mailwoman 1.0.0 → 2.0.0

package/out/commands/corpus/build.d.ts

@@ -0,0 +1,28 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `mailwoman corpus build --version 0.1.0 --output /data/corpus/versioned/ --inputs '{ "wof-admin":
+ *   "/data/wof/admin.db", "wof-postalcode": "/data/wof/post.db" }'`
+ *
+ *   End-to-end corpus build. Drives every registered adapter (or the filtered subset) per `--inputs`,
+ *   runs synthesis + alignment, computes the locality-holdout split, and writes the final JSONL
+ *   shards + per-stage manifests under `<output>/corpus-v<version>/`.
+ *
+ *   Adapters whose id is missing from `--inputs` are skipped (and noted in the manifest); this is how
+ *   the CLI handles partial builds during development.
+ */
+import zod from "zod";
+import type { CommandComponent } from "../../sdk/cli.js";
+declare const BuildConfigSchema: zod.ZodObject<{
+    corpusVersion: zod.ZodDefault<zod.ZodString>;
+    output: zod.ZodString;
+    inputs: zod.ZodString;
+    synthesize: zod.ZodDefault<zod.ZodOptional<zod.z.ZodCoercedBoolean<unknown>>>;
+    rowsPerShard: zod.ZodDefault<zod.ZodOptional<zod.z.ZodCoercedNumber<unknown>>>;
+}, zod.z.core.$strip>;
+export { BuildConfigSchema as options };
+declare const CorpusBuild: CommandComponent<typeof BuildConfigSchema>;
+export default CorpusBuild;
+//# sourceMappingURL=build.d.ts.map

package/out/commands/corpus/build.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"build.d.ts","sourceRoot":"","sources":["../../../commands/corpus/build.tsx"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAOH,OAAO,GAAG,MAAM,KAAK,CAAA;AACrB,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,kBAAkB,CAAA;AAuBxD,QAAA,MAAM,iBAAiB;;;;;;qBAkBrB,CAAA;AAEF,OAAO,EAAE,iBAAiB,IAAI,OAAO,EAAE,CAAA;AAEvC,QAAA,MAAM,WAAW,EAAE,gBAAgB,CAAC,OAAO,iBAAiB,CAwE3D,CAAA;AAED,eAAe,WAAW,CAAA"}

package/out/commands/corpus/build.js

@@ -0,0 +1,107 @@
+import { jsx as _jsx, jsxs as _jsxs } from "react/jsx-runtime";
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `mailwoman corpus build --version 0.1.0 --output /data/corpus/versioned/ --inputs '{ "wof-admin":
+ *   "/data/wof/admin.db", "wof-postalcode": "/data/wof/post.db" }'`
+ *
+ *   End-to-end corpus build. Drives every registered adapter (or the filtered subset) per `--inputs`,
+ *   runs synthesis + alignment, computes the locality-holdout split, and writes the final JSONL
+ *   shards + per-stage manifests under `<output>/corpus-v<version>/`.
+ *
+ *   Adapters whose id is missing from `--inputs` are skipped (and noted in the manifest); this is how
+ *   the CLI handles partial builds during development.
+ */
+import { buildCorpus, defaultAdapterRegistry } from "@mailwoman/corpus";
+import { Box, Text } from "ink";
+import { setImmediate } from "node:timers/promises";
+import { useEffect, useState } from "react";
+import zod from "zod";
+/**
+ * `--inputs` accepts either:
+ *
+ * - A bare string (path-only, for adapters that need no extra options): `"wof-admin": "/data/wof.db"`
+ * - A full AdapterOptions object: `"openaddresses": { "inputPath": "/data/oa.geojsonl", "country":
+ *   "US" }`
+ *
+ * The object form is required by adapters that need a country filter (OpenAddresses), or for
+ * fixture runs that want a `limit`.
+ */
+const AdapterInputSchema = zod.union([
+    zod.string(),
+    zod.object({
+        inputPath: zod.string(),
+        outputDir: zod.string().optional(),
+        country: zod.string().optional(),
+        limit: zod.number().int().positive().optional(),
+    }),
+]);
+const InputsSchema = zod.record(zod.string(), AdapterInputSchema);
+const BuildConfigSchema = zod.object({
+    corpusVersion: zod
+        .string()
+        .default("0.1.0-dev")
+        .describe("Corpus version stamped onto every row + into the output dir name (--version is reserved by Pastel for the CLI's own version)"),
+    output: zod.string().describe("Root output directory; everything lands beneath corpus-v<corpus-version>/"),
+    inputs: zod
+        .string()
+        .describe('JSON map "adapter-id" → input. Each value is either an input path string or a full ' +
+        'AdapterOptions object {"inputPath","country?","limit?"}. Adapters requiring a country ' +
+        "filter (e.g. openaddresses) need the object form. Example: " +
+        '{"wof-admin":"/data/wof.db","openaddresses":{"inputPath":"/data/oa.geojsonl","country":"US"}}'),
+    synthesize: zod.coerce.boolean().optional().default(true).describe("Enable augmentation pass (default true)"),
+    rowsPerShard: zod.coerce.number().int().positive().optional().default(1_000_000).describe("Max rows per JSONL shard"),
+});
+export { BuildConfigSchema as options };
+const CorpusBuild = ({ options }) => {
+    const [error, setError] = useState();
+    const [done, setDone] = useState();
+    const [stage, setStage] = useState();
+    useEffect(() => {
+        if (error) {
+            setImmediate().then(() => process.exit(1));
+        }
+        else if (done) {
+            setImmediate().then(() => process.exit(0));
+        }
+    }, [error, done]);
+    useEffect(() => {
+        let inputsParsed;
+        try {
+            inputsParsed = InputsSchema.parse(JSON.parse(options.inputs));
+        }
+        catch (err) {
+            // eslint-disable-next-line react-hooks/set-state-in-effect -- one-shot mount validation; refactor pending
+            setError(`invalid --inputs JSON: ${err.message}`);
+            return;
+        }
+        const adapterInputs = Object.fromEntries(Object.entries(inputsParsed).map(([id, value]) => [id, typeof value === "string" ? { inputPath: value } : value]));
+        const adapters = defaultAdapterRegistry.list();
+        buildCorpus({
+            outputDir: options.output,
+            corpusVersion: options.corpusVersion,
+            adapters,
+            adapterInputs,
+            synthesize: options.synthesize,
+            rowsPerShard: options.rowsPerShard,
+            onProgress: (name, message) => setStage({ name, message }),
+        })
+            .then((m) => setDone({
+            total: m.shards.total_rows,
+            aligned: m.total_aligned_rows,
+            quarantined: m.quarantine_count,
+            adapters: m.adapters.length,
+        }))
+            .catch((err) => setError(err.message));
+    }, [options]);
+    if (error)
+        return _jsx(Text, { color: "red", children: error });
+    if (done) {
+        return (_jsxs(Box, { flexDirection: "column", children: [_jsxs(Text, { children: ["corpus-v", options.corpusVersion, ": ", _jsx(Text, { color: "green", children: done.total }), " rows (", done.adapters, " adapters,", " ", _jsxs(Text, { dimColor: true, children: [done.quarantined, " quarantined"] }), ")"] }), _jsx(Text, { dimColor: true, children: options.output })] }));
+    }
+    return (_jsxs(Box, { flexDirection: "column", children: [_jsxs(Text, { children: ["building corpus-v", options.corpusVersion, "\u2026"] }), stage ? (_jsxs(Text, { dimColor: true, children: ["[", stage.name, "] ", stage.message] })) : null] }));
+};
+export default CorpusBuild;
+//# sourceMappingURL=build.js.map

package/out/commands/corpus/build.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"build.js","sourceRoot":"","sources":["../../../commands/corpus/build.tsx"],"names":[],"mappings":";AAAA;;;;;;;;;;;;;;GAcG;AAEH,OAAO,EAAE,WAAW,EAAE,sBAAsB,EAAmB,MAAM,mBAAmB,CAAA;AAExF,OAAO,EAAE,GAAG,EAAE,IAAI,EAAE,MAAM,KAAK,CAAA;AAC/B,OAAO,EAAE,YAAY,EAAE,MAAM,sBAAsB,CAAA;AACnD,OAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,MAAM,OAAO,CAAA;AAC3C,OAAO,GAAG,MAAM,KAAK,CAAA;AAGrB;;;;;;;;;GASG;AACH,MAAM,kBAAkB,GAAG,GAAG,CAAC,KAAK,CAAC;IACpC,GAAG,CAAC,MAAM,EAAE;IACZ,GAAG,CAAC,MAAM,CAAC;QACV,SAAS,EAAE,GAAG,CAAC,MAAM,EAAE;QACvB,SAAS,EAAE,GAAG,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;QAClC,OAAO,EAAE,GAAG,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;QAChC,KAAK,EAAE,GAAG,CAAC,MAAM,EAAE,CAAC,GAAG,EAAE,CAAC,QAAQ,EAAE,CAAC,QAAQ,EAAE;KAC/C,CAAC;CACF,CAAC,CAAA;AACF,MAAM,YAAY,GAAG,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,MAAM,EAAE,EAAE,kBAAkB,CAAC,CAAA;AAEjE,MAAM,iBAAiB,GAAG,GAAG,CAAC,MAAM,CAAC;IACpC,aAAa,EAAE,GAAG;SAChB,MAAM,EAAE;SACR,OAAO,CAAC,WAAW,CAAC;SACpB,QAAQ,CACR,8HAA8H,CAC9H;IACF,MAAM,EAAE,GAAG,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,2EAA2E,CAAC;IAC1G,MAAM,EAAE,GAAG;SACT,MAAM,EAAE;SACR,QAAQ,CACR,qFAAqF;QACpF,wFAAwF;QACxF,6DAA6D;QAC7D,+FAA+F,CAChG;IACF,UAAU,EAAE,GAAG,CAAC,MAAM,CAAC,OAAO,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,QAAQ,CAAC,yCAAyC,CAAC;IAC7G,YAAY,EAAE,GAAG,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC,GAAG,EAAE,CAAC,QAAQ,EAAE,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC,QAAQ,CAAC,0BAA0B,CAAC;CACrH,CAAC,CAAA;AAEF,OAAO,EAAE,iBAAiB,IAAI,OAAO,EAAE,CAAA;AAEvC,MAAM,WAAW,GAA+C,CAAC,EAAE,OAAO,EAAE,EAAE,EAAE;IAC/E,MAAM,CAAC,KAAK,EAAE,QAAQ,CAAC,GAAG,QAAQ,EAAU,CAAA;IAC5C,MAAM,CAAC,IAAI,EAAE,OAAO,CAAC,GAAG,QAAQ,EAA6E,CAAA;IAC7G,MAAM,CAAC,KAAK,EAAE,QAAQ,CAAC,GAAG,QAAQ,EAAyC,CAAA;IAE3E,SAAS,CAAC,GAAG,EAAE;QACd,IAAI,KAAK,EAAE,CAAC;YACX,YAAY,EAAE,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;QAC3C,CAAC;aAAM,IAAI,IAAI,EAAE,CAAC;YACjB,YAAY,EAAE,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;QAC3C,CAAC;IACF,CAAC,EAAE,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,CAAA;IAEjB,SAAS,CAAC,GAAG,EAAE;QACd,IAAI,YAAkE,CAAA;QACtE,IAAI,CAAC;YACJ,YAAY,GAAG,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAA;QAC9D,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACd,0GAA0G;YAC1G,QAAQ,CAAC,0BAA2B,GAAa,CAAC,OAAO,EAAE,CAAC,CAAA;YAC5D,OAAM;QACP,CAAC;QAED,MAAM,aAAa,GAAmC,MAAM,CAAC,WAAW,CACvE,MAAM,CAAC,OAAO,CAAC,YAAY,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,KAAK,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,EAAE,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CACjH,CAAA;QAED,MAAM,QAAQ,GAAG,sBAAsB,CAAC,IAAI,EAAE,CAAA;QAC9C,WAAW,CAAC;YACX,SAAS,EAAE,OAAO,CAAC,MAAM;YACzB,aAAa,EAAE,OAAO,CAAC,aAAa;YACpC,QAAQ;YACR,aAAa;YACb,UAAU,EAAE,OAAO,CAAC,UAAU;YAC9B,YAAY,EAAE,OAAO,CAAC,YAAY;YAClC,UAAU,EAAE,CAAC,IAAI,EAAE,OAAO,EAAE,EAAE,CAAC,QAAQ,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC;SAC1D,CAAC;aACA,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CACX,OAAO,CAAC;YACP,KAAK,EAAE,CAAC,CAAC,MAAM,CAAC,UAAU;YAC1B,OAAO,EAAE,CAAC,CAAC,kBAAkB;YAC7B,WAAW,EAAE,CAAC,CAAC,gBAAgB;YAC/B,QAAQ,EAAE,CAAC,CAAC,QAAQ,CAAC,MAAM;SAC3B,CAAC,CACF;aACA,KAAK,CAAC,CAAC,GAAU,EAAE,EAAE,CAAC,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAA;IAC/C,CAAC,EAAE,CAAC,OAAO,CAAC,CAAC,CAAA;IAEb,IAAI,KAAK;QAAE,OAAO,KAAC,IAAI,IAAC,KAAK,EAAC,KAAK,YAAE,KAAK,GAAQ,CAAA;IAElD,IAAI,IAAI,EAAE,CAAC;QACV,OAAO,CACN,MAAC,GAAG,IAAC,aAAa,EAAC,QAAQ,aAC1B,MAAC,IAAI,2BACK,OAAO,CAAC,aAAa,QAAG,KAAC,IAAI,IAAC,KAAK,EAAC,OAAO,YAAE,IAAI,CAAC,KAAK,GAAQ,aAAQ,IAAI,CAAC,QAAQ,gBAAY,GAAG,EAC5G,MAAC,IAAI,IAAC,QAAQ,mBAAE,IAAI,CAAC,WAAW,oBAAoB,SAC9C,EACP,KAAC,IAAI,IAAC,QAAQ,kBAAE,OAAO,CAAC,MAAM,GAAQ,IACjC,CACN,CAAA;IACF,CAAC;IAED,OAAO,CACN,MAAC,GAAG,IAAC,aAAa,EAAC,QAAQ,aAC1B,MAAC,IAAI,oCAAmB,OAAO,CAAC,aAAa,cAAS,EACrD,KAAK,CAAC,CAAC,CAAC,CACR,MAAC,IAAI,IAAC,QAAQ,wBACX,KAAK,CAAC,IAAI,QAAI,KAAK,CAAC,OAAO,IACvB,CACP,CAAC,CAAC,CAAC,IAAI,IACH,CACN,CAAA;AACF,CAAC,CAAA;AAED,eAAe,WAAW,CAAA"}

package/out/commands/corpus/list.d.ts

@@ -0,0 +1,19 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `mailwoman corpus list` — print every adapter known to the default registry, one per line.
+ *
+ *   Used by humans and by scripts that want to fan out a build over adapters. Prints nothing (zero
+ *   lines + exit 0) when no adapter has been registered yet; that's the expected state during early
+ *   Phase 1 while adapters are still being authored.
+ */
+/**
+ * Per-line output is rendered as a single `Text` node so Ink does not column-wrap the adapter id
+ * when the host stdout is non-TTY (CI, spawned tests). The list is meant to be grep-friendly, not
+ * pretty.
+ */
+declare const CorpusList: () => import("react/jsx-runtime").JSX.Element;
+export default CorpusList;
+//# sourceMappingURL=list.d.ts.map

package/out/commands/corpus/list.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"list.d.ts","sourceRoot":"","sources":["../../../commands/corpus/list.tsx"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAKH;;;;GAIG;AACH,QAAA,MAAM,UAAU,+CAcf,CAAA;AAED,eAAe,UAAU,CAAA"}

package/out/commands/corpus/list.js

@@ -0,0 +1,28 @@
+import { jsx as _jsx } from "react/jsx-runtime";
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `mailwoman corpus list` — print every adapter known to the default registry, one per line.
+ *
+ *   Used by humans and by scripts that want to fan out a build over adapters. Prints nothing (zero
+ *   lines + exit 0) when no adapter has been registered yet; that's the expected state during early
+ *   Phase 1 while adapters are still being authored.
+ */
+import { defaultAdapterRegistry } from "@mailwoman/corpus";
+import { Box, Text } from "ink";
+/**
+ * Per-line output is rendered as a single `Text` node so Ink does not column-wrap the adapter id
+ * when the host stdout is non-TTY (CI, spawned tests). The list is meant to be grep-friendly, not
+ * pretty.
+ */
+const CorpusList = () => {
+    const adapters = defaultAdapterRegistry.list();
+    if (adapters.length === 0) {
+        return _jsx(Text, { dimColor: true, children: "No adapters registered." });
+    }
+    return (_jsx(Box, { flexDirection: "column", children: adapters.map((a) => (_jsx(Text, { children: `${a.id}\t${a.defaultLicense}\t${a.description}` }, a.id))) }));
+};
+export default CorpusList;
+//# sourceMappingURL=list.js.map

package/out/commands/corpus/list.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"list.js","sourceRoot":"","sources":["../../../commands/corpus/list.tsx"],"names":[],"mappings":";AAAA;;;;;;;;;;GAUG;AAEH,OAAO,EAAE,sBAAsB,EAAE,MAAM,mBAAmB,CAAA;AAC1D,OAAO,EAAE,GAAG,EAAE,IAAI,EAAE,MAAM,KAAK,CAAA;AAE/B;;;;GAIG;AACH,MAAM,UAAU,GAAG,GAAG,EAAE;IACvB,MAAM,QAAQ,GAAG,sBAAsB,CAAC,IAAI,EAAE,CAAA;IAE9C,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC3B,OAAO,KAAC,IAAI,IAAC,QAAQ,8CAA+B,CAAA;IACrD,CAAC;IAED,OAAO,CACN,KAAC,GAAG,IAAC,aAAa,EAAC,QAAQ,YACzB,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CACpB,KAAC,IAAI,cAAa,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,cAAc,KAAK,CAAC,CAAC,WAAW,EAAE,IAAvD,CAAC,CAAC,EAAE,CAA2D,CAC1E,CAAC,GACG,CACN,CAAA;AACF,CAAC,CAAA;AAED,eAAe,UAAU,CAAA"}

package/out/commands/corpus/run.d.ts

@@ -0,0 +1,26 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `mailwoman corpus run <adapter-id> --input <path> --output <dir> [--country XX] [--limit N]`
+ *
+ *   CLI shim around `runAdapter` from `@mailwoman/corpus`. Resolves `<adapter-id>` against the
+ *   default registry; refuses with a clear, non-zero exit if the id isn't known (and lists the
+ *   registered ids). On success, prints a one-line summary and the path to the manifest file.
+ */
+import zod from "zod";
+import type { CommandComponent } from "../../sdk/cli.js";
+declare const ArgumentsSchema: zod.ZodArray<zod.ZodString>;
+declare const RunConfigSchema: zod.ZodObject<{
+    input: zod.ZodString;
+    output: zod.ZodString;
+    country: zod.ZodOptional<zod.ZodString>;
+    limit: zod.ZodOptional<zod.z.ZodCoercedNumber<unknown>>;
+    corpusVersion: zod.ZodDefault<zod.ZodOptional<zod.ZodString>>;
+    progressEvery: zod.ZodDefault<zod.ZodOptional<zod.z.ZodCoercedNumber<unknown>>>;
+}, zod.z.core.$strip>;
+export { ArgumentsSchema as args, RunConfigSchema as options };
+declare const CorpusRun: CommandComponent<typeof RunConfigSchema, typeof ArgumentsSchema>;
+export default CorpusRun;
+//# sourceMappingURL=run.d.ts.map

package/out/commands/corpus/run.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"run.d.ts","sourceRoot":"","sources":["../../../commands/corpus/run.tsx"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAOH,OAAO,GAAG,MAAM,KAAK,CAAA;AACrB,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,kBAAkB,CAAA;AAExD,QAAA,MAAM,eAAe,6BAAsF,CAAA;AAE3G,QAAA,MAAM,eAAe;;;;;;;qBA0BnB,CAAA;AAEF,OAAO,EAAE,eAAe,IAAI,IAAI,EAAE,eAAe,IAAI,OAAO,EAAE,CAAA;AAE9D,QAAA,MAAM,SAAS,EAAE,gBAAgB,CAAC,OAAO,eAAe,EAAE,OAAO,eAAe,CA2F/E,CAAA;AAED,eAAe,SAAS,CAAA"}

package/out/commands/corpus/run.js

@@ -0,0 +1,109 @@
+import { jsx as _jsx, jsxs as _jsxs } from "react/jsx-runtime";
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `mailwoman corpus run <adapter-id> --input <path> --output <dir> [--country XX] [--limit N]`
+ *
+ *   CLI shim around `runAdapter` from `@mailwoman/corpus`. Resolves `<adapter-id>` against the
+ *   default registry; refuses with a clear, non-zero exit if the id isn't known (and lists the
+ *   registered ids). On success, prints a one-line summary and the path to the manifest file.
+ */
+import { ProgressBar } from "@inkjs/ui";
+import { defaultAdapterRegistry, runAdapter } from "@mailwoman/corpus";
+import { Box, Text } from "ink";
+import { setImmediate } from "node:timers/promises";
+import { useEffect, useState } from "react";
+import zod from "zod";
+const ArgumentsSchema = zod.array(zod.string().describe("Adapter id (e.g. wof-admin, ban, openaddresses)"));
+const RunConfigSchema = zod.object({
+    input: zod.string().describe("Path to the adapter's input data (file, directory, or URL — adapter-specific)"),
+    output: zod.string().describe("Output root directory; the runner creates <output>/<adapter-id>/ under it"),
+    country: zod
+        .string()
+        .regex(/^[A-Z]{2}$/u, "Expected ISO 3166-1 alpha-2 country code (e.g. US, FR)")
+        .optional()
+        .describe("Filter to a single ISO 3166-1 alpha-2 country"),
+    limit: zod.coerce
+        .number()
+        .int()
+        .positive()
+        .optional()
+        .describe("Soft cap on rows the adapter is allowed to emit (smoke runs, fixtures)"),
+    corpusVersion: zod
+        .string()
+        .optional()
+        .default("0.1.0-dev")
+        .describe("Corpus version string stamped onto every row; locks tokenizer pairing"),
+    progressEvery: zod.coerce
+        .number()
+        .int()
+        .positive()
+        .optional()
+        .default(1_000)
+        .describe("Yield count between progress ticks (smaller = chattier, larger = quieter)"),
+});
+export { ArgumentsSchema as args, RunConfigSchema as options };
+const CorpusRun = ({ options, args }) => {
+    const [error, setError] = useState();
+    const [progress, setProgress] = useState({
+        yielded: 0,
+        written: 0,
+        bytes: 0,
+    });
+    const [manifest, setManifest] = useState();
+    useEffect(() => {
+        if (error) {
+            setImmediate().then(() => process.exit(1));
+        }
+        else if (manifest) {
+            setImmediate().then(() => process.exit(0));
+        }
+    }, [error, manifest]);
+    useEffect(() => {
+        const adapterId = args[0];
+        if (!adapterId) {
+            // eslint-disable-next-line react-hooks/set-state-in-effect -- one-shot mount validation; refactor pending
+            setError("missing positional argument: <adapter-id>");
+            return;
+        }
+        const adapter = defaultAdapterRegistry.get(adapterId);
+        if (!adapter) {
+            const ids = defaultAdapterRegistry.ids();
+            const hint = ids.length === 0 ? "(no adapters registered yet)" : `registered: ${ids.join(", ")}`;
+            setError(`unknown adapter id ${JSON.stringify(adapterId)}; ${hint}`);
+            return;
+        }
+        const ac = new AbortController();
+        runAdapter({
+            adapter,
+            adapterOptions: {
+                inputPath: options.input,
+                outputDir: options.output,
+                country: options.country,
+                limit: options.limit,
+                signal: ac.signal,
+            },
+            outputDir: options.output,
+            corpusVersion: options.corpusVersion,
+            progressEvery: options.progressEvery,
+            onProgress: (snap) => {
+                setProgress({ yielded: snap.yielded, written: snap.written, bytes: snap.bytes });
+            },
+        })
+            .then((m) => setManifest(m))
+            .catch((err) => setError(err.message));
+        return () => ac.abort();
+    }, [args, options]);
+    if (error) {
+        return _jsx(Text, { color: "red", children: error });
+    }
+    if (manifest) {
+        return (_jsxs(Box, { flexDirection: "column", children: [_jsxs(Text, { children: [_jsx(Text, { bold: true, children: manifest.adapter_id }), ": wrote ", _jsx(Text, { color: "green", children: manifest.written }), " rows (", _jsxs(Text, { dimColor: true, children: [manifest.deduped, " deduped"] }), ") in ", (manifest.elapsed_ms / 1000).toFixed(2), "s"] }), _jsx(Text, { dimColor: true, children: manifest.jsonl_path }), _jsxs(Text, { dimColor: true, children: ["sha256=", manifest.sha256] })] }));
+    }
+    const ratio = progress.written && progress.yielded ? (progress.written / progress.yielded) * 100 : 0;
+    return (_jsxs(Box, { flexDirection: "column", children: [_jsxs(Text, { children: ["yielded=", progress.yielded, "  ", "written=", progress.written, "  ", "bytes=", progress.bytes] }), _jsx(Box, { paddingX: 1, children: _jsx(ProgressBar, { value: Math.min(100, ratio) }) })] }));
+};
+export default CorpusRun;
+//# sourceMappingURL=run.js.map

package/out/commands/corpus/run.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"run.js","sourceRoot":"","sources":["../../../commands/corpus/run.tsx"],"names":[],"mappings":";AAAA;;;;;;;;;;GAUG;AAEH,OAAO,EAAE,WAAW,EAAE,MAAM,WAAW,CAAA;AACvC,OAAO,EAAE,sBAAsB,EAAE,UAAU,EAA2B,MAAM,mBAAmB,CAAA;AAC/F,OAAO,EAAE,GAAG,EAAE,IAAI,EAAE,MAAM,KAAK,CAAA;AAC/B,OAAO,EAAE,YAAY,EAAE,MAAM,sBAAsB,CAAA;AACnD,OAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,MAAM,OAAO,CAAA;AAC3C,OAAO,GAAG,MAAM,KAAK,CAAA;AAGrB,MAAM,eAAe,GAAG,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,iDAAiD,CAAC,CAAC,CAAA;AAE3G,MAAM,eAAe,GAAG,GAAG,CAAC,MAAM,CAAC;IAClC,KAAK,EAAE,GAAG,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,+EAA+E,CAAC;IAC7G,MAAM,EAAE,GAAG,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,2EAA2E,CAAC;IAC1G,OAAO,EAAE,GAAG;SACV,MAAM,EAAE;SACR,KAAK,CAAC,aAAa,EAAE,wDAAwD,CAAC;SAC9E,QAAQ,EAAE;SACV,QAAQ,CAAC,+CAA+C,CAAC;IAC3D,KAAK,EAAE,GAAG,CAAC,MAAM;SACf,MAAM,EAAE;SACR,GAAG,EAAE;SACL,QAAQ,EAAE;SACV,QAAQ,EAAE;SACV,QAAQ,CAAC,wEAAwE,CAAC;IACpF,aAAa,EAAE,GAAG;SAChB,MAAM,EAAE;SACR,QAAQ,EAAE;SACV,OAAO,CAAC,WAAW,CAAC;SACpB,QAAQ,CAAC,uEAAuE,CAAC;IACnF,aAAa,EAAE,GAAG,CAAC,MAAM;SACvB,MAAM,EAAE;SACR,GAAG,EAAE;SACL,QAAQ,EAAE;SACV,QAAQ,EAAE;SACV,OAAO,CAAC,KAAK,CAAC;SACd,QAAQ,CAAC,2EAA2E,CAAC;CACvF,CAAC,CAAA;AAEF,OAAO,EAAE,eAAe,IAAI,IAAI,EAAE,eAAe,IAAI,OAAO,EAAE,CAAA;AAE9D,MAAM,SAAS,GAAqE,CAAC,EAAE,OAAO,EAAE,IAAI,EAAE,EAAE,EAAE;IACzG,MAAM,CAAC,KAAK,EAAE,QAAQ,CAAC,GAAG,QAAQ,EAAU,CAAA;IAC5C,MAAM,CAAC,QAAQ,EAAE,WAAW,CAAC,GAAG,QAAQ,CAAsD;QAC7F,OAAO,EAAE,CAAC;QACV,OAAO,EAAE,CAAC;QACV,KAAK,EAAE,CAAC;KACR,CAAC,CAAA;IACF,MAAM,CAAC,QAAQ,EAAE,WAAW,CAAC,GAAG,QAAQ,EAAsB,CAAA;IAE9D,SAAS,CAAC,GAAG,EAAE;QACd,IAAI,KAAK,EAAE,CAAC;YACX,YAAY,EAAE,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;QAC3C,CAAC;aAAM,IAAI,QAAQ,EAAE,CAAC;YACrB,YAAY,EAAE,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;QAC3C,CAAC;IACF,CAAC,EAAE,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC,CAAA;IAErB,SAAS,CAAC,GAAG,EAAE;QACd,MAAM,SAAS,GAAG,IAAI,CAAC,CAAC,CAAC,CAAA;QACzB,IAAI,CAAC,SAAS,EAAE,CAAC;YAChB,0GAA0G;YAC1G,QAAQ,CAAC,2CAA2C,CAAC,CAAA;YACrD,OAAM;QACP,CAAC;QAED,MAAM,OAAO,GAAG,sBAAsB,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;QACrD,IAAI,CAAC,OAAO,EAAE,CAAC;YACd,MAAM,GAAG,GAAG,sBAAsB,CAAC,GAAG,EAAE,CAAA;YACxC,MAAM,IAAI,GAAG,GAAG,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,8BAA8B,CAAC,CAAC,CAAC,eAAe,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAA;YAChG,QAAQ,CAAC,sBAAsB,IAAI,CAAC,SAAS,CAAC,SAAS,CAAC,KAAK,IAAI,EAAE,CAAC,CAAA;YACpE,OAAM;QACP,CAAC;QAED,MAAM,EAAE,GAAG,IAAI,eAAe,EAAE,CAAA;QAChC,UAAU,CAAC;YACV,OAAO;YACP,cAAc,EAAE;gBACf,SAAS,EAAE,OAAO,CAAC,KAAK;gBACxB,SAAS,EAAE,OAAO,CAAC,MAAM;gBACzB,OAAO,EAAE,OAAO,CAAC,OAAO;gBACxB,KAAK,EAAE,OAAO,CAAC,KAAK;gBACpB,MAAM,EAAE,EAAE,CAAC,MAAM;aACjB;YACD,SAAS,EAAE,OAAO,CAAC,MAAM;YACzB,aAAa,EAAE,OAAO,CAAC,aAAa;YACpC,aAAa,EAAE,OAAO,CAAC,aAAa;YACpC,UAAU,EAAE,CAAC,IAAI,EAAE,EAAE;gBACpB,WAAW,CAAC,EAAE,OAAO,EAAE,IAAI,CAAC,OAAO,EAAE,OAAO,EAAE,IAAI,CAAC,OAAO,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,EAAE,CAAC,CAAA;YACjF,CAAC;SACD,CAAC;aACA,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC;aAC3B,KAAK,CAAC,CAAC,GAAU,EAAE,EAAE,CAAC,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAA;QAE9C,OAAO,GAAG,EAAE,CAAC,EAAE,CAAC,KAAK,EAAE,CAAA;IACxB,CAAC,EAAE,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,CAAA;IAEnB,IAAI,KAAK,EAAE,CAAC;QACX,OAAO,KAAC,IAAI,IAAC,KAAK,EAAC,KAAK,YAAE,KAAK,GAAQ,CAAA;IACxC,CAAC;IAED,IAAI,QAAQ,EAAE,CAAC;QACd,OAAO,CACN,MAAC,GAAG,IAAC,aAAa,EAAC,QAAQ,aAC1B,MAAC,IAAI,eACJ,KAAC,IAAI,IAAC,IAAI,kBAAE,QAAQ,CAAC,UAAU,GAAQ,EACtC,UAAU,EACX,KAAC,IAAI,IAAC,KAAK,EAAC,OAAO,YAAE,QAAQ,CAAC,OAAO,GAAQ,EAC5C,SAAS,EACV,MAAC,IAAI,IAAC,QAAQ,mBAAE,QAAQ,CAAC,OAAO,gBAAgB,EAC/C,OAAO,EACP,CAAC,QAAQ,CAAC,UAAU,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,SAClC,EACP,KAAC,IAAI,IAAC,QAAQ,kBAAE,QAAQ,CAAC,UAAU,GAAQ,EAC3C,MAAC,IAAI,IAAC,QAAQ,8BAAS,QAAQ,CAAC,MAAM,IAAQ,IACzC,CACN,CAAA;IACF,CAAC;IAED,MAAM,KAAK,GAAG,QAAQ,CAAC,OAAO,IAAI,QAAQ,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,OAAO,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,CAAA;IACpG,OAAO,CACN,MAAC,GAAG,IAAC,aAAa,EAAC,QAAQ,aAC1B,MAAC,IAAI,2BACK,QAAQ,CAAC,OAAO,EACxB,IAAI,cAAU,QAAQ,CAAC,OAAO,EAC9B,IAAI,YAAQ,QAAQ,CAAC,KAAK,IACrB,EACP,KAAC,GAAG,IAAC,QAAQ,EAAE,CAAC,YACf,KAAC,WAAW,IAAC,KAAK,EAAE,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,KAAK,CAAC,GAAI,GACvC,IACD,CACN,CAAA;AACF,CAAC,CAAA;AAED,eAAe,SAAS,CAAA"}

package/out/commands/debug.d.ts

@@ -3,10 +3,13 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  */
-import { PositionalCommandComponent } from "mailwoman/sdk/cli";
 import zod from "zod";
-declare const ArgumentsSchema: zod.ZodArray<zod.ZodString, "many">;
-export { ArgumentsSchema as args };
-declare const DebugCommand: PositionalCommandComponent<typeof ArgumentsSchema>;
+import type { CommandComponent } from "../sdk/cli.js";
+declare const ArgumentsSchema: zod.ZodArray<zod.ZodString>;
+declare const DebugConfigSchema: zod.ZodObject<{
+    locale: zod.ZodOptional<zod.ZodString>;
+}, zod.z.core.$strip>;
+export { ArgumentsSchema as args, DebugConfigSchema as options };
+declare const DebugCommand: CommandComponent<typeof DebugConfigSchema, typeof ArgumentsSchema>;
 export default DebugCommand;
 //# sourceMappingURL=debug.d.ts.map

package/out/commands/debug.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"debug.d.ts","sourceRoot":"","sources":["../../commands/debug.tsx"],"names":[],"mappings":"AAAA;;;;GAIG;AAKH,OAAO,EAAE,0BAA0B,EAAE,MAAM,mBAAmB,CAAA;AAE9D,OAAO,GAAG,MAAM,KAAK,CAAA;AAErB,QAAA,MAAM,eAAe,qCAAiE,CAAA;AACtF,OAAO,EAAE,eAAe,IAAI,IAAI,EAAE,CAAA;AAElC,QAAA,MAAM,YAAY,EAAE,0BAA0B,CAAC,OAAO,eAAe,CAyBpE,CAAA;AAED,eAAe,YAAY,CAAA"}
+{"version":3,"file":"debug.d.ts","sourceRoot":"","sources":["../../commands/debug.tsx"],"names":[],"mappings":"AAAA;;;;GAIG;AAMH,OAAO,GAAG,MAAM,KAAK,CAAA;AACrB,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,eAAe,CAAA;AAErD,QAAA,MAAM,eAAe,6BAAiE,CAAA;AACtF,QAAA,MAAM,iBAAiB;;qBAMrB,CAAA;AACF,OAAO,EAAE,eAAe,IAAI,IAAI,EAAE,iBAAiB,IAAI,OAAO,EAAE,CAAA;AAEhE,QAAA,MAAM,YAAY,EAAE,gBAAgB,CAAC,OAAO,iBAAiB,EAAE,OAAO,eAAe,CA0BpF,CAAA;AAED,eAAe,YAAY,CAAA"}

package/out/commands/debug.js

@@ -10,19 +10,27 @@ import { createAddressParser, createDiagnosticReport } from "mailwoman";
 import { useEffect, useState } from "react";
 import zod from "zod";
 const ArgumentsSchema = zod.array(zod.string().describe("A formatted postal address"));
-export { ArgumentsSchema as args };
-const DebugCommand = ({ args }) => {
+const DebugConfigSchema = zod.object({
+    locale: zod
+        .string()
+        .regex(/^[a-z]{2}(-[A-Z]{2})?$/u, "Expected a BCP-47 tag like en-US or fr-FR")
+        .optional()
+        .describe("BCP-47 locale tag (e.g. en-US, fr-FR). Reserved for the upcoming neural pipeline."),
+});
+export { ArgumentsSchema as args, DebugConfigSchema as options };
+const DebugCommand = ({ options, args }) => {
     const [output, setOutput] = useState();
     const [error, setError] = useState();
     useEffect(() => {
         const parser = createAddressParser();
+        const parseOpts = options.locale ? { verbose: true, locale: options.locale } : { verbose: true };
         parser
-            .parse(args[0], { verbose: true })
+            .parse(args[0], parseOpts)
             .then(createDiagnosticReport)
             .then(setOutput)
             .catch((err) => setError(err.message));
         return;
-    }, [args]);
+    }, [args, options.locale]);
     if (error) {
         return _jsx(Text, { color: "red", children: error });
     }

package/out/commands/debug.js.map

@@ -1 +1 @@
-{"version":3,"file":"debug.js","sourceRoot":"","sources":["../../commands/debug.tsx"],"names":[],"mappings":";AAAA;;;;GAIG;AAEH,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAA;AACnC,OAAO,EAAE,IAAI,EAAE,MAAM,KAAK,CAAA;AAC1B,OAAO,EAAE,mBAAmB,EAAE,sBAAsB,EAAE,MAAM,WAAW,CAAA;AAEvE,OAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,MAAM,OAAO,CAAA;AAC3C,OAAO,GAAG,MAAM,KAAK,CAAA;AAErB,MAAM,eAAe,GAAG,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,4BAA4B,CAAC,CAAC,CAAA;AACtF,OAAO,EAAE,eAAe,IAAI,IAAI,EAAE,CAAA;AAElC,MAAM,YAAY,GAAuD,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE;IACrF,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,GAAG,QAAQ,EAAU,CAAA;IAC9C,MAAM,CAAC,KAAK,EAAE,QAAQ,CAAC,GAAG,QAAQ,EAAU,CAAA;IAE5C,SAAS,CAAC,GAAG,EAAE;QACd,MAAM,MAAM,GAAG,mBAAmB,EAAE,CAAA;QAEpC,MAAM;aACJ,KAAK,CAAC,IAAI,CAAC,CAAC,CAAE,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC;aAClC,IAAI,CAAC,sBAAsB,CAAC;aAC5B,IAAI,CAAC,SAAS,CAAC;aACf,KAAK,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAA;QAEvC,OAAM;IACP,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,CAAA;IAEV,IAAI,KAAK,EAAE,CAAC;QACX,OAAO,KAAC,IAAI,IAAC,KAAK,EAAC,KAAK,YAAE,KAAK,GAAQ,CAAA;IACxC,CAAC;IAED,IAAI,CAAC,MAAM,EAAE,CAAC;QACb,OAAO,KAAC,OAAO,KAAG,CAAA;IACnB,CAAC;IAED,OAAO,KAAC,IAAI,cAAE,MAAM,GAAQ,CAAA;AAC7B,CAAC,CAAA;AAED,eAAe,YAAY,CAAA"}
+{"version":3,"file":"debug.js","sourceRoot":"","sources":["../../commands/debug.tsx"],"names":[],"mappings":";AAAA;;;;GAIG;AAEH,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAA;AACnC,OAAO,EAAE,IAAI,EAAE,MAAM,KAAK,CAAA;AAC1B,OAAO,EAAE,mBAAmB,EAAE,sBAAsB,EAAE,MAAM,WAAW,CAAA;AACvE,OAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,MAAM,OAAO,CAAA;AAC3C,OAAO,GAAG,MAAM,KAAK,CAAA;AAGrB,MAAM,eAAe,GAAG,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,4BAA4B,CAAC,CAAC,CAAA;AACtF,MAAM,iBAAiB,GAAG,GAAG,CAAC,MAAM,CAAC;IACpC,MAAM,EAAE,GAAG;SACT,MAAM,EAAE;SACR,KAAK,CAAC,yBAAyB,EAAE,2CAA2C,CAAC;SAC7E,QAAQ,EAAE;SACV,QAAQ,CAAC,mFAAmF,CAAC;CAC/F,CAAC,CAAA;AACF,OAAO,EAAE,eAAe,IAAI,IAAI,EAAE,iBAAiB,IAAI,OAAO,EAAE,CAAA;AAEhE,MAAM,YAAY,GAAuE,CAAC,EAAE,OAAO,EAAE,IAAI,EAAE,EAAE,EAAE;IAC9G,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,GAAG,QAAQ,EAAU,CAAA;IAC9C,MAAM,CAAC,KAAK,EAAE,QAAQ,CAAC,GAAG,QAAQ,EAAU,CAAA;IAE5C,SAAS,CAAC,GAAG,EAAE;QACd,MAAM,MAAM,GAAG,mBAAmB,EAAE,CAAA;QACpC,MAAM,SAAS,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,IAAa,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,IAAa,EAAE,CAAA;QAElH,MAAM;aACJ,KAAK,CAAC,IAAI,CAAC,CAAC,CAAE,EAAE,SAAS,CAAC;aAC1B,IAAI,CAAC,sBAAsB,CAAC;aAC5B,IAAI,CAAC,SAAS,CAAC;aACf,KAAK,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAA;QAEvC,OAAM;IACP,CAAC,EAAE,CAAC,IAAI,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,CAAA;IAE1B,IAAI,KAAK,EAAE,CAAC;QACX,OAAO,KAAC,IAAI,IAAC,KAAK,EAAC,KAAK,YAAE,KAAK,GAAQ,CAAA;IACxC,CAAC;IAED,IAAI,CAAC,MAAM,EAAE,CAAC;QACb,OAAO,KAAC,OAAO,KAAG,CAAA;IACnB,CAAC;IAED,OAAO,KAAC,IAAI,cAAE,MAAM,GAAQ,CAAA;AAC7B,CAAC,CAAA;AAED,eAAe,YAAY,CAAA"}

package/out/commands/parse.d.ts

@@ -3,17 +3,23 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  */
-import { CommandComponent } from "mailwoman/sdk/cli";
 import zod from "zod";
-declare const ArgumentsSchema: zod.ZodArray<zod.ZodString, "many">;
+import type { CommandComponent } from "../sdk/cli.js";
+declare const ArgumentsSchema: zod.ZodArray<zod.ZodString>;
 export { ArgumentsSchema as args, ParseConfigSchema as options };
 declare const ParseConfigSchema: zod.ZodObject<{
     debug: zod.ZodDefault<zod.ZodOptional<zod.ZodBoolean>>;
-}, "strip", zod.ZodTypeAny, {
-    debug: boolean;
-}, {
-    debug?: boolean | undefined;
-}>;
+    locale: zod.ZodDefault<zod.ZodOptional<zod.ZodString>>;
+    neural: zod.ZodDefault<zod.ZodOptional<zod.ZodBoolean>>;
+    format: zod.ZodDefault<zod.ZodOptional<zod.ZodEnum<{
+        tuple: "tuple";
+        json: "json";
+        xml: "xml";
+    }>>>;
+    model: zod.ZodOptional<zod.ZodString>;
+    tokenizer: zod.ZodOptional<zod.ZodString>;
+    policy: zod.ZodOptional<zod.ZodArray<zod.ZodString>>;
+}, zod.z.core.$strip>;
 declare const ParseCommand: CommandComponent<typeof ParseConfigSchema, typeof ArgumentsSchema>;
 export default ParseCommand;
 //# sourceMappingURL=parse.d.ts.map

package/out/commands/parse.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"parse.d.ts","sourceRoot":"","sources":["../../commands/parse.tsx"],"names":[],"mappings":"AAAA;;;;GAIG;AAKH,OAAO,EAAE,gBAAgB,EAAE,MAAM,mBAAmB,CAAA;AAEpD,OAAO,GAAG,MAAM,KAAK,CAAA;AAErB,QAAA,MAAM,eAAe,qCAAiE,CAAA;AACtF,OAAO,EAAE,eAAe,IAAI,IAAI,EAAE,iBAAiB,IAAI,OAAO,EAAE,CAAA;AAEhE,QAAA,MAAM,iBAAiB;;;;;;EAErB,CAAA;AAEF,QAAA,MAAM,YAAY,EAAE,gBAAgB,CAAC,OAAO,iBAAiB,EAAE,OAAO,eAAe,CA+BpF,CAAA;AAED,eAAe,YAAY,CAAA"}
+{"version":3,"file":"parse.d.ts","sourceRoot":"","sources":["../../commands/parse.tsx"],"names":[],"mappings":"AAAA;;;;GAIG;AAWH,OAAO,GAAG,MAAM,KAAK,CAAA;AACrB,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,eAAe,CAAA;AAKrD,QAAA,MAAM,eAAe,6BAAiE,CAAA;AACtF,OAAO,EAAE,eAAe,IAAI,IAAI,EAAE,iBAAiB,IAAI,OAAO,EAAE,CAAA;AAEhE,QAAA,MAAM,iBAAiB;;;;;;;;;;;;qBA+BrB,CAAA;AAqBF,QAAA,MAAM,YAAY,EAAE,gBAAgB,CAAC,OAAO,iBAAiB,EAAE,OAAO,eAAe,CAwDpF,CAAA;AAsDD,eAAe,YAAY,CAAA"}

package/out/commands/parse.js

@@ -5,35 +5,104 @@ import { jsx as _jsx } from "react/jsx-runtime";
  * @author Teffen Ellis, et al.
  */
 import { Spinner } from "@inkjs/ui";
+import { decodeAsJson, decodeAsTuples, decodeAsXml, proposalsToTree } from "@mailwoman/core/decoder";
+import { collectProposals, filterByPolicy } from "@mailwoman/core/parser";
+import { InMemoryPolicyRegistry } from "@mailwoman/core/policy";
+import { createNeuralProposalClassifier, NeuralAddressClassifier } from "@mailwoman/neural";
 import { Text } from "ink";
 import { createAddressParser, createDiagnosticReport } from "mailwoman";
 import { useEffect, useState } from "react";
 import zod from "zod";
+const POLICY_MODES = ["rule_only", "neural_only", "both", "neural_preferred", "rule_preferred"];
+const POLICY_SPEC_RE = /^([a-z_]+)=([a-z_]+)$/u;
 const ArgumentsSchema = zod.array(zod.string().describe("A formatted postal address"));
 export { ArgumentsSchema as args, ParseConfigSchema as options };
 const ParseConfigSchema = zod.object({
     debug: zod.boolean().optional().default(false).describe("Enable verbose debugging output"),
+    locale: zod
+        .string()
+        .regex(/^[a-z]{2}(-[A-Z]{2})?$/u, "Expected a BCP-47 tag like en-US or fr-FR")
+        .optional()
+        .default("en-US")
+        .describe("Locale tag matching a weights package (en-US, fr-FR). Default en-US."),
+    neural: zod
+        .boolean()
+        .optional()
+        .default(false)
+        .describe("Route through the neural classifier instead of the rule-based parser."),
+    format: zod
+        .enum(["json", "tuple", "xml"])
+        .optional()
+        .default("json")
+        .describe("Output projection for --neural. Ignored without --neural."),
+    model: zod.string().optional().describe("Explicit model.onnx path (--neural only). Overrides --locale resolution."),
+    tokenizer: zod
+        .string()
+        .optional()
+        .describe("Explicit tokenizer.model path (--neural only). Overrides --locale resolution."),
+    policy: zod
+        .array(zod.string().regex(POLICY_SPEC_RE, "Expected <component>=<mode> e.g. postcode=neural_preferred"))
+        .optional()
+        .describe("Per-component policy override, repeatable. <component>=<mode> where mode is one of: " +
+        POLICY_MODES.join(", ") +
+        ". Requires --neural."),
 });
+function parsePolicySpecs(specs) {
+    const out = [];
+    for (const spec of specs) {
+        const m = POLICY_SPEC_RE.exec(spec);
+        if (!m)
+            throw new Error(`Invalid --policy spec ${spec}; expected <component>=<mode>`);
+        const [, component, mode] = m;
+        if (!POLICY_MODES.includes(mode)) {
+            throw new Error(`Unknown policy mode ${mode}; valid: ${POLICY_MODES.join(", ")}`);
+        }
+        out.push({ component: component, mode: mode });
+    }
+    return out;
+}
 const ParseCommand = ({ options, args }) => {
     const [output, setOutput] = useState();
     const [error, setError] = useState();
     useEffect(() => {
-        const parser = createAddressParser();
         const input = args[0];
+        if (options.policy && options.policy.length > 0 && !options.neural) {
+            // eslint-disable-next-line react-hooks/set-state-in-effect -- one-shot mount validation; refactor pending
+            setError("--policy requires --neural in this version (rule-side policy integration is a follow-up).");
+            return;
+        }
+        if (options.neural) {
+            const policyOverrides = options.policy ? parsePolicySpecs(options.policy) : [];
+            runNeural(input, options, policyOverrides)
+                .then(setOutput)
+                .catch((err) => setError(err.message));
+            return;
+        }
+        const parser = createAddressParser();
+        const parseOpts = options.locale ? { locale: options.locale } : {};
         if (options.debug) {
             parser
-                .parse(input, { verbose: true })
+                .parse(input, { verbose: true, ...parseOpts })
                 .then(createDiagnosticReport)
                 .then(setOutput)
                 .catch((err) => setError(err.message));
         }
         else {
             parser
-                .parse(input)
+                .parse(input, parseOpts)
                 .then((results) => setOutput(JSON.stringify(results, null, 2)))
                 .catch((err) => setError(err.message));
         }
-    }, [args, options.debug]);
+    }, [
+        args,
+        options.debug,
+        options.locale,
+        options.neural,
+        options.format,
+        options.model,
+        options.tokenizer,
+        options.policy,
+    ]);
     if (error) {
         return _jsx(Text, { color: "red", children: error });
     }
@@ -42,5 +111,48 @@ const ParseCommand = ({ options, args }) => {
     }
     return _jsx(Text, { children: output });
 };
+async function runNeural(input, options, policyOverrides) {
+    const neural = await NeuralAddressClassifier.loadFromWeights({
+        locale: options.locale,
+        modelPath: options.model,
+        tokenizerPath: options.tokenizer,
+    });
+    // Fast path: no policy → use the existing direct projection, preserves containment nesting.
+    if (policyOverrides.length === 0) {
+        switch (options.format) {
+            case "xml":
+                return neural.parseXml(input);
+            case "tuple":
+                return JSON.stringify(await neural.parseTuples(input), null, 2);
+            default:
+                return JSON.stringify(await neural.parseJson(input), null, 2);
+        }
+    }
+    // Policy path: run the proposal pipeline so per-component overrides are honored. Containment
+    // nesting is lost in this projection — see proposals-to-tree.ts for why.
+    const proposalCls = createNeuralProposalClassifier({ id: `neural-cli-${options.locale}`, classifier: neural });
+    // Without rule classifiers in the CLI loop, the registry's default rule_only would drop every
+    // neural proposal and produce empty output. Default every component to neural_only when
+    // --neural --policy is used, then layer the user's overrides on top.
+    const policy = InMemoryPolicyRegistry.withDefaults();
+    for (const entry of policy.entries()) {
+        policy.set({ component: entry.component, mode: "neural_only" });
+    }
+    for (const o of policyOverrides) {
+        policy.set({ component: o.component, mode: o.mode });
+    }
+    const wholeInputSection = { body: input, start: 0, end: input.length };
+    const proposals = await collectProposals([wholeInputSection], [proposalCls], { locale: options.locale });
+    const filtered = filterByPolicy(proposals, policy, options.locale);
+    const tree = proposalsToTree(input, filtered);
+    switch (options.format) {
+        case "xml":
+            return decodeAsXml(tree);
+        case "tuple":
+            return JSON.stringify(decodeAsTuples(tree), null, 2);
+        default:
+            return JSON.stringify(decodeAsJson(tree), null, 2);
+    }
+}
 export default ParseCommand;
 //# sourceMappingURL=parse.js.map