@mailwoman/corpus 9.3.0 → 10.0.0

package/out/kr/adapters/localdata/key-index.d.ts

@@ -0,0 +1,71 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The LABEL register's key sets, the only thing a permit string is allowed to align against (#2204 §5).
+ *
+ *   The permit registry is a NOISY source: a clerk typed each address, in both address systems, with no validation. So
+ *   nothing in a permit string is taken on its own word. The region must be a 시도 the register lists, the 시군구 one
+ *   that region lists, the road one that 시군구 lists, the 동 one that 시군구 lists, the 리 one that 동 lists. A string
+ *   that satisfies the whole key becomes a training row; one that does not is a BOARD row — an address the model will
+ *   be read on and never trained on.
+ *
+ *   Built by one pass over the LABEL rows, which is why it lives beside them rather than inside the aligner.
+ */
+import { type JusoLabelRow } from "#kr/adapters/juso/label-rows";
+/**
+ * The key sets, keyed as the register nests them.
+ *
+ * A composite string key where the register's own key is a tuple, because a nested map costs a second lookup on the hot
+ * path and every permit row does several.
+ */
+export interface KeyIndex {
+    regions: Set<string>;
+    /**
+     * 시도 → its 시군구 names.
+     */
+    sigunguByRegion: Map<string, Set<string>>;
+    /**
+     * 시도 and 시군구 → the unit's road names.
+     */
+    roadsByUnit: Map<string, Set<string>>;
+    /**
+     * 시도 and 시군구 → the unit's 동 and 읍/면 names.
+     */
+    dongsByUnit: Map<string, Set<string>>;
+    /**
+     * 시도, 시군구 and 동 → the 동's 리 names.
+     */
+    risByDong: Map<string, Set<string>>;
+}
+/**
+ * The parts of a composite key, joined so no two different part lists produce the same key.
+ */
+export declare function unitKey(...parts: readonly string[]): string;
+export declare function emptyKeyIndex(): KeyIndex;
+/**
+ * Add one LABEL row's names to the key sets.
+ *
+ * The 읍/면 joins the road set's unit as a road-form token too, since the road address writes it between the 시군구 and the
+ * road.
+ */
+export declare function indexLabelRow(index: KeyIndex, row: JusoLabelRow): void;
+/**
+ * Admit the pre-merger region names as aliases of the register's current one, SHARING its key sets.
+ *
+ * The sets are shared by reference rather than copied: an alias is the same place under an older name, and a later
+ * `indexLabelRow` under either name must reach the same set.
+ */
+export declare function aliasKeyIndex(index: KeyIndex): void;
+/**
+ * How many tokens from `start` form one of the region's 시군구 — 2 for `수원시 장안구`, 1 for `종로구`, 0 for none.
+ *
+ * Two-token first: `수원시` alone is also a listed name, and taking it would leave `장안구` to be read as a road.
+ */
+export declare function sigunguSpan(index: KeyIndex, region: string, tokens: readonly string[], start: number): number;
+/**
+ * The key sets the permit aligner reads, from one pass over the LABEL rows.
+ */
+export declare function buildKeyIndex(rows: AsyncIterable<JusoLabelRow>): Promise<KeyIndex>;
+//# sourceMappingURL=key-index.d.ts.map

package/out/kr/adapters/localdata/key-index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"key-index.d.ts","sourceRoot":"","sources":["../../../../lib/kr/adapters/localdata/key-index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAEH,OAAO,EAAE,KAAK,YAAY,EAAkB,MAAM,8BAA8B,CAAA;AAUhF;;;;;GAKG;AACH,MAAM,WAAW,QAAQ;IACxB,OAAO,EAAE,GAAG,CAAC,MAAM,CAAC,CAAA;IACpB;;OAEG;IACH,eAAe,EAAE,GAAG,CAAC,MAAM,EAAE,GAAG,CAAC,MAAM,CAAC,CAAC,CAAA;IACzC;;OAEG;IACH,WAAW,EAAE,GAAG,CAAC,MAAM,EAAE,GAAG,CAAC,MAAM,CAAC,CAAC,CAAA;IACrC;;OAEG;IACH,WAAW,EAAE,GAAG,CAAC,MAAM,EAAE,GAAG,CAAC,MAAM,CAAC,CAAC,CAAA;IACrC;;OAEG;IACH,SAAS,EAAE,GAAG,CAAC,MAAM,EAAE,GAAG,CAAC,MAAM,CAAC,CAAC,CAAA;CACnC;AAED;;GAEG;AACH,wBAAgB,OAAO,CAAC,GAAG,KAAK,EAAE,SAAS,MAAM,EAAE,GAAG,MAAM,CAE3D;AAED,wBAAgB,aAAa,IAAI,QAAQ,CAQxC;AAYD;;;;;GAKG;AACH,wBAAgB,aAAa,CAAC,KAAK,EAAE,QAAQ,EAAE,GAAG,EAAE,YAAY,GAAG,IAAI,CAmBtE;AAED;;;;;GAKG;AACH,wBAAgB,aAAa,CAAC,KAAK,EAAE,QAAQ,GAAG,IAAI,CAsCnD;AAED;;;;GAIG;AACH,wBAAgB,WAAW,CAAC,KAAK,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,SAAS,MAAM,EAAE,EAAE,KAAK,EAAE,MAAM,GAAG,MAAM,CAY7G;AAED;;GAEG;AACH,wBAAsB,aAAa,CAAC,IAAI,EAAE,aAAa,CAAC,YAAY,CAAC,GAAG,OAAO,CAAC,QAAQ,CAAC,CAUxF"}

package/out/kr/adapters/localdata/key-index.js

@@ -0,0 +1,133 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The LABEL register's key sets, the only thing a permit string is allowed to align against (#2204 §5).
+ *
+ *   The permit registry is a NOISY source: a clerk typed each address, in both address systems, with no validation. So
+ *   nothing in a permit string is taken on its own word. The region must be a 시도 the register lists, the 시군구 one
+ *   that region lists, the road one that 시군구 lists, the 동 one that 시군구 lists, the 리 one that 동 lists. A string
+ *   that satisfies the whole key becomes a training row; one that does not is a BOARD row — an address the model will
+ *   be read on and never trained on.
+ *
+ *   Built by one pass over the LABEL rows, which is why it lives beside them rather than inside the aligner.
+ */
+import { REGION_ALIASES } from "#kr/adapters/juso/label-rows";
+/**
+ * The separator between the parts of a composite key, written as an escape so no NUL byte enters the source.
+ *
+ * A space cannot serve. A 시군구 is written with one (`수원시 장안구`), so `unitKey("A", "B C")` and `unitKey("A B", "C")` would
+ * produce one key for two different places. No place name contains NUL.
+ */
+const UNIT_SEPARATOR = "\0";
+/**
+ * The parts of a composite key, joined so no two different part lists produce the same key.
+ */
+export function unitKey(...parts) {
+    return parts.join(UNIT_SEPARATOR);
+}
+export function emptyKeyIndex() {
+    return {
+        regions: new Set(),
+        sigunguByRegion: new Map(),
+        roadsByUnit: new Map(),
+        dongsByUnit: new Map(),
+        risByDong: new Map(),
+    };
+}
+function add(map, key, value) {
+    const existing = map.get(key);
+    if (existing) {
+        existing.add(value);
+    }
+    else {
+        map.set(key, new Set([value]));
+    }
+}
+/**
+ * Add one LABEL row's names to the key sets.
+ *
+ * The 읍/면 joins the road set's unit as a road-form token too, since the road address writes it between the 시군구 and the
+ * road.
+ */
+export function indexLabelRow(index, row) {
+    index.regions.add(row.region);
+    add(index.sigunguByRegion, row.region, row.sigungu);
+    const unit = unitKey(row.region, row.sigungu);
+    add(index.roadsByUnit, unit, row.road);
+    if (row.eupmyeon) {
+        add(index.dongsByUnit, unit, row.eupmyeon);
+    }
+    if (row.dong) {
+        add(index.dongsByUnit, unit, row.dong);
+        if (row.ri) {
+            add(index.risByDong, unitKey(row.region, row.sigungu, row.dong), row.ri);
+        }
+    }
+}
+/**
+ * Admit the pre-merger region names as aliases of the register's current one, SHARING its key sets.
+ *
+ * The sets are shared by reference rather than copied: an alias is the same place under an older name, and a later
+ * `indexLabelRow` under either name must reach the same set.
+ */
+export function aliasKeyIndex(index) {
+    for (const [legacy, current] of Object.entries(REGION_ALIASES)) {
+        if (!index.regions.has(current))
+            continue;
+        index.regions.add(legacy);
+        const sigungu = index.sigunguByRegion.get(current);
+        if (sigungu) {
+            index.sigunguByRegion.set(legacy, sigungu);
+        }
+        // oxlint-disable no-useless-spread -- each loop writes into the map it reads, so it walks a snapshot.
+        for (const [key, roads] of [...index.roadsByUnit]) {
+            const [region, unit] = key.split(UNIT_SEPARATOR);
+            if (region === current) {
+                index.roadsByUnit.set(unitKey(legacy, unit), roads);
+            }
+        }
+        for (const [key, dongs] of [...index.dongsByUnit]) {
+            const [region, unit] = key.split(UNIT_SEPARATOR);
+            if (region === current) {
+                index.dongsByUnit.set(unitKey(legacy, unit), dongs);
+            }
+        }
+        for (const [key, ris] of [...index.risByDong]) {
+            const [region, unit, dong] = key.split(UNIT_SEPARATOR);
+            if (region === current) {
+                index.risByDong.set(unitKey(legacy, unit, dong), ris);
+            }
+        }
+        // oxlint-enable no-useless-spread
+    }
+}
+/**
+ * How many tokens from `start` form one of the region's 시군구 — 2 for `수원시 장안구`, 1 for `종로구`, 0 for none.
+ *
+ * Two-token first: `수원시` alone is also a listed name, and taking it would leave `장안구` to be read as a road.
+ */
+export function sigunguSpan(index, region, tokens, start) {
+    const candidates = index.sigunguByRegion.get(region);
+    if (!candidates)
+        return 0;
+    for (const width of [2, 1]) {
+        if (start + width <= tokens.length && candidates.has(tokens.slice(start, start + width).join(" "))) {
+            return width;
+        }
+    }
+    return 0;
+}
+/**
+ * The key sets the permit aligner reads, from one pass over the LABEL rows.
+ */
+export async function buildKeyIndex(rows) {
+    const index = emptyKeyIndex();
+    for await (const row of rows) {
+        indexLabelRow(index, row);
+    }
+    aliasKeyIndex(index);
+    return index;
+}
+//# sourceMappingURL=key-index.js.map

package/out/kr/adapters/localdata/key-index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"key-index.js","sourceRoot":"","sources":["../../../../lib/kr/adapters/localdata/key-index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAEH,OAAO,EAAqB,cAAc,EAAE,MAAM,8BAA8B,CAAA;AAEhF;;;;;GAKG;AACH,MAAM,cAAc,GAAG,IAAI,CAAA;AA4B3B;;GAEG;AACH,MAAM,UAAU,OAAO,CAAC,GAAG,KAAwB;IAClD,OAAO,KAAK,CAAC,IAAI,CAAC,cAAc,CAAC,CAAA;AAClC,CAAC;AAED,MAAM,UAAU,aAAa;IAC5B,OAAO;QACN,OAAO,EAAE,IAAI,GAAG,EAAE;QAClB,eAAe,EAAE,IAAI,GAAG,EAAE;QAC1B,WAAW,EAAE,IAAI,GAAG,EAAE;QACtB,WAAW,EAAE,IAAI,GAAG,EAAE;QACtB,SAAS,EAAE,IAAI,GAAG,EAAE;KACpB,CAAA;AACF,CAAC;AAED,SAAS,GAAG,CAAC,GAA6B,EAAE,GAAW,EAAE,KAAa;IACrE,MAAM,QAAQ,GAAG,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;IAE7B,IAAI,QAAQ,EAAE,CAAC;QACd,QAAQ,CAAC,GAAG,CAAC,KAAK,CAAC,CAAA;IACpB,CAAC;SAAM,CAAC;QACP,GAAG,CAAC,GAAG,CAAC,GAAG,EAAE,IAAI,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAA;IAC/B,CAAC;AACF,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,aAAa,CAAC,KAAe,EAAE,GAAiB;IAC/D,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC,MAAM,CAAC,CAAA;IAC7B,GAAG,CAAC,KAAK,CAAC,eAAe,EAAE,GAAG,CAAC,MAAM,EAAE,GAAG,CAAC,OAAO,CAAC,CAAA;IAEnD,MAAM,IAAI,GAAG,OAAO,CAAC,GAAG,CAAC,MAAM,EAAE,GAAG,CAAC,OAAO,CAAC,CAAA;IAE7C,GAAG,CAAC,KAAK,CAAC,WAAW,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,CAAC,CAAA;IAEtC,IAAI,GAAG,CAAC,QAAQ,EAAE,CAAC;QAClB,GAAG,CAAC,KAAK,CAAC,WAAW,EAAE,IAAI,EAAE,GAAG,CAAC,QAAQ,CAAC,CAAA;IAC3C,CAAC;IAED,IAAI,GAAG,CAAC,IAAI,EAAE,CAAC;QACd,GAAG,CAAC,KAAK,CAAC,WAAW,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,CAAC,CAAA;QAEtC,IAAI,GAAG,CAAC,EAAE,EAAE,CAAC;YACZ,GAAG,CAAC,KAAK,CAAC,SAAS,EAAE,OAAO,CAAC,GAAG,CAAC,MAAM,EAAE,GAAG,CAAC,OAAO,EAAE,GAAG,CAAC,IAAI,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC,CAAA;QACzE,CAAC;IACF,CAAC;AACF,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,aAAa,CAAC,KAAe;IAC5C,KAAK,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,cAAc,CAAC,EAAE,CAAC;QAChE,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC;YAAE,SAAQ;QAEzC,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,CAAA;QAEzB,MAAM,OAAO,GAAG,KAAK,CAAC,eAAe,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;QAElD,IAAI,OAAO,EAAE,CAAC;YACb,KAAK,CAAC,eAAe,CAAC,GAAG,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;QAC3C,CAAC;QAED,sGAAsG;QACtG,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC,WAAW,CAAC,EAAE,CAAC;YACnD,MAAM,CAAC,MAAM,EAAE,IAAI,CAAC,GAAG,GAAG,CAAC,KAAK,CAAC,cAAc,CAAC,CAAA;YAEhD,IAAI,MAAM,KAAK,OAAO,EAAE,CAAC;gBACxB,KAAK,CAAC,WAAW,CAAC,GAAG,CAAC,OAAO,CAAC,MAAM,EAAE,IAAK,CAAC,EAAE,KAAK,CAAC,CAAA;YACrD,CAAC;QACF,CAAC;QAED,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC,WAAW,CAAC,EAAE,CAAC;YACnD,MAAM,CAAC,MAAM,EAAE,IAAI,CAAC,GAAG,GAAG,CAAC,KAAK,CAAC,cAAc,CAAC,CAAA;YAEhD,IAAI,MAAM,KAAK,OAAO,EAAE,CAAC;gBACxB,KAAK,CAAC,WAAW,CAAC,GAAG,CAAC,OAAO,CAAC,MAAM,EAAE,IAAK,CAAC,EAAE,KAAK,CAAC,CAAA;YACrD,CAAC;QACF,CAAC;QAED,KAAK,MAAM,CAAC,GAAG,EAAE,GAAG,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC,SAAS,CAAC,EAAE,CAAC;YAC/C,MAAM,CAAC,MAAM,EAAE,IAAI,EAAE,IAAI,CAAC,GAAG,GAAG,CAAC,KAAK,CAAC,cAAc,CAAC,CAAA;YAEtD,IAAI,MAAM,KAAK,OAAO,EAAE,CAAC;gBACxB,KAAK,CAAC,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,MAAM,EAAE,IAAK,EAAE,IAAK,CAAC,EAAE,GAAG,CAAC,CAAA;YACxD,CAAC;QACF,CAAC;QACD,kCAAkC;IACnC,CAAC;AACF,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,WAAW,CAAC,KAAe,EAAE,MAAc,EAAE,MAAyB,EAAE,KAAa;IACpG,MAAM,UAAU,GAAG,KAAK,CAAC,eAAe,CAAC,GAAG,CAAC,MAAM,CAAC,CAAA;IAEpD,IAAI,CAAC,UAAU;QAAE,OAAO,CAAC,CAAA;IAEzB,KAAK,MAAM,KAAK,IAAI,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC;QAC5B,IAAI,KAAK,GAAG,KAAK,IAAI,MAAM,CAAC,MAAM,IAAI,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,KAAK,EAAE,KAAK,GAAG,KAAK,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC;YACpG,OAAO,KAAK,CAAA;QACb,CAAC;IACF,CAAC;IAED,OAAO,CAAC,CAAA;AACT,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,aAAa,CAAC,IAAiC;IACpE,MAAM,KAAK,GAAG,aAAa,EAAE,CAAA;IAE7B,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QAC9B,aAAa,CAAC,KAAK,EAAE,GAAG,CAAC,CAAA;IAC1B,CAAC;IAED,aAAa,CAAC,KAAK,CAAC,CAAA;IAEpB,OAAO,KAAK,CAAA;AACb,CAAC"}

package/out/kr/adapters/localdata/permit-rows.d.ts

@@ -0,0 +1,66 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The Korean permit registry (지방행정인허가데이터) as a NOISY corpus source: reading the delivered CSVs, and turning
+ *   an aligned string into a corpus row (#2204 §5).
+ *
+ *   The publisher delivers CP949, one file per business category, with both address systems typed by a clerk plus a
+ *   planar coordinate in EPSG:5174 and both postcodes. Reading is separated from ALIGNING so the alignment rate per
+ *   file can be measured before any row enters a corpus.
+ */
+import { type PathBuilderLike } from "path-ts";
+import type { Aligned } from "#kr/adapters/localdata/align";
+/**
+ * The source label every row of this adapter carries, and the name a config's `source_weights` addresses it by.
+ */
+export declare const SOURCE = "localdata-kr";
+/**
+ * Every permit in this registry is Korean; the publisher issues no others.
+ */
+export declare const COUNTRY = "KR";
+/**
+ * The 영업상태명 of a business still trading. A closed one's address records somewhere that was.
+ */
+export declare const OPEN_STATUS = "\uC601\uC5C5/\uC815\uC0C1";
+/**
+ * One permit, with both address systems as the clerk typed them.
+ */
+export interface PermitRow {
+    /**
+     * The file's basename: the business category the publisher splits by.
+     */
+    category: string;
+    name: string;
+    status: string;
+    roadAddress: string;
+    lotAddress: string;
+    roadPostcode: string;
+    lotPostcode: string;
+    x: number | null;
+    y: number | null;
+}
+export interface ReadPermitOptions {
+    /**
+     * Keep only these 영업상태명 values. Empty admits every status.
+     */
+    statuses?: readonly string[];
+}
+/**
+ * Stream every permit of one CSV.
+ */
+export declare function readPermitFile(path: PathBuilderLike, options?: ReadPermitOptions): AsyncGenerator<PermitRow>;
+/**
+ * Stream every permit of every CSV in a directory, in filename order.
+ */
+export declare function readPermitDirectory(directory: PathBuilderLike, options?: ReadPermitOptions): AsyncGenerator<PermitRow>;
+/**
+ * One aligned string as a row in the CJK slice schema.
+ *
+ * The token labels are derived from the spans rather than carried beside them: a token is `B-<tag>` when its FIRST
+ * character falls inside a span, `O` otherwise. A span covering several tokens therefore labels only the token it
+ * starts in, which is what the char-path trainer reads the spans for.
+ */
+export declare function alignedToSliceRow(aligned: Aligned): Record<string, unknown>;
+//# sourceMappingURL=permit-rows.d.ts.map

package/out/kr/adapters/localdata/permit-rows.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"permit-rows.d.ts","sourceRoot":"","sources":["../../../../lib/kr/adapters/localdata/permit-rows.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAGH,OAAO,EAA2B,KAAK,eAAe,EAAE,MAAM,SAAS,CAAA;AAIvE,OAAO,KAAK,EAAE,OAAO,EAAE,MAAM,8BAA8B,CAAA;AAE3D;;GAEG;AACH,eAAO,MAAM,MAAM,iBAAiB,CAAA;AAEpC;;GAEG;AACH,eAAO,MAAM,OAAO,OAAO,CAAA;AAE3B;;GAEG;AACH,eAAO,MAAM,WAAW,8BAAU,CAAA;AAOlC;;GAEG;AACH,MAAM,WAAW,SAAS;IACzB;;OAEG;IACH,QAAQ,EAAE,MAAM,CAAA;IAChB,IAAI,EAAE,MAAM,CAAA;IACZ,MAAM,EAAE,MAAM,CAAA;IACd,WAAW,EAAE,MAAM,CAAA;IACnB,UAAU,EAAE,MAAM,CAAA;IAClB,YAAY,EAAE,MAAM,CAAA;IACpB,WAAW,EAAE,MAAM,CAAA;IACnB,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;IAChB,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;CAChB;AA2BD,MAAM,WAAW,iBAAiB;IACjC;;OAEG;IACH,QAAQ,CAAC,EAAE,SAAS,MAAM,EAAE,CAAA;CAC5B;AAED;;GAEG;AACH,wBAAuB,cAAc,CACpC,IAAI,EAAE,eAAe,EACrB,OAAO,GAAE,iBAAsB,GAC7B,cAAc,CAAC,SAAS,CAAC,CA2B3B;AAED;;GAEG;AACH,wBAAuB,mBAAmB,CACzC,SAAS,EAAE,eAAe,EAC1B,OAAO,GAAE,iBAAsB,GAC7B,cAAc,CAAC,SAAS,CAAC,CAS3B;AAED;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAAC,OAAO,EAAE,OAAO,GAAG,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAoC3E"}

package/out/kr/adapters/localdata/permit-rows.js

@@ -0,0 +1,131 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The Korean permit registry (지방행정인허가데이터) as a NOISY corpus source: reading the delivered CSVs, and turning
+ *   an aligned string into a corpus row (#2204 §5).
+ *
+ *   The publisher delivers CP949, one file per business category, with both address systems typed by a clerk plus a
+ *   planar coordinate in EPSG:5174 and both postcodes. Reading is separated from ALIGNING so the alignment rate per
+ *   file can be measured before any row enters a corpus.
+ */
+import { decodeByteStream, openReadStream } from "@mailwoman/core/fs/streams";
+import { basename, extname, join } from "path-ts";
+import { CSVSpliterator } from "spliterator";
+import { Globerator } from "spliterator/node/fs";
+/**
+ * The source label every row of this adapter carries, and the name a config's `source_weights` addresses it by.
+ */
+export const SOURCE = "localdata-kr";
+/**
+ * Every permit in this registry is Korean; the publisher issues no others.
+ */
+export const COUNTRY = "KR";
+/**
+ * The 영업상태명 of a business still trading. A closed one's address records somewhere that was.
+ */
+export const OPEN_STATUS = "영업/정상";
+/**
+ * The publisher's encoding. `cp949` rather than `euc-kr` for the reason `decodeByteStream` states.
+ */
+const ENCODING = "cp949";
+/**
+ * The first of these column spellings the row carries, trimmed, or the empty string.
+ *
+ * The publisher's older and newer exports name the same field differently — `도로명주소` against `도로명전체주소`, `좌표정보(X)`
+ * against `좌표정보(x)` — and a directory holds both vintages, so each field names every spelling it has been delivered
+ * under.
+ */
+function firstColumn(row, ...names) {
+    for (const name of names) {
+        const value = row[name];
+        if (value !== undefined && value !== null)
+            return String(value).trim();
+    }
+    return "";
+}
+function toNumber(value) {
+    if (!value.trim())
+        return null;
+    const parsed = Number(value);
+    return Number.isFinite(parsed) ? parsed : null;
+}
+/**
+ * Stream every permit of one CSV.
+ */
+export async function* readPermitFile(path, options = {}) {
+    const wanted = new Set(options.statuses ?? [OPEN_STATUS]);
+    const category = basename(String(path), extname(String(path)));
+    const bytes = openReadStream(path);
+    for await (const row of CSVSpliterator.fromAsync(decodeByteStream(bytes, ENCODING))) {
+        const status = firstColumn(row, "영업상태명");
+        if (wanted.size && !wanted.has(status))
+            continue;
+        const roadAddress = firstColumn(row, "도로명주소", "도로명전체주소");
+        const lotAddress = firstColumn(row, "지번주소", "소재지전체주소");
+        if (!roadAddress && !lotAddress)
+            continue;
+        yield {
+            category,
+            name: firstColumn(row, "사업장명"),
+            status,
+            roadAddress,
+            lotAddress,
+            roadPostcode: firstColumn(row, "도로명우편번호"),
+            lotPostcode: firstColumn(row, "소재지우편번호"),
+            x: toNumber(firstColumn(row, "좌표정보(X)", "좌표정보(x)")),
+            y: toNumber(firstColumn(row, "좌표정보(Y)", "좌표정보(y)")),
+        };
+    }
+}
+/**
+ * Stream every permit of every CSV in a directory, in filename order.
+ */
+export async function* readPermitDirectory(directory, options = {}) {
+    const files = await Globerator.from("*", { cwd: directory, withFileTypes: true, onlyFiles: false })
+        .filter((entry) => entry.isFile() && entry.name.endsWith(".csv"))
+        .map((entry) => entry.name)
+        .toSorted();
+    for (const name of files) {
+        yield* readPermitFile(join(directory, name), options);
+    }
+}
+/**
+ * One aligned string as a row in the CJK slice schema.
+ *
+ * The token labels are derived from the spans rather than carried beside them: a token is `B-<tag>` when its FIRST
+ * character falls inside a span, `O` otherwise. A span covering several tokens therefore labels only the token it
+ * starts in, which is what the char-path trainer reads the spans for.
+ */
+export function alignedToSliceRow(aligned) {
+    const raw = aligned.raw;
+    const tokens = [];
+    const labels = [];
+    let cursor = 0;
+    for (const token of raw.split(/\s+/u).filter((piece) => piece.length)) {
+        const index = raw.indexOf(token, cursor);
+        cursor = index + token.length;
+        let label = "O";
+        for (let span = 0; span < aligned.spanStarts.length; span += 1) {
+            if (aligned.spanStarts[span] <= index && index < aligned.spanEnds[span]) {
+                label = `B-${aligned.spanTags[span]}`;
+                break;
+            }
+        }
+        tokens.push(token);
+        labels.push(label);
+    }
+    return {
+        raw,
+        tokens,
+        labels,
+        span_starts: [...aligned.spanStarts],
+        span_ends: [...aligned.spanEnds],
+        span_tags: [...aligned.spanTags],
+        country: COUNTRY,
+        source: SOURCE,
+        register: aligned.register,
+    };
+}
+//# sourceMappingURL=permit-rows.js.map

package/out/kr/adapters/localdata/permit-rows.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"permit-rows.js","sourceRoot":"","sources":["../../../../lib/kr/adapters/localdata/permit-rows.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,OAAO,EAAE,gBAAgB,EAAE,cAAc,EAAE,MAAM,4BAA4B,CAAA;AAC7E,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,IAAI,EAAwB,MAAM,SAAS,CAAA;AACvE,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAA;AAC5C,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAA;AAIhD;;GAEG;AACH,MAAM,CAAC,MAAM,MAAM,GAAG,cAAc,CAAA;AAEpC;;GAEG;AACH,MAAM,CAAC,MAAM,OAAO,GAAG,IAAI,CAAA;AAE3B;;GAEG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,OAAO,CAAA;AAElC;;GAEG;AACH,MAAM,QAAQ,GAAG,OAAO,CAAA;AAoBxB;;;;;;GAMG;AACH,SAAS,WAAW,CAAC,GAAuC,EAAE,GAAG,KAAwB;IACxF,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QAC1B,MAAM,KAAK,GAAG,GAAG,CAAC,IAAI,CAAC,CAAA;QAEvB,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,KAAK,IAAI;YAAE,OAAO,MAAM,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,CAAA;IACvE,CAAC;IAED,OAAO,EAAE,CAAA;AACV,CAAC;AAED,SAAS,QAAQ,CAAC,KAAa;IAC9B,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE;QAAE,OAAO,IAAI,CAAA;IAE9B,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,CAAA;IAE5B,OAAO,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAA;AAC/C,CAAC;AASD;;GAEG;AACH,MAAM,CAAC,KAAK,SAAS,CAAC,CAAC,cAAc,CACpC,IAAqB,EACrB,OAAO,GAAsB,EAAE;IAE/B,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,OAAO,CAAC,QAAQ,IAAI,CAAC,WAAW,CAAC,CAAC,CAAA;IACzD,MAAM,QAAQ,GAAG,QAAQ,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,OAAO,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;IAC9D,MAAM,KAAK,GAAG,cAAc,CAAC,IAAI,CAAC,CAAA;IAElC,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,cAAc,CAAC,SAAS,CAAyB,gBAAgB,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC,EAAE,CAAC;QAC7G,MAAM,MAAM,GAAG,WAAW,CAAC,GAAG,EAAE,OAAO,CAAC,CAAA;QAExC,IAAI,MAAM,CAAC,IAAI,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC;YAAE,SAAQ;QAEhD,MAAM,WAAW,GAAG,WAAW,CAAC,GAAG,EAAE,OAAO,EAAE,SAAS,CAAC,CAAA;QACxD,MAAM,UAAU,GAAG,WAAW,CAAC,GAAG,EAAE,MAAM,EAAE,SAAS,CAAC,CAAA;QAEtD,IAAI,CAAC,WAAW,IAAI,CAAC,UAAU;YAAE,SAAQ;QAEzC,MAAM;YACL,QAAQ;YACR,IAAI,EAAE,WAAW,CAAC,GAAG,EAAE,MAAM,CAAC;YAC9B,MAAM;YACN,WAAW;YACX,UAAU;YACV,YAAY,EAAE,WAAW,CAAC,GAAG,EAAE,SAAS,CAAC;YACzC,WAAW,EAAE,WAAW,CAAC,GAAG,EAAE,SAAS,CAAC;YACxC,CAAC,EAAE,QAAQ,CAAC,WAAW,CAAC,GAAG,EAAE,SAAS,EAAE,SAAS,CAAC,CAAC;YACnD,CAAC,EAAE,QAAQ,CAAC,WAAW,CAAC,GAAG,EAAE,SAAS,EAAE,SAAS,CAAC,CAAC;SACnD,CAAA;IACF,CAAC;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,KAAK,SAAS,CAAC,CAAC,mBAAmB,CACzC,SAA0B,EAC1B,OAAO,GAAsB,EAAE;IAE/B,MAAM,KAAK,GAAG,MAAM,UAAU,CAAC,IAAI,CAAC,GAAG,EAAE,EAAE,GAAG,EAAE,SAAS,EAAE,aAAa,EAAE,IAAI,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC;SACjG,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,MAAM,EAAE,IAAI,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;SAChE,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC;SAC1B,QAAQ,EAAE,CAAA;IAEZ,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QAC1B,KAAK,CAAC,CAAC,cAAc,CAAC,IAAI,CAAC,SAAS,EAAE,IAAI,CAAC,EAAE,OAAO,CAAC,CAAA;IACtD,CAAC;AACF,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,iBAAiB,CAAC,OAAgB;IACjD,MAAM,GAAG,GAAG,OAAO,CAAC,GAAG,CAAA;IACvB,MAAM,MAAM,GAAa,EAAE,CAAA;IAC3B,MAAM,MAAM,GAAa,EAAE,CAAA;IAC3B,IAAI,MAAM,GAAG,CAAC,CAAA;IAEd,KAAK,MAAM,KAAK,IAAI,GAAG,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,CAAC;QACvE,MAAM,KAAK,GAAG,GAAG,CAAC,OAAO,CAAC,KAAK,EAAE,MAAM,CAAC,CAAA;QAExC,MAAM,GAAG,KAAK,GAAG,KAAK,CAAC,MAAM,CAAA;QAE7B,IAAI,KAAK,GAAG,GAAG,CAAA;QAEf,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,GAAG,OAAO,CAAC,UAAU,CAAC,MAAM,EAAE,IAAI,IAAI,CAAC,EAAE,CAAC;YAChE,IAAI,OAAO,CAAC,UAAU,CAAC,IAAI,CAAE,IAAI,KAAK,IAAI,KAAK,GAAG,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAE,EAAE,CAAC;gBAC3E,KAAK,GAAG,KAAK,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAA;gBAErC,MAAK;YACN,CAAC;QACF,CAAC;QAED,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QAClB,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;IACnB,CAAC;IAED,OAAO;QACN,GAAG;QACH,MAAM;QACN,MAAM;QACN,WAAW,EAAE,CAAC,GAAG,OAAO,CAAC,UAAU,CAAC;QACpC,SAAS,EAAE,CAAC,GAAG,OAAO,CAAC,QAAQ,CAAC;QAChC,SAAS,EAAE,CAAC,GAAG,OAAO,CAAC,QAAQ,CAAC;QAChC,OAAO,EAAE,OAAO;QAChB,MAAM,EAAE,MAAM;QACd,QAAQ,EAAE,OAAO,CAAC,QAAQ;KAC1B,CAAA;AACF,CAAC"}

package/out/kr/tools/fetch/juso.d.ts

@@ -0,0 +1,30 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Fetch the Korean road-name address register (도로명주소 주소DB) from the ministry's address-industry
+ *   portal, business.juso.go.kr. This is the LABEL source for Korean addresses: about 6,000,000
+ *   road-name addresses and 8,000,000 lot-number (지번) records, split by the publisher into four
+ *   pipe-delimited CP949 text files per region (도로명코드 / 주소 / 지번 / 부가정보), with the postcode in
+ *   the supplementary file. No coordinates. The public-data portal labels it "이용허락범위 제한 없음"
+ *   (no restriction on the scope of use), and it is a direct download with no application step — the
+ *   distinction that matters against the ministry's entrance-coordinate products, which are provided
+ *   only after a purpose-of-use review (`docs/superpowers/plans/counsel-dossier.md` §4).
+ *
+ *   The portal is a single-page app; the file behind it is reached in two calls. `selectAttrbDBDwldList`
+ *   lists the monthly full files for one product (`rtlDtaDtlSn` 8 is the 주소DB, 2 the English
+ *   road-name DB), and `/api/jst/download` streams one by the parameters the listing carried. The
+ *   English DB rides along because it is the Latin-script half of the same register.
+ *
+ *   Invoke via `mailwoman corpus fetch juso-kr --out-root <path>`.
+ */
+import type { BaseFetchOptions, FetchSummary } from "#tools/fetch/download/index";
+export interface FetchJusoKROptions extends BaseFetchOptions {
+    /**
+     * The month to fetch as `YYYYMM`. Defaults to the latest month the portal lists as present.
+     */
+    month?: string;
+}
+export declare function fetchJusoKR(options: FetchJusoKROptions, report?: (line: string) => void): Promise<FetchSummary>;
+//# sourceMappingURL=juso.d.ts.map

package/out/kr/tools/fetch/juso.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"juso.d.ts","sourceRoot":"","sources":["../../../../lib/kr/tools/fetch/juso.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;GAoBG;AAMH,OAAO,KAAK,EACX,gBAAgB,EAChB,YAAY,EAGZ,MAAM,6BAA6B,CAAA;AAqCpC,MAAM,WAAW,kBAAmB,SAAQ,gBAAgB;IAC3D;;OAEG;IACH,KAAK,CAAC,EAAE,MAAM,CAAA;CACd;AAsCD,wBAAsB,WAAW,CAAC,OAAO,EAAE,kBAAkB,EAAE,MAAM,CAAC,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,GAAG,OAAO,CAAC,YAAY,CAAC,CA4DrH"}

package/out/kr/tools/fetch/juso.js

@@ -0,0 +1,120 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Fetch the Korean road-name address register (도로명주소 주소DB) from the ministry's address-industry
+ *   portal, business.juso.go.kr. This is the LABEL source for Korean addresses: about 6,000,000
+ *   road-name addresses and 8,000,000 lot-number (지번) records, split by the publisher into four
+ *   pipe-delimited CP949 text files per region (도로명코드 / 주소 / 지번 / 부가정보), with the postcode in
+ *   the supplementary file. No coordinates. The public-data portal labels it "이용허락범위 제한 없음"
+ *   (no restriction on the scope of use), and it is a direct download with no application step — the
+ *   distinction that matters against the ministry's entrance-coordinate products, which are provided
+ *   only after a purpose-of-use review (`docs/superpowers/plans/counsel-dossier.md` §4).
+ *
+ *   The portal is a single-page app; the file behind it is reached in two calls. `selectAttrbDBDwldList`
+ *   lists the monthly full files for one product (`rtlDtaDtlSn` 8 is the 주소DB, 2 the English
+ *   road-name DB), and `/api/jst/download` streams one by the parameters the listing carried. The
+ *   English DB rides along because it is the Latin-script half of the same register.
+ *
+ *   Invoke via `mailwoman corpus fetch juso-kr --out-root <path>`.
+ */
+import { makeDirectories } from "@mailwoman/core/fs/writers";
+import { sha256File } from "@mailwoman/core/hash";
+import { join } from "path-ts";
+import { resumableDownload, writeManifest } from "#tools/fetch/download/index";
+const SLUG = "juso-kr";
+const PORTAL = "https://business.juso.go.kr";
+const LIST_URL = `${PORTAL}/api/jst/selectAttrbDBDwldList`;
+const DOWNLOAD_URL = `${PORTAL}/api/jst/download`;
+const REFERER = `${PORTAL}/jst/jstAddressDownload?menu=30`;
+/**
+ * The portal's product ids for the two registers this module fetches.
+ */
+const PRODUCTS = [
+    { sn: "8", name: "주소DB" },
+    { sn: "2", name: "도로명주소 영문" },
+];
+const LICENSE = "이용허락범위 제한 없음 (data.go.kr label for 행정안전부_도로명주소 주소DB); 공공누리 attribution to 행정안전부";
+const ATTRIBUTION = "행정안전부 (Ministry of the Interior and Safety), 도로명주소 (juso.go.kr)";
+const BROWSER_HEADERS = {
+    "user-agent": "Mozilla/5.0 (X11; Linux x86_64) mailwoman-corpus-fetch",
+    referer: REFERER,
+};
+/**
+ * The monthly full files the portal lists for one product over the twelve months ending at `year`/`month`.
+ */
+async function listMonthlyFiles(sn, year, month) {
+    const res = await fetch(LIST_URL, {
+        method: "POST",
+        headers: { ...BROWSER_HEADERS, accept: "application/json", "content-type": "application/json;charset=UTF-8" },
+        body: JSON.stringify({ rtlDtaDtlSn: sn, year, month, expand: "Y" }),
+    });
+    if (!res.ok)
+        throw new Error(`juso-kr: listing product ${sn} answered HTTP ${res.status}`);
+    const json = (await res.json());
+    return (json.results?.allMonthFileList ?? []).filter((file) => file.isExist === "Y");
+}
+function downloadURL(file, regYmd) {
+    const query = new URLSearchParams({
+        reqType: file.fileTypeNm,
+        ctprvnCd: file.ctpvClsfCd,
+        stdde: file.crtrYm,
+        fileName: file.fileNm,
+        realFileName: file.tmprFileNm,
+        intFileNo: file.fileSn ?? "0",
+        intNum: file.atflNo ?? "0",
+        regYmd,
+    });
+    return `${DOWNLOAD_URL}?${query.toString()}`;
+}
+export async function fetchJusoKR(options, report) {
+    const destDir = join(options.outRoot, SLUG);
+    await makeDirectories(destDir);
+    const now = new Date();
+    const files = [];
+    const failedCodes = [];
+    for (const product of PRODUCTS) {
+        const listed = await listMonthlyFiles(product.sn, now.getFullYear(), now.getMonth() + 1);
+        const file = options.month ? listed.find((entry) => entry.crtrYm === options.month) : listed.at(-1);
+        if (!file) {
+            report?.(`  ✗ ${product.name}: no full file listed${options.month ? ` for ${options.month}` : ""}`);
+            failedCodes.push(product.name);
+            continue;
+        }
+        const url = downloadURL(file, String(now.getFullYear()));
+        const dest = join(destDir, file.tmprFileNm);
+        report?.(`=== ${SLUG} / ${product.name} ${file.crtrYm} → ${file.tmprFileNm}`);
+        let bytes;
+        try {
+            // The portal drops a connection every few megabytes and answers Range with 206, so the transfer resumes
+            // from what landed rather than starting the 181 MB over.
+            bytes = await resumableDownload({ url, dest, headers: BROWSER_HEADERS, report });
+        }
+        catch (error) {
+            report?.(`  ✗ ${error.message}`);
+            failedCodes.push(product.name);
+            continue;
+        }
+        const sha = await sha256File(dest);
+        files.push({
+            source_url: url,
+            downloaded_at: new Date().toISOString(),
+            filename: file.tmprFileNm,
+            sha256: sha,
+            bytes,
+        });
+        report?.(`  ✓ ${(bytes / 1024 / 1024).toFixed(1)} MB  sha256=${sha}`);
+    }
+    const manifest = {
+        source: SLUG,
+        source_url: REFERER,
+        license: LICENSE,
+        attribution: ATTRIBUTION,
+        downloaded_at: new Date().toISOString(),
+        files,
+    };
+    await writeManifest(join(destDir, "MANIFEST.json"), manifest);
+    return { fetched: files.length, skipped: 0, failed: failedCodes.length, failedCodes };
+}
+//# sourceMappingURL=juso.js.map

package/out/kr/tools/fetch/juso.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"juso.js","sourceRoot":"","sources":["../../../../lib/kr/tools/fetch/juso.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;GAoBG;AAEH,OAAO,EAAE,eAAe,EAAE,MAAM,4BAA4B,CAAA;AAC5D,OAAO,EAAE,UAAU,EAAE,MAAM,sBAAsB,CAAA;AACjD,OAAO,EAAE,IAAI,EAAE,MAAM,SAAS,CAAA;AAQ9B,OAAO,EAAE,iBAAiB,EAAE,aAAa,EAAE,MAAM,6BAA6B,CAAA;AAE9E,MAAM,IAAI,GAAG,SAAS,CAAA;AACtB,MAAM,MAAM,GAAG,6BAA6B,CAAA;AAC5C,MAAM,QAAQ,GAAG,GAAG,MAAM,gCAAgC,CAAA;AAC1D,MAAM,YAAY,GAAG,GAAG,MAAM,mBAAmB,CAAA;AACjD,MAAM,OAAO,GAAG,GAAG,MAAM,iCAAiC,CAAA;AAE1D;;GAEG;AACH,MAAM,QAAQ,GAAG;IAChB,EAAE,EAAE,EAAE,GAAG,EAAE,IAAI,EAAE,MAAM,EAAE;IACzB,EAAE,EAAE,EAAE,GAAG,EAAE,IAAI,EAAE,UAAU,EAAE;CACpB,CAAA;AAEV,MAAM,OAAO,GACZ,iFAAiF,CAAA;AAElF,MAAM,WAAW,GAAG,iEAAiE,CAAA;AAwBrF,MAAM,eAAe,GAAG;IACvB,YAAY,EAAE,wDAAwD;IACtE,OAAO,EAAE,OAAO;CAChB,CAAA;AAED;;GAEG;AACH,KAAK,UAAU,gBAAgB,CAAC,EAAU,EAAE,IAAY,EAAE,KAAa;IACtE,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,QAAQ,EAAE;QACjC,MAAM,EAAE,MAAM;QACd,OAAO,EAAE,EAAE,GAAG,eAAe,EAAE,MAAM,EAAE,kBAAkB,EAAE,cAAc,EAAE,gCAAgC,EAAE;QAC7G,IAAI,EAAE,IAAI,CAAC,SAAS,CAAC,EAAE,WAAW,EAAE,EAAE,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,GAAG,EAAE,CAAC;KACnE,CAAC,CAAA;IAEF,IAAI,CAAC,GAAG,CAAC,EAAE;QAAE,MAAM,IAAI,KAAK,CAAC,4BAA4B,EAAE,kBAAkB,GAAG,CAAC,MAAM,EAAE,CAAC,CAAA;IAC1F,MAAM,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,CAAiB,CAAA;IAE/C,OAAO,CAAC,IAAI,CAAC,OAAO,EAAE,gBAAgB,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,OAAO,KAAK,GAAG,CAAC,CAAA;AACrF,CAAC;AAED,SAAS,WAAW,CAAC,IAAgB,EAAE,MAAc;IACpD,MAAM,KAAK,GAAG,IAAI,eAAe,CAAC;QACjC,OAAO,EAAE,IAAI,CAAC,UAAU;QACxB,QAAQ,EAAE,IAAI,CAAC,UAAU;QACzB,KAAK,EAAE,IAAI,CAAC,MAAM;QAClB,QAAQ,EAAE,IAAI,CAAC,MAAM;QACrB,YAAY,EAAE,IAAI,CAAC,UAAU;QAC7B,SAAS,EAAE,IAAI,CAAC,MAAM,IAAI,GAAG;QAC7B,MAAM,EAAE,IAAI,CAAC,MAAM,IAAI,GAAG;QAC1B,MAAM;KACN,CAAC,CAAA;IAEF,OAAO,GAAG,YAAY,IAAI,KAAK,CAAC,QAAQ,EAAE,EAAE,CAAA;AAC7C,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,WAAW,CAAC,OAA2B,EAAE,MAA+B;IAC7F,MAAM,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,IAAI,CAAC,CAAA;IAC3C,MAAM,eAAe,CAAC,OAAO,CAAC,CAAA;IAE9B,MAAM,GAAG,GAAG,IAAI,IAAI,EAAE,CAAA;IACtB,MAAM,KAAK,GAAqB,EAAE,CAAA;IAClC,MAAM,WAAW,GAAa,EAAE,CAAA;IAEhC,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAChC,MAAM,MAAM,GAAG,MAAM,gBAAgB,CAAC,OAAO,CAAC,EAAE,EAAE,GAAG,CAAC,WAAW,EAAE,EAAE,GAAG,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC,CAAA;QACxF,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,MAAM,KAAK,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAA;QAEnG,IAAI,CAAC,IAAI,EAAE,CAAC;YACX,MAAM,EAAE,CAAC,OAAO,OAAO,CAAC,IAAI,wBAAwB,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,QAAQ,OAAO,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAA;YACnG,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;YAE9B,SAAQ;QACT,CAAC;QAED,MAAM,GAAG,GAAG,WAAW,CAAC,IAAI,EAAE,MAAM,CAAC,GAAG,CAAC,WAAW,EAAE,CAAC,CAAC,CAAA;QACxD,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,EAAE,IAAI,CAAC,UAAU,CAAC,CAAA;QAC3C,MAAM,EAAE,CAAC,OAAO,IAAI,MAAM,OAAO,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,MAAM,IAAI,CAAC,UAAU,EAAE,CAAC,CAAA;QAC7E,IAAI,KAAa,CAAA;QAEjB,IAAI,CAAC;YACJ,wGAAwG;YACxG,yDAAyD;YACzD,KAAK,GAAG,MAAM,iBAAiB,CAAC,EAAE,GAAG,EAAE,IAAI,EAAE,OAAO,EAAE,eAAe,EAAE,MAAM,EAAE,CAAC,CAAA;QACjF,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YAChB,MAAM,EAAE,CAAC,OAAQ,KAAe,CAAC,OAAO,EAAE,CAAC,CAAA;YAC3C,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;YAE9B,SAAQ;QACT,CAAC;QAED,MAAM,GAAG,GAAG,MAAM,UAAU,CAAC,IAAI,CAAC,CAAA;QAElC,KAAK,CAAC,IAAI,CAAC;YACV,UAAU,EAAE,GAAG;YACf,aAAa,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;YACvC,QAAQ,EAAE,IAAI,CAAC,UAAU;YACzB,MAAM,EAAE,GAAG;YACX,KAAK;SACL,CAAC,CAAA;QAEF,MAAM,EAAE,CAAC,OAAO,CAAC,KAAK,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,eAAe,GAAG,EAAE,CAAC,CAAA;IACtE,CAAC;IAED,MAAM,QAAQ,GAA6B;QAC1C,MAAM,EAAE,IAAI;QACZ,UAAU,EAAE,OAAO;QACnB,OAAO,EAAE,OAAO;QAChB,WAAW,EAAE,WAAW;QACxB,aAAa,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;QACvC,KAAK;KACL,CAAA;IAED,MAAM,aAAa,CAAC,IAAI,CAAC,OAAO,EAAE,eAAe,CAAC,EAAE,QAAQ,CAAC,CAAA;IAE7D,OAAO,EAAE,OAAO,EAAE,KAAK,CAAC,MAAM,EAAE,OAAO,EAAE,CAAC,EAAE,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,WAAW,EAAE,CAAA;AACtF,CAAC"}

package/out/kr/tools/fetch/localdata.d.ts

@@ -0,0 +1,30 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Fetch Korea's local-government permit registry (지방행정인허가데이터): one CSV per permit category,
+ *   every business a local authority has licensed, about 195 categories. Each row carries BOTH address
+ *   forms of the same premises — `소재지전체주소` (the lot-number form, 지번) and `도로명전체주소` (the
+ *   road-name form) — plus both postcodes and a planar coordinate (`좌표정보(x/y)`, EPSG:5174). That
+ *   pairing is the cheapest dual-format signal Korean addresses offer, and the coordinate is the
+ *   board's coordinate half, which is why this is the NOISY source for Korean and not a side dish.
+ *
+ *   The public-data portal labels the category files "이용허락범위 제한 없음" and links them to
+ *   `file.localdata.go.kr`, which serves them behind a session: the category page sets the XSRF cookie,
+ *   `/file/validate/download-count` is the portal's own rate check (429 when it wants a pause), and
+ *   `/file/download/<slug>/info` streams the CSV. The files are CP949 as delivered; the adapter
+ *   decodes. The restaurant category alone is about 700 MB.
+ *
+ *   Invoke via `mailwoman corpus fetch localdata-kr --out-root <path>`, optionally with
+ *   `--categories general_restaurants,rest_cafes`.
+ */
+import type { BaseFetchOptions, FetchSummary } from "#tools/fetch/download/index";
+export interface FetchLocaldataKROptions extends BaseFetchOptions {
+    /**
+     * Category slugs to fetch (the path segment of `/file/<slug>/info`). Defaults to every category the portal lists.
+     */
+    categories?: string[];
+}
+export declare function fetchLocaldataKR(options: FetchLocaldataKROptions, report?: (line: string) => void): Promise<FetchSummary>;
+//# sourceMappingURL=localdata.d.ts.map

package/out/kr/tools/fetch/localdata.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"localdata.d.ts","sourceRoot":"","sources":["../../../../lib/kr/tools/fetch/localdata.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;GAoBG;AASH,OAAO,KAAK,EACX,gBAAgB,EAChB,YAAY,EAGZ,MAAM,6BAA6B,CAAA;AAiBpC,MAAM,WAAW,uBAAwB,SAAQ,gBAAgB;IAChE;;OAEG;IACH,UAAU,CAAC,EAAE,MAAM,EAAE,CAAA;CACrB;AAsDD,wBAAsB,gBAAgB,CACrC,OAAO,EAAE,uBAAuB,EAChC,MAAM,CAAC,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,GAC7B,OAAO,CAAC,YAAY,CAAC,CA4FvB"}