@mailwoman/corpus 9.4.0 → 10.1.0

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

@@ -0,0 +1,54 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * Assemble a sampled, component-labeled Australian address set from G-NAF (the Geocoded National
+ * Address File, Geoscape Australia, Open G-NAF license). Any derived artifact must attribute
+ * "Geoscape Australia". G-NAF is a relational PSV distribution whose street address reconstructs
+ * by joining ADDRESS_DETAIL (number, postcode, the PIDs) to STREET_LOCALITY (street name and type)
+ * and then to locality (suburb). State is the per-file prefix (ACT/NSW/…).
+ *
+ * The two lookup tables load as Maps. ADDRESS_DETAIL streams once and is reservoir-sampled, so
+ * memory stays bounded.
+ *
+ * No coordinates. The output feeds the parser. The parser needs the address string and its component
+ * labels. Output is component tuples as jsonl. An optional held-out eval set is excluded by
+ * (street, locality, postcode) so the training corpus never overlaps the benchmark.
+ */
+import { type PathBuilderLike } from "path-ts";
+export interface GNAFAssembleOptions {
+    /**
+     * G-NAF `Standard` directory (holds the per-state `*_psv.psv` tables).
+     */
+    standardDir: PathBuilderLike;
+    /**
+     * Target sample size (uniform reservoir → population-proportional across states).
+     */
+    sampleSize: number;
+    /**
+     * Output jsonl path.
+     */
+    out: string;
+    /**
+     * Optional held-out eval jsonl (rows with a `components` field) —
+     * its (street,locality,postcode) are excluded.
+     */
+    holdoutPath?: string;
+    /**
+     * Progress sink (the CLI passes a setter).
+     */
+    onProgress?: (message: string) => void;
+}
+export interface GNAFAssembleResult {
+    written: number;
+    seen: number;
+    heldOut: number;
+    byState: Record<string, number>;
+}
+/**
+ * Holdout/dedup key: a street within a locality+postcode (house-number-agnostic — conservative).
+ */
+export declare function gnafHoldoutKey(street: string, locality: string, postcode: string): string;
+export declare function assembleGNAF(opts: GNAFAssembleOptions): Promise<GNAFAssembleResult>;
+//# sourceMappingURL=assemble.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"assemble.d.ts","sourceRoot":"","sources":["../../../../lib/au/adapters/gnaf/assemble.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAGH,OAAO,EAAe,KAAK,eAAe,EAAE,MAAM,SAAS,CAAA;AAI3D,MAAM,WAAW,mBAAmB;IACnC;;OAEG;IACH,WAAW,EAAE,eAAe,CAAA;IAC5B;;OAEG;IACH,UAAU,EAAE,MAAM,CAAA;IAClB;;OAEG;IACH,GAAG,EAAE,MAAM,CAAA;IACX;;;OAGG;IACH,WAAW,CAAC,EAAE,MAAM,CAAA;IACpB;;OAEG;IACH,UAAU,CAAC,EAAE,CAAC,OAAO,EAAE,MAAM,KAAK,IAAI,CAAA;CACtC;AAED,MAAM,WAAW,kBAAkB;IAClC,OAAO,EAAE,MAAM,CAAA;IACf,IAAI,EAAE,MAAM,CAAA;IACZ,OAAO,EAAE,MAAM,CAAA;IACf,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;CAC/B;AAED;;GAEG;AACH,wBAAgB,cAAc,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,MAAM,CAEzF;AA4CD,wBAAsB,YAAY,CAAC,IAAI,EAAE,mBAAmB,GAAG,OAAO,CAAC,kBAAkB,CAAC,CAoGzF"}

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

@@ -0,0 +1,134 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * Assemble a sampled, component-labeled Australian address set from G-NAF (the Geocoded National
+ * Address File, Geoscape Australia, Open G-NAF license). Any derived artifact must attribute
+ * "Geoscape Australia". G-NAF is a relational PSV distribution whose street address reconstructs
+ * by joining ADDRESS_DETAIL (number, postcode, the PIDs) to STREET_LOCALITY (street name and type)
+ * and then to locality (suburb). State is the per-file prefix (ACT/NSW/…).
+ *
+ * The two lookup tables load as Maps. ADDRESS_DETAIL streams once and is reservoir-sampled, so
+ * memory stays bounded.
+ *
+ * No coordinates. The output feeds the parser. The parser needs the address string and its component
+ * labels. Output is component tuples as jsonl. An optional held-out eval set is excluded by
+ * (street, locality, postcode) so the training corpus never overlaps the benchmark.
+ */
+import { tryParsingJSON, stringifyJSON } from "@mailwoman/core/json";
+import { PathBuilder } from "path-ts";
+import { titleCase, createNewlineWriter, PSVSpliterator, TextSpliterator } from "spliterator";
+import { Globerator } from "spliterator/node/fs";
+/**
+ * Holdout/dedup key: a street within a locality+postcode (house-number-agnostic — conservative).
+ */
+export function gnafHoldoutKey(street, locality, postcode) {
+    return `${street}|${locality}|${postcode}`.toLowerCase();
+}
+async function* psvObjects(path) {
+    yield* PSVSpliterator.fromAsync(path);
+}
+/**
+ * Load a small lookup table fully into a Map keyed by `keyCol`.
+ */
+async function loadMap(paths, keyCol, pick) {
+    const m = new Map();
+    for (const p of paths) {
+        for await (const r of psvObjects(p)) {
+            const k = r[keyCol];
+            if (k != null && k !== "") {
+                m.set(String(k), pick(r));
+            }
+        }
+    }
+    return m;
+}
+/**
+ * Build the held-out key set from an eval jsonl whose rows contain a `components` object.
+ */
+async function loadHoldout(path) {
+    const keys = new Set();
+    for await (const line of TextSpliterator.fromAsync(path)) {
+        const c = tryParsingJSON(line)?.components;
+        if (c?.street && c?.locality && c?.postcode) {
+            keys.add(gnafHoldoutKey(c.street, c.locality, c.postcode));
+        }
+    }
+    return keys;
+}
+export async function assembleGNAF(opts) {
+    const progress = opts.onProgress ?? (() => { });
+    const standardDir = PathBuilder.from(opts.standardDir);
+    const files = await Globerator.from("*", { cwd: standardDir, absolute: false }).toArray();
+    // Strings, because the address loop reads the state code out of each path.
+    const pick = (re, exclude) => files.filter((f) => re.test(f) && !(exclude && exclude.test(f))).map((f) => standardDir(f).toString());
+    // `*_LOCALITY_psv.psv` also globs `*_STREET_LOCALITY_psv.psv` — exclude the latter explicitly.
+    const streetPaths = pick(/_STREET_LOCALITY_psv\.psv$/);
+    const localityPaths = pick(/_LOCALITY_psv\.psv$/, /_STREET_LOCALITY_psv\.psv$/);
+    const addressPaths = pick(/_ADDRESS_DETAIL_psv\.psv$/);
+    const holdout = opts.holdoutPath ? await loadHoldout(opts.holdoutPath) : new Set();
+    if (opts.holdoutPath) {
+        progress(`held-out eval keys: ${holdout.size}`);
+    }
+    progress(`loading STREET_LOCALITY (${streetPaths.length} files) + LOCALITY (${localityPaths.length})…`);
+    const streetMap = await loadMap(streetPaths, "street_locality_pid", (r) => ({
+        name: String(r.street_name ?? ""),
+        type: String(r.street_type_code ?? ""),
+        suffix: String(r.street_suffix_code ?? ""),
+    }));
+    const localityMap = await loadMap(localityPaths, "locality_pid", (r) => String(r.locality_name ?? ""));
+    progress(`streets=${streetMap.size.toLocaleString()} localities=${localityMap.size.toLocaleString()}`);
+    const reservoir = [];
+    let seen = 0;
+    let heldOut = 0;
+    for (const p of addressPaths) {
+        const state = (p.match(/\/([A-Z]+)_ADDRESS_DETAIL/) ?? [])[1] ?? "";
+        for await (const r of psvObjects(p)) {
+            const numberFirst = String(r.number_first ?? "");
+            if (!numberFirst || r.date_retired || !r.postcode)
+                continue;
+            const st = streetMap.get(String(r.street_locality_pid ?? ""));
+            const suburbRaw = localityMap.get(String(r.locality_pid ?? ""));
+            if (!st?.name || !suburbRaw)
+                continue;
+            const street = `${titleCase(st.name)} ${titleCase(st.type)}${st.suffix ? " " + titleCase(st.suffix) : ""}`.trim();
+            const locality = titleCase(suburbRaw);
+            const postcode = String(r.postcode);
+            if (holdout.has(gnafHoldoutKey(street, locality, postcode))) {
+                heldOut++;
+                continue;
+            }
+            let house = numberFirst + (r.number_first_suffix ? String(r.number_first_suffix) : "");
+            if (r.number_last) {
+                house = `${house}-${String(r.number_last)}`;
+            }
+            if (r.flat_number) {
+                house = `${String(r.flat_number)}/${house}`;
+            }
+            const tuple = { house_number: house, street, locality, region: state, postcode };
+            seen++;
+            if (reservoir.length < opts.sampleSize) {
+                reservoir.push(tuple);
+            }
+            else {
+                const j = Math.floor(Math.random() * seen);
+                if (j < opts.sampleSize) {
+                    reservoir[j] = tuple;
+                }
+            }
+        }
+        progress(`${state}: ${seen.toLocaleString()} valid joinable seen`);
+    }
+    const byState = {};
+    {
+        await using out = createNewlineWriter(opts.out);
+        for (const t of reservoir) {
+            await out.write(stringifyJSON(t));
+            byState[t.region] = (byState[t.region] ?? 0) + 1;
+        }
+    }
+    progress(`wrote ${reservoir.length.toLocaleString()} tuples → ${opts.out}`);
+    return { written: reservoir.length, seen, heldOut, byState };
+}
+//# sourceMappingURL=assemble.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"assemble.js","sourceRoot":"","sources":["../../../../lib/au/adapters/gnaf/assemble.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAEH,OAAO,EAAE,cAAc,EAAE,aAAa,EAAE,MAAM,sBAAsB,CAAA;AACpE,OAAO,EAAE,WAAW,EAAwB,MAAM,SAAS,CAAA;AAC3D,OAAO,EAAE,SAAS,EAAE,mBAAmB,EAAE,cAAc,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAC7F,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAA;AAiChD;;GAEG;AACH,MAAM,UAAU,cAAc,CAAC,MAAc,EAAE,QAAgB,EAAE,QAAgB;IAChF,OAAO,GAAG,MAAM,IAAI,QAAQ,IAAI,QAAQ,EAAE,CAAC,WAAW,EAAE,CAAA;AACzD,CAAC;AAID,KAAK,SAAS,CAAC,CAAC,UAAU,CAAC,IAAY;IACtC,KAAK,CAAC,CAAC,cAAc,CAAC,SAAS,CAAC,IAAI,CAAuB,CAAA;AAC5D,CAAC;AAED;;GAEG;AACH,KAAK,UAAU,OAAO,CAAI,KAAe,EAAE,MAAc,EAAE,IAAmB;IAC7E,MAAM,CAAC,GAAG,IAAI,GAAG,EAAa,CAAA;IAE9B,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE,CAAC;QACvB,IAAI,KAAK,EAAE,MAAM,CAAC,IAAI,UAAU,CAAC,CAAC,CAAC,EAAE,CAAC;YACrC,MAAM,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,CAAA;YAEnB,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,KAAK,EAAE,EAAE,CAAC;gBAC3B,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;YAC1B,CAAC;QACF,CAAC;IACF,CAAC;IAED,OAAO,CAAC,CAAA;AACT,CAAC;AAED;;GAEG;AACH,KAAK,UAAU,WAAW,CAAC,IAAY;IACtC,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;IAE9B,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,eAAe,CAAC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC;QAC1D,MAAM,CAAC,GAAG,cAAc,CAA0C,IAAI,CAAC,EAAE,UAAU,CAAA;QAEnF,IAAI,CAAC,EAAE,MAAM,IAAI,CAAC,EAAE,QAAQ,IAAI,CAAC,EAAE,QAAQ,EAAE,CAAC;YAC7C,IAAI,CAAC,GAAG,CAAC,cAAc,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,QAAQ,EAAE,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAA;QAC3D,CAAC;IACF,CAAC;IAED,OAAO,IAAI,CAAA;AACZ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,YAAY,CAAC,IAAyB;IAC3D,MAAM,QAAQ,GAAG,IAAI,CAAC,UAAU,IAAI,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAA;IAC9C,MAAM,WAAW,GAAG,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,WAAW,CAAC,CAAA;IACtD,MAAM,KAAK,GAAG,MAAM,UAAU,CAAC,IAAI,CAAC,GAAG,EAAE,EAAE,GAAG,EAAE,WAAW,EAAE,QAAQ,EAAE,KAAK,EAAE,CAAC,CAAC,OAAO,EAAE,CAAA;IAEzF,2EAA2E;IAC3E,MAAM,IAAI,GAAG,CAAC,EAAU,EAAE,OAAgB,EAAE,EAAE,CAC7C,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,CAAC,CAAA;IAEvG,+FAA+F;IAC/F,MAAM,WAAW,GAAG,IAAI,CAAC,4BAA4B,CAAC,CAAA;IACtD,MAAM,aAAa,GAAG,IAAI,CAAC,qBAAqB,EAAE,4BAA4B,CAAC,CAAA;IAC/E,MAAM,YAAY,GAAG,IAAI,CAAC,2BAA2B,CAAC,CAAA;IAEtD,MAAM,OAAO,GAAG,IAAI,CAAC,WAAW,CAAC,CAAC,CAAC,MAAM,WAAW,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,IAAI,GAAG,EAAU,CAAA;IAE1F,IAAI,IAAI,CAAC,WAAW,EAAE,CAAC;QACtB,QAAQ,CAAC,uBAAuB,OAAO,CAAC,IAAI,EAAE,CAAC,CAAA;IAChD,CAAC;IAED,QAAQ,CAAC,4BAA4B,WAAW,CAAC,MAAM,uBAAuB,aAAa,CAAC,MAAM,IAAI,CAAC,CAAA;IAEvG,MAAM,SAAS,GAAG,MAAM,OAAO,CAAC,WAAW,EAAE,qBAAqB,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;QAC3E,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC,WAAW,IAAI,EAAE,CAAC;QACjC,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC,gBAAgB,IAAI,EAAE,CAAC;QACtC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAC,kBAAkB,IAAI,EAAE,CAAC;KAC1C,CAAC,CAAC,CAAA;IAEH,MAAM,WAAW,GAAG,MAAM,OAAO,CAAC,aAAa,EAAE,cAAc,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,aAAa,IAAI,EAAE,CAAC,CAAC,CAAA;IACtG,QAAQ,CAAC,WAAW,SAAS,CAAC,IAAI,CAAC,cAAc,EAAE,eAAe,WAAW,CAAC,IAAI,CAAC,cAAc,EAAE,EAAE,CAAC,CAAA;IAEtG,MAAM,SAAS,GACd,EAAE,CAAA;IAEH,IAAI,IAAI,GAAG,CAAC,CAAA;IACZ,IAAI,OAAO,GAAG,CAAC,CAAA;IAEf,KAAK,MAAM,CAAC,IAAI,YAAY,EAAE,CAAC;QAC9B,MAAM,KAAK,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,2BAA2B,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAA;QAEnE,IAAI,KAAK,EAAE,MAAM,CAAC,IAAI,UAAU,CAAC,CAAC,CAAC,EAAE,CAAC;YACrC,MAAM,WAAW,GAAG,MAAM,CAAC,CAAC,CAAC,YAAY,IAAI,EAAE,CAAC,CAAA;YAEhD,IAAI,CAAC,WAAW,IAAI,CAAC,CAAC,YAAY,IAAI,CAAC,CAAC,CAAC,QAAQ;gBAAE,SAAQ;YAC3D,MAAM,EAAE,GAAG,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,mBAAmB,IAAI,EAAE,CAAC,CAAC,CAAA;YAC7D,MAAM,SAAS,GAAG,WAAW,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,YAAY,IAAI,EAAE,CAAC,CAAC,CAAA;YAE/D,IAAI,CAAC,EAAE,EAAE,IAAI,IAAI,CAAC,SAAS;gBAAE,SAAQ;YACrC,MAAM,MAAM,GAAG,GAAG,SAAS,CAAC,EAAE,CAAC,IAAI,CAAC,IAAI,SAAS,CAAC,EAAE,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,GAAG,SAAS,CAAC,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,IAAI,EAAE,CAAA;YACjH,MAAM,QAAQ,GAAG,SAAS,CAAC,SAAS,CAAC,CAAA;YACrC,MAAM,QAAQ,GAAG,MAAM,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAA;YAEnC,IAAI,OAAO,CAAC,GAAG,CAAC,cAAc,CAAC,MAAM,EAAE,QAAQ,EAAE,QAAQ,CAAC,CAAC,EAAE,CAAC;gBAC7D,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,IAAI,KAAK,GAAG,WAAW,GAAG,CAAC,CAAC,CAAC,mBAAmB,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,mBAAmB,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAA;YAEtF,IAAI,CAAC,CAAC,WAAW,EAAE,CAAC;gBACnB,KAAK,GAAG,GAAG,KAAK,IAAI,MAAM,CAAC,CAAC,CAAC,WAAW,CAAC,EAAE,CAAA;YAC5C,CAAC;YAED,IAAI,CAAC,CAAC,WAAW,EAAE,CAAC;gBACnB,KAAK,GAAG,GAAG,MAAM,CAAC,CAAC,CAAC,WAAW,CAAC,IAAI,KAAK,EAAE,CAAA;YAC5C,CAAC;YAED,MAAM,KAAK,GAAG,EAAE,YAAY,EAAE,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,EAAE,KAAK,EAAE,QAAQ,EAAE,CAAA;YAEhF,IAAI,EAAE,CAAA;YAEN,IAAI,SAAS,CAAC,MAAM,GAAG,IAAI,CAAC,UAAU,EAAE,CAAC;gBACxC,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;YACtB,CAAC;iBAAM,CAAC;gBACP,MAAM,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,EAAE,GAAG,IAAI,CAAC,CAAA;gBAE1C,IAAI,CAAC,GAAG,IAAI,CAAC,UAAU,EAAE,CAAC;oBACzB,SAAS,CAAC,CAAC,CAAC,GAAG,KAAK,CAAA;gBACrB,CAAC;YACF,CAAC;QACF,CAAC;QAED,QAAQ,CAAC,GAAG,KAAK,KAAK,IAAI,CAAC,cAAc,EAAE,sBAAsB,CAAC,CAAA;IACnE,CAAC;IAED,MAAM,OAAO,GAA2B,EAAE,CAAA;IAE1C,CAAC;QACA,YAAY,GAAG,GAAG,mBAAmB,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;QAE/C,KAAK,MAAM,CAAC,IAAI,SAAS,EAAE,CAAC;YAC3B,MAAM,GAAG,CAAC,KAAK,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAA;YACjC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;QACjD,CAAC;IACF,CAAC;IAED,QAAQ,CAAC,SAAS,SAAS,CAAC,MAAM,CAAC,cAAc,EAAE,aAAa,IAAI,CAAC,GAAG,EAAE,CAAC,CAAA;IAE3E,OAAO,EAAE,OAAO,EAAE,SAAS,CAAC,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;AAC7D,CAAC"}

package/out/be/adapters/brussels/adapter.d.ts

@@ -0,0 +1,212 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `brussels`: the INSPIRE Addresses (AD) theme the Brussels Regional Informatics Center publishes out
+ * of UrbIS.
+ *
+ * Coverage is the Brussels-Capital Region rather than Belgium. Belgium needs three readers, because
+ * each region publishes the theme through its own service: Flanders through a WFS (`vlaanderen`),
+ * Wallonia through an ATOM download service (`wallonie`), and Brussels through the bare HTTP
+ * download below. Every row this adapter emits is `BE`, and any other `opts.country` raises
+ * {@linkcode UnsupportedCountryError}.
+ *
+ * The schema-level reading is `#inspire/address`. This module holds only what is Brussels' own: the
+ * NIS code embedded in every place name, and the empty component reference nine addresses write.
+ *
+ * ## The publication
+ *
+ * `https://urbisdownload.datastore.brussels/INSPIRE/URBIS_ADM_Adresses.zip`, HTTP 200, 11,521,182
+ * bytes, holding one member `UrbAdm_Adresses.gml` of 474,245,978 bytes. The member is streamed and
+ * never buffered.
+ *
+ * Brussels publishes no INSPIRE ATOM feed for this theme, so there is no `<updated>` to poll and the
+ * download is a bare URL. The archive's `last-modified` is `Thu, 01 Dec 2022 07:57:50 GMT`, against
+ * Wallonia's 2025-12-11, so this is the older of the two Belgian regional publications by three
+ * years.
+ *
+ * Two of the publisher's hostnames do not serve it. `urbis.brussels` and `urbisonline.brussels` both
+ * fail TLS with `ERR_TLS_CERT_ALTNAME_INVALID`, because the certificate's altnames are `*.irisnet.be`
+ * and `irisnet.be`. The services live on `irisnet.be`, and the GML's own `base:namespace` states that
+ * host: `http://urbisdownload-v2.gis.irisnet.be/en/temporality#`.
+ *
+ * Counts, read 2026-10-02 with {@linkcode streamMarkupElements} and reconciled against a second
+ * instrument that extracted `gml:featureMember` blocks from the same stream as text: 225,660
+ * `ad:Address`, 6,238 `ad:ThoroughfareName`, 26 `ad:PostalDescriptor` and 23 `ad:AdminUnitName`. It
+ * publishes no `ad:AddressAreaName`. The two instruments agree and they reconcile: 225,660 addresses
+ * plus 6,287 component features is the 231,947 `gml:featureMember` elements the file holds, and
+ * 231,947 is also the number of void `ad:validFrom` elements, one per member.
+ *
+ * ## The join takes one pass
+ *
+ * Brussels writes all 6,287 component features before the first address, the opposite of Wallonia, so
+ * one pass indexes and joins. Each address writes exactly 7 `ad:component` references — country,
+ * region, province, district, municipality, postal zone and street — all of them local `#` fragments.
+ * Of 1,579,620 reference slots, 1,579,611 name a `gml:id` in the same document and every one of those
+ * resolves. The remaining 9 are empty.
+ *
+ * ## The nine empty references
+ *
+ * Nine addresses write `<ad:component xlink:href=""/>` in the postal-zone slot, so those nine state
+ * no postcode: `BE.BRUSSELS.BRIC.ADM.ADPT.220623`, `.220631`, `.220635`, `.256972`, `.257005` and
+ * four more, in Bruxelles and Schaerbeek. `componentHrefs` skips an empty href, so an unwary reader
+ * yields those rows with the postcode simply missing. This adapter refuses them and
+ * counts them under {@link Refusals.noPostcode}, because a postcode the publisher left unstated is
+ * not a postcode this repository may read as absent from the address.
+ *
+ * ## The NIS code embedded in every place name
+ *
+ * An `ad:AdminUnitName` spells its name as the NIS code, a space, and the name in parentheses:
+ * `21019 (Woluwe-Saint-Pierre)`, `21004 (Bruxelles)`, `01000 (Belgique)`,
+ * `21000 (Bruxelles-Capitale)`. All 23 take that shape. The code is the publisher's key rather than
+ * part of the name, so {@linkcode placeNameWithoutNISCode} strips it and a row reads
+ * `Woluwe-Saint-Pierre`. A name not of that shape is returned unchanged rather than guessed at.
+ *
+ * The province the file names is `20001 (Brabant flamand)`, and the Brussels-Capital Region sits in no
+ * province. This adapter reads the province nowhere: the locality comes from the unit at
+ * {@link MUNICIPALITY_LEVEL}.
+ *
+ * ## Voids state no reason at all
+ *
+ * Not one nil in this file states a `nilReason`. Over the whole member, 0 void elements give a
+ * reason and these state none: `ad:validFrom` 231,947, `ad:method` 225,660, `ad:specification`
+ * 225,660, `gn:nativeness`, `gn:nameStatus`, `gn:sourceOfName`, `gn:pronunciation` and `gn:script`
+ * 12,574 each, `ad:beginLifespanVersion` 23 and `ad:adminUnit` 23. A reader keyed on `nilReason`
+ * mistakes every one of them for a populated value. `isVoid` in `#inspire/address` therefore keys
+ * on `xsi:nil`, and `voidReason` is read nowhere here.
+ *
+ * ## Coordinates
+ *
+ * This adapter emits text and components only, because {@linkcode CanonicalRow} declares no position
+ * field. The axis order is recorded here for whoever does emit one, because this publisher
+ * contradicts its own declaration.
+ *
+ * Brussels declares `srsName="http://www.opengis.net/def/crs/EPSG/0/3035"`, whose authority order is
+ * northing then easting, and writes easting then northing. `BE.BRUSSELS.BRIC.ADM.ADPT.169731` states
+ * `gml:pos` `3928985.14831507 3095401.17112511`. Read in the declared order that is latitude
+ * 56.877382, longitude −10.402355, in the Atlantic west of Ireland. Swapped it is latitude 50.835742,
+ * longitude 4.429144, in Woluwe-Saint-Pierre, the municipality the same address's own
+ * components name. The file's `gml:boundedBy` lower corner, `3916355.39160602 3088216.44803091`, puts
+ * the easting first in the same way, and Wallonia's, `2942208.441255148 3816649.462808032`, puts the
+ * northing first. So the two Belgian regions declare one CRS and write two axis orders, and a reader
+ * that emits coordinates must decide per publisher rather than per declaration.
+ *
+ * ## Terms, and why no row of this source may ship yet
+ *
+ * **The address-source register holds no row for Brussels.** Its four `BE` sources are
+ * `be-business-legal-1`, `be-business-register-1`, `be-property-building-1` (Flanders) and
+ * `be-property-building-2` (Wallonia). The word Brussels appears in the register only inside
+ * Flanders' and Wallonia's own prose. Both state that Brussels publishes separately. So no license
+ * decision has been elected for this publisher, `electedLicenseLabel` finds no decision to answer
+ * with, and {@linkcode BRUSSELS_DEFAULT_LICENSE} is a label that reads as unreviewed rather than an
+ * SPDX id that would read as elected.
+ *
+ * `createIneligibilityReader` refuses every row whose source no register row declares, so this
+ * adapter's rows cannot enter a release-profile corpus until a row exists, a license is elected for
+ * it, and its `adapterID` is set to `brussels`. Until then the adapter is a reader without a grant.
+ *
+ * The publisher is reported to state CC BY 4.0 in its GeoNetwork record. This file records that as
+ * reported rather than as observed, because the catalogue was not reached from here on 2026-10-02:
+ * `datastore.brussels/geonetwork/srv/...` answered HTTP 404 on both the API and CSW paths and
+ * `geocatalog.brussels` did not resolve. The register elects the terms, and that election needs the
+ * record itself.
+ */
+import { type AdapterOptions, 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 name is the region, as `vlaanderen`'s and `wallonie`'s are.
+ */
+export declare const BRUSSELS_ADAPTER_ID = "brussels";
+/**
+ * The stable id of the publication behind every row.
+ *
+ * It belongs in `SourceRegister` beside `be-vlaanderen-inspire-ad`, and this constant
+ * stands in until that entry lands, because a built corpus stores the id on every row
+ * and the id must not change afterwards.
+ * Replace the constant with the `SourceRegister` member on integration.
+ */
+export declare const BRUSSELS_SOURCE_REGISTER: "be-brussels-inspire-ad";
+/**
+ * The one jurisdiction this adapter emits.
+ *
+ * The download publishes the Brussels-Capital Region, to which ISO 3166-1 assigns
+ * no code of its own, so the set holds one code and a caller naming another gets
+ * {@linkcode UnsupportedCountryError} rather than zero rows.
+ */
+export declare const BRUSSELS_COUNTRIES: readonly string[];
+/**
+ * The license label recorded on every row.
+ *
+ * The register's source `be-property-building-3` elects `CC-BY-4.0` for Paradigm, read from
+ * the `gmd:useConstraints` of ISO 19139 record `260abd6f-c5b4-4a79-aa80-d6c265002d65`,
+ * which names `URBIS_ADM_Adresses.zip` among its distributions.
+ * That record governs this archive rather than Paradigm's other address product,
+ * the BeSt Brussels register, which is CC0-1.0.
+ *
+ * The value is the SPDX identifier rather than the license's title, because
+ * `licenseVerdict` resolves an identifier and reads a title as unrecognized.
+ * This constant previously held `unreviewed-be-brussels-urbis`, which resolved to no expression,
+ * so these rows carried unknown obligations into a build that had no way to filter them.
+ */
+export declare const BRUSSELS_DEFAULT_LICENSE = "CC-BY-4.0";
+/**
+/**
+ * A place name with the publisher's NIS code removed.
+ *
+ * `21019 (Woluwe-Saint-Pierre)` reads `Woluwe-Saint-Pierre`.
+ * A value not of that shape is returned trimmed and otherwise unchanged, because this strips
+ * a prefix the publisher writes rather than parsing a name: a changed shape must reach a
+ * reader as the publisher's own value instead of as a guess at what the name inside might be.
+ *
+ * A value of that shape whose parentheses hold an empty string answers `undefined`,
+ * so the caller refuses the row rather than writing a NIS code into a locality.
+ *
+ * @returns The name, or `undefined` for an absent value and for a code with an empty name.
+ */
+export declare function placeNameWithoutNISCode(value: string | undefined): string | undefined;
+/**
+ * The components an address references, keyed by the `gml:id` its `#` fragment names.
+ */
+export interface BrusselsComponentIndex {
+    /**
+     * `ad:ThoroughfareName` → the street name the publisher spells.
+     *
+     * A street writes a French and a Dutch `ad:name`, in that order, and `thoroughfareName`
+     * reads the first: `BE.BRUSSELS.BRIC.ADM.PN.2110` is `Avenue Jules de Trooz`
+     * and `Jules de Troozlaan`, and the row takes the French spelling.
+     */
+    streets: ReadonlyMap<string, string>;
+    /**
+     * `ad:PostalDescriptor` → its `ad:postCode`.
+     *
+     * Twenty-six descriptors over nineteen municipalities, so a code may appear twice:
+     * `BE.BRUSSELS.BRIC.ADM.MZ.13` and `.MZ.18` both hold `1050`.
+     */
+    postcodes: ReadonlyMap<string, string>;
+    /**
+     * `ad:AdminUnitName` at {@link MUNICIPALITY_LEVEL} → the municipality name, NIS code removed.
+     *
+     * A unit at another level is left out, so the country, region, province
+     * and district references cannot answer for a locality.
+     */
+    municipalities: ReadonlyMap<string, string>;
+}
+/**
+ * Indexes the three component types in one pass over the member's head.
+ *
+ * The pass runs to the end of the member rather than stopping at the first address, because the
+ * component-before-address order is a measurement of this file rather than a rule of the schema.
+ *
+ * @throws {@linkcode InspireArchiveError} when a component states no `gml:id`,
+ * when two components share one, or when the pass ends with an empty index.
+ */
+export declare function readBrusselsComponents(inputPath: AdapterOptions["inputPath"], signal?: AbortSignal): Promise<BrusselsComponentIndex>;
+export declare function createBrusselsAdapter(): CorpusAdapter;
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export declare const brusselsAdapter: CorpusAdapter;
+//# sourceMappingURL=adapter.d.ts.map

package/out/be/adapters/brussels/adapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/be/adapters/brussels/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAiHG;AAsBH,OAAO,EAAe,KAAK,cAAc,EAAqB,KAAK,aAAa,EAAiB,MAAM,QAAQ,CAAA;AAE/G;;;;;GAKG;AACH,eAAO,MAAM,mBAAmB,aAAa,CAAA;AAE7C;;;;;;;GAOG;AACH,eAAO,MAAM,wBAAwB,0BAA0C,CAAA;AAE/E;;;;;;GAMG;AACH,eAAO,MAAM,kBAAkB,EAAE,SAAS,MAAM,EAAW,CAAA;AAE3D;;;;;;;;;;;;;GAaG;AACH,eAAO,MAAM,wBAAwB,cAAc,CAAA;AA6DnD;;;;;;;;;;;;;GAaG;AACH,wBAAgB,uBAAuB,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,GAAG,MAAM,GAAG,SAAS,CAUrF;AAED;;GAEG;AACH,MAAM,WAAW,sBAAsB;IACtC;;;;;;OAMG;IACH,OAAO,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IACpC;;;;;OAKG;IACH,SAAS,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IACtC;;;;;OAKG;IACH,cAAc,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;CAC3C;AAED;;;;;;;;GAQG;AACH,wBAAsB,sBAAsB,CAC3C,SAAS,EAAE,cAAc,CAAC,WAAW,CAAC,EACtC,MAAM,CAAC,EAAE,WAAW,GAClB,OAAO,CAAC,sBAAsB,CAAC,CAwEjC;AAkCD,wBAAgB,qBAAqB,IAAI,aAAa,CAqJrD;AAED;;GAEG;AACH,eAAO,MAAM,eAAe,eAA0B,CAAA"}