@mailwoman/corpus 10.0.0 → 10.2.0

package/out/adapters/dedup-key.d.ts

@@ -0,0 +1,30 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The key that decides two rows describe the same address.
+ */
+import type { CanonicalRow } from "#types";
+/**
+ * Canonical dedup key for a row.
+ *
+ * Two rows that share this key are duplicates and the first one wins.
+ * The key combines `country` with the sorted `components` dictionary and the row's
+ * `raw` text, lower-cased with whitespace collapsed.
+ *
+ * License and provenance fields stay out of the key, so two adapters emitting
+ * the same address produce the same key.
+ * The key store does not deduplicate across adapters: `runAdapter` holds its own as a local
+ * and discards it when the adapter returns, so each adapter is deduplicated only against itself.
+ *
+ * A cross-adapter pass would hold every adapter's keys at once and would have to choose
+ * which adapter's copy of an address survives.
+ * That choice sets the row's `source`, `license` and `register`, which makes it a
+ * corpus-level decision the runner does not take today.
+ *
+ * An augmented row is never deduplicated against the row it was fanned from, because
+ * `recipe.recipe` joins the key when present and each augmentation variant therefore survives.
+ */
+export declare function canonicalDedupKey(row: CanonicalRow): string;
+//# sourceMappingURL=dedup-key.d.ts.map

package/out/adapters/dedup-key.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"dedup-key.d.ts","sourceRoot":"","sources":["../../lib/adapters/dedup-key.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAIH,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,QAAQ,CAAA;AAE1C;;;;;;;;;;;;;;;;;;;GAmBG;AACH,wBAAgB,iBAAiB,CAAC,GAAG,EAAE,YAAY,GAAG,MAAM,CAO3D"}

package/out/adapters/dedup-key.js

@@ -0,0 +1,35 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The key that decides two rows describe the same address.
+ */
+/**
+ * Canonical dedup key for a row.
+ *
+ * Two rows that share this key are duplicates and the first one wins.
+ * The key combines `country` with the sorted `components` dictionary and the row's
+ * `raw` text, lower-cased with whitespace collapsed.
+ *
+ * License and provenance fields stay out of the key, so two adapters emitting
+ * the same address produce the same key.
+ * The key store does not deduplicate across adapters: `runAdapter` holds its own as a local
+ * and discards it when the adapter returns, so each adapter is deduplicated only against itself.
+ *
+ * A cross-adapter pass would hold every adapter's keys at once and would have to choose
+ * which adapter's copy of an address survives.
+ * That choice sets the row's `source`, `license` and `register`, which makes it a
+ * corpus-level decision the runner does not take today.
+ *
+ * An augmented row is never deduplicated against the row it was fanned from, because
+ * `recipe.recipe` joins the key when present and each augmentation variant therefore survives.
+ */
+export function canonicalDedupKey(row) {
+    const sortedKeys = Object.keys(row.components).toSorted();
+    const compPart = sortedKeys.map((k) => `${k}=${row.components[k] ?? ""}`).join("\u001F");
+    const rawNorm = row.raw.toLowerCase().replaceAll(/\s+/g, " ").trim();
+    const recipePart = row.recipe ? `\u001E${row.recipe.recipe}` : "";
+    return `${row.country}\u001E${rawNorm}\u001E${compPart}${recipePart}`;
+}
+//# sourceMappingURL=dedup-key.js.map

package/out/adapters/dedup-key.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"dedup-key.js","sourceRoot":"","sources":["../../lib/adapters/dedup-key.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAMH;;;;;;;;;;;;;;;;;;;GAmBG;AACH,MAAM,UAAU,iBAAiB,CAAC,GAAiB;IAClD,MAAM,UAAU,GAAG,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC,QAAQ,EAAoB,CAAA;IAC3E,MAAM,QAAQ,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,IAAI,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;IACxF,MAAM,OAAO,GAAG,GAAG,CAAC,GAAG,CAAC,WAAW,EAAE,CAAC,UAAU,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;IACpE,MAAM,UAAU,GAAG,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,SAAS,GAAG,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,EAAE,CAAA;IAEjE,OAAO,GAAG,GAAG,CAAC,OAAO,SAAS,OAAO,SAAS,QAAQ,GAAG,UAAU,EAAE,CAAA;AACtE,CAAC"}

package/out/adapters/errors.d.ts

@@ -0,0 +1,17 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The errors an adapter raises for a caller's own input.
+ */
+/**
+ * Raised when a caller asks for a country outside the set the adapter's source publishes.
+ *
+ * An adapter whose source spans several jurisdictions checks `opts.country` against its own set,
+ * so a caller reads which jurisdictions the source publishes instead of receiving zero rows.
+ */
+export declare class UnsupportedCountryError extends Error {
+    constructor(adapterID: string, covered: readonly string[], requested: string);
+}
+//# sourceMappingURL=errors.d.ts.map

package/out/adapters/errors.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"errors.d.ts","sourceRoot":"","sources":["../../lib/adapters/errors.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH;;;;;GAKG;AACH,qBAAa,uBAAwB,SAAQ,KAAK;IACjD,YAAY,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,SAAS,MAAM,EAAE,EAAE,SAAS,EAAE,MAAM,EAI3E;CACD"}

package/out/adapters/errors.js

@@ -0,0 +1,20 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The errors an adapter raises for a caller's own input.
+ */
+/**
+ * Raised when a caller asks for a country outside the set the adapter's source publishes.
+ *
+ * An adapter whose source spans several jurisdictions checks `opts.country` against its own set,
+ * so a caller reads which jurisdictions the source publishes instead of receiving zero rows.
+ */
+export class UnsupportedCountryError extends Error {
+    constructor(adapterID, covered, requested) {
+        super(`${adapterID} adapter: the dataset covers ${covered.join(", ")}, got country=${requested}`);
+        this.name = "UnsupportedCountryError";
+    }
+}
+//# sourceMappingURL=errors.js.map

package/out/adapters/errors.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"errors.js","sourceRoot":"","sources":["../../lib/adapters/errors.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH;;;;;GAKG;AACH,MAAM,OAAO,uBAAwB,SAAQ,KAAK;IACjD,YAAY,SAAiB,EAAE,OAA0B,EAAE,SAAiB;QAC3E,KAAK,CAAC,GAAG,SAAS,gCAAgC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,iBAAiB,SAAS,EAAE,CAAC,CAAA;QAEjG,IAAI,CAAC,IAAI,GAAG,yBAAyB,CAAA;IACtC,CAAC;CACD"}

package/out/adapters/geonames/adapter.d.ts

@@ -3,44 +3,42 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   `geonames`: GeoNames populated-places consumer (https://www.geonames.org/, CC-BY-4.0).
+ *   Reads populated places from a GeoNames country dump and emits locality rows.
  *
- *   GeoNames is a global gazetteer of ~12M features. This adapter ingests the POPULATED PLACES
- *   (`feature_class = "P"`, excluding historical/abandoned/destroyed variants) from a per-country
- *   dump file — global locality coverage, including the small towns and villages a coarser admin
- *   gazetteer (WOF) lacks. It's the cheapest path to broadening the corpus's LOCALE coverage.
- *
- *   Input: a per-country tab-separated dump (e.g. `US.txt` from
- *   `https://download.geonames.org/export/dump/`, 19 columns, no header). Two sibling files in the
- *   same directory supply human-readable names (downloaded once from the same place):
- *
- *   - `admin1CodesASCII.txt` — `<CC>.<admin1_code>` → region name (e.g. `US.VT` → "Vermont").
- *   - `countryInfo.txt` — ISO alpha-2 → country name (e.g. `US` → "United States"); `#`-commented. If a
- *       sibling is missing, the corresponding component is simply omitted (graceful degradation).
- *
- *   Output: per place, up to two hierarchy variants (mirroring `wof-admin`'s with/without-country
- *   balance so the model sees both domestic and international order) —
- *
- *   1. `{ locality, region }` → "City, Region"
- *   2. `{ locality, region, country }` → "City, Region, Country" `componentsPresentIn` drops any
- *        component that didn't survive into the rendered `raw`.
- *
- *   License: stamped `"CC-BY-4.0"` per row (GeoNames' terms); provenance is the `geonames-<id>` key.
+ *   Region and country names come from `admin1CodesASCII.txt` and `countryInfo.txt` in the same directory. When a
+ *   file is missing, rows omit that component.
  */
-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.
+ * The source ID on rows from this adapter.
  */
 export declare const GEONAMES_ADAPTER_ID = "geonames";
 /**
- * License carried by this source (CC-BY-4.0), attached to each row so downstream consumers inherit the terms rather
- * than having to look them up.
+ * The license on rows from this adapter.
  */
 export declare const GEONAMES_DEFAULT_LICENSE = "CC-BY-4.0";
+/**
+ * Zero-based column indices in a GeoNames main-table dump.
+ */
+export declare const GEONAMES_MAIN_COLUMNS: {
+    readonly geonameid: 0;
+    readonly name: 1;
+    readonly asciiname: 2;
+    readonly alternatenames: 3;
+    readonly latitude: 4;
+    readonly longitude: 5;
+    readonly featureClass: 6;
+    readonly featureCode: 7;
+    readonly country: 8;
+    readonly admin1: 10;
+    readonly population: 14;
+};
+/**
+ * Creates the GeoNames populated-place adapter.
+ */
 export declare function createGeonamesAdapter(): CorpusAdapter;
 /**
- * The configured adapter instance registered with the corpus builder.
+ * The adapter instance registered with the corpus builder.
  */
 export declare const geonamesAdapter: CorpusAdapter;
 //# sourceMappingURL=adapter.d.ts.map

package/out/adapters/geonames/adapter.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../lib/adapters/geonames/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AAQH,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEzE;;;GAGG;AACH,eAAO,MAAM,mBAAmB,aAAa,CAAA;AAC7C;;;GAGG;AACH,eAAO,MAAM,wBAAwB,cAAc,CAAA;AA8DnD,wBAAgB,qBAAqB,IAAI,aAAa,CAoFrD;AAED;;GAEG;AACH,eAAO,MAAM,eAAe,eAA0B,CAAA"}
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../lib/adapters/geonames/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AASH,OAAO,EAAuD,KAAK,aAAa,EAAiB,MAAM,QAAQ,CAAA;AAE/G;;GAEG;AACH,eAAO,MAAM,mBAAmB,aAAa,CAAA;AAC7C;;GAEG;AACH,eAAO,MAAM,wBAAwB,cAAc,CAAA;AAEnD;;GAEG;AACH,eAAO,MAAM,qBAAqB;aACjC,SAAS,EAAE,CAAC;aACZ,IAAI,EAAE,CAAC;aACP,SAAS,EAAE,CAAC;aACZ,cAAc,EAAE,CAAC;aACjB,QAAQ,EAAE,CAAC;aACX,SAAS,EAAE,CAAC;aACZ,YAAY,EAAE,CAAC;aACf,WAAW,EAAE,CAAC;aACd,OAAO,EAAE,CAAC;aACV,MAAM,EAAE,EAAE;aACV,UAAU,EAAE,EAAE;CACL,CAAA;AAkDV;;GAEG;AACH,wBAAgB,qBAAqB,IAAI,aAAa,CAsFrD;AAED;;GAEG;AACH,eAAO,MAAM,eAAe,eAA0B,CAAA"}

package/out/adapters/geonames/adapter.js

@@ -3,71 +3,57 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   `geonames`: GeoNames populated-places consumer (https://www.geonames.org/, CC-BY-4.0).
+ *   Reads populated places from a GeoNames country dump and emits locality rows.
  *
- *   GeoNames is a global gazetteer of ~12M features. This adapter ingests the POPULATED PLACES
- *   (`feature_class = "P"`, excluding historical/abandoned/destroyed variants) from a per-country
- *   dump file — global locality coverage, including the small towns and villages a coarser admin
- *   gazetteer (WOF) lacks. It's the cheapest path to broadening the corpus's LOCALE coverage.
- *
- *   Input: a per-country tab-separated dump (e.g. `US.txt` from
- *   `https://download.geonames.org/export/dump/`, 19 columns, no header). Two sibling files in the
- *   same directory supply human-readable names (downloaded once from the same place):
- *
- *   - `admin1CodesASCII.txt` — `<CC>.<admin1_code>` → region name (e.g. `US.VT` → "Vermont").
- *   - `countryInfo.txt` — ISO alpha-2 → country name (e.g. `US` → "United States"); `#`-commented. If a
- *       sibling is missing, the corresponding component is simply omitted (graceful degradation).
- *
- *   Output: per place, up to two hierarchy variants (mirroring `wof-admin`'s with/without-country
- *   balance so the model sees both domestic and international order) —
- *
- *   1. `{ locality, region }` → "City, Region"
- *   2. `{ locality, region, country }` → "City, Region, Country" `componentsPresentIn` drops any
- *        component that didn't survive into the rendered `raw`.
- *
- *   License: stamped `"CC-BY-4.0"` per row (GeoNames' terms); provenance is the `geonames-<id>` key.
+ *   Region and country names come from `admin1CodesASCII.txt` and `countryInfo.txt` in the same directory. When a
+ *   file is missing, rows omit that component.
  */
-import { componentsPresentIn } from "@mailwoman/codex/address-format";
-import { pathExists } from "@mailwoman/core/fs/readers";
-import { dirname, join } from "path-ts";
-import { TSVSpliterator } from "spliterator";
-import { stableSourceID } from "#adapters/utils";
+import { componentsPresentIn } from "@mailwoman/codex/address/format";
+import { readUnquotedTSV } from "@mailwoman/core/fs/delimited";
+import { pathExists } from "@mailwoman/core/fs/readers/stat";
+import { PathBuilder } from "path-ts";
+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.
+ * The source ID on rows from this adapter.
  */
 export const GEONAMES_ADAPTER_ID = "geonames";
 /**
- * License carried by this source (CC-BY-4.0), attached to each row so downstream consumers inherit the terms rather
- * than having to look them up.
+ * The license on rows from this adapter.
  */
 export const GEONAMES_DEFAULT_LICENSE = "CC-BY-4.0";
 /**
- * GeoNames main-table column indices (0-based; see the export README).
+ * Zero-based column indices in a GeoNames main-table dump.
  */
-const COL = {
+export const GEONAMES_MAIN_COLUMNS = {
     geonameid: 0,
     name: 1,
+    asciiname: 2,
     alternatenames: 3,
+    latitude: 4,
+    longitude: 5,
     featureClass: 6,
     featureCode: 7,
     country: 8,
     admin1: 10,
+    population: 14,
 };
+const COL = GEONAMES_MAIN_COLUMNS;
 /**
- * Populated-place feature codes that are NOT current real places — skip them.
+ * The adapter skips these feature codes for historical, abandoned, or destroyed populated places.
  */
 const NON_CURRENT_PPL = new Set(["PPLH", "PPLQ", "PPLW", "PPLCH"]);
 /**
- * Load `admin1CodesASCII.txt` → Map("<CC>.<admin1>" → region name). Empty map if absent.
+ * Loads region names keyed by `<CC>.<admin1>`.
+ * Returns an empty map when the file is missing.
  */
 async function loadAdmin1(dir) {
     const map = new Map();
-    const fp = join(dir, "admin1CodesASCII.txt");
+    const fp = dir("admin1CodesASCII.txt");
     if (!(await pathExists(fp)))
         return map;
-    // `header: false` — the file is headerless, and the spliterator eats row 1 as a header otherwise.
-    for await (const cols of TSVSpliterator.fromAsync(fp, { header: false })) {
+    for await (const cols of readUnquotedTSV(fp)) {
         if (cols[0] && cols[1]) {
             map.set(cols[0], cols[1]);
         }
@@ -75,36 +61,40 @@ async function loadAdmin1(dir) {
     return map;
 }
 /**
- * Load `countryInfo.txt` → Map(ISO → country name). Empty map if absent. The file is `#`-commented.
+ * Loads country names keyed by ISO code.
+ * Returns an empty map when the file is missing.
  */
 async function loadCountries(dir) {
     const map = new Map();
-    const fp = join(dir, "countryInfo.txt");
+    const fp = dir("countryInfo.txt");
     if (!(await pathExists(fp)))
         return map;
-    // `header: false` — the file's header IS a `#` comment, so it falls out with the other comments
-    // rather than being consumed as column names.
-    for await (const cols of TSVSpliterator.fromAsync(fp, { header: false })) {
+    for await (const cols of readUnquotedTSV(fp)) {
         if (cols[0]?.startsWith("#"))
             continue;
-        // ISO(0), ISO3(1), iso-numeric(2), fips(3), Country(4), ...
+        // Column 0 holds the ISO code and column 4 the country name.
         if (cols[0] && cols[4]) {
             map.set(cols[0], cols[4]);
         }
     }
     return map;
 }
+/**
+ * Creates the GeoNames populated-place adapter.
+ */
 export function createGeonamesAdapter() {
     return {
         id: GEONAMES_ADAPTER_ID,
         defaultLicense: GEONAMES_DEFAULT_LICENSE,
+        addressRole: AddressRole.Premise,
+        register: SourceRegister.GeoNames,
+        surface: SurfaceOrigin.Rendered,
         description: "GeoNames populated places (CC-BY-4.0) — global locality coverage incl. small towns, with region/country names from the sibling admin1/countryInfo files.",
         async *rows(opts) {
-            const dir = dirname(opts.inputPath);
+            const dir = PathBuilder.from(opts.inputPath).dirname();
             const admin1 = await loadAdmin1(dir);
             const countries = await loadCountries(dir);
-            // `header: false` — the per-country dump is headerless.
-            const rows = TSVSpliterator.fromAsync(opts.inputPath, { header: false });
+            const rows = readUnquotedTSV(opts.inputPath);
             let emitted = 0;
             for await (const rec of rows) {
                 if (opts.signal?.aborted)
@@ -126,8 +116,8 @@ export function createGeonamesAdapter() {
                 const geonameid = (rec[COL.geonameid] ?? "").trim();
                 const region = admin1.get(`${cc}.${(rec[COL.admin1] ?? "").trim()}`);
                 const country = countries.get(cc);
-                // Two hierarchy variants (domestic + international order) — but only emit the
-                // distinct ones the available names support.
+                // With a region, the adapter emits locality-region and locality-region-country rows.
+                // Without one, it emits a single row with the country when known.
                 const variants = [];
                 if (region) {
                     variants.push({ slot: "lr", comp: { locality, region }, raw: `${locality}, ${region}` });
@@ -170,7 +160,7 @@ export function createGeonamesAdapter() {
     };
 }
 /**
- * The configured adapter instance registered with the corpus builder.
+ * The adapter instance registered with the corpus builder.
  */
 export const geonamesAdapter = createGeonamesAdapter();
 //# sourceMappingURL=adapter.js.map

package/out/adapters/geonames/adapter.js.map

@@ -1 +1 @@
-{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../lib/adapters/geonames/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AAEH,OAAO,EAAE,mBAAmB,EAAE,MAAM,iCAAiC,CAAA;AACrE,OAAO,EAAE,UAAU,EAAE,MAAM,4BAA4B,CAAA;AACvD,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,SAAS,CAAA;AACvC,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAA;AAE5C,OAAO,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAGhD;;;GAGG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAG,UAAU,CAAA;AAC7C;;;GAGG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,WAAW,CAAA;AAEnD;;GAEG;AACH,MAAM,GAAG,GAAG;IACX,SAAS,EAAE,CAAC;IACZ,IAAI,EAAE,CAAC;IACP,cAAc,EAAE,CAAC;IACjB,YAAY,EAAE,CAAC;IACf,WAAW,EAAE,CAAC;IACd,OAAO,EAAE,CAAC;IACV,MAAM,EAAE,EAAE;CACD,CAAA;AAEV;;GAEG;AACH,MAAM,eAAe,GAAG,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC,CAAA;AAElE;;GAEG;AACH,KAAK,UAAU,UAAU,CAAC,GAAW;IACpC,MAAM,GAAG,GAAG,IAAI,GAAG,EAAkB,CAAA;IACrC,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,EAAE,sBAAsB,CAAC,CAAA;IAE5C,IAAI,CAAC,CAAC,MAAM,UAAU,CAAC,EAAE,CAAC,CAAC;QAAE,OAAO,GAAG,CAAA;IAEvC,kGAAkG;IAClG,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,cAAc,CAAC,SAAS,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC;QAC1E,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;YACxB,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;QAC1B,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;GAEG;AACH,KAAK,UAAU,aAAa,CAAC,GAAW;IACvC,MAAM,GAAG,GAAG,IAAI,GAAG,EAAkB,CAAA;IACrC,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,EAAE,iBAAiB,CAAC,CAAA;IAEvC,IAAI,CAAC,CAAC,MAAM,UAAU,CAAC,EAAE,CAAC,CAAC;QAAE,OAAO,GAAG,CAAA;IAEvC,gGAAgG;IAChG,8CAA8C;IAC9C,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,cAAc,CAAC,SAAS,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC;QAC1E,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,UAAU,CAAC,GAAG,CAAC;YAAE,SAAQ;QAEtC,4DAA4D;QAC5D,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;YACxB,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;QAC1B,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED,MAAM,UAAU,qBAAqB;IACpC,OAAO;QACN,EAAE,EAAE,mBAAmB;QACvB,cAAc,EAAE,wBAAwB;QACxC,WAAW,EACV,0JAA0J;QAE3J,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,MAAM,GAAG,GAAG,OAAO,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;YACnC,MAAM,MAAM,GAAG,MAAM,UAAU,CAAC,GAAG,CAAC,CAAA;YACpC,MAAM,SAAS,GAAG,MAAM,aAAa,CAAC,GAAG,CAAC,CAAA;YAE1C,wDAAwD;YACxD,MAAM,IAAI,GAAG,cAAc,CAAC,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAA;YAExE,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;gBAC9B,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,IAAI,GAAG,CAAC,GAAG,CAAC,YAAY,CAAC,KAAK,GAAG;oBAAE,SAAQ;gBAE3C,IAAI,eAAe,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC;oBAAE,SAAQ;gBAE7D,MAAM,EAAE,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE1C,IAAI,CAAC,EAAE;oBAAE,SAAQ;gBAEjB,IAAI,IAAI,CAAC,OAAO,IAAI,EAAE,KAAK,IAAI,CAAC,OAAO;oBAAE,SAAQ;gBAEjD,MAAM,QAAQ,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE7C,IAAI,CAAC,QAAQ;oBAAE,SAAQ;gBACvB,MAAM,SAAS,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACnD,MAAM,MAAM,GAAG,MAAM,CAAC,GAAG,CAAC,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA;gBACpE,MAAM,OAAO,GAAG,SAAS,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;gBAEjC,8EAA8E;gBAC9E,6CAA6C;gBAC7C,MAAM,QAAQ,GAA2E,EAAE,CAAA;gBAE3F,IAAI,MAAM,EAAE,CAAC;oBACZ,QAAQ,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,EAAE,GAAG,EAAE,GAAG,QAAQ,KAAK,MAAM,EAAE,EAAE,CAAC,CAAA;oBAExF,IAAI,OAAO,EAAE,CAAC;wBACb,QAAQ,CAAC,IAAI,CAAC;4BACb,IAAI,EAAE,KAAK;4BACX,IAAI,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,OAAO,EAAE;4BACnC,GAAG,EAAE,GAAG,QAAQ,KAAK,MAAM,KAAK,OAAO,EAAE;yBACzC,CAAC,CAAA;oBACH,CAAC;gBACF,CAAC;qBAAM,IAAI,OAAO,EAAE,CAAC;oBACpB,QAAQ,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,OAAO,EAAE,EAAE,GAAG,EAAE,GAAG,QAAQ,KAAK,OAAO,EAAE,EAAE,CAAC,CAAA;gBAC3F,CAAC;qBAAM,CAAC;oBACP,QAAQ,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,EAAE,GAAG,EAAE,QAAQ,EAAE,CAAC,CAAA;gBAChE,CAAC;gBAED,KAAK,MAAM,CAAC,IAAI,QAAQ,EAAE,CAAC;oBAC1B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;wBAAE,MAAK;oBAC5D,MAAM,OAAO,GAAG,mBAAmB,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAA;oBAElD,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM;wBAAE,SAAQ;oBAE1C,MAAM,QAAQ,GAAG,SAAS;wBACzB,CAAC,CAAC,GAAG,mBAAmB,IAAI,SAAS,IAAI,CAAC,CAAC,IAAI,EAAE;wBACjD,CAAC,CAAC,cAAc,CAAC,mBAAmB,EAAE,OAAO,CAAC,CAAA;oBAE/C,MAAM;wBACL,GAAG,EAAE,CAAC,CAAC,GAAG;wBACV,UAAU,EAAE,OAAO;wBACnB,OAAO,EAAE,EAAE;wBACX,MAAM,EAAE,mBAAmB;wBAC3B,SAAS,EAAE,QAAQ;wBACnB,cAAc,EAAE,EAAE;wBAClB,OAAO,EAAE,wBAAwB;qBACjC,CAAA;oBAED,OAAO,EAAE,CAAA;gBACV,CAAC;YACF,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,qBAAqB,EAAE,CAAA"}
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../lib/adapters/geonames/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EAAE,mBAAmB,EAAE,MAAM,iCAAiC,CAAA;AACrE,OAAO,EAAE,eAAe,EAAE,MAAM,8BAA8B,CAAA;AAC9D,OAAO,EAAE,UAAU,EAAE,MAAM,iCAAiC,CAAA;AAC5D,OAAO,EAAE,WAAW,EAAE,MAAM,SAAS,CAAA;AAErC,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;;GAEG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAG,UAAU,CAAA;AAC7C;;GAEG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,WAAW,CAAA;AAEnD;;GAEG;AACH,MAAM,CAAC,MAAM,qBAAqB,GAAG;IACpC,SAAS,EAAE,CAAC;IACZ,IAAI,EAAE,CAAC;IACP,SAAS,EAAE,CAAC;IACZ,cAAc,EAAE,CAAC;IACjB,QAAQ,EAAE,CAAC;IACX,SAAS,EAAE,CAAC;IACZ,YAAY,EAAE,CAAC;IACf,WAAW,EAAE,CAAC;IACd,OAAO,EAAE,CAAC;IACV,MAAM,EAAE,EAAE;IACV,UAAU,EAAE,EAAE;CACL,CAAA;AAEV,MAAM,GAAG,GAAG,qBAAqB,CAAA;AAEjC;;GAEG;AACH,MAAM,eAAe,GAAG,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC,CAAA;AAElE;;;GAGG;AACH,KAAK,UAAU,UAAU,CAAC,GAAgB;IACzC,MAAM,GAAG,GAAG,IAAI,GAAG,EAAkB,CAAA;IACrC,MAAM,EAAE,GAAG,GAAG,CAAC,sBAAsB,CAAC,CAAA;IAEtC,IAAI,CAAC,CAAC,MAAM,UAAU,CAAC,EAAE,CAAC,CAAC;QAAE,OAAO,GAAG,CAAA;IAEvC,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,eAAe,CAAC,EAAE,CAAC,EAAE,CAAC;QAC9C,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;YACxB,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;QAC1B,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;;GAGG;AACH,KAAK,UAAU,aAAa,CAAC,GAAgB;IAC5C,MAAM,GAAG,GAAG,IAAI,GAAG,EAAkB,CAAA;IACrC,MAAM,EAAE,GAAG,GAAG,CAAC,iBAAiB,CAAC,CAAA;IAEjC,IAAI,CAAC,CAAC,MAAM,UAAU,CAAC,EAAE,CAAC,CAAC;QAAE,OAAO,GAAG,CAAA;IAEvC,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,eAAe,CAAC,EAAE,CAAC,EAAE,CAAC;QAC9C,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,UAAU,CAAC,GAAG,CAAC;YAAE,SAAQ;QAEtC,6DAA6D;QAC7D,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;YACxB,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;QAC1B,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,qBAAqB;IACpC,OAAO;QACN,EAAE,EAAE,mBAAmB;QACvB,cAAc,EAAE,wBAAwB;QACxC,WAAW,EAAE,WAAW,CAAC,OAAO;QAChC,QAAQ,EAAE,cAAc,CAAC,QAAQ;QACjC,OAAO,EAAE,aAAa,CAAC,QAAQ;QAC/B,WAAW,EACV,0JAA0J;QAE3J,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,MAAM,GAAG,GAAG,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,OAAO,EAAE,CAAA;YACtD,MAAM,MAAM,GAAG,MAAM,UAAU,CAAC,GAAG,CAAC,CAAA;YACpC,MAAM,SAAS,GAAG,MAAM,aAAa,CAAC,GAAG,CAAC,CAAA;YAE1C,MAAM,IAAI,GAAG,eAAe,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;YAE5C,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;gBAC9B,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,IAAI,GAAG,CAAC,GAAG,CAAC,YAAY,CAAC,KAAK,GAAG;oBAAE,SAAQ;gBAE3C,IAAI,eAAe,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC;oBAAE,SAAQ;gBAE7D,MAAM,EAAE,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE1C,IAAI,CAAC,EAAE;oBAAE,SAAQ;gBAEjB,IAAI,IAAI,CAAC,OAAO,IAAI,EAAE,KAAK,IAAI,CAAC,OAAO;oBAAE,SAAQ;gBAEjD,MAAM,QAAQ,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE7C,IAAI,CAAC,QAAQ;oBAAE,SAAQ;gBACvB,MAAM,SAAS,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACnD,MAAM,MAAM,GAAG,MAAM,CAAC,GAAG,CAAC,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA;gBACpE,MAAM,OAAO,GAAG,SAAS,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;gBAEjC,qFAAqF;gBACrF,kEAAkE;gBAClE,MAAM,QAAQ,GAA2E,EAAE,CAAA;gBAE3F,IAAI,MAAM,EAAE,CAAC;oBACZ,QAAQ,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,EAAE,GAAG,EAAE,GAAG,QAAQ,KAAK,MAAM,EAAE,EAAE,CAAC,CAAA;oBAExF,IAAI,OAAO,EAAE,CAAC;wBACb,QAAQ,CAAC,IAAI,CAAC;4BACb,IAAI,EAAE,KAAK;4BACX,IAAI,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,OAAO,EAAE;4BACnC,GAAG,EAAE,GAAG,QAAQ,KAAK,MAAM,KAAK,OAAO,EAAE;yBACzC,CAAC,CAAA;oBACH,CAAC;gBACF,CAAC;qBAAM,IAAI,OAAO,EAAE,CAAC;oBACpB,QAAQ,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,OAAO,EAAE,EAAE,GAAG,EAAE,GAAG,QAAQ,KAAK,OAAO,EAAE,EAAE,CAAC,CAAA;gBAC3F,CAAC;qBAAM,CAAC;oBACP,QAAQ,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,EAAE,GAAG,EAAE,QAAQ,EAAE,CAAC,CAAA;gBAChE,CAAC;gBAED,KAAK,MAAM,CAAC,IAAI,QAAQ,EAAE,CAAC;oBAC1B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;wBAAE,MAAK;oBAC5D,MAAM,OAAO,GAAG,mBAAmB,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAA;oBAElD,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM;wBAAE,SAAQ;oBAE1C,MAAM,QAAQ,GAAG,SAAS;wBACzB,CAAC,CAAC,GAAG,mBAAmB,IAAI,SAAS,IAAI,CAAC,CAAC,IAAI,EAAE;wBACjD,CAAC,CAAC,cAAc,CAAC,mBAAmB,EAAE,OAAO,CAAC,CAAA;oBAE/C,MAAM;wBACL,GAAG,EAAE,CAAC,CAAC,GAAG;wBACV,UAAU,EAAE,OAAO;wBACnB,OAAO,EAAE,EAAE;wBACX,MAAM,EAAE,mBAAmB;wBAC3B,SAAS,EAAE,QAAQ;wBACnB,cAAc,EAAE,EAAE;wBAClB,OAAO,EAAE,wBAAwB;qBACjC,CAAA;oBAED,OAAO,EAAE,CAAA;gBACV,CAAC;YACF,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,qBAAqB,EAAE,CAAA"}

package/out/adapters/geonames/postal/adapter.d.ts

@@ -3,41 +3,23 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   `geonames-postal`: GeoNames postal-code dump consumer (https://www.geonames.org/, CC-BY-4.0).
- *
- *   The GeoNames postal export (`https://download.geonames.org/export/zip/<CC>.zip`) is a clean,
- *   per-country `postcode → place → admin1` table with the place + region NAMES inline (no aux-file
- *   join needed). It broadens the corpus's postcode→locality→region coverage to ~80 countries, well
- *   beyond `wof-postalcode`/the coordinate-first table — forward coverage for the multi-locale
- *   goal.
- *
- *   Input: a per-country postal dump (`<CC>.txt`, 12 tab-separated columns, no header): country,
- *   postcode, place, admin1_name, admin1_code, admin2__, admin3__, lat, lon, accuracy.
- *
- *   Output: per row, postcode-FIRST (international) variants — the common order for the non-US
- *   locales this fills (US postcodes are already covered by TIGER/WOF, which use postcode-LAST):
- *
- *   1. `{ postcode, locality }` → "AD100 Canillo"
- *   2. `{ postcode, locality, region }` → "AD100 Canillo, Canillo" Prefer configuring this adapter for
- *        non-US countries; for US, the postcode-last sources are the right order. License:
- *        `"CC-BY-4.0"` per row (attribute "GeoNames").
+ *   Reads a per-country GeoNames postal dump and emits postcode-first locality rows. Each dump row includes its place and
+ *   region names inline.
  */
-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.
+ * The source ID on rows from this adapter.
  */
 export declare const GEONAMES_POSTAL_ADAPTER_ID = "geonames-postal";
 /**
- * License carried by this source (CC-BY-4.0), attached to each row so downstream consumers inherit the terms rather
- * than having to look them up.
+ * The license on rows from this adapter.
  */
 export declare const GEONAMES_POSTAL_DEFAULT_LICENSE = "CC-BY-4.0";
 /**
- * GeoNames postal-dump columns (0-based): country, postcode, place, admin1_name, admin1_code, admin2_name, ….
+ * Zero-based column indices in a GeoNames postal dump.
  *
- * Shared with `tools/postcode-triples.ts`, which additionally reads `admin2Name` — the CITY for the IN/MX/PT-shaped
- * exports whose `place` column is a street or colonia.
+ * The coordinates locate the postcode.
+ * That position may differ from the locality's position.
  */
 export declare const GEONAMES_POSTAL_COLUMNS: {
     readonly country: 0;
@@ -45,10 +27,15 @@ export declare const GEONAMES_POSTAL_COLUMNS: {
     readonly place: 2;
     readonly admin1Name: 3;
     readonly admin2Name: 5;
+    readonly latitude: 9;
+    readonly longitude: 10;
 };
+/**
+ * Creates the GeoNames postal adapter.
+ */
 export declare function createGeonamesPostalAdapter(): CorpusAdapter;
 /**
- * The configured adapter instance registered with the corpus builder.
+ * The adapter instance registered with the corpus builder.
  */
 export declare const geonamesPostalAdapter: CorpusAdapter;
 //# sourceMappingURL=adapter.d.ts.map

package/out/adapters/geonames/postal/adapter.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/adapters/geonames/postal/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AAMH,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEzE;;;GAGG;AACH,eAAO,MAAM,0BAA0B,oBAAoB,CAAA;AAC3D;;;GAGG;AACH,eAAO,MAAM,+BAA+B,cAAc,CAAA;AAE1D;;;;;GAKG;AACH,eAAO,MAAM,uBAAuB;aAAK,OAAO,EAAE,CAAC;aAAE,QAAQ,EAAE,CAAC;aAAE,KAAK,EAAE,CAAC;aAAE,UAAU,EAAE,CAAC;aAAE,UAAU,EAAE,CAAC;CAAW,CAAA;AAEnH,wBAAgB,2BAA2B,IAAI,aAAa,CAkE3D;AAED;;GAEG;AACH,eAAO,MAAM,qBAAqB,eAAgC,CAAA"}
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/adapters/geonames/postal/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAOH,OAAO,EAAuD,KAAK,aAAa,EAAiB,MAAM,QAAQ,CAAA;AAE/G;;GAEG;AACH,eAAO,MAAM,0BAA0B,oBAAoB,CAAA;AAC3D;;GAEG;AACH,eAAO,MAAM,+BAA+B,cAAc,CAAA;AAE1D;;;;;GAKG;AACH,eAAO,MAAM,uBAAuB;aACnC,OAAO,EAAE,CAAC;aACV,QAAQ,EAAE,CAAC;aACX,KAAK,EAAE,CAAC;aACR,UAAU,EAAE,CAAC;aACb,UAAU,EAAE,CAAC;aACb,QAAQ,EAAE,CAAC;aACX,SAAS,EAAE,EAAE;CACJ,CAAA;AAEV;;GAEG;AACH,wBAAgB,2BAA2B,IAAI,aAAa,CAkE3D;AAED;;GAEG;AACH,eAAO,MAAM,qBAAqB,eAAgC,CAAA"}

package/out/adapters/geonames/postal/adapter.js

@@ -3,54 +3,50 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   `geonames-postal`: GeoNames postal-code dump consumer (https://www.geonames.org/, CC-BY-4.0).
- *
- *   The GeoNames postal export (`https://download.geonames.org/export/zip/<CC>.zip`) is a clean,
- *   per-country `postcode → place → admin1` table with the place + region NAMES inline (no aux-file
- *   join needed). It broadens the corpus's postcode→locality→region coverage to ~80 countries, well
- *   beyond `wof-postalcode`/the coordinate-first table — forward coverage for the multi-locale
- *   goal.
- *
- *   Input: a per-country postal dump (`<CC>.txt`, 12 tab-separated columns, no header): country,
- *   postcode, place, admin1_name, admin1_code, admin2__, admin3__, lat, lon, accuracy.
- *
- *   Output: per row, postcode-FIRST (international) variants — the common order for the non-US
- *   locales this fills (US postcodes are already covered by TIGER/WOF, which use postcode-LAST):
- *
- *   1. `{ postcode, locality }` → "AD100 Canillo"
- *   2. `{ postcode, locality, region }` → "AD100 Canillo, Canillo" Prefer configuring this adapter for
- *        non-US countries; for US, the postcode-last sources are the right order. License:
- *        `"CC-BY-4.0"` per row (attribute "GeoNames").
+ *   Reads a per-country GeoNames postal dump and emits postcode-first locality rows. Each dump row includes its place and
+ *   region names inline.
  */
-import { componentsPresentIn } from "@mailwoman/codex/address-format";
-import { TSVSpliterator } from "spliterator";
-import { stableSourceID } from "#adapters/utils";
+import { componentsPresentIn } from "@mailwoman/codex/address/format";
+import { readUnquotedTSV } from "@mailwoman/core/fs/delimited";
+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.
+ * The source ID on rows from this adapter.
  */
 export const GEONAMES_POSTAL_ADAPTER_ID = "geonames-postal";
 /**
- * License carried by this source (CC-BY-4.0), attached to each row so downstream consumers inherit the terms rather
- * than having to look them up.
+ * The license on rows from this adapter.
  */
 export const GEONAMES_POSTAL_DEFAULT_LICENSE = "CC-BY-4.0";
 /**
- * GeoNames postal-dump columns (0-based): country, postcode, place, admin1_name, admin1_code, admin2_name, ….
+ * Zero-based column indices in a GeoNames postal dump.
  *
- * Shared with `tools/postcode-triples.ts`, which additionally reads `admin2Name` — the CITY for the IN/MX/PT-shaped
- * exports whose `place` column is a street or colonia.
+ * The coordinates locate the postcode.
+ * That position may differ from the locality's position.
+ */
+export const GEONAMES_POSTAL_COLUMNS = {
+    country: 0,
+    postcode: 1,
+    place: 2,
+    admin1Name: 3,
+    admin2Name: 5,
+    latitude: 9,
+    longitude: 10,
+};
+/**
+ * Creates the GeoNames postal adapter.
  */
-export const GEONAMES_POSTAL_COLUMNS = { country: 0, postcode: 1, place: 2, admin1Name: 3, admin2Name: 5 };
 export function createGeonamesPostalAdapter() {
     return {
         id: GEONAMES_POSTAL_ADAPTER_ID,
         defaultLicense: GEONAMES_POSTAL_DEFAULT_LICENSE,
+        addressRole: AddressRole.Premise,
+        register: SourceRegister.GeoNamesPostal,
+        surface: SurfaceOrigin.Rendered,
         description: "GeoNames postcodes (CC-BY-4.0) — multi-locale postcode→locality→region, names inline; international postcode-first order.",
         async *rows(opts) {
-            // `header: false` — the GeoNames postal dump is headerless, and the spliterator would
-            // otherwise consume row 1 as column names and lose its first postcode.
-            const rows = TSVSpliterator.fromAsync(opts.inputPath, { header: false });
+            const rows = readUnquotedTSV(opts.inputPath);
             let emitted = 0;
             for await (const rec of rows) {
                 if (opts.signal?.aborted)
@@ -67,8 +63,7 @@ export function createGeonamesPostalAdapter() {
                 if (!postcode || !locality)
                     continue;
                 const region = (rec[GEONAMES_POSTAL_COLUMNS.admin1Name] ?? "").trim();
-                // Postcode-first (international) variants. Skip the region variant when admin1 just
-                // repeats the place (common for city-states / micro-admin) to avoid "X X" noise.
+                // The region row is skipped when the region name equals the locality name.
                 const variants = [
                     { slot: "pl", comp: { postcode, locality }, raw: `${postcode} ${locality}` },
                 ];
@@ -101,7 +96,7 @@ export function createGeonamesPostalAdapter() {
     };
 }
 /**
- * The configured adapter instance registered with the corpus builder.
+ * The adapter instance registered with the corpus builder.
  */
 export const geonamesPostalAdapter = createGeonamesPostalAdapter();
 //# sourceMappingURL=adapter.js.map

package/out/adapters/geonames/postal/adapter.js.map

@@ -1 +1 @@
-{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../lib/adapters/geonames/postal/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AAEH,OAAO,EAAE,mBAAmB,EAAE,MAAM,iCAAiC,CAAA;AACrE,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAA;AAE5C,OAAO,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAGhD;;;GAGG;AACH,MAAM,CAAC,MAAM,0BAA0B,GAAG,iBAAiB,CAAA;AAC3D;;;GAGG;AACH,MAAM,CAAC,MAAM,+BAA+B,GAAG,WAAW,CAAA;AAE1D;;;;;GAKG;AACH,MAAM,CAAC,MAAM,uBAAuB,GAAG,EAAE,OAAO,EAAE,CAAC,EAAE,QAAQ,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,UAAU,EAAE,CAAC,EAAE,UAAU,EAAE,CAAC,EAAW,CAAA;AAEnH,MAAM,UAAU,2BAA2B;IAC1C,OAAO;QACN,EAAE,EAAE,0BAA0B;QAC9B,cAAc,EAAE,+BAA+B;QAC/C,WAAW,EACV,2HAA2H;QAE5H,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,sFAAsF;YACtF,uEAAuE;YACvE,MAAM,IAAI,GAAG,cAAc,CAAC,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAA;YAExE,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,IAA+B,EAAE,CAAC;gBACzD,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,EAAE,GAAG,CAAC,GAAG,CAAC,uBAAuB,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE9D,IAAI,CAAC,EAAE;oBAAE,SAAQ;gBAEjB,IAAI,IAAI,CAAC,OAAO,IAAI,EAAE,KAAK,IAAI,CAAC,OAAO;oBAAE,SAAQ;gBAEjD,MAAM,QAAQ,GAAG,CAAC,GAAG,CAAC,uBAAuB,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACrE,MAAM,QAAQ,GAAG,CAAC,GAAG,CAAC,uBAAuB,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAElE,IAAI,CAAC,QAAQ,IAAI,CAAC,QAAQ;oBAAE,SAAQ;gBACpC,MAAM,MAAM,GAAG,CAAC,GAAG,CAAC,uBAAuB,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAErE,oFAAoF;gBACpF,iFAAiF;gBACjF,MAAM,QAAQ,GAA2E;oBACxF,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,QAAQ,EAAE,EAAE,GAAG,EAAE,GAAG,QAAQ,IAAI,QAAQ,EAAE,EAAE;iBAC5E,CAAA;gBAED,IAAI,MAAM,IAAI,MAAM,CAAC,WAAW,EAAE,KAAK,QAAQ,CAAC,WAAW,EAAE,EAAE,CAAC;oBAC/D,QAAQ,CAAC,IAAI,CAAC;wBACb,IAAI,EAAE,KAAK;wBACX,IAAI,EAAE,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE;wBACpC,GAAG,EAAE,GAAG,QAAQ,IAAI,QAAQ,KAAK,MAAM,EAAE;qBACzC,CAAC,CAAA;gBACH,CAAC;gBAED,KAAK,MAAM,CAAC,IAAI,QAAQ,EAAE,CAAC;oBAC1B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;wBAAE,MAAK;oBAC5D,MAAM,OAAO,GAAG,mBAAmB,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAA;oBAElD,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM,GAAG,CAAC;wBAAE,SAAQ;oBAE7C,MAAM;wBACL,GAAG,EAAE,CAAC,CAAC,GAAG;wBACV,UAAU,EAAE,OAAO;wBACnB,OAAO,EAAE,EAAE;wBACX,MAAM,EAAE,0BAA0B;wBAClC,SAAS,EAAE,GAAG,cAAc,CAAC,0BAA0B,EAAE,OAAO,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE;wBAC7E,cAAc,EAAE,EAAE;wBAClB,OAAO,EAAE,+BAA+B;qBACxC,CAAA;oBAED,OAAO,EAAE,CAAA;gBACV,CAAC;YACF,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,qBAAqB,GAAG,2BAA2B,EAAE,CAAA"}
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../lib/adapters/geonames/postal/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,EAAE,mBAAmB,EAAE,MAAM,iCAAiC,CAAA;AACrE,OAAO,EAAE,eAAe,EAAE,MAAM,8BAA8B,CAAA;AAE9D,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;;GAEG;AACH,MAAM,CAAC,MAAM,0BAA0B,GAAG,iBAAiB,CAAA;AAC3D;;GAEG;AACH,MAAM,CAAC,MAAM,+BAA+B,GAAG,WAAW,CAAA;AAE1D;;;;;GAKG;AACH,MAAM,CAAC,MAAM,uBAAuB,GAAG;IACtC,OAAO,EAAE,CAAC;IACV,QAAQ,EAAE,CAAC;IACX,KAAK,EAAE,CAAC;IACR,UAAU,EAAE,CAAC;IACb,UAAU,EAAE,CAAC;IACb,QAAQ,EAAE,CAAC;IACX,SAAS,EAAE,EAAE;CACJ,CAAA;AAEV;;GAEG;AACH,MAAM,UAAU,2BAA2B;IAC1C,OAAO;QACN,EAAE,EAAE,0BAA0B;QAC9B,cAAc,EAAE,+BAA+B;QAC/C,WAAW,EAAE,WAAW,CAAC,OAAO;QAChC,QAAQ,EAAE,cAAc,CAAC,cAAc;QACvC,OAAO,EAAE,aAAa,CAAC,QAAQ;QAC/B,WAAW,EACV,2HAA2H;QAE5H,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,MAAM,IAAI,GAAG,eAAe,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;YAE5C,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;gBAC9B,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,EAAE,GAAG,CAAC,GAAG,CAAC,uBAAuB,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE9D,IAAI,CAAC,EAAE;oBAAE,SAAQ;gBAEjB,IAAI,IAAI,CAAC,OAAO,IAAI,EAAE,KAAK,IAAI,CAAC,OAAO;oBAAE,SAAQ;gBAEjD,MAAM,QAAQ,GAAG,CAAC,GAAG,CAAC,uBAAuB,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACrE,MAAM,QAAQ,GAAG,CAAC,GAAG,CAAC,uBAAuB,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAElE,IAAI,CAAC,QAAQ,IAAI,CAAC,QAAQ;oBAAE,SAAQ;gBACpC,MAAM,MAAM,GAAG,CAAC,GAAG,CAAC,uBAAuB,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAErE,2EAA2E;gBAC3E,MAAM,QAAQ,GAA2E;oBACxF,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,QAAQ,EAAE,EAAE,GAAG,EAAE,GAAG,QAAQ,IAAI,QAAQ,EAAE,EAAE;iBAC5E,CAAA;gBAED,IAAI,MAAM,IAAI,MAAM,CAAC,WAAW,EAAE,KAAK,QAAQ,CAAC,WAAW,EAAE,EAAE,CAAC;oBAC/D,QAAQ,CAAC,IAAI,CAAC;wBACb,IAAI,EAAE,KAAK;wBACX,IAAI,EAAE,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE;wBACpC,GAAG,EAAE,GAAG,QAAQ,IAAI,QAAQ,KAAK,MAAM,EAAE;qBACzC,CAAC,CAAA;gBACH,CAAC;gBAED,KAAK,MAAM,CAAC,IAAI,QAAQ,EAAE,CAAC;oBAC1B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;wBAAE,MAAK;oBAC5D,MAAM,OAAO,GAAG,mBAAmB,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAA;oBAElD,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM,GAAG,CAAC;wBAAE,SAAQ;oBAE7C,MAAM;wBACL,GAAG,EAAE,CAAC,CAAC,GAAG;wBACV,UAAU,EAAE,OAAO;wBACnB,OAAO,EAAE,EAAE;wBACX,MAAM,EAAE,0BAA0B;wBAClC,SAAS,EAAE,GAAG,cAAc,CAAC,0BAA0B,EAAE,OAAO,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE;wBAC7E,cAAc,EAAE,EAAE;wBAClB,OAAO,EAAE,+BAA+B;qBACxC,CAAA;oBAED,OAAO,EAAE,CAAA;gBACV,CAAC;YACF,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,qBAAqB,GAAG,2BAA2B,EAAE,CAAA"}