@mailwoman/corpus 5.9.0 → 5.10.0

package/out/src/shard-recipes/index.d.ts

@@ -7,8 +7,8 @@
  *   `build-*-shard.mjs` script used to be; the `mailwoman corpus shard <recipe>` command dispatches
  *   to them. A new shard is a new entry here, not a new script.
  */
-import type { ShardRecipe } from "./scaffold.js";
-export * from "./scaffold.js";
+import type { ShardRecipe } from "./scaffold.ts";
+export * from "./scaffold.ts";
 /** Recipe name → recipe. */
 export declare const SHARD_RECIPES: ReadonlyMap<string, ShardRecipe>;
 /** Look up a recipe by its `<recipe>` name. */

package/out/src/shard-recipes/intersection.d.ts

@@ -33,6 +33,6 @@
  *       /tmp/tiger-edges)
  *   - /tmp/oa-cache/us__il__cook.zip (ZIP→city tails)
  */
-import { type ShardRecipe } from "./scaffold.js";
+import { type ShardRecipe } from "./scaffold.ts";
 export declare const intersectionRecipe: ShardRecipe;
 //# sourceMappingURL=intersection.d.ts.map

package/out/src/shard-recipes/locale.d.ts

@@ -24,8 +24,8 @@
  *   source shape, the rest the spaced conventional one. These draws are consumed ONLY for their
  *   country, so DE/FR emit streams are unchanged for a given seed.
  */
-import { type LocaleBaseTuple } from "../synthesize-german.js";
-import { type ShardRecipe } from "./scaffold.js";
+import { type LocaleBaseTuple } from "../synthesize-german.ts";
+import { type ShardRecipe } from "./scaffold.ts";
 /**
  * One per-country OA source part: either a cached `zip` + `csv` member (streamed via `unzip -p`) or an extracted plain
  * `path` (streamed via `createReadStream`). Both carry the standard OA header

package/out/src/shard-recipes/nl-postcode.d.ts

@@ -17,6 +17,6 @@
  *   emitted so the model learns the digits-first postcode regardless of spacing; the three orders
  *   keep polarity balanced (the v1.9.9 lesson).
  */
-import { type ShardRecipe } from "./scaffold.js";
+import { type ShardRecipe } from "./scaffold.ts";
 export declare const nlPostcodeRecipe: ShardRecipe;
 //# sourceMappingURL=nl-postcode.d.ts.map

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

@@ -15,6 +15,6 @@
  *   2. city-first "«city», «pc», «st» «n»"
  *   3. pc-first   "«pc» «city», «st» «n»"   (the 7% floor — the anchor the others converge to)
  */
-import { type ShardRecipe } from "./scaffold.js";
+import { type ShardRecipe } from "./scaffold.ts";
 export declare const noStreetLedRecipe: ShardRecipe;
 //# sourceMappingURL=no-street-led.d.ts.map

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

@@ -9,6 +9,6 @@
  *   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";
+import { type ShardRecipe } from "./scaffold.ts";
 export declare const noStreetRecipe: ShardRecipe;
 //# sourceMappingURL=no-street.d.ts.map

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

@@ -29,6 +29,6 @@
  *   (`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";
+import { type ShardRecipe } from "./scaffold.ts";
 export declare const poBoxCedexRecipe: ShardRecipe;
 //# sourceMappingURL=po-box-cedex.d.ts.map

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

@@ -8,6 +8,6 @@
  *   `--military-ratio`. Region is required EXCEPT region-less locales (NZ). Ported from
  *   scripts/build-po-box-shard.mjs.
  */
-import { type ShardRecipe } from "./scaffold.js";
+import { type ShardRecipe } from "./scaffold.ts";
 export declare const poBoxRecipe: ShardRecipe;
 //# sourceMappingURL=po-box.d.ts.map

package/out/src/shard-recipes/si-bare-village.d.ts

@@ -23,6 +23,6 @@
  *   «n» = house_number (NEVER split mid-digits), «pc» = postcode (never swallowing the neighbor),
  *   trailing «V» = locality (the binding the resolver needs).
  */
-import { type ShardRecipe } from "./scaffold.js";
+import { type ShardRecipe } from "./scaffold.ts";
 export declare const siBareVillageRecipe: ShardRecipe;
 //# sourceMappingURL=si-bare-village.d.ts.map

package/out/src/shard-recipes/street-affix.d.ts

@@ -26,6 +26,6 @@
  *   to keep the postcode-ORDER distribution multi-locale so a US-heavy affix shard doesn't dilute
  *   FR/DE postcode (the v0.9.8 blemish).
  */
-import { type ShardRecipe } from "./scaffold.js";
+import { type ShardRecipe } from "./scaffold.ts";
 export declare const streetAffixRecipe: ShardRecipe;
 //# sourceMappingURL=street-affix.d.ts.map

package/out/src/shard-recipes/street-bare.d.ts

@@ -16,6 +16,6 @@
  *   (`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";
+import { type ShardRecipe } from "./scaffold.ts";
 export declare const streetBareRecipe: ShardRecipe;
 //# sourceMappingURL=street-bare.d.ts.map

package/out/src/shard-recipes/street.d.ts

@@ -6,6 +6,6 @@
  *   `street` shard recipe — synthetic street-decomposition rows for Stage 3 (US-only): tuples →
  *   {@link synthesizeStreetRow} → aligned LabeledRow. Ported from scripts/build-street-shard.mjs.
  */
-import { type ShardRecipe } from "./scaffold.js";
+import { type ShardRecipe } from "./scaffold.ts";
 export declare const streetRecipe: ShardRecipe;
 //# sourceMappingURL=street.d.ts.map

package/out/src/shard-recipes/unit.d.ts

@@ -20,6 +20,6 @@
  *   cached OA zips) — `--count` bounds the OUTPUT, not the input. The passed `random` (the
  *   framework LCG) is consumed in the exact call order the legacy script used.
  */
-import { type ShardRecipe } from "./scaffold.js";
+import { type ShardRecipe } from "./scaffold.ts";
 export declare const unitRecipe: ShardRecipe;
 //# sourceMappingURL=unit.d.ts.map

package/out/src/split.d.ts

@@ -23,7 +23,7 @@
  *   The output is a `SplitManifest`: three `string[]` arrays of `source_id`. Manifests live in git
  *   (under `corpus/splits/<version>/`) so reruns are reproducible bit-for-bit.
  */
-import type { CanonicalRow, LabeledRow } from "./types.js";
+import type { CanonicalRow, LabeledRow } from "./types.ts";
 export type SplitName = "train" | "val" | "test";
 export interface SplitOptions {
     /**

package/out/src/split.js

@@ -28,7 +28,7 @@ import { createHash } from "node:crypto";
 import { createWriteStream } from "node:fs";
 import { mkdir, unlink, writeFile } from "node:fs/promises";
 import { join } from "node:path";
-import { childEnv } from "@mailwoman/core/utils";
+import { childEnv } from "@mailwoman/core/scripting/utils";
 import { JSONSpliterator } from "spliterator";
 /**
  * Phase 1 default holdouts (per plan).

package/out/src/split.js.map

@@ -1 +1 @@
-{"version":3,"file":"split.js","sourceRoot":"","sources":["../../src/split.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AAEH,OAAO,EAAE,KAAK,EAAE,MAAM,oBAAoB,CAAA;AAC1C,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAA;AACxC,OAAO,EAAE,iBAAiB,EAAE,MAAM,SAAS,CAAA;AAC3C,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAA;AAC3D,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAA;AAEhC,OAAO,EAAE,QAAQ,EAAE,MAAM,uBAAuB,CAAA;AAChD,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AA4B7C;;;;;;;;;GASG;AACH,MAAM,UAAU,eAAe;IAC9B,OAAO;QACN,EAAE,EAAE,CAAC,SAAS,EAAE,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,cAAc,EAAE,IAAI,CAAC;QAC5D,EAAE,EAAE,CAAC,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,CAAC;QAC3C,EAAE,EAAE,CAAC,UAAU,EAAE,IAAI,EAAE,wBAAwB,EAAE,IAAI,CAAC;KACtD,CAAA;AACF,CAAC;AAID;;;;GAIG;AACH,MAAM,UAAU,WAAW,CAC1B,GAAgE,EAChE,WAA8C,eAAe,EAAE;IAE/D,MAAM,MAAM,GAAG,GAAG,CAAC,UAAU,CAAC,MAAM,CAAA;IACpC,MAAM,eAAe,GAAG,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,CAAA;IACnD,MAAM,SAAS,GAAG,MAAM,KAAK,SAAS,IAAI,eAAe,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAA;IAE1E,IAAI,CAAC,SAAS;QAAE,OAAO,OAAO,CAAA;IAE9B,oFAAoF;IACpF,OAAO,UAAU,CAAC,GAAG,CAAC,SAAS,EAAE,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAA;AAC3D,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,SAAS,CAAC,IAA6B,EAAE,OAAqB,EAAE;IAC/E,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,IAAI,eAAe,EAAE,CAAA;IACnD,MAAM,KAAK,GAAa,EAAE,CAAA;IAC1B,MAAM,GAAG,GAAa,EAAE,CAAA;IACxB,MAAM,IAAI,GAAa,EAAE,CAAA;IACzB,IAAI,cAAc,GAAG,EAAE,CAAA;IAEvB,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACxB,IAAI,CAAC,cAAc,IAAI,GAAG,CAAC,cAAc,EAAE,CAAC;YAC3C,cAAc,GAAG,GAAG,CAAC,cAAc,CAAA;QACpC,CAAC;QACD,MAAM,KAAK,GAAG,WAAW,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAA;QAExC,IAAI,KAAK,KAAK,OAAO,EAAE,CAAC;YACvB,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;QAC1B,CAAC;aAAM,IAAI,KAAK,KAAK,KAAK,EAAE,CAAC;YAC5B,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;QACxB,CAAC;aAAM,CAAC;YACP,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;QACzB,CAAC;IACF,CAAC;IAED,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,GAAG,GAAG,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,CAAA;IAErD,OAAO;QACN,KAAK;QACL,GAAG;QACH,IAAI;QACJ,QAAQ;QACR,cAAc;QACd,MAAM,EAAE,EAAE,KAAK,EAAE,KAAK,CAAC,MAAM,EAAE,GAAG,EAAE,GAAG,CAAC,MAAM,EAAE,IAAI,EAAE,IAAI,CAAC,MAAM,EAAE,KAAK,EAAE;KAC1E,CAAA;AACF,CAAC;AAED,kEAAkE;AAClE,MAAM,UAAU,UAAU,CAAC,EAAU,EAAE,CAAS;IAC/C,MAAM,MAAM,GAAG,UAAU,CAAC,QAAQ,CAAC,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,MAAM,EAAE,CAAA;IACvD,mDAAmD;IACnD,MAAM,CAAC,GAAG,MAAM,CAAC,CAAC,CAAE,GAAG,aAAa,GAAG,MAAM,CAAC,CAAC,CAAE,GAAG,UAAU,GAAG,MAAM,CAAC,CAAC,CAAE,GAAG,OAAO,GAAG,MAAM,CAAC,CAAC,CAAE,CAAA;IAElG,OAAO,CAAC,GAAG,CAAC,CAAA;AACb,CAAC;AAED;;;;;;GAMG;AACH,MAAM,CAAC,KAAK,UAAU,mBAAmB,CAAC,QAAuB,EAAE,SAAiB;IACnF,MAAM,KAAK,CAAC,SAAS,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;IAE3C,KAAK,MAAM,IAAI,IAAI,CAAC,OAAO,EAAE,KAAK,EAAE,MAAM,CAAU,EAAE,CAAC;QACtD,MAAM,MAAM,GAAG,CAAC,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,CAAA;QACzC,MAAM,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE,GAAG,IAAI,MAAM,CAAC,EAAE,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,MAAM,CAAC,CAAA;IACzG,CAAC;IACD,MAAM,OAAO,GAAG;QACf,cAAc,EAAE,QAAQ,CAAC,cAAc;QACvC,QAAQ,EAAE,QAAQ,CAAC,QAAQ;QAC3B,MAAM,EAAE,QAAQ,CAAC,MAAM;KACvB,CAAA;IACD,MAAM,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE,qBAAqB,CAAC,EAAE,GAAG,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,MAAM,CAAC,CAAA;AACzG,CAAC;AAKD;;;;;;;GAOG;AACH,MAAM,CAAC,KAAK,UAAU,mCAAmC,CAAC,IAMzD;IACA,MAAM,KAAK,CAAC,IAAI,CAAC,SAAS,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;IAChD,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,IAAI,eAAe,EAAE,CAAA;IAEnD,KAAK,MAAM,KAAK,IAAI,CAAC,OAAO,EAAE,KAAK,EAAE,MAAM,CAAU,EAAE,CAAC;QACvD,MAAM,WAAW,GAAG,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC,CAAA;QAC5C,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,GAAG,KAAK,MAAM,CAAC,CAAA;QACpD,MAAM,qBAAqB,CAAC,WAAW,EAAE,OAAO,CAAC,CAAA;IAClD,CAAC;IAED,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,GAAG,IAAI,CAAC,MAAM,CAAC,IAAI,CAAA;IACpE,MAAM,OAAO,GAAG;QACf,cAAc,EAAE,IAAI,CAAC,aAAa;QAClC,QAAQ;QACR,MAAM,EAAE,EAAE,GAAG,IAAI,CAAC,MAAM,EAAE,KAAK,EAAE;KACjC,CAAA;IACD,MAAM,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,qBAAqB,CAAC,EAAE,GAAG,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,MAAM,CAAC,CAAA;IAE7G,OAAO,OAAO,CAAC,MAAM,CAAA;AACtB,CAAC;AAED;;;GAGG;AACH,KAAK,UAAU,qBAAqB,CAAC,gBAAwB,EAAE,OAAe;IAC7E,MAAM,YAAY,GAAG,GAAG,OAAO,WAAW,CAAA;IAC1C,MAAM,GAAG,GAAG,iBAAiB,CAAC,YAAY,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAA;IACjE,MAAM,SAAS,GAAG,IAAI,OAAO,CAAO,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;QACvD,GAAG,CAAC,EAAE,CAAC,OAAO,EAAE,GAAG,EAAE,CAAC,OAAO,EAAE,CAAC,CAAA;QAChC,GAAG,CAAC,EAAE,CAAC,OAAO,EAAE,MAAM,CAAC,CAAA;IACxB,CAAC,CAAC,CAAA;IAEF,wFAAwF;IACxF,8FAA8F;IAC9F,wFAAwF;IACxF,IAAI,CAAC;QACJ,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,eAAe,CAAC,SAAS,CAAyB,gBAAgB,CAAC,EAAE,CAAC;YAC7F,IAAI,OAAO,GAAG,CAAC,SAAS,KAAK,QAAQ,EAAE,CAAC;gBACvC,GAAG,CAAC,KAAK,CAAC,GAAG,GAAG,CAAC,SAAS,IAAI,CAAC,CAAA;YAChC,CAAC;QACF,CAAC;IACF,CAAC;YAAS,CAAC;QACV,GAAG,CAAC,GAAG,EAAE,CAAA;IACV,CAAC;IACD,MAAM,SAAS,CAAA;IAEf,MAAM,IAAI,OAAO,CAAO,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;QAC3C,wEAAwE;QACxE,MAAM,IAAI,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,YAAY,EAAE,IAAI,EAAE,OAAO,CAAC,EAAE,EAAE,GAAG,EAAE,QAAQ,CAAC,EAAE,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC,CAAA;QAC7F,IAAI,CAAC,EAAE,CAAC,OAAO,EAAE,MAAM,CAAC,CAAA;QACxB,IAAI,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE;YACxB,IAAI,IAAI,KAAK,CAAC,EAAE,CAAC;gBAChB,OAAO,EAAE,CAAA;YACV,CAAC;iBAAM,CAAC;gBACP,MAAM,CAAC,IAAI,KAAK,CAAC,yBAAyB,IAAI,EAAE,CAAC,CAAC,CAAA;YACnD,CAAC;QACF,CAAC,CAAC,CAAA;IACH,CAAC,CAAC,CAAA;IACF,MAAM,MAAM,CAAC,YAAY,CAAC,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAA;AAC3C,CAAC"}
+{"version":3,"file":"split.js","sourceRoot":"","sources":["../../src/split.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AAEH,OAAO,EAAE,KAAK,EAAE,MAAM,oBAAoB,CAAA;AAC1C,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAA;AACxC,OAAO,EAAE,iBAAiB,EAAE,MAAM,SAAS,CAAA;AAC3C,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAA;AAC3D,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAA;AAEhC,OAAO,EAAE,QAAQ,EAAE,MAAM,iCAAiC,CAAA;AAC1D,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AA4B7C;;;;;;;;;GASG;AACH,MAAM,UAAU,eAAe;IAC9B,OAAO;QACN,EAAE,EAAE,CAAC,SAAS,EAAE,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,cAAc,EAAE,IAAI,CAAC;QAC5D,EAAE,EAAE,CAAC,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,CAAC;QAC3C,EAAE,EAAE,CAAC,UAAU,EAAE,IAAI,EAAE,wBAAwB,EAAE,IAAI,CAAC;KACtD,CAAA;AACF,CAAC;AAID;;;;GAIG;AACH,MAAM,UAAU,WAAW,CAC1B,GAAgE,EAChE,WAA8C,eAAe,EAAE;IAE/D,MAAM,MAAM,GAAG,GAAG,CAAC,UAAU,CAAC,MAAM,CAAA;IACpC,MAAM,eAAe,GAAG,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,CAAA;IACnD,MAAM,SAAS,GAAG,MAAM,KAAK,SAAS,IAAI,eAAe,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAA;IAE1E,IAAI,CAAC,SAAS;QAAE,OAAO,OAAO,CAAA;IAE9B,oFAAoF;IACpF,OAAO,UAAU,CAAC,GAAG,CAAC,SAAS,EAAE,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAA;AAC3D,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,SAAS,CAAC,IAA6B,EAAE,OAAqB,EAAE;IAC/E,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,IAAI,eAAe,EAAE,CAAA;IACnD,MAAM,KAAK,GAAa,EAAE,CAAA;IAC1B,MAAM,GAAG,GAAa,EAAE,CAAA;IACxB,MAAM,IAAI,GAAa,EAAE,CAAA;IACzB,IAAI,cAAc,GAAG,EAAE,CAAA;IAEvB,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACxB,IAAI,CAAC,cAAc,IAAI,GAAG,CAAC,cAAc,EAAE,CAAC;YAC3C,cAAc,GAAG,GAAG,CAAC,cAAc,CAAA;QACpC,CAAC;QACD,MAAM,KAAK,GAAG,WAAW,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAA;QAExC,IAAI,KAAK,KAAK,OAAO,EAAE,CAAC;YACvB,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;QAC1B,CAAC;aAAM,IAAI,KAAK,KAAK,KAAK,EAAE,CAAC;YAC5B,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;QACxB,CAAC;aAAM,CAAC;YACP,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;QACzB,CAAC;IACF,CAAC;IAED,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,GAAG,GAAG,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,CAAA;IAErD,OAAO;QACN,KAAK;QACL,GAAG;QACH,IAAI;QACJ,QAAQ;QACR,cAAc;QACd,MAAM,EAAE,EAAE,KAAK,EAAE,KAAK,CAAC,MAAM,EAAE,GAAG,EAAE,GAAG,CAAC,MAAM,EAAE,IAAI,EAAE,IAAI,CAAC,MAAM,EAAE,KAAK,EAAE;KAC1E,CAAA;AACF,CAAC;AAED,kEAAkE;AAClE,MAAM,UAAU,UAAU,CAAC,EAAU,EAAE,CAAS;IAC/C,MAAM,MAAM,GAAG,UAAU,CAAC,QAAQ,CAAC,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,MAAM,EAAE,CAAA;IACvD,mDAAmD;IACnD,MAAM,CAAC,GAAG,MAAM,CAAC,CAAC,CAAE,GAAG,aAAa,GAAG,MAAM,CAAC,CAAC,CAAE,GAAG,UAAU,GAAG,MAAM,CAAC,CAAC,CAAE,GAAG,OAAO,GAAG,MAAM,CAAC,CAAC,CAAE,CAAA;IAElG,OAAO,CAAC,GAAG,CAAC,CAAA;AACb,CAAC;AAED;;;;;;GAMG;AACH,MAAM,CAAC,KAAK,UAAU,mBAAmB,CAAC,QAAuB,EAAE,SAAiB;IACnF,MAAM,KAAK,CAAC,SAAS,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;IAE3C,KAAK,MAAM,IAAI,IAAI,CAAC,OAAO,EAAE,KAAK,EAAE,MAAM,CAAU,EAAE,CAAC;QACtD,MAAM,MAAM,GAAG,CAAC,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE,CAAA;QACzC,MAAM,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE,GAAG,IAAI,MAAM,CAAC,EAAE,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,MAAM,CAAC,CAAA;IACzG,CAAC;IACD,MAAM,OAAO,GAAG;QACf,cAAc,EAAE,QAAQ,CAAC,cAAc;QACvC,QAAQ,EAAE,QAAQ,CAAC,QAAQ;QAC3B,MAAM,EAAE,QAAQ,CAAC,MAAM;KACvB,CAAA;IACD,MAAM,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE,qBAAqB,CAAC,EAAE,GAAG,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,MAAM,CAAC,CAAA;AACzG,CAAC;AAKD;;;;;;;GAOG;AACH,MAAM,CAAC,KAAK,UAAU,mCAAmC,CAAC,IAMzD;IACA,MAAM,KAAK,CAAC,IAAI,CAAC,SAAS,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;IAChD,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,IAAI,eAAe,EAAE,CAAA;IAEnD,KAAK,MAAM,KAAK,IAAI,CAAC,OAAO,EAAE,KAAK,EAAE,MAAM,CAAU,EAAE,CAAC;QACvD,MAAM,WAAW,GAAG,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC,CAAA;QAC5C,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,GAAG,KAAK,MAAM,CAAC,CAAA;QACpD,MAAM,qBAAqB,CAAC,WAAW,EAAE,OAAO,CAAC,CAAA;IAClD,CAAC;IAED,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,GAAG,IAAI,CAAC,MAAM,CAAC,IAAI,CAAA;IACpE,MAAM,OAAO,GAAG;QACf,cAAc,EAAE,IAAI,CAAC,aAAa;QAClC,QAAQ;QACR,MAAM,EAAE,EAAE,GAAG,IAAI,CAAC,MAAM,EAAE,KAAK,EAAE;KACjC,CAAA;IACD,MAAM,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,qBAAqB,CAAC,EAAE,GAAG,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,MAAM,CAAC,CAAA;IAE7G,OAAO,OAAO,CAAC,MAAM,CAAA;AACtB,CAAC;AAED;;;GAGG;AACH,KAAK,UAAU,qBAAqB,CAAC,gBAAwB,EAAE,OAAe;IAC7E,MAAM,YAAY,GAAG,GAAG,OAAO,WAAW,CAAA;IAC1C,MAAM,GAAG,GAAG,iBAAiB,CAAC,YAAY,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAA;IACjE,MAAM,SAAS,GAAG,IAAI,OAAO,CAAO,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;QACvD,GAAG,CAAC,EAAE,CAAC,OAAO,EAAE,GAAG,EAAE,CAAC,OAAO,EAAE,CAAC,CAAA;QAChC,GAAG,CAAC,EAAE,CAAC,OAAO,EAAE,MAAM,CAAC,CAAA;IACxB,CAAC,CAAC,CAAA;IAEF,wFAAwF;IACxF,8FAA8F;IAC9F,wFAAwF;IACxF,IAAI,CAAC;QACJ,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,eAAe,CAAC,SAAS,CAAyB,gBAAgB,CAAC,EAAE,CAAC;YAC7F,IAAI,OAAO,GAAG,CAAC,SAAS,KAAK,QAAQ,EAAE,CAAC;gBACvC,GAAG,CAAC,KAAK,CAAC,GAAG,GAAG,CAAC,SAAS,IAAI,CAAC,CAAA;YAChC,CAAC;QACF,CAAC;IACF,CAAC;YAAS,CAAC;QACV,GAAG,CAAC,GAAG,EAAE,CAAA;IACV,CAAC;IACD,MAAM,SAAS,CAAA;IAEf,MAAM,IAAI,OAAO,CAAO,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;QAC3C,wEAAwE;QACxE,MAAM,IAAI,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,YAAY,EAAE,IAAI,EAAE,OAAO,CAAC,EAAE,EAAE,GAAG,EAAE,QAAQ,CAAC,EAAE,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC,CAAA;QAC7F,IAAI,CAAC,EAAE,CAAC,OAAO,EAAE,MAAM,CAAC,CAAA;QACxB,IAAI,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE;YACxB,IAAI,IAAI,KAAK,CAAC,EAAE,CAAC;gBAChB,OAAO,EAAE,CAAA;YACV,CAAC;iBAAM,CAAC;gBACP,MAAM,CAAC,IAAI,KAAK,CAAC,yBAAyB,IAAI,EAAE,CAAC,CAAC,CAAA;YACnD,CAAC;QACF,CAAC,CAAC,CAAA;IACH,CAAC,CAAC,CAAA;IACF,MAAM,MAAM,CAAC,YAAY,CAAC,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAA;AAC3C,CAAC"}

package/out/src/synthesize-boundary-stress.d.ts

@@ -48,7 +48,7 @@
  *   separately-scoped AU/NZ/UK boundary-coverage shard that ALSO adds AU base coverage, not in this
  *   base-locale shard. `synthesize-boundary-stress.test.ts` proves the alignments.
  */
-import type { CanonicalRow } from "./types.js";
+import type { CanonicalRow } from "./types.ts";
 export type BoundaryStressTemplate = "street-eats-affix" | "comma-less-city-state" | "fr-prefix" | "house-number-after-street" | "bare-locality" | "house-number-before-street";
 export interface BoundaryStressBaseTuple {
     locality: string;

package/out/src/synthesize-german.d.ts

@@ -21,7 +21,7 @@
  *   aligner turns the row into BIO labels; every emitted component surface form occurs verbatim in
  *   `raw` so alignment lands.
  */
-import type { CanonicalRow } from "./types.js";
+import type { CanonicalRow } from "./types.ts";
 /** A real address tuple (e.g. one OpenAddresses row): street + locality required, rest optional. */
 export interface LocaleBaseTuple {
     house_number?: string;

package/out/src/synthesize-house-venue.d.ts

@@ -27,7 +27,7 @@
  *   are deliberately NOT used here — the point is to teach co-occurrence, not to re-introduce
  *   decompose-mode pressure.
  */
-import type { CanonicalRow } from "./types.js";
+import type { CanonicalRow } from "./types.ts";
 export interface HouseVenueBaseTuple {
     locality: string;
     region: string;

package/out/src/synthesize-intersection.d.ts

@@ -18,7 +18,7 @@
  *
  *   US-idiomatic only (the harness intersection cases are US: "X & Y, City, ST ZIP").
  */
-import type { CanonicalRow } from "./types.js";
+import type { CanonicalRow } from "./types.ts";
 export interface IntersectionBaseTuple {
     locality: string;
     region: string;

package/out/src/synthesize-no-street.d.ts

@@ -36,7 +36,7 @@
  *   This complements (does not replace) the existing US-base-tuple source used by
  *   `synthesize-po-box.ts`; the same `NoStreetBaseTuple` shape is consumed.
  */
-import type { CanonicalRow } from "./types.js";
+import type { CanonicalRow } from "./types.ts";
 export interface NoStreetBaseTuple {
     locality: string;
     region: string;

package/out/src/synthesize-po-box.d.ts

@@ -23,7 +23,7 @@
  *   - USPS Pub 28 §28C2.040 — Private Mailbox formatting
  *   - USPS DMM 508 §4.1.4 / §4.5.4 — PO Box and street-addressed PO Box
  */
-import type { CanonicalRow } from "./types.js";
+import type { CanonicalRow } from "./types.ts";
 export interface PoBoxBaseTuple {
     locality: string;
     region: string;

package/out/src/synthesize-street.d.ts

@@ -18,7 +18,7 @@
  *   distribution matches what the model would see if/when TIGER shards are rebuilt with the new
  *   adapter.
  */
-import type { CanonicalRow } from "./types.js";
+import type { CanonicalRow } from "./types.ts";
 export interface StreetBaseTuple {
     locality: string;
     region: string;

package/out/src/synthesize.d.ts

@@ -25,8 +25,8 @@
  *   is kept OUT of the default set ({@link defaultAugmentationsForCountry}) until its on-model
  *   effect is measured; see the note there.
  */
-import { type Tokenizer } from "./tokenize.js";
-import type { CanonicalRow, LabeledRow, QuarantinedRow } from "./types.js";
+import { type Tokenizer } from "./tokenize.ts";
+import type { CanonicalRow, LabeledRow, QuarantinedRow } from "./types.ts";
 /**
  * An augmentation transforms a single row. Return `null` if the augmentation doesn't apply (e.g. accent-strip on a row
  * that has no accents; particle-strip on a US row).

package/out/src/tools/align-shard.d.ts

@@ -0,0 +1,8 @@
+#!/usr/bin/env node
+export interface AlignShardOptions {
+    input: string;
+    output: string;
+    corpusVersion: string;
+}
+export declare function alignCanonicalShard(args: AlignShardOptions): Promise<void>;
+//# sourceMappingURL=align-shard.d.ts.map

package/out/src/tools/align-shard.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"align-shard.d.ts","sourceRoot":"","sources":["../../../src/tools/align-shard.ts"],"names":[],"mappings":";AA+BA,MAAM,WAAW,iBAAiB;IACjC,KAAK,EAAE,MAAM,CAAA;IACb,MAAM,EAAE,MAAM,CAAA;IACd,aAAa,EAAE,MAAM,CAAA;CACrB;AAED,wBAAsB,mBAAmB,CAAC,IAAI,EAAE,iBAAiB,GAAG,OAAO,CAAC,IAAI,CAAC,CA6BhF"}

package/out/src/tools/align-shard.js

@@ -0,0 +1,58 @@
+#!/usr/bin/env node
+import { createWriteStream } from "node:fs";
+/**
+ * Re-emit a CANONICAL jsonl ({raw, components, country, source, ...}) as a LABELED jsonl in the CURRENT align format,
+ * by running every row through `alignRow` (corpus/src/align.ts).
+ *
+ * ## Why this exists
+ *
+ * Most synthetic shards are GENERATED on demand by a `build-*-shard` recipe (parametrized by --count), so re-emitting
+ * them in a new label format is just a re-run. A few shards are FIXED corpora with a hand/DeepSeek-authored canonical
+ * source that is never regenerated — notably `deepseek-kryptonite` (the adversarial hard-case set) and the
+ * `deepseek-translit-*` variants. Their committed parquets carry whatever label format was current when they were first
+ * built.
+ *
+ * When the corpus label format changes (the v0.5.0 char-offset triple, #519), those fixed shards must be RE-ALIGNED,
+ * not regenerated — feed the canonical source back through the same `alignRow` the from-source build uses, so the spans
+ * land in the new format with zero drift. That is exactly what this does: canonical jsonl in → labeled jsonl out, one
+ * `alignRow` per row, quarantine on miss.
+ *
+ * It is the uniform counterpart to corpus/src/tools/shard-kryptonite.ts (which couples to a base manifest and writes
+ * parquet directly). Output goes to jsonl so it joins the SAME jsonl-to-parquet path every other overlay shard uses.
+ *
+ * Usage: node scripts/align-canonical-shard.ts\
+ * --input /path/canonical-kryptonite.jsonl\
+ * --output /tmp/kryptonite-labeled.jsonl\
+ * --corpus-version 0.5.0
+ */
+import { alignRow } from "@mailwoman/corpus";
+import { TextSpliterator } from "spliterator";
+export async function alignCanonicalShard(args) {
+    // Read phase only — the write path stays on createWriteStream. TextSpliterator + JSON.parse keeps the
+    // original tolerance: the `!line.trim()` guard skips blank lines and a trailing CR is valid JSON whitespace.
+    const outStream = createWriteStream(args.output, { encoding: "utf8" });
+    let labeled = 0;
+    let quarantined = 0;
+    const quarantineReasons = {};
+    for await (const line of TextSpliterator.fromAsync(args.input)) {
+        if (!line.trim())
+            continue;
+        const canonical = JSON.parse(line);
+        // Stamp the target corpus version so the emitted row's provenance matches the run it joins.
+        canonical.corpus_version = args.corpusVersion;
+        const result = alignRow(canonical);
+        if (result.kind === "labeled") {
+            outStream.write(JSON.stringify(result.row) + "\n");
+            labeled++;
+        }
+        else {
+            quarantined++;
+            const r = result.row.reason ?? "unknown";
+            quarantineReasons[r] = (quarantineReasons[r] ?? 0) + 1;
+        }
+    }
+    await new Promise((res) => outStream.end(res));
+    console.error(`align-canonical-shard: ${labeled} labeled, ${quarantined} quarantined → ${args.output}\n` +
+        `  quarantine reasons: ${JSON.stringify(quarantineReasons)}`);
+}
+//# sourceMappingURL=align-shard.js.map

package/out/src/tools/align-shard.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"align-shard.js","sourceRoot":"","sources":["../../../src/tools/align-shard.ts"],"names":[],"mappings":";AACA,OAAO,EAAE,iBAAiB,EAAE,MAAM,SAAS,CAAA;AAE3C;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AACH,OAAO,EAAE,QAAQ,EAAE,MAAM,mBAAmB,CAAA;AAC5C,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAQ7C,MAAM,CAAC,KAAK,UAAU,mBAAmB,CAAC,IAAuB;IAChE,sGAAsG;IACtG,6GAA6G;IAC7G,MAAM,SAAS,GAAG,iBAAiB,CAAC,IAAI,CAAC,MAAM,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAA;IACtE,IAAI,OAAO,GAAG,CAAC,CAAA;IACf,IAAI,WAAW,GAAG,CAAC,CAAA;IACnB,MAAM,iBAAiB,GAA2B,EAAE,CAAA;IAEpD,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,eAAe,CAAC,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;QAChE,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE;YAAE,SAAQ;QAC1B,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAmC,CAAA;QACpE,4FAA4F;QAC5F,SAAS,CAAC,cAAc,GAAG,IAAI,CAAC,aAAa,CAAA;QAC7C,MAAM,MAAM,GAAG,QAAQ,CAAC,SAAS,CAAC,CAAA;QAElC,IAAI,MAAM,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;YAC/B,SAAS,CAAC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,CAAA;YAClD,OAAO,EAAE,CAAA;QACV,CAAC;aAAM,CAAC;YACP,WAAW,EAAE,CAAA;YACb,MAAM,CAAC,GAAG,MAAM,CAAC,GAAG,CAAC,MAAM,IAAI,SAAS,CAAA;YACxC,iBAAiB,CAAC,CAAC,CAAC,GAAG,CAAC,iBAAiB,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;QACvD,CAAC;IACF,CAAC;IACD,MAAM,IAAI,OAAO,CAAO,CAAC,GAAG,EAAE,EAAE,CAAC,SAAS,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAA;IACpD,OAAO,CAAC,KAAK,CACZ,0BAA0B,OAAO,aAAa,WAAW,kBAAkB,IAAI,CAAC,MAAM,IAAI;QACzF,yBAAyB,IAAI,CAAC,SAAS,CAAC,iBAAiB,CAAC,EAAE,CAC7D,CAAA;AACF,CAAC"}

package/out/src/tools/audit.d.ts

@@ -0,0 +1,28 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `mailwoman corpus audit` — per-source shard-count vs source_weight diagnostic.
+ *
+ *   Reads a corpus dir's MANIFEST.json (or scans shards directly), counts shards per source,
+ *   optionally loads a training config to pair the counts with the configured source_weights, and
+ *   reports the estimated sampled-row distribution at training time.
+ *
+ *   Would have caught v0.3.0's "NAD = 411/674 train shards × 2.0 weight = ~75% of sampled mix"
+ *   finding before the v0.3.0 retrospective surfaced it.
+ *
+ *   Emits warnings to stderr and the audit table to stdout; never throws on an empty corpus.
+ */
+export interface AuditOpts {
+    corpusDir: string;
+    configPath?: string;
+    /**
+     * Sample at most N shards per split when counting sources. Default 100 for speed; bump to read the full set on a slow
+     * run. The first row of each shard determines its source — corpus-v0.2.0+ shards are 100% source-segregated, so a
+     * one-row read is authoritative.
+     */
+    sampleShardCount?: number;
+}
+export declare function audit(opts: AuditOpts): void;
+//# sourceMappingURL=audit.d.ts.map

package/out/src/tools/audit.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"audit.d.ts","sourceRoot":"","sources":["../../../src/tools/audit.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAKH,MAAM,WAAW,SAAS;IACzB,SAAS,EAAE,MAAM,CAAA;IACjB,UAAU,CAAC,EAAE,MAAM,CAAA;IACnB;;;;OAIG;IACH,gBAAgB,CAAC,EAAE,MAAM,CAAA;CACzB;AAyTD,wBAAgB,KAAK,CAAC,IAAI,EAAE,SAAS,GAAG,IAAI,CAS3C"}

package/out/src/tools/audit.js

@@ -0,0 +1,273 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `mailwoman corpus audit` — per-source shard-count vs source_weight diagnostic.
+ *
+ *   Reads a corpus dir's MANIFEST.json (or scans shards directly), counts shards per source,
+ *   optionally loads a training config to pair the counts with the configured source_weights, and
+ *   reports the estimated sampled-row distribution at training time.
+ *
+ *   Would have caught v0.3.0's "NAD = 411/674 train shards × 2.0 weight = ~75% of sampled mix"
+ *   finding before the v0.3.0 retrospective surfaced it.
+ *
+ *   Emits warnings to stderr and the audit table to stdout; never throws on an empty corpus.
+ */
+import { existsSync, readFileSync, readdirSync } from "node:fs";
+import { basename, join } from "node:path";
+/**
+ * Try parsing a training YAML's source_weights as a minimal regex-based extract. We don't pull in a YAML lib for this
+ * script — the syntax is so small that a regex over the source_weights block is sufficient + keeps the script
+ * dep-free.
+ */
+function parseConfig(configPath) {
+    if (!existsSync(configPath))
+        return null;
+    const text = readFileSync(configPath, "utf8");
+    const lines = text.split("\n");
+    const weights = {};
+    let inBlock = false;
+    let blockIndent = -1;
+    for (const raw of lines) {
+        const sourceWeightsMatch = raw.match(/^([\t ]*)source_weights:\s*$/);
+        if (sourceWeightsMatch) {
+            inBlock = true;
+            blockIndent = sourceWeightsMatch[1].length;
+            continue;
+        }
+        if (!inBlock)
+            continue;
+        // Skip blank lines and comments.
+        if (/^[\t ]*(#|$)/.test(raw))
+            continue;
+        // Lines indented MORE than `source_weights:` are entries; lines with ≤ indent end the block.
+        const indent = raw.match(/^[\t ]*/)[0].length;
+        if (indent <= blockIndent) {
+            inBlock = false;
+            continue;
+        }
+        const m = raw.match(/^[\t ]+([\w-]+):\s*([\d.]+)/);
+        if (m) {
+            weights[m[1]] = parseFloat(m[2]);
+        }
+    }
+    return { sourceWeights: weights };
+}
+/**
+ * Scan a corpus directory's shards (typically under <corpus_dir>/train, /val, /test) and count shards per source per
+ * split.
+ */
+function scanShards(corpusDir, sampleCount) {
+    const stats = { bySplit: {}, totalShards: 0, totalFiles: 0 };
+    for (const split of ["train", "val", "test"]) {
+        const splitDir = join(corpusDir, split);
+        if (!existsSync(splitDir))
+            continue;
+        const files = readdirSync(splitDir)
+            .filter((f) => f.endsWith(".parquet"))
+            .sort();
+        stats.totalFiles += files.length;
+        const sampleEvery = Math.max(1, Math.floor(files.length / sampleCount));
+        const sampled = files.filter((_, i) => i % sampleEvery === 0).slice(0, sampleCount);
+        const splitMap = {};
+        // We can't read parquet without a dep, so we infer source from filenames where possible.
+        // The corpus build typically writes deterministically by source — fall back to "<unknown>"
+        // when filename gives no hint. For accurate per-source counts on real corpora, the
+        // MANIFEST.json route below is preferred.
+        for (const f of sampled) {
+            const inferred = inferSourceFromFilename(f);
+            splitMap[inferred] = (splitMap[inferred] ?? 0) + 1;
+        }
+        // Scale to estimated full-shard counts.
+        const scale = files.length / Math.max(sampled.length, 1);
+        for (const k of Object.keys(splitMap)) {
+            splitMap[k] = Math.round(splitMap[k] * scale);
+        }
+        stats.bySplit[split] = splitMap;
+        stats.totalShards += files.length;
+    }
+    return stats;
+}
+function inferSourceFromFilename(filename) {
+    // Many corpus builds write part-<source>-<n>.parquet or part-<n>.parquet. The latter (current
+    // build at corpus-v0.3.0) gives no source signal in the filename — see manifestScan() for the
+    // authoritative path. Return "<unknown>" so the caller flags this case.
+    const m = basename(filename).match(/part-([\w-]+)-\d+\.parquet$/);
+    if (m && m[1] !== undefined)
+        return m[1];
+    return "<unknown>";
+}
+/**
+ * Known source name prefixes. Corpus-v0.3.0 uses these as `source_id` prefixes; matching against the longest prefix
+ * that fits a given `first_source_id` recovers the canonical source name.
+ *
+ * Order matters: longer prefixes must be tried first so `usgov-nad-...` matches `usgov-nad` rather than `usgov`. Sorted
+ * descending by length at use site.
+ */
+const KNOWN_SOURCE_PREFIXES = [
+    "wof-admin",
+    "wof-postalcode",
+    "ban",
+    "tiger",
+    "usgov-nad",
+    "usgov-nppes",
+    "usgov-hrsa-fqhc",
+    "usgov-imls-pls",
+    "state-ia-contractors",
+    "state-tx-notaries",
+    "state-ny-notaries",
+    "openaddresses",
+    // Synthetic adversarial sources (corpus-v0.4.0+, Thread B).
+    "deepseek-kryptonite",
+    "deepseek-translit-cyrl",
+    "deepseek-translit-jpan",
+    "deepseek-translit-hans",
+    "deepseek-translit-hang",
+    "deepseek-translit-armn",
+];
+/** Extract the source-name prefix from a `first_source_id` value. */
+function sourceFromID(sourceID, knownPrefixes) {
+    // Sort longest-first so usgov-nad beats usgov, wof-admin beats wof.
+    const sorted = [...knownPrefixes].sort((a, b) => b.length - a.length);
+    for (const prefix of sorted) {
+        if (sourceID.startsWith(prefix + "-") || sourceID === prefix)
+            return prefix;
+    }
+    return "<unknown>";
+}
+/**
+ * Prefer reading MANIFEST.json when present — uses each shard's `first_source_id` + prefix matching to recover the
+ * source name. Falls back to scanShards when MANIFEST is absent.
+ *
+ * NOTE: corpus-v0.3.0 shards can mix sources (see `last_source_id` differing from `first_source_id`). The first-row
+ * source is an approximation; reading the parquet's full source column would be authoritative but requires a parquet
+ * dep. For audit purposes the first-row approximation is accurate within ~5% for the corpus-v0.3.0 shape (most shards
+ * are >95% one source).
+ */
+function manifestScan(corpusDir, knownPrefixes) {
+    const manifestPath = join(corpusDir, "MANIFEST.json");
+    if (!existsSync(manifestPath))
+        return null;
+    const manifest = JSON.parse(readFileSync(manifestPath, "utf8"));
+    if (!Array.isArray(manifest.shards))
+        return null;
+    const bySplit = {};
+    for (const shard of manifest.shards) {
+        const split = shard.split;
+        const src = shard.source ?? sourceFromID(shard.first_source_id ?? "", knownPrefixes);
+        bySplit[split] ??= {};
+        bySplit[split][src] = (bySplit[split][src] ?? 0) + 1;
+    }
+    const total = Object.values(bySplit).reduce((sum, m) => sum + Object.values(m).reduce((a, b) => a + b, 0), 0);
+    return { bySplit, totalShards: total, totalFiles: total };
+}
+function buildAuditRows(stats, weights) {
+    const totalShards = Object.values(stats).reduce((a, b) => a + b, 0);
+    const allSources = new Set([...Object.keys(stats), ...Object.keys(weights)]);
+    const rows = [];
+    // Compute effective sample weight: shard_count × source_weight. Sources with no weight get the
+    // "—" marker (loader skips them).
+    const sampleWeights = [];
+    for (const src of allSources) {
+        const shards = stats[src] ?? 0;
+        const weight = weights[src];
+        const effective = weight !== undefined ? shards * weight : 0;
+        sampleWeights.push([src, effective]);
+    }
+    const totalSampleWeight = sampleWeights.reduce((a, [, w]) => a + w, 0);
+    for (const src of allSources) {
+        const shards = stats[src] ?? 0;
+        const weight = weights[src] ?? "—";
+        const effective = typeof weight === "number" ? (shards * weight) / Math.max(totalSampleWeight, 1) : "—";
+        rows.push({
+            source: src,
+            shards,
+            shardPct: totalShards > 0 ? shards / totalShards : 0,
+            weight,
+            effectiveSamplePct: typeof effective === "number" ? effective : "—",
+        });
+    }
+    // Flag the dominator: empirically calibrated against the v0.3.0 → v0.4.0 retrospective.
+    // v0.3.0 had usgov-nad at 52% effective sample (1.9× ban); the resulting label-space dilution
+    // was responsible for the coarse-F1 regression. So flag a source as "concentration warning"
+    // when it's above 40% effective sample OR more than 1.5× the next-highest.
+    const numeric = rows.filter((r) => typeof r.effectiveSamplePct === "number");
+    numeric.sort((a, b) => b.effectiveSamplePct - a.effectiveSamplePct);
+    if (numeric.length >= 1) {
+        const top = numeric[0];
+        const next = numeric[1]?.effectiveSamplePct ?? 0;
+        if (top.effectiveSamplePct > 0.4 || (next > 0 && top.effectiveSamplePct / next > 1.5)) {
+            top.overweightFactor = next > 0 ? top.effectiveSamplePct / next : Infinity;
+        }
+    }
+    rows.sort((a, b) => b.shards - a.shards);
+    return rows;
+}
+function formatPct(v) {
+    if (v === "—")
+        return "—";
+    return `${(v * 100).toFixed(1)}%`;
+}
+function printReport(corpusDir, configPath, stats, rows) {
+    console.log(`\nCorpus audit — ${corpusDir}`);
+    if (configPath) {
+        console.log(`Config:        ${configPath}`);
+    }
+    console.log(`Total shards:  ${stats.totalShards}${stats.totalFiles !== stats.totalShards ? ` (${stats.totalFiles} files on disk)` : ""}`);
+    console.log("");
+    const trainStats = stats.bySplit["train"];
+    if (trainStats) {
+        const total = Object.values(trainStats).reduce((a, b) => a + b, 0);
+        console.log(`Train split: ${total} shards`);
+        console.log("");
+        const headers = ["source", "shards", "shard %", "weight", "eff. sample %"];
+        const widths = [22, 8, 10, 8, 14];
+        const fmtRow = (cells) => cells.map((c, i) => c.padEnd(widths[i])).join("  ");
+        console.log(fmtRow(headers));
+        console.log(fmtRow(widths.map((w) => "─".repeat(w))));
+        for (const row of rows) {
+            console.log(fmtRow([
+                row.source,
+                String(row.shards),
+                formatPct(row.shardPct),
+                typeof row.weight === "number" ? row.weight.toFixed(2) : "—",
+                formatPct(row.effectiveSamplePct),
+            ]));
+        }
+        console.log("");
+        const dominator = rows.find((r) => r.overweightFactor !== undefined);
+        if (dominator) {
+            const factor = dominator.overweightFactor;
+            const factorStr = factor === Infinity ? "∞" : factor?.toFixed(1);
+            console.error(`⚠ Concentration: ${dominator.source} would sample ${formatPct(dominator.effectiveSamplePct)} ` +
+                `of training rows (${factorStr}× the next-highest). ` +
+                `Past lesson: v0.3.0's NAD at ~52% caused the 21-label coarse regression. ` +
+                `Consider lowering this source's weight or boosting others.`);
+        }
+        else {
+            console.log("✓ No single-source concentration (top source < 40% effective sample AND < 1.5× next).");
+        }
+        const missingWeights = rows.filter((r) => r.weight === "—" && r.shards > 0);
+        if (missingWeights.length > 0 && configPath) {
+            console.error(`⚠ Sources present in corpus but absent from config.source_weights ` +
+                `(loader will skip them): ${missingWeights.map((r) => r.source).join(", ")}`);
+        }
+        const orphanWeights = rows.filter((r) => typeof r.weight === "number" && r.shards === 0);
+        if (orphanWeights.length > 0) {
+            console.error(`⚠ Sources weighted in config but no shards found in corpus ` +
+                `(no-op weights): ${orphanWeights.map((r) => r.source).join(", ")}`);
+        }
+    }
+}
+export function audit(opts) {
+    const config = opts.configPath ? parseConfig(opts.configPath) : null;
+    // Compose the known-prefix list from both the hardcoded set and any extra names in the config
+    // (forward-compat for future adapters added before this file is updated).
+    const prefixes = [...new Set([...KNOWN_SOURCE_PREFIXES, ...Object.keys(config?.sourceWeights ?? {})])];
+    const stats = manifestScan(opts.corpusDir, prefixes) ?? scanShards(opts.corpusDir, opts.sampleShardCount ?? 100);
+    const trainStats = stats.bySplit["train"] ?? {};
+    const rows = buildAuditRows(trainStats, config?.sourceWeights ?? {});
+    printReport(opts.corpusDir, opts.configPath, stats, rows);
+}
+//# sourceMappingURL=audit.js.map