@mailwoman/corpus 9.3.0 → 10.0.0

package/out/codex/us-fips-state.js

@@ -1,100 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   US Census FIPS state codes → two-letter abbreviation + full display name.
- *
- *   Covers all 50 states + DC + the five primary territories (PR, GU, VI, MP, AS).
- *
- *   The data is public-domain US Census reference (FIPS PUB 5-2, withdrawn but still the canonical
- *   numeric ID used by every TIGER product). It is reproduced here so the TIGER adapter can resolve
- *   `statefp` columns (e.g. `"50"`) into a `region` component (e.g. `"VT"`) without an extra DB
- *   join.
- */
-/**
- * FIPS state-or-territory code → `{ abbreviation, name }`. Includes all 50 states, DC, and the five primary territories
- * (PR, GU, VI, MP, AS). Codes are two-digit zero-padded strings to match TIGER column `statefp`.
- */
-export const US_FIPS_STATE = Object.freeze({
-    "01": { abbreviation: "AL", name: "Alabama" },
-    "02": { abbreviation: "AK", name: "Alaska" },
-    "04": { abbreviation: "AZ", name: "Arizona" },
-    "05": { abbreviation: "AR", name: "Arkansas" },
-    "06": { abbreviation: "CA", name: "California" },
-    "08": { abbreviation: "CO", name: "Colorado" },
-    "09": { abbreviation: "CT", name: "Connecticut" },
-    "10": { abbreviation: "DE", name: "Delaware" },
-    "11": { abbreviation: "DC", name: "District of Columbia" },
-    "12": { abbreviation: "FL", name: "Florida" },
-    "13": { abbreviation: "GA", name: "Georgia" },
-    "15": { abbreviation: "HI", name: "Hawaii" },
-    "16": { abbreviation: "ID", name: "Idaho" },
-    "17": { abbreviation: "IL", name: "Illinois" },
-    "18": { abbreviation: "IN", name: "Indiana" },
-    "19": { abbreviation: "IA", name: "Iowa" },
-    "20": { abbreviation: "KS", name: "Kansas" },
-    "21": { abbreviation: "KY", name: "Kentucky" },
-    "22": { abbreviation: "LA", name: "Louisiana" },
-    "23": { abbreviation: "ME", name: "Maine" },
-    "24": { abbreviation: "MD", name: "Maryland" },
-    "25": { abbreviation: "MA", name: "Massachusetts" },
-    "26": { abbreviation: "MI", name: "Michigan" },
-    "27": { abbreviation: "MN", name: "Minnesota" },
-    "28": { abbreviation: "MS", name: "Mississippi" },
-    "29": { abbreviation: "MO", name: "Missouri" },
-    "30": { abbreviation: "MT", name: "Montana" },
-    "31": { abbreviation: "NE", name: "Nebraska" },
-    "32": { abbreviation: "NV", name: "Nevada" },
-    "33": { abbreviation: "NH", name: "New Hampshire" },
-    "34": { abbreviation: "NJ", name: "New Jersey" },
-    "35": { abbreviation: "NM", name: "New Mexico" },
-    "36": { abbreviation: "NY", name: "New York" },
-    "37": { abbreviation: "NC", name: "North Carolina" },
-    "38": { abbreviation: "ND", name: "North Dakota" },
-    "39": { abbreviation: "OH", name: "Ohio" },
-    "40": { abbreviation: "OK", name: "Oklahoma" },
-    "41": { abbreviation: "OR", name: "Oregon" },
-    "42": { abbreviation: "PA", name: "Pennsylvania" },
-    "44": { abbreviation: "RI", name: "Rhode Island" },
-    "45": { abbreviation: "SC", name: "South Carolina" },
-    "46": { abbreviation: "SD", name: "South Dakota" },
-    "47": { abbreviation: "TN", name: "Tennessee" },
-    "48": { abbreviation: "TX", name: "Texas" },
-    "49": { abbreviation: "UT", name: "Utah" },
-    "50": { abbreviation: "VT", name: "Vermont" },
-    "51": { abbreviation: "VA", name: "Virginia" },
-    "53": { abbreviation: "WA", name: "Washington" },
-    "54": { abbreviation: "WV", name: "West Virginia" },
-    "55": { abbreviation: "WI", name: "Wisconsin" },
-    "56": { abbreviation: "WY", name: "Wyoming" },
-    // Territories
-    "60": { abbreviation: "AS", name: "American Samoa" },
-    "66": { abbreviation: "GU", name: "Guam" },
-    "69": { abbreviation: "MP", name: "Northern Mariana Islands" },
-    "72": { abbreviation: "PR", name: "Puerto Rico" },
-    "78": { abbreviation: "VI", name: "Virgin Islands" },
-});
-/**
- * Lookup helper. Returns null when the FIPS code isn't recognized.
- */
-export function lookupFipsState(statefp) {
-    if (!statefp)
-        return null;
-    return US_FIPS_STATE[statefp] ?? null;
-}
-/**
- * Inverted view: two-letter postal abbreviation → `USStateInfo`. Built once at module load. Used by adapters whose
- * source data ships the abbreviation rather than the FIPS code (FCC BDC, most federal CSVs).
- */
-export const US_STATE_BY_ABBREVIATION = Object.freeze(Object.fromEntries(Object.values(US_FIPS_STATE).map((info) => [info.abbreviation, info])));
-/**
- * Lookup helper for adapters carrying 2-char USPS abbreviations (`"CA"`, `"VT"`). Case-folded; null for any value
- * outside the 50 states + DC + the five primary territories.
- */
-export function lookupStateAbbreviation(abbreviation) {
-    if (!abbreviation)
-        return null;
-    return US_STATE_BY_ABBREVIATION[abbreviation.toUpperCase()] ?? null;
-}
-//# sourceMappingURL=us-fips-state.js.map

package/out/codex/us-fips-state.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"us-fips-state.js","sourceRoot":"","sources":["../../lib/codex/us-fips-state.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAUH;;;GAGG;AACH,MAAM,CAAC,MAAM,aAAa,GAA0C,MAAM,CAAC,MAAM,CAAC;IACjF,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE;IAC7C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,EAAE;IAC5C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE;IAC7C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE;IAC9C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,YAAY,EAAE;IAChD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE;IAC9C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,aAAa,EAAE;IACjD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE;IAC9C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,sBAAsB,EAAE;IAC1D,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE;IAC7C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE;IAC7C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,EAAE;IAC5C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE;IAC3C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE;IAC9C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE;IAC7C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE;IAC1C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,EAAE;IAC5C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE;IAC9C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,WAAW,EAAE;IAC/C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE;IAC3C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE;IAC9C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,eAAe,EAAE;IACnD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE;IAC9C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,WAAW,EAAE;IAC/C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,aAAa,EAAE;IACjD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE;IAC9C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE;IAC7C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE;IAC9C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,EAAE;IAC5C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,eAAe,EAAE;IACnD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,YAAY,EAAE;IAChD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,YAAY,EAAE;IAChD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE;IAC9C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,gBAAgB,EAAE;IACpD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,cAAc,EAAE;IAClD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE;IAC1C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE;IAC9C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,EAAE;IAC5C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,cAAc,EAAE;IAClD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,cAAc,EAAE;IAClD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,gBAAgB,EAAE;IACpD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,cAAc,EAAE;IAClD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,WAAW,EAAE;IAC/C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE;IAC3C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE;IAC1C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE;IAC7C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE;IAC9C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,YAAY,EAAE;IAChD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,eAAe,EAAE;IACnD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,WAAW,EAAE;IAC/C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE;IAC7C,cAAc;IACd,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,gBAAgB,EAAE;IACpD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE;IAC1C,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,0BAA0B,EAAE;IAC9D,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,aAAa,EAAE;IACjD,IAAI,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,gBAAgB,EAAE;CACpD,CAAC,CAAA;AAEF;;GAEG;AACH,MAAM,UAAU,eAAe,CAAC,OAAkC;IACjE,IAAI,CAAC,OAAO;QAAE,OAAO,IAAI,CAAA;IAEzB,OAAO,aAAa,CAAC,OAAO,CAAC,IAAI,IAAI,CAAA;AACtC,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAA0C,MAAM,CAAC,MAAM,CAC3F,MAAM,CAAC,WAAW,CAAC,MAAM,CAAC,MAAM,CAAC,aAAa,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,YAAY,EAAE,IAAI,CAAC,CAAC,CAAC,CACzF,CAAA;AAED;;;GAGG;AACH,MAAM,UAAU,uBAAuB,CAAC,YAAuC;IAC9E,IAAI,CAAC,YAAY;QAAE,OAAO,IAAI,CAAA;IAE9B,OAAO,wBAAwB,CAAC,YAAY,CAAC,WAAW,EAAE,CAAC,IAAI,IAAI,CAAA;AACpE,CAAC"}

package/out/name-prone-us-suffixes.d.ts

@@ -1,12 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- * Re-export shim: the name-prone set moved to `@mailwoman/codex/us` (data in
- * `codex/us/street-suffix.json`, 2026-08-10) so the golden relabel flags, the #1569 slice
- * recipe, AND the Python relabel pass (via the `gazetteer affix-relabel` v2 lexicon) all read
- * one record. Import from `@mailwoman/codex/us` in new code.
- */
-export { NAME_PRONE_US_SUFFIXES } from "@mailwoman/codex/us";
-//# sourceMappingURL=name-prone-us-suffixes.d.ts.map

package/out/name-prone-us-suffixes.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"name-prone-us-suffixes.d.ts","sourceRoot":"","sources":["../lib/name-prone-us-suffixes.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EAAE,sBAAsB,EAAE,MAAM,qBAAqB,CAAA"}

package/out/name-prone-us-suffixes.js

@@ -1,12 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- * Re-export shim: the name-prone set moved to `@mailwoman/codex/us` (data in
- * `codex/us/street-suffix.json`, 2026-08-10) so the golden relabel flags, the #1569 slice
- * recipe, AND the Python relabel pass (via the `gazetteer affix-relabel` v2 lexicon) all read
- * one record. Import from `@mailwoman/codex/us` in new code.
- */
-export { NAME_PRONE_US_SUFFIXES } from "@mailwoman/codex/us";
-//# sourceMappingURL=name-prone-us-suffixes.js.map

package/out/name-prone-us-suffixes.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"name-prone-us-suffixes.js","sourceRoot":"","sources":["../lib/name-prone-us-suffixes.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EAAE,sBAAsB,EAAE,MAAM,qBAAqB,CAAA"}

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

@@ -1,25 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `bare-country` — the country name as the WHOLE query. Addresses almost never consist of one bare
- *   country name, so the class is out of the model's training distribution and the tag is a coin
- *   flip: measured on the 2026-08-13 panel, France/Germany/United States/New Zealand parsed
- *   `country` while Japan/China/Nigeria/Australia/Deutschland parsed `locality`. The retrieval-side
- *   bare-country race (#1651) covers the answer either way; this slice closes the PARSE half so the
- *   tag itself is right.
- *
- *   Source is the codex country table — every ISO canonical English name plus the curated surface
- *   forms (endonyms and long-form names). The 2–3 letter CODES are deliberately excluded: a bare
- *   `JP` or `GER` is an abbreviation register with its own ambiguity surface (`IN`, `DE`, `TO` are
- *   English words), and teaching it here would be an unmeasured claim. Case augmentation is the
- *   loader's job, as everywhere.
- */
-import { type CorpusRecipe } from "#recipes/scaffold";
-/**
- * Slice recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
- * and `description` below for the surface form it generates.
- */
-export declare const bareCountryRecipe: CorpusRecipe;
-//# sourceMappingURL=bare-country.d.ts.map

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

@@ -1 +0,0 @@
-{"version":3,"file":"bare-country.d.ts","sourceRoot":"","sources":["../../lib/recipes/bare-country.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAIH,OAAO,EAAiC,KAAK,YAAY,EAAiB,MAAM,mBAAmB,CAAA;AAgCnG;;;GAGG;AACH,eAAO,MAAM,iBAAiB,EAAE,YAqC/B,CAAA"}

package/out/recipes/bare-country.js

@@ -1,87 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `bare-country` — the country name as the WHOLE query. Addresses almost never consist of one bare
- *   country name, so the class is out of the model's training distribution and the tag is a coin
- *   flip: measured on the 2026-08-13 panel, France/Germany/United States/New Zealand parsed
- *   `country` while Japan/China/Nigeria/Australia/Deutschland parsed `locality`. The retrieval-side
- *   bare-country race (#1651) covers the answer either way; this slice closes the PARSE half so the
- *   tag itself is right.
- *
- *   Source is the codex country table — every ISO canonical English name plus the curated surface
- *   forms (endonyms and long-form names). The 2–3 letter CODES are deliberately excluded: a bare
- *   `JP` or `GER` is an abbreviation register with its own ambiguity surface (`IN`, `DE`, `TO` are
- *   English words), and teaching it here would be an unmeasured claim. Case augmentation is the
- *   loader's job, as everywhere.
- */
-import { COUNTRY_SURFACE_FORMS, CountryNames, matchCountry } from "@mailwoman/codex/country";
-import { alignAndWrite, makeMulberry32, sliceSourceID } from "#recipes/scaffold";
-/**
- * Surfaces shorter than this are the code register (`JP`, `GER`), not a name — excluded (see the module doc).
- */
-const MIN_NAME_LENGTH = 4;
-/**
- * Every (surface, iso2) pair the slice emits: the ISO canonical names plus each country's curated surface forms,
- * deduplicated on the surface string (a form shared across countries — none known — would keep its first bearer).
- */
-function* bareCountrySurfaces() {
-    const seen = new Set();
-    for (const [iso2, forms] of Object.entries(COUNTRY_SURFACE_FORMS)) {
-        for (const surface of forms) {
-            if (surface.length < MIN_NAME_LENGTH || seen.has(surface))
-                continue;
-            seen.add(surface);
-            yield { surface, iso2 };
-        }
-    }
-    for (const name of CountryNames) {
-        if (name.length < MIN_NAME_LENGTH || seen.has(name))
-            continue;
-        const matched = matchCountry(name);
-        if (!matched)
-            continue;
-        seen.add(name);
-        yield { surface: name, iso2: matched.iso2 };
-    }
-}
-/**
- * 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 bareCountryRecipe = {
-    name: "bare-country",
-    description: "The country name as the whole query (#1651 parse half): ISO names + curated endonyms, no codes",
-    mode: "generate",
-    async run(opts, write) {
-        // Seeded for parity with the other recipes; the surfaces themselves drive the content.
-        makeMulberry32(opts.seed);
-        let read = 0;
-        let emitted = 0;
-        let skipped = 0;
-        for (const { surface, iso2 } of bareCountrySurfaces()) {
-            read++;
-            const components = { country: surface };
-            const source_id = sliceSourceID("synth-bare-country", { country: surface, cc: iso2, v: String(read) });
-            const canonical = {
-                raw: surface,
-                components,
-                country: iso2,
-                locale: "und",
-                source: "synth-bare-country",
-                source_id,
-                corpus_version: "0.11.0",
-                license: "Synthetic — bare-country; surfaces from the codex country table (ISO 3166 + curated forms)",
-            };
-            if (alignAndWrite(write, canonical, "bare-country")) {
-                emitted++;
-            }
-            else {
-                skipped++;
-            }
-        }
-        return { read, emitted, skipped };
-    },
-};
-//# sourceMappingURL=bare-country.js.map

package/out/recipes/bare-country.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"bare-country.js","sourceRoot":"","sources":["../../lib/recipes/bare-country.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAEH,OAAO,EAAE,qBAAqB,EAAE,YAAY,EAAE,YAAY,EAAE,MAAM,0BAA0B,CAAA;AAE5F,OAAO,EAAE,aAAa,EAAE,cAAc,EAAqB,aAAa,EAAE,MAAM,mBAAmB,CAAA;AAEnG;;GAEG;AACH,MAAM,eAAe,GAAG,CAAC,CAAA;AAEzB;;;GAGG;AACH,QAAQ,CAAC,CAAC,mBAAmB;IAC5B,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;IAE9B,KAAK,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,qBAAqB,CAAC,EAAE,CAAC;QACnE,KAAK,MAAM,OAAO,IAAI,KAAK,EAAE,CAAC;YAC7B,IAAI,OAAO,CAAC,MAAM,GAAG,eAAe,IAAI,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC;gBAAE,SAAQ;YACnE,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;YACjB,MAAM,EAAE,OAAO,EAAE,IAAI,EAAE,CAAA;QACxB,CAAC;IACF,CAAC;IAED,KAAK,MAAM,IAAI,IAAI,YAAY,EAAE,CAAC;QACjC,IAAI,IAAI,CAAC,MAAM,GAAG,eAAe,IAAI,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC;YAAE,SAAQ;QAC7D,MAAM,OAAO,GAAG,YAAY,CAAC,IAAI,CAAC,CAAA;QAElC,IAAI,CAAC,OAAO;YAAE,SAAQ;QACtB,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;QACd,MAAM,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,CAAC,IAAI,EAAE,CAAA;IAC5C,CAAC;AACF,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAAiB;IAC9C,IAAI,EAAE,cAAc;IACpB,WAAW,EAAE,gGAAgG;IAC7G,IAAI,EAAE,UAAU;IAChB,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,uFAAuF;QACvF,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,KAAK,MAAM,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,mBAAmB,EAAE,EAAE,CAAC;YACvD,IAAI,EAAE,CAAA;YAEN,MAAM,UAAU,GAA2B,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;YAC/D,MAAM,SAAS,GAAG,aAAa,CAAC,oBAAoB,EAAE,EAAE,OAAO,EAAE,OAAO,EAAE,EAAE,EAAE,IAAI,EAAE,CAAC,EAAE,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;YAEtG,MAAM,SAAS,GAAG;gBACjB,GAAG,EAAE,OAAO;gBACZ,UAAU;gBACV,OAAO,EAAE,IAAI;gBACb,MAAM,EAAE,KAAK;gBACb,MAAM,EAAE,oBAAoB;gBAC5B,SAAS;gBACT,cAAc,EAAE,QAAQ;gBACxB,OAAO,EAAE,4FAA4F;aACrG,CAAA;YAED,IAAI,aAAa,CAAC,KAAK,EAAE,SAAS,EAAE,cAAc,CAAC,EAAE,CAAC;gBACrD,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/recipes/cn-organizational-units.d.ts

@@ -1,36 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- * @file `cn-organizational-units` recipe (#2034) — real CN address strings whose settlement is an ORGANIZATIONAL
- *   ladder (`赵光三分场二十九队`: the Zhaoguang farm, No. 3 sub-farm, No. 29 production team), labeled by rule and aligned
- *   character by character for the CJK sibling model.
- *
- *   THE LABELS ARE A READING OF THE SUFFIX, NOT A GUESS. Every generic that ends an ordinal unit (`分场`, `大队`, `队`,
- *   `连`, `团`, `组`, `场部`) is in `@mailwoman/core/locale/zh-cn-units`' one table, and the same table reads the span back
- *   after decode. The whole ordinal chain is ONE `locality_unit` span; the named head it belongs to (`赵光`, `孟定农场`) is
- *   `dependent_locality`; a province, city or county written in front of it takes `region`, `locality`, `subregion`; a
- *   Latin admin tail (`, Heilongjiang, China`) takes `region` and `country`. A row with no chain is skipped, not labeled:
- *   `红卫大队` is a village name whose generic carries no ordinal, and `苗辽林场` is a named forest farm.
- *
- *   WHERE THE ROWS COME FROM. `--input` is a JSONL of `{ raw, country }` rows — the shape of `data/coarse-placer/*.jsonl`,
- *   whose 50,000 CN rows hold 328 with unit vocabulary. That file is a local artifact and carries no per-row source; the
- *   `<name>, <admin1>, <country>` shape is the corpus's GeoNames adapter's, so the rows are stamped with GeoNames'
- *   licence and the inference is stated in the `license` field rather than hidden behind it.
- *
- *   FOR THE CJK MODEL ONLY. The tokenizer is {@link cjkAwareTokenizer}: one token per Han character, because the
- *   whitespace tokenizer reads `三分场八队` as one word and could never give it two labels. The Latin model never trains
- *   on this recipe's rows; its label set has no `locality_unit`.
- */
-import { type CorpusRecipe } from "#recipes/scaffold";
-/**
- * The components a row's string supports, every value a verbatim substring of `raw`, or `null` when the string carries
- * no organizational chain and so teaches nothing this recipe exists for.
- */
-export declare function labelCNOrganizationalRow(raw: string): Record<string, string> | null;
-/**
- * Slice recipe registered with the corpus builder — see the file header for the rows it labels and why every label is a
- * reading of a generic rather than a guess.
- */
-export declare const cnOrganizationalUnitsRecipe: CorpusRecipe;
-//# sourceMappingURL=cn-organizational-units.d.ts.map

package/out/recipes/cn-organizational-units.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"cn-organizational-units.d.ts","sourceRoot":"","sources":["../../lib/recipes/cn-organizational-units.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AAIH,OAAO,EAAE,KAAK,YAAY,EAA6B,MAAM,mBAAmB,CAAA;AAmBhF;;;GAGG;AACH,wBAAgB,wBAAwB,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,GAAG,IAAI,CAkDnF;AAmBD;;;GAGG;AACH,eAAO,MAAM,2BAA2B,EAAE,YA2EzC,CAAA"}

package/out/recipes/cn-organizational-units.js

@@ -1,159 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- * @file `cn-organizational-units` recipe (#2034) — real CN address strings whose settlement is an ORGANIZATIONAL
- *   ladder (`赵光三分场二十九队`: the Zhaoguang farm, No. 3 sub-farm, No. 29 production team), labeled by rule and aligned
- *   character by character for the CJK sibling model.
- *
- *   THE LABELS ARE A READING OF THE SUFFIX, NOT A GUESS. Every generic that ends an ordinal unit (`分场`, `大队`, `队`,
- *   `连`, `团`, `组`, `场部`) is in `@mailwoman/core/locale/zh-cn-units`' one table, and the same table reads the span back
- *   after decode. The whole ordinal chain is ONE `locality_unit` span; the named head it belongs to (`赵光`, `孟定农场`) is
- *   `dependent_locality`; a province, city or county written in front of it takes `region`, `locality`, `subregion`; a
- *   Latin admin tail (`, Heilongjiang, China`) takes `region` and `country`. A row with no chain is skipped, not labeled:
- *   `红卫大队` is a village name whose generic carries no ordinal, and `苗辽林场` is a named forest farm.
- *
- *   WHERE THE ROWS COME FROM. `--input` is a JSONL of `{ raw, country }` rows — the shape of `data/coarse-placer/*.jsonl`,
- *   whose 50,000 CN rows hold 328 with unit vocabulary. That file is a local artifact and carries no per-row source; the
- *   `<name>, <admin1>, <country>` shape is the corpus's GeoNames adapter's, so the rows are stamped with GeoNames'
- *   licence and the inference is stated in the `license` field rather than hidden behind it.
- *
- *   FOR THE CJK MODEL ONLY. The tokenizer is {@link cjkAwareTokenizer}: one token per Han character, because the
- *   whitespace tokenizer reads `三分场八队` as one word and could never give it two labels. The Latin model never trains
- *   on this recipe's rows; its label set has no `locality_unit`.
- */
-import { splitCNUnitChain } from "@mailwoman/core/locale/zh-cn-units";
-import { readTuples, sliceSourceID } from "#recipes/scaffold";
-import { alignRow } from "#utils/align";
-import { cjkAwareTokenizer } from "#utils/tokenize";
-/**
- * The leading run of Han characters (plus the digits an ordinal may be written with): the Chinese half of a row.
- */
-const LEADING_HAN = /^[\p{Script=Han}〇\d]+/u;
-/**
- * Admin prefixes a CJK address writes in front of the settlement, coarsest first. Each is matched at the START of what
- * remains, so `云南省临沧市孟定农场三分场二队` peels `云南省` (province), `临沧市` (city), and hands `孟定农场三分场二队` to the unit reader.
- */
-const ADMIN_PREFIXES = [
-    [/^(.+?(?:省|自治区))/u, "region"],
-    [/^(.+?市)/u, "locality"],
-    [/^(.+?(?:自治县|县|旗))/u, "subregion"],
-];
-/**
- * The components a row's string supports, every value a verbatim substring of `raw`, or `null` when the string carries
- * no organizational chain and so teaches nothing this recipe exists for.
- */
-export function labelCNOrganizationalRow(raw) {
-    const han = LEADING_HAN.exec(raw)?.[0];
-    if (!han)
-        return null;
-    const components = {};
-    let rest = han;
-    for (const [pattern, tag] of ADMIN_PREFIXES) {
-        const match = pattern.exec(rest);
-        // A prefix must leave something behind it, or the whole run was the admin name and there is no settlement.
-        if (match && match[1].length < rest.length) {
-            components[tag] = match[1];
-            rest = rest.slice(match[1].length);
-        }
-    }
-    const split = splitCNUnitChain(rest);
-    if (!split)
-        return null;
-    if (split.head) {
-        components.dependent_locality = split.head;
-    }
-    components.locality_unit = split.chain;
-    const tail = raw.slice(han.length).trim();
-    if (tail) {
-        // `, Heilongjiang, China` or `Hunan China` or `Inner Mongolia`: comma segments when there are commas, else the last
-        // word is the country when it says so and the rest is the admin1 name.
-        const segments = tail.includes(",")
-            ? tail
-                .split(",")
-                .map((segment) => segment.trim())
-                .filter((segment) => segment.length > 0)
-            : tailWithoutCommas(tail);
-        for (const segment of segments) {
-            if (/^china$/iu.test(segment)) {
-                components.country = segment;
-            }
-            else if (!components.region && !/^\d+$/u.test(segment)) {
-                components.region = segment;
-            }
-        }
-    }
-    return components;
-}
-/**
- * A space-separated tail: `Hunan China` → [`Hunan`, `China`]; `Inner Mongolia` → [`Inner Mongolia`]; `Xinjiang Uyghur`
- * → [`Xinjiang Uyghur`].
- */
-function tailWithoutCommas(tail) {
-    const words = tail.split(/\s+/u).filter((word) => word.length > 0);
-    const last = words.at(-1);
-    if (last && /^china$/iu.test(last) && words.length > 1) {
-        return [words.slice(0, -1).join(" "), last];
-    }
-    return [words.join(" ")];
-}
-const SOURCE = "coarse-placer-cn-units";
-/**
- * Slice recipe registered with the corpus builder — see the file header for the rows it labels and why every label is a
- * reading of a generic rather than a guess.
- */
-export const cnOrganizationalUnitsRecipe = {
-    name: "cn-organizational-units",
-    description: "CN rows whose settlement is an organizational ladder (分场/队/连/组), labeled by the suffix grammar as one locality_unit span, per-character BIO",
-    mode: "tuples",
-    options: [],
-    async run(opts, write) {
-        if (!opts.input) {
-            throw new Error("cn-organizational-units: --input <rows.jsonl> is required ({ raw, country } rows)");
-        }
-        const tokenizer = cjkAwareTokenizer();
-        let emitted = 0;
-        let skipped = 0;
-        let quarantined = 0;
-        for await (const tuple of readTuples(opts.input)) {
-            const raw = typeof tuple["raw"] === "string" ? tuple["raw"].trim() : "";
-            const country = typeof tuple["country"] === "string" ? tuple["country"].toUpperCase() : "CN";
-            if (!raw || country !== "CN") {
-                skipped++;
-                continue;
-            }
-            const components = labelCNOrganizationalRow(raw);
-            if (!components) {
-                skipped++;
-                continue;
-            }
-            if (opts.golden) {
-                write(JSON.stringify({ raw, components, country: "CN", locale: "zh-CN" }) + "\n");
-                emitted++;
-                continue;
-            }
-            const canonical = {
-                raw,
-                components,
-                country: "CN",
-                locale: "zh-CN",
-                source: opts.sourceName ?? SOURCE,
-                source_id: sliceSourceID(opts.sourceName ?? SOURCE, { raw }),
-                corpus_version: "0.4.0",
-                license: "CC-BY-4.0 — GeoNames populated places, INFERRED from the `<name>, <admin1>, <country>` row shape; data/coarse-placer carries no per-row source",
-            };
-            // Verbatim only: every value above was sliced out of `raw`, so an edit-distance match would mean this file
-            // has a bug, not that the source spells something differently.
-            const aligned = alignRow(canonical, { tokenizer, maxEditDistance: 0 });
-            if (aligned.kind !== "labeled" || !aligned.row) {
-                quarantined++;
-                continue;
-            }
-            write(JSON.stringify({ ...aligned.row, synth_method: "cn-organizational-units", synth_base_id: null }) + "\n");
-            emitted++;
-        }
-        console.error(`  cn-organizational-units: ${emitted} labeled, ${skipped} without a chain, ${quarantined} quarantined`);
-        return { emitted, skipped, quarantined };
-    },
-};
-//# sourceMappingURL=cn-organizational-units.js.map

package/out/recipes/cn-organizational-units.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"cn-organizational-units.js","sourceRoot":"","sources":["../../lib/recipes/cn-organizational-units.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AAEH,OAAO,EAAE,gBAAgB,EAAE,MAAM,oCAAoC,CAAA;AAErE,OAAO,EAAqB,UAAU,EAAE,aAAa,EAAE,MAAM,mBAAmB,CAAA;AAChF,OAAO,EAAE,QAAQ,EAAE,MAAM,cAAc,CAAA;AACvC,OAAO,EAAE,iBAAiB,EAAE,MAAM,iBAAiB,CAAA;AAEnD;;GAEG;AACH,MAAM,WAAW,GAAG,wBAAwB,CAAA;AAE5C;;;GAGG;AACH,MAAM,cAAc,GAAwF;IAC3G,CAAC,kBAAkB,EAAE,QAAQ,CAAC;IAC9B,CAAC,UAAU,EAAE,UAAU,CAAC;IACxB,CAAC,oBAAoB,EAAE,WAAW,CAAC;CACnC,CAAA;AAED;;;GAGG;AACH,MAAM,UAAU,wBAAwB,CAAC,GAAW;IACnD,MAAM,GAAG,GAAG,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAEtC,IAAI,CAAC,GAAG;QAAE,OAAO,IAAI,CAAA;IAErB,MAAM,UAAU,GAA2B,EAAE,CAAA;IAC7C,IAAI,IAAI,GAAG,GAAG,CAAA;IAEd,KAAK,MAAM,CAAC,OAAO,EAAE,GAAG,CAAC,IAAI,cAAc,EAAE,CAAC;QAC7C,MAAM,KAAK,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QAEhC,2GAA2G;QAC3G,IAAI,KAAK,IAAI,KAAK,CAAC,CAAC,CAAE,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC;YAC7C,UAAU,CAAC,GAAG,CAAC,GAAG,KAAK,CAAC,CAAC,CAAE,CAAA;YAC3B,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAE,CAAC,MAAM,CAAC,CAAA;QACpC,CAAC;IACF,CAAC;IAED,MAAM,KAAK,GAAG,gBAAgB,CAAC,IAAI,CAAC,CAAA;IAEpC,IAAI,CAAC,KAAK;QAAE,OAAO,IAAI,CAAA;IAEvB,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC;QAChB,UAAU,CAAC,kBAAkB,GAAG,KAAK,CAAC,IAAI,CAAA;IAC3C,CAAC;IAED,UAAU,CAAC,aAAa,GAAG,KAAK,CAAC,KAAK,CAAA;IAEtC,MAAM,IAAI,GAAG,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,CAAA;IAEzC,IAAI,IAAI,EAAE,CAAC;QACV,oHAAoH;QACpH,uEAAuE;QACvE,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC;YAClC,CAAC,CAAC,IAAI;iBACH,KAAK,CAAC,GAAG,CAAC;iBACV,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;iBAChC,MAAM,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC;YAC1C,CAAC,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAA;QAE1B,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;YAChC,IAAI,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;gBAC/B,UAAU,CAAC,OAAO,GAAG,OAAO,CAAA;YAC7B,CAAC;iBAAM,IAAI,CAAC,UAAU,CAAC,MAAM,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;gBAC1D,UAAU,CAAC,MAAM,GAAG,OAAO,CAAA;YAC5B,CAAC;QACF,CAAC;IACF,CAAC;IAED,OAAO,UAAU,CAAA;AAClB,CAAC;AAED;;;GAGG;AACH,SAAS,iBAAiB,CAAC,IAAY;IACtC,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAClE,MAAM,IAAI,GAAG,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAA;IAEzB,IAAI,IAAI,IAAI,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACxD,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,IAAI,CAAC,CAAA;IAC5C,CAAC;IAED,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAA;AACzB,CAAC;AAED,MAAM,MAAM,GAAG,wBAAwB,CAAA;AAEvC;;;GAGG;AACH,MAAM,CAAC,MAAM,2BAA2B,GAAiB;IACxD,IAAI,EAAE,yBAAyB;IAC/B,WAAW,EACV,6IAA6I;IAC9I,IAAI,EAAE,QAAQ;IACd,OAAO,EAAE,EAAE;IACX,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC;YACjB,MAAM,IAAI,KAAK,CAAC,mFAAmF,CAAC,CAAA;QACrG,CAAC;QAED,MAAM,SAAS,GAAG,iBAAiB,EAAE,CAAA;QACrC,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,MAAM,GAAG,GAAG,OAAO,KAAK,CAAC,KAAK,CAAC,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAA;YACvE,MAAM,OAAO,GAAG,OAAO,KAAK,CAAC,SAAS,CAAC,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,IAAI,CAAA;YAE5F,IAAI,CAAC,GAAG,IAAI,OAAO,KAAK,IAAI,EAAE,CAAC;gBAC9B,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,MAAM,UAAU,GAAG,wBAAwB,CAAC,GAAG,CAAC,CAAA;YAEhD,IAAI,CAAC,UAAU,EAAE,CAAC;gBACjB,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;gBACjB,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,GAAG,EAAE,UAAU,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC,GAAG,IAAI,CAAC,CAAA;gBAEjF,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,MAAM,SAAS,GAAG;gBACjB,GAAG;gBACH,UAAU;gBACV,OAAO,EAAE,IAAI;gBACb,MAAM,EAAE,OAAO;gBACf,MAAM,EAAE,IAAI,CAAC,UAAU,IAAI,MAAM;gBACjC,SAAS,EAAE,aAAa,CAAC,IAAI,CAAC,UAAU,IAAI,MAAM,EAAE,EAAE,GAAG,EAAE,CAAC;gBAC5D,cAAc,EAAE,OAAO;gBACvB,OAAO,EACN,gJAAgJ;aACjJ,CAAA;YAED,2GAA2G;YAC3G,+DAA+D;YAC/D,MAAM,OAAO,GAAG,QAAQ,CAAC,SAA2C,EAAE,EAAE,SAAS,EAAE,eAAe,EAAE,CAAC,EAAE,CAAC,CAAA;YAExG,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC;gBAChD,WAAW,EAAE,CAAA;gBAEb,SAAQ;YACT,CAAC;YAED,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,GAAG,OAAO,CAAC,GAAG,EAAE,YAAY,EAAE,yBAAyB,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,GAAG,IAAI,CAAC,CAAA;YAE9G,OAAO,EAAE,CAAA;QACV,CAAC;QAED,OAAO,CAAC,KAAK,CACZ,8BAA8B,OAAO,aAAa,OAAO,qBAAqB,WAAW,cAAc,CACvG,CAAA;QAED,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,WAAW,EAAE,CAAA;IACzC,CAAC;CACD,CAAA"}

package/out/recipes/cz-pcfirst-preposition.d.ts

@@ -1,22 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `cz-pcfirst-preposition` — the Czech pc-first prepositional-locality slice, fourth orthography
- *   of the #901 leading-name-boundary family. The #897 close-out read all 8 residual CZ rows as one
- *   class: a LEADING postcode mis-assigned as house_number while the multi-word "nad/pod/u X"
- *   locality shatters ("51244 Rokytnice nad Jizerou, Dolní Rokytnice 111" → street
- *   'RokytnicenadJizerou' + house_number '51244'). That leading-5-digit confusion is the
- *   anchor-pollution class whose decode-time OVERRIDE was correctly killed in #723 — this slice is
- *   the model-first fix as DATA: real prepositional localities in the order that breaks, so the
- *   model learns that a leading postcode before a multi-word name is a postcode. pc-first leads the
- *   cycle (the lesson); canonical and city-first keep the polarity balanced (the v1.9.9 lesson).
- */
-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 czPcFirstPrepositionRecipe: CorpusRecipe;
-//# sourceMappingURL=cz-pcfirst-preposition.d.ts.map

package/out/recipes/cz-pcfirst-preposition.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"cz-pcfirst-preposition.d.ts","sourceRoot":"","sources":["../../lib/recipes/cz-pcfirst-preposition.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAEH,OAAO,EAA6C,KAAK,YAAY,EAAiB,MAAM,mBAAmB,CAAA;AAS/G;;;GAGG;AACH,eAAO,MAAM,0BAA0B,EAAE,YAiFxC,CAAA"}

package/out/recipes/cz-pcfirst-preposition.js

@@ -1,100 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `cz-pcfirst-preposition` — the Czech pc-first prepositional-locality slice, fourth orthography
- *   of the #901 leading-name-boundary family. The #897 close-out read all 8 residual CZ rows as one
- *   class: a LEADING postcode mis-assigned as house_number while the multi-word "nad/pod/u X"
- *   locality shatters ("51244 Rokytnice nad Jizerou, Dolní Rokytnice 111" → street
- *   'RokytnicenadJizerou' + house_number '51244'). That leading-5-digit confusion is the
- *   anchor-pollution class whose decode-time OVERRIDE was correctly killed in #723 — this slice is
- *   the model-first fix as DATA: real prepositional localities in the order that breaks, so the
- *   model learns that a leading postcode before a multi-word name is a postcode. pc-first leads the
- *   cycle (the lesson); canonical and city-first keep the polarity balanced (the v1.9.9 lesson).
- */
-import { alignAndWrite, makeMulberry32, readTuples, sliceSourceID } from "#recipes/scaffold";
-/**
- * The order-cycle slot for the STREET-LESS form (`«city» «pc», Česko`) — the exact surface of the
- * `cz-full-praha-100-00` board row, whose absence from the street-bearing orders was the v4.5.0 no-promote's measured
- * gap.
- */
-const STREETLESS_ORDER = 3;
-/**
- * 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 czPcFirstPrepositionRecipe = {
-    name: "cz-pcfirst-preposition",
-    description: "CZ pc-first + prepositional locality (#901 family): '«pc» «city nad X», «st» «n»' — the #723 class as data",
-    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 street = String(t.street ?? "").trim();
-            const city = String(t.locality ?? "").trim();
-            const number = String(t.number ?? "").trim();
-            const postcode = String(t.postcode ?? "").trim();
-            if (!street || !city || !number || !postcode) {
-                skipped++;
-                continue;
-            }
-            const order = read % 4;
-            // The OFFICIAL Czech rendering spaces the PSČ as `NNN NN` ('512 44'); OpenAddresses stores it
-            // unspaced ('51244'), and a model trained only on the source form reads the spaced surface as
-            // house_number + garbage (the 'Praha 100 00' mangle). Alternate the two renderings so both
-            // orthographies are attested — the LABEL is the postcode either way.
-            const spaced = read % 2 === 0 && /^\d{5}$/.test(postcode);
-            const postcodeSurface = spaced ? `${postcode.slice(0, 3)} ${postcode.slice(3)}` : postcode;
-            let raw;
-            const components = order === STREETLESS_ORDER
-                ? { locality: city, postcode: postcodeSurface, country: "Česko" }
-                : { street, house_number: number, postcode: postcodeSurface, locality: city };
-            if (order === 0) {
-                raw = `${postcodeSurface} ${city}, ${street} ${number}`;
-            }
-            else if (order === 1) {
-                raw = `${street} ${number}, ${postcodeSurface} ${city}`;
-            }
-            else if (order === 2) {
-                raw = `${city}, ${postcodeSurface}, ${street} ${number}`;
-            }
-            else {
-                // order === STREETLESS_ORDER: the street-less form — `«city» «pc», Česko` — the exact surface of the
-                // cz-full-praha-100-00 board row. The v4.5.0 no-promote receipt measured the gap: every
-                // prior order was street-bearing, so the model never saw a spaced PSČ beside a bare
-                // locality and mangled 'Praha 100 00, Czechia' into house_number spans. Street/number
-                // stay OUT of the components for this form (they are not in the surface).
-                raw = `${city} ${postcodeSurface}, Česko`;
-            }
-            const source_id = sliceSourceID("synth-cz-pcfirst-preposition", {
-                ...components,
-                o: String(order),
-                s: spaced ? "1" : "0",
-                v: String(read),
-            });
-            const canonical = {
-                raw,
-                components,
-                country: "CZ",
-                locale: "cs-CZ",
-                source: "synth-cz-pcfirst-preposition",
-                source_id,
-                corpus_version: "0.10.0",
-                license: "Synthetic — cz-pcfirst-preposition; (street, number, postcode, city) from OpenAddresses CZ (per-source attribution in the model card)",
-            };
-            if (alignAndWrite(write, canonical, "cz-pcfirst-preposition")) {
-                emitted++;
-            }
-            else {
-                skipped++;
-            }
-        }
-        return { read, emitted, skipped };
-    },
-};
-//# sourceMappingURL=cz-pcfirst-preposition.js.map

package/out/recipes/cz-pcfirst-preposition.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"cz-pcfirst-preposition.js","sourceRoot":"","sources":["../../lib/recipes/cz-pcfirst-preposition.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAEH,OAAO,EAAE,aAAa,EAAE,cAAc,EAAE,UAAU,EAAqB,aAAa,EAAE,MAAM,mBAAmB,CAAA;AAE/G;;;;GAIG;AACH,MAAM,gBAAgB,GAAG,CAAC,CAAA;AAE1B;;;GAGG;AACH,MAAM,CAAC,MAAM,0BAA0B,GAAiB;IACvD,IAAI,EAAE,wBAAwB;IAC9B,WAAW,EACV,4GAA4G;IAC7G,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,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAC5C,MAAM,IAAI,GAAG,MAAM,CAAC,CAAC,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAC5C,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,MAAM,IAAI,CAAC,IAAI,IAAI,CAAC,MAAM,IAAI,CAAC,QAAQ,EAAE,CAAC;gBAC9C,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,MAAM,KAAK,GAAG,IAAI,GAAG,CAAC,CAAA;YACtB,8FAA8F;YAC9F,8FAA8F;YAC9F,2FAA2F;YAC3F,qEAAqE;YACrE,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,IAAI,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;YACzD,MAAM,eAAe,GAAG,MAAM,CAAC,CAAC,CAAC,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAA;YAC1F,IAAI,GAAW,CAAA;YAEf,MAAM,UAAU,GACf,KAAK,KAAK,gBAAgB;gBACzB,CAAC,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE,QAAQ,EAAE,eAAe,EAAE,OAAO,EAAE,OAAO,EAAE;gBACjE,CAAC,CAAC,EAAE,MAAM,EAAE,YAAY,EAAE,MAAM,EAAE,QAAQ,EAAE,eAAe,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAA;YAE/E,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;gBACjB,GAAG,GAAG,GAAG,eAAe,IAAI,IAAI,KAAK,MAAM,IAAI,MAAM,EAAE,CAAA;YACxD,CAAC;iBAAM,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;gBACxB,GAAG,GAAG,GAAG,MAAM,IAAI,MAAM,KAAK,eAAe,IAAI,IAAI,EAAE,CAAA;YACxD,CAAC;iBAAM,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;gBACxB,GAAG,GAAG,GAAG,IAAI,KAAK,eAAe,KAAK,MAAM,IAAI,MAAM,EAAE,CAAA;YACzD,CAAC;iBAAM,CAAC;gBACP,qGAAqG;gBACrG,wFAAwF;gBACxF,oFAAoF;gBACpF,sFAAsF;gBACtF,0EAA0E;gBAC1E,GAAG,GAAG,GAAG,IAAI,IAAI,eAAe,SAAS,CAAA;YAC1C,CAAC;YAED,MAAM,SAAS,GAAG,aAAa,CAAC,8BAA8B,EAAE;gBAC/D,GAAG,UAAU;gBACb,CAAC,EAAE,MAAM,CAAC,KAAK,CAAC;gBAChB,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG;gBACrB,CAAC,EAAE,MAAM,CAAC,IAAI,CAAC;aACf,CAAC,CAAA;YAEF,MAAM,SAAS,GAAG;gBACjB,GAAG;gBACH,UAAU;gBACV,OAAO,EAAE,IAAI;gBACb,MAAM,EAAE,OAAO;gBACf,MAAM,EAAE,8BAA8B;gBACtC,SAAS;gBACT,cAAc,EAAE,QAAQ;gBACxB,OAAO,EACN,uIAAuI;aACxI,CAAA;YAED,IAAI,aAAa,CAAC,KAAK,EAAE,SAAS,EAAE,wBAAwB,CAAC,EAAE,CAAC;gBAC/D,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"}