Nomad changes

e2e/wasm/setup.ts

@@ -0,0 +1,173 @@
+// This file is auto-generated by alef — DO NOT EDIT.
+// alef:hash:4e15143f4af1ae8bafbdb1506ef057da924484c66a19483966333558ad437e75
+// To regenerate: alef generate
+// To verify freshness: alef verify --exit-code
+// Issues & docs: https://github.com/kreuzberg-dev/alef
+import { createRequire } from 'module';
+import { readFileSync } from 'fs';
+import { fileURLToPath } from 'url';
+import { dirname, join } from 'path';
+
+// Pre-initialize the wasm-bindgen module so that exports are callable
+// in every vitest worker. The async default export uses fetch() which
+// does not support file:// URLs in Node.js; use initSync with a
+// readFileSync buffer instead.
+try {
+  const _require = createRequire(import.meta.url);
+  const wasmPkgDir = _require.resolve('@kreuzberg/wasm');
+  const wasmModule = await import(/* @vite-ignore */ wasmPkgDir);
+  const initSync = (wasmModule as unknown as Record<string, unknown>).initSync as ((mod: WebAssembly.Module | BufferSource) => unknown) | undefined;
+  if (typeof initSync === 'function') {
+    // Locate the .wasm binary next to the JS entry.
+    const wasmJsPath = fileURLToPath(new URL(wasmPkgDir, 'file://'));
+    const wasmBinPath = wasmJsPath.replace(/\.js$/, '_bg.wasm');
+    const wasmBytes = readFileSync(wasmBinPath);
+    // Pass as object form to avoid wasm-bindgen deprecation warning.
+    initSync({ module: wasmBytes });
+  } else {
+    // Fallback: try the async default init (wasm-pack --target nodejs bundles).
+    const initDefault = (wasmModule as unknown as Record<string, unknown>).default as (() => Promise<unknown>) | undefined;
+    if (typeof initDefault === 'function') await initDefault();
+  }
+} catch (err) {
+  // Module may not require explicit init — continue anyway.
+  console.warn('[alef wasm setup] init skipped:', (err as Error).message);
+}
+
+// Patch CommonJS `require('env')` and `require('wasi_snapshot_preview1')` to
+// return shim objects. wasm-pack `--target nodejs` emits bare `require()`
+// calls for these from getrandom/wasi transitives, but they are not real
+// Node modules — the WASM module imports them by name and the host is
+// expected to satisfy them. Patch Module._load BEFORE the wasm bundle is
+// imported by any test file.
+// Note: setupFiles run per-test-worker; vitest imports the test files
+// AFTER setupFiles complete, so this hook installs in time.
+{
+  const _require = createRequire(import.meta.url);
+  const Module = _require('module');
+  // env.system / env.mkstemp come from C-runtime calls embedded in some
+  // WASM-compiled deps (e.g. tesseract-wasm). Tests that don't exercise
+  // those paths only need the imports to be callable for module instantiation.
+  const env = {
+    system: (_cmd: number) => -1,
+    mkstemp: (_template: number) => -1,
+  };
+  // WASI shims. Critical: clock_time_get and random_get must produce realistic
+  // values — returning 0 for all clock calls causes WASM-side timing loops to
+  // spin forever (e.g. getrandom's spin-until-elapsed retry), and zero-filled
+  // random buffers can cause init loops in deps expecting non-zero entropy.
+  const _wasiMemoryView = (): DataView | null => {
+    // Imports are wired before the WASM is instantiated; the bundle stashes
+    // its instance on a runtime-known global once available. We try to grab
+    // it lazily so writes to wasm memory go to the right place.
+    const g = globalThis as unknown as { __alef_wasm_memory__?: WebAssembly.Memory };
+    return g.__alef_wasm_memory__ ? new DataView(g.__alef_wasm_memory__.buffer) : null;
+  };
+  const _cryptoFill = (buf: Uint8Array) => {
+    const c = globalThis.crypto;
+    if (c && typeof c.getRandomValues === 'function') c.getRandomValues(buf);
+    else for (let i = 0; i < buf.length; i++) buf[i] = Math.floor(Math.random() * 256);
+  };
+  const wasi_snapshot_preview1 = {
+    proc_exit: () => {},
+    environ_get: () => 0,
+    environ_sizes_get: (countOut: number, _sizeOut: number) => {
+      const v = _wasiMemoryView();
+      if (v) v.setUint32(countOut, 0, true);
+      return 0;
+    },
+    // WASI fd_write must update `nwritten_ptr` with the total bytes consumed,
+    // otherwise libc-style callers (e.g. tesseract-compiled-to-wasm fputs)
+    // see 0 of N bytes written and retry forever, hanging the host.
+    fd_write: (_fd: number, iovsPtr: number, iovsLen: number, nwrittenPtr: number) => {
+      const v = _wasiMemoryView();
+      if (!v) return 0;
+      let total = 0;
+      for (let i = 0; i < iovsLen; i++) {
+        const off = iovsPtr + i * 8;
+        total += v.getUint32(off + 4, true);
+      }
+      v.setUint32(nwrittenPtr, total, true);
+      return 0;
+    },
+    // Mirror fd_write: callers retry on partial reads. Reporting 0 bytes
+    // read (EOF) is fine; just make sure `nread_ptr` is written.
+    fd_read: (_fd: number, _iovsPtr: number, _iovsLen: number, nreadPtr: number) => {
+      const v = _wasiMemoryView();
+      if (v) v.setUint32(nreadPtr, 0, true);
+      return 0;
+    },
+    fd_seek: () => 0,
+    fd_close: () => 0,
+    fd_prestat_get: () => 8, // EBADF — no preopens.
+    fd_prestat_dir_name: () => 0,
+    fd_fdstat_get: () => 0,
+    fd_fdstat_set_flags: () => 0,
+    path_open: () => 44, // ENOENT.
+    path_create_directory: () => 0,
+    path_remove_directory: () => 0,
+    path_unlink_file: () => 0,
+    path_filestat_get: () => 44, // ENOENT.
+    path_rename: () => 0,
+    clock_time_get: (_clockId: number, _precision: bigint, timeOut: number) => {
+      const ns = BigInt(Date.now()) * 1_000_000n + BigInt(performance.now() | 0) % 1_000_000n;
+      const v = _wasiMemoryView();
+      if (v) v.setBigUint64(timeOut, ns, true);
+      return 0;
+    },
+    clock_res_get: (_clockId: number, resOut: number) => {
+      const v = _wasiMemoryView();
+      if (v) v.setBigUint64(resOut, 1_000n, true);
+      return 0;
+    },
+    random_get: (bufPtr: number, bufLen: number) => {
+      const g = globalThis as unknown as { __alef_wasm_memory__?: WebAssembly.Memory };
+      if (!g.__alef_wasm_memory__) return 0;
+      _cryptoFill(new Uint8Array(g.__alef_wasm_memory__.buffer, bufPtr, bufLen));
+      return 0;
+    },
+    args_get: () => 0,
+    args_sizes_get: (countOut: number, _sizeOut: number) => {
+      const v = _wasiMemoryView();
+      if (v) v.setUint32(countOut, 0, true);
+      return 0;
+    },
+    poll_oneoff: () => 0,
+    sched_yield: () => 0,
+  };
+  const _origResolve = Module._resolveFilename;
+  Module._resolveFilename = function(request: string, parent: unknown, ...rest: unknown[]) {
+    if (request === 'env' || request === 'wasi_snapshot_preview1') return request;
+    return _origResolve.call(this, request, parent, ...rest);
+  };
+  const _origLoad = Module._load;
+  Module._load = function(request: string, parent: unknown, ...rest: unknown[]) {
+    if (request === 'env') return env;
+    if (request === 'wasi_snapshot_preview1') return wasi_snapshot_preview1;
+    return _origLoad.call(this, request, parent, ...rest);
+  };
+  // Capture the WASM linear memory at instantiation time so the WASI shims
+  // can read/write into it. Without this, every shim that needs memory
+  // (fd_write nwritten, clock_time_get, random_get, etc.) silently no-ops
+  // and the host-side C runtime hangs in a retry loop.
+  const _OrigInstance = WebAssembly.Instance;
+  const PatchedInstance = function(this: WebAssembly.Instance, mod: WebAssembly.Module, imports?: WebAssembly.Imports) {
+    const inst = new _OrigInstance(mod, imports);
+    const exportsMem = (inst.exports as Record<string, unknown>).memory;
+    if (exportsMem instanceof WebAssembly.Memory) {
+      (globalThis as unknown as { __alef_wasm_memory__?: WebAssembly.Memory }).__alef_wasm_memory__ = exportsMem;
+    }
+    return inst;
+  } as unknown as typeof WebAssembly.Instance;
+  PatchedInstance.prototype = _OrigInstance.prototype;
+  (WebAssembly as unknown as { Instance: typeof WebAssembly.Instance }).Instance = PatchedInstance;
+}
+
+// Change to the configured test-documents directory so that fixture file paths like
+// "pdf/fake_memo.pdf" resolve correctly when vitest runs from e2e/wasm/.
+// setup.ts lives in e2e/wasm/; the fixtures dir lives at the repository root,
+// two directories up: e2e/wasm/ -> e2e/ -> repo root.
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = dirname(__filename);
+const testDocumentsDir = join(__dirname, '..', '..', 'test_documents');
+process.chdir(testDocumentsDir);