@bio-rs/biors-wasm 0.47.4

package/README.md

@@ -0,0 +1,48 @@
+# biors-wasm
+
+WebAssembly/JavaScript bindings for [bio-rs](https://github.com/bio-rs/bio-rs) core biological sequence processing.
+
+## Usage
+
+```javascript
+import init, {
+  parseFasta,
+  validateFasta,
+  tokenize,
+  buildModelInputWithPolicy,
+  runWorkflow,
+} from "@bio-rs/biors-wasm";
+
+await init();
+
+const fastaText = ">seq1\nACDE\n>seq2\nFGHI\n";
+const bytes = new TextEncoder().encode(fastaText);
+
+const parsed = parseFasta(bytes);
+const validated = validateFasta(bytes, "protein");
+const tokenized = tokenize(parsed, "protein-20");
+const modelInput = buildModelInputWithPolicy(tokenized.records, 8, 0, "fixed_length");
+const workflow = runWorkflow({
+  fastaBytes: bytes,
+  maxLength: 8,
+  padding: "fixed_length",
+  padTokenId: 0,
+});
+```
+
+## Development
+
+```bash
+# Check compilation
+cargo check -p biors-wasm --target wasm32-unknown-unknown
+
+# Build for bundlers
+wasm-pack build packages/rust/biors-wasm --target bundler
+
+# Run browser tests
+wasm-pack test --headless --chrome packages/rust/biors-wasm
+```
+
+## License
+
+MIT OR Apache-2.0

package/biors_wasm.d.ts

@@ -0,0 +1,176 @@
+/* tslint:disable */
+/* eslint-disable */
+
+export interface FastaRecord {
+    id: string;
+    sequence: string;
+}
+
+
+
+export interface ModelInputRecord {
+    id: string;
+    input_ids: number[];
+    attention_mask: number[];
+    truncated: boolean;
+}
+
+export interface ModelInputOutput {
+    policy: {
+        max_length: number;
+        pad_token_id: number;
+        padding: "fixed_length" | "no_padding";
+    };
+    records: ModelInputRecord[];
+}
+
+
+
+export interface ResidueIssue {
+    residue: string;
+    position: number;
+}
+
+
+
+export interface SequenceKindCounts {
+    protein: number;
+    dna: number;
+    rna: number;
+}
+
+
+
+export interface SequenceValidationIssue {
+    symbol: string;
+    position: number;
+    kind: "protein" | "dna" | "rna";
+    code: "ambiguous_symbol" | "invalid_symbol";
+    message: string;
+}
+
+
+
+export interface TokenizedRecord {
+    id: string;
+    tokens: number[];
+    length: number;
+    alphabet: string;
+    valid: boolean;
+    warnings: ResidueIssue[];
+    errors: ResidueIssue[];
+}
+
+export interface TokenizeOutput {
+    inputIds: number[][];
+    attentionMask: number[][];
+    ids: string[];
+    records: TokenizedRecord[];
+}
+
+
+
+export interface ValidatedSequence {
+    id: string;
+    sequence: string;
+    kind: "protein" | "dna" | "rna";
+    alphabet: string;
+    valid: boolean;
+    warnings: SequenceValidationIssue[];
+    errors: SequenceValidationIssue[];
+}
+
+
+
+export interface ValidationReport {
+    records: number;
+    valid_records: number;
+    warning_count: number;
+    error_count: number;
+    kind_counts: SequenceKindCounts;
+    sequences: ValidatedSequence[];
+}
+
+
+
+export interface WorkflowConfig {
+    fastaBytes: Uint8Array;
+    kind?: "auto" | "protein" | "dna" | "rna";
+    profile?: "protein-20" | "protein-20-special";
+    maxLength: number;
+    padding?: "fixed_length" | "no_padding";
+    padTokenId?: number;
+}
+
+export interface WorkflowReadinessIssue {
+    code: string;
+    id: string;
+    warning_count: number;
+    error_count: number;
+    message: string;
+}
+
+export interface WorkflowOutput {
+    workflow: string;
+    model_ready: boolean;
+    validation: ValidationReport;
+    tokenization: {
+        summary: {
+            records: number;
+            total_length: number;
+            valid_records: number;
+            warning_count: number;
+            error_count: number;
+        };
+        records: TokenizedRecord[];
+    };
+    model_input: ModelInputOutput | null;
+    readiness_issues: WorkflowReadinessIssue[];
+    provenance: {
+        biors_core_version: string;
+        invocation: {
+            command: string;
+            arguments: string[];
+        };
+        input_hash: string;
+        normalization: string;
+        validation_alphabet: string;
+        tokenizer: {
+            name: string;
+            vocab_size: number;
+            unknown_token_id: number;
+            unknown_token_policy: string;
+        };
+        model_input_policy: {
+            max_length: number;
+            pad_token_id: number;
+            padding: string;
+        };
+        hashes: {
+            vocabulary_sha256: string;
+            output_data_sha256: string;
+        };
+    };
+}
+
+
+
+export function buildModelInput(tokenized: any, max_length: number): any;
+
+export function buildModelInputWithPolicy(tokenized: any, max_length: number, pad_token_id: number, padding: string): any;
+
+export function init_panic_hook(): void;
+
+/**
+ * Parse FASTA bytes into an array of sequence records.
+ */
+export function parseFasta(bytes: Uint8Array): any;
+
+export function runWorkflow(config: any): any;
+
+export function tokenize(records: any, profile: string): any;
+
+/**
+ * Validate FASTA bytes and return a structured validation report.
+ */
+export function validateFasta(bytes: Uint8Array, kind: string): any;

package/biors_wasm.js

@@ -0,0 +1,9 @@
+/* @ts-self-types="./biors_wasm.d.ts" */
+import * as wasm from "./biors_wasm_bg.wasm";
+import { __wbg_set_wasm } from "./biors_wasm_bg.js";
+
+__wbg_set_wasm(wasm);
+wasm.__wbindgen_start();
+export {
+    buildModelInput, buildModelInputWithPolicy, init_panic_hook, parseFasta, runWorkflow, tokenize, validateFasta
+} from "./biors_wasm_bg.js";

package/index.d.ts

@@ -0,0 +1,44 @@
+// index.d.ts
+// Hand-written public TypeScript declarations for @bio-rs/biors-wasm.
+// Re-exports and refines the auto-generated wasm-pack types.
+
+export {
+    parseFasta,
+    validateFasta,
+    tokenize,
+    buildModelInput,
+    buildModelInputWithPolicy,
+    runWorkflow,
+    default as init,
+} from "./biors_wasm.js";
+
+export type {
+    FastaRecord,
+    SequenceValidationIssue,
+    ValidatedSequence,
+    SequenceKindCounts,
+    ValidationReport,
+    ResidueIssue,
+    TokenizedRecord,
+    TokenizeOutput,
+    ModelInputRecord,
+    ModelInputOutput,
+    WorkflowConfig,
+    WorkflowReadinessIssue,
+    WorkflowOutput,
+} from "./biors_wasm.d.ts";
+
+// Refine function signatures for consumers
+declare module "./biors_wasm.js" {
+    export function parseFasta(bytes: Uint8Array): FastaRecord[];
+    export function validateFasta(bytes: Uint8Array, kind: string): ValidationReport;
+    export function tokenize(records: FastaRecord[], profile: string): TokenizeOutput;
+    export function buildModelInput(tokenized: TokenizedRecord[], maxLength: number): ModelInputOutput;
+    export function buildModelInputWithPolicy(
+        tokenized: TokenizedRecord[],
+        maxLength: number,
+        padTokenId: number,
+        padding: "fixed_length" | "no_padding"
+    ): ModelInputOutput;
+    export function runWorkflow(config: WorkflowConfig): WorkflowOutput;
+}

package/package.json

@@ -0,0 +1,30 @@
+{
+  "name": "@bio-rs/biors-wasm",
+  "type": "module",
+  "description": "WebAssembly bindings for bio-rs biological sequence processing",
+  "version": "0.47.4",
+  "license": "MIT OR Apache-2.0",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/bio-rs/bio-rs.git",
+    "directory": "packages/rust/biors-wasm"
+  },
+  "files": [
+    "biors_wasm_bg.wasm",
+    "biors_wasm.js",
+    "biors_wasm.d.ts",
+    "index.d.ts",
+    "README.md"
+  ],
+  "main": "biors_wasm.js",
+  "types": "index.d.ts",
+  "sideEffects": false,
+  "keywords": [
+    "bioinformatics",
+    "protein",
+    "tokenizer",
+    "wasm",
+    "rust"
+  ],
+  "module": "biors_wasm.js"
+}