@mailwoman/corpus 9.4.0 → 10.1.0

package/out/sg/tools/fetch/acra.js

@@ -0,0 +1,123 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Fetches Singapore ACRA corporate-entity CSVs from data.gov.sg.
+ *   The publisher includes fielded addresses in these files.
+ */
+import { BYTES_PER_KIB } from "@mailwoman/core/fs/formatters";
+import { pathExists } from "@mailwoman/core/fs/readers";
+import { makeDirectories } from "@mailwoman/core/fs/writers";
+import { sha256File } from "@mailwoman/core/hash";
+import { sleep } from "@mailwoman/core/utils/sleep";
+import { loadCollectionFiles, streamBodyToFile, withRetries, writeManifest } from "#tools/fetch/download";
+const SLUG = "acra-sg";
+const COLLECTION_URL = "https://api-production.data.gov.sg/v2/public/api/collections/2/metadata";
+const DATASET_API = "https://api-production.data.gov.sg/v2/public/api/datasets";
+const DOWNLOAD_API = "https://api-open.data.gov.sg/v1/public/api/datasets";
+const LICENSE = "Singapore Open Data Licence version 1.0 — https://data.gov.sg/open-data-licence";
+// The license prescribes this attribution sentence.
+// The manifest records it.
+const ATTRIBUTION = "Contains information from ACRA Information on Corporate Entities accessed on <date> from data.gov.sg which is made available under the terms of the Singapore Open Data Licence version 1.0 https://data.gov.sg/open-data-licence";
+const POLL_INTERVAL_MS = 3000;
+const POLL_ATTEMPTS = 40;
+async function readJSON(url) {
+    const res = await fetch(url, { headers: { accept: "application/json" } });
+    if (!res.ok)
+        throw new Error(`acra-sg: ${url} answered HTTP ${res.status}`);
+    return (await res.json());
+}
+/**
+ * Asks data.gov.sg to prepare a CSV export and polls until the export's signed URL is ready.
+ *
+ * The function returns `undefined` when polling runs out of attempts.
+ */
+async function signedURLFor(datasetID) {
+    await readJSON(`${DOWNLOAD_API}/${datasetID}/initiate-download`);
+    for (let attempt = 0; attempt < POLL_ATTEMPTS; attempt++) {
+        await sleep(POLL_INTERVAL_MS);
+        const poll = await readJSON(`${DOWNLOAD_API}/${datasetID}/poll-download`);
+        if (poll.data?.status === "DOWNLOAD_SUCCESS" && poll.data.url)
+            return poll.data.url;
+    }
+    return undefined;
+}
+function filenameFor(name, datasetID) {
+    const letter = /\('([^']+)'\)/.exec(name)?.[1];
+    return `acra-entities-${(letter ?? datasetID).toLowerCase().replaceAll(/[^a-z0-9]+/g, "-")}.csv`;
+}
+/**
+ * Downloads every dataset in the ACRA collection and writes a collection manifest after each file.
+ *
+ * The function skips a file that the previous manifest lists and that still exists on disk.
+ */
+export async function fetchACRASG(options, report) {
+    const destDir = options.outRoot(SLUG);
+    await makeDirectories(destDir);
+    const manifestPath = destDir("MANIFEST.json");
+    const collection = await readJSON(COLLECTION_URL);
+    const datasetIDs = collection.data?.collectionMetadata?.childDatasets ?? [];
+    report?.(`=== ${SLUG}: ${datasetIDs.length} datasets in collection 2`);
+    const previous = await loadCollectionFiles(manifestPath);
+    const files = new Map();
+    let fetched = 0;
+    let skipped = 0;
+    const failedCodes = [];
+    for (const datasetID of datasetIDs) {
+        const metadata = await readJSON(`${DATASET_API}/${datasetID}/metadata`);
+        const name = metadata.data?.name ?? datasetID;
+        const filename = filenameFor(name, datasetID);
+        const dest = destDir(filename);
+        const before = previous.get(filename);
+        if (before && (await pathExists(dest))) {
+            files.set(filename, before);
+            skipped++;
+            continue;
+        }
+        report?.(`--- ${name} (${datasetID})`);
+        let bytes;
+        try {
+            bytes = await withRetries(async () => {
+                const url = await signedURLFor(datasetID);
+                if (!url)
+                    throw new Error(`the export never reached DOWNLOAD_SUCCESS for ${datasetID}`);
+                const res = await fetch(url, { signal: AbortSignal.timeout(1_800_000) });
+                if (!res.ok)
+                    throw new Error(`HTTP ${res.status} on the signed URL for ${datasetID}`);
+                return streamBodyToFile(res, dest);
+            }, { report, label: name });
+        }
+        catch (error) {
+            report?.(`  ✗ ${error.message}`);
+            failedCodes.push(datasetID);
+            continue;
+        }
+        if (bytes < BYTES_PER_KIB) {
+            report?.(`  ✗ ${bytes} bytes — an error page, not the register`);
+            failedCodes.push(datasetID);
+            continue;
+        }
+        const sha = await sha256File(dest);
+        files.set(filename, {
+            source_url: `https://data.gov.sg/datasets/${datasetID}/view`,
+            downloaded_at: new Date().toISOString(),
+            filename,
+            sha256: sha,
+            bytes,
+        });
+        fetched++;
+        report?.(`  ✓ ${(bytes / 1024 / 1024).toFixed(1)} MB  sha256=${sha}`);
+        const manifest = {
+            source: SLUG,
+            source_url: "https://data.gov.sg/collections/2/view",
+            license: LICENSE,
+            attribution: ATTRIBUTION,
+            downloaded_at: new Date().toISOString(),
+            files: [...files.values()],
+        };
+        await writeManifest(manifestPath, manifest);
+    }
+    return { fetched, skipped, failed: failedCodes.length, failedCodes };
+}
+//# sourceMappingURL=acra.js.map

package/out/sg/tools/fetch/acra.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"acra.js","sourceRoot":"","sources":["../../../../lib/sg/tools/fetch/acra.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,EAAE,aAAa,EAAE,MAAM,+BAA+B,CAAA;AAC7D,OAAO,EAAE,UAAU,EAAE,MAAM,4BAA4B,CAAA;AACvD,OAAO,EAAE,eAAe,EAAE,MAAM,4BAA4B,CAAA;AAC5D,OAAO,EAAE,UAAU,EAAE,MAAM,sBAAsB,CAAA;AACjD,OAAO,EAAE,KAAK,EAAE,MAAM,6BAA6B,CAAA;AAGnD,OAAO,EAAE,mBAAmB,EAAE,gBAAgB,EAAE,WAAW,EAAE,aAAa,EAAE,MAAM,uBAAuB,CAAA;AAEzG,MAAM,IAAI,GAAG,SAAS,CAAA;AACtB,MAAM,cAAc,GAAG,yEAAyE,CAAA;AAChG,MAAM,WAAW,GAAG,2DAA2D,CAAA;AAC/E,MAAM,YAAY,GAAG,qDAAqD,CAAA;AAC1E,MAAM,OAAO,GAAG,iFAAiF,CAAA;AAEjG,oDAAoD;AACpD,2BAA2B;AAC3B,MAAM,WAAW,GAChB,mOAAmO,CAAA;AAEpO,MAAM,gBAAgB,GAAG,IAAI,CAAA;AAC7B,MAAM,aAAa,GAAG,EAAE,CAAA;AAmBxB,KAAK,UAAU,QAAQ,CAAI,GAAW;IACrC,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,GAAG,EAAE,EAAE,OAAO,EAAE,EAAE,MAAM,EAAE,kBAAkB,EAAE,EAAE,CAAC,CAAA;IAEzE,IAAI,CAAC,GAAG,CAAC,EAAE;QAAE,MAAM,IAAI,KAAK,CAAC,YAAY,GAAG,kBAAkB,GAAG,CAAC,MAAM,EAAE,CAAC,CAAA;IAE3E,OAAO,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,CAAM,CAAA;AAC/B,CAAC;AAED;;;;GAIG;AACH,KAAK,UAAU,YAAY,CAAC,SAAiB;IAC5C,MAAM,QAAQ,CAAC,GAAG,YAAY,IAAI,SAAS,oBAAoB,CAAC,CAAA;IAEhE,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,OAAO,GAAG,aAAa,EAAE,OAAO,EAAE,EAAE,CAAC;QAC1D,MAAM,KAAK,CAAC,gBAAgB,CAAC,CAAA;QAC7B,MAAM,IAAI,GAAG,MAAM,QAAQ,CAAe,GAAG,YAAY,IAAI,SAAS,gBAAgB,CAAC,CAAA;QAEvF,IAAI,IAAI,CAAC,IAAI,EAAE,MAAM,KAAK,kBAAkB,IAAI,IAAI,CAAC,IAAI,CAAC,GAAG;YAAE,OAAO,IAAI,CAAC,IAAI,CAAC,GAAG,CAAA;IACpF,CAAC;IAED,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,SAAS,WAAW,CAAC,IAAY,EAAE,SAAiB;IACnD,MAAM,MAAM,GAAG,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAE9C,OAAO,iBAAiB,CAAC,MAAM,IAAI,SAAS,CAAC,CAAC,WAAW,EAAE,CAAC,UAAU,CAAC,aAAa,EAAE,GAAG,CAAC,MAAM,CAAA;AACjG,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,WAAW,CAAC,OAA2B,EAAE,MAA+B;IAC7F,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;IACrC,MAAM,eAAe,CAAC,OAAO,CAAC,CAAA;IAC9B,MAAM,YAAY,GAAG,OAAO,CAAC,eAAe,CAAC,CAAA;IAE7C,MAAM,UAAU,GAAG,MAAM,QAAQ,CAAqB,cAAc,CAAC,CAAA;IACrE,MAAM,UAAU,GAAG,UAAU,CAAC,IAAI,EAAE,kBAAkB,EAAE,aAAa,IAAI,EAAE,CAAA;IAC3E,MAAM,EAAE,CAAC,OAAO,IAAI,KAAK,UAAU,CAAC,MAAM,2BAA2B,CAAC,CAAA;IAEtE,MAAM,QAAQ,GAAG,MAAM,mBAAmB,CAAC,YAAY,CAAC,CAAA;IACxD,MAAM,KAAK,GAAG,IAAI,GAAG,EAA0B,CAAA;IAC/C,IAAI,OAAO,GAAG,CAAC,CAAA;IACf,IAAI,OAAO,GAAG,CAAC,CAAA;IACf,MAAM,WAAW,GAAa,EAAE,CAAA;IAEhC,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;QACpC,MAAM,QAAQ,GAAG,MAAM,QAAQ,CAAkB,GAAG,WAAW,IAAI,SAAS,WAAW,CAAC,CAAA;QACxF,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,EAAE,IAAI,IAAI,SAAS,CAAA;QAC7C,MAAM,QAAQ,GAAG,WAAW,CAAC,IAAI,EAAE,SAAS,CAAC,CAAA;QAC7C,MAAM,IAAI,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAA;QAC9B,MAAM,MAAM,GAAG,QAAQ,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;QAErC,IAAI,MAAM,IAAI,CAAC,MAAM,UAAU,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;YACxC,KAAK,CAAC,GAAG,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAA;YAE3B,OAAO,EAAE,CAAA;YAET,SAAQ;QACT,CAAC;QAED,MAAM,EAAE,CAAC,OAAO,IAAI,KAAK,SAAS,GAAG,CAAC,CAAA;QACtC,IAAI,KAAa,CAAA;QAEjB,IAAI,CAAC;YACJ,KAAK,GAAG,MAAM,WAAW,CACxB,KAAK,IAAI,EAAE;gBACV,MAAM,GAAG,GAAG,MAAM,YAAY,CAAC,SAAS,CAAC,CAAA;gBAEzC,IAAI,CAAC,GAAG;oBAAE,MAAM,IAAI,KAAK,CAAC,iDAAiD,SAAS,EAAE,CAAC,CAAA;gBACvF,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,GAAG,EAAE,EAAE,MAAM,EAAE,WAAW,CAAC,OAAO,CAAC,SAAS,CAAC,EAAE,CAAC,CAAA;gBAExE,IAAI,CAAC,GAAG,CAAC,EAAE;oBAAE,MAAM,IAAI,KAAK,CAAC,QAAQ,GAAG,CAAC,MAAM,0BAA0B,SAAS,EAAE,CAAC,CAAA;gBAErF,OAAO,gBAAgB,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;YACnC,CAAC,EACD,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,CACvB,CAAA;QACF,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YAChB,MAAM,EAAE,CAAC,OAAQ,KAAe,CAAC,OAAO,EAAE,CAAC,CAAA;YAC3C,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;YAE3B,SAAQ;QACT,CAAC;QAED,IAAI,KAAK,GAAG,aAAa,EAAE,CAAC;YAC3B,MAAM,EAAE,CAAC,OAAO,KAAK,0CAA0C,CAAC,CAAA;YAChE,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;YAE3B,SAAQ;QACT,CAAC;QAED,MAAM,GAAG,GAAG,MAAM,UAAU,CAAC,IAAI,CAAC,CAAA;QAElC,KAAK,CAAC,GAAG,CAAC,QAAQ,EAAE;YACnB,UAAU,EAAE,gCAAgC,SAAS,OAAO;YAC5D,aAAa,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;YACvC,QAAQ;YACR,MAAM,EAAE,GAAG;YACX,KAAK;SACL,CAAC,CAAA;QAEF,OAAO,EAAE,CAAA;QACT,MAAM,EAAE,CAAC,OAAO,CAAC,KAAK,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,eAAe,GAAG,EAAE,CAAC,CAAA;QAErE,MAAM,QAAQ,GAA6B;YAC1C,MAAM,EAAE,IAAI;YACZ,UAAU,EAAE,wCAAwC;YACpD,OAAO,EAAE,OAAO;YAChB,WAAW,EAAE,WAAW;YACxB,aAAa,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;YACvC,KAAK,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;SAC1B,CAAA;QAED,MAAM,aAAa,CAAC,YAAY,EAAE,QAAQ,CAAC,CAAA;IAC5C,CAAC;IAED,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,WAAW,EAAE,CAAA;AACrE,CAAC"}

package/out/si/recipes/bare-village.d.ts

@@ -0,0 +1,31 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `si-bare-village`, the Slovenian no-street counter-recipe. Slovenia's rural addressing has no
+ *   street line, so the village name is the street-level token and repeats as the locality
+ *   ("Zabiče 8, 6250 Zabiče"). The leading name must keep its number whole. The trailing mention
+ *   must stay locality-bound. This recipe is the paired counter-distribution, with the same lesson
+ *   ("name before number, comma, then admin") and the opposite polarity on the trailing mention.
+ *
+ *   Three real-order templates cycle per tuple (mirrors the coord-golden orders so the eval and the
+ *   training distribution agree):
+ *
+ *   1. canonical  "«V» «n», «pc» «V»"
+ *   2. bare       "«V» «n», «V»"        (no postcode, the anchor-free form)
+ *   3. pc-first   "«pc» «V», «V» «n»"
+ *
+ *   Gold spans: leading «V» = street (matches OA ground truth, the village is the address line),
+ *   «n» = house_number (never split mid-digits), «pc» = postcode (never swallowing the neighbor),
+ *   trailing «V» = locality (the binding the resolver needs).
+ */
+import { type CorpusRecipe } from "#recipes/scaffold";
+/**
+ * Recipe registered with the corpus builder.
+ *
+ * See the file header for the parse behavior this recipe exercises
+ * and `description` below for its surface form.
+ */
+export declare const siBareVillageRecipe: CorpusRecipe;
+//# sourceMappingURL=bare-village.d.ts.map

package/out/si/recipes/bare-village.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"bare-village.d.ts","sourceRoot":"","sources":["../../../lib/si/recipes/bare-village.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;GAqBG;AAIH,OAAO,EAA8C,KAAK,YAAY,EAAkB,MAAM,mBAAmB,CAAA;AAIjH;;;;;GAKG;AACH,eAAO,MAAM,mBAAmB,EAAE,YAgFjC,CAAA"}

package/out/si/recipes/bare-village.js

@@ -0,0 +1,103 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `si-bare-village`, the Slovenian no-street counter-recipe. Slovenia's rural addressing has no
+ *   street line, so the village name is the street-level token and repeats as the locality
+ *   ("Zabiče 8, 6250 Zabiče"). The leading name must keep its number whole. The trailing mention
+ *   must stay locality-bound. This recipe is the paired counter-distribution, with the same lesson
+ *   ("name before number, comma, then admin") and the opposite polarity on the trailing mention.
+ *
+ *   Three real-order templates cycle per tuple (mirrors the coord-golden orders so the eval and the
+ *   training distribution agree):
+ *
+ *   1. canonical  "«V» «n», «pc» «V»"
+ *   2. bare       "«V» «n», «V»"        (no postcode, the anchor-free form)
+ *   3. pc-first   "«pc» «V», «V» «n»"
+ *
+ *   Gold spans: leading «V» = street (matches OA ground truth, the village is the address line),
+ *   «n» = house_number (never split mid-digits), «pc» = postcode (never swallowing the neighbor),
+ *   trailing «V» = locality (the binding the resolver needs).
+ */
+import { mulberry32 as makeMulberry32 } from "@mailwoman/core/utils";
+import { alignAndWrite, readTuples, requireRegister, recipeSourceID } from "#recipes/scaffold";
+import { defaultRecipeSource } from "#recipes/sources";
+import { SurfaceOrigin } from "#types";
+/**
+ * Recipe registered with the corpus builder.
+ *
+ * See the file header for the parse behavior this recipe exercises
+ * and `description` below for its surface form.
+ */
+export const siBareVillageRecipe = {
+    name: "si-bare-village",
+    description: "SI no-street village form (#901 run-2): '«V» «n», «pc» «V»' — the fr-bare-street counter-distribution",
+    mode: "tuples",
+    async run(opts, write) {
+        makeMulberry32(opts.seed);
+        let read = 0;
+        let emitted = 0;
+        let skipped = 0;
+        const SI_BARE_VILLAGE_PROVENANCE = {
+            register: requireRegister(opts, "si-bare-village"),
+            surface: SurfaceOrigin.Composed,
+        };
+        for await (const t of readTuples(opts.input)) {
+            read++;
+            const village = String(t.locality ?? "").trim();
+            const number = String(t.number ?? "").trim();
+            const postcode = String(t.postcode ?? "").trim();
+            if (!village || !number || !postcode) {
+                skipped++;
+                continue;
+            }
+            const order = read % 3;
+            let raw;
+            const components = {
+                street: village,
+                house_number: number,
+                locality: village,
+            };
+            if (order === 0) {
+                components.postcode = postcode;
+                raw = `${village} ${number}, ${postcode} ${village}`;
+            }
+            else if (order === 1) {
+                raw = `${village} ${number}, ${village}`;
+            }
+            else {
+                components.postcode = postcode;
+                raw = `${postcode} ${village}, ${village} ${number}`;
+            }
+            // One resolution for both, because `source_id` includes the source as its prefix
+            // and a pair that disagreed would name a source no row of this output declares.
+            const source = defaultRecipeSource("synth-si-bare-village");
+            const source_id = recipeSourceID(source, { ...components, o: String(order), v: String(read) });
+            const canonical = {
+                raw,
+                components,
+                country: "SI",
+                locale: "sl-SI",
+                source,
+                source_id,
+                corpus_version: "0.9.9",
+                license: "Synthetic — si-bare-village; (village, number, postcode) from OpenAddresses SI (per-source attribution in the model card)",
+            };
+            // Per row rather than on the shared literal, because the id names this tuple's record.
+            // A tuples file written before `sourceID` existed has no id.
+            // `null` records that case.
+            if (alignAndWrite(write, canonical, "si-bare-village", {
+                ...SI_BARE_VILLAGE_PROVENANCE,
+                baseSourceID: t.sourceID ?? null,
+            })) {
+                emitted++;
+            }
+            else {
+                skipped++;
+            }
+        }
+        return { read, emitted, skipped };
+    },
+};
+//# sourceMappingURL=bare-village.js.map

package/out/si/recipes/bare-village.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"bare-village.js","sourceRoot":"","sources":["../../../lib/si/recipes/bare-village.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;GAqBG;AAEH,OAAO,EAAE,UAAU,IAAI,cAAc,EAAE,MAAM,uBAAuB,CAAA;AAEpE,OAAO,EAAE,aAAa,EAAE,UAAU,EAAE,eAAe,EAAqB,cAAc,EAAE,MAAM,mBAAmB,CAAA;AACjH,OAAO,EAAE,mBAAmB,EAAE,MAAM,kBAAkB,CAAA;AACtD,OAAO,EAAE,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEtC;;;;;GAKG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAiB;IAChD,IAAI,EAAE,iBAAiB;IACvB,WAAW,EAAE,uGAAuG;IACpH,IAAI,EAAE,QAAQ;IACd,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACzB,IAAI,IAAI,GAAG,CAAC,CAAA;QACZ,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QAEf,MAAM,0BAA0B,GAAG;YAClC,QAAQ,EAAE,eAAe,CAAC,IAAI,EAAE,iBAAiB,CAAC;YAClD,OAAO,EAAE,aAAa,CAAC,QAAQ;SAC/B,CAAA;QAED,IAAI,KAAK,EAAE,MAAM,CAAC,IAAI,UAAU,CAAC,IAAI,CAAC,KAAM,CAAC,EAAE,CAAC;YAC/C,IAAI,EAAE,CAAA;YACN,MAAM,OAAO,GAAG,MAAM,CAAC,CAAC,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAC/C,MAAM,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAC5C,MAAM,QAAQ,GAAG,MAAM,CAAC,CAAC,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAEhD,IAAI,CAAC,OAAO,IAAI,CAAC,MAAM,IAAI,CAAC,QAAQ,EAAE,CAAC;gBACtC,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,MAAM,KAAK,GAAG,IAAI,GAAG,CAAC,CAAA;YACtB,IAAI,GAAW,CAAA;YAEf,MAAM,UAAU,GAA2B;gBAC1C,MAAM,EAAE,OAAO;gBACf,YAAY,EAAE,MAAM;gBACpB,QAAQ,EAAE,OAAO;aACjB,CAAA;YAED,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;gBACjB,UAAU,CAAC,QAAQ,GAAG,QAAQ,CAAA;gBAC9B,GAAG,GAAG,GAAG,OAAO,IAAI,MAAM,KAAK,QAAQ,IAAI,OAAO,EAAE,CAAA;YACrD,CAAC;iBAAM,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;gBACxB,GAAG,GAAG,GAAG,OAAO,IAAI,MAAM,KAAK,OAAO,EAAE,CAAA;YACzC,CAAC;iBAAM,CAAC;gBACP,UAAU,CAAC,QAAQ,GAAG,QAAQ,CAAA;gBAC9B,GAAG,GAAG,GAAG,QAAQ,IAAI,OAAO,KAAK,OAAO,IAAI,MAAM,EAAE,CAAA;YACrD,CAAC;YAED,iFAAiF;YACjF,gFAAgF;YAChF,MAAM,MAAM,GAAG,mBAAmB,CAAC,uBAAuB,CAAC,CAAA;YAC3D,MAAM,SAAS,GAAG,cAAc,CAAC,MAAM,EAAE,EAAE,GAAG,UAAU,EAAE,CAAC,EAAE,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,EAAE,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;YAE9F,MAAM,SAAS,GAAG;gBACjB,GAAG;gBACH,UAAU;gBACV,OAAO,EAAE,IAAI;gBACb,MAAM,EAAE,OAAO;gBACf,MAAM;gBACN,SAAS;gBACT,cAAc,EAAE,OAAO;gBACvB,OAAO,EACN,2HAA2H;aAC5H,CAAA;YAED,uFAAuF;YACvF,6DAA6D;YAC7D,4BAA4B;YAC5B,IACC,aAAa,CAAC,KAAK,EAAE,SAAS,EAAE,iBAAiB,EAAE;gBAClD,GAAG,0BAA0B;gBAC7B,YAAY,EAAE,CAAC,CAAC,QAAQ,IAAI,IAAI;aAChC,CAAC,EACD,CAAC;gBACF,OAAO,EAAE,CAAA;YACV,CAAC;iBAAM,CAAC;gBACP,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;QAED,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;IAClC,CAAC;CACD,CAAA"}

package/out/sk/adapters/inspire/adapter.d.ts

@@ -0,0 +1,119 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `sk-inspire`: the INSPIRE Addresses theme the Slovak Ministry of the Interior publishes out of the
+ * national address register, read from saved pages of its Web Feature Service.
+ *
+ * The service is `https://rageo.minv.sk/geoserver/ad/wfs`, a GeoServer WFS 2.0 that advertises
+ * `ImplementsResultPaging` and serves all five AD feature types. Measured 2026-10-02, a
+ * `GetFeature` for `ad:Address` answers `numberMatched="1704196"`, and the other four types hold
+ * 29,315 `ad:ThoroughfareName`, 3,015 `ad:AdminUnitName`, 1,769 `ad:AddressAreaName` and 1,415
+ * `ad:PostalDescriptor`. Acquisition pages each type into its own document and the adapter reads a
+ * directory of those pages, so a build harvests the service once and reads the saved bytes.
+ *
+ * The reference form is a WFS stored-query URL: an `ad:component` href is
+ * `…/ad/ows?…&storedQuery_id=urn:ogc:def:query:OGC-WFS::GetFeatureById&id=AdminUnitName.199`, and
+ * that `id` parameter equals the target's `gml:id` exactly, which a `GetFeatureById` fetch of
+ * `AdminUnitName.199` confirms. `componentJoinKey` reads it. Each reference also repeats its target's
+ * value in `xlink:title`, as Czechia's does, and the adapter reads none of those titles. Three of an
+ * address's references are `ad:AdminUnitName` features whose titles are a region, a district and a
+ * municipality, and only the resolved feature's `ad:level` separates the three.
+ *
+ * That is why the reader takes two passes over the input rather than one. A component page is a
+ * different document from an address page, so a single pass would have to defer an address until the
+ * page holding its municipality arrived, and with 1,704,196 addresses against 35,514 component
+ * features the deferred list is the wrong side of the join to hold. The first pass indexes the four
+ * component types from every document, and the second streams `ad:Address`. The index is therefore
+ * complete before the first address is read, so a reference resolving to no feature is a value the
+ * reader asked for and could not read, and it raises.
+ *
+ * The locator vocabulary is two types, and the publisher's own rendering fixes how they compose.
+ * `ad:designator` repeats and states no position, so selection goes by `ad:type/@xlink:href`:
+ * `addressNumber` is the *číslo popisné* and is present on every address, and `buildingIdentifier`
+ * is the *orientačné číslo* and appears only where the address also references a street. Measured
+ * over 1,000 addresses across ten windows spread through the 1,704,196, `addressNumber` appeared
+ * 1,000 times, `buildingIdentifier` 334 times, and the street count was also exactly 334.
+ *
+ * The same publisher serves its register's own rendered address through `ra:address` on the same
+ * GeoServer, which is what pins the composition. Three records, matched by street, number and postcode:
+ *
+ * | `addressNumber` | `buildingIdentifier` | street | `fulladdress` |
+ * | --- | --- | --- | --- |
+ * | `1760` | `29` | `Lipová` | `Lipová,1760/29,Hurbanovo,94701` |
+ * | `1771` | `4` | `Skalická` | `Skalická,1771/4,Košice-Krásna,04018` |
+ * | `142` | (absent) | (absent) | `142,Sikenička,94359` |
+ *
+ * So the number is `addressNumber` by itself, or `addressNumber/buildingIdentifier` where the
+ * orientation number exists, which is the publisher's own `annotation` field verbatim. A street-less
+ * address is ordinary rather than a defect: 666 of those 1,000 referenced no `ad:ThoroughfareName`.
+ *
+ * The place names come from two types. `ad:AdminUnitName` at `4thOrder` is the municipality and is
+ * referenced by every address measured, 1,080 of 1,080. The `3rdOrder` district and the `2ndOrder`
+ * region are referenced beside it and are not part of a written address. `ad:AddressAreaName` is the
+ * *časť obce*, the settlement part, and 426 of those 1,080 addresses reference one. Of those 426, 325
+ * repeat the municipality and 101 state another place, such as `Lidér Tejed` in the municipality
+ * of `Povoda`, so the adapter writes a dependent locality only where the two differ. The publisher's
+ * own `fulladdress` omits the settlement part in both cases, so a row holding one states a component
+ * that string does not.
+ *
+ * The postcode is `ad:PostalDescriptor/ad:postCode`, written as five digits with its leading zero
+ * intact (`04018`). Slovak usage splits it as `040 18`. The publisher writes it closed up, and the
+ * row stores what the publisher wrote, as the Czech reader does with `66463`.
+ *
+ * The Ministry states one condition of use and states it twice, in its own register record
+ * `https://rpi.gov.sk/api/collection_record/86dea70e-a55b-4241-bd63-4024b3f76b72` and in that
+ * record's INSPIRE export: `Neuplatňujú sa žiadne podmienky (CC0 Voľné dielo)`. That phrase renders
+ * INSPIRE's controlled value for no conditions applying to access and use, in Slovak, with the
+ * publisher's own identification of it as CC0. No version accompanies the label, so the
+ * address-source register writes no SPDX identifier and the rows record the publisher's words.
+ *
+ * The adapter honors `opts.limit` and `opts.signal`. `opts.country` is optional and accepts only `SK`.
+ */
+import { type CorpusAdapter } from "#types";
+/**
+ * Registry id for this adapter.
+ *
+ * Stamped into every row it emits, so a corpus record can be traced back to the dataset it came from.
+ */
+export declare const SK_INSPIRE_ADAPTER_ID = "sk-inspire";
+/**
+ * The publication this adapter reads.
+ *
+ * The literal stands in for a `SourceRegister` member until `lib/registers.ts` declares
+ * `SlovakiaInspireAddresses: "sk-minv-inspire-ad"`, which is the id this value spells.
+ */
+export declare const SK_INSPIRE_REGISTER: "sk-minv-inspire-ad";
+/**
+ * The one jurisdiction this adapter emits.
+ *
+ * The service's register record states a national spatial scope with a `footprint_geo`
+ * of `Slovenská Republika`, so the set has one member.
+ */
+export declare const SK_INSPIRE_COUNTRIES: readonly string[];
+/**
+ * The license the address-source register elected for this publisher.
+ *
+ * `unchecked-access-free-sk-ministerstvo-vn-tra-slovenskej-republiky` is elected with
+ * `electedTerms` `Neuplatňujú sa žiadne podmienky (CC0 Voľné dielo)` and no `spdx`,
+ * because the publisher names CC0 without a version.
+ * The rows therefore record the publisher's own words, as Poland's and Czechia's do.
+ */
+export declare const SK_INSPIRE_DEFAULT_LICENSE = "Neuplat\u0148uj\u00FA sa \u017Eiadne podmienky (CC0 Vo\u013En\u00E9 dielo)";
+/**
+ * The house number as the Register adries writes it in its own `annotation` field.
+ *
+ * The descriptive number is the whole value where the publisher states no orientation
+ * number, and the two are joined by a solidus where it does.
+ * The table in this module's header reads that off three records of the publisher's `ra:address` layer.
+ *
+ * @returns The number, or undefined when the publisher states no `addressNumber`.
+ */
+export declare function slovakHouseNumber(byType: ReadonlyMap<string, readonly string[]>): string | undefined;
+export declare function createSKInspireAdapter(): CorpusAdapter;
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export declare const skInspireAdapter: CorpusAdapter;
+//# sourceMappingURL=adapter.d.ts.map

package/out/sk/adapters/inspire/adapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/sk/adapters/inspire/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAwEG;AAuBH,OAAO,EAAuD,KAAK,aAAa,EAAiB,MAAM,QAAQ,CAAA;AAE/G;;;;GAIG;AACH,eAAO,MAAM,qBAAqB,eAAe,CAAA;AAEjD;;;;;GAKG;AACH,eAAO,MAAM,mBAAmB,sBAA0C,CAAA;AAE1E;;;;;GAKG;AACH,eAAO,MAAM,oBAAoB,EAAE,SAAS,MAAM,EAAW,CAAA;AAE7D;;;;;;;GAOG;AACH,eAAO,MAAM,0BAA0B,+EAAsD,CAAA;AA+D7F;;;;;;;;GAQG;AACH,wBAAgB,iBAAiB,CAAC,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC,GAAG,MAAM,GAAG,SAAS,CAQpG;AAED,wBAAgB,sBAAsB,IAAI,aAAa,CA8DtD;AAuHD;;GAEG;AACH,eAAO,MAAM,gBAAgB,eAA2B,CAAA"}