@stll/anonymize-wasm 2.0.0-alpha.1 → 2.0.1

package/dist/native/wasi-worker.mjs

@@ -0,0 +1,63 @@
+import fs from "node:fs";
+import { createRequire } from "node:module";
+import { parse } from "node:path";
+import { WASI } from "node:wasi";
+import { parentPort, Worker } from "node:worker_threads";
+
+const require = createRequire(import.meta.url);
+
+const { instantiateNapiModuleSync, MessageHandler, getDefaultContext } = require("@napi-rs/wasm-runtime");
+
+if (parentPort) {
+  parentPort.on("message", (data) => {
+    globalThis.onmessage({ data });
+  });
+}
+
+Object.assign(globalThis, {
+  self: globalThis,
+  require,
+  Worker,
+  importScripts: function (f) {
+    ;(0, eval)(fs.readFileSync(f, "utf8") + "//# sourceURL=" + f);
+  },
+  postMessage: function (msg) {
+    if (parentPort) {
+      parentPort.postMessage(msg);
+    }
+  },
+});
+
+const emnapiContext = getDefaultContext();
+
+const __rootDir = parse(process.cwd()).root;
+
+const handler = new MessageHandler({
+  onLoad({ wasmModule, wasmMemory }) {
+    const wasi = new WASI({
+      version: 'preview1',
+      env: process.env,
+      preopens: {
+        [__rootDir]: __rootDir,
+      },
+    });
+
+    return instantiateNapiModuleSync(wasmModule, {
+      childThread: true,
+      wasi,
+      context: emnapiContext,
+      overwriteImports(importObject) {
+        importObject.env = {
+          ...importObject.env,
+          ...importObject.napi,
+          ...importObject.emnapi,
+          memory: wasmMemory
+        };
+      },
+    });
+  },
+});
+
+globalThis.onmessage = function (e) {
+  handler.handle(e);
+};

package/dist/vite.d.mts

@@ -1,7 +1,23 @@
 import { Plugin } from "vite";
 
 //#region src/vite.d.ts
-declare function stllAnonymizeWasmVite(): Plugin;
+/** Which bundled `.stlanonpkg` prepared packages the plugin emits into a
+ * production build. The wasm binary, glue, and workers are always emitted; only
+ * the prepared packages are selectable because they dominate the output size.
+ *
+ * - `"all"` (default): every bundled package, preserving prior behavior.
+ * - `"none"`: no prepared packages. The consumer must supply its own package to
+ *   `loadPipeline`; `loadDefaultPipeline()` then fails at runtime with a fetch
+ *   error because the asset was never emitted.
+ * - a name list: only these packages. Use `"default"` for the full-dictionary
+ *   package and a language code (e.g. `"cs"`, `"de"`, `"en"`) for a scoped one,
+ *   for example `["cs"]` or `["default", "en"]`. A requested package that is
+ *   not bundled fails the build. */
+type AnonymizeWasmPackages = "all" | "none" | readonly string[];
+type AnonymizeWasmPluginOptions = {
+  packages?: AnonymizeWasmPackages;
+};
+declare function stllAnonymizeWasmVite(options?: AnonymizeWasmPluginOptions): Plugin;
 //#endregion
-export { stllAnonymizeWasmVite as default };
+export { AnonymizeWasmPackages, AnonymizeWasmPluginOptions, stllAnonymizeWasmVite as default };
 //# sourceMappingURL=vite.d.mts.map

package/dist/vite.mjs

@@ -1,19 +1,96 @@
+import { readFile, readdir } from "node:fs/promises";
+import { basename, dirname, join } from "node:path";
 //#region src/vite.ts
-const PACKAGES = [
-	"@stll/text-search-wasm",
-	"@stll/aho-corasick-wasm",
-	"@stll/aho-corasick-wasm32-wasi",
-	"@stll/fuzzy-search-wasm",
-	"@stll/fuzzy-search-wasm32-wasi",
-	"@stll/regex-set-wasm",
-	"@stll/regex-set-wasm32-wasi"
-];
-function stllAnonymizeWasmVite() {
+const OPTIMIZE_EXCLUDE = ["@stll/anonymize-wasm", "@napi-rs/wasm-runtime"];
+const NATIVE_DIR = "native";
+/** The Node glue is always shipped in `native/`; any file there works as the
+* anchor. `new URL(fileName, <anchor>)` resolves to `native/<fileName>` because
+* URL resolution replaces the anchor's last path segment. */
+const ANCHOR_FILE = "index.wasi.cjs";
+/** The exact `assetUrl` base expression emitted by the build (see wasm.ts).
+* Kept in sync with the compiled `wasm.mjs`; the transform fails loudly if the
+* dist shape drifts so this cannot silently ship broken asset paths. */
+const ASSET_URL_BASE = "`./${NATIVE_ASSET_DIR}/${fileName}`, import.meta.url";
+/** Prepared packages are named `native-pipeline.stlanonpkg` (the full-dictionary
+* default) and `native-pipeline.<language>.stlanonpkg` (scoped variants). */
+const PACKAGE_SUFFIX = ".stlanonpkg";
+const PACKAGE_PREFIX = "native-pipeline.";
+const DEFAULT_PACKAGE_FILE = `${PACKAGE_PREFIX}stlanonpkg`;
+/** Selection name for the full-dictionary default package. Every other name is
+* a language code that maps to `native-pipeline.<name>.stlanonpkg`. */
+const DEFAULT_PACKAGE_NAME = "default";
+const isWasmEntry = (id) => id.includes("anonymize-wasm") && basename(id) === "wasm.mjs";
+const isPackageFile = (file) => file.endsWith(PACKAGE_SUFFIX);
+/** Map a selection name to its on-disk package file name. */
+const packageFileName = (name) => name === DEFAULT_PACKAGE_NAME ? DEFAULT_PACKAGE_FILE : `${PACKAGE_PREFIX}${name}${PACKAGE_SUFFIX}`;
+/** Resolve which prepared-package files to emit for the given selection,
+* validating that every explicitly requested name exists in `nativeFiles`. */
+const resolvePackageSelection = (nativeFiles, packages) => {
+	const packageFiles = nativeFiles.filter(isPackageFile);
+	if (packages === "all") return {
+		status: "ok",
+		emit: new Set(packageFiles)
+	};
+	if (packages === "none") return {
+		status: "ok",
+		emit: /* @__PURE__ */ new Set()
+	};
+	if (!Array.isArray(packages)) return {
+		status: "invalid",
+		value: String(packages)
+	};
+	const available = new Set(packageFiles);
+	const emit = /* @__PURE__ */ new Set();
+	const missing = [];
+	for (const name of packages) {
+		const fileName = packageFileName(name);
+		if (available.has(fileName)) emit.add(fileName);
+		else missing.push(name);
+	}
+	if (missing.length > 0) return {
+		status: "missing",
+		names: missing,
+		available: packageFiles
+	};
+	return {
+		status: "ok",
+		emit
+	};
+};
+function stllAnonymizeWasmVite(options = {}) {
+	const packages = options.packages ?? "all";
+	let isBuild = false;
 	return {
 		name: "stll-anonymize-wasm",
-		apply: "serve",
 		config() {
-			return { optimizeDeps: { exclude: PACKAGES } };
+			return { optimizeDeps: { exclude: OPTIMIZE_EXCLUDE } };
+		},
+		configResolved(resolved) {
+			isBuild = resolved.command === "build";
+		},
+		async transform(code, id) {
+			if (!(isBuild && isWasmEntry(id))) return null;
+			if (!code.includes(ASSET_URL_BASE)) this.error(`stll-anonymize-wasm: could not find the assetUrl base in ${id}. The @stll/anonymize-wasm dist shape changed; update ASSET_URL_BASE in the Vite plugin.`);
+			const nativeDir = join(dirname(id), NATIVE_DIR);
+			const files = await readdir(nativeDir);
+			const selection = resolvePackageSelection(files, packages);
+			if (selection.status === "invalid") this.error(`stll-anonymize-wasm: invalid packages option ${JSON.stringify(selection.value)}; expected "all", "none", or an array like ["default", "cs"].`);
+			if (selection.status === "missing") this.error(`stll-anonymize-wasm: requested package(s) not bundled: ${selection.names.join(", ")}. Available: ${selection.available.join(", ") || "(none)"}.`);
+			let anchorRef;
+			for (const file of files) {
+				if (isPackageFile(file) && !selection.emit.has(file)) continue;
+				const referenceId = this.emitFile({
+					type: "asset",
+					fileName: `${NATIVE_DIR}/${file}`,
+					source: await readFile(join(nativeDir, file))
+				});
+				if (file === ANCHOR_FILE) anchorRef = referenceId;
+			}
+			if (anchorRef === void 0) this.error(`stll-anonymize-wasm: ${ANCHOR_FILE} is missing from ${nativeDir}; cannot anchor the native asset base.`);
+			return {
+				code: code.replace(ASSET_URL_BASE, `fileName, import.meta.ROLLUP_FILE_URL_${anchorRef}`),
+				map: null
+			};
 		}
 	};
 }

package/dist/vite.mjs.map

@@ -1 +1 @@
-{"version":3,"file":"vite.mjs","names":[],"sources":["../../src/vite.ts"],"sourcesContent":["/* Vite plugin that wires up @stll/anonymize-wasm so its transitive\n * napi-rs-generated wasm loaders survive Vite's dep pre-bundler. Adds\n * this package plus its @stll/*-wasm dependencies to optimizeDeps.exclude\n * so the loader modules are served with their original paths, keeping\n * `new URL(\"./foo.wasm\", import.meta.url)` correct. */\nimport type { Plugin } from \"vite\";\n\n// Only the napi-rs WASM packages need excluding — they use\n// `new URL(\"./foo.wasm\", import.meta.url)` which Vite's optimizer\n// would rewrite. @stll/anonymize-wasm itself is pure JS and should\n// NOT be excluded.\nconst PACKAGES = [\n  \"@stll/text-search-wasm\",\n  \"@stll/aho-corasick-wasm\",\n  \"@stll/aho-corasick-wasm32-wasi\",\n  \"@stll/fuzzy-search-wasm\",\n  \"@stll/fuzzy-search-wasm32-wasi\",\n  \"@stll/regex-set-wasm\",\n  \"@stll/regex-set-wasm32-wasi\",\n];\n\nexport default function stllAnonymizeWasmVite(): Plugin {\n  return {\n    name: \"stll-anonymize-wasm\",\n    apply: \"serve\",\n    config() {\n      return {\n        optimizeDeps: { exclude: PACKAGES },\n      };\n    },\n  };\n}\n"],"mappings":";AAWA,MAAM,WAAW;CACf;CACA;CACA;CACA;CACA;CACA;CACA;AACF;AAEA,SAAwB,wBAAgC;CACtD,OAAO;EACL,MAAM;EACN,OAAO;EACP,SAAS;GACP,OAAO,EACL,cAAc,EAAE,SAAS,SAAS,EACpC;EACF;CACF;AACF"}
+{"version":3,"file":"vite.mjs","names":[],"sources":["../../src/vite.ts"],"sourcesContent":["/* Vite plugin that wires up @stll/anonymize-wasm so its wasm binding and\n * prepared-package assets survive both Vite's dev pre-bundler and a production\n * `vite build`.\n *\n * The package loads its napi-rs binding and bundled `.stlanonpkg` packages at\n * runtime from its own `native/` asset directory via\n * `new URL(\"./native/…\", import.meta.url)` (the wasm module, the WASI worker,\n * the glue, and the packages). The paths are computed at runtime, so neither\n * Vite's optimizer nor Rollup's static `new URL(…, import.meta.url)` handling\n * can follow them on their own.\n *\n * Dev (`vite serve`): the package plus the `@napi-rs/wasm-runtime` loader are\n * added to `optimizeDeps.exclude`, keeping their original on-disk paths so the\n * relative URLs resolve next to the package's own files.\n *\n * Build (`vite build`): Rollup bundles `wasm.mjs` into a chunk, rewriting\n * `import.meta.url`, and never copies the `native/` files. This plugin emits\n * the wasm binary + glue as build assets under a stable `native/` subdir, emits\n * the selected `.stlanonpkg` prepared packages (see {@link\n * AnonymizeWasmPluginOptions.packages}), and re-anchors the package's\n * `assetUrl` base onto a Rollup file-URL reference, so the runtime URLs resolve\n * to the emitted assets from whatever chunk ends up referencing them. The\n * glue's own internal `new URL(…, import.meta.url)` references (its `.wasm` and\n * worker) keep working because the glue directory is emitted side by side.\n *\n * The bundled prepared packages dominate the emitted size (the default\n * full-dictionary package alone is ~20 MB, plus per-language variants). An app\n * that only ever calls `loadPipeline` with its own package, or\n * `loadDefaultPipeline(language)` for a known subset of languages, can drop the\n * rest with the `packages` option instead of shipping every bundled package.\n */\nimport { readdir, readFile } from \"node:fs/promises\";\nimport { basename, dirname, join } from \"node:path\";\nimport type { Plugin } from \"vite\";\n\nconst OPTIMIZE_EXCLUDE = [\"@stll/anonymize-wasm\", \"@napi-rs/wasm-runtime\"];\nconst NATIVE_DIR = \"native\";\n/** The Node glue is always shipped in `native/`; any file there works as the\n * anchor. `new URL(fileName, <anchor>)` resolves to `native/<fileName>` because\n * URL resolution replaces the anchor's last path segment. */\nconst ANCHOR_FILE = \"index.wasi.cjs\";\n/** The exact `assetUrl` base expression emitted by the build (see wasm.ts).\n * Kept in sync with the compiled `wasm.mjs`; the transform fails loudly if the\n * dist shape drifts so this cannot silently ship broken asset paths. */\nconst ASSET_URL_BASE = \"`./${NATIVE_ASSET_DIR}/${fileName}`, import.meta.url\";\n\n/** Prepared packages are named `native-pipeline.stlanonpkg` (the full-dictionary\n * default) and `native-pipeline.<language>.stlanonpkg` (scoped variants). */\nconst PACKAGE_SUFFIX = \".stlanonpkg\";\nconst PACKAGE_PREFIX = \"native-pipeline.\";\nconst DEFAULT_PACKAGE_FILE = `${PACKAGE_PREFIX}stlanonpkg`;\n/** Selection name for the full-dictionary default package. Every other name is\n * a language code that maps to `native-pipeline.<name>.stlanonpkg`. */\nconst DEFAULT_PACKAGE_NAME = \"default\";\n\n/** Which bundled `.stlanonpkg` prepared packages the plugin emits into a\n * production build. The wasm binary, glue, and workers are always emitted; only\n * the prepared packages are selectable because they dominate the output size.\n *\n * - `\"all\"` (default): every bundled package, preserving prior behavior.\n * - `\"none\"`: no prepared packages. The consumer must supply its own package to\n *   `loadPipeline`; `loadDefaultPipeline()` then fails at runtime with a fetch\n *   error because the asset was never emitted.\n * - a name list: only these packages. Use `\"default\"` for the full-dictionary\n *   package and a language code (e.g. `\"cs\"`, `\"de\"`, `\"en\"`) for a scoped one,\n *   for example `[\"cs\"]` or `[\"default\", \"en\"]`. A requested package that is\n *   not bundled fails the build. */\nexport type AnonymizeWasmPackages = \"all\" | \"none\" | readonly string[];\n\nexport type AnonymizeWasmPluginOptions = {\n  packages?: AnonymizeWasmPackages;\n};\n\nconst isWasmEntry = (id: string): boolean =>\n  id.includes(\"anonymize-wasm\") && basename(id) === \"wasm.mjs\";\n\nconst isPackageFile = (file: string): boolean => file.endsWith(PACKAGE_SUFFIX);\n\n/** Map a selection name to its on-disk package file name. */\nconst packageFileName = (name: string): string =>\n  name === DEFAULT_PACKAGE_NAME\n    ? DEFAULT_PACKAGE_FILE\n    : `${PACKAGE_PREFIX}${name}${PACKAGE_SUFFIX}`;\n\ntype PackageSelection =\n  | { status: \"ok\"; emit: ReadonlySet<string> }\n  | { status: \"missing\"; names: string[]; available: string[] }\n  | { status: \"invalid\"; value: string };\n\n/** Resolve which prepared-package files to emit for the given selection,\n * validating that every explicitly requested name exists in `nativeFiles`. */\nconst resolvePackageSelection = (\n  nativeFiles: readonly string[],\n  packages: AnonymizeWasmPackages,\n): PackageSelection => {\n  const packageFiles = nativeFiles.filter(isPackageFile);\n  if (packages === \"all\") {\n    return { status: \"ok\", emit: new Set(packageFiles) };\n  }\n  if (packages === \"none\") {\n    return { status: \"ok\", emit: new Set() };\n  }\n  if (!Array.isArray(packages)) {\n    // A bare string like \"cs\" would otherwise iterate per character below.\n    return { status: \"invalid\", value: String(packages) };\n  }\n  const available = new Set(packageFiles);\n  const emit = new Set<string>();\n  const missing: string[] = [];\n  for (const name of packages) {\n    const fileName = packageFileName(name);\n    if (available.has(fileName)) {\n      emit.add(fileName);\n    } else {\n      missing.push(name);\n    }\n  }\n  if (missing.length > 0) {\n    return { status: \"missing\", names: missing, available: packageFiles };\n  }\n  return { status: \"ok\", emit };\n};\n\nexport default function stllAnonymizeWasmVite(\n  options: AnonymizeWasmPluginOptions = {},\n): Plugin {\n  const packages = options.packages ?? \"all\";\n  let isBuild = false;\n  return {\n    name: \"stll-anonymize-wasm\",\n    // optimizeDeps only affects dev; harmless (ignored) during build.\n    config() {\n      return { optimizeDeps: { exclude: OPTIMIZE_EXCLUDE } };\n    },\n    configResolved(resolved) {\n      isBuild = resolved.command === \"build\";\n    },\n    async transform(code, id) {\n      if (!(isBuild && isWasmEntry(id))) {\n        return null;\n      }\n      if (!code.includes(ASSET_URL_BASE)) {\n        this.error(\n          `stll-anonymize-wasm: could not find the assetUrl base in ${id}. ` +\n            \"The @stll/anonymize-wasm dist shape changed; update ASSET_URL_BASE \" +\n            \"in the Vite plugin.\",\n        );\n      }\n      const nativeDir = join(dirname(id), NATIVE_DIR);\n      const files = await readdir(nativeDir);\n      const selection = resolvePackageSelection(files, packages);\n      if (selection.status === \"invalid\") {\n        this.error(\n          `stll-anonymize-wasm: invalid packages option ${JSON.stringify(selection.value)}; ` +\n            `expected \"all\", \"none\", or an array like [\"default\", \"cs\"].`,\n        );\n      }\n      if (selection.status === \"missing\") {\n        this.error(\n          `stll-anonymize-wasm: requested package(s) not bundled: ` +\n            `${selection.names.join(\", \")}. Available: ${\n              selection.available.join(\", \") || \"(none)\"\n            }.`,\n        );\n      }\n      let anchorRef: string | undefined;\n      for (const file of files) {\n        if (isPackageFile(file) && !selection.emit.has(file)) {\n          continue;\n        }\n        const referenceId = this.emitFile({\n          type: \"asset\",\n          fileName: `${NATIVE_DIR}/${file}`,\n          source: await readFile(join(nativeDir, file)),\n        });\n        if (file === ANCHOR_FILE) {\n          anchorRef = referenceId;\n        }\n      }\n      if (anchorRef === undefined) {\n        this.error(\n          `stll-anonymize-wasm: ${ANCHOR_FILE} is missing from ${nativeDir}; ` +\n            \"cannot anchor the native asset base.\",\n        );\n      }\n      return {\n        code: code.replace(\n          ASSET_URL_BASE,\n          `fileName, import.meta.ROLLUP_FILE_URL_${anchorRef}`,\n        ),\n        map: null,\n      };\n    },\n  };\n}\n"],"mappings":";;;AAmCA,MAAM,mBAAmB,CAAC,wBAAwB,uBAAuB;AACzE,MAAM,aAAa;;;;AAInB,MAAM,cAAc;;;;AAIpB,MAAM,iBAAiB;;;AAIvB,MAAM,iBAAiB;AACvB,MAAM,iBAAiB;AACvB,MAAM,uBAAuB,GAAG,eAAe;;;AAG/C,MAAM,uBAAuB;AAoB7B,MAAM,eAAe,OACnB,GAAG,SAAS,gBAAgB,KAAK,SAAS,EAAE,MAAM;AAEpD,MAAM,iBAAiB,SAA0B,KAAK,SAAS,cAAc;;AAG7E,MAAM,mBAAmB,SACvB,SAAS,uBACL,uBACA,GAAG,iBAAiB,OAAO;;;AASjC,MAAM,2BACJ,aACA,aACqB;CACrB,MAAM,eAAe,YAAY,OAAO,aAAa;CACrD,IAAI,aAAa,OACf,OAAO;EAAE,QAAQ;EAAM,MAAM,IAAI,IAAI,YAAY;CAAE;CAErD,IAAI,aAAa,QACf,OAAO;EAAE,QAAQ;EAAM,sBAAM,IAAI,IAAI;CAAE;CAEzC,IAAI,CAAC,MAAM,QAAQ,QAAQ,GAEzB,OAAO;EAAE,QAAQ;EAAW,OAAO,OAAO,QAAQ;CAAE;CAEtD,MAAM,YAAY,IAAI,IAAI,YAAY;CACtC,MAAM,uBAAO,IAAI,IAAY;CAC7B,MAAM,UAAoB,CAAC;CAC3B,KAAK,MAAM,QAAQ,UAAU;EAC3B,MAAM,WAAW,gBAAgB,IAAI;EACrC,IAAI,UAAU,IAAI,QAAQ,GACxB,KAAK,IAAI,QAAQ;OAEjB,QAAQ,KAAK,IAAI;CAErB;CACA,IAAI,QAAQ,SAAS,GACnB,OAAO;EAAE,QAAQ;EAAW,OAAO;EAAS,WAAW;CAAa;CAEtE,OAAO;EAAE,QAAQ;EAAM;CAAK;AAC9B;AAEA,SAAwB,sBACtB,UAAsC,CAAC,GAC/B;CACR,MAAM,WAAW,QAAQ,YAAY;CACrC,IAAI,UAAU;CACd,OAAO;EACL,MAAM;EAEN,SAAS;GACP,OAAO,EAAE,cAAc,EAAE,SAAS,iBAAiB,EAAE;EACvD;EACA,eAAe,UAAU;GACvB,UAAU,SAAS,YAAY;EACjC;EACA,MAAM,UAAU,MAAM,IAAI;GACxB,IAAI,EAAE,WAAW,YAAY,EAAE,IAC7B,OAAO;GAET,IAAI,CAAC,KAAK,SAAS,cAAc,GAC/B,KAAK,MACH,4DAA4D,GAAG,yFAGjE;GAEF,MAAM,YAAY,KAAK,QAAQ,EAAE,GAAG,UAAU;GAC9C,MAAM,QAAQ,MAAM,QAAQ,SAAS;GACrC,MAAM,YAAY,wBAAwB,OAAO,QAAQ;GACzD,IAAI,UAAU,WAAW,WACvB,KAAK,MACH,gDAAgD,KAAK,UAAU,UAAU,KAAK,EAAE,8DAElF;GAEF,IAAI,UAAU,WAAW,WACvB,KAAK,MACH,0DACK,UAAU,MAAM,KAAK,IAAI,EAAE,eAC5B,UAAU,UAAU,KAAK,IAAI,KAAK,SACnC,EACL;GAEF,IAAI;GACJ,KAAK,MAAM,QAAQ,OAAO;IACxB,IAAI,cAAc,IAAI,KAAK,CAAC,UAAU,KAAK,IAAI,IAAI,GACjD;IAEF,MAAM,cAAc,KAAK,SAAS;KAChC,MAAM;KACN,UAAU,GAAG,WAAW,GAAG;KAC3B,QAAQ,MAAM,SAAS,KAAK,WAAW,IAAI,CAAC;IAC9C,CAAC;IACD,IAAI,SAAS,aACX,YAAY;GAEhB;GACA,IAAI,cAAc,KAAA,GAChB,KAAK,MACH,wBAAwB,YAAY,mBAAmB,UAAU,uCAEnE;GAEF,OAAO;IACL,MAAM,KAAK,QACT,gBACA,yCAAyC,WAC3C;IACA,KAAK;GACP;EACF;CACF;AACF"}