src/sp-fs-shim.ts

/**
 * Browser-side shim for `fs` — used only in the Vite browser build.
 *
 * sentencepiece-js calls `fs.readFileSync(path)` internally (line 122 of its
 * dist) to read the .model bytes before writing them into its Emscripten
 * virtual FS.  In the browser there is no real filesystem, so we maintain a
 * small in-process registry keyed by virtual path strings.
 *
 * Usage (in tokenizer.ts browser path):
 *   import { registerBytes } from './sp-fs-shim';
 *   registerBytes('needle.model', modelBytes);
 *   await spp.load('needle.model');   // sentencepiece-js calls readFileSync('needle.model') → shim
 */

// Use a globalThis-anchored registry so that even if Vite's CJS/ESM interop
// causes two evaluations of this module (once for sentencepiece-js's
// `require('fs')` alias, once for tokenizer.ts's `import('./sp-fs-shim')`),
// both see the same Map.
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const _g = globalThis as any;
if (!_g.__sp_fs_registry__) {
  _g.__sp_fs_registry__ = new Map<string, Uint8Array>();
}
const _registry: Map<string, Uint8Array> = _g.__sp_fs_registry__;

/**
 * Register `bytes` under `path` so that `readFileSync(path)` returns them.
 * Call this before `spp.load(path)`.
 */
export function registerBytes(path: string, bytes: Uint8Array): void {
  _registry.set(path, bytes);
}

/**
 * Drop-in replacement for `fs.readFileSync(path)` used by sentencepiece-js.
 * Returns the Uint8Array that was previously registered via `registerBytes`.
 */
export function readFileSync(path: string): Uint8Array {
  const bytes = _registry.get(path);
  if (!bytes) {
    throw new Error(
      `sp-fs-shim: no bytes registered for "${path}". ` +
        'Call registerBytes(path, bytes) before spp.load(path).',
    );
  }
  return bytes;
}

// Default export mirrors the shape that sentencepiece-js sees after
// _interopNamespace wraps the CJS `require('fs')` result.
export default { readFileSync };