAAA;AAGzE,OAAO,EAAE,gBAAgB,EAAkB,MAAM,QAAQ,CAAA;AAEzD;;;;;;;;;GASG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAA2B;IAC3D,EAAE,EAAE,0BAA0B;IAC9B,EAAE,EAAE,QAAQ;CACZ,CAAA;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAA2B;IACxD,EAAE,EAAE,OAAO;IACX,EAAE,EAAE,OAAO;CACX,CAAA;AAeD;;;GAGG;AACH,MAAM,UAAU,YAAY,CAAC,GAAc,EAAE,UAA2B,EAAE;IACzE,MAAM,iBAAiB,GAAG,OAAO,CAAC,aAAa,EAAE,CAAC,GAAG,CAAC,IAAI,GAAG,CAAC,IAAI,CAAA;IAElE,MAAM,IAAI,GAAG,IAAI,GAAG,CAAS,CAAC,iBAAiB,CAAC,CAAC,CAAA;IACjD,MAAM,KAAK,GAA0C,CAAC,EAAE,GAAG,EAAE,SAAS,EAAE,KAAK,EAAE,iBAAiB,EAAE,CAAC,CAAA;IAEnG,KAAK,MAAM,CAAC,MAAM,EAAE,KAAK,CAAC,IAAI,GAAG,CAAC,YAAY,EAAE,CAAC;QAChD,IAAI,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC;YAAE,SAAQ;QAC7B,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAA;QACf,KAAK,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,gBAAgB,CAAC,MAAM,CAAC,EAAE,KAAK,EAAE,CAAC,CAAA;IACrD,CAAC;IAED,OAAO,KAAK,CAAA;AACb,CAAC;AAaD;;GAEG;AACH,MAAM,SAAS,CAAC,CAAC,eAAe,CAAC,OAA+B;IAC/D,MAAM,EACL,OAAO,EACP,QAAQ,EACR,cAAc,EACd,SAAS,EACT,eAAe,EACf,UAAU,GAAG,GAAG,EAAE,CAAC,IAAI,EACvB,YAAY,EAAE,cAAc,EAC5B,WAAW,GACX,GAAG,OAAO,CAAA;IAEX,IAAI,OAAO,GAAG,CAAC,CAAA;IAEf,KAAK,MAAM,EAAE,IAAI,CAAC,GAAG,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC;QAChE,IAAI,cAAc,CAAC,MAAM,EAAE,OAAO;YAAE,OAAM;QAC1C,MAAM,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,EAAE,CAAE,CAAA;QAE/B,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC;YAAE,SAAQ;QAEjC,KAAK,MAAM,IAAI,IAAI,cAAc,CAAC,MAAM,CAAC,EAAE,CAAC;YAC3C,KAAK,MAAM,OAAO,IAAI,WAAW,CAAC,MAAM,EAAE,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,EAAE,EAAE,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;gBAC/E,IAAI,cAAc,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,cAAc,CAAC,KAAK;oBAAE,OAAM;gBAEjF,MAAM,GAAG,GAAG,aAAa,CAAC,OAAO,CAAC,UAAU,EAAE,MAAM,CAAC,OAAO,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;gBAElF,IAAI,CAAC,GAAG;oBAAE,SAAQ;gBAClB,MAAM,UAAU,GAAG,mBAAmB,CAAC,OAAO,CAAC,UAAU,EAAE,GAAG,CAAC,CAAA;gBAE/D,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC,MAAM;oBAAE,SAAQ;gBAE7C,MAAM;oBACL,GAAG;oBACH,UAAU;oBACV,OAAO,EAAE,MAAM,CAAC,OAAO;oBACvB,MAAM,EAAE,eAAe,CAAC,MAAM,CAAC,OAAO,CAAC;oBACvC,MAAM,EAAE,SAAS;oBACjB,SAAS,EAAE,GAAG,SAAS,IAAI,MAAM,CAAC,EAAE,IAAI,IAAI,CAAC,GAAG,IAAI,OAAO,CAAC,MAAM,EAAE;oBACpE,cAAc,EAAE,EAAE;oBAClB,OAAO,EAAE,SAAS;iBAClB,CAAA;gBAED,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;IACF,CAAC;AACF,CAAC"}

package/out/adapters/wof-postalcode-json/adapter.d.ts

@@ -1,65 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `wof-postalcode`: Who's On First postalcode GeoJSON-bundle adapter.
- *
- *   **Phase 1.5.1 pivot.** Replaces the previous SpatiaLite-backed implementation (formerly at
- *   `packages/corpus/lib/adapters/wof-postalcode/`, removed in this same change). The rationale is
- *   in `wof-admin-json/adapter.ts` and in `DECISIONS.md` — short version: the SQLite distribution
- *   mirror is dead, the live distro tags every postcode row `mz:is_current = -1` which the old
- *   `is_current = 1` predicate excluded, and localized `name:*` variants don't ship in the SQLite
- *   export at all.
- *
- *   Input: a directory containing one or more cloned `whosonfirst-data-postalcode-<cc>` repos plus
- *   the relevant `whosonfirst-data-admin-<cc>` repos (postcode records reference admin ancestry by
- *   `wof:parent_id`, so the locality / region / country records must be in the same walk for the
- *   ancestry chain to resolve). The corpus pipeline clones all four repos under
- *   `/data/corpus/sources/wof/repos/` and points the adapter at that root.
- *
- *   Per live postalcode record, the adapter emits one row per `(name-variant, hierarchy-variant)`
- *   pair:
- *
- *   - **Name variants**: canonical `wof:name` (slot key `default`, typically the postcode digits
- *       themselves) plus any `name:*` variants on the postcode feature. In practice WOF postcode
- *       records rarely carry localized name variants, so this expansion is usually a no-op — but
- *       the code path stays symmetric with the admin adapter for consistency.
- *   - **Hierarchy variants** (unchanged from the SQLite adapter): self, +locality, +locality+region,
- *       +locality+region+country.
- *
- *   `source_id` is `wof-postalcode-<wof_id>-<name-slot>-<hierarchy-variant>`. Ancestor names always
- *   come from the ancestor's canonical `wof:name`; this adapter does NOT iterate ancestor name
- *   variants (e.g. it does not emit `"75008 Париж"` even when Paris has a `name:rus_x_preferred`).
- *   That cross-product belongs to a future synthesis pass; emitting it here would multiply row
- *   counts ~10× without a clear training-value story.
- *
- *   License: CC0.
- */
-import { type WOFVariantSpec } from "#adapters/wof-json-rows";
-import type { CorpusAdapter } from "#types";
-import { type WOFRecord } from "#utils";
-/**
- * Compute hierarchy variants for a postcode record. `selfName` is the postcode surface form (canonical `wof:name` for
- * the `default` slot, a `name:*` localized variant otherwise).
- */
-export declare function postcodeVariantsFor(row: WOFRecord, ancestry: WOFRecord[], selfName: string): WOFVariantSpec[];
-/**
- * Build the per-record name-slot list. The `default` slot uses `wof:name` verbatim (postcode digits); subsequent slots
- * come from `name:*` variants dedup'd against the default.
- */
-export declare function nameSlotsFor(rec: WOFRecord): Array<{
-    key: string;
-    value: string;
-}>;
-/**
- * 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 WOF_POSTALCODE_ADAPTER_ID = "wof-postalcode";
-export declare function createWOFPostalcodeAdapter(): CorpusAdapter;
-/**
- * The configured adapter instance registered with the corpus builder.
- */
-export declare const wofPostalcodeAdapter: CorpusAdapter;
-//# sourceMappingURL=adapter.d.ts.map

package/out/adapters/wof-postalcode-json/adapter.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../lib/adapters/wof-postalcode-json/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAqCG;AAKH,OAAO,EAKN,KAAK,cAAc,EACnB,MAAM,yBAAyB,CAAA;AAChC,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,QAAQ,CAAA;AACzE,OAAO,EAAoC,KAAK,SAAS,EAAE,MAAM,QAAQ,CAAA;AA0BzE;;;GAGG;AACH,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,SAAS,EAAE,QAAQ,EAAE,SAAS,EAAE,EAAE,QAAQ,EAAE,MAAM,GAAG,cAAc,EAAE,CAqC7G;AAED;;;GAGG;AACH,wBAAgB,YAAY,CAAC,GAAG,EAAE,SAAS,GAAG,KAAK,CAAC;IAAE,GAAG,EAAE,MAAM,CAAC;IAAC,KAAK,EAAE,MAAM,CAAA;CAAE,CAAC,CAElF;AAED;;;GAGG;AACH,eAAO,MAAM,yBAAyB,mBAAmB,CAAA;AAEzD,wBAAgB,0BAA0B,IAAI,aAAa,CAqC1D;AAED;;GAEG;AACH,eAAO,MAAM,oBAAoB,eAA+B,CAAA"}

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

@@ -1,151 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `wof-postalcode`: Who's On First postalcode GeoJSON-bundle adapter.
- *
- *   **Phase 1.5.1 pivot.** Replaces the previous SpatiaLite-backed implementation (formerly at
- *   `packages/corpus/lib/adapters/wof-postalcode/`, removed in this same change). The rationale is
- *   in `wof-admin-json/adapter.ts` and in `DECISIONS.md` — short version: the SQLite distribution
- *   mirror is dead, the live distro tags every postcode row `mz:is_current = -1` which the old
- *   `is_current = 1` predicate excluded, and localized `name:*` variants don't ship in the SQLite
- *   export at all.
- *
- *   Input: a directory containing one or more cloned `whosonfirst-data-postalcode-<cc>` repos plus
- *   the relevant `whosonfirst-data-admin-<cc>` repos (postcode records reference admin ancestry by
- *   `wof:parent_id`, so the locality / region / country records must be in the same walk for the
- *   ancestry chain to resolve). The corpus pipeline clones all four repos under
- *   `/data/corpus/sources/wof/repos/` and points the adapter at that root.
- *
- *   Per live postalcode record, the adapter emits one row per `(name-variant, hierarchy-variant)`
- *   pair:
- *
- *   - **Name variants**: canonical `wof:name` (slot key `default`, typically the postcode digits
- *       themselves) plus any `name:*` variants on the postcode feature. In practice WOF postcode
- *       records rarely carry localized name variants, so this expansion is usually a no-op — but
- *       the code path stays symmetric with the admin adapter for consistency.
- *   - **Hierarchy variants** (unchanged from the SQLite adapter): self, +locality, +locality+region,
- *       +locality+region+country.
- *
- *   `source_id` is `wof-postalcode-<wof_id>-<name-slot>-<hierarchy-variant>`. Ancestor names always
- *   come from the ancestor's canonical `wof:name`; this adapter does NOT iterate ancestor name
- *   variants (e.g. it does not emit `"75008 Париж"` even when Paris has a `name:rus_x_preferred`).
- *   That cross-product belongs to a future synthesis pass; emitting it here would multiply row
- *   counts ~10× without a clear training-value story.
- *
- *   License: CC0.
- */
-import { COUNTRY_DISPLAY_NAME, emitWOFJSONRows, LOCALE_BY_COUNTRY, nameSlotsFor as wofNameSlotsFor, } from "#adapters/wof-json-rows";
-import { buildAncestryIndex, walkFeatures } from "#utils";
-/**
- * Map a WOF placetype to a Mailwoman `ComponentTag`, or `undefined` to skip.
- *
- * Per-adapter deliberately (the admin adapter carries its own): each table is a record FILTER for its adapter's
- * emission set — this one keeps `postalcode` plus the ancestry placetypes its variants render — not a shared
- * vocabulary.
- */
-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;
-    }
-}
-/**
- * Compute hierarchy variants for a postcode record. `selfName` is the postcode surface form (canonical `wof:name` for
- * the `default` slot, a `name:*` localized variant otherwise).
- */
-export function postcodeVariantsFor(row, ancestry, selfName) {
-    if (placetypeToTag(row.placetype) !== "postcode")
-        return [];
-    const locality = ancestry.find((a) => placetypeToTag(a.placetype) === "locality");
-    const region = ancestry.find((a) => placetypeToTag(a.placetype) === "region");
-    const country = ancestry.find((a) => placetypeToTag(a.placetype) === "country");
-    const countryDisplay = COUNTRY_DISPLAY_NAME[row.country] ?? country?.name ?? row.country;
-    const variants = [{ suffix: "self", components: { postcode: selfName } }];
-    if (locality) {
-        variants.push({
-            suffix: "with-locality",
-            components: { postcode: selfName, locality: locality.name },
-        });
-    }
-    if (locality && region) {
-        variants.push({
-            suffix: "with-locality-region",
-            components: { postcode: selfName, locality: locality.name, region: region.name },
-        });
-    }
-    if (locality && region && country) {
-        variants.push({
-            suffix: "with-locality-region-country",
-            components: {
-                postcode: selfName,
-                locality: locality.name,
-                region: region.name,
-                country: countryDisplay,
-            },
-        });
-    }
-    return variants;
-}
-/**
- * Build the per-record name-slot list. The `default` slot uses `wof:name` verbatim (postcode digits); subsequent slots
- * come from `name:*` variants dedup'd against the default.
- */
-export function nameSlotsFor(rec) {
-    return wofNameSlotsFor(rec);
-}
-/**
- * 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 WOF_POSTALCODE_ADAPTER_ID = "wof-postalcode";
-export function createWOFPostalcodeAdapter() {
-    return {
-        id: WOF_POSTALCODE_ADAPTER_ID,
-        defaultLicense: "CC0-1.0",
-        description: "Who's On First postalcode GeoJSON bundles (postcode → locality/region pairs). Ancestor names from sibling admin repos.",
-        async *rows(opts) {
-            // Pass 1: full walk. We keep every record whose placetype maps to a ComponentTag — the
-            // postcode adapter needs locality / region / country admin records in the index so it
-            // can resolve postcode ancestry, even though it only emits rows for postcode records.
-            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 = buildAncestryIndex(byID);
-            // Pass 2: emit postcode rows only, sorted by id for determinism.
-            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

@@ -1 +0,0 @@
-{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../lib/adapters/wof-postalcode-json/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAqCG;AAKH,OAAO,EACN,oBAAoB,EACpB,eAAe,EACf,iBAAiB,EACjB,YAAY,IAAI,eAAe,GAE/B,MAAM,yBAAyB,CAAA;AAEhC,OAAO,EAAE,kBAAkB,EAAE,YAAY,EAAkB,MAAM,QAAQ,CAAA;AAEzE;;;;;;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;;;GAGG;AACH,MAAM,UAAU,mBAAmB,CAAC,GAAc,EAAE,QAAqB,EAAE,QAAgB;IAC1F,IAAI,cAAc,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,UAAU;QAAE,OAAO,EAAE,CAAA;IAE3D,MAAM,QAAQ,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,cAAc,CAAC,CAAC,CAAC,SAAS,CAAC,KAAK,UAAU,CAAC,CAAA;IACjF,MAAM,MAAM,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,cAAc,CAAC,CAAC,CAAC,SAAS,CAAC,KAAK,QAAQ,CAAC,CAAA;IAC7E,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,cAAc,CAAC,CAAC,CAAC,SAAS,CAAC,KAAK,SAAS,CAAC,CAAA;IAC/E,MAAM,cAAc,GAAG,oBAAoB,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,OAAO,EAAE,IAAI,IAAI,GAAG,CAAC,OAAO,CAAA;IAExF,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,QAAQ,CAAC,IAAI,EAAE;SAC3D,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,QAAQ,CAAC,IAAI,EAAE,MAAM,EAAE,MAAM,CAAC,IAAI,EAAE;SAChF,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,EAAE,QAAQ,CAAC,IAAI;gBACvB,MAAM,EAAE,MAAM,CAAC,IAAI;gBACnB,OAAO,EAAE,cAAc;aACvB;SACD,CAAC,CAAA;IACH,CAAC;IAED,OAAO,QAAQ,CAAA;AAChB,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,YAAY,CAAC,GAAc;IAC1C,OAAO,eAAe,CAAC,GAAG,CAAC,CAAA;AAC5B,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,MAAM,yBAAyB,GAAG,gBAAgB,CAAA;AAEzD,MAAM,UAAU,0BAA0B;IACzC,OAAO;QACN,EAAE,EAAE,yBAAyB;QAC7B,cAAc,EAAE,SAAS;QACzB,WAAW,EACV,wHAAwH;QAEzH,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,uFAAuF;YACvF,sFAAsF;YACtF,sFAAsF;YACtF,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,kBAAkB,CAAC,IAAI,CAAC,CAAA;YAEzC,iEAAiE;YACjE,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/codex/us-fips-state.d.ts

@@ -1,41 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   US Census FIPS state codes → two-letter abbreviation + full display name.
- *
- *   Covers all 50 states + DC + the five primary territories (PR, GU, VI, MP, AS).
- *
- *   The data is public-domain US Census reference (FIPS PUB 5-2, withdrawn but still the canonical
- *   numeric ID used by every TIGER product). It is reproduced here so the TIGER adapter can resolve
- *   `statefp` columns (e.g. `"50"`) into a `region` component (e.g. `"VT"`) without an extra DB
- *   join.
- */
-/**
- * Per-state record: two-letter postal abbreviation + full canonical display name.
- */
-export interface USStateInfo {
-    abbreviation: string;
-    name: string;
-}
-/**
- * FIPS state-or-territory code → `{ abbreviation, name }`. Includes all 50 states, DC, and the five primary territories
- * (PR, GU, VI, MP, AS). Codes are two-digit zero-padded strings to match TIGER column `statefp`.
- */
-export declare const US_FIPS_STATE: Readonly<Record<string, USStateInfo>>;
-/**
- * Lookup helper. Returns null when the FIPS code isn't recognized.
- */
-export declare function lookupFipsState(statefp: string | null | undefined): USStateInfo | null;
-/**
- * Inverted view: two-letter postal abbreviation → `USStateInfo`. Built once at module load. Used by adapters whose
- * source data ships the abbreviation rather than the FIPS code (FCC BDC, most federal CSVs).
- */
-export declare const US_STATE_BY_ABBREVIATION: Readonly<Record<string, USStateInfo>>;
-/**
- * Lookup helper for adapters carrying 2-char USPS abbreviations (`"CA"`, `"VT"`). Case-folded; null for any value
- * outside the 50 states + DC + the five primary territories.
- */
-export declare function lookupStateAbbreviation(abbreviation: string | null | undefined): USStateInfo | null;
-//# sourceMappingURL=us-fips-state.d.ts.map

package/out/codex/us-fips-state.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"us-fips-state.d.ts","sourceRoot":"","sources":["../../lib/codex/us-fips-state.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH;;GAEG;AACH,MAAM,WAAW,WAAW;IAC3B,YAAY,EAAE,MAAM,CAAA;IACpB,IAAI,EAAE,MAAM,CAAA;CACZ;AAED;;;GAGG;AACH,eAAO,MAAM,aAAa,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,WAAW,CAAC,CA0D9D,CAAA;AAEF;;GAEG;AACH,wBAAgB,eAAe,CAAC,OAAO,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,GAAG,WAAW,GAAG,IAAI,CAItF;AAED;;;GAGG;AACH,eAAO,MAAM,wBAAwB,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,WAAW,CAAC,CAE1E,CAAA;AAED;;;GAGG;AACH,wBAAgB,uBAAuB,CAAC,YAAY,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,GAAG,WAAW,GAAG,IAAI,CAInG"}