@mailwoman/corpus 4.16.2 → 5.1.0

package/out/src/shard-recipes/locale.js

@@ -0,0 +1,257 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `locale` shard recipe — the multi-locale generalization of the `german` recipe. Reads REAL
+ *   OpenAddresses tuples for a `--country` (DE/FR/NL/IT/ES), renders each via
+ *   {@link synthesizeLocaleRow} in BOTH orders (`--intl-fraction`, default 0.4 international / the
+ *   rest country-native), aligns to BIO, and emits a labeled JSONL. Generate-mode: it STREAMS each
+ *   source CSV (`unzip -p | readline`) and reservoir-samples to {@link RESERVOIR_CAP} (so FR/ES
+ *   countrywide work in bounded memory), then draws `--count` rows from the pool with the passed
+ *   `random`. Ported from scripts/build-locale-shard.mjs.
+ *
+ *   The reservoir uses its OWN seeded PRNG ({@link mulberry32}, per part), independent of the emit
+ *   `random`, so the input sample is reproducible WITHOUT perturbing the synth/order draws.
+ */
+import { spawn } from "node:child_process";
+import { createInterface } from "node:readline";
+import { stableSourceID } from "../adapter.js";
+import { alignRow } from "../align.js";
+import { synthesizeLocaleRow } from "../synthesize-german.js";
+import { makeMulberry32 } from "./scaffold.js";
+/**
+ * Per-country OA sources (cached zips) + the source name used in the corpus. A part may carry a `region` fallback (the
+ * admin region the file covers) for countries whose OA REGION column is empty — DE's is, so the international-order
+ * tail needs it set per-state (#327). FR/NL/IT leave it unset (their REGION column is populated, used per-row). ES
+ * carries a `conform` map (the raw CNIG schema).
+ */
+const COUNTRY_SOURCES = {
+    DE: {
+        source: "synth-german",
+        parts: [
+            { zip: "/tmp/oa-cache/de__berlin.zip", csv: "de/berlin.csv", region: "Berlin" },
+            { zip: "/tmp/oa-cache/de__sn__statewide.zip", csv: "de/sn/statewide.csv", region: "Sachsen" },
+        ],
+    },
+    FR: { source: "synth-fr", parts: [{ zip: "/tmp/oa-cache/fr__countrywide.zip", csv: "fr/countrywide.csv" }] },
+    NL: { source: "synth-nl", parts: [{ zip: "/tmp/oa-cache/nl__countrywide.zip", csv: "nl/countrywide.csv" }] },
+    // IT: 5-digit postcode; OA REGION populated (the regione) → international rows carry the tail.
+    IT: { source: "synth-it", parts: [{ zip: "/tmp/oa-cache/it__countrywide.zip", csv: "it/countrywide.csv" }] },
+    // ES: the raw CNIG/IGN national set (NOT OA-conformed), via a `conform` map: street = join(tipo_vial,
+    // nombre_via); region = comunidad_autonoma (POPULATED → international "City, Comunidad Postcode" tail).
+    ES: {
+        source: "synth-es",
+        parts: [
+            {
+                zip: "/tmp/oa-cache/es__countrywide.zip",
+                csv: "es_addresses.csv",
+                conform: {
+                    number: "numero",
+                    street: ["tipo_vial", "nombre_via"],
+                    city: "municipio",
+                    region: "comunidad_autonoma",
+                    postcode: "cod_postal",
+                },
+            },
+        ],
+    },
+};
+/**
+ * Per-part reservoir cap. Streaming + Algorithm-R reservoir sampling to this size keeps memory bounded regardless of
+ * source size, where buffering the whole CSV OOMs / overflows on FR/ES countrywide (~2.5 GB, ~25M rows). DE/NL-scale
+ * sources (≤ ~1.2M) fit entirely, so they're sampled losslessly.
+ */
+const RESERVOIR_CAP = 1_200_000;
+/** Minimal RFC-4180-ish splitter (handles quoted fields). */
+function splitCSV(line) {
+    const out = [];
+    let cur = "";
+    let inQ = false;
+    for (let i = 0; i < line.length; i++) {
+        const c = line[i];
+        if (inQ) {
+            if (c === '"') {
+                if (line[i + 1] === '"') {
+                    cur += '"';
+                    i++;
+                }
+                else
+                    inQ = false;
+            }
+            else
+                cur += c;
+        }
+        else if (c === '"')
+            inQ = true;
+        else if (c === ",") {
+            out.push(cur);
+            cur = "";
+        }
+        else
+            cur += c;
+    }
+    out.push(cur);
+    return out;
+}
+/**
+ * Stream real tuples out of a cached OA zip and reservoir-sample to {@link RESERVOIR_CAP}. Reads the CSV line-by-line
+ * via `unzip -p | readline` (bounded memory) and keeps a uniform random sample (Algorithm R) seeded by `rng` — separate
+ * from the emit loop's PRNG. NO global dedup (a 25M-key Set would OOM; OA rows are near-unique). The region falls back
+ * to `part.region` when the row's REGION cell is empty (DE).
+ */
+function readTuples(part, rng) {
+    return new Promise((resolve) => {
+        const child = spawn("unzip", ["-p", part.zip, part.csv]);
+        const rl = createInterface({ input: child.stdout, crlfDelay: Infinity });
+        const get = (cells, i) => (i >= 0 && i < cells.length ? (cells[i] ?? "").trim() : "");
+        const reservoir = [];
+        let cols = null;
+        let header = null;
+        let seen = 0;
+        rl.on("line", (line) => {
+            if (!line)
+                return;
+            if (header === null) {
+                header = splitCSV(line).map((h) => h.trim().toLowerCase());
+                const ix = (name) => header.indexOf(String(name).toLowerCase());
+                // OA-standard columns (IT/FR/NL/DE), unless the part carries a `conform` map for a raw
+                // upstream schema (ES — CNIG columns, street split across `tipo_vial` + `nombre_via`).
+                const c = part.conform;
+                cols = c
+                    ? {
+                        num: ix(c.number),
+                        streetParts: (Array.isArray(c.street) ? c.street : [c.street]).map(ix),
+                        city: ix(c.city),
+                        region: c.region ? ix(c.region) : -1,
+                        post: ix(c.postcode),
+                    }
+                    : {
+                        num: ix("number"),
+                        streetParts: [ix("street")],
+                        city: ix("city"),
+                        region: ix("region"),
+                        post: ix("postcode"),
+                    };
+                return;
+            }
+            if (cols === null)
+                return;
+            const cells = splitCSV(line);
+            const street = cols.streetParts
+                .map((i) => get(cells, i))
+                .filter(Boolean)
+                .join(" ")
+                .trim();
+            const locality = get(cells, cols.city);
+            if (!street || !locality)
+                return;
+            const tuple = {
+                house_number: get(cells, cols.num),
+                street,
+                locality,
+                region: get(cells, cols.region) || part.region || "",
+                postcode: get(cells, cols.post),
+            };
+            seen++;
+            if (reservoir.length < RESERVOIR_CAP) {
+                reservoir.push(tuple);
+            }
+            else {
+                const j = Math.floor(rng() * seen);
+                // 0 .. seen-1
+                if (j < RESERVOIR_CAP)
+                    reservoir[j] = tuple;
+            }
+        });
+        rl.on("close", () => {
+            console.error(`  ${part.csv}: ${reservoir.length} sampled of ${seen} rows`);
+            resolve(reservoir);
+        });
+        child.on("error", (err) => {
+            console.error(`  WARN: unzip failed for ${part.zip}: ${err.message}`);
+            resolve([]);
+        });
+    });
+}
+export const localeRecipe = {
+    name: "locale",
+    description: "Per-locale coverage rows (DE/FR/NL/IT/ES) from real OA tuples, both orders → synthesizeLocaleRow",
+    mode: "generate",
+    options: [
+        { flag: "--country <cc>", description: "Target country (DE|FR|NL|IT|ES). Default DE" },
+        { flag: "--intl-fraction <f>", description: "Fraction rendered international order. Default 0.4" },
+    ],
+    async run(opts, write) {
+        // Emit PRNG: the legacy build-locale-shard.mjs seeded mulberry32(opts.seed). The reservoir uses a
+        // SEPARATE per-part mulberry32 (below) so input sampling never perturbs this emit stream.
+        const random = makeMulberry32(opts.seed);
+        const country = (opts.country ?? "DE").toUpperCase();
+        const countrySource = COUNTRY_SOURCES[country];
+        if (!countrySource) {
+            throw new Error(`No OA sources registered for --country ${country}. Known: ${Object.keys(COUNTRY_SOURCES).join(", ")}.`);
+        }
+        const intlFraction = opts.intlFraction ?? 0.4;
+        if (!(intlFraction >= 0 && intlFraction <= 1)) {
+            throw new Error(`--intl-fraction must be in [0, 1], got ${intlFraction}`);
+        }
+        const source = opts.sourceName ?? countrySource.source;
+        const count = opts.count ?? 4000;
+        const { parts } = countrySource;
+        const pool = [];
+        for (let pi = 0; pi < parts.length; pi++) {
+            // A reservoir PRNG per part, seeded but independent of the emit loop's `random`, so the sample is
+            // reproducible without perturbing the synth/order draws.
+            const reservoirRng = makeMulberry32((opts.seed ^ (0x9e3779b9 * (pi + 1))) >>> 0);
+            const t = await readTuples(parts[pi], reservoirRng);
+            for (const x of t)
+                pool.push(x); // NOT pool.push(...t) — spreading huge arrays overflows the stack
+        }
+        if (pool.length === 0) {
+            throw new Error(`No ${country} tuples found — are the cached zips present in /tmp/oa-cache?`);
+        }
+        let emitted = 0;
+        let skipped = 0;
+        let guard = 0;
+        const N = pool.length;
+        while (emitted < count && guard++ < count * 6) {
+            const base = pool[Math.floor(random() * N)];
+            const order = random() < intlFraction ? "international" : "native";
+            const synth = synthesizeLocaleRow(base, country, { random, order });
+            if (!synth) {
+                skipped++;
+                continue;
+            }
+            if (opts.golden) {
+                write(JSON.stringify({ raw: synth.raw, components: synth.components, country, order }) + "\n");
+                emitted++;
+                continue;
+            }
+            const sourceID = stableSourceID(source, {
+                street: synth.components.street,
+                house_number: synth.components.house_number,
+                locality: synth.components.locality,
+                postcode: synth.components.postcode,
+            });
+            const canonical = {
+                raw: synth.raw,
+                components: synth.components,
+                country,
+                locale: synth.locale,
+                source,
+                source_id: sourceID,
+                corpus_version: "0.4.0",
+                license: `OpenAddresses ${country} tuples, rendered ${order}-order — see ingest SOURCES`,
+            };
+            const aligned = alignRow(canonical);
+            if (aligned.kind !== "labeled" || !aligned.row) {
+                skipped++;
+                continue;
+            }
+            write(JSON.stringify({ ...aligned.row, synth_method: source, synth_order: order, synth_base_id: null }) + "\n");
+            emitted++;
+        }
+        return { emitted, skipped };
+    },
+};
+//# sourceMappingURL=locale.js.map

package/out/src/shard-recipes/locale.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"locale.js","sourceRoot":"","sources":["../../../src/shard-recipes/locale.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAEH,OAAO,EAAE,KAAK,EAAE,MAAM,oBAAoB,CAAA;AAC1C,OAAO,EAAE,eAAe,EAAE,MAAM,eAAe,CAAA;AAE/C,OAAO,EAAE,cAAc,EAAE,MAAM,eAAe,CAAA;AAC9C,OAAO,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAA;AACtC,OAAO,EAAwB,mBAAmB,EAAE,MAAM,yBAAyB,CAAA;AACnF,OAAO,EAAE,cAAc,EAAoB,MAAM,eAAe,CAAA;AA2BhE;;;;;GAKG;AACH,MAAM,eAAe,GAAwC;IAC5D,EAAE,EAAE;QACH,MAAM,EAAE,cAAc;QACtB,KAAK,EAAE;YACN,EAAE,GAAG,EAAE,8BAA8B,EAAE,GAAG,EAAE,eAAe,EAAE,MAAM,EAAE,QAAQ,EAAE;YAC/E,EAAE,GAAG,EAAE,qCAAqC,EAAE,GAAG,EAAE,qBAAqB,EAAE,MAAM,EAAE,SAAS,EAAE;SAC7F;KACD;IACD,EAAE,EAAE,EAAE,MAAM,EAAE,UAAU,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,mCAAmC,EAAE,GAAG,EAAE,oBAAoB,EAAE,CAAC,EAAE;IAC5G,EAAE,EAAE,EAAE,MAAM,EAAE,UAAU,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,mCAAmC,EAAE,GAAG,EAAE,oBAAoB,EAAE,CAAC,EAAE;IAC5G,+FAA+F;IAC/F,EAAE,EAAE,EAAE,MAAM,EAAE,UAAU,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,mCAAmC,EAAE,GAAG,EAAE,oBAAoB,EAAE,CAAC,EAAE;IAC5G,sGAAsG;IACtG,wGAAwG;IACxG,EAAE,EAAE;QACH,MAAM,EAAE,UAAU;QAClB,KAAK,EAAE;YACN;gBACC,GAAG,EAAE,mCAAmC;gBACxC,GAAG,EAAE,kBAAkB;gBACvB,OAAO,EAAE;oBACR,MAAM,EAAE,QAAQ;oBAChB,MAAM,EAAE,CAAC,WAAW,EAAE,YAAY,CAAC;oBACnC,IAAI,EAAE,WAAW;oBACjB,MAAM,EAAE,oBAAoB;oBAC5B,QAAQ,EAAE,YAAY;iBACtB;aACD;SACD;KACD;CACD,CAAA;AAED;;;;GAIG;AACH,MAAM,aAAa,GAAG,SAAS,CAAA;AAE/B,6DAA6D;AAC7D,SAAS,QAAQ,CAAC,IAAY;IAC7B,MAAM,GAAG,GAAa,EAAE,CAAA;IACxB,IAAI,GAAG,GAAG,EAAE,CAAA;IACZ,IAAI,GAAG,GAAG,KAAK,CAAA;IAEf,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACtC,MAAM,CAAC,GAAG,IAAI,CAAC,CAAC,CAAE,CAAA;QAElB,IAAI,GAAG,EAAE,CAAC;YACT,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC;gBACf,IAAI,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;oBACzB,GAAG,IAAI,GAAG,CAAA;oBACV,CAAC,EAAE,CAAA;gBACJ,CAAC;;oBAAM,GAAG,GAAG,KAAK,CAAA;YACnB,CAAC;;gBAAM,GAAG,IAAI,CAAC,CAAA;QAChB,CAAC;aAAM,IAAI,CAAC,KAAK,GAAG;YAAE,GAAG,GAAG,IAAI,CAAA;aAC3B,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC;YACpB,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;YACb,GAAG,GAAG,EAAE,CAAA;QACT,CAAC;;YAAM,GAAG,IAAI,CAAC,CAAA;IAChB,CAAC;IACD,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;IAEb,OAAO,GAAG,CAAA;AACX,CAAC;AAUD;;;;;GAKG;AACH,SAAS,UAAU,CAAC,IAAgB,EAAE,GAAiB;IACtD,OAAO,IAAI,OAAO,CAAC,CAAC,OAAO,EAAE,EAAE;QAC9B,MAAM,KAAK,GAAG,KAAK,CAAC,OAAO,EAAE,CAAC,IAAI,EAAE,IAAI,CAAC,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,CAAA;QACxD,MAAM,EAAE,GAAG,eAAe,CAAC,EAAE,KAAK,EAAE,KAAK,CAAC,MAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,CAAC,CAAA;QACzE,MAAM,GAAG,GAAG,CAAC,KAAe,EAAE,CAAS,EAAU,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAA;QAC/G,MAAM,SAAS,GAAsB,EAAE,CAAA;QACvC,IAAI,IAAI,GAAuB,IAAI,CAAA;QACnC,IAAI,MAAM,GAAoB,IAAI,CAAA;QAClC,IAAI,IAAI,GAAG,CAAC,CAAA;QACZ,EAAE,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE;YACtB,IAAI,CAAC,IAAI;gBAAE,OAAM;YAEjB,IAAI,MAAM,KAAK,IAAI,EAAE,CAAC;gBACrB,MAAM,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC,CAAA;gBAC1D,MAAM,EAAE,GAAG,CAAC,IAAY,EAAU,EAAE,CAAC,MAAO,CAAC,OAAO,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,WAAW,EAAE,CAAC,CAAA;gBAChF,uFAAuF;gBACvF,uFAAuF;gBACvF,MAAM,CAAC,GAAG,IAAI,CAAC,OAAO,CAAA;gBACtB,IAAI,GAAG,CAAC;oBACP,CAAC,CAAC;wBACA,GAAG,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC;wBACjB,WAAW,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC;wBACtE,IAAI,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC;wBAChB,MAAM,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;wBACpC,IAAI,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAC;qBACpB;oBACF,CAAC,CAAC;wBACA,GAAG,EAAE,EAAE,CAAC,QAAQ,CAAC;wBACjB,WAAW,EAAE,CAAC,EAAE,CAAC,QAAQ,CAAC,CAAC;wBAC3B,IAAI,EAAE,EAAE,CAAC,MAAM,CAAC;wBAChB,MAAM,EAAE,EAAE,CAAC,QAAQ,CAAC;wBACpB,IAAI,EAAE,EAAE,CAAC,UAAU,CAAC;qBACpB,CAAA;gBAEH,OAAM;YACP,CAAC;YAED,IAAI,IAAI,KAAK,IAAI;gBAAE,OAAM;YACzB,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAA;YAC5B,MAAM,MAAM,GAAG,IAAI,CAAC,WAAW;iBAC7B,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC;iBACzB,MAAM,CAAC,OAAO,CAAC;iBACf,IAAI,CAAC,GAAG,CAAC;iBACT,IAAI,EAAE,CAAA;YACR,MAAM,QAAQ,GAAG,GAAG,CAAC,KAAK,EAAE,IAAI,CAAC,IAAI,CAAC,CAAA;YAEtC,IAAI,CAAC,MAAM,IAAI,CAAC,QAAQ;gBAAE,OAAM;YAChC,MAAM,KAAK,GAAoB;gBAC9B,YAAY,EAAE,GAAG,CAAC,KAAK,EAAE,IAAI,CAAC,GAAG,CAAC;gBAClC,MAAM;gBACN,QAAQ;gBACR,MAAM,EAAE,GAAG,CAAC,KAAK,EAAE,IAAI,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,MAAM,IAAI,EAAE;gBACpD,QAAQ,EAAE,GAAG,CAAC,KAAK,EAAE,IAAI,CAAC,IAAI,CAAC;aAC/B,CAAA;YACD,IAAI,EAAE,CAAA;YAEN,IAAI,SAAS,CAAC,MAAM,GAAG,aAAa,EAAE,CAAC;gBACtC,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;YACtB,CAAC;iBAAM,CAAC;gBACP,MAAM,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,EAAE,GAAG,IAAI,CAAC,CAAA;gBAElC,cAAc;gBACd,IAAI,CAAC,GAAG,aAAa;oBAAE,SAAS,CAAC,CAAC,CAAC,GAAG,KAAK,CAAA;YAC5C,CAAC;QACF,CAAC,CAAC,CAAA;QACF,EAAE,CAAC,EAAE,CAAC,OAAO,EAAE,GAAG,EAAE;YACnB,OAAO,CAAC,KAAK,CAAC,KAAK,IAAI,CAAC,GAAG,KAAK,SAAS,CAAC,MAAM,eAAe,IAAI,OAAO,CAAC,CAAA;YAC3E,OAAO,CAAC,SAAS,CAAC,CAAA;QACnB,CAAC,CAAC,CAAA;QACF,KAAK,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE;YACzB,OAAO,CAAC,KAAK,CAAC,4BAA4B,IAAI,CAAC,GAAG,KAAK,GAAG,CAAC,OAAO,EAAE,CAAC,CAAA;YACrE,OAAO,CAAC,EAAE,CAAC,CAAA;QACZ,CAAC,CAAC,CAAA;IACH,CAAC,CAAC,CAAA;AACH,CAAC;AAED,MAAM,CAAC,MAAM,YAAY,GAAgB;IACxC,IAAI,EAAE,QAAQ;IACd,WAAW,EAAE,kGAAkG;IAC/G,IAAI,EAAE,UAAU;IAChB,OAAO,EAAE;QACR,EAAE,IAAI,EAAE,gBAAgB,EAAE,WAAW,EAAE,6CAA6C,EAAE;QACtF,EAAE,IAAI,EAAE,qBAAqB,EAAE,WAAW,EAAE,oDAAoD,EAAE;KAClG;IACD,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,kGAAkG;QAClG,0FAA0F;QAC1F,MAAM,MAAM,GAAG,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACxC,MAAM,OAAO,GAAG,CAAC,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,CAAC,WAAW,EAAE,CAAA;QACpD,MAAM,aAAa,GAAG,eAAe,CAAC,OAAO,CAAC,CAAA;QAE9C,IAAI,CAAC,aAAa,EAAE,CAAC;YACpB,MAAM,IAAI,KAAK,CACd,0CAA0C,OAAO,YAAY,MAAM,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CACvG,CAAA;QACF,CAAC;QACD,MAAM,YAAY,GAAG,IAAI,CAAC,YAAY,IAAI,GAAG,CAAA;QAE7C,IAAI,CAAC,CAAC,YAAY,IAAI,CAAC,IAAI,YAAY,IAAI,CAAC,CAAC,EAAE,CAAC;YAC/C,MAAM,IAAI,KAAK,CAAC,0CAA0C,YAAY,EAAE,CAAC,CAAA;QAC1E,CAAC;QACD,MAAM,MAAM,GAAG,IAAI,CAAC,UAAU,IAAI,aAAa,CAAC,MAAM,CAAA;QACtD,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,IAAI,IAAI,CAAA;QAChC,MAAM,EAAE,KAAK,EAAE,GAAG,aAAa,CAAA;QAE/B,MAAM,IAAI,GAAsB,EAAE,CAAA;QAElC,KAAK,IAAI,EAAE,GAAG,CAAC,EAAE,EAAE,GAAG,KAAK,CAAC,MAAM,EAAE,EAAE,EAAE,EAAE,CAAC;YAC1C,kGAAkG;YAClG,yDAAyD;YACzD,MAAM,YAAY,GAAG,cAAc,CAAC,CAAC,IAAI,CAAC,IAAI,GAAG,CAAC,UAAU,GAAG,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAA;YAChF,MAAM,CAAC,GAAG,MAAM,UAAU,CAAC,KAAK,CAAC,EAAE,CAAE,EAAE,YAAY,CAAC,CAAA;YAEpD,KAAK,MAAM,CAAC,IAAI,CAAC;gBAAE,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA,CAAC,kEAAkE;QACnG,CAAC;QAED,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACvB,MAAM,IAAI,KAAK,CAAC,MAAM,OAAO,+DAA+D,CAAC,CAAA;QAC9F,CAAC;QAED,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,KAAK,GAAG,CAAC,CAAA;QACb,MAAM,CAAC,GAAG,IAAI,CAAC,MAAM,CAAA;QAErB,OAAO,OAAO,GAAG,KAAK,IAAI,KAAK,EAAE,GAAG,KAAK,GAAG,CAAC,EAAE,CAAC;YAC/C,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,CAAE,CAAA;YAC5C,MAAM,KAAK,GAAG,MAAM,EAAE,GAAG,YAAY,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,QAAQ,CAAA;YAClE,MAAM,KAAK,GAAG,mBAAmB,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAA;YAEnE,IAAI,CAAC,KAAK,EAAE,CAAC;gBACZ,OAAO,EAAE,CAAA;gBACT,SAAQ;YACT,CAAC;YAED,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;gBACjB,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,GAAG,EAAE,KAAK,CAAC,GAAG,EAAE,UAAU,EAAE,KAAK,CAAC,UAAU,EAAE,OAAO,EAAE,KAAK,EAAE,CAAC,GAAG,IAAI,CAAC,CAAA;gBAC9F,OAAO,EAAE,CAAA;gBACT,SAAQ;YACT,CAAC;YACD,MAAM,QAAQ,GAAG,cAAc,CAAC,MAAM,EAAE;gBACvC,MAAM,EAAE,KAAK,CAAC,UAAU,CAAC,MAAM;gBAC/B,YAAY,EAAE,KAAK,CAAC,UAAU,CAAC,YAAY;gBAC3C,QAAQ,EAAE,KAAK,CAAC,UAAU,CAAC,QAAQ;gBACnC,QAAQ,EAAE,KAAK,CAAC,UAAU,CAAC,QAAQ;aACnC,CAAC,CAAA;YACF,MAAM,SAAS,GAAG;gBACjB,GAAG,EAAE,KAAK,CAAC,GAAG;gBACd,UAAU,EAAE,KAAK,CAAC,UAAU;gBAC5B,OAAO;gBACP,MAAM,EAAE,KAAK,CAAC,MAAM;gBACpB,MAAM;gBACN,SAAS,EAAE,QAAQ;gBACnB,cAAc,EAAE,OAAO;gBACvB,OAAO,EAAE,iBAAiB,OAAO,qBAAqB,KAAK,6BAA6B;aACxF,CAAA;YACD,MAAM,OAAO,GAAG,QAAQ,CAAC,SAA2C,CAAC,CAAA;YAErE,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC;gBAChD,OAAO,EAAE,CAAA;gBACT,SAAQ;YACT,CAAC;YACD,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,GAAG,OAAO,CAAC,GAAG,EAAE,YAAY,EAAE,MAAM,EAAE,WAAW,EAAE,KAAK,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,GAAG,IAAI,CAAC,CAAA;YAC/G,OAAO,EAAE,CAAA;QACV,CAAC;QAED,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;IAC5B,CAAC;CACD,CAAA"}

package/out/src/shard-recipes/no-street.d.ts

@@ -0,0 +1,14 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `no-street` shard recipe — synthetic NO-street counter-example rows: tuples →
+ *   {@link synthesizeNoStreetRow} → aligned LabeledRow. The corpus-side counterweight to the
+ *   synth-street shard that drove v0.6.1's `dependent_locality` regression — venue+admin and
+ *   admin-only rows with explicit absence of any street-side component. Ported from
+ *   scripts/build-no-street-shard.mjs.
+ */
+import { type ShardRecipe } from "./scaffold.js";
+export declare const noStreetRecipe: ShardRecipe;
+//# sourceMappingURL=no-street.d.ts.map

package/out/src/shard-recipes/no-street.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"no-street.d.ts","sourceRoot":"","sources":["../../../src/shard-recipes/no-street.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAGH,OAAO,EAAqD,KAAK,WAAW,EAAE,MAAM,eAAe,CAAA;AAInG,eAAO,MAAM,cAAc,EAAE,WAyD5B,CAAA"}

package/out/src/shard-recipes/no-street.js

@@ -0,0 +1,66 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `no-street` shard recipe — synthetic NO-street counter-example rows: tuples →
+ *   {@link synthesizeNoStreetRow} → aligned LabeledRow. The corpus-side counterweight to the
+ *   synth-street shard that drove v0.6.1's `dependent_locality` regression — venue+admin and
+ *   admin-only rows with explicit absence of any street-side component. Ported from
+ *   scripts/build-no-street-shard.mjs.
+ */
+import { synthesizeNoStreetRow } from "../synthesize-no-street.js";
+import { alignAndWrite, makeLcg, readTuples, shardSourceID } from "./scaffold.js";
+const LICENSE = "Synthetic — derived from CC-BY / public-domain input tuples";
+export const noStreetRecipe = {
+    name: "no-street",
+    description: "No-street counter-example rows: tuples → synthesizeNoStreetRow → aligned LabeledRow",
+    mode: "tuples",
+    async run(opts, write) {
+        if (!opts.input)
+            throw new Error("no-street recipe requires --input <tuples.jsonl>");
+        // Legacy build-no-street-shard.mjs seeded the LCG via makeRandom(opts.seed) (s = seed).
+        const random = makeLcg(opts.seed);
+        const source = opts.sourceName ?? "synth-no-street";
+        let read = 0;
+        let emitted = 0;
+        let skipped = 0;
+        for await (const tuple of readTuples(opts.input)) {
+            read++;
+            if (!tuple.locality || !tuple.region || !tuple.postcode || !tuple.country) {
+                skipped++;
+                continue;
+            }
+            for (let v = 0; v < opts.variants; v++) {
+                const synth = synthesizeNoStreetRow(tuple, { random });
+                if (!synth) {
+                    skipped++;
+                    continue;
+                }
+                const ok = alignAndWrite(write, {
+                    raw: synth.raw,
+                    components: synth.components,
+                    country: tuple.country,
+                    locale: synth.locale,
+                    source,
+                    source_id: shardSourceID(source, {
+                        locality: tuple.locality,
+                        region: tuple.region,
+                        postcode: tuple.postcode,
+                        country: tuple.country,
+                        template: synth.template,
+                        v: String(v),
+                    }),
+                    corpus_version: "0.4.0",
+                    license: LICENSE,
+                }, synth.template);
+                if (ok)
+                    emitted++;
+                else
+                    skipped++;
+            }
+        }
+        return { read, emitted, skipped };
+    },
+};
+//# sourceMappingURL=no-street.js.map

package/out/src/shard-recipes/no-street.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"no-street.js","sourceRoot":"","sources":["../../../src/shard-recipes/no-street.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAEH,OAAO,EAAE,qBAAqB,EAA0B,MAAM,4BAA4B,CAAA;AAC1F,OAAO,EAAE,aAAa,EAAE,OAAO,EAAE,UAAU,EAAE,aAAa,EAAoB,MAAM,eAAe,CAAA;AAEnG,MAAM,OAAO,GAAG,6DAA6D,CAAA;AAE7E,MAAM,CAAC,MAAM,cAAc,GAAgB;IAC1C,IAAI,EAAE,WAAW;IACjB,WAAW,EAAE,qFAAqF;IAClG,IAAI,EAAE,QAAQ;IACd,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,IAAI,CAAC,IAAI,CAAC,KAAK;YAAE,MAAM,IAAI,KAAK,CAAC,kDAAkD,CAAC,CAAA;QACpF,wFAAwF;QACxF,MAAM,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACjC,MAAM,MAAM,GAAG,IAAI,CAAC,UAAU,IAAI,iBAAiB,CAAA;QACnD,IAAI,IAAI,GAAG,CAAC,CAAA;QACZ,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QAEf,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,UAAU,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;YAClD,IAAI,EAAE,CAAA;YAEN,IAAI,CAAC,KAAK,CAAC,QAAQ,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,CAAC,KAAK,CAAC,QAAQ,IAAI,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC;gBAC3E,OAAO,EAAE,CAAA;gBACT,SAAQ;YACT,CAAC;YAED,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;gBACxC,MAAM,KAAK,GAAG,qBAAqB,CAAC,KAA0B,EAAE,EAAE,MAAM,EAAE,CAAC,CAAA;gBAE3E,IAAI,CAAC,KAAK,EAAE,CAAC;oBACZ,OAAO,EAAE,CAAA;oBACT,SAAQ;gBACT,CAAC;gBACD,MAAM,EAAE,GAAG,aAAa,CACvB,KAAK,EACL;oBACC,GAAG,EAAE,KAAK,CAAC,GAAG;oBACd,UAAU,EAAE,KAAK,CAAC,UAAU;oBAC5B,OAAO,EAAE,KAAK,CAAC,OAAO;oBACtB,MAAM,EAAE,KAAK,CAAC,MAAM;oBACpB,MAAM;oBACN,SAAS,EAAE,aAAa,CAAC,MAAM,EAAE;wBAChC,QAAQ,EAAE,KAAK,CAAC,QAAQ;wBACxB,MAAM,EAAE,KAAK,CAAC,MAAM;wBACpB,QAAQ,EAAE,KAAK,CAAC,QAAQ;wBACxB,OAAO,EAAE,KAAK,CAAC,OAAO;wBACtB,QAAQ,EAAE,KAAK,CAAC,QAAQ;wBACxB,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC;qBACZ,CAAC;oBACF,cAAc,EAAE,OAAO;oBACvB,OAAO,EAAE,OAAO;iBAChB,EACD,KAAK,CAAC,QAAQ,CACd,CAAA;gBAED,IAAI,EAAE;oBAAE,OAAO,EAAE,CAAA;;oBACZ,OAAO,EAAE,CAAA;YACf,CAAC;QACF,CAAC;QAED,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;IAClC,CAAC;CACD,CAAA"}

package/out/src/shard-recipes/po-box-cedex.d.ts

@@ -0,0 +1,34 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `po-box-cedex` shard recipe — the po_box / cedex coverage lever of the parity campaign (the
+ *   unit-shard playbook applied to the two starved tags `po_box` and `cedex`). Generate-mode,
+ *   self-contained. Ported from scripts/build-po-box-cedex-shard.mjs.
+ *
+ *   Surface vocabulary is provenance-first: US designators come from `@mailwoman/codex/us`
+ *   (`isPOBox`, USPS Pub 28 §29); non-US leaders from the corpus `PO_BOX_LOCALE_TEMPLATES`;
+ *   Canadian postcodes are synthesized to the `@mailwoman/codex/ca` pattern; CEDEX rides
+ *   `@mailwoman/codex/fr` (`isCedex`); AU/NZ delivery services round-trip
+ *   `@mailwoman/codex/{au,nz}`. Span convention: the WHOLE designator+number phrase is the `po_box`
+ *   span, and "CEDEX 08" is a SEPARATE `cedex` span.
+ *
+ *   Classes (CLASS_MIX): po-box-us, po-box-us-military (#517), pmb-us, bp-fr, cedex-fr, cp-ca-fr,
+ *   po-box-ca-en, po-box-au (#517), po-box-nz (#517). `--golden` emits the leakage-safe holdout
+ *   variant ({raw, components, country}); US golden = Vermont only, FR/CA/AU/NZ = a stable
+ *   locality-hash holdout (hash%10===0).
+ *
+ *   Prerequisites (read once, before the generation loop — these do NOT consume `random`): the cached
+ *   OA zips in /tmp/oa-cache, the GeoNames Canada dump at /tmp/geonames-cache/CA.zip, and the
+ *   GeoNames POSTAL-CODE dumps for AU/NZ (/tmp/geonames-cache/{AU,NZ}-postal.zip). See the legacy
+ *   script header for the exact curl commands.
+ *
+ *   Byte-fidelity: the legacy script seeded its own mulberry32 from `--seed`
+ *   (`mulberry32(opts.seed)`); this recipe re-creates the SAME generator
+ *   (`makeMulberry32(opts.seed)`) and preserves the synthesis call order exactly, so `--seed N`
+ *   reproduces the legacy run byte-for-byte.
+ */
+import { type ShardRecipe } from "./scaffold.js";
+export declare const poBoxCedexRecipe: ShardRecipe;
+//# sourceMappingURL=po-box-cedex.d.ts.map

package/out/src/shard-recipes/po-box-cedex.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"po-box-cedex.d.ts","sourceRoot":"","sources":["../../../src/shard-recipes/po-box-cedex.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8BG;AAiBH,OAAO,EAAyD,KAAK,WAAW,EAAE,MAAM,eAAe,CAAA;AAonBvG,eAAO,MAAM,gBAAgB,EAAE,WA8J9B,CAAA"}