@mailwoman/corpus 9.2.0 → 9.3.0

package/out/adapters/wof-postalcode-json/adapter.d.ts

@@ -0,0 +1,65 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `wof-postalcode`: Who's On First postalcode GeoJSON-bundle adapter.
+ *
+ *   **Phase 1.5.1 pivot.** Replaces the previous SpatiaLite-backed implementation (formerly at
+ *   `packages/corpus/lib/adapters/wof-postalcode/`, removed in this same change). The rationale is
+ *   in `wof-admin-json/adapter.ts` and in `DECISIONS.md` — short version: the SQLite distribution
+ *   mirror is dead, the live distro tags every postcode row `mz:is_current = -1` which the old
+ *   `is_current = 1` predicate excluded, and localized `name:*` variants don't ship in the SQLite
+ *   export at all.
+ *
+ *   Input: a directory containing one or more cloned `whosonfirst-data-postalcode-<cc>` repos plus
+ *   the relevant `whosonfirst-data-admin-<cc>` repos (postcode records reference admin ancestry by
+ *   `wof:parent_id`, so the locality / region / country records must be in the same walk for the
+ *   ancestry chain to resolve). The corpus pipeline clones all four repos under
+ *   `/data/corpus/sources/wof/repos/` and points the adapter at that root.
+ *
+ *   Per live postalcode record, the adapter emits one row per `(name-variant, hierarchy-variant)`
+ *   pair:
+ *
+ *   - **Name variants**: canonical `wof:name` (slot key `default`, typically the postcode digits
+ *       themselves) plus any `name:*` variants on the postcode feature. In practice WOF postcode
+ *       records rarely carry localized name variants, so this expansion is usually a no-op — but
+ *       the code path stays symmetric with the admin adapter for consistency.
+ *   - **Hierarchy variants** (unchanged from the SQLite adapter): self, +locality, +locality+region,
+ *       +locality+region+country.
+ *
+ *   `source_id` is `wof-postalcode-<wof_id>-<name-slot>-<hierarchy-variant>`. Ancestor names always
+ *   come from the ancestor's canonical `wof:name`; this adapter does NOT iterate ancestor name
+ *   variants (e.g. it does not emit `"75008 Париж"` even when Paris has a `name:rus_x_preferred`).
+ *   That cross-product belongs to a future synthesis pass; emitting it here would multiply row
+ *   counts ~10× without a clear training-value story.
+ *
+ *   License: CC0.
+ */
+import { type WOFVariantSpec } from "#adapters/wof-json-rows";
+import type { CorpusAdapter } from "#types";
+import { type WOFRecord } from "#utils";
+/**
+ * Compute hierarchy variants for a postcode record. `selfName` is the postcode surface form (canonical `wof:name` for
+ * the `default` slot, a `name:*` localized variant otherwise).
+ */
+export declare function postcodeVariantsFor(row: WOFRecord, ancestry: WOFRecord[], selfName: string): WOFVariantSpec[];
+/**
+ * Build the per-record name-slot list. The `default` slot uses `wof:name` verbatim (postcode digits); subsequent slots
+ * come from `name:*` variants dedup'd against the default.
+ */
+export declare function nameSlotsFor(rec: WOFRecord): Array<{
+    key: string;
+    value: string;
+}>;
+/**
+ * Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset
+ * it came from.
+ */
+export declare const WOF_POSTALCODE_ADAPTER_ID = "wof-postalcode";
+export declare function createWOFPostalcodeAdapter(): CorpusAdapter;
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export declare const wofPostalcodeAdapter: CorpusAdapter;
+//# sourceMappingURL=adapter.d.ts.map

package/out/adapters/wof-postalcode-json/adapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../lib/adapters/wof-postalcode-json/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAqCG;AAKH,OAAO,EAKN,KAAK,cAAc,EACnB,MAAM,yBAAyB,CAAA;AAChC,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,QAAQ,CAAA;AACzE,OAAO,EAAoC,KAAK,SAAS,EAAE,MAAM,QAAQ,CAAA;AA0BzE;;;GAGG;AACH,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,SAAS,EAAE,QAAQ,EAAE,SAAS,EAAE,EAAE,QAAQ,EAAE,MAAM,GAAG,cAAc,EAAE,CAqC7G;AAED;;;GAGG;AACH,wBAAgB,YAAY,CAAC,GAAG,EAAE,SAAS,GAAG,KAAK,CAAC;IAAE,GAAG,EAAE,MAAM,CAAC;IAAC,KAAK,EAAE,MAAM,CAAA;CAAE,CAAC,CAElF;AAED;;;GAGG;AACH,eAAO,MAAM,yBAAyB,mBAAmB,CAAA;AAEzD,wBAAgB,0BAA0B,IAAI,aAAa,CAqC1D;AAED;;GAEG;AACH,eAAO,MAAM,oBAAoB,eAA+B,CAAA"}

package/out/adapters/wof-postalcode-json/adapter.js

@@ -0,0 +1,151 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `wof-postalcode`: Who's On First postalcode GeoJSON-bundle adapter.
+ *
+ *   **Phase 1.5.1 pivot.** Replaces the previous SpatiaLite-backed implementation (formerly at
+ *   `packages/corpus/lib/adapters/wof-postalcode/`, removed in this same change). The rationale is
+ *   in `wof-admin-json/adapter.ts` and in `DECISIONS.md` — short version: the SQLite distribution
+ *   mirror is dead, the live distro tags every postcode row `mz:is_current = -1` which the old
+ *   `is_current = 1` predicate excluded, and localized `name:*` variants don't ship in the SQLite
+ *   export at all.
+ *
+ *   Input: a directory containing one or more cloned `whosonfirst-data-postalcode-<cc>` repos plus
+ *   the relevant `whosonfirst-data-admin-<cc>` repos (postcode records reference admin ancestry by
+ *   `wof:parent_id`, so the locality / region / country records must be in the same walk for the
+ *   ancestry chain to resolve). The corpus pipeline clones all four repos under
+ *   `/data/corpus/sources/wof/repos/` and points the adapter at that root.
+ *
+ *   Per live postalcode record, the adapter emits one row per `(name-variant, hierarchy-variant)`
+ *   pair:
+ *
+ *   - **Name variants**: canonical `wof:name` (slot key `default`, typically the postcode digits
+ *       themselves) plus any `name:*` variants on the postcode feature. In practice WOF postcode
+ *       records rarely carry localized name variants, so this expansion is usually a no-op — but
+ *       the code path stays symmetric with the admin adapter for consistency.
+ *   - **Hierarchy variants** (unchanged from the SQLite adapter): self, +locality, +locality+region,
+ *       +locality+region+country.
+ *
+ *   `source_id` is `wof-postalcode-<wof_id>-<name-slot>-<hierarchy-variant>`. Ancestor names always
+ *   come from the ancestor's canonical `wof:name`; this adapter does NOT iterate ancestor name
+ *   variants (e.g. it does not emit `"75008 Париж"` even when Paris has a `name:rus_x_preferred`).
+ *   That cross-product belongs to a future synthesis pass; emitting it here would multiply row
+ *   counts ~10× without a clear training-value story.
+ *
+ *   License: CC0.
+ */
+import { COUNTRY_DISPLAY_NAME, emitWOFJSONRows, LOCALE_BY_COUNTRY, nameSlotsFor as wofNameSlotsFor, } from "#adapters/wof-json-rows";
+import { buildAncestryIndex, walkFeatures } from "#utils";
+/**
+ * Map a WOF placetype to a Mailwoman `ComponentTag`, or `undefined` to skip.
+ *
+ * Per-adapter deliberately (the admin adapter carries its own): each table is a record FILTER for its adapter's
+ * emission set — this one keeps `postalcode` plus the ancestry placetypes its variants render — not a shared
+ * vocabulary.
+ */
+function placetypeToTag(placetype) {
+    switch (placetype) {
+        case "country":
+        case "nation":
+            return "country";
+        case "macroregion":
+        case "region":
+            return "region";
+        case "locality":
+            return "locality";
+        case "postalcode":
+            return "postcode";
+        default:
+            return undefined;
+    }
+}
+/**
+ * Compute hierarchy variants for a postcode record. `selfName` is the postcode surface form (canonical `wof:name` for
+ * the `default` slot, a `name:*` localized variant otherwise).
+ */
+export function postcodeVariantsFor(row, ancestry, selfName) {
+    if (placetypeToTag(row.placetype) !== "postcode")
+        return [];
+    const locality = ancestry.find((a) => placetypeToTag(a.placetype) === "locality");
+    const region = ancestry.find((a) => placetypeToTag(a.placetype) === "region");
+    const country = ancestry.find((a) => placetypeToTag(a.placetype) === "country");
+    const countryDisplay = COUNTRY_DISPLAY_NAME[row.country] ?? country?.name ?? row.country;
+    const variants = [{ suffix: "self", components: { postcode: selfName } }];
+    if (locality) {
+        variants.push({
+            suffix: "with-locality",
+            components: { postcode: selfName, locality: locality.name },
+        });
+    }
+    if (locality && region) {
+        variants.push({
+            suffix: "with-locality-region",
+            components: { postcode: selfName, locality: locality.name, region: region.name },
+        });
+    }
+    if (locality && region && country) {
+        variants.push({
+            suffix: "with-locality-region-country",
+            components: {
+                postcode: selfName,
+                locality: locality.name,
+                region: region.name,
+                country: countryDisplay,
+            },
+        });
+    }
+    return variants;
+}
+/**
+ * Build the per-record name-slot list. The `default` slot uses `wof:name` verbatim (postcode digits); subsequent slots
+ * come from `name:*` variants dedup'd against the default.
+ */
+export function nameSlotsFor(rec) {
+    return wofNameSlotsFor(rec);
+}
+/**
+ * Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset
+ * it came from.
+ */
+export const WOF_POSTALCODE_ADAPTER_ID = "wof-postalcode";
+export function createWOFPostalcodeAdapter() {
+    return {
+        id: WOF_POSTALCODE_ADAPTER_ID,
+        defaultLicense: "CC0-1.0",
+        description: "Who's On First postalcode GeoJSON bundles (postcode → locality/region pairs). Ancestor names from sibling admin repos.",
+        async *rows(opts) {
+            // Pass 1: full walk. We keep every record whose placetype maps to a ComponentTag — the
+            // postcode adapter needs locality / region / country admin records in the index so it
+            // can resolve postcode ancestry, even though it only emits rows for postcode records.
+            const byID = new Map();
+            for await (const rec of walkFeatures(opts.inputPath, { signal: opts.signal })) {
+                if (opts.signal?.aborted)
+                    return;
+                if (opts.country && rec.country !== opts.country)
+                    continue;
+                if (!placetypeToTag(rec.placetype))
+                    continue;
+                byID.set(rec.id, rec);
+            }
+            const ancestry = buildAncestryIndex(byID);
+            // Pass 2: emit postcode rows only, sorted by id for determinism.
+            yield* emitWOFJSONRows({
+                records: byID,
+                ancestry,
+                adapterOptions: opts,
+                adapterID: WOF_POSTALCODE_ADAPTER_ID,
+                localeByCountry: LOCALE_BY_COUNTRY,
+                shouldEmit: (record) => placetypeToTag(record.placetype) === "postcode",
+                nameSlotsFor,
+                variantsFor: postcodeVariantsFor,
+            });
+        },
+    };
+}
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export const wofPostalcodeAdapter = createWOFPostalcodeAdapter();
+//# sourceMappingURL=adapter.js.map

package/out/adapters/wof-postalcode-json/adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../lib/adapters/wof-postalcode-json/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAqCG;AAKH,OAAO,EACN,oBAAoB,EACpB,eAAe,EACf,iBAAiB,EACjB,YAAY,IAAI,eAAe,GAE/B,MAAM,yBAAyB,CAAA;AAEhC,OAAO,EAAE,kBAAkB,EAAE,YAAY,EAAkB,MAAM,QAAQ,CAAA;AAEzE;;;;;;GAMG;AACH,SAAS,cAAc,CAAC,SAAwC;IAC/D,QAAQ,SAAS,EAAE,CAAC;QACnB,KAAK,SAAS,CAAC;QACf,KAAK,QAAQ;YACZ,OAAO,SAAS,CAAA;QACjB,KAAK,aAAa,CAAC;QACnB,KAAK,QAAQ;YACZ,OAAO,QAAQ,CAAA;QAChB,KAAK,UAAU;YACd,OAAO,UAAU,CAAA;QAClB,KAAK,YAAY;YAChB,OAAO,UAAU,CAAA;QAClB;YACC,OAAO,SAAS,CAAA;IAClB,CAAC;AACF,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,mBAAmB,CAAC,GAAc,EAAE,QAAqB,EAAE,QAAgB;IAC1F,IAAI,cAAc,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,UAAU;QAAE,OAAO,EAAE,CAAA;IAE3D,MAAM,QAAQ,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,cAAc,CAAC,CAAC,CAAC,SAAS,CAAC,KAAK,UAAU,CAAC,CAAA;IACjF,MAAM,MAAM,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,cAAc,CAAC,CAAC,CAAC,SAAS,CAAC,KAAK,QAAQ,CAAC,CAAA;IAC7E,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,cAAc,CAAC,CAAC,CAAC,SAAS,CAAC,KAAK,SAAS,CAAC,CAAA;IAC/E,MAAM,cAAc,GAAG,oBAAoB,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,OAAO,EAAE,IAAI,IAAI,GAAG,CAAC,OAAO,CAAA;IAExF,MAAM,QAAQ,GAAqB,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,EAAE,QAAQ,EAAE,QAAQ,EAAE,EAAE,CAAC,CAAA;IAE3F,IAAI,QAAQ,EAAE,CAAC;QACd,QAAQ,CAAC,IAAI,CAAC;YACb,MAAM,EAAE,eAAe;YACvB,UAAU,EAAE,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,CAAC,IAAI,EAAE;SAC3D,CAAC,CAAA;IACH,CAAC;IAED,IAAI,QAAQ,IAAI,MAAM,EAAE,CAAC;QACxB,QAAQ,CAAC,IAAI,CAAC;YACb,MAAM,EAAE,sBAAsB;YAC9B,UAAU,EAAE,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,CAAC,IAAI,EAAE,MAAM,EAAE,MAAM,CAAC,IAAI,EAAE;SAChF,CAAC,CAAA;IACH,CAAC;IAED,IAAI,QAAQ,IAAI,MAAM,IAAI,OAAO,EAAE,CAAC;QACnC,QAAQ,CAAC,IAAI,CAAC;YACb,MAAM,EAAE,8BAA8B;YACtC,UAAU,EAAE;gBACX,QAAQ,EAAE,QAAQ;gBAClB,QAAQ,EAAE,QAAQ,CAAC,IAAI;gBACvB,MAAM,EAAE,MAAM,CAAC,IAAI;gBACnB,OAAO,EAAE,cAAc;aACvB;SACD,CAAC,CAAA;IACH,CAAC;IAED,OAAO,QAAQ,CAAA;AAChB,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,YAAY,CAAC,GAAc;IAC1C,OAAO,eAAe,CAAC,GAAG,CAAC,CAAA;AAC5B,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,MAAM,yBAAyB,GAAG,gBAAgB,CAAA;AAEzD,MAAM,UAAU,0BAA0B;IACzC,OAAO;QACN,EAAE,EAAE,yBAAyB;QAC7B,cAAc,EAAE,SAAS;QACzB,WAAW,EACV,wHAAwH;QAEzH,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,uFAAuF;YACvF,sFAAsF;YACtF,sFAAsF;YACtF,MAAM,IAAI,GAAG,IAAI,GAAG,EAAqB,CAAA;YAEzC,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,YAAY,CAAC,IAAI,CAAC,SAAS,EAAE,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,CAAC;gBAC/E,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;oBAAE,OAAM;gBAEhC,IAAI,IAAI,CAAC,OAAO,IAAI,GAAG,CAAC,OAAO,KAAK,IAAI,CAAC,OAAO;oBAAE,SAAQ;gBAE1D,IAAI,CAAC,cAAc,CAAC,GAAG,CAAC,SAAS,CAAC;oBAAE,SAAQ;gBAC5C,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,EAAE,GAAG,CAAC,CAAA;YACtB,CAAC;YAED,MAAM,QAAQ,GAAG,kBAAkB,CAAC,IAAI,CAAC,CAAA;YAEzC,iEAAiE;YACjE,KAAK,CAAC,CAAC,eAAe,CAAC;gBACtB,OAAO,EAAE,IAAI;gBACb,QAAQ;gBACR,cAAc,EAAE,IAAI;gBACpB,SAAS,EAAE,yBAAyB;gBACpC,eAAe,EAAE,iBAAiB;gBAClC,UAAU,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,cAAc,CAAC,MAAM,CAAC,SAAS,CAAC,KAAK,UAAU;gBACvE,YAAY;gBACZ,WAAW,EAAE,mBAAmB;aAChC,CAAC,CAAA;QACH,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAG,0BAA0B,EAAE,CAAA"}

package/out/build.d.ts

@@ -0,0 +1,132 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   End-to-end corpus build (Phase 1 task #10 in the plan).
+ *
+ *   `buildCorpus(opts)` orchestrates every stage of the pipeline:
+ *
+ *   1. **Adapter runs** — drives every adapter in turn (via `runAdapter`), writing
+ *        `<intermediate>/<adapter.id>/canonical.jsonl` slices.
+ *   2. **Synthesis** — optional. For each canonical row, every applicable augmentation in the row's
+ *        country-default policy emits an augmented row alongside the original.
+ *   3. **Alignment** — every row (original + augmented) is aligned via `alignRow`. Successes go to
+ *        `labeled.jsonl`; quarantines are appended to `quarantine.jsonl` with reasons.
+ *   4. **Splits** — `splitRows` partitions labeled `source_id`s into train/val/test by locality holdout.
+ *        Manifest written to `splits/SPLIT_MANIFEST.json` + per-split `train.txt` / `val.txt` /
+ *        `test.txt`.
+ *   5. **Parquet slices** — `writeSlices` streams labeled rows into 1M-row `.parquet` slices per split
+ *        under `corpus-v<version>/{train,val,test}/part-NNNN.parquet` (SNAPPY-compressed, 50k-row
+ *        row groups), with per-slice checksums + per-stage manifest in
+ *        `corpus-v<version>/MANIFEST.json`.
+ *   6. **Top-level manifest** — `<outputDir>/MANIFEST.json` ties every per-stage manifest together with
+ *        a top-level corpus_version, built_at, and aggregate counts.
+ *
+ *   Output layout:
+ *
+ *   ```
+ *   <outputDir>/
+ *   MANIFEST.json
+ *   intermediate/
+ *     <adapter.id>/canonical.jsonl   # one per adapter
+ *     labeled.jsonl                  # post-alignment, pre-slice
+ *     quarantine.jsonl               # rows that failed alignment
+ *   splits/
+ *     SPLIT_MANIFEST.json
+ *     train.txt / val.txt / test.txt
+ *   corpus-v<version>/
+ *     MANIFEST.json
+ *     train/part-NNNN.parquet
+ *     val/part-NNNN.parquet
+ *     test/part-NNNN.parquet
+ * ```
+ *
+ *   The intermediate files live alongside the final slices for reproducibility + debugging. Operators
+ *   can `rm -rf intermediate/` after the build if disk is tight; the final `corpus-v<version>/` is
+ *   self-contained.
+ */
+import { type AdapterRunManifest } from "#runner";
+import type { AdapterOptions, CorpusAdapter } from "#types";
+import { type SliceManifest } from "#utils/parquet";
+import { type SplitManifest } from "#utils/split";
+/**
+ * Stage tags surfaced to `onProgress`.
+ */
+export type BuildStage = "adapter-run" | "align" | "split" | "slice" | "manifest";
+/**
+ * Per-invocation options for `buildCorpus`.
+ */
+export interface BuildCorpusOptions {
+    /**
+     * Root output directory. All build artifacts land beneath it.
+     */
+    outputDir: string;
+    /**
+     * Corpus version (e.g. `"0.1.0"`). Stamped onto every row + into the output dir name.
+     */
+    corpusVersion: string;
+    /**
+     * Adapters to drive, in order. Defaults to `defaultAdapterRegistry.list()`. Pass an explicit list to filter (e.g.
+     * `[wofAdminAdapter]` for a smoke run).
+     */
+    adapters?: readonly CorpusAdapter[];
+    /**
+     * Per-adapter `AdapterOptions` — looked up by adapter id. Adapters whose id is missing from this map are skipped (and
+     * noted in the manifest).
+     */
+    adapterInputs: Record<string, AdapterOptions>;
+    /**
+     * Enable synthesis pass. Default `true`. Set `false` for fixture-driven smoke tests.
+     */
+    synthesize?: boolean;
+    /**
+     * Forwarded to `writeSlices`. Default 1_000_000.
+     */
+    rowsPerSlice?: number;
+    /**
+     * Progress hook. Errors thrown abort the build.
+     */
+    onProgress?: (stage: BuildStage, message: string) => void;
+    /**
+     * License kinds to PURPOSELY exclude from this build (#26). Compiled patterns (see `compileLicenseExcludes` /
+     * `SHARE_ALIKE_PATTERN` in `license.ts`); a row whose `license` matches any is dropped at ingest. Default (omitted)
+     * includes EVERYTHING — exclusion is a deliberate act, not a silent default. A proprietary-weights build passes the
+     * share-alike set (`--exclude-share-alike`).
+     */
+    excludeLicenses?: readonly RegExp[];
+}
+/**
+ * Top-level manifest tying every stage together.
+ */
+export interface BuildCorpusManifest {
+    corpus_version: string;
+    built_at: string;
+    adapters: AdapterRunManifest[];
+    skipped_adapters: string[];
+    splits: {
+        counts: SplitManifest["counts"];
+        holdouts: SplitManifest["holdouts"];
+    };
+    slices: {
+        counts: SliceManifest["counts"];
+        total_rows: number;
+    };
+    quarantine_count: number;
+    total_aligned_rows: number;
+    /**
+     * Resolved license set across all INCLUDED rows (license string → row count), + the count dropped by
+     * `excludeLicenses` (#26). The model card derives its data-attribution table from `licenses`.
+     */
+    licenses: Record<string, number>;
+    excluded_by_license: number;
+}
+/**
+ * Drive the full corpus build to completion.
+ *
+ * Memory profile: the function maintains an in-memory `Map<source_id, SplitName>` to bridge the align → slice hand-off.
+ * For Phase 1 fixture-scale runs (≤ 10⁴ rows) this is trivial. For real 5M+ runs, the map fits comfortably in a few
+ * hundred MB; the canonical.jsonl and labeled.jsonl payloads stream and never sit in memory.
+ */
+export declare function buildCorpus(opts: BuildCorpusOptions): Promise<BuildCorpusManifest>;
+//# sourceMappingURL=build.d.ts.map

package/out/build.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"build.d.ts","sourceRoot":"","sources":["../lib/build.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA+CG;AAUH,OAAO,EAAoB,KAAK,kBAAkB,EAAE,MAAM,SAAS,CAAA;AAEnE,OAAO,KAAK,EAAE,cAAc,EAAgB,aAAa,EAAc,MAAM,QAAQ,CAAA;AAGrF,OAAO,EAAe,KAAK,aAAa,EAAE,MAAM,gBAAgB,CAAA;AAChE,OAAO,EAIN,KAAK,aAAa,EAElB,MAAM,cAAc,CAAA;AAErB;;GAEG;AACH,MAAM,MAAM,UAAU,GAAG,aAAa,GAAG,OAAO,GAAG,OAAO,GAAG,OAAO,GAAG,UAAU,CAAA;AAEjF;;GAEG;AACH,MAAM,WAAW,kBAAkB;IAClC;;OAEG;IACH,SAAS,EAAE,MAAM,CAAA;IAEjB;;OAEG;IACH,aAAa,EAAE,MAAM,CAAA;IAErB;;;OAGG;IACH,QAAQ,CAAC,EAAE,SAAS,aAAa,EAAE,CAAA;IAEnC;;;OAGG;IACH,aAAa,EAAE,MAAM,CAAC,MAAM,EAAE,cAAc,CAAC,CAAA;IAE7C;;OAEG;IACH,UAAU,CAAC,EAAE,OAAO,CAAA;IAEpB;;OAEG;IACH,YAAY,CAAC,EAAE,MAAM,CAAA;IAErB;;OAEG;IACH,UAAU,CAAC,EAAE,CAAC,KAAK,EAAE,UAAU,EAAE,OAAO,EAAE,MAAM,KAAK,IAAI,CAAA;IAEzD;;;;;OAKG;IACH,eAAe,CAAC,EAAE,SAAS,MAAM,EAAE,CAAA;CACnC;AAED;;GAEG;AACH,MAAM,WAAW,mBAAmB;IACnC,cAAc,EAAE,MAAM,CAAA;IACtB,QAAQ,EAAE,MAAM,CAAA;IAChB,QAAQ,EAAE,kBAAkB,EAAE,CAAA;IAC9B,gBAAgB,EAAE,MAAM,EAAE,CAAA;IAC1B,MAAM,EAAE;QAAE,MAAM,EAAE,aAAa,CAAC,QAAQ,CAAC,CAAC;QAAC,QAAQ,EAAE,aAAa,CAAC,UAAU,CAAC,CAAA;KAAE,CAAA;IAChF,MAAM,EAAE;QAAE,MAAM,EAAE,aAAa,CAAC,QAAQ,CAAC,CAAC;QAAC,UAAU,EAAE,MAAM,CAAA;KAAE,CAAA;IAC/D,gBAAgB,EAAE,MAAM,CAAA;IACxB,kBAAkB,EAAE,MAAM,CAAA;IAC1B;;;OAGG;IACH,QAAQ,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IAChC,mBAAmB,EAAE,MAAM,CAAA;CAC3B;AAED;;;;;;GAMG;AACH,wBAAsB,WAAW,CAAC,IAAI,EAAE,kBAAkB,GAAG,OAAO,CAAC,mBAAmB,CAAC,CAyNxF"}

package/out/build.js

@@ -0,0 +1,245 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   End-to-end corpus build (Phase 1 task #10 in the plan).
+ *
+ *   `buildCorpus(opts)` orchestrates every stage of the pipeline:
+ *
+ *   1. **Adapter runs** — drives every adapter in turn (via `runAdapter`), writing
+ *        `<intermediate>/<adapter.id>/canonical.jsonl` slices.
+ *   2. **Synthesis** — optional. For each canonical row, every applicable augmentation in the row's
+ *        country-default policy emits an augmented row alongside the original.
+ *   3. **Alignment** — every row (original + augmented) is aligned via `alignRow`. Successes go to
+ *        `labeled.jsonl`; quarantines are appended to `quarantine.jsonl` with reasons.
+ *   4. **Splits** — `splitRows` partitions labeled `source_id`s into train/val/test by locality holdout.
+ *        Manifest written to `splits/SPLIT_MANIFEST.json` + per-split `train.txt` / `val.txt` /
+ *        `test.txt`.
+ *   5. **Parquet slices** — `writeSlices` streams labeled rows into 1M-row `.parquet` slices per split
+ *        under `corpus-v<version>/{train,val,test}/part-NNNN.parquet` (SNAPPY-compressed, 50k-row
+ *        row groups), with per-slice checksums + per-stage manifest in
+ *        `corpus-v<version>/MANIFEST.json`.
+ *   6. **Top-level manifest** — `<outputDir>/MANIFEST.json` ties every per-stage manifest together with
+ *        a top-level corpus_version, built_at, and aggregate counts.
+ *
+ *   Output layout:
+ *
+ *   ```
+ *   <outputDir>/
+ *   MANIFEST.json
+ *   intermediate/
+ *     <adapter.id>/canonical.jsonl   # one per adapter
+ *     labeled.jsonl                  # post-alignment, pre-slice
+ *     quarantine.jsonl               # rows that failed alignment
+ *   splits/
+ *     SPLIT_MANIFEST.json
+ *     train.txt / val.txt / test.txt
+ *   corpus-v<version>/
+ *     MANIFEST.json
+ *     train/part-NNNN.parquet
+ *     val/part-NNNN.parquet
+ *     test/part-NNNN.parquet
+ * ```
+ *
+ *   The intermediate files live alongside the final slices for reproducibility + debugging. Operators
+ *   can `rm -rf intermediate/` after the build if disk is tight; the final `corpus-v<version>/` is
+ *   self-contained.
+ */
+import { pathExists, readLocalJSONFile } from "@mailwoman/core/fs/readers";
+import { openWriteStream } from "@mailwoman/core/fs/streams";
+import { writeLocalJSONFile, makeDirectories } from "@mailwoman/core/fs/writers";
+import { join } from "path-ts";
+import { JSONSpliterator } from "spliterator";
+import { defaultAdapterRegistry } from "#adapters/utils";
+import { $public } from "#env";
+import { once, runAdapter } from "#runner";
+import { defaultAugmentationsForCountry, synthesizeRow } from "#synthesizers/utils";
+import { alignRow } from "#utils/align";
+import { licenseExcluded } from "#utils/license";
+import { writeSlices } from "#utils/parquet";
+import { defaultHoldouts, splitForRow, writeSplitManifestsFromLabeledFiles, } from "#utils/split";
+/**
+ * Drive the full corpus build to completion.
+ *
+ * Memory profile: the function maintains an in-memory `Map<source_id, SplitName>` to bridge the align → slice hand-off.
+ * For Phase 1 fixture-scale runs (≤ 10⁴ rows) this is trivial. For real 5M+ runs, the map fits comfortably in a few
+ * hundred MB; the canonical.jsonl and labeled.jsonl payloads stream and never sit in memory.
+ */
+export async function buildCorpus(opts) {
+    const adapters = opts.adapters ?? defaultAdapterRegistry.list();
+    const synthesize = opts.synthesize ?? true;
+    const rowsPerSlice = opts.rowsPerSlice ?? 1_000_000;
+    const built_at = new Date().toISOString();
+    await makeDirectories(opts.outputDir);
+    const intermediateDir = join(opts.outputDir, "intermediate");
+    await makeDirectories(intermediateDir);
+    // 1. Adapter runs.
+    const adapterRuns = [];
+    const skipped = [];
+    for (const adapter of adapters) {
+        const adapterOptions = opts.adapterInputs[adapter.id];
+        if (!adapterOptions) {
+            skipped.push(adapter.id);
+            opts.onProgress?.("adapter-run", `skipped ${adapter.id} (no input configured)`);
+            continue;
+        }
+        // Opt-in resume (MAILWOMAN_RESUME=1): if a complete per-adapter canonical.jsonl + MANIFEST.json
+        // already exist, reuse them instead of re-emitting. The MANIFEST is written only after the
+        // canonical is fully flushed, so its presence guarantees completeness; row order is identical,
+        // so downstream holdout-split determinism is preserved. Recovers an align-phase crash without
+        // redoing the (expensive) emit phase. Default (unset) re-emits, preserving correctness. (2026-06-12.)
+        const adapterDir = join(intermediateDir, adapter.id);
+        const cachedManifest = join(adapterDir, "MANIFEST.json");
+        if ($public.MAILWOMAN_RESUME === "1" &&
+            (await pathExists(cachedManifest)) &&
+            (await pathExists(join(adapterDir, "canonical.jsonl")))) {
+            const cached = await readLocalJSONFile(cachedManifest);
+            opts.onProgress?.("adapter-run", `resumed ${adapter.id} (reused ${cached.yielded} canonical rows)`);
+            adapterRuns.push(cached);
+            continue;
+        }
+        opts.onProgress?.("adapter-run", `running ${adapter.id}`);
+        const m = await runAdapter({
+            adapter,
+            adapterOptions,
+            outputDir: intermediateDir,
+            corpusVersion: opts.corpusVersion,
+        });
+        adapterRuns.push(m);
+    }
+    // 2 + 3. Synthesis + alignment: stream every canonical.jsonl, optionally augment, align,
+    // and route each labeled row directly to its split-specific JSONL (`labeled-{train,val,test}.
+    // jsonl`). Memory cost is O(1) — the prior in-memory `splitInputs` array + `splitByIDMap`
+    // + `SplitManifest.{train,val,test}` arrays are gone; per-row split is decided inline via
+    // `splitForRow` (a pure function of source_id + region + holdout policy).
+    const labeledPaths = {
+        train: join(intermediateDir, "labeled-train.jsonl"),
+        val: join(intermediateDir, "labeled-val.jsonl"),
+        test: join(intermediateDir, "labeled-test.jsonl"),
+    };
+    const labeledStreams = {
+        train: openWriteStream(labeledPaths.train, { encoding: "utf8" }),
+        val: openWriteStream(labeledPaths.val, { encoding: "utf8" }),
+        test: openWriteStream(labeledPaths.test, { encoding: "utf8" }),
+    };
+    const quarantinePath = join(intermediateDir, "quarantine.jsonl");
+    const quarantineStream = openWriteStream(quarantinePath, { encoding: "utf8" });
+    let aligned = 0;
+    let quarantined = 0;
+    const counts = { train: 0, val: 0, test: 0 };
+    const holdouts = defaultHoldouts();
+    // License accounting + the deliberate exclusion filter (#26). `licenseCounts` is the resolved
+    // license set (→ manifest + model-card attribution); `excludeLicenses` (empty by default → include
+    // everything) is the operator's PURPOSEFUL exclusion, never a silent drop.
+    const excludeLicenses = opts.excludeLicenses ?? [];
+    const licenseCounts = new Map();
+    let excludedByLicense = 0;
+    const writeQuarantine = (row, reason) => {
+        quarantineStream.write(`${JSON.stringify({ row, reason })}\n`);
+    };
+    for (const adapterRun of adapterRuns) {
+        opts.onProgress?.("align", `aligning ${adapterRun.adapter_id}`);
+        for await (const row of streamJSONL(adapterRun.jsonl_path)) {
+            licenseCounts.set(row.license, (licenseCounts.get(row.license) ?? 0) + 1);
+            // Deliberate license exclusion (#26): drop a row ONLY when the operator named its license
+            // kind via `excludeLicenses`. Default (no patterns) keeps everything — exclusion is a
+            // purposeful act, not a silent default. Counted BEFORE the drop so the manifest's license
+            // set reflects what the corpus actually CONTAINED, and `excluded_by_license` what was removed.
+            if (licenseExcluded(row.license, excludeLicenses)) {
+                excludedByLicense++;
+                continue;
+            }
+            const fanned = [row];
+            if (synthesize) {
+                for (const aug of synthesizeRow(row, defaultAugmentationsForCountry(row.country))) {
+                    fanned.push(aug);
+                }
+            }
+            for (const r of fanned) {
+                let result;
+                try {
+                    result = alignRow(r);
+                }
+                catch (error) {
+                    // Last-resort robustness (2026-06-12): no single row may crash a multi-hour build.
+                    // alignRow's targeted paths normalize/quarantine known issues with specific reasons;
+                    // this catches any UNKNOWN throw (e.g. assertSpanInvariants on an unforeseen span
+                    // shape) → quarantine + continue. A spike in `align-threw` reasons is a finding.
+                    writeQuarantine(r, `align-threw:${error.message.slice(0, 160)}`);
+                    quarantined++;
+                    continue;
+                }
+                if (result.kind === "labeled") {
+                    const split = splitForRow(result.row, holdouts);
+                    labeledStreams[split].write(`${JSON.stringify(result.row)}\n`);
+                    counts[split]++;
+                    aligned++;
+                }
+                else {
+                    writeQuarantine(r, result.row.reason);
+                    quarantined++;
+                }
+            }
+        }
+    }
+    for (const s of Object.values(labeledStreams)) {
+        s.end();
+    }
+    quarantineStream.end();
+    await Promise.all([...Object.values(labeledStreams).map((s) => once(s, "close")), once(quarantineStream, "close")]);
+    // 4. Splits — manifest derived by streaming the per-split labeled files; no in-memory
+    // source-id arrays. `sort(1)` from coreutils produces the deterministic per-split .txt
+    // manifests with disk spill for splits that exceed in-memory thresholds.
+    opts.onProgress?.("split", `splitting ${aligned} aligned rows`);
+    const splitsDir = join(opts.outputDir, "splits");
+    const splitCounts = await writeSplitManifestsFromLabeledFiles({
+        labeledPaths,
+        outputDir: splitsDir,
+        corpusVersion: opts.corpusVersion,
+        counts,
+        holdouts,
+    });
+    // 5. Parquet slices — per-split labeled JSONL streams in, sliced `.parquet` out. The prior
+    // `splitFor(source_id)` callback (and the `Map<source_id, SplitName>` behind it) is gone.
+    opts.onProgress?.("slice", "writing parquet slices");
+    const sliceManifest = await writeSlices({
+        train: streamJSONL(labeledPaths.train),
+        val: streamJSONL(labeledPaths.val),
+        test: streamJSONL(labeledPaths.test),
+    }, {
+        outputDir: opts.outputDir,
+        corpusVersion: opts.corpusVersion,
+        rowsPerSlice,
+    });
+    // License-set visibility (#26): loudly report the resolved license set so a build is an obvious
+    // deliberate act — especially a proprietary-weights build (did you pass --exclude-share-alike?).
+    const licenseSummary = [...licenseCounts.entries()].toSorted((a, b) => b[1] - a[1]);
+    opts.onProgress?.("manifest", `license set: ${licenseSummary.map(([l, c]) => `${l}=${c}`).join(", ")}` +
+        (excludedByLicense > 0
+            ? ` | EXCLUDED ${excludedByLicense} rows by --exclude-licenses`
+            : " | NO license exclusion applied (all rows kept)"));
+    // 6. Top-level manifest.
+    opts.onProgress?.("manifest", "writing top-level MANIFEST.json");
+    const manifest = {
+        corpus_version: opts.corpusVersion,
+        built_at,
+        adapters: adapterRuns,
+        skipped_adapters: skipped,
+        splits: { counts: splitCounts, holdouts },
+        slices: { counts: sliceManifest.counts, total_rows: sliceManifest.total_rows },
+        quarantine_count: quarantined,
+        total_aligned_rows: aligned,
+        licenses: Object.fromEntries(licenseSummary),
+        excluded_by_license: excludedByLicense,
+    };
+    await writeLocalJSONFile(manifest, opts.outputDir, "MANIFEST.json");
+    return manifest;
+}
+async function* streamJSONL(path) {
+    // JSONSpliterator yields already-parsed rows (skipEmpty is on by default, so blank
+    // lines are dropped at the row level) and throws SyntaxError on a malformed row —
+    // same fail-loud behavior as the prior readline + bare `JSON.parse`.
+    yield* JSONSpliterator.fromAsync(path);
+}
+//# sourceMappingURL=build.js.map

package/out/build.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"build.js","sourceRoot":"","sources":["../lib/build.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA+CG;AAEH,OAAO,EAAE,UAAU,EAAE,iBAAiB,EAAE,MAAM,4BAA4B,CAAA;AAC1E,OAAO,EAAE,eAAe,EAAoB,MAAM,4BAA4B,CAAA;AAC9E,OAAO,EAAE,kBAAkB,EAAE,eAAe,EAAE,MAAM,4BAA4B,CAAA;AAChF,OAAO,EAAE,IAAI,EAAE,MAAM,SAAS,CAAA;AAC9B,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAE7C,OAAO,EAAE,sBAAsB,EAAE,MAAM,iBAAiB,CAAA;AACxD,OAAO,EAAE,OAAO,EAAE,MAAM,MAAM,CAAA;AAC9B,OAAO,EAAE,IAAI,EAAE,UAAU,EAA2B,MAAM,SAAS,CAAA;AACnE,OAAO,EAAE,8BAA8B,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAA;AAEnF,OAAO,EAAE,QAAQ,EAAE,MAAM,cAAc,CAAA;AACvC,OAAO,EAAE,eAAe,EAAE,MAAM,gBAAgB,CAAA;AAChD,OAAO,EAAE,WAAW,EAAsB,MAAM,gBAAgB,CAAA;AAChE,OAAO,EACN,eAAe,EACf,WAAW,EACX,mCAAmC,GAGnC,MAAM,cAAc,CAAA;AA6ErB;;;;;;GAMG;AACH,MAAM,CAAC,KAAK,UAAU,WAAW,CAAC,IAAwB;IACzD,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,IAAI,sBAAsB,CAAC,IAAI,EAAE,CAAA;IAC/D,MAAM,UAAU,GAAG,IAAI,CAAC,UAAU,IAAI,IAAI,CAAA;IAC1C,MAAM,YAAY,GAAG,IAAI,CAAC,YAAY,IAAI,SAAS,CAAA;IACnD,MAAM,QAAQ,GAAG,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE,CAAA;IAEzC,MAAM,eAAe,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;IACrC,MAAM,eAAe,GAAG,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,cAAc,CAAC,CAAA;IAC5D,MAAM,eAAe,CAAC,eAAe,CAAC,CAAA;IAEtC,mBAAmB;IACnB,MAAM,WAAW,GAAyB,EAAE,CAAA;IAC5C,MAAM,OAAO,GAAa,EAAE,CAAA;IAE5B,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAChC,MAAM,cAAc,GAAG,IAAI,CAAC,aAAa,CAAC,OAAO,CAAC,EAAE,CAAC,CAAA;QAErD,IAAI,CAAC,cAAc,EAAE,CAAC;YACrB,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC,CAAA;YACxB,IAAI,CAAC,UAAU,EAAE,CAAC,aAAa,EAAE,WAAW,OAAO,CAAC,EAAE,wBAAwB,CAAC,CAAA;YAE/E,SAAQ;QACT,CAAC;QAED,gGAAgG;QAChG,2FAA2F;QAC3F,+FAA+F;QAC/F,8FAA8F;QAC9F,sGAAsG;QACtG,MAAM,UAAU,GAAG,IAAI,CAAC,eAAe,EAAE,OAAO,CAAC,EAAE,CAAC,CAAA;QACpD,MAAM,cAAc,GAAG,IAAI,CAAC,UAAU,EAAE,eAAe,CAAC,CAAA;QAExD,IACC,OAAO,CAAC,gBAAgB,KAAK,GAAG;YAChC,CAAC,MAAM,UAAU,CAAC,cAAc,CAAC,CAAC;YAClC,CAAC,MAAM,UAAU,CAAC,IAAI,CAAC,UAAU,EAAE,iBAAiB,CAAC,CAAC,CAAC,EACtD,CAAC;YACF,MAAM,MAAM,GAAG,MAAM,iBAAiB,CAAqB,cAAc,CAAC,CAAA;YAC1E,IAAI,CAAC,UAAU,EAAE,CAAC,aAAa,EAAE,WAAW,OAAO,CAAC,EAAE,YAAY,MAAM,CAAC,OAAO,kBAAkB,CAAC,CAAA;YACnG,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YAExB,SAAQ;QACT,CAAC;QAED,IAAI,CAAC,UAAU,EAAE,CAAC,aAAa,EAAE,WAAW,OAAO,CAAC,EAAE,EAAE,CAAC,CAAA;QAEzD,MAAM,CAAC,GAAG,MAAM,UAAU,CAAC;YAC1B,OAAO;YACP,cAAc;YACd,SAAS,EAAE,eAAe;YAC1B,aAAa,EAAE,IAAI,CAAC,aAAa;SACjC,CAAC,CAAA;QAEF,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;IACpB,CAAC;IAED,yFAAyF;IACzF,8FAA8F;IAC9F,0FAA0F;IAC1F,0FAA0F;IAC1F,0EAA0E;IAC1E,MAAM,YAAY,GAA8B;QAC/C,KAAK,EAAE,IAAI,CAAC,eAAe,EAAE,qBAAqB,CAAC;QACnD,GAAG,EAAE,IAAI,CAAC,eAAe,EAAE,mBAAmB,CAAC;QAC/C,IAAI,EAAE,IAAI,CAAC,eAAe,EAAE,oBAAoB,CAAC;KACjD,CAAA;IAED,MAAM,cAAc,GAAmC;QACtD,KAAK,EAAE,eAAe,CAAC,YAAY,CAAC,KAAK,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC;QAChE,GAAG,EAAE,eAAe,CAAC,YAAY,CAAC,GAAG,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC;QAC5D,IAAI,EAAE,eAAe,CAAC,YAAY,CAAC,IAAI,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC;KAC9D,CAAA;IAED,MAAM,cAAc,GAAG,IAAI,CAAC,eAAe,EAAE,kBAAkB,CAAC,CAAA;IAChE,MAAM,gBAAgB,GAAG,eAAe,CAAC,cAAc,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAA;IAE9E,IAAI,OAAO,GAAG,CAAC,CAAA;IACf,IAAI,WAAW,GAAG,CAAC,CAAA;IACnB,MAAM,MAAM,GAA8B,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,IAAI,EAAE,CAAC,EAAE,CAAA;IACvE,MAAM,QAAQ,GAAG,eAAe,EAAE,CAAA;IAClC,8FAA8F;IAC9F,mGAAmG;IACnG,2EAA2E;IAC3E,MAAM,eAAe,GAAG,IAAI,CAAC,eAAe,IAAI,EAAE,CAAA;IAClD,MAAM,aAAa,GAAG,IAAI,GAAG,EAAkB,CAAA;IAC/C,IAAI,iBAAiB,GAAG,CAAC,CAAA;IAEzB,MAAM,eAAe,GAAG,CAAC,GAAiB,EAAE,MAAc,EAAQ,EAAE;QACnE,gBAAgB,CAAC,KAAK,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,CAAC,IAAI,CAAC,CAAA;IAC/D,CAAC,CAAA;IAED,KAAK,MAAM,UAAU,IAAI,WAAW,EAAE,CAAC;QACtC,IAAI,CAAC,UAAU,EAAE,CAAC,OAAO,EAAE,YAAY,UAAU,CAAC,UAAU,EAAE,CAAC,CAAA;QAE/D,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,WAAW,CAAe,UAAU,CAAC,UAAU,CAAC,EAAE,CAAC;YAC1E,aAAa,CAAC,GAAG,CAAC,GAAG,CAAC,OAAO,EAAE,CAAC,aAAa,CAAC,GAAG,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAA;YAEzE,0FAA0F;YAC1F,sFAAsF;YACtF,0FAA0F;YAC1F,+FAA+F;YAC/F,IAAI,eAAe,CAAC,GAAG,CAAC,OAAO,EAAE,eAAe,CAAC,EAAE,CAAC;gBACnD,iBAAiB,EAAE,CAAA;gBAEnB,SAAQ;YACT,CAAC;YAED,MAAM,MAAM,GAAmB,CAAC,GAAG,CAAC,CAAA;YAEpC,IAAI,UAAU,EAAE,CAAC;gBAChB,KAAK,MAAM,GAAG,IAAI,aAAa,CAAC,GAAG,EAAE,8BAA8B,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,EAAE,CAAC;oBACnF,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;gBACjB,CAAC;YACF,CAAC;YAED,KAAK,MAAM,CAAC,IAAI,MAAM,EAAE,CAAC;gBACxB,IAAI,MAAmC,CAAA;gBAEvC,IAAI,CAAC;oBACJ,MAAM,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAA;gBACrB,CAAC;gBAAC,OAAO,KAAK,EAAE,CAAC;oBAChB,mFAAmF;oBACnF,qFAAqF;oBACrF,kFAAkF;oBAClF,iFAAiF;oBACjF,eAAe,CAAC,CAAC,EAAE,eAAgB,KAAe,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC,CAAA;oBAE3E,WAAW,EAAE,CAAA;oBAEb,SAAQ;gBACT,CAAC;gBAED,IAAI,MAAM,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;oBAC/B,MAAM,KAAK,GAAG,WAAW,CAAC,MAAM,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAA;oBAC/C,cAAc,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;oBAE9D,MAAM,CAAC,KAAK,CAAC,EAAE,CAAA;oBAEf,OAAO,EAAE,CAAA;gBACV,CAAC;qBAAM,CAAC;oBACP,eAAe,CAAC,CAAC,EAAE,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,CAAA;oBAErC,WAAW,EAAE,CAAA;gBACd,CAAC;YACF,CAAC;QACF,CAAC;IACF,CAAC;IAED,KAAK,MAAM,CAAC,IAAI,MAAM,CAAC,MAAM,CAAC,cAAc,CAAC,EAAE,CAAC;QAC/C,CAAC,CAAC,GAAG,EAAE,CAAA;IACR,CAAC;IAED,gBAAgB,CAAC,GAAG,EAAE,CAAA;IACtB,MAAM,OAAO,CAAC,GAAG,CAAC,CAAC,GAAG,MAAM,CAAC,MAAM,CAAC,cAAc,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC,EAAE,IAAI,CAAC,gBAAgB,EAAE,OAAO,CAAC,CAAC,CAAC,CAAA;IAEnH,sFAAsF;IACtF,uFAAuF;IACvF,yEAAyE;IACzE,IAAI,CAAC,UAAU,EAAE,CAAC,OAAO,EAAE,aAAa,OAAO,eAAe,CAAC,CAAA;IAC/D,MAAM,SAAS,GAAG,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,QAAQ,CAAC,CAAA;IAEhD,MAAM,WAAW,GAAG,MAAM,mCAAmC,CAAC;QAC7D,YAAY;QACZ,SAAS,EAAE,SAAS;QACpB,aAAa,EAAE,IAAI,CAAC,aAAa;QACjC,MAAM;QACN,QAAQ;KACR,CAAC,CAAA;IAEF,2FAA2F;IAC3F,0FAA0F;IAC1F,IAAI,CAAC,UAAU,EAAE,CAAC,OAAO,EAAE,wBAAwB,CAAC,CAAA;IAEpD,MAAM,aAAa,GAAG,MAAM,WAAW,CACtC;QACC,KAAK,EAAE,WAAW,CAAa,YAAY,CAAC,KAAK,CAAC;QAClD,GAAG,EAAE,WAAW,CAAa,YAAY,CAAC,GAAG,CAAC;QAC9C,IAAI,EAAE,WAAW,CAAa,YAAY,CAAC,IAAI,CAAC;KAChD,EACD;QACC,SAAS,EAAE,IAAI,CAAC,SAAS;QACzB,aAAa,EAAE,IAAI,CAAC,aAAa;QACjC,YAAY;KACZ,CACD,CAAA;IAED,gGAAgG;IAChG,iGAAiG;IACjG,MAAM,cAAc,GAAG,CAAC,GAAG,aAAa,CAAC,OAAO,EAAE,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;IAEnF,IAAI,CAAC,UAAU,EAAE,CAChB,UAAU,EACV,gBAAgB,cAAc,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE;QACvE,CAAC,iBAAiB,GAAG,CAAC;YACrB,CAAC,CAAC,eAAe,iBAAiB,6BAA6B;YAC/D,CAAC,CAAC,iDAAiD,CAAC,CACtD,CAAA;IAED,yBAAyB;IACzB,IAAI,CAAC,UAAU,EAAE,CAAC,UAAU,EAAE,iCAAiC,CAAC,CAAA;IAEhE,MAAM,QAAQ,GAAwB;QACrC,cAAc,EAAE,IAAI,CAAC,aAAa;QAClC,QAAQ;QACR,QAAQ,EAAE,WAAW;QACrB,gBAAgB,EAAE,OAAO;QACzB,MAAM,EAAE,EAAE,MAAM,EAAE,WAAW,EAAE,QAAQ,EAAE;QACzC,MAAM,EAAE,EAAE,MAAM,EAAE,aAAa,CAAC,MAAM,EAAE,UAAU,EAAE,aAAa,CAAC,UAAU,EAAE;QAC9E,gBAAgB,EAAE,WAAW;QAC7B,kBAAkB,EAAE,OAAO;QAC3B,QAAQ,EAAE,MAAM,CAAC,WAAW,CAAC,cAAc,CAAC;QAC5C,mBAAmB,EAAE,iBAAiB;KACtC,CAAA;IAED,MAAM,kBAAkB,CAAC,QAAQ,EAAE,IAAI,CAAC,SAAS,EAAE,eAAe,CAAC,CAAA;IAEnE,OAAO,QAAQ,CAAA;AAChB,CAAC;AAED,KAAK,SAAS,CAAC,CAAC,WAAW,CAAI,IAAY;IAC1C,mFAAmF;IACnF,kFAAkF;IAClF,qEAAqE;IACrE,KAAK,CAAC,CAAC,eAAe,CAAC,SAAS,CAAI,IAAI,CAAC,CAAA;AAC1C,CAAC"}

package/out/codex/us-fips-state.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"us-fips-state.d.ts","sourceRoot":"","sources":["../../lib/codex/us-fips-state.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH;;GAEG;AACH,MAAM,WAAW,WAAW;IAC3B,YAAY,EAAE,MAAM,CAAA;IACpB,IAAI,EAAE,MAAM,CAAA;CACZ;AAED;;;GAGG;AACH,eAAO,MAAM,aAAa,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,WAAW,CAAC,CA0D9D,CAAA;AAEF;;GAEG;AACH,wBAAgB,eAAe,CAAC,OAAO,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,GAAG,WAAW,GAAG,IAAI,CAItF;AAED;;;GAGG;AACH,eAAO,MAAM,wBAAwB,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,WAAW,CAAC,CAE1E,CAAA;AAED;;;GAGG;AACH,wBAAgB,uBAAuB,CAAC,YAAY,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,GAAG,WAAW,GAAG,IAAI,CAInG"}

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

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