@mailwoman/corpus 10.0.0 → 10.2.0

package/out/test-kit/corpus-recipe.d.ts

@@ -3,25 +3,25 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   Shared harness for the slice-recipe tests. Each of them was carrying its own copy of the same three
- *   pieces — a row type, a scratch-directory writer, and a runner that differed only in which recipe and
- *   which seed it passed.
+ *   Shared harness for the corpus-recipe tests.
  *
- *   Lives under `test-kit/` (the convention `mailwoman/test-kit/` set) and is excluded from corpus's
- *   build project, so it is never emitted into `out/` and never reaches the published tarball.
+ *   It lives under `test-kit/` and is excluded from corpus's build project, so it is never emitted into
+ *   `out/` and never reaches the published tarball.
  */
 import { type TemporaryDirectory } from "@mailwoman/core/fs/temporary";
-import type { SliceRecipeOpts } from "#recipes/scaffold";
+import type { PathBuilder } from "path-ts";
+import type { RecipeOptions } from "#recipes/scaffold";
 /**
- * The fields a slice-recipe assertion reads off an emitted row.
+ * The fields a recipe assertion reads off an emitted row.
  *
- * `parseJSONStrict` without a type argument hands back `unknown`; naming the shape is what lets the assertions be
- * checked at all. The previous `Record<string, never>` typed every property as `never`, so nothing could be read
- * without a cast and nothing was ever verified.
+ * A declared shape lets the code check `parseJSONStrict`'s `unknown` value.
  */
-export interface SliceRow {
+export interface RecipeRow {
     raw: string;
-    synth_method?: string;
+    recipe?: string;
+    register?: string | null;
+    surface?: string;
+    base_source_id?: string | null;
     source?: string;
     source_id?: string;
     components?: Partial<Record<string, string>>;
@@ -32,28 +32,34 @@ export interface SliceRow {
  * A recipe's `run` surface, as the tests drive it.
  */
 export interface CorpusRecipe<TStats> {
-    run(options: SliceRecipeOpts, emit: (line: string) => void): Promise<TStats>;
+    run(options: RecipeOptions, emit: (line: string) => void): Promise<TStats>;
 }
 /**
  * The two input paths a recipe reads, plus the directory holding them.
  *
- * The CALLER owns it: a recipe opens both files by path well after this function returns, so the directory has to
- * outlive the call. Bind it with `using` and it goes when the test does.
+ * The caller owns it, because a recipe opens both files by path well after this function returns.
  */
-export type SliceRecipeInputs = TemporaryDirectory & {
-    input: string;
-    exclude: string;
+export type RecipeInputs = TemporaryDirectory & {
+    input: PathBuilder;
+    exclude: PathBuilder;
 };
 /**
  * Write the tuple + reserved-surface inputs a recipe reads, into a fresh temporary directory.
  */
-export declare function scratch(prefix: string, tuples: object[], surfaces: string[]): Promise<SliceRecipeInputs>;
+export declare function scratch(prefix: string, tuples: object[], surfaces: string[]): Promise<RecipeInputs>;
 /**
- * Bind a recipe and its seed to a runner the tests call with just the tuples and reserved surfaces. The seed is
- * per-recipe and required — these suites assert on generated distributions.
+ * The register id included in a harness-built tuple set.
+ *
+ * It names no publisher, so a row written under it cannot be mistaken for a real register's record.
+ */
+export declare const TEST_REGISTER = "test-harness";
+/**
+ * Bind a recipe and its seed to a runner the tests call with just the tuples and reserved surfaces.
+ *
+ * The seed is required because these suites assert on generated distributions.
  */
-export declare function sliceRunner<TStats>(prefix: string, recipe: CorpusRecipe<TStats>, seed: number): (tuples: object[], surfaces: string[], opts?: Partial<SliceRecipeOpts>) => Promise<{
+export declare function recipeRunner<TStats>(prefix: string, recipe: CorpusRecipe<TStats>, seed: number): (tuples: object[], surfaces: string[], opts?: Partial<RecipeOptions>) => Promise<{
     stats: TStats;
-    rows: SliceRow[];
+    rows: RecipeRow[];
 }>;
 //# sourceMappingURL=corpus-recipe.d.ts.map

package/out/test-kit/corpus-recipe.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"corpus-recipe.d.ts","sourceRoot":"","sources":["../../lib/test-kit/corpus-recipe.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,OAAO,EAAsB,KAAK,kBAAkB,EAAE,MAAM,8BAA8B,CAAA;AAI1F,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,mBAAmB,CAAA;AAExD;;;;;;GAMG;AACH,MAAM,WAAW,QAAQ;IACxB,GAAG,EAAE,MAAM,CAAA;IACX,YAAY,CAAC,EAAE,MAAM,CAAA;IACrB,MAAM,CAAC,EAAE,MAAM,CAAA;IACf,SAAS,CAAC,EAAE,MAAM,CAAA;IAClB,UAAU,CAAC,EAAE,OAAO,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;IAC5C,MAAM,CAAC,EAAE,MAAM,EAAE,CAAA;IACjB,MAAM,CAAC,EAAE,MAAM,EAAE,CAAA;CACjB;AAED;;GAEG;AACH,MAAM,WAAW,YAAY,CAAC,MAAM;IACnC,GAAG,CAAC,OAAO,EAAE,eAAe,EAAE,IAAI,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,GAAG,OAAO,CAAC,MAAM,CAAC,CAAA;CAC5E;AAED;;;;;GAKG;AACH,MAAM,MAAM,iBAAiB,GAAG,kBAAkB,GAAG;IAAE,KAAK,EAAE,MAAM,CAAC;IAAC,OAAO,EAAE,MAAM,CAAA;CAAE,CAAA;AAEvF;;GAEG;AACH,wBAAsB,OAAO,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC,iBAAiB,CAAC,CAS9G;AAED;;;GAGG;AACH,wBAAgB,WAAW,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,YAAY,CAAC,MAAM,CAAC,EAAE,IAAI,EAAE,MAAM,YAEpF,MAAM,EAAE,YACN,MAAM,EAAE,SACZ,OAAO,CAAC,eAAe,CAAC,KAC5B,OAAO,CAAC;IAAE,KAAK,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,QAAQ,EAAE,CAAA;CAAE,CAAC,CAW/C"}
+{"version":3,"file":"corpus-recipe.d.ts","sourceRoot":"","sources":["../../lib/test-kit/corpus-recipe.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EAAsB,KAAK,kBAAkB,EAAE,MAAM,8BAA8B,CAAA;AAG1F,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,SAAS,CAAA;AAE1C,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,mBAAmB,CAAA;AAEtD;;;;GAIG;AACH,MAAM,WAAW,SAAS;IACzB,GAAG,EAAE,MAAM,CAAA;IACX,MAAM,CAAC,EAAE,MAAM,CAAA;IACf,QAAQ,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;IACxB,OAAO,CAAC,EAAE,MAAM,CAAA;IAChB,cAAc,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;IAC9B,MAAM,CAAC,EAAE,MAAM,CAAA;IACf,SAAS,CAAC,EAAE,MAAM,CAAA;IAClB,UAAU,CAAC,EAAE,OAAO,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;IAC5C,MAAM,CAAC,EAAE,MAAM,EAAE,CAAA;IACjB,MAAM,CAAC,EAAE,MAAM,EAAE,CAAA;CACjB;AAED;;GAEG;AACH,MAAM,WAAW,YAAY,CAAC,MAAM;IACnC,GAAG,CAAC,OAAO,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,GAAG,OAAO,CAAC,MAAM,CAAC,CAAA;CAC1E;AAED;;;;GAIG;AACH,MAAM,MAAM,YAAY,GAAG,kBAAkB,GAAG;IAAE,KAAK,EAAE,WAAW,CAAC;IAAC,OAAO,EAAE,WAAW,CAAA;CAAE,CAAA;AAE5F;;GAEG;AACH,wBAAsB,OAAO,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC,YAAY,CAAC,CASzG;AAED;;;;GAIG;AACH,eAAO,MAAM,aAAa,iBAAiB,CAAA;AAE3C;;;;GAIG;AACH,wBAAgB,YAAY,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,YAAY,CAAC,MAAM,CAAC,EAAE,IAAI,EAAE,MAAM,YAErF,MAAM,EAAE,YACN,MAAM,EAAE,SACZ,OAAO,CAAC,aAAa,CAAC,KAC1B,OAAO,CAAC;IAAE,KAAK,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,SAAS,EAAE,CAAA;CAAE,CAAC,CAqBhD"}

package/out/test-kit/corpus-recipe.js

@@ -3,36 +3,51 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   Shared harness for the slice-recipe tests. Each of them was carrying its own copy of the same three
- *   pieces — a row type, a scratch-directory writer, and a runner that differed only in which recipe and
- *   which seed it passed.
+ *   Shared harness for the corpus-recipe tests.
  *
- *   Lives under `test-kit/` (the convention `mailwoman/test-kit/` set) and is excluded from corpus's
- *   build project, so it is never emitted into `out/` and never reaches the published tarball.
+ *   It lives under `test-kit/` and is excluded from corpus's build project, so it is never emitted into
+ *   `out/` and never reaches the published tarball.
  */
 import { temporaryDirectory } from "@mailwoman/core/fs/temporary";
-import { writeLocalTextFile } from "@mailwoman/core/fs/writers";
+import { writeLocalTextFile, writeLocalJSONLFile } from "@mailwoman/core/fs/writers";
 import { parseJSONStrict } from "@mailwoman/core/json";
 /**
  * Write the tuple + reserved-surface inputs a recipe reads, into a fresh temporary directory.
  */
 export async function scratch(prefix, tuples, surfaces) {
     const dir = await temporaryDirectory(`${prefix}-`);
-    const input = dir.resolve("tuples.jsonl");
-    const exclude = dir.resolve("surfaces.txt");
-    await writeLocalTextFile(tuples.map((t) => JSON.stringify(t)).join("\n") + "\n", input);
+    const input = dir.path("tuples.jsonl");
+    const exclude = dir.path("surfaces.txt");
+    await writeLocalJSONLFile(tuples, input);
     await writeLocalTextFile("# reserved\n" + surfaces.join("\n") + "\n", exclude);
     return dir.moveWith({ input, exclude });
 }
 /**
- * Bind a recipe and its seed to a runner the tests call with just the tuples and reserved surfaces. The seed is
- * per-recipe and required — these suites assert on generated distributions.
+ * The register id included in a harness-built tuple set.
+ *
+ * It names no publisher, so a row written under it cannot be mistaken for a real register's record.
+ */
+export const TEST_REGISTER = "test-harness";
+/**
+ * Bind a recipe and its seed to a runner the tests call with just the tuples and reserved surfaces.
+ *
+ * The seed is required because these suites assert on generated distributions.
  */
-export function sliceRunner(prefix, recipe, seed) {
+export function recipeRunner(prefix, recipe, seed) {
     return async function run(tuples, surfaces, opts = {}) {
         await using inputs = await scratch(prefix, tuples, surfaces);
         const lines = [];
-        const stats = await recipe.run({ output: "", seed, variants: 1, input: inputs.input, excludeSurfaces: inputs.exclude, ...opts }, (line) => lines.push(line));
+        const stats = await recipe.run({
+            output: "",
+            seed,
+            variants: 1,
+            input: inputs.input,
+            excludeSurfaces: inputs.exclude,
+            // The tuples come from this harness rather than a publisher.
+            // A recipe that reads them refuses to run without a register.
+            register: TEST_REGISTER,
+            ...opts,
+        }, (line) => lines.push(line));
         return { stats, rows: lines.map((line) => parseJSONStrict(line)) };
     };
 }

package/out/test-kit/corpus-recipe.js.map

@@ -1 +1 @@
-{"version":3,"file":"corpus-recipe.js","sourceRoot":"","sources":["../../lib/test-kit/corpus-recipe.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,OAAO,EAAE,kBAAkB,EAA2B,MAAM,8BAA8B,CAAA;AAC1F,OAAO,EAAE,kBAAkB,EAAE,MAAM,4BAA4B,CAAA;AAC/D,OAAO,EAAE,eAAe,EAAE,MAAM,sBAAsB,CAAA;AAoCtD;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,OAAO,CAAC,MAAc,EAAE,MAAgB,EAAE,QAAkB;IACjF,MAAM,GAAG,GAAG,MAAM,kBAAkB,CAAC,GAAG,MAAM,GAAG,CAAC,CAAA;IAClD,MAAM,KAAK,GAAG,GAAG,CAAC,OAAO,CAAC,cAAc,CAAC,CAAA;IACzC,MAAM,OAAO,GAAG,GAAG,CAAC,OAAO,CAAC,cAAc,CAAC,CAAA;IAE3C,MAAM,kBAAkB,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,KAAK,CAAC,CAAA;IACvF,MAAM,kBAAkB,CAAC,cAAc,GAAG,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,OAAO,CAAC,CAAA;IAE9E,OAAO,GAAG,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC,CAAA;AACxC,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,WAAW,CAAS,MAAc,EAAE,MAA4B,EAAE,IAAY;IAC7F,OAAO,KAAK,UAAU,GAAG,CACxB,MAAgB,EAChB,QAAkB,EAClB,IAAI,GAA6B,EAAE;QAEnC,YAAY,MAAM,GAAG,MAAM,OAAO,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,CAAC,CAAA;QAC5D,MAAM,KAAK,GAAa,EAAE,CAAA;QAE1B,MAAM,KAAK,GAAG,MAAM,MAAM,CAAC,GAAG,CAC7B,EAAE,MAAM,EAAE,EAAE,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,EAAE,KAAK,EAAE,MAAM,CAAC,KAAK,EAAE,eAAe,EAAE,MAAM,CAAC,OAAO,EAAE,GAAG,IAAI,EAAE,EAChG,CAAC,IAAI,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAC1B,CAAA;QAED,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,eAAe,CAAW,IAAI,CAAC,CAAC,EAAE,CAAA;IAC7E,CAAC,CAAA;AACF,CAAC"}
+{"version":3,"file":"corpus-recipe.js","sourceRoot":"","sources":["../../lib/test-kit/corpus-recipe.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EAAE,kBAAkB,EAA2B,MAAM,8BAA8B,CAAA;AAC1F,OAAO,EAAE,kBAAkB,EAAE,mBAAmB,EAAE,MAAM,4BAA4B,CAAA;AACpF,OAAO,EAAE,eAAe,EAAE,MAAM,sBAAsB,CAAA;AAqCtD;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,OAAO,CAAC,MAAc,EAAE,MAAgB,EAAE,QAAkB;IACjF,MAAM,GAAG,GAAG,MAAM,kBAAkB,CAAC,GAAG,MAAM,GAAG,CAAC,CAAA;IAClD,MAAM,KAAK,GAAG,GAAG,CAAC,IAAI,CAAC,cAAc,CAAC,CAAA;IACtC,MAAM,OAAO,GAAG,GAAG,CAAC,IAAI,CAAC,cAAc,CAAC,CAAA;IAExC,MAAM,mBAAmB,CAAC,MAAM,EAAE,KAAK,CAAC,CAAA;IACxC,MAAM,kBAAkB,CAAC,cAAc,GAAG,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,EAAE,OAAO,CAAC,CAAA;IAE9E,OAAO,GAAG,CAAC,QAAQ,CAAC,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC,CAAA;AACxC,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,MAAM,aAAa,GAAG,cAAc,CAAA;AAE3C;;;;GAIG;AACH,MAAM,UAAU,YAAY,CAAS,MAAc,EAAE,MAA4B,EAAE,IAAY;IAC9F,OAAO,KAAK,UAAU,GAAG,CACxB,MAAgB,EAChB,QAAkB,EAClB,IAAI,GAA2B,EAAE;QAEjC,YAAY,MAAM,GAAG,MAAM,OAAO,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,CAAC,CAAA;QAC5D,MAAM,KAAK,GAAa,EAAE,CAAA;QAE1B,MAAM,KAAK,GAAG,MAAM,MAAM,CAAC,GAAG,CAC7B;YACC,MAAM,EAAE,EAAE;YACV,IAAI;YACJ,QAAQ,EAAE,CAAC;YACX,KAAK,EAAE,MAAM,CAAC,KAAK;YACnB,eAAe,EAAE,MAAM,CAAC,OAAO;YAC/B,6DAA6D;YAC7D,8DAA8D;YAC9D,QAAQ,EAAE,aAAa;YACvB,GAAG,IAAI;SACP,EACD,CAAC,IAAI,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAC1B,CAAA;QAED,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,eAAe,CAAY,IAAI,CAAC,CAAC,EAAE,CAAA;IAC9E,CAAC,CAAA;AACF,CAAC"}

package/out/test-kit.d.ts

@@ -0,0 +1,63 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The corpus adapter-test harness — the sibling of `mailwoman/test-kit/`.
+ *
+ *   `corpus/package.json` excludes this directory from the published tarball.
+ *   That entry predates this file. The test harness can therefore import `vitest`.
+ *   `mailwoman/test-kit/index.ts` uses the same packaging boundary.
+ */
+import { PathBuilder, type PathBuilderLike } from "path-ts";
+import type { CanonicalRow } from "#types";
+/**
+ * A per-test scratch directory.
+ *
+ * `path` exists only inside a test body.
+ * A read before `beforeEach` or after `afterEach` throws.
+ */
+export interface ScratchDir {
+    readonly path: PathBuilder;
+}
+/**
+ * Register a fresh scratch directory for each test in the current suite, removed afterwards.
+ *
+ * `slug` names the directory (`mailwoman-<slug>-xxxxxx` under the OS temp dir)
+ * so a stray leftover is traceable to its suite.
+ * Teardown swallows its own errors, because a test that already removed the directory
+ * or a platform holding a handle open must not turn a passing assertion into a failing suite.
+ */
+export declare function useScratchDir(slug: string): ScratchDir;
+/**
+ * Read back the canonical rows a `runAdapter` call wrote at `<outputDir>/<adapterID>/canonical.jsonl`.
+ */
+export declare function readCanonicalRows(outputDir: PathBuilderLike, adapterID: string): Promise<CanonicalRow[]>;
+/**
+ * Write a delimited fixture (a header line plus the given rows) and answer its path;
+ * `createNewlineWriter` terminates every line, so a caller passes content without a delimiter.
+ */
+export declare function writeDelimitedFixture<P extends PathBuilderLike>(filePath: P, header: string, rows: readonly string[]): Promise<P>;
+/**
+ * Whether a test that calls a publisher's live service should run.
+ *
+ * CI's `unit-slow` leg runs every `*.integration.test.ts`, and the required
+ * `test` context fails when that leg does.
+ * A test calling a foreign government host therefore makes a merge depend on that host's
+ * availability: Poland's address service answered `ECONNRESET` mid-run on 2026-10-02
+ * and reddened a local preflight that had passed an hour earlier.
+ *
+ * So a live-service test is opt-in, and the suite reports it as skipped
+ * rather than passing on a check it did not make.
+ * Run one with `MAILWOMAN_LIVE_PUBLISHER_TESTS=1 yarn vitest --run --config vitest.slow.config.ts <file>`.
+ *
+ * A test reading a committed fixture needs this switch unset, and most of
+ * each harvester's coverage is there.
+ * The switch is for the assertions only the publisher can answer: that a URL still serves,
+ * that a feed still states what the register recorded, that a harvest the adapter can read comes back.
+ *
+ * It reads `process.env` directly, because it is a per-run test switch rather than
+ * project configuration and so has no entry in `@mailwoman/core/env`.
+ */
+export declare const LIVE_PUBLISHER_TESTS: boolean;
+//# sourceMappingURL=test-kit.d.ts.map

package/out/test-kit.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"test-kit.d.ts","sourceRoot":"","sources":["../lib/test-kit.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAGH,OAAO,EAAE,WAAW,EAAE,KAAK,eAAe,EAAE,MAAM,SAAS,CAAA;AAI3D,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,QAAQ,CAAA;AAE1C;;;;;GAKG;AACH,MAAM,WAAW,UAAU;IAC1B,QAAQ,CAAC,IAAI,EAAE,WAAW,CAAA;CAC1B;AAED;;;;;;;GAOG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,UAAU,CAwBtD;AAED;;GAEG;AACH,wBAAgB,iBAAiB,CAAC,SAAS,EAAE,eAAe,EAAE,SAAS,EAAE,MAAM,GAAG,OAAO,CAAC,YAAY,EAAE,CAAC,CAIxG;AAED;;;GAGG;AACH,wBAAsB,qBAAqB,CAAC,CAAC,SAAS,eAAe,EACpE,QAAQ,EAAE,CAAC,EACX,MAAM,EAAE,MAAM,EACd,IAAI,EAAE,SAAS,MAAM,EAAE,GACrB,OAAO,CAAC,CAAC,CAAC,CAUZ;AAED;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,eAAO,MAAM,oBAAoB,SAEkB,CAAA"}

package/out/test-kit.js

@@ -0,0 +1,88 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The corpus adapter-test harness — the sibling of `mailwoman/test-kit/`.
+ *
+ *   `corpus/package.json` excludes this directory from the published tarball.
+ *   That entry predates this file. The test harness can therefore import `vitest`.
+ *   `mailwoman/test-kit/index.ts` uses the same packaging boundary.
+ */
+import { temporaryDirectory } from "@mailwoman/core/fs/temporary";
+import { PathBuilder } from "path-ts";
+import { createNewlineWriter, JSONSpliterator } from "spliterator";
+import { afterEach, beforeEach } from "vitest";
+/**
+ * Register a fresh scratch directory for each test in the current suite, removed afterwards.
+ *
+ * `slug` names the directory (`mailwoman-<slug>-xxxxxx` under the OS temp dir)
+ * so a stray leftover is traceable to its suite.
+ * Teardown swallows its own errors, because a test that already removed the directory
+ * or a platform holding a handle open must not turn a passing assertion into a failing suite.
+ */
+export function useScratchDir(slug) {
+    let owned = null;
+    beforeEach(async () => {
+        owned = await temporaryDirectory(`mailwoman-${slug}-`);
+    });
+    // The directory is owned by the test, never by a module-scoped stack: under `isolate: false`
+    // this module is shared across every corpus adapter suite in a fork.
+    afterEach(async () => {
+        try {
+            await owned?.[Symbol.asyncDispose]();
+        }
+        catch { }
+        owned = null;
+    });
+    return {
+        get path() {
+            if (!owned)
+                throw new Error(`useScratchDir("${slug}"): the scratch directory exists only inside a test body`);
+            return owned.path;
+        },
+    };
+}
+/**
+ * Read back the canonical rows a `runAdapter` call wrote at `<outputDir>/<adapterID>/canonical.jsonl`.
+ */
+export function readCanonicalRows(outputDir, adapterID) {
+    return Array.fromAsync(JSONSpliterator.fromAsync(PathBuilder.from(outputDir)(adapterID, "canonical.jsonl")));
+}
+/**
+ * Write a delimited fixture (a header line plus the given rows) and answer its path;
+ * `createNewlineWriter` terminates every line, so a caller passes content without a delimiter.
+ */
+export async function writeDelimitedFixture(filePath, header, rows) {
+    await using out = createNewlineWriter(filePath);
+    await out.write(header);
+    for (const row of rows) {
+        await out.write(row);
+    }
+    return filePath;
+}
+/**
+ * Whether a test that calls a publisher's live service should run.
+ *
+ * CI's `unit-slow` leg runs every `*.integration.test.ts`, and the required
+ * `test` context fails when that leg does.
+ * A test calling a foreign government host therefore makes a merge depend on that host's
+ * availability: Poland's address service answered `ECONNRESET` mid-run on 2026-10-02
+ * and reddened a local preflight that had passed an hour earlier.
+ *
+ * So a live-service test is opt-in, and the suite reports it as skipped
+ * rather than passing on a check it did not make.
+ * Run one with `MAILWOMAN_LIVE_PUBLISHER_TESTS=1 yarn vitest --run --config vitest.slow.config.ts <file>`.
+ *
+ * A test reading a committed fixture needs this switch unset, and most of
+ * each harvester's coverage is there.
+ * The switch is for the assertions only the publisher can answer: that a URL still serves,
+ * that a feed still states what the register recorded, that a harvest the adapter can read comes back.
+ *
+ * It reads `process.env` directly, because it is a per-run test switch rather than
+ * project configuration and so has no entry in `@mailwoman/core/env`.
+ */
+export const LIVE_PUBLISHER_TESTS = 
+// oxlint-disable-next-line sister-software/no-process-globals -- stated in the block above.
+process.env.MAILWOMAN_LIVE_PUBLISHER_TESTS === "1";
+//# sourceMappingURL=test-kit.js.map

package/out/test-kit.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"test-kit.js","sourceRoot":"","sources":["../lib/test-kit.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAEH,OAAO,EAAE,kBAAkB,EAA2B,MAAM,8BAA8B,CAAA;AAC1F,OAAO,EAAE,WAAW,EAAwB,MAAM,SAAS,CAAA;AAC3D,OAAO,EAAE,mBAAmB,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAClE,OAAO,EAAE,SAAS,EAAE,UAAU,EAAE,MAAM,QAAQ,CAAA;AAc9C;;;;;;;GAOG;AACH,MAAM,UAAU,aAAa,CAAC,IAAY;IACzC,IAAI,KAAK,GAA8B,IAAI,CAAA;IAE3C,UAAU,CAAC,KAAK,IAAI,EAAE;QACrB,KAAK,GAAG,MAAM,kBAAkB,CAAC,aAAa,IAAI,GAAG,CAAC,CAAA;IACvD,CAAC,CAAC,CAAA;IAEF,6FAA6F;IAC7F,qEAAqE;IACrE,SAAS,CAAC,KAAK,IAAI,EAAE;QACpB,IAAI,CAAC;YACJ,MAAM,KAAK,EAAE,CAAC,MAAM,CAAC,YAAY,CAAC,EAAE,CAAA;QACrC,CAAC;QAAC,MAAM,CAAC,CAAA,CAAC;QAEV,KAAK,GAAG,IAAI,CAAA;IACb,CAAC,CAAC,CAAA;IAEF,OAAO;QACN,IAAI,IAAI;YACP,IAAI,CAAC,KAAK;gBAAE,MAAM,IAAI,KAAK,CAAC,kBAAkB,IAAI,0DAA0D,CAAC,CAAA;YAE7G,OAAO,KAAK,CAAC,IAAI,CAAA;QAClB,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,iBAAiB,CAAC,SAA0B,EAAE,SAAiB;IAC9E,OAAO,KAAK,CAAC,SAAS,CACrB,eAAe,CAAC,SAAS,CAAe,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,SAAS,EAAE,iBAAiB,CAAC,CAAC,CAClG,CAAA;AACF,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,KAAK,UAAU,qBAAqB,CAC1C,QAAW,EACX,MAAc,EACd,IAAuB;IAEvB,YAAY,GAAG,GAAG,mBAAmB,CAAC,QAAQ,CAAC,CAAA;IAE/C,MAAM,GAAG,CAAC,KAAK,CAAC,MAAM,CAAC,CAAA;IAEvB,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACxB,MAAM,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAA;IACrB,CAAC;IAED,OAAO,QAAQ,CAAA;AAChB,CAAC;AAED;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,MAAM,CAAC,MAAM,oBAAoB;AAChC,4FAA4F;AAC5F,OAAO,CAAC,GAAG,CAAC,8BAA8B,KAAK,GAAG,CAAA"}

package/out/tools/align-canonical.d.ts

@@ -0,0 +1,7 @@
+export interface AlignCanonicalOptions {
+    input: string;
+    output: string;
+    corpusVersion: string;
+}
+export declare function alignCanonicalRows(args: AlignCanonicalOptions): Promise<void>;
+//# sourceMappingURL=align-canonical.d.ts.map

package/out/tools/align-canonical.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"align-canonical.d.ts","sourceRoot":"","sources":["../../lib/tools/align-canonical.ts"],"names":[],"mappings":"AA+BA,MAAM,WAAW,qBAAqB;IACrC,KAAK,EAAE,MAAM,CAAA;IACb,MAAM,EAAE,MAAM,CAAA;IACd,aAAa,EAAE,MAAM,CAAA;CACrB;AAED,wBAAsB,kBAAkB,CAAC,IAAI,EAAE,qBAAqB,GAAG,OAAO,CAAC,IAAI,CAAC,CA0BnF"}

package/out/tools/align-canonical.js

@@ -0,0 +1,52 @@
+import { stringifyJSON } from "@mailwoman/core/json";
+import { createNewlineWriter, JSONSpliterator } from "spliterator";
+/**
+ * Re-emit a canonical jsonl ({raw, components, country, source, ...}) as a labeled jsonl in
+ * the current align format, by running every row through `alignRow` (corpus/src/align.ts).
+ *
+ * Most synthetic recipe outputs are generated on demand by a recipe (parametrized by --count),
+ * so re-emitting them in a new label format is a re-run.
+ * A few are fixed corpora with a hand/DeepSeek-authored canonical source that is never regenerated,
+ * notably `deepseek-kryptonite` (the adversarial hard-case set) and the `deepseek-translit-*` variants.
+ *
+ * Their committed parquets use the label format that was current when they were first built.
+ *
+ * When the corpus label format changes (the char-offset span triple), those fixed
+ * corpora are aligned again from their canonical source.
+ * Feed the canonical source back through the same `alignRow` the from-source build uses,
+ * so the spans land in the new format with zero drift.
+ *
+ * That is what this does.
+ * Canonical jsonl in → labeled jsonl out, one `alignRow` per row, quarantine on miss.
+ *
+ * It is the uniform counterpart to `tools/overlay/kryptonite.ts`
+ * (which couples to a base manifest and writes parquet directly).
+ * Output goes to jsonl so it joins the same jsonl-to-parquet path every other overlay uses.
+ *
+ * Usage: mailwoman corpus align-slice\ --input /path/canonical-kryptonite.jsonl\
+ * --out /tmp/kryptonite-labeled.jsonl\ --corpus-version 0.5.0
+ */
+import { alignRow } from "#utils";
+export async function alignCanonicalRows(args) {
+    await using outStream = createNewlineWriter(args.output);
+    let labeled = 0;
+    let quarantined = 0;
+    const quarantineReasons = {};
+    for await (const canonical of JSONSpliterator.fromAsync(args.input)) {
+        // Stamp the target corpus version so the emitted row's provenance matches the run it joins.
+        canonical.corpus_version = args.corpusVersion;
+        const result = alignRow(canonical);
+        if (result.kind === "labeled") {
+            await outStream.write(stringifyJSON(result.row));
+            labeled++;
+        }
+        else {
+            quarantined++;
+            const r = result.row.reason ?? "unknown";
+            quarantineReasons[r] = (quarantineReasons[r] ?? 0) + 1;
+        }
+    }
+    console.error(`align-canonical: ${labeled} labeled, ${quarantined} quarantined → ${args.output}\n` +
+        `  quarantine reasons: ${stringifyJSON(quarantineReasons)}`);
+}
+//# sourceMappingURL=align-canonical.js.map

package/out/tools/align-canonical.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"align-canonical.js","sourceRoot":"","sources":["../../lib/tools/align-canonical.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,aAAa,EAAE,MAAM,sBAAsB,CAAA;AACpD,OAAO,EAAE,mBAAmB,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAElE;;;;;;;;;;;;;;;;;;;;;;;;;GAyBG;AACH,OAAO,EAAE,QAAQ,EAAE,MAAM,QAAQ,CAAA;AAQjC,MAAM,CAAC,KAAK,UAAU,kBAAkB,CAAC,IAA2B;IACnE,YAAY,SAAS,GAAG,mBAAmB,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;IACxD,IAAI,OAAO,GAAG,CAAC,CAAA;IACf,IAAI,WAAW,GAAG,CAAC,CAAA;IACnB,MAAM,iBAAiB,GAA2B,EAAE,CAAA;IAEpD,IAAI,KAAK,EAAE,MAAM,SAAS,IAAI,eAAe,CAAC,SAAS,CAAiC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;QACrG,4FAA4F;QAC5F,SAAS,CAAC,cAAc,GAAG,IAAI,CAAC,aAAa,CAAA;QAC7C,MAAM,MAAM,GAAG,QAAQ,CAAC,SAAS,CAAC,CAAA;QAElC,IAAI,MAAM,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;YAC/B,MAAM,SAAS,CAAC,KAAK,CAAC,aAAa,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAA;YAEhD,OAAO,EAAE,CAAA;QACV,CAAC;aAAM,CAAC;YACP,WAAW,EAAE,CAAA;YACb,MAAM,CAAC,GAAG,MAAM,CAAC,GAAG,CAAC,MAAM,IAAI,SAAS,CAAA;YACxC,iBAAiB,CAAC,CAAC,CAAC,GAAG,CAAC,iBAAiB,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;QACvD,CAAC;IACF,CAAC;IAED,OAAO,CAAC,KAAK,CACZ,oBAAoB,OAAO,aAAa,WAAW,kBAAkB,IAAI,CAAC,MAAM,IAAI;QACnF,yBAAyB,aAAa,CAAC,iBAAiB,CAAC,EAAE,CAC5D,CAAA;AACF,CAAC"}

package/out/tools/audit.d.ts

@@ -3,27 +3,35 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   `mailwoman corpus audit` — per-source slice-count vs source_weight diagnostic.
+ *   `mailwoman corpus audit` — per-source parquet-file count vs source_weight diagnostic.
  *
- *   Reads a corpus dir's MANIFEST.json (or scans slices directly), counts slices per source,
- *   optionally loads a training config to pair the counts with the configured source_weights, and
- *   reports the estimated sampled-row distribution at training time.
+ *   Reads a corpus dir's manifest.json (or scans the parquet files directly), counts files per
+ *   source, optionally loads a training config to pair the counts with the configured
+ *   source_weights. It reports the estimated sampled-row distribution at training time.
  *
- *   Would have caught v0.3.0's "NAD = 411/674 train slices × 2.0 weight = ~75% of sampled mix"
+ *   Would have caught v0.3.0's "NAD = 411/674 train files × 2.0 weight = ~75% of sampled mix"
  *   finding before the v0.3.0 retrospective surfaced it.
  *
- *   Emits warnings to stderr and the audit table to stdout; never throws on an empty corpus.
+ *   Emits warnings to stderr and the audit table to stdout. never throws on an empty corpus.
  */
 import { type PathBuilderLike } from "path-ts";
+/**
+ * Options for {@linkcode audit}.
+ */
 export interface AuditOpts {
     corpusDir: PathBuilderLike;
-    configPath?: string;
+    configPath?: PathBuilderLike;
     /**
-     * Sample at most N slices per split when counting sources. Default 100 for speed; bump to read the full set on a slow
-     * run. The first row of each slice determines its source — corpus-v0.2.0+ slices are 100% source-segregated, so a
-     * one-row read is authoritative.
+     * Sample at most N parquet files per split when counting sources.
+     *
+     * Bump to read the full set on a slow run.
+     *
+     * The first row of each file determines its source — corpus-v0.2.0+ files are 100%
+     * source-segregated, so a one-row read is authoritative.
+     *
+     * @defaultValue `100` for speed
      */
-    sampleSliceCount?: number;
+    sampleFileCount?: number;
 }
 export declare function audit(opts: AuditOpts): Promise<void>;
 //# sourceMappingURL=audit.d.ts.map

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

@@ -1 +1 @@
-{"version":3,"file":"audit.d.ts","sourceRoot":"","sources":["../../lib/tools/audit.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAGH,OAAO,EAAkB,KAAK,eAAe,EAAE,MAAM,SAAS,CAAA;AAe9D,MAAM,WAAW,SAAS;IACzB,SAAS,EAAE,eAAe,CAAA;IAC1B,UAAU,CAAC,EAAE,MAAM,CAAA;IACnB;;;;OAIG;IACH,gBAAgB,CAAC,EAAE,MAAM,CAAA;CACzB;AA6VD,wBAAsB,KAAK,CAAC,IAAI,EAAE,SAAS,GAAG,OAAO,CAAC,IAAI,CAAC,CAY1D"}
+{"version":3,"file":"audit.d.ts","sourceRoot":"","sources":["../../lib/tools/audit.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAIH,OAAO,EAAY,KAAK,eAAe,EAAsB,MAAM,SAAS,CAAA;AAkB5E;;GAEG;AACH,MAAM,WAAW,SAAS;IACzB,SAAS,EAAE,eAAe,CAAA;IAC1B,UAAU,CAAC,EAAE,eAAe,CAAA;IAC5B;;;;;;;;;OASG;IACH,eAAe,CAAC,EAAE,MAAM,CAAA;CACxB;AA6UD,wBAAsB,KAAK,CAAC,IAAI,EAAE,SAAS,GAAG,OAAO,CAAC,IAAI,CAAC,CAa1D"}