@mailwoman/corpus 10.0.0 → 10.1.0

package/out/pl/tools/fetch/emuia.d.ts

@@ -0,0 +1,165 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * Harvests Poland's INSPIRE Addresses (AD) theme into the GML file that
+ * `#pl/adapters/emuia/adapter` reads.
+ *
+ * Główny Urząd Geodezji i Kartografii publishes the Państwowy Rejestr Granic through MapServer at
+ * {@linkcode PL_EMUIA_WFS}. Its dataset record states one condition of use in `otherConstraints`,
+ * `Brak warunków dostępu i użytkowania` — no conditions for access and use — which the
+ * address-source register elected. No license instrument is named, so the publisher's own words are
+ * the label, and that statement grants the reproduction a paged harvest performs.
+ *
+ * `#tools/fetch/wfs-harvest` owns the page read, the appending, the manifest and the resumption.
+ * What is particular to Poland is five measurements, taken on 2026-10-02:
+ *
+ * 1. `ms:AD.Address` is the only feature type the service publishes, and the payload is the Polish
+ *    register's own schema with every address value inline, so the harvest takes that one type and
+ *    resolves no reference.
+ * 2. The service serves GML and no JSON. Its `GetFeature` `outputFormat` parameter advertises four
+ *    values — `application/gml+xml; version=3.2`, `text/xml; subtype=gml/3.2.1`, `text/xml;
+ *    subtype=gml/3.1.1` and `text/xml; subtype=gml/2.1.2` — and answers with
+ *    `text/xml; subtype="gml/3.2.1"`.
+ * 3. A page holds at most 1,000 features whatever `count` asks for. `count=5000` answers
+ *    `numberReturned="1000"`, and the capabilities document states `CountDefault` as 1000. The page
+ *    size is therefore the service's cap rather than a choice, and a full harvest is one request per
+ *    thousand features.
+ * 4. The service states no count of its own. `resultType=hits` answers `numberMatched="1000"` both
+ *    bare and at `startIndex=1`, which is that same cap, and a feature page answers
+ *    `numberMatched="unknown"`. `readCheckedWFSFeatureCount` proves a reported count against a page
+ *    of the type, and a count equal to the page cap passes that check because no page can exceed the
+ *    cap — a 10-feature probe agrees with 1000. `statedFeatureCount` refuses a count equal to the
+ *    advertised `CountDefault` for that reason, so this service is harvested with no count and the
+ *    harvest ends on the first page that returns no features. The size is read from the service when
+ *    a caller asks for it, by `countWFSFeaturesByPaging`, and no count is written into this file.
+ * 5. The service honors a sort. `ImplementsSorting` is `TRUE`, `sortBy=ms:id` answers HTTP 200 and
+ *    `sortBy=ms:id D` returns the register's last rows first, which proves the parameter is applied.
+ *    `PagingIsTransactionSafe` is `FALSE`, which is the service stating that a paged sequence may
+ *    shift under edits made while it is being read. The sort bounds that shift to the rows whose
+ *    `ms:id` changed rather than leaving the whole sequence unordered.
+ *
+ * A measurement by paging rests on monotone presence: that an index holding a feature implies every
+ * smaller index holds one. A WFS offers no such guarantee, and a bisection over a service where it
+ * fails reports the first gap rather than the extent. The figure is therefore reported with the date
+ * it was taken, never stored as a constant: the register recorded 8,625,921 on 2026-10-01, and on
+ * 2026-10-02 `startIndex=8626950` returned one feature while `startIndex=8626951` returned none, so
+ * the service then held 8,626,951. That is 1,030 more than the day before, which is ordinary
+ * business for a municipal address register.
+ *
+ * At 1,000 features per page and 1,607,180 bytes per page, a full harvest at that size is 8,627
+ * requests and about 13.9 GB.
+ *
+ * The harvest assembles one GML document rather than one file per page, because the adapter opens
+ * `opts.inputPath` as a single markup stream. The first page's root tag opens the file with its
+ * namespace declarations kept and its per-page `numberReturned`, `numberMatched` and `timeStamp`
+ * dropped, since those describe one page and the manifest records them per page. The closing tag is
+ * written when the harvest reaches the end of the type, so a partial harvest is an unclosed document
+ * — which the adapter's scanner still reads member by member, and which a resumed run appends to
+ * after truncating back to the last page boundary.
+ */
+import { type APIClient } from "@mailwoman/core/api";
+import type { PathBuilderLike } from "path-ts";
+import type { BaseFetchOptions, FetchSummary } from "#tools/fetch/download";
+import { type WFSCapabilities } from "#tools/fetch/inspire-addresses";
+import { type FeatureCountStatement, type WFSHarvestManifest } from "#tools/fetch/wfs-harvest";
+/**
+ * The service this module harvests.
+ */
+export declare const PL_EMUIA_WFS = "https://mapy.geoportal.gov.pl/wss/service/INSPIRE/Addresses";
+/**
+ * The one feature type the service publishes, spelled as its capabilities document spells it.
+ */
+export declare const PL_EMUIA_TYPE_NAME = "ms:AD.Address";
+/**
+ * The output format the harvest asks for, which is the first the service advertises.
+ */
+export declare const PL_EMUIA_OUTPUT_FORMAT = "application/gml+xml; version=3.2";
+/**
+ * The property the pages are ordered by.
+ */
+export declare const PL_EMUIA_SORT_BY = "ms:id";
+/**
+ * Features per page.
+ *
+ * The service's own cap rather than a choice: `count=5000` answers with 1,000 features,
+ * and the capabilities document states `CountDefault` as 1000.
+ */
+export declare const PL_EMUIA_PAGE_SIZE = 1000;
+/**
+ * The file the adapter's `inputPath` points at, inside the harvest directory.
+ */
+export declare const PL_EMUIA_HARVEST_FILE = "address.gml";
+export interface HarvestEMUiAOptions {
+    /**
+     * Where the harvest is written.
+     *
+     * The adapter reads {@linkcode PL_EMUIA_HARVEST_FILE} inside it.
+     */
+    outputDir: PathBuilderLike;
+    /**
+     * Features per page, capped at {@linkcode PL_EMUIA_PAGE_SIZE} because the service caps it there.
+     */
+    pageSize?: number;
+    /**
+     * Stop once the manifest holds this many pages, for a probe rather than a full harvest.
+     */
+    maxPages?: number;
+    /**
+     * Measure the service's extent by paging before the harvest starts.
+     *
+     * Off by default.
+     * The measurement costs about 2·log2(n) requests — 51 for a register of this size — and the
+     * harvest does not need it: paging to the first empty page reads the type whole either way.
+     *
+     * It buys a denominator for the progress lines and a figure for the manifest,
+     * and it rests on the monotone-presence assumption this module's header states.
+     */
+    measureCount?: boolean;
+    signal?: AbortSignal;
+    report?: (line: string) => void;
+}
+/**
+ * The count the service states, with how it was obtained.
+ *
+ * The service states none that can be used, so this reports why rather than
+ * carrying its page cap forward as a count.
+ * A caller that asked for {@linkcode HarvestEMUiAOptions.measureCount} gets the
+ * paged measurement instead, with the requests it cost.
+ */
+export declare function readEMUiAFeatureCount(client: Pick<APIClient, "fetch">, options: {
+    measureCount?: boolean;
+    pageCap: number | null;
+}): Promise<FeatureCountStatement>;
+/**
+ * What the service states about itself, with the three claims this harvest depends on checked.
+ *
+ * The type name and the output format are read from the document rather than taken
+ * from this file's constants, so a service that stops publishing either raises here
+ * instead of producing a harvest of something else.
+ *
+ * @throws When the service publishes no Addresses type, does not advertise the GML
+ * format the adapter reads, or does not advertise result paging.
+ */
+export declare function readEMUiACapabilities(client: Pick<APIClient, "fetch">): Promise<WFSCapabilities>;
+/**
+ * Harvests the service into `options.outputDir` and writes the manifest beside the data file.
+ *
+ * @returns The manifest as written.
+ */
+export declare function harvestEMUiAPL(client: Pick<APIClient, "fetch">, options: HarvestEMUiAOptions): Promise<WFSHarvestManifest>;
+export type FetchEMUiAPLOptions = BaseFetchOptions & Pick<HarvestEMUiAOptions, "maxPages" | "measureCount" | "pageSize" | "signal"> & {
+    /**
+     * An injected client, for a test that supplies its own transport.
+     */
+    client?: Pick<APIClient, "fetch">;
+};
+/**
+ * Harvests Poland's addresses under `<outRoot>/emuia/`, for `mailwoman corpus fetch emuia-pl`.
+ *
+ * Re-runnable: a complete harvest whose file still hashes to the manifest's digest makes
+ * no request, and an interrupted one resumes at the page after the last that reached disk.
+ */
+export declare function fetchEMUiAPL(options: FetchEMUiAPLOptions, report?: (line: string) => void): Promise<FetchSummary>;
+//# sourceMappingURL=emuia.d.ts.map

package/out/pl/tools/fetch/emuia.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"emuia.d.ts","sourceRoot":"","sources":["../../../../lib/pl/tools/fetch/emuia.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4DG;AAEH,OAAO,EAAwD,KAAK,SAAS,EAAE,MAAM,qBAAqB,CAAA;AAC1G,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,SAAS,CAAA;AAE9C,OAAO,KAAK,EAAE,gBAAgB,EAAE,YAAY,EAAE,MAAM,uBAAuB,CAAA;AAC3E,OAAO,EAAuB,KAAK,eAAe,EAAE,MAAM,gCAAgC,CAAA;AAC1F,OAAO,EAON,KAAK,qBAAqB,EAE1B,KAAK,kBAAkB,EACvB,MAAM,0BAA0B,CAAA;AAEjC;;GAEG;AACH,eAAO,MAAM,YAAY,gEAAgE,CAAA;AAEzF;;GAEG;AACH,eAAO,MAAM,kBAAkB,kBAAkB,CAAA;AAEjD;;GAEG;AACH,eAAO,MAAM,sBAAsB,qCAAqC,CAAA;AAExE;;GAEG;AACH,eAAO,MAAM,gBAAgB,UAAU,CAAA;AAEvC;;;;;GAKG;AACH,eAAO,MAAM,kBAAkB,OAAO,CAAA;AAEtC;;GAEG;AACH,eAAO,MAAM,qBAAqB,gBAAgB,CAAA;AAkClD,MAAM,WAAW,mBAAmB;IACnC;;;;OAIG;IACH,SAAS,EAAE,eAAe,CAAA;IAC1B;;OAEG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAA;IACjB;;OAEG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAA;IACjB;;;;;;;;;OASG;IACH,YAAY,CAAC,EAAE,OAAO,CAAA;IACtB,MAAM,CAAC,EAAE,WAAW,CAAA;IACpB,MAAM,CAAC,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,CAAA;CAC/B;AAED;;;;;;;GAOG;AACH,wBAAsB,qBAAqB,CAC1C,MAAM,EAAE,IAAI,CAAC,SAAS,EAAE,OAAO,CAAC,EAChC,OAAO,EAAE;IAAE,YAAY,CAAC,EAAE,OAAO,CAAC;IAAC,OAAO,EAAE,MAAM,GAAG,IAAI,CAAA;CAAE,GACzD,OAAO,CAAC,qBAAqB,CAAC,CA8BhC;AAED;;;;;;;;;GASG;AACH,wBAAsB,qBAAqB,CAAC,MAAM,EAAE,IAAI,CAAC,SAAS,EAAE,OAAO,CAAC,GAAG,OAAO,CAAC,eAAe,CAAC,CAsBtG;AAED;;;;GAIG;AACH,wBAAsB,cAAc,CACnC,MAAM,EAAE,IAAI,CAAC,SAAS,EAAE,OAAO,CAAC,EAChC,OAAO,EAAE,mBAAmB,GAC1B,OAAO,CAAC,kBAAkB,CAAC,CAsD7B;AAED,MAAM,MAAM,mBAAmB,GAAG,gBAAgB,GACjD,IAAI,CAAC,mBAAmB,EAAE,UAAU,GAAG,cAAc,GAAG,UAAU,GAAG,QAAQ,CAAC,GAAG;IAChF;;OAEG;IACH,MAAM,CAAC,EAAE,IAAI,CAAC,SAAS,EAAE,OAAO,CAAC,CAAA;CACjC,CAAA;AAEF;;;;;GAKG;AACH,wBAAsB,YAAY,CACjC,OAAO,EAAE,mBAAmB,EAC5B,MAAM,CAAC,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,GAC7B,OAAO,CAAC,YAAY,CAAC,CA2BvB"}

package/out/pl/tools/fetch/emuia.js

@@ -0,0 +1,260 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * Harvests Poland's INSPIRE Addresses (AD) theme into the GML file that
+ * `#pl/adapters/emuia/adapter` reads.
+ *
+ * Główny Urząd Geodezji i Kartografii publishes the Państwowy Rejestr Granic through MapServer at
+ * {@linkcode PL_EMUIA_WFS}. Its dataset record states one condition of use in `otherConstraints`,
+ * `Brak warunków dostępu i użytkowania` — no conditions for access and use — which the
+ * address-source register elected. No license instrument is named, so the publisher's own words are
+ * the label, and that statement grants the reproduction a paged harvest performs.
+ *
+ * `#tools/fetch/wfs-harvest` owns the page read, the appending, the manifest and the resumption.
+ * What is particular to Poland is five measurements, taken on 2026-10-02:
+ *
+ * 1. `ms:AD.Address` is the only feature type the service publishes, and the payload is the Polish
+ *    register's own schema with every address value inline, so the harvest takes that one type and
+ *    resolves no reference.
+ * 2. The service serves GML and no JSON. Its `GetFeature` `outputFormat` parameter advertises four
+ *    values — `application/gml+xml; version=3.2`, `text/xml; subtype=gml/3.2.1`, `text/xml;
+ *    subtype=gml/3.1.1` and `text/xml; subtype=gml/2.1.2` — and answers with
+ *    `text/xml; subtype="gml/3.2.1"`.
+ * 3. A page holds at most 1,000 features whatever `count` asks for. `count=5000` answers
+ *    `numberReturned="1000"`, and the capabilities document states `CountDefault` as 1000. The page
+ *    size is therefore the service's cap rather than a choice, and a full harvest is one request per
+ *    thousand features.
+ * 4. The service states no count of its own. `resultType=hits` answers `numberMatched="1000"` both
+ *    bare and at `startIndex=1`, which is that same cap, and a feature page answers
+ *    `numberMatched="unknown"`. `readCheckedWFSFeatureCount` proves a reported count against a page
+ *    of the type, and a count equal to the page cap passes that check because no page can exceed the
+ *    cap — a 10-feature probe agrees with 1000. `statedFeatureCount` refuses a count equal to the
+ *    advertised `CountDefault` for that reason, so this service is harvested with no count and the
+ *    harvest ends on the first page that returns no features. The size is read from the service when
+ *    a caller asks for it, by `countWFSFeaturesByPaging`, and no count is written into this file.
+ * 5. The service honors a sort. `ImplementsSorting` is `TRUE`, `sortBy=ms:id` answers HTTP 200 and
+ *    `sortBy=ms:id D` returns the register's last rows first, which proves the parameter is applied.
+ *    `PagingIsTransactionSafe` is `FALSE`, which is the service stating that a paged sequence may
+ *    shift under edits made while it is being read. The sort bounds that shift to the rows whose
+ *    `ms:id` changed rather than leaving the whole sequence unordered.
+ *
+ * A measurement by paging rests on monotone presence: that an index holding a feature implies every
+ * smaller index holds one. A WFS offers no such guarantee, and a bisection over a service where it
+ * fails reports the first gap rather than the extent. The figure is therefore reported with the date
+ * it was taken, never stored as a constant: the register recorded 8,625,921 on 2026-10-01, and on
+ * 2026-10-02 `startIndex=8626950` returned one feature while `startIndex=8626951` returned none, so
+ * the service then held 8,626,951. That is 1,030 more than the day before, which is ordinary
+ * business for a municipal address register.
+ *
+ * At 1,000 features per page and 1,607,180 bytes per page, a full harvest at that size is 8,627
+ * requests and about 13.9 GB.
+ *
+ * The harvest assembles one GML document rather than one file per page, because the adapter opens
+ * `opts.inputPath` as a single markup stream. The first page's root tag opens the file with its
+ * namespace declarations kept and its per-page `numberReturned`, `numberMatched` and `timeStamp`
+ * dropped, since those describe one page and the manifest records them per page. The closing tag is
+ * written when the harvest reaches the end of the type, so a partial harvest is an unclosed document
+ * — which the adapter's scanner still reads member by member, and which a resumed run appends to
+ * after truncating back to the last page boundary.
+ */
+import { countWFSFeaturesByPaging, readCheckedWFSFeatureCount } from "@mailwoman/core/api";
+import { readWFSCapabilities } from "#tools/fetch/inspire-addresses";
+import { collectionOpeningTag, harvestPagedWFS, runWFSHarvest, readWFSMarkupPage, rootContent, statedFeatureCount, } from "#tools/fetch/wfs-harvest";
+/**
+ * The service this module harvests.
+ */
+export const PL_EMUIA_WFS = "https://mapy.geoportal.gov.pl/wss/service/INSPIRE/Addresses";
+/**
+ * The one feature type the service publishes, spelled as its capabilities document spells it.
+ */
+export const PL_EMUIA_TYPE_NAME = "ms:AD.Address";
+/**
+ * The output format the harvest asks for, which is the first the service advertises.
+ */
+export const PL_EMUIA_OUTPUT_FORMAT = "application/gml+xml; version=3.2";
+/**
+ * The property the pages are ordered by.
+ */
+export const PL_EMUIA_SORT_BY = "ms:id";
+/**
+ * Features per page.
+ *
+ * The service's own cap rather than a choice: `count=5000` answers with 1,000 features,
+ * and the capabilities document states `CountDefault` as 1000.
+ */
+export const PL_EMUIA_PAGE_SIZE = 1000;
+/**
+ * The file the adapter's `inputPath` points at, inside the harvest directory.
+ */
+export const PL_EMUIA_HARVEST_FILE = "address.gml";
+/**
+ * The XML declaration the assembled document opens with.
+ *
+ * The service's own pages declare UTF-8, and the harvest is written as UTF-8.
+ */
+const XML_DECLARATION = '<?xml version="1.0" encoding="UTF-8"?>\n';
+/**
+ * The adapter slug this module fetches for, and the directory it writes under.
+ */
+const SLUG = "emuia";
+/**
+ * Names the service in every refusal this module raises.
+ */
+const CONTEXT = "pl emuia wfs";
+/**
+ * The terms the address-source register elected for this publisher.
+ */
+const LICENSE = "Brak warunków dostępu i użytkowania";
+/**
+ * The spacing between requests.
+ *
+ * The service answers a 1,000-feature page in about 2 seconds and a page at a deep `startIndex`
+ * in about 6, so its own latency already spaces a serial harvest further apart than this.
+ * The interval is the ceiling a run is held to rather than the rate it achieves, and one
+ * request per second against a government host whose terms state no rate is the ceiling chosen.
+ */
+const MIN_REQUEST_INTERVAL_MS = 1000;
+/**
+ * The count the service states, with how it was obtained.
+ *
+ * The service states none that can be used, so this reports why rather than
+ * carrying its page cap forward as a count.
+ * A caller that asked for {@linkcode HarvestEMUiAOptions.measureCount} gets the
+ * paged measurement instead, with the requests it cost.
+ */
+export async function readEMUiAFeatureCount(client, options) {
+    const stated = statedFeatureCount(await readCheckedWFSFeatureCount(client, {
+        wfsURL: PL_EMUIA_WFS,
+        typeNames: PL_EMUIA_TYPE_NAME,
+        context: CONTEXT,
+    }), options.pageCap);
+    if (stated.count !== null)
+        return stated;
+    if (!options.measureCount) {
+        return {
+            count: null,
+            because: `${stated.because}, and the harvest ends on the first page that returns no features`,
+        };
+    }
+    const measured = await countWFSFeaturesByPaging(client, {
+        wfsURL: PL_EMUIA_WFS,
+        typeNames: PL_EMUIA_TYPE_NAME,
+        outputFormat: PL_EMUIA_OUTPUT_FORMAT,
+        context: CONTEXT,
+    });
+    return {
+        count: measured.count,
+        because: `${stated.because}, so the extent was measured by paging in ${measured.requests} requests, ${measured.confirmed ? "confirmed by a page straddling the last index" : "unconfirmed by the straddling page, which returned fewer features than the count implies"}, and the measurement assumes that an index holding a feature implies every smaller index holds one`,
+    };
+}
+/**
+ * What the service states about itself, with the three claims this harvest depends on checked.
+ *
+ * The type name and the output format are read from the document rather than taken
+ * from this file's constants, so a service that stops publishing either raises here
+ * instead of producing a harvest of something else.
+ *
+ * @throws When the service publishes no Addresses type, does not advertise the GML
+ * format the adapter reads, or does not advertise result paging.
+ */
+export async function readEMUiACapabilities(client) {
+    const capabilities = await readWFSCapabilities(client, { wfsURL: PL_EMUIA_WFS, context: CONTEXT });
+    if (!capabilities.typeNames.Address) {
+        throw new Error(`${CONTEXT}: the capabilities document publishes no Addresses feature type, so ${PL_EMUIA_TYPE_NAME} could not be confirmed`);
+    }
+    if (!capabilities.outputFormats.includes(PL_EMUIA_OUTPUT_FORMAT)) {
+        throw new Error(`${CONTEXT}: the service does not advertise ${PL_EMUIA_OUTPUT_FORMAT}, which is the format the adapter reads — it advertises ${capabilities.outputFormats.join(", ")}`);
+    }
+    if (!capabilities.supportsPaging) {
+        throw new Error(`${CONTEXT}: the service does not advertise ImplementsResultPaging, and such a service answers every page with the first, so a paged harvest would store one page repeated`);
+    }
+    return capabilities;
+}
+/**
+ * Harvests the service into `options.outputDir` and writes the manifest beside the data file.
+ *
+ * @returns The manifest as written.
+ */
+export async function harvestEMUiAPL(client, options) {
+    const capabilities = await readEMUiACapabilities(client);
+    const typeName = capabilities.typeNames.Address ?? PL_EMUIA_TYPE_NAME;
+    const featureCount = await readEMUiAFeatureCount(client, {
+        measureCount: options.measureCount,
+        pageCap: capabilities.countDefault,
+    });
+    options.report?.(`  Feature count: ${featureCount.count ?? "not stated"} — ${featureCount.because}`);
+    return await harvestPagedWFS({
+        context: CONTEXT,
+        source: SLUG,
+        outputDir: options.outputDir,
+        filename: PL_EMUIA_HARVEST_FILE,
+        request: {
+            wfsURL: PL_EMUIA_WFS,
+            typeName,
+            outputFormat: PL_EMUIA_OUTPUT_FORMAT,
+            sortBy: PL_EMUIA_SORT_BY,
+        },
+        license: LICENSE,
+        // The elected statement names no attribution clause.
+        // Crediting GUGiK stays good practice.
+        attribution: "",
+        featureCount,
+        pageSize: Math.min(options.pageSize ?? PL_EMUIA_PAGE_SIZE, PL_EMUIA_PAGE_SIZE),
+        maxPages: options.maxPages,
+        signal: options.signal,
+        report: options.report,
+        readPage: async ({ startIndex, count }) => {
+            const page = await readWFSMarkupPage(client, {
+                wfsURL: PL_EMUIA_WFS,
+                context: CONTEXT,
+                params: {
+                    typeNames: typeName,
+                    outputFormat: PL_EMUIA_OUTPUT_FORMAT,
+                    sortBy: PL_EMUIA_SORT_BY,
+                    count: String(count),
+                    startIndex: String(startIndex),
+                },
+            });
+            return {
+                payload: rootContent(page.body, page.root, CONTEXT),
+                header: `${XML_DECLARATION}${collectionOpeningTag(page.root)}`,
+                footer: `</${page.root.name}>\n`,
+                numberReturned: page.numberReturned,
+                numberMatched: page.numberMatched,
+                retrievedAt: page.timeStamp,
+            };
+        },
+    });
+}
+/**
+ * Harvests Poland's addresses under `<outRoot>/emuia/`, for `mailwoman corpus fetch emuia-pl`.
+ *
+ * Re-runnable: a complete harvest whose file still hashes to the manifest's digest makes
+ * no request, and an interrupted one resumes at the page after the last that reached disk.
+ */
+export async function fetchEMUiAPL(options, report) {
+    // The service caps a page at 1,000, so a larger request is answered with 1,000 either way
+    // and the manifest records what was asked for rather than what arrived.
+    const pageSize = Math.min(options.pageSize ?? PL_EMUIA_PAGE_SIZE, PL_EMUIA_PAGE_SIZE);
+    return await runWFSHarvest({
+        slug: SLUG,
+        outRoot: options.outRoot,
+        filename: PL_EMUIA_HARVEST_FILE,
+        expect: { wfsURL: PL_EMUIA_WFS, sortBy: PL_EMUIA_SORT_BY, pageSize },
+        displayName: "pl-emuia",
+        minRequestIntervalMs: MIN_REQUEST_INTERVAL_MS,
+        client: options.client,
+        maxPages: options.maxPages,
+        harvest: async (client) => await harvestEMUiAPL(client, {
+            outputDir: options.outRoot(SLUG),
+            pageSize,
+            maxPages: options.maxPages,
+            measureCount: options.measureCount,
+            signal: options.signal,
+            report,
+        }),
+    }, report);
+}
+//# sourceMappingURL=emuia.js.map

package/out/pl/tools/fetch/emuia.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"emuia.js","sourceRoot":"","sources":["../../../../lib/pl/tools/fetch/emuia.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4DG;AAEH,OAAO,EAAE,wBAAwB,EAAE,0BAA0B,EAAkB,MAAM,qBAAqB,CAAA;AAI1G,OAAO,EAAE,mBAAmB,EAAwB,MAAM,gCAAgC,CAAA;AAC1F,OAAO,EACN,oBAAoB,EACpB,eAAe,EACf,aAAa,EACb,iBAAiB,EACjB,WAAW,EACX,kBAAkB,GAIlB,MAAM,0BAA0B,CAAA;AAEjC;;GAEG;AACH,MAAM,CAAC,MAAM,YAAY,GAAG,6DAA6D,CAAA;AAEzF;;GAEG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAG,eAAe,CAAA;AAEjD;;GAEG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAAG,kCAAkC,CAAA;AAExE;;GAEG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAG,OAAO,CAAA;AAEvC;;;;;GAKG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAG,IAAI,CAAA;AAEtC;;GAEG;AACH,MAAM,CAAC,MAAM,qBAAqB,GAAG,aAAa,CAAA;AAElD;;;;GAIG;AACH,MAAM,eAAe,GAAG,0CAA0C,CAAA;AAElE;;GAEG;AACH,MAAM,IAAI,GAAG,OAAO,CAAA;AAEpB;;GAEG;AACH,MAAM,OAAO,GAAG,cAAc,CAAA;AAE9B;;GAEG;AACH,MAAM,OAAO,GAAG,qCAAqC,CAAA;AAErD;;;;;;;GAOG;AACH,MAAM,uBAAuB,GAAG,IAAI,CAAA;AAgCpC;;;;;;;GAOG;AACH,MAAM,CAAC,KAAK,UAAU,qBAAqB,CAC1C,MAAgC,EAChC,OAA2D;IAE3D,MAAM,MAAM,GAAG,kBAAkB,CAChC,MAAM,0BAA0B,CAAC,MAAM,EAAE;QACxC,MAAM,EAAE,YAAY;QACpB,SAAS,EAAE,kBAAkB;QAC7B,OAAO,EAAE,OAAO;KAChB,CAAC,EACF,OAAO,CAAC,OAAO,CACf,CAAA;IAED,IAAI,MAAM,CAAC,KAAK,KAAK,IAAI;QAAE,OAAO,MAAM,CAAA;IAExC,IAAI,CAAC,OAAO,CAAC,YAAY,EAAE,CAAC;QAC3B,OAAO;YACN,KAAK,EAAE,IAAI;YACX,OAAO,EAAE,GAAG,MAAM,CAAC,OAAO,mEAAmE;SAC7F,CAAA;IACF,CAAC;IAED,MAAM,QAAQ,GAAG,MAAM,wBAAwB,CAAC,MAAM,EAAE;QACvD,MAAM,EAAE,YAAY;QACpB,SAAS,EAAE,kBAAkB;QAC7B,YAAY,EAAE,sBAAsB;QACpC,OAAO,EAAE,OAAO;KAChB,CAAC,CAAA;IAEF,OAAO;QACN,KAAK,EAAE,QAAQ,CAAC,KAAK;QACrB,OAAO,EAAE,GAAG,MAAM,CAAC,OAAO,6CAA6C,QAAQ,CAAC,QAAQ,cAAc,QAAQ,CAAC,SAAS,CAAC,CAAC,CAAC,+CAA+C,CAAC,CAAC,CAAC,0FAA0F,qGAAqG;KAC5W,CAAA;AACF,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,CAAC,KAAK,UAAU,qBAAqB,CAAC,MAAgC;IAC3E,MAAM,YAAY,GAAG,MAAM,mBAAmB,CAAC,MAAM,EAAE,EAAE,MAAM,EAAE,YAAY,EAAE,OAAO,EAAE,OAAO,EAAE,CAAC,CAAA;IAElG,IAAI,CAAC,YAAY,CAAC,SAAS,CAAC,OAAO,EAAE,CAAC;QACrC,MAAM,IAAI,KAAK,CACd,GAAG,OAAO,uEAAuE,kBAAkB,yBAAyB,CAC5H,CAAA;IACF,CAAC;IAED,IAAI,CAAC,YAAY,CAAC,aAAa,CAAC,QAAQ,CAAC,sBAAsB,CAAC,EAAE,CAAC;QAClE,MAAM,IAAI,KAAK,CACd,GAAG,OAAO,oCAAoC,sBAAsB,2DAA2D,YAAY,CAAC,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CACtK,CAAA;IACF,CAAC;IAED,IAAI,CAAC,YAAY,CAAC,cAAc,EAAE,CAAC;QAClC,MAAM,IAAI,KAAK,CACd,GAAG,OAAO,iKAAiK,CAC3K,CAAA;IACF,CAAC;IAED,OAAO,YAAY,CAAA;AACpB,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,cAAc,CACnC,MAAgC,EAChC,OAA4B;IAE5B,MAAM,YAAY,GAAG,MAAM,qBAAqB,CAAC,MAAM,CAAC,CAAA;IACxD,MAAM,QAAQ,GAAG,YAAY,CAAC,SAAS,CAAC,OAAO,IAAI,kBAAkB,CAAA;IAErE,MAAM,YAAY,GAAG,MAAM,qBAAqB,CAAC,MAAM,EAAE;QACxD,YAAY,EAAE,OAAO,CAAC,YAAY;QAClC,OAAO,EAAE,YAAY,CAAC,YAAY;KAClC,CAAC,CAAA;IAEF,OAAO,CAAC,MAAM,EAAE,CAAC,oBAAoB,YAAY,CAAC,KAAK,IAAI,YAAY,MAAM,YAAY,CAAC,OAAO,EAAE,CAAC,CAAA;IAEpG,OAAO,MAAM,eAAe,CAAC;QAC5B,OAAO,EAAE,OAAO;QAChB,MAAM,EAAE,IAAI;QACZ,SAAS,EAAE,OAAO,CAAC,SAAS;QAC5B,QAAQ,EAAE,qBAAqB;QAC/B,OAAO,EAAE;YACR,MAAM,EAAE,YAAY;YACpB,QAAQ;YACR,YAAY,EAAE,sBAAsB;YACpC,MAAM,EAAE,gBAAgB;SACxB;QACD,OAAO,EAAE,OAAO;QAChB,qDAAqD;QACrD,uCAAuC;QACvC,WAAW,EAAE,EAAE;QACf,YAAY;QACZ,QAAQ,EAAE,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,QAAQ,IAAI,kBAAkB,EAAE,kBAAkB,CAAC;QAC9E,QAAQ,EAAE,OAAO,CAAC,QAAQ;QAC1B,MAAM,EAAE,OAAO,CAAC,MAAM;QACtB,MAAM,EAAE,OAAO,CAAC,MAAM;QACtB,QAAQ,EAAE,KAAK,EAAE,EAAE,UAAU,EAAE,KAAK,EAAE,EAA6B,EAAE;YACpE,MAAM,IAAI,GAAG,MAAM,iBAAiB,CAAC,MAAM,EAAE;gBAC5C,MAAM,EAAE,YAAY;gBACpB,OAAO,EAAE,OAAO;gBAChB,MAAM,EAAE;oBACP,SAAS,EAAE,QAAQ;oBACnB,YAAY,EAAE,sBAAsB;oBACpC,MAAM,EAAE,gBAAgB;oBACxB,KAAK,EAAE,MAAM,CAAC,KAAK,CAAC;oBACpB,UAAU,EAAE,MAAM,CAAC,UAAU,CAAC;iBAC9B;aACD,CAAC,CAAA;YAEF,OAAO;gBACN,OAAO,EAAE,WAAW,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,OAAO,CAAC;gBACnD,MAAM,EAAE,GAAG,eAAe,GAAG,oBAAoB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE;gBAC9D,MAAM,EAAE,KAAK,IAAI,CAAC,IAAI,CAAC,IAAI,KAAK;gBAChC,cAAc,EAAE,IAAI,CAAC,cAAc;gBACnC,aAAa,EAAE,IAAI,CAAC,aAAa;gBACjC,WAAW,EAAE,IAAI,CAAC,SAAS;aAC3B,CAAA;QACF,CAAC;KACD,CAAC,CAAA;AACH,CAAC;AAUD;;;;;GAKG;AACH,MAAM,CAAC,KAAK,UAAU,YAAY,CACjC,OAA4B,EAC5B,MAA+B;IAE/B,0FAA0F;IAC1F,wEAAwE;IACxE,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,QAAQ,IAAI,kBAAkB,EAAE,kBAAkB,CAAC,CAAA;IAErF,OAAO,MAAM,aAAa,CACzB;QACC,IAAI,EAAE,IAAI;QACV,OAAO,EAAE,OAAO,CAAC,OAAO;QACxB,QAAQ,EAAE,qBAAqB;QAC/B,MAAM,EAAE,EAAE,MAAM,EAAE,YAAY,EAAE,MAAM,EAAE,gBAAgB,EAAE,QAAQ,EAAE;QACpE,WAAW,EAAE,UAAU;QACvB,oBAAoB,EAAE,uBAAuB;QAC7C,MAAM,EAAE,OAAO,CAAC,MAAM;QACtB,QAAQ,EAAE,OAAO,CAAC,QAAQ;QAC1B,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,EAAE,CACzB,MAAM,cAAc,CAAC,MAAM,EAAE;YAC5B,SAAS,EAAE,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC;YAChC,QAAQ;YACR,QAAQ,EAAE,OAAO,CAAC,QAAQ;YAC1B,YAAY,EAAE,OAAO,CAAC,YAAY;YAClC,MAAM,EAAE,OAAO,CAAC,MAAM;YACtB,MAAM;SACN,CAAC;KACH,EACD,MAAM,CACN,CAAA;AACF,CAAC"}

package/out/recipes/anchor-absorption.d.ts

@@ -3,18 +3,18 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   `anchor-absorption` slice recipe (#220/#723, Probe A1) — self-generates `--count` rows from
- *   {@link synthesizeAnchorAbsorptionRow}'s 6-slice mix (CASE-H / CASE-P-us-rural / CASE-P-de /
- *   anchor-fp / locale-ambig / standard), aligns each to BIO, and emits a labeled JSONL. The
- *   LEADING 5-digit on CASE-H/anchor-fp/locale-ambig is sampled from the REAL US ZIPs in the
- *   postcode-anchor lookup, so the shaped-painted anchor fires on it exactly as inference does —
- *   teaching the model to OVERRIDE a present anchor from context. Ported from
- *   scripts/build-anchor-absorption-slice.mjs.
+ *   `anchor-absorption` recipe, a self-generated `--count`-row source built from
+ *   {@link synthesizeAnchorAbsorptionRow}'s six-template mix (case-H / case-P-us-rural / case-P-de /
+ *   anchor-fp / locale-ambig / standard), aligned to BIO and emitted as a labeled jsonl. The
+ *   leading 5-digit on case-H/anchor-fp/locale-ambig is sampled from the real US ZIPs in the
+ *   postcode-anchor lookup, so the shaped-painted anchor fires on it exactly as inference does.
  */
 import { type CorpusRecipe } from "#recipes/scaffold";
 /**
- * Slice recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
- * and `description` below for the surface form it generates.
+ * Recipe registered with the corpus builder.
+ *
+ * See the file header for the parse behavior it exercises.
+ * See `description` below for the generated surface form.
  */
 export declare const anchorAbsorptionRecipe: CorpusRecipe;
 //# sourceMappingURL=anchor-absorption.d.ts.map

package/out/recipes/anchor-absorption.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"anchor-absorption.d.ts","sourceRoot":"","sources":["../../lib/recipes/anchor-absorption.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AAOH,OAAO,EAAiB,KAAK,YAAY,EAAE,MAAM,mBAAmB,CAAA;AAyBpE;;;GAGG;AACH,eAAO,MAAM,sBAAsB,EAAE,YAoDpC,CAAA"}
+{"version":3,"file":"anchor-absorption.d.ts","sourceRoot":"","sources":["../../lib/recipes/anchor-absorption.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAQH,OAAO,EAAkB,KAAK,YAAY,EAAE,MAAM,mBAAmB,CAAA;AA6BrE;;;;;GAKG;AACH,eAAO,MAAM,sBAAsB,EAAE,YA0DpC,CAAA"}

package/out/recipes/anchor-absorption.js

@@ -3,23 +3,25 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   `anchor-absorption` slice recipe (#220/#723, Probe A1) — self-generates `--count` rows from
- *   {@link synthesizeAnchorAbsorptionRow}'s 6-slice mix (CASE-H / CASE-P-us-rural / CASE-P-de /
- *   anchor-fp / locale-ambig / standard), aligns each to BIO, and emits a labeled JSONL. The
- *   LEADING 5-digit on CASE-H/anchor-fp/locale-ambig is sampled from the REAL US ZIPs in the
- *   postcode-anchor lookup, so the shaped-painted anchor fires on it exactly as inference does —
- *   teaching the model to OVERRIDE a present anchor from context. Ported from
- *   scripts/build-anchor-absorption-slice.mjs.
+ *   `anchor-absorption` recipe, a self-generated `--count`-row source built from
+ *   {@link synthesizeAnchorAbsorptionRow}'s six-template mix (case-H / case-P-us-rural / case-P-de /
+ *   anchor-fp / locale-ambig / standard), aligned to BIO and emitted as a labeled jsonl. The
+ *   leading 5-digit on case-H/anchor-fp/locale-ambig is sampled from the real US ZIPs in the
+ *   postcode-anchor lookup, so the shaped-painted anchor fires on it exactly as inference does.
  */
 import { dataRootPath } from "@mailwoman/core/data-root";
 import { readLocalJSONFile } from "@mailwoman/core/fs/readers";
+import { stringifyJSON } from "@mailwoman/core/json";
 import { makeLcg } from "@mailwoman/core/utils";
-import { sliceSourceID } from "#recipes/scaffold";
+import { recipeSourceID } from "#recipes/scaffold";
+import { defaultRecipeSource } from "#recipes/sources";
 import { synthesizeAnchorAbsorptionRow } from "#synthesizers/anchor-absorption";
+import { SurfaceOrigin } from "#types";
 import { alignRow } from "#utils";
 /**
- * The leading-5-digit source: the real US ZIPs in the postcode anchor's pilot lookup. Resolved through the data-root
- * helper (the lab default is `$MAILWOMAN_DATA_ROOT`), never re-hardcoded.
+ * The leading-5-digit source: the real US ZIPs in the postcode anchor's pilot lookup.
+ *
+ * Resolved through the data-root helper (the lab default is `$MAILWOMAN_DATA_ROOT`), never re-hardcoded.
  */
 const ANCHOR_LOOKUP = dataRootPath("anchor", "pilot-anchor-lookup.json");
 /**
@@ -36,17 +38,18 @@ async function loadRealUsZips(path) {
     return zips;
 }
 /**
- * Slice recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
- * and `description` below for the surface form it generates.
+ * Recipe registered with the corpus builder.
+ *
+ * See the file header for the parse behavior it exercises.
+ * See `description` below for the generated surface form.
  */
 export const anchorAbsorptionRecipe = {
     name: "anchor-absorption",
-    description: "Anchor-absorption counter-augmentation (#220/#723): 6-slice mix → synthesizeAnchorAbsorptionRow",
+    description: "Anchor-absorption counter-augmentation (#220/#723): six-template mix → synthesizeAnchorAbsorptionRow",
     mode: "generate",
     async run(opts, write) {
-        // Emit PRNG: the legacy build-anchor-absorption-slice.mjs seeded an LCG (lcg(opts.seed)).
         const random = makeLcg(opts.seed);
-        const source = opts.sourceName ?? "synth-anchor-absorption";
+        const source = opts.sourceName ?? defaultRecipeSource("synth-anchor-absorption");
         const count = opts.count ?? 50_000;
         const realZips = await loadRealUsZips(ANCHOR_LOOKUP);
         console.error(`Loaded ${realZips.length} real US ZIPs from the anchor lookup (the leading-5-digit source).`);
@@ -56,26 +59,33 @@ export const anchorAbsorptionRecipe = {
         for (let i = 0; i < count; i++) {
             const synth = synthesizeAnchorAbsorptionRow({ random, realZips });
             const country = synth.locale.split("-")[1];
-            // "en-US" -> "US", "de-DE" -> "DE"
             const canonical = {
                 raw: synth.raw,
                 components: synth.components,
                 country,
                 locale: synth.locale,
                 source,
-                source_id: sliceSourceID(source, { v: String(i) }),
+                source_id: recipeSourceID(source, { v: String(i) }),
             };
             const aligned = alignRow(canonical);
             if (aligned.kind !== "labeled") {
                 quarantined++;
                 continue;
             }
-            write(JSON.stringify({ ...aligned.row, synth_method: "anchor-absorption", synth_template: synth.template }) + "\n");
+            write(stringifyJSON({
+                ...aligned.row,
+                recipe: "anchor-absorption",
+                template: synth.template,
+                // The components are drawn from weighted template tables rather than read from a
+                // register, so the row teaches the absorption shape and names no published address.
+                register: null,
+                surface: SurfaceOrigin.Invented,
+            }));
             written++;
             byTemplate[synth.template] = (byTemplate[synth.template] ?? 0) + 1;
         }
         console.error(`\nwrote ${written} rows (${quarantined} quarantined)`);
-        console.error("  by slice:", JSON.stringify(byTemplate));
+        console.error("  by template:", stringifyJSON(byTemplate));
         return { emitted: written, skipped: quarantined };
     },
 };

package/out/recipes/anchor-absorption.js.map

@@ -1 +1 @@
-{"version":3,"file":"anchor-absorption.js","sourceRoot":"","sources":["../../lib/recipes/anchor-absorption.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AAEH,OAAO,EAAE,YAAY,EAAE,MAAM,2BAA2B,CAAA;AACxD,OAAO,EAAE,iBAAiB,EAAE,MAAM,4BAA4B,CAAA;AAC9D,OAAO,EAAE,OAAO,EAAE,MAAM,uBAAuB,CAAA;AAG/C,OAAO,EAAE,aAAa,EAAqB,MAAM,mBAAmB,CAAA;AACpE,OAAO,EAAE,6BAA6B,EAAE,MAAM,iCAAiC,CAAA;AAC/E,OAAO,EAAE,QAAQ,EAAE,MAAM,QAAQ,CAAA;AACjC;;;GAGG;AACH,MAAM,aAAa,GAAG,YAAY,CAAC,QAAQ,EAAE,0BAA0B,CAAC,CAAA;AAExE;;GAEG;AACH,KAAK,UAAU,cAAc,CAAC,IAAqB;IAClD,MAAM,CAAC,GAAG,MAAM,iBAAiB,CAA0B,IAAI,CAAC,CAAA;IAChE,MAAM,IAAI,GAAa,EAAE,CAAA;IAEzB,KAAK,MAAM,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC;QACzC,IAAI,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,OAAO,CAAC,CAAC,CAAC,CAAC,KAAK,QAAQ,IAAI,IAAI,IAAK,CAAC,CAAC,CAAC,CAAY,IAAI,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,CAAC;YAC5G,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;QACd,CAAC;IACF,CAAC;IAED,OAAO,IAAI,CAAA;AACZ,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAAiB;IACnD,IAAI,EAAE,mBAAmB;IACzB,WAAW,EAAE,iGAAiG;IAC9G,IAAI,EAAE,UAAU;IAChB,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,0FAA0F;QAC1F,MAAM,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACjC,MAAM,MAAM,GAAG,IAAI,CAAC,UAAU,IAAI,yBAAyB,CAAA;QAC3D,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,IAAI,MAAM,CAAA;QAClC,MAAM,QAAQ,GAAG,MAAM,cAAc,CAAC,aAAa,CAAC,CAAA;QAEpD,OAAO,CAAC,KAAK,CAAC,UAAU,QAAQ,CAAC,MAAM,oEAAoE,CAAC,CAAA;QAE5G,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,WAAW,GAAG,CAAC,CAAA;QACnB,MAAM,UAAU,GAA2B,EAAE,CAAA;QAE7C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,EAAE,CAAC,EAAE,EAAE,CAAC;YAChC,MAAM,KAAK,GAAG,6BAA6B,CAAC,EAAE,MAAM,EAAE,QAAQ,EAAE,CAAC,CAAA;YACjE,MAAM,OAAO,GAAG,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAA;YAE1C,mCAAmC;YACnC,MAAM,SAAS,GAAG;gBACjB,GAAG,EAAE,KAAK,CAAC,GAAG;gBACd,UAAU,EAAE,KAAK,CAAC,UAAU;gBAC5B,OAAO;gBACP,MAAM,EAAE,KAAK,CAAC,MAAM;gBACpB,MAAM;gBACN,SAAS,EAAE,aAAa,CAAC,MAAM,EAAE,EAAE,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC;aAClD,CAAA;YAED,MAAM,OAAO,GAAG,QAAQ,CAAC,SAA2C,CAAC,CAAA;YAErE,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;gBAChC,WAAW,EAAE,CAAA;gBAEb,SAAQ;YACT,CAAC;YAED,KAAK,CACJ,IAAI,CAAC,SAAS,CAAC,EAAE,GAAG,OAAO,CAAC,GAAG,EAAE,YAAY,EAAE,mBAAmB,EAAE,cAAc,EAAE,KAAK,CAAC,QAAQ,EAAE,CAAC,GAAG,IAAI,CAC5G,CAAA;YAED,OAAO,EAAE,CAAA;YACT,UAAU,CAAC,KAAK,CAAC,QAAQ,CAAC,GAAG,CAAC,UAAU,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;QACnE,CAAC;QAED,OAAO,CAAC,KAAK,CAAC,WAAW,OAAO,UAAU,WAAW,eAAe,CAAC,CAAA;QACrE,OAAO,CAAC,KAAK,CAAC,aAAa,EAAE,IAAI,CAAC,SAAS,CAAC,UAAU,CAAC,CAAC,CAAA;QAExD,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,WAAW,EAAE,CAAA;IAClD,CAAC;CACD,CAAA"}
+{"version":3,"file":"anchor-absorption.js","sourceRoot":"","sources":["../../lib/recipes/anchor-absorption.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAEH,OAAO,EAAE,YAAY,EAAE,MAAM,2BAA2B,CAAA;AACxD,OAAO,EAAE,iBAAiB,EAAE,MAAM,4BAA4B,CAAA;AAC9D,OAAO,EAAE,aAAa,EAAE,MAAM,sBAAsB,CAAA;AACpD,OAAO,EAAE,OAAO,EAAE,MAAM,uBAAuB,CAAA;AAG/C,OAAO,EAAE,cAAc,EAAqB,MAAM,mBAAmB,CAAA;AACrE,OAAO,EAAE,mBAAmB,EAAE,MAAM,kBAAkB,CAAA;AACtD,OAAO,EAAE,6BAA6B,EAAE,MAAM,iCAAiC,CAAA;AAC/E,OAAO,EAAE,aAAa,EAAE,MAAM,QAAQ,CAAA;AACtC,OAAO,EAAE,QAAQ,EAAE,MAAM,QAAQ,CAAA;AAEjC;;;;GAIG;AACH,MAAM,aAAa,GAAG,YAAY,CAAC,QAAQ,EAAE,0BAA0B,CAAC,CAAA;AAExE;;GAEG;AACH,KAAK,UAAU,cAAc,CAAC,IAAqB;IAClD,MAAM,CAAC,GAAG,MAAM,iBAAiB,CAA0B,IAAI,CAAC,CAAA;IAChE,MAAM,IAAI,GAAa,EAAE,CAAA;IAEzB,KAAK,MAAM,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC;QACzC,IAAI,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,OAAO,CAAC,CAAC,CAAC,CAAC,KAAK,QAAQ,IAAI,IAAI,IAAK,CAAC,CAAC,CAAC,CAAY,IAAI,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,CAAC;YAC5G,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;QACd,CAAC;IACF,CAAC;IAED,OAAO,IAAI,CAAA;AACZ,CAAC;AAED;;;;;GAKG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAAiB;IACnD,IAAI,EAAE,mBAAmB;IACzB,WAAW,EAAE,sGAAsG;IACnH,IAAI,EAAE,UAAU;IAChB,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,MAAM,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACjC,MAAM,MAAM,GAAG,IAAI,CAAC,UAAU,IAAI,mBAAmB,CAAC,yBAAyB,CAAC,CAAA;QAChF,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,IAAI,MAAM,CAAA;QAClC,MAAM,QAAQ,GAAG,MAAM,cAAc,CAAC,aAAa,CAAC,CAAA;QAEpD,OAAO,CAAC,KAAK,CAAC,UAAU,QAAQ,CAAC,MAAM,oEAAoE,CAAC,CAAA;QAE5G,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,WAAW,GAAG,CAAC,CAAA;QACnB,MAAM,UAAU,GAA2B,EAAE,CAAA;QAE7C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,EAAE,CAAC,EAAE,EAAE,CAAC;YAChC,MAAM,KAAK,GAAG,6BAA6B,CAAC,EAAE,MAAM,EAAE,QAAQ,EAAE,CAAC,CAAA;YACjE,MAAM,OAAO,GAAG,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAA;YAE1C,MAAM,SAAS,GAAG;gBACjB,GAAG,EAAE,KAAK,CAAC,GAAG;gBACd,UAAU,EAAE,KAAK,CAAC,UAAU;gBAC5B,OAAO;gBACP,MAAM,EAAE,KAAK,CAAC,MAAM;gBACpB,MAAM;gBACN,SAAS,EAAE,cAAc,CAAC,MAAM,EAAE,EAAE,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC;aACnD,CAAA;YAED,MAAM,OAAO,GAAG,QAAQ,CAAC,SAA2C,CAAC,CAAA;YAErE,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;gBAChC,WAAW,EAAE,CAAA;gBAEb,SAAQ;YACT,CAAC;YAED,KAAK,CACJ,aAAa,CAAC;gBACb,GAAG,OAAO,CAAC,GAAG;gBACd,MAAM,EAAE,mBAAmB;gBAC3B,QAAQ,EAAE,KAAK,CAAC,QAAQ;gBACxB,iFAAiF;gBACjF,oFAAoF;gBACpF,QAAQ,EAAE,IAAI;gBACd,OAAO,EAAE,aAAa,CAAC,QAAQ;aAC/B,CAAC,CACF,CAAA;YAED,OAAO,EAAE,CAAA;YACT,UAAU,CAAC,KAAK,CAAC,QAAQ,CAAC,GAAG,CAAC,UAAU,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;QACnE,CAAC;QAED,OAAO,CAAC,KAAK,CAAC,WAAW,OAAO,UAAU,WAAW,eAAe,CAAC,CAAA;QACrE,OAAO,CAAC,KAAK,CAAC,gBAAgB,EAAE,aAAa,CAAC,UAAU,CAAC,CAAC,CAAA;QAE1D,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,WAAW,EAAE,CAAA;IAClD,CAAC;CACD,CAAA"}

package/out/recipes/bare/country.d.ts

@@ -1,25 +1,7 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `bare-country` — the country name as the WHOLE query. Addresses almost never consist of one bare
- *   country name, so the class is out of the model's training distribution and the tag is a coin
- *   flip: measured on the 2026-08-13 panel, France/Germany/United States/New Zealand parsed
- *   `country` while Japan/China/Nigeria/Australia/Deutschland parsed `locality`. The retrieval-side
- *   bare-country race (#1651) covers the answer either way; this slice closes the PARSE half so the
- *   tag itself is right.
- *
- *   Source is the codex country table — every ISO canonical English name plus the curated surface
- *   forms (endonyms and long-form names). The 2–3 letter CODES are deliberately excluded: a bare
- *   `JP` or `GER` is an abbreviation register with its own ambiguity surface (`IN`, `DE`, `TO` are
- *   English words), and teaching it here would be an unmeasured claim. Case augmentation is the
- *   loader's job, as everywhere.
- */
 import { type CorpusRecipe } from "#recipes/scaffold";
 /**
- * Slice recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
- * and `description` below for the surface form it generates.
+ * Generates queries that consist only of a country name, drawn from ISO names
+ * and curated surface forms of at least four characters, with no country codes.
  */
 export declare const bareCountryRecipe: CorpusRecipe;
 //# sourceMappingURL=country.d.ts.map

package/out/recipes/bare/country.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"country.d.ts","sourceRoot":"","sources":["../../../lib/recipes/bare/country.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAKH,OAAO,EAAiB,KAAK,YAAY,EAAiB,MAAM,mBAAmB,CAAA;AA+BnF;;;GAGG;AACH,eAAO,MAAM,iBAAiB,EAAE,YAqC/B,CAAA"}
+{"version":3,"file":"country.d.ts","sourceRoot":"","sources":["../../../lib/recipes/bare/country.ts"],"names":[],"mappings":"AAGA,OAAO,EAAiB,KAAK,YAAY,EAAkB,MAAM,mBAAmB,CAAA;AA0CpF;;;GAGG;AACH,eAAO,MAAM,iBAAiB,EAAE,YAqC/B,CAAA"}