@mailwoman/corpus 9.4.0 → 10.0.0

package/out/sg/recipes/register.d.ts

@@ -0,0 +1,69 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `sg-register` slice recipe — the forms a person in Singapore types, rendered over the Overture-SG register rows
+ *   (#2204 §4). The base `overture` source carries the register's own shape, `990 OLD CHOA CHU KANG ROAD, 699814,
+ *   Singapore`, and the Latin model has no register for the three forms that dominate typed Singapore addresses:
+ *
+ *   - the HDB block line, `Blk 123 Ang Mo Kio Ave 3 #05-67 Singapore 560123` — `Blk` in front of the block number (the
+ *     register's `number`), a floor-unit `#NN-NN` after the street, the country name as the locality;
+ *   - the bracketed postcode, `123 Ang Mo Kio Ave 3 S(560123)` — the form the postal service prints;
+ *   - the building-led line, `National Shooting Centre, 990 Old Choa Chu Kang Road, Singapore 699814` — the register's
+ *     `unit` field holds the BUILDING or ESTATE name on 91,818 of 142,210 rows, which the adapter refuses as a unit and
+ *     this recipe renders as a `venue` when it reads as a building rather than an estate.
+ *
+ *   Every value in `components` is a verbatim substring of `raw`, aligned by {@link alignAndWrite}. The floor-unit is
+ *   SYNTHESIZED (the register carries one on a single row): floor 01–30, unit 01–399, zero-padded — the shape, not a
+ *   real occupancy. `Blk` and the `S(` `)` around a postcode are untagged, as the board's `sg-cs-blk-12-kallang-ave`
+ *   row expects (`house_number: "12"`). Streets arrive upper-case and are rendered title-cased; a third abbreviate the
+ *   generic (`Avenue` → `Ave`, `Road` → `Rd`, …) the way a typed line does.
+ *
+ *   Run: mailwoman corpus slice sg-register --input <overture-sg.corpus.jsonl> --count N --seed S
+ */
+import { type CorpusRecipe } from "#recipes/scaffold";
+/**
+ * `OLD CHOA CHU KANG ROAD` → `Old Choa Chu Kang Road`; an already-mixed value is returned as it is. The register writes
+ * every street and building name upper-case, the same convention as G-NAF, so the G-NAF title-caser serves both.
+ */
+export declare function titleCaseSGName(name: string): string;
+/**
+ * Abbreviate the street's generic, wherever it sits (`Ang Mo Kio Avenue 3` → `Ang Mo Kio Ave 3`).
+ */
+export declare function abbreviateSGStreet(street: string): string;
+/**
+ * Whether the register's `unit` value names a building rather than an estate — the building-led register's lead. A name
+ * carrying a digit, a parenthesized abbreviation (`DIABETES & METABOLISM CENTRE (DMC)`) or one word only is not one.
+ */
+export declare function isBuildingName(unit: string): boolean;
+interface SGRow {
+    street: string;
+    number: string;
+    unit: string;
+    postcode: string;
+}
+/**
+ * The four typed forms this recipe renders.
+ */
+export type SGRegister = "official" | "block" | "bracket_postcode" | "building_led";
+/**
+ * One rendered register: the raw line plus the components it carries, every value a substring of the line.
+ */
+export interface SGRegisterRendering {
+    raw: string;
+    components: Record<string, string>;
+    register: SGRegister;
+}
+/**
+ * Render one register form for one register row. Pure: every random draw comes through `random`. A `register` names the
+ * form to render instead of drawing one, for a board that wants a balanced spread; `building_led` still needs a
+ * building-shaped `unit` and falls back to `official` without one.
+ */
+export declare function renderSGRegister(row: SGRow, random: () => number, register?: SGRegister): SGRegisterRendering;
+/**
+ * Slice recipe registered with the corpus builder — see the file header for the registers and why each is here.
+ */
+export declare const sgRegisterRecipe: CorpusRecipe;
+export {};
+//# sourceMappingURL=register.d.ts.map

package/out/sg/recipes/register.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"register.d.ts","sourceRoot":"","sources":["../../../lib/sg/recipes/register.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AAKH,OAAO,EAAiB,KAAK,YAAY,EAA6B,MAAM,mBAAmB,CAAA;AAmD/F;;;GAGG;AACH,wBAAgB,eAAe,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAEpD;AAED;;GAEG;AACH,wBAAgB,kBAAkB,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,CAQzD;AAED;;;GAGG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAMpD;AAED,UAAU,KAAK;IACd,MAAM,EAAE,MAAM,CAAA;IACd,MAAM,EAAE,MAAM,CAAA;IACd,IAAI,EAAE,MAAM,CAAA;IACZ,QAAQ,EAAE,MAAM,CAAA;CAChB;AAED;;GAEG;AACH,MAAM,MAAM,UAAU,GAAG,UAAU,GAAG,OAAO,GAAG,kBAAkB,GAAG,cAAc,CAAA;AAEnF;;GAEG;AACH,MAAM,WAAW,mBAAmB;IACnC,GAAG,EAAE,MAAM,CAAA;IACX,UAAU,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IAClC,QAAQ,EAAE,UAAU,CAAA;CACpB;AAED;;;;GAIG;AACH,wBAAgB,gBAAgB,CAAC,GAAG,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,MAAM,EAAE,QAAQ,CAAC,EAAE,UAAU,GAAG,mBAAmB,CA2C7G;AAED;;GAEG;AACH,eAAO,MAAM,gBAAgB,EAAE,YAmF9B,CAAA"}

package/out/sg/recipes/register.js

@@ -0,0 +1,211 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `sg-register` slice recipe — the forms a person in Singapore types, rendered over the Overture-SG register rows
+ *   (#2204 §4). The base `overture` source carries the register's own shape, `990 OLD CHOA CHU KANG ROAD, 699814,
+ *   Singapore`, and the Latin model has no register for the three forms that dominate typed Singapore addresses:
+ *
+ *   - the HDB block line, `Blk 123 Ang Mo Kio Ave 3 #05-67 Singapore 560123` — `Blk` in front of the block number (the
+ *     register's `number`), a floor-unit `#NN-NN` after the street, the country name as the locality;
+ *   - the bracketed postcode, `123 Ang Mo Kio Ave 3 S(560123)` — the form the postal service prints;
+ *   - the building-led line, `National Shooting Centre, 990 Old Choa Chu Kang Road, Singapore 699814` — the register's
+ *     `unit` field holds the BUILDING or ESTATE name on 91,818 of 142,210 rows, which the adapter refuses as a unit and
+ *     this recipe renders as a `venue` when it reads as a building rather than an estate.
+ *
+ *   Every value in `components` is a verbatim substring of `raw`, aligned by {@link alignAndWrite}. The floor-unit is
+ *   SYNTHESIZED (the register carries one on a single row): floor 01–30, unit 01–399, zero-padded — the shape, not a
+ *   real occupancy. `Blk` and the `S(` `)` around a postcode are untagged, as the board's `sg-cs-blk-12-kallang-ave`
+ *   row expects (`house_number: "12"`). Streets arrive upper-case and are rendered title-cased; a third abbreviate the
+ *   generic (`Avenue` → `Ave`, `Road` → `Rd`, …) the way a typed line does.
+ *
+ *   Run: mailwoman corpus slice sg-register --input <overture-sg.corpus.jsonl> --count N --seed S
+ */
+import { mulberry32 as makeMulberry32 } from "@mailwoman/core/utils";
+import { titlecase } from "#au/adapters/gnaf/assemble";
+import { alignAndWrite, readTuples, sliceSourceID } from "#recipes/scaffold";
+const SOURCE = "synth-sg-register";
+/**
+ * The register's generics and the abbreviation a typed line uses for each.
+ */
+const GENERIC_ABBREVIATIONS = [
+    ["Avenue", "Ave"],
+    ["Road", "Rd"],
+    ["Street", "St"],
+    ["Drive", "Dr"],
+    ["Crescent", "Cres"],
+    ["Lane", "Ln"],
+    ["Boulevard", "Blvd"],
+    ["Terrace", "Ter"],
+    ["Place", "Pl"],
+    ["Close", "Cl"],
+    ["Walk", "Wk"],
+];
+/**
+ * Estate and area names the register's `unit` field carries beside building names. A name ending in one of these is an
+ * estate, which no one writes in front of an address line; a name without one reads as a building and is rendered as a
+ * `venue`.
+ */
+const ESTATE_TAILS = ["ESTATE", "CONSERVATION AREA", "HILLS", "PARK", "GARDENS", "GARDEN", "HEIGHTS", "GROVE", "VILLE"];
+/**
+ * Share of rows per register. The block line and the bracketed postcode are the two the Latin model has never seen; the
+ * official line keeps the register's own shape in the mix so the block forms train beside it.
+ */
+const P_BLOCK = 0.4;
+const P_BRACKET_POSTCODE = 0.25;
+const P_BUILDING_LED = 0.15;
+const P_ABBREVIATE = 0.35;
+const P_UPPER = 0.1;
+const FLOOR_MAX = 30;
+const UNIT_MAX = 399;
+/**
+ * The draw value that lands each register in the branch order below, for a caller that names the form.
+ */
+const REGISTER_DRAW = {
+    block: 0,
+    bracket_postcode: P_BLOCK,
+    building_led: P_BLOCK + P_BRACKET_POSTCODE,
+    official: 1,
+};
+/**
+ * `OLD CHOA CHU KANG ROAD` → `Old Choa Chu Kang Road`; an already-mixed value is returned as it is. The register writes
+ * every street and building name upper-case, the same convention as G-NAF, so the G-NAF title-caser serves both.
+ */
+export function titleCaseSGName(name) {
+    return name === name.toUpperCase() ? titlecase(name) : name;
+}
+/**
+ * Abbreviate the street's generic, wherever it sits (`Ang Mo Kio Avenue 3` → `Ang Mo Kio Ave 3`).
+ */
+export function abbreviateSGStreet(street) {
+    for (const [full, short] of GENERIC_ABBREVIATIONS) {
+        const pattern = new RegExp(`\\b${full}\\b`, "u");
+        if (pattern.test(street))
+            return street.replace(pattern, short);
+    }
+    return street;
+}
+/**
+ * Whether the register's `unit` value names a building rather than an estate — the building-led register's lead. A name
+ * carrying a digit, a parenthesized abbreviation (`DIABETES & METABOLISM CENTRE (DMC)`) or one word only is not one.
+ */
+export function isBuildingName(unit) {
+    const value = unit.trim().toUpperCase();
+    if (!value || value === "NIL" || /[\d()]/u.test(value) || !/\s/u.test(value))
+        return false;
+    return !ESTATE_TAILS.some((tail) => value.endsWith(tail));
+}
+/**
+ * Render one register form for one register row. Pure: every random draw comes through `random`. A `register` names the
+ * form to render instead of drawing one, for a board that wants a balanced spread; `building_led` still needs a
+ * building-shaped `unit` and falls back to `official` without one.
+ */
+export function renderSGRegister(row, random, register) {
+    const street0 = titleCaseSGName(row.street);
+    const street = random() < P_ABBREVIATE ? abbreviateSGStreet(street0) : street0;
+    const drawn = random();
+    const building = isBuildingName(row.unit) ? titleCaseSGName(row.unit) : null;
+    const draw = register === undefined ? drawn : REGISTER_DRAW[register];
+    let raw;
+    let rendered;
+    const components = { house_number: row.number, street, postcode: row.postcode };
+    if (draw < P_BLOCK && /^\d+[A-Z]?$/u.test(row.number)) {
+        const floor = String(1 + Math.floor(random() * FLOOR_MAX)).padStart(2, "0");
+        const unitNo = String(1 + Math.floor(random() * UNIT_MAX)).padStart(2, "0");
+        const unit = `#${floor}-${unitNo}`;
+        raw = `Blk ${row.number} ${street} ${unit} Singapore ${row.postcode}`;
+        components.unit = unit;
+        components.locality = "Singapore";
+        rendered = "block";
+    }
+    else if (draw < P_BLOCK + P_BRACKET_POSTCODE) {
+        raw = `${row.number} ${street} S(${row.postcode})`;
+        rendered = "bracket_postcode";
+    }
+    else if (draw < P_BLOCK + P_BRACKET_POSTCODE + P_BUILDING_LED && building) {
+        raw = `${building}, ${row.number} ${street}, Singapore ${row.postcode}`;
+        components.venue = building;
+        components.locality = "Singapore";
+        rendered = "building_led";
+    }
+    else {
+        raw = `${row.number} ${street} Singapore ${row.postcode}`;
+        components.locality = "Singapore";
+        rendered = "official";
+    }
+    if (random() < P_UPPER) {
+        raw = raw.toUpperCase();
+        for (const key of Object.keys(components)) {
+            components[key] = components[key].toUpperCase();
+        }
+    }
+    return { raw, components, register: rendered };
+}
+/**
+ * Slice recipe registered with the corpus builder — see the file header for the registers and why each is here.
+ */
+export const sgRegisterRecipe = {
+    name: "sg-register",
+    description: "Singapore typed registers over the Overture-SG rows: the HDB block line with a #NN-NN unit, the S(NNNNNN) postcode, the building-led line",
+    mode: "tuples",
+    options: [
+        { flag: "--count", description: "Rows to emit (default: every input row once)" },
+        { flag: "--input", description: "The Overture-SG corpus JSONL ({ street, number, unit, postcode, locality })" },
+    ],
+    async run(opts, write) {
+        if (!opts.input) {
+            throw new Error("sg-register: --input <overture-sg.corpus.jsonl> is required");
+        }
+        const random = makeMulberry32(opts.seed);
+        const source = opts.sourceName ?? SOURCE;
+        const registers = {};
+        let read = 0;
+        let emitted = 0;
+        let skipped = 0;
+        let quarantined = 0;
+        for await (const tuple of readTuples(opts.input)) {
+            read++;
+            if (opts.count !== undefined && emitted >= opts.count)
+                break;
+            const street = typeof tuple["street"] === "string" ? tuple["street"].trim() : "";
+            const number = typeof tuple["number"] === "string" ? tuple["number"].trim() : "";
+            const postcode = typeof tuple["postcode"] === "string" ? tuple["postcode"].trim() : "";
+            const unit = typeof tuple["unit"] === "string" ? tuple["unit"].trim() : "";
+            if (!street || !/^\d+[A-Z]?$/u.test(number) || !/^\d{6}$/u.test(postcode)) {
+                skipped++;
+                continue;
+            }
+            const rendering = renderSGRegister({ street, number, unit, postcode }, random);
+            if (opts.golden) {
+                const golden = { raw: rendering.raw, components: rendering.components, country: "SG", locale: "en-SG" };
+                write(JSON.stringify(golden) + "\n");
+                emitted++;
+                registers[rendering.register] = (registers[rendering.register] ?? 0) + 1;
+                continue;
+            }
+            const ok = alignAndWrite(write, {
+                raw: rendering.raw,
+                components: rendering.components,
+                country: "SG",
+                locale: "en-SG",
+                source,
+                source_id: sliceSourceID(source, { raw: rendering.raw }),
+                corpus_version: "0.1.0",
+                license: "CDLA-Permissive-2.0 — Overture Maps addresses over the Singapore Open Data Licence 1.0 (OneMap / Singapore Land Authority)",
+            }, `sg-register:${rendering.register}`);
+            if (ok) {
+                emitted++;
+                registers[rendering.register] = (registers[rendering.register] ?? 0) + 1;
+            }
+            else {
+                quarantined++;
+            }
+        }
+        console.error(`  sg-register: ${emitted} emitted (${Object.entries(registers)
+            .map(([k, v]) => `${k} ${v}`)
+            .join(", ")}), ${skipped} skipped, ${quarantined} quarantined of ${read} read`);
+        return { read, emitted, skipped };
+    },
+};
+//# sourceMappingURL=register.js.map

package/out/sg/recipes/register.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"register.js","sourceRoot":"","sources":["../../../lib/sg/recipes/register.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AAEH,OAAO,EAAE,UAAU,IAAI,cAAc,EAAE,MAAM,uBAAuB,CAAA;AAEpE,OAAO,EAAE,SAAS,EAAE,MAAM,4BAA4B,CAAA;AACtD,OAAO,EAAE,aAAa,EAAqB,UAAU,EAAE,aAAa,EAAE,MAAM,mBAAmB,CAAA;AAE/F,MAAM,MAAM,GAAG,mBAAmB,CAAA;AAElC;;GAEG;AACH,MAAM,qBAAqB,GAA0D;IACpF,CAAC,QAAQ,EAAE,KAAK,CAAC;IACjB,CAAC,MAAM,EAAE,IAAI,CAAC;IACd,CAAC,QAAQ,EAAE,IAAI,CAAC;IAChB,CAAC,OAAO,EAAE,IAAI,CAAC;IACf,CAAC,UAAU,EAAE,MAAM,CAAC;IACpB,CAAC,MAAM,EAAE,IAAI,CAAC;IACd,CAAC,WAAW,EAAE,MAAM,CAAC;IACrB,CAAC,SAAS,EAAE,KAAK,CAAC;IAClB,CAAC,OAAO,EAAE,IAAI,CAAC;IACf,CAAC,OAAO,EAAE,IAAI,CAAC;IACf,CAAC,MAAM,EAAE,IAAI,CAAC;CACd,CAAA;AAED;;;;GAIG;AACH,MAAM,YAAY,GAAG,CAAC,QAAQ,EAAE,mBAAmB,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,QAAQ,EAAE,SAAS,EAAE,OAAO,EAAE,OAAO,CAAC,CAAA;AAEvH;;;GAGG;AACH,MAAM,OAAO,GAAG,GAAG,CAAA;AACnB,MAAM,kBAAkB,GAAG,IAAI,CAAA;AAC/B,MAAM,cAAc,GAAG,IAAI,CAAA;AAC3B,MAAM,YAAY,GAAG,IAAI,CAAA;AACzB,MAAM,OAAO,GAAG,GAAG,CAAA;AAEnB,MAAM,SAAS,GAAG,EAAE,CAAA;AACpB,MAAM,QAAQ,GAAG,GAAG,CAAA;AAEpB;;GAEG;AACH,MAAM,aAAa,GAA+B;IACjD,KAAK,EAAE,CAAC;IACR,gBAAgB,EAAE,OAAO;IACzB,YAAY,EAAE,OAAO,GAAG,kBAAkB;IAC1C,QAAQ,EAAE,CAAC;CACX,CAAA;AAED;;;GAGG;AACH,MAAM,UAAU,eAAe,CAAC,IAAY;IAC3C,OAAO,IAAI,KAAK,IAAI,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;AAC5D,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,kBAAkB,CAAC,MAAc;IAChD,KAAK,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,qBAAqB,EAAE,CAAC;QACnD,MAAM,OAAO,GAAG,IAAI,MAAM,CAAC,MAAM,IAAI,KAAK,EAAE,GAAG,CAAC,CAAA;QAEhD,IAAI,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC;YAAE,OAAO,MAAM,CAAC,OAAO,CAAC,OAAO,EAAE,KAAK,CAAC,CAAA;IAChE,CAAC;IAED,OAAO,MAAM,CAAA;AACd,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,cAAc,CAAC,IAAY;IAC1C,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAA;IAEvC,IAAI,CAAC,KAAK,IAAI,KAAK,KAAK,KAAK,IAAI,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,OAAO,KAAK,CAAA;IAE1F,OAAO,CAAC,YAAY,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAA;AAC1D,CAAC;AAuBD;;;;GAIG;AACH,MAAM,UAAU,gBAAgB,CAAC,GAAU,EAAE,MAAoB,EAAE,QAAqB;IACvF,MAAM,OAAO,GAAG,eAAe,CAAC,GAAG,CAAC,MAAM,CAAC,CAAA;IAC3C,MAAM,MAAM,GAAG,MAAM,EAAE,GAAG,YAAY,CAAC,CAAC,CAAC,kBAAkB,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,OAAO,CAAA;IAC9E,MAAM,KAAK,GAAG,MAAM,EAAE,CAAA;IACtB,MAAM,QAAQ,GAAG,cAAc,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,eAAe,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;IAC5E,MAAM,IAAI,GAAG,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,aAAa,CAAC,QAAQ,CAAC,CAAA;IAErE,IAAI,GAAW,CAAA;IACf,IAAI,QAAoB,CAAA;IACxB,MAAM,UAAU,GAA2B,EAAE,YAAY,EAAE,GAAG,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,CAAC,QAAQ,EAAE,CAAA;IAEvG,IAAI,IAAI,GAAG,OAAO,IAAI,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC;QACvD,MAAM,KAAK,GAAG,MAAM,CAAC,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,CAAA;QAC3E,MAAM,MAAM,GAAG,MAAM,CAAC,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,QAAQ,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,CAAA;QAC3E,MAAM,IAAI,GAAG,IAAI,KAAK,IAAI,MAAM,EAAE,CAAA;QAElC,GAAG,GAAG,OAAO,GAAG,CAAC,MAAM,IAAI,MAAM,IAAI,IAAI,cAAc,GAAG,CAAC,QAAQ,EAAE,CAAA;QACrE,UAAU,CAAC,IAAI,GAAG,IAAI,CAAA;QACtB,UAAU,CAAC,QAAQ,GAAG,WAAW,CAAA;QACjC,QAAQ,GAAG,OAAO,CAAA;IACnB,CAAC;SAAM,IAAI,IAAI,GAAG,OAAO,GAAG,kBAAkB,EAAE,CAAC;QAChD,GAAG,GAAG,GAAG,GAAG,CAAC,MAAM,IAAI,MAAM,MAAM,GAAG,CAAC,QAAQ,GAAG,CAAA;QAClD,QAAQ,GAAG,kBAAkB,CAAA;IAC9B,CAAC;SAAM,IAAI,IAAI,GAAG,OAAO,GAAG,kBAAkB,GAAG,cAAc,IAAI,QAAQ,EAAE,CAAC;QAC7E,GAAG,GAAG,GAAG,QAAQ,KAAK,GAAG,CAAC,MAAM,IAAI,MAAM,eAAe,GAAG,CAAC,QAAQ,EAAE,CAAA;QACvE,UAAU,CAAC,KAAK,GAAG,QAAQ,CAAA;QAC3B,UAAU,CAAC,QAAQ,GAAG,WAAW,CAAA;QACjC,QAAQ,GAAG,cAAc,CAAA;IAC1B,CAAC;SAAM,CAAC;QACP,GAAG,GAAG,GAAG,GAAG,CAAC,MAAM,IAAI,MAAM,cAAc,GAAG,CAAC,QAAQ,EAAE,CAAA;QACzD,UAAU,CAAC,QAAQ,GAAG,WAAW,CAAA;QACjC,QAAQ,GAAG,UAAU,CAAA;IACtB,CAAC;IAED,IAAI,MAAM,EAAE,GAAG,OAAO,EAAE,CAAC;QACxB,GAAG,GAAG,GAAG,CAAC,WAAW,EAAE,CAAA;QAEvB,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,EAAE,CAAC;YAC3C,UAAU,CAAC,GAAG,CAAC,GAAG,UAAU,CAAC,GAAG,CAAE,CAAC,WAAW,EAAE,CAAA;QACjD,CAAC;IACF,CAAC;IAED,OAAO,EAAE,GAAG,EAAE,UAAU,EAAE,QAAQ,EAAE,QAAQ,EAAE,CAAA;AAC/C,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAiB;IAC7C,IAAI,EAAE,aAAa;IACnB,WAAW,EACV,2IAA2I;IAC5I,IAAI,EAAE,QAAQ;IACd,OAAO,EAAE;QACR,EAAE,IAAI,EAAE,SAAS,EAAE,WAAW,EAAE,8CAA8C,EAAE;QAChF,EAAE,IAAI,EAAE,SAAS,EAAE,WAAW,EAAE,6EAA6E,EAAE;KAC/G;IACD,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC;YACjB,MAAM,IAAI,KAAK,CAAC,6DAA6D,CAAC,CAAA;QAC/E,CAAC;QAED,MAAM,MAAM,GAAG,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACxC,MAAM,MAAM,GAAG,IAAI,CAAC,UAAU,IAAI,MAAM,CAAA;QACxC,MAAM,SAAS,GAA2B,EAAE,CAAA;QAC5C,IAAI,IAAI,GAAG,CAAC,CAAA;QACZ,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,WAAW,GAAG,CAAC,CAAA;QAEnB,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,UAAU,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;YAClD,IAAI,EAAE,CAAA;YAEN,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;gBAAE,MAAK;YAE5D,MAAM,MAAM,GAAG,OAAO,KAAK,CAAC,QAAQ,CAAC,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAA;YAChF,MAAM,MAAM,GAAG,OAAO,KAAK,CAAC,QAAQ,CAAC,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAA;YAChF,MAAM,QAAQ,GAAG,OAAO,KAAK,CAAC,UAAU,CAAC,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAA;YACtF,MAAM,IAAI,GAAG,OAAO,KAAK,CAAC,MAAM,CAAC,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAA;YAE1E,IAAI,CAAC,MAAM,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,QAAQ,CAAC,EAAE,CAAC;gBAC3E,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,MAAM,SAAS,GAAG,gBAAgB,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,EAAE,MAAM,CAAC,CAAA;YAE9E,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;gBACjB,MAAM,MAAM,GAAG,EAAE,GAAG,EAAE,SAAS,CAAC,GAAG,EAAE,UAAU,EAAE,SAAS,CAAC,UAAU,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,CAAA;gBAEvG,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,GAAG,IAAI,CAAC,CAAA;gBAEpC,OAAO,EAAE,CAAA;gBACT,SAAS,CAAC,SAAS,CAAC,QAAQ,CAAC,GAAG,CAAC,SAAS,CAAC,SAAS,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;gBAExE,SAAQ;YACT,CAAC;YAED,MAAM,EAAE,GAAG,aAAa,CACvB,KAAK,EACL;gBACC,GAAG,EAAE,SAAS,CAAC,GAAG;gBAClB,UAAU,EAAE,SAAS,CAAC,UAAU;gBAChC,OAAO,EAAE,IAAI;gBACb,MAAM,EAAE,OAAO;gBACf,MAAM;gBACN,SAAS,EAAE,aAAa,CAAC,MAAM,EAAE,EAAE,GAAG,EAAE,SAAS,CAAC,GAAG,EAAE,CAAC;gBACxD,cAAc,EAAE,OAAO;gBACvB,OAAO,EACN,4HAA4H;aAC7H,EACD,eAAe,SAAS,CAAC,QAAQ,EAAE,CACnC,CAAA;YAED,IAAI,EAAE,EAAE,CAAC;gBACR,OAAO,EAAE,CAAA;gBACT,SAAS,CAAC,SAAS,CAAC,QAAQ,CAAC,GAAG,CAAC,SAAS,CAAC,SAAS,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;YACzE,CAAC;iBAAM,CAAC;gBACP,WAAW,EAAE,CAAA;YACd,CAAC;QACF,CAAC;QAED,OAAO,CAAC,KAAK,CACZ,kBAAkB,OAAO,aAAa,MAAM,CAAC,OAAO,CAAC,SAAS,CAAC;aAC7D,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;aAC5B,IAAI,CAAC,IAAI,CAAC,MAAM,OAAO,aAAa,WAAW,mBAAmB,IAAI,OAAO,CAC/E,CAAA;QAED,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;IAClC,CAAC;CACD,CAAA"}

package/out/sg/tools/fetch/acra.d.ts

@@ -0,0 +1,23 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Fetch Singapore's ACRA corporate-entity register from data.gov.sg: 27 monthly CSVs (one per first
+ *   letter of the entity name plus "others"), 53 columns, the registered address FIELDED — `Block`,
+ *   `Street Name`, `Level No`, `Unit No`, `Building Name`, `Postal Code` — which is the NPPES shape,
+ *   so the corpus adapter renders the string and the spans fall out by construction. A Singapore
+ *   postcode names one building, so the postcode column is also the join to the OneMap/Overture rows.
+ *
+ *   License: Singapore Open Data Licence version 1.0 (attribution; commercial use, modification and
+ *   adaptation permitted; no share-alike). The prescribed attribution sentence goes in the manifest.
+ *
+ *   data.gov.sg serves a dataset in two calls: `initiate-download` prepares a signed URL and
+ *   `poll-download` answers it once the export is ready.
+ *
+ *   Invoke via `mailwoman corpus fetch acra-sg --out-root <path>`.
+ */
+import type { BaseFetchOptions, FetchSummary } from "#tools/fetch/download/index";
+export type FetchACRASGOptions = BaseFetchOptions;
+export declare function fetchACRASG(options: FetchACRASGOptions, report?: (line: string) => void): Promise<FetchSummary>;
+//# sourceMappingURL=acra.d.ts.map

package/out/sg/tools/fetch/acra.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"acra.d.ts","sourceRoot":"","sources":["../../../../lib/sg/tools/fetch/acra.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;GAkBG;AASH,OAAO,KAAK,EACX,gBAAgB,EAChB,YAAY,EAGZ,MAAM,6BAA6B,CAAA;AAepC,MAAM,MAAM,kBAAkB,GAAG,gBAAgB,CAAA;AA4CjD,wBAAsB,WAAW,CAAC,OAAO,EAAE,kBAAkB,EAAE,MAAM,CAAC,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,GAAG,OAAO,CAAC,YAAY,CAAC,CAuFrH"}

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

@@ -0,0 +1,126 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Fetch Singapore's ACRA corporate-entity register from data.gov.sg: 27 monthly CSVs (one per first
+ *   letter of the entity name plus "others"), 53 columns, the registered address FIELDED — `Block`,
+ *   `Street Name`, `Level No`, `Unit No`, `Building Name`, `Postal Code` — which is the NPPES shape,
+ *   so the corpus adapter renders the string and the spans fall out by construction. A Singapore
+ *   postcode names one building, so the postcode column is also the join to the OneMap/Overture rows.
+ *
+ *   License: Singapore Open Data Licence version 1.0 (attribution; commercial use, modification and
+ *   adaptation permitted; no share-alike). The prescribed attribution sentence goes in the manifest.
+ *
+ *   data.gov.sg serves a dataset in two calls: `initiate-download` prepares a signed URL and
+ *   `poll-download` answers it once the export is ready.
+ *
+ *   Invoke via `mailwoman corpus fetch acra-sg --out-root <path>`.
+ */
+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 { join } from "path-ts";
+import { loadCollectionFiles, streamBodyToFile, withRetries, writeManifest } from "#tools/fetch/download/index";
+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";
+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 ${res.status}`);
+    return (await res.json());
+}
+/**
+ * Ask the portal to prepare the CSV export, then poll until it names the signed URL.
+ */
+async function signedURLFor(datasetID) {
+    await readJSON(`${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`;
+}
+export async function fetchACRASG(options, report) {
+    const destDir = join(options.outRoot, SLUG);
+    await makeDirectories(destDir);
+    const manifestPath = join(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 = join(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;;;;;;;;;;;;;;;;;;GAkBG;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;AACnD,OAAO,EAAE,IAAI,EAAE,MAAM,SAAS,CAAA;AAQ9B,OAAO,EAAE,mBAAmB,EAAE,gBAAgB,EAAE,WAAW,EAAE,aAAa,EAAE,MAAM,6BAA6B,CAAA;AAE/G,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,MAAM,WAAW,GAChB,mOAAmO,CAAA;AAEpO,MAAM,gBAAgB,GAAG,IAAI,CAAA;AAC7B,MAAM,aAAa,GAAG,EAAE,CAAA;AAgBxB,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;;GAEG;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,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;IAC9B,MAAM,YAAY,GAAG,IAAI,CAAC,OAAO,EAAE,eAAe,CAAC,CAAA;IAEnD,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,IAAI,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAA;QACpC,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 +1,32 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `si-bare-village` — the Slovenian no-street counter-slice (#901 run-2). Slovenia's rural
+ *   addressing has NO street line: the village name is the street-level token and repeats as the
+ *   locality ("Zabiče 8, 6250 Zabiče"). The v1.9.8 check FALSIFIED the fr-bare-street slice alone on
+ *   exactly this class (SI resolve −3.4pp; "Apače 108" split into street "Apače 10" + house "8") —
+ *   the bare-street boundary lesson generalizes onto a form where the leading name must keep its
+ *   number whole and the trailing mention must stay locality-bound. This slice is the paired
+ *   counter-distribution: same lesson ("name before number, comma, then admin"), 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, the fr-bare lesson)
+ *   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";
+/**
+ * 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.
+ */
+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;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AAIH,OAAO,EAA6B,KAAK,YAAY,EAAiB,MAAM,mBAAmB,CAAA;AAE/F;;;GAGG;AACH,eAAO,MAAM,mBAAmB,EAAE,YAgEjC,CAAA"}

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

@@ -0,0 +1,89 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `si-bare-village` — the Slovenian no-street counter-slice (#901 run-2). Slovenia's rural
+ *   addressing has NO street line: the village name is the street-level token and repeats as the
+ *   locality ("Zabiče 8, 6250 Zabiče"). The v1.9.8 check FALSIFIED the fr-bare-street slice alone on
+ *   exactly this class (SI resolve −3.4pp; "Apače 108" split into street "Apače 10" + house "8") —
+ *   the bare-street boundary lesson generalizes onto a form where the leading name must keep its
+ *   number whole and the trailing mention must stay locality-bound. This slice is the paired
+ *   counter-distribution: same lesson ("name before number, comma, then admin"), 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, the fr-bare lesson)
+ *   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, sliceSourceID } 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.
+ */
+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;
+        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}`;
+            }
+            const source_id = sliceSourceID("synth-si-bare-village", { ...components, o: String(order), v: String(read) });
+            const canonical = {
+                raw,
+                components,
+                country: "SI",
+                locale: "sl-SI",
+                source: "synth-si-bare-village",
+                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)",
+            };
+            if (alignAndWrite(write, canonical, "si-bare-village")) {
+                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;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AAEH,OAAO,EAAE,UAAU,IAAI,cAAc,EAAE,MAAM,uBAAuB,CAAA;AAEpE,OAAO,EAAE,aAAa,EAAE,UAAU,EAAqB,aAAa,EAAE,MAAM,mBAAmB,CAAA;AAE/F;;;GAGG;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,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,MAAM,SAAS,GAAG,aAAa,CAAC,uBAAuB,EAAE,EAAE,GAAG,UAAU,EAAE,CAAC,EAAE,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,EAAE,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;YAE9G,MAAM,SAAS,GAAG;gBACjB,GAAG;gBACH,UAAU;gBACV,OAAO,EAAE,IAAI;gBACb,MAAM,EAAE,OAAO;gBACf,MAAM,EAAE,uBAAuB;gBAC/B,SAAS;gBACT,cAAc,EAAE,OAAO;gBACvB,OAAO,EACN,2HAA2H;aAC5H,CAAA;YAED,IAAI,aAAa,CAAC,KAAK,EAAE,SAAS,EAAE,iBAAiB,CAAC,EAAE,CAAC;gBACxD,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/south-asia/recipes/register.d.ts

@@ -0,0 +1,51 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `pk-register` and `bd-register` slice recipes — the typed forms of Pakistan and Bangladesh the country templates do
+ *   not render, over the OpenStreetMap corpus JSONL the `osm` adapter reads (#1744, the F8 family). The adapter's own
+ *   rows carry the template shape (`4 38th Street, DHA Phase 6, Karachi`; `24 Road 104, Dhaka - 1207`); these two carry
+ *   the register:
+ *
+ *   - the HOUSE line, `House 4, Street 25, F-7/2, Islamabad` and `House 34, Road 4, Sector 9, Uttara, Dhaka 1230` —
+ *     `House` in front of the number (untagged, the same rule as Singapore's `Blk`), the numbered street, the scheme or
+ *     sector as the dependent locality, the city;
+ *   - the PLAIN line with the trailing postcode and no dash, `58 Kalabagan 1st Ln, Dhaka 1205` — what a person types,
+ *     where the template writes `Dhaka - 1205`.
+ *
+ *   Islamabad's sector codes (`F-7/2`, `G-10/3`) are the one synthesized value: the register carries them on a handful
+ *   of rows, mostly inside `addr:city` (`F7/2 Islamabad`), so a row whose city is Islamabad and whose scheme is unknown
+ *   draws one from the capital's grid. Every other value is the row's own. The components are the `osm` adapter's
+ *   mapping (`componentsForOSMRow`), so the two surfaces cannot disagree about what a field means.
+ *
+ *   ⚠ ODbL: the rows inherit OpenStreetMap's share-alike license, and `--exclude-share-alike` drops them.
+ *
+ *   Run: mailwoman corpus slice pk-register --input <osm-pk.corpus.jsonl> --count N --seed S
+ */
+import { type CorpusRecipe } from "#recipes/scaffold";
+import type { CanonicalRow } from "#types";
+export type SouthAsiaRegister = "house" | "plain";
+export interface SouthAsiaRendering {
+    raw: string;
+    components: Record<string, string>;
+    register: SouthAsiaRegister;
+}
+/**
+ * Draw one Islamabad sector code, `F-7/2`.
+ */
+export declare function drawIslamabadSector(random: () => number): string;
+/**
+ * Render one register line from the adapter's components. Returns null when the row has no house number or no locality:
+ * both registers need the two.
+ */
+export declare function renderSouthAsiaRegister(components: CanonicalRow["components"], random: () => number, register?: SouthAsiaRegister): SouthAsiaRendering | null;
+/**
+ * Pakistan: `House 4, Street 25, F-7/2, Islamabad` and `B230 Street 5, Block 3, Karachi`.
+ */
+export declare const pkRegisterRecipe: CorpusRecipe;
+/**
+ * Bangladesh: `House 34, Road 4, Sector 9, Uttara, Dhaka 1230` and `58 Kalabagan 1st Ln, Dhaka 1205`.
+ */
+export declare const bdRegisterRecipe: CorpusRecipe;
+//# sourceMappingURL=register.d.ts.map