@mailwoman/corpus 9.4.0 → 10.1.0

package/out/recipes/sources.js

@@ -0,0 +1,348 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ * @file The `source` identifiers that recipe outputs write, under their retired and current spellings.
+ *
+ *   A recipe-output `source` is a wire identifier rather than prose. It is the literal value of the `source` column on
+ *   every row of every built corpus, the key a training config's `source_weights`, `source_reps`,
+ *   `augment_exclude_sources` and `required_corpus_receipts[].source` address this value. `overlay-manifest` records
+ *   the `--source` label per file. Published model cards also quote the string. A rename requires a data migration
+ *   across every corpus that stores it. This table retains both spellings while an archived corpus or historical
+ *   config records the older one.
+ *
+ *   The retired spelling is `synth-<tail>`. It was read as "synthetic", and most of the rows it labels are real
+ *   published records written in a layout: `synth-german` stores `register: openaddresses` and `surface: composed`.
+ *   The current spelling is `<operation>-<tail>`, where the operation states what the recipe did to attested data.
+ *   The tail is kept byte-for-byte, so the mapping is a prefix swap in either direction.
+ *
+ *   The prefix states the recipe's operation. That value stays constant per source. It omits `surface`, a row property that
+ *   varies within a source. It also omits the register because `requireRegister` makes that an invocation property.
+ *
+ *   Every recipe reads its default `source` from this table through {@linkcode defaultRecipeSource}, and
+ *   {@linkcode WRITE_CURRENT_SOURCE_NAMES} decides which spelling it answers with. An assembly rewrites its routed
+ *   overlays through this table before `overlay-manifest` runs, so the rows and the recipe defaults move together.
+ */
+/**
+ * What a recipe did to attested data to produce its rows.
+ *
+ * This describes the operation behind a source's id.
+ * The row's `surface` is a separate field that `SurfaceOrigin` in `#types` records,
+ * and the two spell `rendered` for different classifications.
+ *
+ * A `rendered-*` source records `SurfaceOrigin.Composed`, because `renderLocaleRow` draws
+ * on the row's house number and postcode and takes an order, so it varies the written form
+ * rather than writing one canonical form per record.
+ * `SurfaceOrigin.Rendered` belongs to an adapter that writes that one form.
+ */
+export const SourceOperation = {
+    /**
+     * One real record, written in a layout its country uses, with the form varied across rows.
+     */
+    Rendered: "rendered",
+    /**
+     * Part of one real record.
+     */
+    Fragment: "fragment",
+    /**
+     * A real record with a drawn component added.
+     */
+    Spliced: "spliced",
+    /**
+     * No real record behind the row.
+     */
+    Invented: "invented",
+    /**
+     * A string copied verbatim from `@mailwoman/codex`.
+     */
+    Codex: "codex",
+    /**
+     * Rows a person wrote and checked.
+     */
+    Reviewed: "reviewed",
+};
+/**
+ * Builds the entries of one operation, so each line below reads as `retired tail → producer`.
+ */
+function entries(operation, prefix, rows) {
+    return rows.map(([tail, producer, note]) => ({
+        retired: `synth-${tail}`,
+        current: `${prefix}-${tail}`,
+        operation,
+        producer,
+        ...(note ? { note } : {}),
+    }));
+}
+const LOCALE_RECIPE = "international/recipes/locale.ts";
+const TRAILING_REGION = "recipes/trailing-region.ts";
+/**
+ * Every recipe-output source that a corpus on disk, a training config
+ * or a recipe default has spelled `synth-*`.
+ *
+ * Measured on 2026-09-26 over `/mnt/mw/corpus/versioned/`: 46 values on disk,
+ * 47 keys across 226 configs and the recipe defaults.
+ * This list is their union.
+ */
+export const RECIPE_SOURCES = [
+    ...entries(SourceOperation.Rendered, "rendered", [
+        ["fr", LOCALE_RECIPE],
+        ["nl", LOCALE_RECIPE],
+        ["it", LOCALE_RECIPE],
+        ["pt", LOCALE_RECIPE],
+        ["ch", LOCALE_RECIPE],
+        ["hr", LOCALE_RECIPE],
+        ["sk", LOCALE_RECIPE],
+        ["lu", LOCALE_RECIPE],
+        ["es", LOCALE_RECIPE],
+        ["es-pedania", LOCALE_RECIPE, "The ES source run with `--district-as-locality`, under `--source-name`."],
+        ["nz", LOCALE_RECIPE],
+        ["gb", LOCALE_RECIPE],
+        ["fr-order", "fr/recipes/order.ts"],
+        ["fr-admin-split", "fr/recipes/admin-split.ts", "The département is derived from the record's own postcode."],
+        ["fr-lieudit", "fr/recipes/lieudit.ts"],
+        ["sg-register", "sg/recipes/register.ts"],
+        ["pk-register", "south-asia/recipes/register.ts"],
+        ["bd-register", "south-asia/recipes/register.ts"],
+        ["cz-pcfirst-preposition", "cz/recipes/pcfirst-preposition.ts"],
+        ["cz-pcfirst-preposition-v20", "cz/recipes/pcfirst-preposition.ts"],
+        ["cz-pcfirst-preposition-v21", "cz/recipes/pcfirst-preposition.ts"],
+        ["no-street-led", "no/recipes/street/led.ts"],
+        ["si-bare-village", "si/recipes/bare-village.ts"],
+        ["nl-postcode", "nl/recipes/postcode.ts"],
+    ]),
+    // Germany is the one locale the retired names spelled as a language word
+    // where every other reads as an ISO code, so its tail becomes `de`.
+    // Its producer already sits in `de/recipes/`, and a reader comparing `rendered-de`
+    // against `rendered-fr` and `rendered-es` sees one shape for all three.
+    //
+    // The table keeps every other tail byte-for-byte, so three conventions for where a locale sits in a
+    // tail survive: the locale by itself (`rendered-fr`), the locale first (`invented-no-street`).
+    // The locale comes last (`spliced-trailing-region-us`).
+    // A second migration over the rows each entry names would be required to normalize those tails.
+    // This migration preserves them.
+    {
+        retired: "synth-german",
+        current: "rendered-de",
+        operation: SourceOperation.Rendered,
+        producer: "de/recipes/locale.ts",
+    },
+    // Three tails include the word `fragment` themselves.
+    // It folds into the prefix instead of repeating.
+    {
+        retired: "synth-fragment",
+        current: "fragment-assay",
+        operation: SourceOperation.Fragment,
+        producer: "corpus-python/src/mailwoman_train/corpora/fragment/build.py",
+        note: "The rows carry `recipe: fragment-assay`, and no registered recipe emits this source: the producer " +
+            "named above is out of the tree, so the rows cannot be rebuilt. 4,765 of the 200,348 on disk carry " +
+            "`country: ZZ`, of which the candidate gazetteer resolves 3,115 to one country and 1,650 to none or " +
+            "several (measured 2026-09-28). The source survives that decision, so the name is settled (#2358).",
+    },
+    {
+        retired: "synth-fr-fragment",
+        current: "fragment-fr",
+        operation: SourceOperation.Fragment,
+        producer: "fr/recipes/fragment.ts",
+    },
+    {
+        retired: "synth-no-fragment",
+        current: "fragment-no",
+        operation: SourceOperation.Fragment,
+        producer: "no/recipes/fragment.ts",
+    },
+    ...entries(SourceOperation.Fragment, "fragment", [
+        ["fr-bare-street", "fr/recipes/bare-street.ts"],
+        ["fr-bare-street-v20", "fr/recipes/bare-street.ts"],
+        ["fr-bare-street-v21", "fr/recipes/bare-street.ts"],
+        ["fr-bare-street-v22", "fr/recipes/bare-street.ts"],
+    ]),
+    ...entries(SourceOperation.Spliced, "spliced", [
+        ["unit", "recipes/unit.ts", "The unit designator and number are drawn."],
+        ["unit-v30", "recipes/unit.ts"],
+        ["affix", "recipes/street/affix.ts"],
+        ["suffix-boundary", "recipes/street/affix.ts"],
+        ["country", "recipes/country-balanced.ts", "Country names are drawn without regard to the address's own country."],
+        ["sub-venue", "recipes/sub/venue/index.ts"],
+        ["trailing-region", TRAILING_REGION, "The house number is drawn on a quarter of the postcode rows."],
+        ["trailing-region-structured", TRAILING_REGION],
+        ["trailing-region-us", TRAILING_REGION],
+        ["trailing-region-ca-bare", TRAILING_REGION],
+        ["trailing-region-es-v23", TRAILING_REGION],
+        ["trailing-region-gb-v23", TRAILING_REGION],
+        ["trailing-region-es-v28", TRAILING_REGION],
+    ]),
+    ...entries(SourceOperation.Invented, "invented", [
+        ["po-box", "recipes/po/box/index.ts"],
+        ["po-box-cedex", "recipes/po/box/cedex/recipe.ts"],
+        ["po-box-military", "recipes/po/box/index.ts"],
+        ["intersection", "recipes/intersection.ts"],
+        ["anchor-absorption", "recipes/anchor-absorption.ts"],
+        ["boundary-stress", "recipes/boundary-stress.ts"],
+        ["street", "recipes/street.ts"],
+        [
+            "street-bare",
+            "recipes/street/bare.ts",
+            "`synthesizeStreetRow` composes the street from `STREET_NAMES`, `DIRECTIONAL_PREFIXES` and " +
+                "`STREET_SUFFIXES`, all hand-written word lists, over the hand-written `DEFAULT_US_BASES` city " +
+                "tuples. No register supplies any part of the row, so the operation is invented rather than composed.",
+        ],
+        ["no-street", "no/recipes/street/index.ts"],
+        ["no-street-v063", "no/recipes/street/index.ts"],
+        ["house-venue", "recipes/house-venue.ts"],
+        ["house-venue-v063", "recipes/house-venue.ts"],
+    ]),
+    ...entries(SourceOperation.Codex, "codex", [
+        ["bare-postcode", "recipes/bare/postcode/index.ts"],
+        ["bare-country", "recipes/bare/country.ts"],
+        ["bare-country-v23", "recipes/bare/country.ts"],
+    ]),
+    {
+        retired: "synth-reviewed-postcode-tail",
+        current: "reviewed-postcode-tail",
+        operation: SourceOperation.Reviewed,
+        producer: "ve/recipes/reviewed-postcode-tail.ts",
+        note: "The tail's own `reviewed` folds into the prefix.",
+    },
+];
+/**
+ * Overlay sources that are not recipe outputs and keep their spelling.
+ *
+ * Each is either a register's own name, used by an adapter or builder that reads that
+ * register by construction, or an adversarial set whose producer chose the name.
+ * They are listed so a check over training configs can tell a source that
+ * exists from one that a sweep invented.
+ */
+export const CARRIED_SOURCES = [
+    "overture",
+    "overture-latam",
+    "gnaf",
+    "osm",
+    // One id per country for the 2026-10-02 Geofabrik extracts, written by `corpus run osm --source-name`.
+    // A country's OSM rows then carry their own weight, so admitting a country,
+    // admitting its OSM rows and sampling them stay three settings.
+    // The September PK, VN and BD rows keep `osm`.
+    "osm-ru",
+    "osm-cn",
+    "osm-in",
+    "osm-id",
+    "osm-ng",
+    "osm-vn",
+    "osm-pk",
+    "osm-bd",
+    // One id per French overseas jurisdiction for the BAN department files,
+    // written by `corpus run ban --source-name`.
+    // Metropolitan rows keep `ban`, so its weight does not decide how often a territory is drawn.
+    "ban-gp",
+    "ban-mq",
+    "ban-gf",
+    "ban-re",
+    "ban-pm",
+    "ban-yt",
+    "ban-bl",
+    "ban-mf",
+    "ban-pf",
+    "ban-nc",
+    "deepseek-kryptonite",
+    "deepseek-translit-cyrl",
+    "deepseek-translit-jpan",
+    "deepseek-translit-hans",
+    "deepseek-translit-hang",
+    "deepseek-translit-armn",
+    "synthetic-secondary",
+    "overture-jp",
+    "houjin-jp",
+    "juso-kr",
+    "localdata-kr",
+    "overture-tw",
+    "gcis-tw",
+    "coarse-placer-cn-units",
+    // A 2026-07 overlay weighted by `v1.9.6-slavic-anchor.yaml` and `v1.9.7-bsplice.yaml`.
+    // The producer's code is absent from the tree.
+    // The two configs record its runs.
+    "oa-slavic",
+];
+const BY_RETIRED = new Map(RECIPE_SOURCES.map((entry) => [entry.retired, entry]));
+const BY_CURRENT = new Map(RECIPE_SOURCES.map((entry) => [entry.current, entry]));
+/**
+ * The current spelling of a recipe-output source given either spelling,
+ * or `null` for a source this table does not record.
+ *
+ * A caller rewriting a corpus must refuse on `null` rather than pass the value through, because a
+ * pass-through leaves a corpus carrying two vocabularies with no record of which files were touched.
+ */
+export function currentSourceName(source) {
+    return BY_RETIRED.get(source)?.current ?? BY_CURRENT.get(source)?.current ?? null;
+}
+/**
+ * The table entry a source belongs to, under either spelling.
+ */
+export function recipeSource(source) {
+    return BY_RETIRED.get(source) ?? BY_CURRENT.get(source) ?? null;
+}
+/**
+ * The spelling a recipe writes today, given the retired spelling it has always written.
+ *
+ * Every recipe's default `source` reads this rather than holding a literal, so the
+ * vocabulary is one constant rather than 26 files. {@linkcode WRITE_CURRENT_SOURCE_NAMES}
+ * decides which spelling it answers with.
+ * It must agree with the corpus that a recipe output joins.
+ *
+ * @throws When the table records no entry for `retired`.
+ * A pass-through would let a typo become a source ID on every row of a built corpus.
+ * `wire-identifiers` catches that failure later.
+ * This function catches it at the call.
+ */
+export function defaultRecipeSource(retired) {
+    const entry = BY_RETIRED.get(retired);
+    if (!entry) {
+        throw new Error(`RECIPE_SOURCES records no source ${retired}. A recipe's default \`source\` is a wire identifier ` +
+            `stored on every row it writes, so add the entry with its operation and producer before writing it.`);
+    }
+    return WRITE_CURRENT_SOURCE_NAMES ? entry.current : entry.retired;
+}
+/**
+ * Whether a recipe writes the operation spelling or the retired one.
+ *
+ * True since 2026-09-28, because the rewrite has run.
+ * `v0.7.0-overlay-staging` records 43 distinct sources, each under an operation prefix.
+ *
+ * Its `source-names.json` records the mapping applied per file with the row count
+ * and the ordered-`source_id` digest verified on both sides.
+ * `v0.7.0-de-holdout/corpus-v0.7.0-de-holdout` is assembled from those bytes over 766 slices.
+ *
+ * A recipe writing the retired spelling from here on would produce an output
+ * disagreeing with the corpus it joins.
+ *
+ * `v0.6.0-register-surface` and the staging directory's `.pre-rename` copy are never rewritten in place.
+ * The configs that target them keep the spelling their corpus stores.
+ */
+export const WRITE_CURRENT_SOURCE_NAMES = true;
+/**
+ * The retired spelling of a recipe-output source given either spelling, or the value unchanged.
+ *
+ * `RECIPE_SURFACES` and `OVERLAY_REGISTERS` key on the retired spelling.
+ * Both refuse a source they do not record.
+ *
+ * A corpus rewritten to the current spelling would therefore make each of them
+ * throw on every row in the corpus.
+ *
+ * Callers of those two tables resolve the key through this, so a lookup answers under
+ * either spelling and a genuinely unknown source still refuses.
+ *
+ * The value is returned unchanged for a source outside the table — an adapter id,
+ * or an overlay source id — because those tables are keyed by whatever the producer emits.
+ */
+export function retiredSourceName(source) {
+    return BY_CURRENT.get(source)?.retired ?? source;
+}
+/**
+ * Every recipe-output spelling and adapter source id this table knows.
+ *
+ * Adapter ids are not here.
+ * They are declared by each adapter as `<NAME>_ADAPTER_ID`, and a reader that needs
+ * the full set of stored `source` values joins the two.
+ */
+export function knownOverlaySourceNames() {
+    return new Set([...BY_RETIRED.keys(), ...BY_CURRENT.keys(), ...CARRIED_SOURCES]);
+}
+//# sourceMappingURL=sources.js.map

package/out/recipes/sources.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"sources.js","sourceRoot":"","sources":["../../lib/recipes/sources.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AAEH;;;;;;;;;;;GAWG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG;IAC9B;;OAEG;IACH,QAAQ,EAAE,UAAU;IACpB;;OAEG;IACH,QAAQ,EAAE,UAAU;IACpB;;OAEG;IACH,OAAO,EAAE,SAAS;IAClB;;OAEG;IACH,QAAQ,EAAE,UAAU;IACpB;;OAEG;IACH,KAAK,EAAE,OAAO;IACd;;OAEG;IACH,QAAQ,EAAE,UAAU;CACX,CAAA;AAiCV;;GAEG;AACH,SAAS,OAAO,CACf,SAA0B,EAC1B,MAAc,EACd,IAA6E;IAE7E,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,QAAQ,EAAE,IAAI,CAAC,EAAE,EAAE,CAAC,CAAC;QAC5C,OAAO,EAAE,SAAS,IAAI,EAAE;QACxB,OAAO,EAAE,GAAG,MAAM,IAAI,IAAI,EAAE;QAC5B,SAAS;QACT,QAAQ;QACR,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;KACzB,CAAC,CAAC,CAAA;AACJ,CAAC;AAED,MAAM,aAAa,GAAG,iCAAiC,CAAA;AACvD,MAAM,eAAe,GAAG,4BAA4B,CAAA;AAEpD;;;;;;;GAOG;AACH,MAAM,CAAC,MAAM,cAAc,GAAgC;IAC1D,GAAG,OAAO,CAAC,eAAe,CAAC,QAAQ,EAAE,UAAU,EAAE;QAChD,CAAC,IAAI,EAAE,aAAa,CAAC;QACrB,CAAC,IAAI,EAAE,aAAa,CAAC;QACrB,CAAC,IAAI,EAAE,aAAa,CAAC;QACrB,CAAC,IAAI,EAAE,aAAa,CAAC;QACrB,CAAC,IAAI,EAAE,aAAa,CAAC;QACrB,CAAC,IAAI,EAAE,aAAa,CAAC;QACrB,CAAC,IAAI,EAAE,aAAa,CAAC;QACrB,CAAC,IAAI,EAAE,aAAa,CAAC;QACrB,CAAC,IAAI,EAAE,aAAa,CAAC;QACrB,CAAC,YAAY,EAAE,aAAa,EAAE,yEAAyE,CAAC;QACxG,CAAC,IAAI,EAAE,aAAa,CAAC;QACrB,CAAC,IAAI,EAAE,aAAa,CAAC;QACrB,CAAC,UAAU,EAAE,qBAAqB,CAAC;QACnC,CAAC,gBAAgB,EAAE,2BAA2B,EAAE,4DAA4D,CAAC;QAC7G,CAAC,YAAY,EAAE,uBAAuB,CAAC;QACvC,CAAC,aAAa,EAAE,wBAAwB,CAAC;QACzC,CAAC,aAAa,EAAE,gCAAgC,CAAC;QACjD,CAAC,aAAa,EAAE,gCAAgC,CAAC;QACjD,CAAC,wBAAwB,EAAE,mCAAmC,CAAC;QAC/D,CAAC,4BAA4B,EAAE,mCAAmC,CAAC;QACnE,CAAC,4BAA4B,EAAE,mCAAmC,CAAC;QACnE,CAAC,eAAe,EAAE,0BAA0B,CAAC;QAC7C,CAAC,iBAAiB,EAAE,4BAA4B,CAAC;QACjD,CAAC,aAAa,EAAE,wBAAwB,CAAC;KACzC,CAAC;IACF,yEAAyE;IACzE,oEAAoE;IACpE,mFAAmF;IACnF,wEAAwE;IACxE,EAAE;IACF,oGAAoG;IACpG,+FAA+F;IAC/F,wDAAwD;IACxD,gGAAgG;IAChG,iCAAiC;IACjC;QACC,OAAO,EAAE,cAAc;QACvB,OAAO,EAAE,aAAa;QACtB,SAAS,EAAE,eAAe,CAAC,QAAQ;QACnC,QAAQ,EAAE,sBAAsB;KAChC;IACD,sDAAsD;IACtD,iDAAiD;IACjD;QACC,OAAO,EAAE,gBAAgB;QACzB,OAAO,EAAE,gBAAgB;QACzB,SAAS,EAAE,eAAe,CAAC,QAAQ;QACnC,QAAQ,EAAE,6DAA6D;QACvE,IAAI,EACH,oGAAoG;YACpG,oGAAoG;YACpG,qGAAqG;YACrG,mGAAmG;KACpG;IACD;QACC,OAAO,EAAE,mBAAmB;QAC5B,OAAO,EAAE,aAAa;QACtB,SAAS,EAAE,eAAe,CAAC,QAAQ;QACnC,QAAQ,EAAE,wBAAwB;KAClC;IACD;QACC,OAAO,EAAE,mBAAmB;QAC5B,OAAO,EAAE,aAAa;QACtB,SAAS,EAAE,eAAe,CAAC,QAAQ;QACnC,QAAQ,EAAE,wBAAwB;KAClC;IACD,GAAG,OAAO,CAAC,eAAe,CAAC,QAAQ,EAAE,UAAU,EAAE;QAChD,CAAC,gBAAgB,EAAE,2BAA2B,CAAC;QAC/C,CAAC,oBAAoB,EAAE,2BAA2B,CAAC;QACnD,CAAC,oBAAoB,EAAE,2BAA2B,CAAC;QACnD,CAAC,oBAAoB,EAAE,2BAA2B,CAAC;KACnD,CAAC;IACF,GAAG,OAAO,CAAC,eAAe,CAAC,OAAO,EAAE,SAAS,EAAE;QAC9C,CAAC,MAAM,EAAE,iBAAiB,EAAE,2CAA2C,CAAC;QACxE,CAAC,UAAU,EAAE,iBAAiB,CAAC;QAC/B,CAAC,OAAO,EAAE,yBAAyB,CAAC;QACpC,CAAC,iBAAiB,EAAE,yBAAyB,CAAC;QAC9C,CAAC,SAAS,EAAE,6BAA6B,EAAE,sEAAsE,CAAC;QAClH,CAAC,WAAW,EAAE,4BAA4B,CAAC;QAC3C,CAAC,iBAAiB,EAAE,eAAe,EAAE,8DAA8D,CAAC;QACpG,CAAC,4BAA4B,EAAE,eAAe,CAAC;QAC/C,CAAC,oBAAoB,EAAE,eAAe,CAAC;QACvC,CAAC,yBAAyB,EAAE,eAAe,CAAC;QAC5C,CAAC,wBAAwB,EAAE,eAAe,CAAC;QAC3C,CAAC,wBAAwB,EAAE,eAAe,CAAC;QAC3C,CAAC,wBAAwB,EAAE,eAAe,CAAC;KAC3C,CAAC;IACF,GAAG,OAAO,CAAC,eAAe,CAAC,QAAQ,EAAE,UAAU,EAAE;QAChD,CAAC,QAAQ,EAAE,yBAAyB,CAAC;QACrC,CAAC,cAAc,EAAE,gCAAgC,CAAC;QAClD,CAAC,iBAAiB,EAAE,yBAAyB,CAAC;QAC9C,CAAC,cAAc,EAAE,yBAAyB,CAAC;QAC3C,CAAC,mBAAmB,EAAE,8BAA8B,CAAC;QACrD,CAAC,iBAAiB,EAAE,4BAA4B,CAAC;QACjD,CAAC,QAAQ,EAAE,mBAAmB,CAAC;QAC/B;YACC,aAAa;YACb,wBAAwB;YACxB,4FAA4F;gBAC3F,gGAAgG;gBAChG,sGAAsG;SACvG;QACD,CAAC,WAAW,EAAE,4BAA4B,CAAC;QAC3C,CAAC,gBAAgB,EAAE,4BAA4B,CAAC;QAChD,CAAC,aAAa,EAAE,wBAAwB,CAAC;QACzC,CAAC,kBAAkB,EAAE,wBAAwB,CAAC;KAC9C,CAAC;IACF,GAAG,OAAO,CAAC,eAAe,CAAC,KAAK,EAAE,OAAO,EAAE;QAC1C,CAAC,eAAe,EAAE,gCAAgC,CAAC;QACnD,CAAC,cAAc,EAAE,yBAAyB,CAAC;QAC3C,CAAC,kBAAkB,EAAE,yBAAyB,CAAC;KAC/C,CAAC;IACF;QACC,OAAO,EAAE,8BAA8B;QACvC,OAAO,EAAE,wBAAwB;QACjC,SAAS,EAAE,eAAe,CAAC,QAAQ;QACnC,QAAQ,EAAE,sCAAsC;QAChD,IAAI,EAAE,kDAAkD;KACxD;CACD,CAAA;AAED;;;;;;;GAOG;AACH,MAAM,CAAC,MAAM,eAAe,GAA0B;IACrD,UAAU;IACV,gBAAgB;IAChB,MAAM;IACN,KAAK;IACL,uGAAuG;IACvG,4EAA4E;IAC5E,gEAAgE;IAChE,+CAA+C;IAC/C,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,wEAAwE;IACxE,6CAA6C;IAC7C,8FAA8F;IAC9F,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,QAAQ;IACR,qBAAqB;IACrB,wBAAwB;IACxB,wBAAwB;IACxB,wBAAwB;IACxB,wBAAwB;IACxB,wBAAwB;IACxB,qBAAqB;IACrB,aAAa;IACb,WAAW;IACX,SAAS;IACT,cAAc;IACd,aAAa;IACb,SAAS;IACT,wBAAwB;IACxB,uFAAuF;IACvF,+CAA+C;IAC/C,mCAAmC;IACnC,WAAW;CACX,CAAA;AAED,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,cAAc,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC,CAAA;AACjF,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,cAAc,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC,CAAA;AAEjF;;;;;;GAMG;AACH,MAAM,UAAU,iBAAiB,CAAC,MAAc;IAC/C,OAAO,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,OAAO,IAAI,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,OAAO,IAAI,IAAI,CAAA;AAClF,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,YAAY,CAAC,MAAc;IAC1C,OAAO,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,CAAA;AAChE,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,UAAU,mBAAmB,CAAC,OAAe;IAClD,MAAM,KAAK,GAAG,UAAU,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;IAErC,IAAI,CAAC,KAAK,EAAE,CAAC;QACZ,MAAM,IAAI,KAAK,CACd,oCAAoC,OAAO,uDAAuD;YACjG,oGAAoG,CACrG,CAAA;IACF,CAAC;IAED,OAAO,0BAA0B,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAA;AAClE,CAAC;AAED;;;;;;;;;;;;;;;GAeG;AACH,MAAM,CAAC,MAAM,0BAA0B,GAAG,IAAI,CAAA;AAE9C;;;;;;;;;;;;;;GAcG;AACH,MAAM,UAAU,iBAAiB,CAAC,MAAc;IAC/C,OAAO,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,OAAO,IAAI,MAAM,CAAA;AACjD,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,uBAAuB;IACtC,OAAO,IAAI,GAAG,CAAC,CAAC,GAAG,UAAU,CAAC,IAAI,EAAE,EAAE,GAAG,UAAU,CAAC,IAAI,EAAE,EAAE,GAAG,eAAe,CAAC,CAAC,CAAA;AACjF,CAAC"}

package/out/recipes/street/affix.d.ts

@@ -0,0 +1,59 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import type { ComponentTag } from "@mailwoman/codex/component";
+import { type CorpusRecipe } from "#recipes/scaffold";
+interface USTuple {
+    house_number: string;
+    street: string;
+    locality: string;
+    region: string;
+    postcode: string;
+    base_source_id: string;
+}
+/**
+ * The two street shapes that the suffix-boundary recipe balances.
+ *
+ * In a `terminal-only` street, a name-prone suffix word precedes the real suffix and belongs to the name.
+ * In a `terminal-contrast` street, the name-prone word is itself the final suffix.
+ */
+export type SuffixBoundaryClass = "terminal-only" | "terminal-contrast";
+/**
+ * Classifies a street as `terminal-only` or `terminal-contrast`, or returns `null` when it is neither.
+ *
+ * A street whose final word is name-prone is `terminal-contrast` even when the word
+ * before it is also name-prone, because only the final token is the suffix.
+ */
+export declare function classifySuffixBoundaryStreet(street: string): SuffixBoundaryClass | null;
+interface RenderRowOpts {
+    venues?: readonly string[];
+    cutoffs?: readonly [number, number, number];
+}
+/**
+ * Places a rendered street in a full, bare, street-only, or venue-prefixed address.
+ *
+ * The `cutoffs` option sets the cumulative probabilities of the first three formats.
+ * The remainder produces venue-prefixed rows.
+ */
+export declare function renderRow(random: () => number, base: USTuple, street: string, streetComponents: Partial<Record<ComponentTag, string>>, opts?: RenderRowOpts): {
+    fmt: string;
+    raw: string;
+    components: Partial<Record<ComponentTag, string>>;
+};
+/**
+ * Generates US rows that split OpenAddresses streets into `street_prefix`, `street` and `street_suffix`.
+ *
+ * The `--multilocale-count` option appends FR, DE, IT and NL rows in native order without affix labels.
+ */
+export declare const streetAffixRecipe: CorpusRecipe;
+/**
+ * Generates US rows that teach where a street name ends and its suffix begins.
+ *
+ * The recipe targets 80% `terminal-only` and 20% `terminal-contrast` streets.
+ * It draws training rows from non-Vermont OpenAddresses streets and reserves Vermont for the golden set.
+ */
+export declare const suffixBoundaryRecipe: CorpusRecipe;
+export {};
+//# sourceMappingURL=affix.d.ts.map

package/out/recipes/street/affix.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"affix.d.ts","sourceRoot":"","sources":["../../../lib/recipes/street/affix.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAA;AAmB9D,OAAO,EAAgD,KAAK,YAAY,EAAE,MAAM,mBAAmB,CAAA;AA0EnG,UAAU,OAAO;IAChB,YAAY,EAAE,MAAM,CAAA;IACpB,MAAM,EAAE,MAAM,CAAA;IACd,QAAQ,EAAE,MAAM,CAAA;IAChB,MAAM,EAAE,MAAM,CAAA;IACd,QAAQ,EAAE,MAAM,CAAA;IAChB,cAAc,EAAE,MAAM,CAAA;CACtB;AAmED;;;;;GAKG;AACH,MAAM,MAAM,mBAAmB,GAAG,eAAe,GAAG,mBAAmB,CAAA;AAIvE;;;;;GAKG;AACH,wBAAgB,4BAA4B,CAAC,MAAM,EAAE,MAAM,GAAG,mBAAmB,GAAG,IAAI,CAevF;AA8ED,UAAU,aAAa;IACtB,MAAM,CAAC,EAAE,SAAS,MAAM,EAAE,CAAA;IAC1B,OAAO,CAAC,EAAE,SAAS,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC,CAAA;CAC3C;AAED;;;;;GAKG;AACH,wBAAgB,SAAS,CACxB,MAAM,EAAE,MAAM,MAAM,EACpB,IAAI,EAAE,OAAO,EACb,MAAM,EAAE,MAAM,EACd,gBAAgB,EAAE,OAAO,CAAC,MAAM,CAAC,YAAY,EAAE,MAAM,CAAC,CAAC,EACvD,IAAI,GAAE,aAAkB,GACtB;IAAE,GAAG,EAAE,MAAM,CAAC;IAAC,GAAG,EAAE,MAAM,CAAC;IAAC,UAAU,EAAE,OAAO,CAAC,MAAM,CAAC,YAAY,EAAE,MAAM,CAAC,CAAC,CAAA;CAAE,CA8BjF;AAyBD;;;;GAIG;AACH,eAAO,MAAM,iBAAiB,EAAE,YA2L/B,CAAA;AAMD;;;;;GAKG;AACH,eAAO,MAAM,oBAAoB,EAAE,YAoJlC,CAAA"}