@nirs4all/io-wasm 0.1.11 → 0.1.13

package/nirs4all_io_wasm.d.ts

@@ -42,6 +42,16 @@ export function inferFiles(files: any, options: any): any;
  */
 export function inferRecords(record_sets: any): any;
 
+/**
+ * Materialize the same fs-free inputs as [`assemble_dataset`] and return the
+ * canonical structural summary used by the native CLI and C ABI.
+ *
+ * Keeping this as canonical JSON (rather than a re-serialized JS object) makes
+ * the cross-language qualification byte-exact and preserves source, sample,
+ * observation, repetition, group, and fold provenance without host-side logic.
+ */
+export function loadSummary(files: any, record_sets: any, spec: string): string;
+
 /**
  * Iterative, user-validatable dataset inference.
  *

package/nirs4all_io_wasm.js

@@ -84,6 +84,40 @@ function inferRecords(record_sets) {
 }
 exports.inferRecords = inferRecords;
 
+/**
+ * Materialize the same fs-free inputs as [`assemble_dataset`] and return the
+ * canonical structural summary used by the native CLI and C ABI.
+ *
+ * Keeping this as canonical JSON (rather than a re-serialized JS object) makes
+ * the cross-language qualification byte-exact and preserves source, sample,
+ * observation, repetition, group, and fold provenance without host-side logic.
+ * @param {any} files
+ * @param {any} record_sets
+ * @param {string} spec
+ * @returns {string}
+ */
+function loadSummary(files, record_sets, spec) {
+    let deferred3_0;
+    let deferred3_1;
+    try {
+        const ptr0 = passStringToWasm0(spec, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        const len0 = WASM_VECTOR_LEN;
+        const ret = wasm.loadSummary(files, record_sets, ptr0, len0);
+        var ptr2 = ret[0];
+        var len2 = ret[1];
+        if (ret[3]) {
+            ptr2 = 0; len2 = 0;
+            throw takeFromExternrefTable0(ret[2]);
+        }
+        deferred3_0 = ptr2;
+        deferred3_1 = len2;
+        return getStringFromWasm0(ptr2, len2);
+    } finally {
+        wasm.__wbindgen_free(deferred3_0, deferred3_1, 1);
+    }
+}
+exports.loadSummary = loadSummary;
+
 /**
  * Iterative, user-validatable dataset inference.
  *

package/package.json

@@ -4,7 +4,7 @@
     "Gregory Beurier <gregory.beurier@cirad.fr>"
   ],
   "description": "wasm-bindgen binding for nirs4all-io: the fs-free JSON surface (to_spec / validate over a spec dict). Excluded from the cargo workspace; built with wasm-pack.",
-  "version": "0.1.11",
+  "version": "0.1.13",
   "license": "CECILL-2.1 OR AGPL-3.0-or-later",
   "repository": {
     "type": "git",
@@ -14,12 +14,36 @@
   "files": [
     "nirs4all_io_wasm_bg.wasm",
     "nirs4all_io_wasm.js",
-    "nirs4all_io_wasm.d.ts"
+    "nirs4all_io_wasm.d.ts",
+    "idiomatic.mjs",
+    "idiomatic.d.ts",
+    "types/nirs4all-io.d.ts",
+    "LICENSE",
+    "LICENSES",
+    "LICENSING.md",
+    "THIRD_PARTY_NOTICES.md",
+    "COPY_PROVENANCE.md",
+    "nirs4all-io-wasm.cdx.json"
   ],
   "main": "nirs4all_io_wasm.js",
   "types": "nirs4all_io_wasm.d.ts",
   "publishConfig": {
     "access": "public",
     "provenance": true
+  },
+  "exports": {
+    ".": {
+      "types": "./nirs4all_io_wasm.d.ts",
+      "require": "./nirs4all_io_wasm.js",
+      "default": "./nirs4all_io_wasm.js"
+    },
+    "./idiomatic": {
+      "types": "./idiomatic.d.ts",
+      "import": "./idiomatic.mjs",
+      "default": "./idiomatic.mjs"
+    },
+    "./types": {
+      "types": "./types/nirs4all-io.d.ts"
+    }
   }
 }

package/types/nirs4all-io.d.ts

@@ -0,0 +1,143 @@
+// SPDX-License-Identifier: CeCILL-2.1 OR AGPL-3.0-or-later
+//
+// Hand-written TypeScript types for the nirs4all-io wasm binding.
+//
+// wasm-bindgen generates `any` for every `JsValue` parameter and return value
+// (see the generated `nirs4all_io_wasm.d.ts`). This file gives the real shapes
+// of the inputs and results so editors and `tsc` can check call sites. Import it
+// alongside the generated module, or re-export through `idiomatic.mjs`.
+
+/** A raw file: its name (used for source matching) and its bytes. */
+export interface NamedFile {
+  name: string;
+  bytes: Uint8Array;
+}
+
+/** A set of decoded `nirs4all-formats` records for one source. */
+export interface RecordSet {
+  source: string;
+  format?: string;
+  records: unknown[];
+}
+
+/** Options for the inference entry points. */
+export interface InferOptions {
+  /** Override the default `nirs4all-classic` convention list. */
+  conventions?: string[];
+}
+
+/** A user lock fed back into `proposeDataset` to honour a confirmed decision. */
+export interface ConfirmedLock {
+  kind: string;
+  target: string;
+  value?: unknown;
+  status?: string;
+}
+
+/** Options for the iterative `proposeDataset` builder. */
+export interface ProposeOptions extends InferOptions {
+  confirmed?: ConfirmedLock[];
+}
+
+/** A single scored inference decision (structure, signal type, task type, ...). */
+export interface Decision<T = unknown> {
+  value: T;
+  score: number;
+  ambiguous: boolean;
+  evidence: string[];
+  alternatives: unknown[];
+}
+
+/** One source of a `DatasetSpec`. */
+export interface SourceSpec {
+  id: string;
+  role: string;
+  input: string;
+  partition?: string;
+  join?: unknown;
+  params?: Record<string, unknown>;
+  [key: string]: unknown;
+}
+
+/** A canonical dataset specification. */
+export interface DatasetSpec {
+  schema_version: number;
+  name?: string;
+  sources: SourceSpec[];
+  params?: Record<string, unknown>;
+  partitions?: unknown;
+  folds?: unknown;
+  [key: string]: unknown;
+}
+
+/** A scored dataset plan produced by the inference entry points. */
+export interface DatasetPlan {
+  resolved_spec: DatasetSpec;
+  overall_score: number;
+  recommendations: string[];
+  warnings: string[];
+  structure: Decision<string>;
+  signal_type: Decision<string>;
+  task_type: Decision<string>;
+  columns: unknown[];
+  input: Record<string, unknown>;
+  [key: string]: unknown;
+}
+
+/** A still-open (or echoed confirmed) decision from `proposeDataset`. */
+export interface Proposal {
+  kind: string;
+  target: string;
+  value: unknown;
+  alternatives: unknown[];
+  evidence: string[];
+  score: number;
+  ambiguous: boolean;
+  status?: string;
+}
+
+/** The iterative builder result. */
+export interface ProposeResult {
+  plan: DatasetPlan;
+  proposals: Proposal[];
+  spec: DatasetSpec;
+  valid: boolean;
+  validation_errors: string[];
+}
+
+/** A materialized dataset (per-partition X/y/metadata blocks). */
+export interface AssembledDataset {
+  name: string;
+  blocks: Record<string, unknown>;
+  folds?: unknown;
+  [key: string]: unknown;
+}
+
+// --- typed views of the wasm-pack exports -------------------------------------
+
+export function to_spec(specJson: string): string;
+export function validate(specJson: string): void;
+export function version(): string;
+export function inferFiles(files: NamedFile[], options?: InferOptions): DatasetPlan;
+export function inferRecords(recordSets: RecordSet[]): DatasetPlan;
+export function inferDataset(
+  files: NamedFile[],
+  recordSets: RecordSet[],
+  options?: InferOptions,
+): DatasetPlan;
+export function proposeDataset(
+  files: NamedFile[],
+  recordSets: RecordSet[],
+  options?: ProposeOptions,
+): ProposeResult;
+export function assembleDataset(
+  files: NamedFile[],
+  recordSets: RecordSet[],
+  specJson: string,
+): AssembledDataset;
+/** Canonical structural summary from the same native fs-free assembler. */
+export function loadSummary(
+  files: NamedFile[],
+  recordSets: RecordSet[],
+  specJson: string,
+): string;