@mailwoman/corpus 8.6.0 → 9.0.0

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

@@ -1 +1 @@
-{"version":3,"file":"shard-kryptonite.js","sourceRoot":"","sources":["../../../src/tools/shard-kryptonite.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AAEH,OAAO,EAAE,UAAU,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,SAAS,CAAA;AACjE,OAAO,EAAE,KAAK,EAAE,MAAM,kBAAkB,CAAA;AACxC,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAA;AAEhC,OAAO,EAAE,YAAY,EAAE,MAAM,uBAAuB,CAAA;AAEpD,OAAO,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAA;AAEtC,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,iBAAiB,EAAE,WAAW,EAAE,MAAM,eAAe,CAAA;AAiB/F,KAAK,SAAS,CAAC,CAAC,aAAa,CAAC,KAAa,EAAE,aAAqB;IACjE,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,YAAY,CAA0B,KAAK,CAAC,EAAE,CAAC;QACtE,oFAAoF;QACpF,MAAM,UAAU,GAAG,GAAG,CAAC,YAAY,CAA2B,CAAA;QAE9D,MAAM;YACL,GAAG,EAAE,GAAG,CAAC,KAAK,CAAW;YACzB,UAAU;YACV,OAAO,EAAG,GAAG,CAAC,SAAS,CAAY,IAAI,IAAI;YAC3C,MAAM,EAAG,GAAG,CAAC,QAAQ,CAAY,IAAI,SAAS;YAC9C,MAAM,EAAG,GAAG,CAAC,QAAQ,CAAY,IAAI,qBAAqB;YAC1D,SAAS,EAAE,GAAG,CAAC,WAAW,CAAW;YACrC,cAAc,EAAE,aAAa;YAC7B,OAAO,EAAG,GAAG,CAAC,SAAS,CAAY,IAAI,gDAAgD;YACvF,KAAK,EAAE,GAAG,CAAC,OAAO,CAA0B;SAC5C,CAAA;IACF,CAAC;AACF,CAAC;AAED,KAAK,SAAS,CAAC,CAAC,WAAW,CAAC,KAAa,EAAE,aAAqB,EAAE,aAAuB;IACxF,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,aAAa,CAAC,KAAK,EAAE,aAAa,CAAC,EAAE,CAAC;QAC7D,MAAM,MAAM,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAA;QAE5B,IAAI,MAAM,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;YAC/B,MAAM,MAAM,CAAC,GAAG,CAAA;QACjB,CAAC;aAAM,CAAC;YACP,aAAa,CAAC,IAAI,CAAC,GAAG,GAAG,CAAC,SAAS,KAAK,MAAM,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC,CAAA;QAC7D,CAAC;IACF,CAAC;AACF,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,oBAAoB,CACzC,OAA+B,EAC/B,MAA+B;IAE/B,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,OAAO,CAAA;IACtD,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,IAAI,qBAAqB,CAAA;IAEtD,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,KAAK,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,oBAAoB,OAAO,CAAC,KAAK,EAAE,CAAC,CAAA;IAEpF,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,YAAY,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,4BAA4B,OAAO,CAAC,YAAY,EAAE,CAAC,CAAA;IAE1G,MAAM,KAAK,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;IAEhD,MAAM,UAAU,GAAa,EAAE,CAAA;IAE/B,MAAM,WAAW,GAAG,MAAM,WAAW,CACpC,EAAE,KAAK,EAAE,WAAW,CAAC,OAAO,CAAC,KAAK,EAAE,aAAa,EAAE,UAAU,CAAC,EAAE,EAChE,EAAE,SAAS,EAAE,OAAO,CAAC,MAAM,EAAE,aAAa,EAAE,CAC5C,CAAA;IAED,MAAM,EAAE,CACP,SAAS,WAAW,CAAC,UAAU,cAAc,WAAW,CAAC,MAAM,CAAC,MAAM,aAAa;QAClF,eAAe,UAAU,CAAC,MAAM,EAAE,CACnC,CAAA;IAED,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC;QACvB,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,WAAW,aAAa,EAAE,EAAE,2BAA2B,CAAC,CAAA;QAC3F,aAAa,CAAC,KAAK,EAAE,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,MAAM,CAAC,CAAA;QAC1D,MAAM,EAAE,CAAC,oBAAoB,KAAK,EAAE,CAAC,CAAA;IACtC,CAAC;IAED,mFAAmF;IACnF,sFAAsF;IACtF,yFAAyF;IACzF,KAAK,MAAM,EAAE,IAAI,WAAW,CAAC,MAAM,EAAE,CAAC;QACrC,CAAC;QAAC,EAAoC,CAAC,MAAM,GAAG,MAAM,CAAA;IACvD,CAAC;IAED,sFAAsF;IACtF,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,OAAO,CAAC,YAAY,EAAE,MAAM,CAAC,CAAkB,CAAA;IAEpF,MAAM,QAAQ,GAAkB;QAC/B,cAAc,EAAE,aAAa;QAC7B,MAAM,EAAE,eAAe;QACvB,cAAc,EAAE,IAAI,CAAC,cAAc;QACnC,cAAc,EAAE,IAAI,CAAC,cAAc,IAAI,cAAc;QACrD,MAAM,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,GAAG,WAAW,CAAC,MAAM,CAAC;QAC/C,MAAM,EAAE;YACP,KAAK,EAAE,IAAI,CAAC,MAAM,CAAC,KAAK,GAAG,CAAC,WAAW,CAAC,MAAM,CAAC,KAAK,IAAI,CAAC,CAAC;YAC1D,GAAG,EAAE,IAAI,CAAC,MAAM,CAAC,GAAG;YACpB,IAAI,EAAE,IAAI,CAAC,MAAM,CAAC,IAAI;SACtB;QACD,UAAU,EAAE,IAAI,CAAC,UAAU,GAAG,WAAW,CAAC,UAAU;KACpD,CAAA;IAED,oFAAoF;IACpF,kFAAkF;IAClF,uFAAuF;IACvF,MAAM,YAAY,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,WAAW,aAAa,EAAE,EAAE,eAAe,CAAC,CAAA;IACtF,aAAa,CAAC,YAAY,EAAE,IAAI,CAAC,SAAS,CAAC,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAC,GAAG,IAAI,EAAE,MAAM,CAAC,CAAA;IAC7E,MAAM,EAAE,CAAC,6BAA6B,YAAY,EAAE,CAAC,CAAA;IACrD,MAAM,EAAE,CAAC,gBAAgB,QAAQ,CAAC,UAAU,UAAU,IAAI,CAAC,UAAU,WAAW,WAAW,CAAC,UAAU,GAAG,CAAC,CAAA;IAC1G,MAAM,EAAE,CAAC,YAAY,QAAQ,CAAC,MAAM,CAAC,MAAM,UAAU,IAAI,CAAC,MAAM,CAAC,MAAM,WAAW,WAAW,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAA;IAC/G,MAAM,EAAE,CAAC,iBAAiB,iBAAiB,EAAE,CAAC,CAAA;AAC/C,CAAC"}
+{"version":3,"file":"shard-kryptonite.js","sourceRoot":"","sources":["../../../src/tools/shard-kryptonite.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AAEH,OAAO,EAAE,UAAU,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,SAAS,CAAA;AACjE,OAAO,EAAE,KAAK,EAAE,MAAM,kBAAkB,CAAA;AACxC,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAA;AAEhC,OAAO,EAAE,eAAe,EAAE,MAAM,yBAAyB,CAAA;AACzD,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAE7C,OAAO,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAA;AAEtC,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,iBAAiB,EAAE,WAAW,EAAE,MAAM,eAAe,CAAA;AAiB/F,KAAK,SAAS,CAAC,CAAC,aAAa,CAAC,KAAa,EAAE,aAAqB;IACjE,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,eAAe,CAAC,SAAS,CAA0B,KAAK,CAAC,EAAE,CAAC;QACnF,oFAAoF;QACpF,MAAM,UAAU,GAAG,GAAG,CAAC,YAAY,CAA2B,CAAA;QAE9D,MAAM;YACL,GAAG,EAAE,GAAG,CAAC,KAAK,CAAW;YACzB,UAAU;YACV,OAAO,EAAG,GAAG,CAAC,SAAS,CAAY,IAAI,IAAI;YAC3C,MAAM,EAAG,GAAG,CAAC,QAAQ,CAAY,IAAI,SAAS;YAC9C,MAAM,EAAG,GAAG,CAAC,QAAQ,CAAY,IAAI,qBAAqB;YAC1D,SAAS,EAAE,GAAG,CAAC,WAAW,CAAW;YACrC,cAAc,EAAE,aAAa;YAC7B,OAAO,EAAG,GAAG,CAAC,SAAS,CAAY,IAAI,gDAAgD;YACvF,KAAK,EAAE,GAAG,CAAC,OAAO,CAA0B;SAC5C,CAAA;IACF,CAAC;AACF,CAAC;AAED,KAAK,SAAS,CAAC,CAAC,WAAW,CAAC,KAAa,EAAE,aAAqB,EAAE,aAAuB;IACxF,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,aAAa,CAAC,KAAK,EAAE,aAAa,CAAC,EAAE,CAAC;QAC7D,MAAM,MAAM,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAA;QAE5B,IAAI,MAAM,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;YAC/B,MAAM,MAAM,CAAC,GAAG,CAAA;QACjB,CAAC;aAAM,CAAC;YACP,aAAa,CAAC,IAAI,CAAC,GAAG,GAAG,CAAC,SAAS,KAAK,MAAM,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC,CAAA;QAC7D,CAAC;IACF,CAAC;AACF,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,oBAAoB,CACzC,OAA+B,EAC/B,MAA+B;IAE/B,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,OAAO,CAAA;IACtD,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,IAAI,qBAAqB,CAAA;IAEtD,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,KAAK,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,oBAAoB,OAAO,CAAC,KAAK,EAAE,CAAC,CAAA;IAEpF,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,YAAY,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,4BAA4B,OAAO,CAAC,YAAY,EAAE,CAAC,CAAA;IAE1G,MAAM,KAAK,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;IAEhD,MAAM,UAAU,GAAa,EAAE,CAAA;IAE/B,MAAM,WAAW,GAAG,MAAM,WAAW,CACpC,EAAE,KAAK,EAAE,WAAW,CAAC,OAAO,CAAC,KAAK,EAAE,aAAa,EAAE,UAAU,CAAC,EAAE,EAChE,EAAE,SAAS,EAAE,OAAO,CAAC,MAAM,EAAE,aAAa,EAAE,CAC5C,CAAA;IAED,MAAM,EAAE,CACP,SAAS,WAAW,CAAC,UAAU,cAAc,WAAW,CAAC,MAAM,CAAC,MAAM,aAAa;QAClF,eAAe,UAAU,CAAC,MAAM,EAAE,CACnC,CAAA;IAED,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC;QACvB,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,WAAW,aAAa,EAAE,EAAE,2BAA2B,CAAC,CAAA;QAC3F,aAAa,CAAC,KAAK,EAAE,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,MAAM,CAAC,CAAA;QAC1D,MAAM,EAAE,CAAC,oBAAoB,KAAK,EAAE,CAAC,CAAA;IACtC,CAAC;IAED,mFAAmF;IACnF,sFAAsF;IACtF,yFAAyF;IACzF,KAAK,MAAM,EAAE,IAAI,WAAW,CAAC,MAAM,EAAE,CAAC;QACrC,CAAC;QAAC,EAAoC,CAAC,MAAM,GAAG,MAAM,CAAA;IACvD,CAAC;IAED,sFAAsF;IACtF,MAAM,IAAI,GAAG,eAAe,CAAgB,YAAY,CAAC,OAAO,CAAC,YAAY,EAAE,MAAM,CAAC,CAAC,CAAA;IAEvF,MAAM,QAAQ,GAAkB;QAC/B,cAAc,EAAE,aAAa;QAC7B,MAAM,EAAE,eAAe;QACvB,cAAc,EAAE,IAAI,CAAC,cAAc;QACnC,cAAc,EAAE,IAAI,CAAC,cAAc,IAAI,cAAc;QACrD,MAAM,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,GAAG,WAAW,CAAC,MAAM,CAAC;QAC/C,MAAM,EAAE;YACP,KAAK,EAAE,IAAI,CAAC,MAAM,CAAC,KAAK,GAAG,CAAC,WAAW,CAAC,MAAM,CAAC,KAAK,IAAI,CAAC,CAAC;YAC1D,GAAG,EAAE,IAAI,CAAC,MAAM,CAAC,GAAG;YACpB,IAAI,EAAE,IAAI,CAAC,MAAM,CAAC,IAAI;SACtB;QACD,UAAU,EAAE,IAAI,CAAC,UAAU,GAAG,WAAW,CAAC,UAAU;KACpD,CAAA;IAED,oFAAoF;IACpF,kFAAkF;IAClF,uFAAuF;IACvF,MAAM,YAAY,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,WAAW,aAAa,EAAE,EAAE,eAAe,CAAC,CAAA;IACtF,aAAa,CAAC,YAAY,EAAE,IAAI,CAAC,SAAS,CAAC,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAC,GAAG,IAAI,EAAE,MAAM,CAAC,CAAA;IAC7E,MAAM,EAAE,CAAC,6BAA6B,YAAY,EAAE,CAAC,CAAA;IACrD,MAAM,EAAE,CAAC,gBAAgB,QAAQ,CAAC,UAAU,UAAU,IAAI,CAAC,UAAU,WAAW,WAAW,CAAC,UAAU,GAAG,CAAC,CAAA;IAC1G,MAAM,EAAE,CAAC,YAAY,QAAQ,CAAC,MAAM,CAAC,MAAM,UAAU,IAAI,CAAC,MAAM,CAAC,MAAM,WAAW,WAAW,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAA;IAC/G,MAAM,EAAE,CAAC,iBAAiB,iBAAiB,EAAE,CAAC,CAAA;AAC/C,CAAC"}

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

@@ -16,10 +16,10 @@
  *   - Each shard is written to `train/part-translit-<slug>.parquet` (distinct from kryptonite's
  *       `part-0000.parquet`, which v0.4.0's first builder already produced).
  *   - Inherits the path-canonicalization fix flagged in Thread B's postmortem: v0.3.0 shard paths are
- *       rewritten from `/mnt/playpen/mailwoman-data/...` to `/data/...` in the combined MANIFEST so
+ *       rewritten from `$MAILWOMAN_DATA_ROOT/...` to `/data/...` in the combined MANIFEST so
  *       all paths share one container-friendly form.
  *
- *   See docs/articles/plan/reference/CORPUS_V0_4_0_GENERATION.md for prompts, model, and the
+ *   See docs/engineering/reference/CORPUS_V0_4_0_GENERATION.md for prompts, model, and the
  *   reproducibility contract.
  *
  *   Invoke via `mailwoman corpus shard translit \
@@ -40,7 +40,10 @@ export interface ShardTranslitOptions {
      */
     canonicalPathPrefix?: string;
     /**
-     * Default `"/mnt/playpen/mailwoman-data/"`.
+     * Prefix the base manifest's shard paths currently carry, to be rewritten to
+     * {@link ShardTranslitOptions.canonicalPathPrefix}. Defaults to `mailwomanDataRoot()` with a trailing slash — the
+     * root that WROTE those paths. Pass it explicitly when translating a manifest generated under a different
+     * `$MAILWOMAN_DATA_ROOT` than the one you are running with.
      */
     legacyPathPrefix?: string;
 }

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

@@ -1 +1 @@
-{"version":3,"file":"shard-translit.d.ts","sourceRoot":"","sources":["../../../src/tools/shard-translit.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AAcH,MAAM,WAAW,oBAAoB;IACpC,KAAK,EAAE,MAAM,CAAA;IACb,YAAY,EAAE,MAAM,CAAA;IACpB,MAAM,EAAE,MAAM,CAAA;IACd;;OAEG;IACH,aAAa,CAAC,EAAE,MAAM,CAAA;IACtB;;OAEG;IACH,mBAAmB,CAAC,EAAE,MAAM,CAAA;IAC5B;;OAEG;IACH,gBAAgB,CAAC,EAAE,MAAM,CAAA;CACzB;AAsGD,wBAAsB,kBAAkB,CACvC,OAAO,EAAE,oBAAoB,EAC7B,MAAM,CAAC,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,GAC7B,OAAO,CAAC,IAAI,CAAC,CA+Ff"}
+{"version":3,"file":"shard-translit.d.ts","sourceRoot":"","sources":["../../../src/tools/shard-translit.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AAgBH,MAAM,WAAW,oBAAoB;IACpC,KAAK,EAAE,MAAM,CAAA;IACb,YAAY,EAAE,MAAM,CAAA;IACpB,MAAM,EAAE,MAAM,CAAA;IACd;;OAEG;IACH,aAAa,CAAC,EAAE,MAAM,CAAA;IACtB;;OAEG;IACH,mBAAmB,CAAC,EAAE,MAAM,CAAA;IAC5B;;;;;OAKG;IACH,gBAAgB,CAAC,EAAE,MAAM,CAAA;CACzB;AAsGD,wBAAsB,kBAAkB,CACvC,OAAO,EAAE,oBAAoB,EAC7B,MAAM,CAAC,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,GAC7B,OAAO,CAAC,IAAI,CAAC,CA+Ff"}

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

@@ -16,10 +16,10 @@
  *   - Each shard is written to `train/part-translit-<slug>.parquet` (distinct from kryptonite's
  *       `part-0000.parquet`, which v0.4.0's first builder already produced).
  *   - Inherits the path-canonicalization fix flagged in Thread B's postmortem: v0.3.0 shard paths are
- *       rewritten from `/mnt/playpen/mailwoman-data/...` to `/data/...` in the combined MANIFEST so
+ *       rewritten from `$MAILWOMAN_DATA_ROOT/...` to `/data/...` in the combined MANIFEST so
  *       all paths share one container-friendly form.
  *
- *   See docs/articles/plan/reference/CORPUS_V0_4_0_GENERATION.md for prompts, model, and the
+ *   See docs/engineering/reference/CORPUS_V0_4_0_GENERATION.md for prompts, model, and the
  *   reproducibility contract.
  *
  *   Invoke via `mailwoman corpus shard translit \
@@ -30,7 +30,9 @@
 import { existsSync, readFileSync, writeFileSync } from "node:fs";
 import { mkdir, stat } from "node:fs/promises";
 import { join } from "node:path";
-import { iterateJSONL, sha256File } from "@mailwoman/core/utils";
+import { parseJSONStrict } from "@mailwoman/core/objects";
+import { mailwomanDataRoot, sha256File } from "@mailwoman/core/utils";
+import { JSONSpliterator } from "spliterator";
 import { alignRow } from "../align.js";
 import { ParquetWriter } from "../parquet-wrapper/index.js";
 import { LABELED_ROW_SCHEMA, PARQUET_COLUMNS, ROW_GROUP_SIZE, rowToParquet, SHARD_COMPRESSION } from "../parquet.js";
@@ -116,7 +118,7 @@ function canonicalizeShardPath(path, legacyPrefix, canonicalPrefix) {
 export async function buildTranslitShard(options, report) {
     const corpusVersion = options.corpusVersion ?? "0.4.0";
     const canonicalPathPrefix = options.canonicalPathPrefix ?? "/data/";
-    const legacyPathPrefix = options.legacyPathPrefix ?? "/mnt/playpen/mailwoman-data/";
+    const legacyPathPrefix = options.legacyPathPrefix ?? `${mailwomanDataRoot()}/`;
     if (!existsSync(options.jsonl))
         throw new Error(`jsonl not found: ${options.jsonl}`);
     if (!existsSync(options.baseManifest))
@@ -128,7 +130,7 @@ export async function buildTranslitShard(options, report) {
     const buckets = new Map();
     const quarantine = [];
     let totalIn = 0;
-    for await (const raw of iterateJSONL(options.jsonl)) {
+    for await (const raw of JSONSpliterator.fromAsync(options.jsonl)) {
         totalIn++;
         const canon = toCanonicalRow(raw, corpusVersion);
         const result = alignRow(canon);
@@ -163,7 +165,7 @@ export async function buildTranslitShard(options, report) {
     // Compose final MANIFEST: rewrite base.shards paths from /mnt/playpen/... → /data/... and append
     // the new translit shards. Kryptonite shard already lives in the base manifest (it was written
     // there by Thread B).
-    const base = JSON.parse(readFileSync(options.baseManifest, "utf8"));
+    const base = parseJSONStrict(readFileSync(options.baseManifest, "utf8"));
     const rewrittenBase = base.shards.map((sh) => ({
         ...sh,
         path: canonicalizeShardPath(sh.path, legacyPathPrefix, canonicalPathPrefix),

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

@@ -1 +1 @@
-{"version":3,"file":"shard-translit.js","sourceRoot":"","sources":["../../../src/tools/shard-translit.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AAEH,OAAO,EAAE,UAAU,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,SAAS,CAAA;AACjE,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,kBAAkB,CAAA;AAC9C,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAA;AAEhC,OAAO,EAAE,YAAY,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAA;AAEhE,OAAO,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAA;AACtC,OAAO,EAAE,aAAa,EAAE,MAAM,6BAA6B,CAAA;AAE3D,OAAO,EAAE,kBAAkB,EAAE,eAAe,EAAE,cAAc,EAAE,YAAY,EAAE,iBAAiB,EAAE,MAAM,eAAe,CAAA;AAqBpH,SAAS,cAAc,CAAC,GAA4B,EAAE,aAAqB;IAC1E,OAAO;QACN,GAAG,EAAE,GAAG,CAAC,KAAK,CAAW;QACzB,UAAU,EAAE,GAAG,CAAC,YAAY,CAA2B;QACvD,OAAO,EAAG,GAAG,CAAC,SAAS,CAAY,IAAI,IAAI;QAC3C,MAAM,EAAG,GAAG,CAAC,QAAQ,CAAY,IAAI,SAAS;QAC9C,MAAM,EAAE,GAAG,CAAC,QAAQ,CAAW;QAC/B,SAAS,EAAE,GAAG,CAAC,WAAW,CAAW;QACrC,cAAc,EAAE,aAAa;QAC7B,OAAO,EAAG,GAAG,CAAC,SAAS,CAAY,IAAI,gDAAgD;QACvF,KAAK,EAAE,GAAG,CAAC,OAAO,CAA0B;KAC5C,CAAA;AACF,CAAC;AAED,SAAS,WAAW,CAAC,GAAe;IACnC,MAAM,GAAG,GAA4B;QACpC,GAAG,EAAE,GAAG,CAAC,GAAG;QACZ,MAAM,EAAE,GAAG,CAAC,MAAM;QAClB,MAAM,EAAE,GAAG,CAAC,MAAM;QAClB,OAAO,EAAE,GAAG,CAAC,OAAO;QACpB,MAAM,EAAE,GAAG,CAAC,MAAM;QAClB,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,cAAc,EAAE,GAAG,CAAC,cAAc;QAClC,OAAO,EAAE,GAAG,CAAC,OAAO;KACpB,CAAA;IAED,IAAI,GAAG,CAAC,MAAM,KAAK,IAAI,EAAE,CAAC;QACzB,GAAG,CAAC,MAAM,GAAG,GAAG,CAAC,MAAM,CAAA;IACxB,CAAC;IAED,IAAI,GAAG,CAAC,YAAY,KAAK,IAAI,EAAE,CAAC;QAC/B,GAAG,CAAC,YAAY,GAAG,GAAG,CAAC,YAAY,CAAA;IACpC,CAAC;IAED,IAAI,GAAG,CAAC,aAAa,KAAK,IAAI,EAAE,CAAC;QAChC,GAAG,CAAC,aAAa,GAAG,GAAG,CAAC,aAAa,CAAA;IACtC,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;;GAGG;AACH,KAAK,UAAU,aAAa,CAC3B,IAA2B,EAC3B,OAAe,EACf,MAAc,EACd,aAAqB;IAErB,MAAM,MAAM,GAAG,MAAM,aAAa,CAAC,QAAQ,CAAa,kBAAkB,EAAE,OAAO,EAAE;QACpF,YAAY,EAAE,cAAc;KAC5B,CAAC,CAAA;IAEF,MAAM,CAAC,WAAW,CAAC,0BAA0B,EAAE,aAAa,CAAC,CAAA;IAC7D,MAAM,CAAC,WAAW,CAAC,iBAAiB,EAAE,OAAO,CAAC,CAAA;IAC9C,MAAM,CAAC,WAAW,CAAC,wBAAwB,EAAE,MAAM,CAAC,CAAA;IAEpD,IAAI,aAAa,GAAG,EAAE,CAAA;IACtB,IAAI,YAAY,GAAG,EAAE,CAAA;IAErB,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACxB,MAAM,EAAE,GAAG,YAAY,CAAC,GAAG,CAAC,CAAA;QAC5B,MAAM,MAAM,CAAC,SAAS,CAAC,WAAW,CAAC,EAAE,CAA0B,CAAC,CAAA;QAEhE,IAAI,aAAa,KAAK,EAAE,EAAE,CAAC;YAC1B,aAAa,GAAG,GAAG,CAAC,SAAS,CAAA;QAC9B,CAAC;QAED,YAAY,GAAG,GAAG,CAAC,SAAS,CAAA;IAC7B,CAAC;IAED,MAAM,MAAM,CAAC,KAAK,EAAE,CAAA;IAEpB,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC,CAAA;IACpC,MAAM,MAAM,GAAG,MAAM,UAAU,CAAC,OAAO,CAAC,CAAA;IAExC,OAAO;QACN,KAAK,EAAE,OAAO;QACd,IAAI,EAAE,OAAO;QACb,MAAM,EAAE,SAAS;QACjB,WAAW,EAAE,iBAAiB;QAC9B,IAAI,EAAE,IAAI,CAAC,MAAM;QACjB,KAAK,EAAE,QAAQ,CAAC,IAAI;QACpB,MAAM;QACN,eAAe,EAAE,aAAa;QAC9B,cAAc,EAAE,YAAY;QAC5B,wFAAwF;QACxF,yEAAyE;QACzE,GAAI,EAAE,MAAM,EAA6B;KACzC,CAAA;AACF,CAAC;AAED,SAAS,qBAAqB,CAAC,IAAY,EAAE,YAAoB,EAAE,eAAuB;IACzF,IAAI,IAAI,CAAC,UAAU,CAAC,YAAY,CAAC;QAAE,OAAO,eAAe,GAAG,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,MAAM,CAAC,CAAA;IAE3F,OAAO,IAAI,CAAA;AACZ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,kBAAkB,CACvC,OAA6B,EAC7B,MAA+B;IAE/B,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,OAAO,CAAA;IACtD,MAAM,mBAAmB,GAAG,OAAO,CAAC,mBAAmB,IAAI,QAAQ,CAAA;IACnE,MAAM,gBAAgB,GAAG,OAAO,CAAC,gBAAgB,IAAI,8BAA8B,CAAA;IAEnF,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,KAAK,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,oBAAoB,OAAO,CAAC,KAAK,EAAE,CAAC,CAAA;IAEpF,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,YAAY,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,4BAA4B,OAAO,CAAC,YAAY,EAAE,CAAC,CAAA;IAE1G,MAAM,SAAS,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,WAAW,aAAa,EAAE,CAAC,CAAA;IAClE,MAAM,QAAQ,GAAG,IAAI,CAAC,SAAS,EAAE,OAAO,CAAC,CAAA;IACzC,MAAM,KAAK,CAAC,QAAQ,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;IAE1C,gEAAgE;IAChE,MAAM,OAAO,GAAG,IAAI,GAAG,EAAwB,CAAA;IAC/C,MAAM,UAAU,GAAa,EAAE,CAAA;IAC/B,IAAI,OAAO,GAAG,CAAC,CAAA;IAEf,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,YAAY,CAA0B,OAAO,CAAC,KAAK,CAAC,EAAE,CAAC;QAC9E,OAAO,EAAE,CAAA;QACT,MAAM,KAAK,GAAG,cAAc,CAAC,GAAG,EAAE,aAAa,CAAC,CAAA;QAChD,MAAM,MAAM,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAA;QAE9B,IAAI,MAAM,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;YAC/B,UAAU,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC,SAAS,KAAK,MAAM,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC,CAAA;YAE3D,SAAQ;QACT,CAAC;QAED,MAAM,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,MAAM,CAAC,CAAA;QAExC,IAAI,MAAM,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,CAAA;QACxB,CAAC;aAAM,CAAC;YACP,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAA;QACxC,CAAC;IACF,CAAC;IAED,MAAM,EAAE,CAAC,QAAQ,OAAO,UAAU,UAAU,CAAC,MAAM,iBAAiB,OAAO,CAAC,IAAI,iBAAiB,CAAC,CAAA;IAElG,MAAM,SAAS,GAAsB,EAAE,CAAA;IACvC,MAAM,UAAU,GAAG,CAAC,GAAG,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,QAAQ,EAAE,CAAA;IAEjD,KAAK,MAAM,MAAM,IAAI,UAAU,EAAE,CAAC;QACjC,MAAM,IAAI,GAAG,OAAO,CAAC,GAAG,CAAC,MAAM,CAAE,CAAA;QACjC,MAAM,IAAI,GAAG,MAAM,CAAC,UAAU,CAAC,oBAAoB,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,oBAAoB,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,MAAM,CAAA;QACzG,MAAM,OAAO,GAAG,IAAI,CAAC,QAAQ,EAAE,iBAAiB,IAAI,UAAU,CAAC,CAAA;QAC/D,MAAM,UAAU,GAAG,MAAM,aAAa,CAAC,IAAI,EAAE,OAAO,EAAE,MAAM,EAAE,aAAa,CAAC,CAAA;QAC5E,SAAS,CAAC,IAAI,CAAC,UAAU,CAAC,CAAA;QAC1B,MAAM,EAAE,CAAC,KAAK,MAAM,KAAK,UAAU,CAAC,IAAI,WAAW,OAAO,KAAK,UAAU,CAAC,KAAK,SAAS,CAAC,CAAA;IAC1F,CAAC;IAED,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC;QACvB,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,EAAE,gCAAgC,CAAC,CAAA;QAC/D,aAAa,CAAC,KAAK,EAAE,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,MAAM,CAAC,CAAA;QAC1D,MAAM,EAAE,CAAC,oBAAoB,KAAK,KAAK,UAAU,CAAC,MAAM,QAAQ,CAAC,CAAA;IAClE,CAAC;IAED,iGAAiG;IACjG,+FAA+F;IAC/F,sBAAsB;IACtB,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,OAAO,CAAC,YAAY,EAAE,MAAM,CAAC,CAAkB,CAAA;IAEpF,MAAM,aAAa,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC;QAC9C,GAAG,EAAE;QACL,IAAI,EAAE,qBAAqB,CAAC,EAAE,CAAC,IAAI,EAAE,gBAAgB,EAAE,mBAAmB,CAAC;KAC3E,CAAC,CAAC,CAAA;IAEH,MAAM,YAAY,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,EAAE,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC,CAAA;IAEpE,MAAM,QAAQ,GAAkB;QAC/B,cAAc,EAAE,aAAa;QAC7B,MAAM,EAAE,eAAe;QACvB,cAAc,EAAE,IAAI,CAAC,cAAc;QACnC,cAAc,EAAE,IAAI,CAAC,cAAc,IAAI,cAAc;QACrD,MAAM,EAAE,CAAC,GAAG,aAAa,EAAE,GAAG,SAAS,CAAC;QACxC,MAAM,EAAE;YACP,KAAK,EAAE,IAAI,CAAC,MAAM,CAAC,KAAK,GAAG,YAAY;YACvC,GAAG,EAAE,IAAI,CAAC,MAAM,CAAC,GAAG;YACpB,IAAI,EAAE,IAAI,CAAC,MAAM,CAAC,IAAI;SACtB;QACD,UAAU,EAAE,IAAI,CAAC,UAAU,GAAG,YAAY;KAC1C,CAAA;IAED,MAAM,YAAY,GAAG,IAAI,CAAC,SAAS,EAAE,eAAe,CAAC,CAAA;IACrD,aAAa,CAAC,YAAY,EAAE,IAAI,CAAC,SAAS,CAAC,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAC,GAAG,IAAI,EAAE,MAAM,CAAC,CAAA;IAC7E,MAAM,EAAE,CAAC,6BAA6B,YAAY,EAAE,CAAC,CAAA;IACrD,MAAM,EAAE,CAAC,gBAAgB,QAAQ,CAAC,UAAU,UAAU,IAAI,CAAC,UAAU,WAAW,YAAY,GAAG,CAAC,CAAA;IAChG,MAAM,EAAE,CAAC,YAAY,QAAQ,CAAC,MAAM,CAAC,MAAM,UAAU,IAAI,CAAC,MAAM,CAAC,MAAM,WAAW,SAAS,CAAC,MAAM,GAAG,CAAC,CAAA;IACtG,MAAM,EAAE,CAAC,iBAAiB,iBAAiB,EAAE,CAAC,CAAA;IAC9C,MAAM,OAAO,GAAG,aAAa,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,IAAI,CAAC,MAAM,CAAC,CAAC,CAAE,CAAC,IAAI,CAAC,CAAC,MAAM,CAAA;IAEtF,IAAI,OAAO,GAAG,CAAC,EAAE,CAAC;QACjB,MAAM,EAAE,CAAC,qCAAqC,OAAO,aAAa,gBAAgB,QAAQ,mBAAmB,IAAI,CAAC,CAAA;IACnH,CAAC;AACF,CAAC"}
+{"version":3,"file":"shard-translit.js","sourceRoot":"","sources":["../../../src/tools/shard-translit.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AAEH,OAAO,EAAE,UAAU,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,SAAS,CAAA;AACjE,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,kBAAkB,CAAA;AAC9C,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAA;AAEhC,OAAO,EAAE,eAAe,EAAE,MAAM,yBAAyB,CAAA;AACzD,OAAO,EAAE,iBAAiB,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAA;AACrE,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAE7C,OAAO,EAAE,QAAQ,EAAE,MAAM,aAAa,CAAA;AACtC,OAAO,EAAE,aAAa,EAAE,MAAM,6BAA6B,CAAA;AAE3D,OAAO,EAAE,kBAAkB,EAAE,eAAe,EAAE,cAAc,EAAE,YAAY,EAAE,iBAAiB,EAAE,MAAM,eAAe,CAAA;AAwBpH,SAAS,cAAc,CAAC,GAA4B,EAAE,aAAqB;IAC1E,OAAO;QACN,GAAG,EAAE,GAAG,CAAC,KAAK,CAAW;QACzB,UAAU,EAAE,GAAG,CAAC,YAAY,CAA2B;QACvD,OAAO,EAAG,GAAG,CAAC,SAAS,CAAY,IAAI,IAAI;QAC3C,MAAM,EAAG,GAAG,CAAC,QAAQ,CAAY,IAAI,SAAS;QAC9C,MAAM,EAAE,GAAG,CAAC,QAAQ,CAAW;QAC/B,SAAS,EAAE,GAAG,CAAC,WAAW,CAAW;QACrC,cAAc,EAAE,aAAa;QAC7B,OAAO,EAAG,GAAG,CAAC,SAAS,CAAY,IAAI,gDAAgD;QACvF,KAAK,EAAE,GAAG,CAAC,OAAO,CAA0B;KAC5C,CAAA;AACF,CAAC;AAED,SAAS,WAAW,CAAC,GAAe;IACnC,MAAM,GAAG,GAA4B;QACpC,GAAG,EAAE,GAAG,CAAC,GAAG;QACZ,MAAM,EAAE,GAAG,CAAC,MAAM;QAClB,MAAM,EAAE,GAAG,CAAC,MAAM;QAClB,OAAO,EAAE,GAAG,CAAC,OAAO;QACpB,MAAM,EAAE,GAAG,CAAC,MAAM;QAClB,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,cAAc,EAAE,GAAG,CAAC,cAAc;QAClC,OAAO,EAAE,GAAG,CAAC,OAAO;KACpB,CAAA;IAED,IAAI,GAAG,CAAC,MAAM,KAAK,IAAI,EAAE,CAAC;QACzB,GAAG,CAAC,MAAM,GAAG,GAAG,CAAC,MAAM,CAAA;IACxB,CAAC;IAED,IAAI,GAAG,CAAC,YAAY,KAAK,IAAI,EAAE,CAAC;QAC/B,GAAG,CAAC,YAAY,GAAG,GAAG,CAAC,YAAY,CAAA;IACpC,CAAC;IAED,IAAI,GAAG,CAAC,aAAa,KAAK,IAAI,EAAE,CAAC;QAChC,GAAG,CAAC,aAAa,GAAG,GAAG,CAAC,aAAa,CAAA;IACtC,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;;GAGG;AACH,KAAK,UAAU,aAAa,CAC3B,IAA2B,EAC3B,OAAe,EACf,MAAc,EACd,aAAqB;IAErB,MAAM,MAAM,GAAG,MAAM,aAAa,CAAC,QAAQ,CAAa,kBAAkB,EAAE,OAAO,EAAE;QACpF,YAAY,EAAE,cAAc;KAC5B,CAAC,CAAA;IAEF,MAAM,CAAC,WAAW,CAAC,0BAA0B,EAAE,aAAa,CAAC,CAAA;IAC7D,MAAM,CAAC,WAAW,CAAC,iBAAiB,EAAE,OAAO,CAAC,CAAA;IAC9C,MAAM,CAAC,WAAW,CAAC,wBAAwB,EAAE,MAAM,CAAC,CAAA;IAEpD,IAAI,aAAa,GAAG,EAAE,CAAA;IACtB,IAAI,YAAY,GAAG,EAAE,CAAA;IAErB,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACxB,MAAM,EAAE,GAAG,YAAY,CAAC,GAAG,CAAC,CAAA;QAC5B,MAAM,MAAM,CAAC,SAAS,CAAC,WAAW,CAAC,EAAE,CAA0B,CAAC,CAAA;QAEhE,IAAI,aAAa,KAAK,EAAE,EAAE,CAAC;YAC1B,aAAa,GAAG,GAAG,CAAC,SAAS,CAAA;QAC9B,CAAC;QAED,YAAY,GAAG,GAAG,CAAC,SAAS,CAAA;IAC7B,CAAC;IAED,MAAM,MAAM,CAAC,KAAK,EAAE,CAAA;IAEpB,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC,CAAA;IACpC,MAAM,MAAM,GAAG,MAAM,UAAU,CAAC,OAAO,CAAC,CAAA;IAExC,OAAO;QACN,KAAK,EAAE,OAAO;QACd,IAAI,EAAE,OAAO;QACb,MAAM,EAAE,SAAS;QACjB,WAAW,EAAE,iBAAiB;QAC9B,IAAI,EAAE,IAAI,CAAC,MAAM;QACjB,KAAK,EAAE,QAAQ,CAAC,IAAI;QACpB,MAAM;QACN,eAAe,EAAE,aAAa;QAC9B,cAAc,EAAE,YAAY;QAC5B,wFAAwF;QACxF,yEAAyE;QACzE,GAAI,EAAE,MAAM,EAA6B;KACzC,CAAA;AACF,CAAC;AAED,SAAS,qBAAqB,CAAC,IAAY,EAAE,YAAoB,EAAE,eAAuB;IACzF,IAAI,IAAI,CAAC,UAAU,CAAC,YAAY,CAAC;QAAE,OAAO,eAAe,GAAG,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,MAAM,CAAC,CAAA;IAE3F,OAAO,IAAI,CAAA;AACZ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,kBAAkB,CACvC,OAA6B,EAC7B,MAA+B;IAE/B,MAAM,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,OAAO,CAAA;IACtD,MAAM,mBAAmB,GAAG,OAAO,CAAC,mBAAmB,IAAI,QAAQ,CAAA;IACnE,MAAM,gBAAgB,GAAG,OAAO,CAAC,gBAAgB,IAAI,GAAG,iBAAiB,EAAE,GAAG,CAAA;IAE9E,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,KAAK,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,oBAAoB,OAAO,CAAC,KAAK,EAAE,CAAC,CAAA;IAEpF,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,YAAY,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,4BAA4B,OAAO,CAAC,YAAY,EAAE,CAAC,CAAA;IAE1G,MAAM,SAAS,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,WAAW,aAAa,EAAE,CAAC,CAAA;IAClE,MAAM,QAAQ,GAAG,IAAI,CAAC,SAAS,EAAE,OAAO,CAAC,CAAA;IACzC,MAAM,KAAK,CAAC,QAAQ,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;IAE1C,gEAAgE;IAChE,MAAM,OAAO,GAAG,IAAI,GAAG,EAAwB,CAAA;IAC/C,MAAM,UAAU,GAAa,EAAE,CAAA;IAC/B,IAAI,OAAO,GAAG,CAAC,CAAA;IAEf,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,eAAe,CAAC,SAAS,CAA0B,OAAO,CAAC,KAAK,CAAC,EAAE,CAAC;QAC3F,OAAO,EAAE,CAAA;QACT,MAAM,KAAK,GAAG,cAAc,CAAC,GAAG,EAAE,aAAa,CAAC,CAAA;QAChD,MAAM,MAAM,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAA;QAE9B,IAAI,MAAM,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;YAC/B,UAAU,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC,SAAS,KAAK,MAAM,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC,CAAA;YAE3D,SAAQ;QACT,CAAC;QAED,MAAM,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,MAAM,CAAC,CAAA;QAExC,IAAI,MAAM,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,CAAA;QACxB,CAAC;aAAM,CAAC;YACP,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAA;QACxC,CAAC;IACF,CAAC;IAED,MAAM,EAAE,CAAC,QAAQ,OAAO,UAAU,UAAU,CAAC,MAAM,iBAAiB,OAAO,CAAC,IAAI,iBAAiB,CAAC,CAAA;IAElG,MAAM,SAAS,GAAsB,EAAE,CAAA;IACvC,MAAM,UAAU,GAAG,CAAC,GAAG,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,QAAQ,EAAE,CAAA;IAEjD,KAAK,MAAM,MAAM,IAAI,UAAU,EAAE,CAAC;QACjC,MAAM,IAAI,GAAG,OAAO,CAAC,GAAG,CAAC,MAAM,CAAE,CAAA;QACjC,MAAM,IAAI,GAAG,MAAM,CAAC,UAAU,CAAC,oBAAoB,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,oBAAoB,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,MAAM,CAAA;QACzG,MAAM,OAAO,GAAG,IAAI,CAAC,QAAQ,EAAE,iBAAiB,IAAI,UAAU,CAAC,CAAA;QAC/D,MAAM,UAAU,GAAG,MAAM,aAAa,CAAC,IAAI,EAAE,OAAO,EAAE,MAAM,EAAE,aAAa,CAAC,CAAA;QAC5E,SAAS,CAAC,IAAI,CAAC,UAAU,CAAC,CAAA;QAC1B,MAAM,EAAE,CAAC,KAAK,MAAM,KAAK,UAAU,CAAC,IAAI,WAAW,OAAO,KAAK,UAAU,CAAC,KAAK,SAAS,CAAC,CAAA;IAC1F,CAAC;IAED,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC;QACvB,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,EAAE,gCAAgC,CAAC,CAAA;QAC/D,aAAa,CAAC,KAAK,EAAE,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,MAAM,CAAC,CAAA;QAC1D,MAAM,EAAE,CAAC,oBAAoB,KAAK,KAAK,UAAU,CAAC,MAAM,QAAQ,CAAC,CAAA;IAClE,CAAC;IAED,iGAAiG;IACjG,+FAA+F;IAC/F,sBAAsB;IACtB,MAAM,IAAI,GAAG,eAAe,CAAgB,YAAY,CAAC,OAAO,CAAC,YAAY,EAAE,MAAM,CAAC,CAAC,CAAA;IAEvF,MAAM,aAAa,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC;QAC9C,GAAG,EAAE;QACL,IAAI,EAAE,qBAAqB,CAAC,EAAE,CAAC,IAAI,EAAE,gBAAgB,EAAE,mBAAmB,CAAC;KAC3E,CAAC,CAAC,CAAA;IAEH,MAAM,YAAY,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,EAAE,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC,CAAA;IAEpE,MAAM,QAAQ,GAAkB;QAC/B,cAAc,EAAE,aAAa;QAC7B,MAAM,EAAE,eAAe;QACvB,cAAc,EAAE,IAAI,CAAC,cAAc;QACnC,cAAc,EAAE,IAAI,CAAC,cAAc,IAAI,cAAc;QACrD,MAAM,EAAE,CAAC,GAAG,aAAa,EAAE,GAAG,SAAS,CAAC;QACxC,MAAM,EAAE;YACP,KAAK,EAAE,IAAI,CAAC,MAAM,CAAC,KAAK,GAAG,YAAY;YACvC,GAAG,EAAE,IAAI,CAAC,MAAM,CAAC,GAAG;YACpB,IAAI,EAAE,IAAI,CAAC,MAAM,CAAC,IAAI;SACtB;QACD,UAAU,EAAE,IAAI,CAAC,UAAU,GAAG,YAAY;KAC1C,CAAA;IAED,MAAM,YAAY,GAAG,IAAI,CAAC,SAAS,EAAE,eAAe,CAAC,CAAA;IACrD,aAAa,CAAC,YAAY,EAAE,IAAI,CAAC,SAAS,CAAC,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAC,GAAG,IAAI,EAAE,MAAM,CAAC,CAAA;IAC7E,MAAM,EAAE,CAAC,6BAA6B,YAAY,EAAE,CAAC,CAAA;IACrD,MAAM,EAAE,CAAC,gBAAgB,QAAQ,CAAC,UAAU,UAAU,IAAI,CAAC,UAAU,WAAW,YAAY,GAAG,CAAC,CAAA;IAChG,MAAM,EAAE,CAAC,YAAY,QAAQ,CAAC,MAAM,CAAC,MAAM,UAAU,IAAI,CAAC,MAAM,CAAC,MAAM,WAAW,SAAS,CAAC,MAAM,GAAG,CAAC,CAAA;IACtG,MAAM,EAAE,CAAC,iBAAiB,iBAAiB,EAAE,CAAC,CAAA;IAC9C,MAAM,OAAO,GAAG,aAAa,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,IAAI,CAAC,MAAM,CAAC,CAAC,CAAE,CAAC,IAAI,CAAC,CAAC,MAAM,CAAA;IAEtF,IAAI,OAAO,GAAG,CAAC,EAAE,CAAC;QACjB,MAAM,EAAE,CAAC,qCAAqC,OAAO,aAAa,gBAAgB,QAAQ,mBAAmB,IAAI,CAAC,CAAA;IACnH,CAAC;AACF,CAAC"}

package/out/src/tools/sub-venue-lexicon.d.ts

@@ -0,0 +1,507 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Build the sub-venue designator lexicon (#35) — the vocabulary a corpus shard and, eventually, the
+ *   span proposer read to recognize `Terminal 5`, `North Terminal`, `Concourse B`, `ターミナル1` as
+ *   venue-INTERIOR structure.
+ *
+ *   Reads the fetch outputs (`mailwoman corpus fetch wikidata-subvenue`, a JSONL of
+ *   `@mailwoman/osm/sdk`'s `SubVenueSourceRow`s per region, and the Overture slice of `poi.db` via
+ *   `overture-subvenue.ts`) and emits one committed JSON table. The shape follows
+ *   `@mailwoman/poi-taxonomy`'s `taxonomy.json` idiom exactly — typed records plus a FLAT phrase array
+ *   keyed back to a record id, which is what makes a longest-match phrase index cheap to build over it.
+ *   {@link SubVenueSurface} is this table's `SynonymEntry`.
+ *
+ *   ── Determinism ──────────────────────────────────────────────────────────────────────────────────
+ *   {@link buildSubVenueLexicon} is a PURE function of its inputs with a stable sort on every array, so
+ *   a regenerate against the same fetch outputs is byte-identical. No timestamp is emitted for the same
+ *   reason `taxonomy.json` carries none — a clock in the artifact makes every regenerate a diff.
+ *   Vintages live in `sources[]`, taken from the fetch manifests.
+ *
+ *   ── The findings this build is SHAPED BY, all measured ───────────────────────────────────────────
+ *
+ *   **1. A feature's `name` is usually the VENUE's name, not a sub-venue phrase.** Measured on the
+ *   Berlin extract (2,060 matched features, 2026-08-04): a `railway=platform` is named `Stendaler
+ *   Straße`, a `railway=station` is named `Bellevue`, an `amenity=university` is named `Hertie School`.
+ *   Across 250,116 named Great Britain features only 6,003 (2.40%) contain a designator token at all.
+ *   So names are NOT harvested wholesale — {@link extractAttestedPhrases} keeps a name only when it
+ *   CONTAINS a known designator surface, which is what makes `Terminal E (Untere Ebene)` evidence and
+ *   `Otto Lilienthal Flughafen Berlin Tegel` not.
+ *
+ *   **2. The identifier lives in `ref`, not in `name`.** Every one of Berlin's 26 `aeroway=gate`
+ *   features is unnamed and carries only a `ref`: `13`, `6`, `0/1`, `14/15`, `16-18`. That means
+ *   `Gate A12` is a RENDERING (`<designator> <ref>`) rather than a string anyone has written down, and a
+ *   shard that wants to generate the designator+identifier form needs the identifier DISTRIBUTION, not
+ *   a list of phrases. {@link IdentifierShape} is that distribution, and it is why the artifact has a
+ *   section for it at all.
+ *
+ *   **3. A matched phrase belongs to the record the PHRASE names, not to the record the ROW carries.**
+ *   Wave 1 attributed every hit to `row.designatorID`, which is the rule that matched the FEATURE. On
+ *   the GB extract that produced `west → platform`, `hall → platform`, `biggin → platform` — 108 of 133
+ *   OSM-derived surfaces had a `phrase` that named a different record than the one they pointed at,
+ *   because a bus stop tagged `public_transport=platform` is named "Village Hall" or "West Kensington".
+ *   {@link extractAttestedPhrases} now takes a phrase → record INDEX and attributes by phrase; the row's
+ *   own designator is kept as `context`, which is exactly the axis a confound board needs (a `hall` seen
+ *   on a platform is a confound; a `hall` seen on a terminal is evidence).
+ *
+ *   ── What `curated: false` means, and how a surface stops being it ────────────────────────────────
+ *   Wikidata gives a CONCEPT NAME per language, not a designator as addressed. Q849706's Spanish label
+ *   is `terminal aeroportuaria`; the addressed form is `Terminal`. Q240854 (`hall`) is `sala` in
+ *   Italian, which names an ordinary room and would fire on half of Italy. So every machine-derived
+ *   surface lands `curated: false`, and {@link SubVenueLexiconTable} consumers that gate parsing MUST
+ *   filter to `curated: true`.
+ *
+ *   A surface becomes curated ONLY by matching a {@link SubVenuePromotion} in
+ *   `sub-venue-promotions.ts` — a per-designator, per-LOCALE decision carrying the census that backs
+ *   it. Promotion is per-locale because the same token is a designator in one language and a disaster
+ *   in another: `hall` is `Halle 2` at Frankfurt and `Village Hall` at 3,205 British bus stops.
+ *
+ *   ── The seed DUPLICATES `neural/venue-structure.ts`, knowingly ───────────────────────────────────
+ *   `@mailwoman/corpus` does not depend on `@mailwoman/neural` (the dependency runs the other way for
+ *   the training path, and pulling onnxruntime into a corpus build to read three string arrays would be
+ *   absurd), so the shipped vocabulary is re-declared below. That is a drift surface and it is stated
+ *   rather than hidden: `sub-venue-lexicon.test.ts` pins the seed's contents literally, so a change in
+ *   either place fails a test rather than passing silently.
+ */
+import { type SubVenuePromotion } from "./sub-venue-promotions.ts";
+/**
+ * This table's own data version. Bump when the source vintages or the build semantics change.
+ *
+ * `0.2.0` — wave 2: per-region attestation, the phrase-attribution fix (finding 3 above), derived head nouns, the
+ * Overture source, and the `promotions[]` receipts section.
+ */
+export declare const SUBVENUE_LEXICON_VERSION = "0.2.0";
+/**
+ * Which side of the containment relation a designator names. Mirrors `@mailwoman/osm/sdk`'s `SubVenueTier`; re-declared
+ * for the same dependency-direction reason as the seed.
+ */
+export declare const LexiconTier: {
+    readonly SubVenue: "subvenue";
+    readonly Venue: "venue";
+};
+export type LexiconTier = (typeof LexiconTier)[keyof typeof LexiconTier];
+/**
+ * One designator record — a venue-interior (or containing-venue) structural noun.
+ */
+export interface SubVenueDesignator {
+    /**
+     * Canonical id, lowercase English. Matches `neural/venue-structure.ts`'s `VENUE_STRUCTURE_DESIGNATORS` wherever the
+     * two overlap.
+     */
+    id: string;
+    tier: LexiconTier;
+    /**
+     * Whether this designator may be preceded by a {@link SubVenueModifier} — the `North Terminal` shape.
+     *
+     * A SUBSET, and the exclusions are load-bearing: `gate` and `building` form ordinary STREET names in exactly this
+     * shape ("East Gate" is a real GB street, "Building Society Place" is a real street), so admitting them turns a
+     * correct street parse into a sub-venue one. Setting this true means claiming no street is named `<modifier> <id>`.
+     * Check before you do.
+     */
+    modifierEligible: boolean;
+    /**
+     * Whether the shipped span proposer already recognizes this designator. `false` means the lexicon proposes it and
+     * nothing consumes it yet.
+     */
+    shipped: boolean;
+    /**
+     * Where the term comes from, one entry per attesting source: `wof:placetype`, `osm:aeroway=terminal`,
+     * `wikidata:Q849706`, `overture:airport_terminal`. Sorted, so a regenerate is stable.
+     */
+    provenance: string[];
+}
+/**
+ * One positional modifier — the `North`/`Upper`/`Main` half of `North Terminal`.
+ */
+export interface SubVenueModifier {
+    id: string;
+    shipped: boolean;
+    provenance: string[];
+}
+/**
+ * One surface form: a phrase, the record it names, and where it was attested.
+ *
+ * This is the table's `SynonymEntry` — the flat array a phrase index is built over.
+ */
+export interface SubVenueSurface {
+    /**
+     * The phrase, lowercased for Latin-script languages and left as written otherwise (case-folding is meaningless for
+     * Japanese, and `toLowerCase` on Turkish `I` is actively wrong).
+     */
+    phrase: string;
+    /**
+     * The {@link SubVenueDesignator.id} or {@link SubVenueModifier.id} this phrase is a surface of.
+     */
+    recordID: string;
+    /**
+     * Which record table `recordID` points into.
+     */
+    recordKind: "designator" | "modifier";
+    /**
+     * BCP-47-ish language subtag as the source wrote it (`en`, `ja`, `zh-Hant`, `pt-BR`), or `und` when the source gave
+     * an untagged default name.
+     */
+    lang: string;
+    /**
+     * ISO 3166-1 alpha-2 of the DATA the phrase was attested in, `""` for vocabulary sources that attest a term's
+     * existence rather than its use anywhere. This is the axis promotion is decided on: `hall` is attested 3,274 times in
+     * `GB` and every promotion of it lives or dies on a per-region census, never a global one.
+     */
+    region: string;
+    /**
+     * `wikidata:label`, `wikidata:alt`, `osm:name`, `osm:name:<lang>`, `overture:name`, `derived:head-noun`, or `seed`.
+     */
+    source: string;
+    /**
+     * Whether a human has approved this surface for parsing use IN ITS REGION. Everything machine-derived starts `false`
+     * and is flipped only by a matching {@link SubVenuePromotion}. A consumer that gates a parse MUST filter on this — see
+     * the module docstring.
+     */
+    curated: boolean;
+    /**
+     * How many source features attested this exact phrase, when the source counts (OSM, Overture). `0` for vocabulary
+     * sources, which attest a term's EXISTENCE rather than its frequency.
+     */
+    observations: number;
+    /**
+     * The rule-assigned designator of the FEATURES that carried this phrase, with a count each — `platform:3205
+     * campus:49` for GB's `hall`. Empty for vocabulary sources.
+     *
+     * This is the confound axis. A `hall` on a `platform` row is a British bus stop named after a village hall; a `hall`
+     * on a `terminal` row is a real German departure hall. Without it, a surface's `observations` count is a magnitude
+     * with no sign — see the repo's "meaning of zero" rule, which applies just as hard to a large number.
+     */
+    context: Record<string, number>;
+}
+/**
+ * The measured shape of a designator's identifier half — what follows `Gate`/`Terminal` in real data.
+ *
+ * Derived from OSM `ref` values, not from names. See the module docstring's finding 2.
+ */
+export interface IdentifierShape {
+    designatorID: string;
+    /**
+     * ISO 3166-1 alpha-2 of the extract this distribution was measured in. Per-region because the shapes differ: GB gates
+     * are 70% bare digits, Japanese platform refs are overwhelmingly bare digits with a different range, and a shard that
+     * generates `Gate <ref>` for a French address should sample France's distribution.
+     */
+    region: string;
+    /**
+     * A coarse class: `digit` (`5`), `letter` (`B`), `letter-digit` (`A12`), `digit-letter` (`2F`), `range` (`16-18`,
+     * `0/1`), or `other`.
+     */
+    shape: string;
+    observations: number;
+    /**
+     * Up to eight real values, sorted, so a shard author can see what the class actually contains.
+     */
+    examples: string[];
+}
+/**
+ * One input source's provenance, copied off its fetch manifest.
+ */
+export interface SubVenueLexiconSource {
+    id: string;
+    origin: string;
+    license: string;
+    retrieved: string;
+    rows: number;
+}
+/**
+ * The committed table.
+ */
+export interface SubVenueLexiconTable {
+    version: string;
+    sources: SubVenueLexiconSource[];
+    designators: SubVenueDesignator[];
+    modifiers: SubVenueModifier[];
+    surfaces: SubVenueSurface[];
+    identifierShapes: IdentifierShape[];
+    /**
+     * Every curation decision taken against this table, promotion AND rejection, each with the census that backs it. A
+     * rejection is as load-bearing as a promotion: it is what stops the next reader re-proposing `hall` for en-GB.
+     */
+    promotions: SubVenuePromotion[];
+}
+/**
+ * The vocabulary that already ships in `neural/venue-structure.ts`, re-declared. See the module docstring for why this
+ * duplication exists.
+ *
+ * `tier` is added here (the shipped list has no such field): the seven WOF placetypes plus `terminal`/`gate` are all
+ * venue-INTERIOR, except `campus` and `building`, which name a whole venue as often as a part of one. They are marked
+ * `subvenue` anyway, because that is the role the span proposer uses them in — `Building 43, Googleplex` is a unit
+ * inside a venue.
+ */
+export declare const SHIPPED_DESIGNATOR_SEED: ReadonlyArray<{
+    id: string;
+    modifierEligible: boolean;
+    provenance: string[];
+}>;
+/**
+ * The shipped positional modifiers, re-declared from `neural/venue-structure.ts`'s `VENUE_STRUCTURE_MODIFIERS`.
+ */
+export declare const SHIPPED_MODIFIER_SEED: readonly string[];
+/**
+ * Designators the lexicon ADDS beyond what ships, each with the source that attests it.
+ *
+ * `platform`, `station` and `airport` come from the OSM extractor's rule table and are the rail/aviation venue-side
+ * vocabulary the corpus line needs. `hall` and `satellite` come from Wikidata concepts and from
+ * `wof-osm-placetype-map.mdx`'s own "plausible additions" note, which lists `hall` explicitly. `pier` joins them in
+ * wave 2 on 282 Overture attestations in the `pier` category plus 162 in the GB extract — the corpus task names `Pier
+ * C` as a target shape, so the record has to exist before a shard can generate it.
+ *
+ * None is `modifierEligible`: that claim needs a confound board per term AND per locale, and `sub-venue-promotions.ts`
+ * is where those live. A promotion marks a SURFACE usable; it does not widen the modifier grammar.
+ */
+export declare const PROPOSED_DESIGNATORS: ReadonlyArray<{
+    id: string;
+    tier: LexiconTier;
+    provenance: string[];
+}>;
+/**
+ * `designatorID` → Wikidata QID, mirroring `fetch/wikidata-subvenue.ts`'s `SUBVENUE_CONCEPTS`. Re-declared here so the
+ * builder stays a pure function over PARSED input rather than reaching into a fetch module for a constant; the test
+ * pins the two against each other.
+ */
+export declare const CONCEPT_QIDS: Readonly<Record<string, string>>;
+/**
+ * Normalize a surface for the table: trim, collapse internal whitespace, and lowercase ONLY when the string is entirely
+ * in a bicameral script. `ターミナルビル` and `航站楼` pass through untouched; `Flughafenterminal` folds.
+ */
+export declare function normalizeSurface(text: string): string;
+/**
+ * Turn the Wikidata designator-label payload into surfaces.
+ *
+ * A row is dropped when its language tag is empty (an untagged literal, which Wikidata occasionally carries), when the
+ * QID maps to no designator in {@link CONCEPT_QIDS}, or when the normalized phrase is empty. Everything that survives
+ * lands `curated: false` — see the module docstring.
+ */
+export declare function surfacesFromWikidata(payload: unknown, conceptQIDs?: Readonly<Record<string, string>>): SubVenueSurface[];
+/**
+ * One row of a harvestable source, as read back off JSONL or out of a layer database.
+ *
+ * SOURCE-NEUTRAL by design, and verified so in wave 2: an Overture Places row from the `airport_terminal` category is
+ * `{ designatorID, name }` and fits unchanged. What did NOT fit was the harvest function's hardcoded `osm:name` source
+ * stamp — see `overture-subvenue.ts`'s docstring. Declared locally so the builder does not import `@mailwoman/osm`
+ * (which `@mailwoman/corpus` does not depend on) just to name a shape it reads from a file.
+ */
+export interface SubVenueHarvestRow {
+    /**
+     * The designator the SOURCE's rule assigned to the FEATURE. Not necessarily the record a matched phrase names — see
+     * the module docstring's finding 3. Carried into {@link SubVenueSurface.context}.
+     */
+    designatorID: string;
+    name?: string | null;
+    ref?: string | null;
+    localizedNames?: Record<string, string>;
+}
+/**
+ * Classify an OSM `ref` into an {@link IdentifierShape} class.
+ *
+ * The classes are the ones Berlin's gates actually produced, plus the two aviation forms the corpus task names
+ * (`Terminal 2F` is digit-letter, `Concourse B` is letter). `range` covers both separators OSM uses for a gate serving
+ * more than one stand: `16-18` and `0/1`.
+ */
+export declare function classifyIdentifier(ref: string): string;
+/**
+ * A phrase → record index, keyed on the normalized phrase. Built by {@link buildSurfaceIndex} from the surfaces present
+ * before the harvest runs, and the reason a matched phrase can be attributed to the record it actually names.
+ */
+export type SurfaceIndex = ReadonlyMap<string, {
+    recordID: string;
+    recordKind: "designator" | "modifier";
+}>;
+/**
+ * Index the surfaces accumulated so far by phrase. First writer wins, so a seed record beats a Wikidata alias that
+ * happens to collide — `terminal` stays the `terminal` designator even though it is also an Italian alias for it.
+ */
+export declare function buildSurfaceIndex(surfaces: readonly SubVenueSurface[]): SurfaceIndex;
+/**
+ * Every known phrase found in `name`, as whole-token runs for spacing scripts and as substrings for non-spacing ones.
+ *
+ * Token-boundary matching for Latin script, not substring: `Nordterminal` is a real German compound in which `terminal`
+ * is a suffix, and a substring test would also fire on `Terminalstraße`. The compound case is a genuine miss and it is
+ * the right miss — admitting suffix matches would fire on every `-hall`/`-gate` compound in Germanic and Nordic street
+ * naming, which is exactly the confound class `Briggate`/`Kirkgate` represents.
+ *
+ * For Han/Kana names that rule finds nothing at all, because the script has no word boundaries: `第1ターミナル` splits into
+ * one token that matches no surface. There the LONGEST known substring is the correct operator, and the compound
+ * objection does not transfer — Japanese has no `-gate` street-name suffix class.
+ */
+export declare function nameContainsSurfaces(name: string, index: SurfaceIndex): string[];
+/**
+ * Options for one harvest pass — which source stamp its surfaces carry and which region they were attested in.
+ *
+ * Both default to the OSM/unknown-region values wave 1 hardcoded, so an existing caller is unchanged.
+ */
+export interface HarvestOptions {
+    /**
+     * Source family: `osm` yields `osm:name` / `osm:name:<lang>`, `overture` yields `overture:name`.
+     */
+    source?: string;
+    /**
+     * ISO 3166-1 alpha-2 of the extract or partition. `""` when unknown.
+     */
+    region?: string;
+}
+/**
+ * Harvest attested phrases and identifier shapes out of a source's rows.
+ *
+ * `index` gates the name harvest — a name contributes only when it CONTAINS a phrase already in the table. That filter
+ * is the whole reason this function is safe to run over raw OSM: see the module docstring's finding 1 for the Berlin
+ * measurement that motivated it. The index also decides ATTRIBUTION (finding 3): a hit is a surface of the record the
+ * PHRASE names, and the row's own designator is recorded as `context`.
+ *
+ * Returns surfaces with real `observations` counts, so the lexicon can rank `terminal` above a phrase attested once.
+ */
+export declare function extractAttestedPhrases(rows: Iterable<SubVenueHarvestRow>, index: SurfaceIndex, options?: HarvestOptions): {
+    surfaces: SubVenueSurface[];
+    identifierShapes: IdentifierShape[];
+};
+/**
+ * Derive the HEAD NOUN of every multi-part surface, so `terminal aeroportuaria` contributes the form anyone actually
+ * writes on an envelope.
+ *
+ * The problem this solves is the whole reason wave 1 shipped 1,014 uncurated surfaces: Wikidata's label for a concept
+ * is the ENCYCLOPAEDIC name (`terminal aeroportuaria`, `letištní terminál`, `havalimanı terminali`), while the
+ * addressed form is the bare head (`Terminal`, `Terminál`, `Terminali`). Nothing can promote the encyclopaedic form, so
+ * the head has to be extracted before the curation pass has anything to decide about.
+ *
+ * Two derivations, because the table holds two kinds of writing:
+ *
+ * - **Latin script — the COGNATE test.** A token is the head when its ASCII fold shares {@link HEAD_NOUN_PREFIX_FLOOR}
+ *   leading characters with the designator's own canonical id. Nothing subtler survived contact with the data: an
+ *   earlier version matched a token against any SINGLE-TOKEN surface of the record, and because Dutch `universiteit` is
+ *   a one-token surface of `campus`, it derived `universitario`, `universitaire`, `üniversite` and twenty more as head
+ *   nouns of `campus`. Those are the MODIFIER half of the label, and admitting them would have taught the harvest to
+ *   read "Ciudad Universitaria" as sub-venue structure.
+ * - **Non-Latin script — the SHARED-SUBSTRING test.** The cognate test cannot reach a script the id is not written in,
+ *   and for Han and Kana a token split finds nothing at all. So every substring of length ≥
+ *   {@link NON_LATIN_HEAD_MIN_LENGTH} occurring in at least two DISTINCT surfaces of the same record and primary
+ *   language becomes a candidate, ranked by how many surfaces carry it. Japanese yields `ターミナル` (in all five `ja`
+ *   terminal labels) ahead of `ターミナルビル` (three); Chinese yields `航站`, `航站楼`, `航站樓`. Where the script DOES space its
+ *   words (Korean, Greek, Cyrillic) a candidate must be a whole token, so `공항 터미널` ∩ `공항터미널` gives `터미널` and never a
+ *   fragment.
+ *
+ * The non-Latin branch deliberately emits SEVERAL candidates instead of picking one. Choosing between `航站` and `航站楼`
+ * from Wikidata alone is guesswork; the Japan extract answers it by counting, and the promotion ledger records which
+ * count won. Everything derived lands `curated: false` — the derivation is a hypothesis about what the addressed form
+ * is, and a locale's own data is what confirms or kills it.
+ */
+export declare function deriveHeadNounSurfaces(surfaces: readonly SubVenueSurface[]): SubVenueSurface[];
+/**
+ * Apply the curation decisions to a surface list, IN PLACE on a copy.
+ *
+ * A promotion binds `(designatorID, phrase, locale)`. A surface matches when it names the same record with the same
+ * phrase and its language is the locale's language OR the untagged `und` — the default `name` tag carries no language,
+ * and a German extract's untagged `Halle 2` is German.
+ *
+ * REGION is the subtle half. A surface attested in an extract carries that extract's region and matches only its own
+ * locale. A surface with `region: ""` is region-FREE — a Wikidata label or a derived head noun — and a promotion
+ * reaches it only when no REJECTION exists for the same designator, phrase and language anywhere else. That guard is
+ * not decoration: `pier` is promoted for en-GB and rejected for en-US, and without it the en-GB decision would curate
+ * the region-free English surface and hand `Pier 1 Imports` the promotion en-US was refused. Where no rejection
+ * competes — `terminal` in `es`, `ターミナル` in `ja` — the region-free surface is the whole point, since a language's
+ * designator does not stop at a border.
+ *
+ * Rejections mark nothing themselves. They exist in `promotions[]` as the record of a decision taken, so the next
+ * reader meets en-GB `hall`'s 3,204 bus stops before re-proposing it, not after.
+ */
+export declare function applyPromotions(surfaces: readonly SubVenueSurface[], promotions: readonly SubVenuePromotion[]): SubVenueSurface[];
+/**
+ * One harvestable input: rows plus the stamp they carry into the table.
+ */
+export interface SubVenueHarvest {
+    rows: readonly SubVenueHarvestRow[];
+    source?: string;
+    region?: string;
+}
+/**
+ * Everything {@link buildSubVenueLexicon} needs, already parsed. Keeping the builder off the filesystem is what makes it
+ * deterministic and testable without fixtures on disk.
+ */
+export interface BuildSubVenueLexiconInput {
+    /**
+     * The raw `designator-labels.json` SPARQL envelope, or `null` to build the seed-only table.
+     */
+    wikidata: unknown | null;
+    /**
+     * Every harvestable source, in the order they should contribute. Order matters only for the surface INDEX: a source
+     * can match a phrase an earlier source introduced, never a later one.
+     */
+    harvests: readonly SubVenueHarvest[];
+    /**
+     * Provenance rows, copied off the fetch manifests by the caller.
+     */
+    sources: readonly SubVenueLexiconSource[];
+    /**
+     * Curation decisions. Defaults to the committed {@link SUBVENUE_PROMOTIONS}; pass an empty array to build the
+     * pre-curation table (which is what the promotion census itself is taken against).
+     */
+    promotions?: readonly SubVenuePromotion[];
+}
+/**
+ * Build the lexicon table. PURE and deterministic — same inputs, byte-identical output.
+ *
+ * Order of operations is load-bearing in three places:
+ *
+ * 1. Seed surfaces are inserted before anything else, so `terminal` indexes to the `terminal` designator rather than to
+ *    whichever Wikidata alias sorts first.
+ * 2. Head nouns are derived AFTER Wikidata and BEFORE the harvests, because `ターミナル` has to exist as a surface before a
+ *    Japanese extract can be searched for it. That ordering is the entire reason the Japan harvest finds anything — see
+ *    `PROVENANCE.md`.
+ * 3. Promotions are applied LAST, over the union, so a decision can promote a surface whichever source produced it.
+ */
+export declare function buildSubVenueLexicon(input: BuildSubVenueLexiconInput): SubVenueLexiconTable;
+/**
+ * Serialize the table the way the committed artifact stores it: pretty-printed, trailing newline. Run `oxfmt` over the
+ * result before committing — repo law is that committed JSON is oxfmt-clean, which `JSON.stringify` cannot reproduce.
+ */
+export declare function serializeSubVenueLexicon(table: SubVenueLexiconTable): string;
+/**
+ * Read a JSONL file of {@link SubVenueHarvestRow}s. Blank lines and unparseable rows are skipped rather than fatal — an
+ * extract is a build output, and one malformed line should not cost the whole lexicon.
+ */
+export declare function readSubVenueJSONL(path: string): SubVenueHarvestRow[];
+/**
+ * One OSM extract to harvest: the JSONL path plus the ISO country its rows describe.
+ */
+export interface SubVenueExtractInput {
+    path: string;
+    region: string;
+}
+export interface GenerateSubVenueLexiconOptions {
+    /**
+     * Directory holding the `mailwoman corpus fetch wikidata-subvenue` output. Omit to build the seed-only table.
+     */
+    wikidataDir?: string;
+    /**
+     * OSM extract JSONLs, one per region.
+     */
+    extracts?: readonly SubVenueExtractInput[];
+    /**
+     * Already-read Overture rows (`readOvertureSubVenues`), grouped by the caller. Kept as a parameter rather than a path
+     * so this function stays free of a 3.9 GB database dependency — the CLI opens `poi.db`, this assembles.
+     */
+    overtureRows?: readonly (SubVenueHarvestRow & {
+        country: string;
+    })[];
+    /**
+     * `poi.db`'s layer vintage, for `sources[]`. Only read when `overtureRows` is non-empty.
+     */
+    overtureVintage?: string;
+    /**
+     * Where the table is written.
+     */
+    outPath: string;
+}
+/**
+ * Read the fetch outputs, build the table, and write it.
+ *
+ * The IO half only — every decision lives in {@link buildSubVenueLexicon}, which is pure. Run `oxfmt` over `outPath`
+ * afterwards; repo law is that committed JSON is oxfmt-clean.
+ */
+export declare function generateSubVenueLexicon(options: GenerateSubVenueLexiconOptions): SubVenueLexiconTable;
+//# sourceMappingURL=sub-venue-lexicon.d.ts.map