gentle-pi 4.3.0 → 4.4.0

package/lib/vim-editor-adapter.ts

@@ -32,6 +32,8 @@ interface PrivateEditor {
   onChange?: (text: string) => void;
   pastes: Map<number, string>;
   pasteCounter: number;
+  /** Host-owned atomic spans (gentle-shell#2464 image chips) that NORMAL motions must not split. */
+  atomicSpans?(line: string): Array<{ start: number; end: number }>;
   undoStack: Stack;
   history: string[];
   historyIndex: number;
@@ -151,12 +153,13 @@ function highlightRenderedRow(row: string, chunk: LayoutLine, left: number, righ
   return result;
 }
 
-function boundaries(line: string, pastes: Map<number, string>): number[] {
+function boundaries(line: string, editor: Pick<PrivateEditor, "pastes" | "atomicSpans">): number[] {
   const units = [...segmenter.segment(line)].map((part) => part.index);
-  for (const match of line.matchAll(marker)) {
-    if (!pastes.has(Number(match[1]))) continue;
-    const start = match.index;
-    const end = start + match[0].length;
+  const spans = [...line.matchAll(marker)]
+    .filter((match) => editor.pastes.has(Number(match[1])))
+    .map((match) => ({ start: match.index, end: match.index + match[0].length }));
+  if (typeof editor.atomicSpans === "function") spans.push(...editor.atomicSpans(line));
+  for (const { start, end } of spans) {
     for (let i = units.length - 1; i >= 0; i--) {
       if (units[i]! > start && units[i]! < end) units.splice(i, 1);
     }
@@ -177,7 +180,7 @@ function assertUniquePastes(editor: PrivateEditor): void {
 function assertPosition(editor: PrivateEditor, pos: Position): void {
   assertUniquePastes(editor);
   const line = editor.state.lines[pos.line];
-  if (line === undefined || !Number.isInteger(pos.col) || !boundaries(line, editor.pastes).includes(pos.col)) {
+  if (line === undefined || !Number.isInteger(pos.col) || !boundaries(line, editor).includes(pos.col)) {
     throw new Error("Cursor must be on a valid grapheme or paste boundary");
   }
 }
@@ -265,7 +268,7 @@ export function createVimEditorAdapter(value: unknown, version: string, editorCl
     },
     motionBoundaries(): number[][] {
       assertUniquePastes(editor);
-      return editor.state.lines.map((line) => boundaries(line, editor.pastes));
+      return editor.state.lines.map((line) => boundaries(line, editor));
     },
     move(pos: Position): void {
       assertPosition(editor, pos);
@@ -277,7 +280,7 @@ export function createVimEditorAdapter(value: unknown, version: string, editorCl
     moveByGraphemes(delta: number): void {
       if (!Number.isInteger(delta)) throw new Error("Invalid motion");
       const positions = editor.state.lines.flatMap((line, index) =>
-        boundaries(line, editor.pastes).map((col) => ({ line: index, col })));
+        boundaries(line, editor).map((col) => ({ line: index, col })));
       const current = editor.getCursor();
       const at = positions.findIndex((p) => p.line === current.line && p.col === current.col);
       if (at < 0) throw new Error("Invalid cursor boundary");

package/lib/voice-artifacts.ts

@@ -0,0 +1,188 @@
+// Pinned artifacts for the optional local voice dictation (S1-S3). Nothing here is
+// fetched or resolved at runtime: every version, size and digest below was read
+// from the public npm registry and Hugging Face metadata for the exact versions,
+// and the tarball layouts were inspected for the facts the comments state.
+// Installation lives in voice-install.ts; this module is data only.
+
+export type VoiceTarget = "darwin-x64" | "darwin-arm64" | "linux-x64" | "linux-arm64" | "win32-x64";
+
+export const VOICE_TARGETS: readonly VoiceTarget[] = ["darwin-x64", "darwin-arm64", "linux-x64", "linux-arm64", "win32-x64"];
+
+// A redirect hop is followed only when its URL is HTTPS on this origin and its
+// path starts with the prefix. Anything else ends the download.
+export interface VoiceRedirectRule {
+	origin: string;
+	pathPrefix: string;
+}
+
+export interface VoiceNpmPackage {
+	name: string;
+	version: string;
+	license: string;
+	tarball: string;
+	// The packed tarball size in bytes and its registry `dist.integrity`.
+	size: number;
+	integrity: string;
+	// Registry `dist.unpackedSize` and `dist.fileCount`: hard limits for extraction.
+	unpackedSize: number;
+	fileCount: number;
+	// Paths relative to the package root that must be present after extraction.
+	requiredFiles: string[];
+}
+
+export interface VoiceRuntimeManifest {
+	// The installed directory name; also part of the install marker.
+	id: string;
+	target: VoiceTarget;
+	packages: VoiceNpmPackage[];
+	redirects: VoiceRedirectRule[];
+}
+
+export interface VoiceModelFile {
+	name: string;
+	size: number;
+	sha256: string;
+}
+
+export interface VoiceModelManifest {
+	id: string;
+	revision: string;
+	// Files are fetched from `${baseUrl}/${name}`.
+	baseUrl: string;
+	files: VoiceModelFile[];
+	redirects: VoiceRedirectRule[];
+}
+
+const SHERPA_VERSION = "1.13.8";
+
+const npmTarball = (name: string, version: string) => `https://registry.npmjs.org/${name}/-/${name}-${version}.tgz`;
+
+// sherpa-onnx-node holds only JavaScript. Its addon-static-import.js loads
+// `../sherpa-onnx-<platform>-<arch>/sherpa-onnx.node` (Windows spelled `win`), so the
+// platform package must be installed as a sibling directory under node_modules.
+export const SHERPA_ONNX_NODE: VoiceNpmPackage = {
+	name: "sherpa-onnx-node",
+	version: SHERPA_VERSION,
+	license: "Apache-2.0",
+	tarball: npmTarball("sherpa-onnx-node", SHERPA_VERSION),
+	size: 11954,
+	integrity: "sha512-MsDMBdhLFTZ1GwvcGSSQhnS7g/EA8OMH6IYysCVUOM7j8Icty9KRc0E6YT1A5fWBsZwRfKOeh88QC95aRvS8ag==",
+	unpackedSize: 61221,
+	fileCount: 19,
+	requiredFiles: ["package.json", "sherpa-onnx.js", "addon.js", "addon-static-import.js", "non-streaming-asr.js"],
+};
+
+function sherpaPlatform(target: VoiceTarget, size: number, integrity: string, unpackedSize: number, fileCount: number): VoiceNpmPackage {
+	const name = `sherpa-onnx-${target.replace(/^win32-/, "win-")}`;
+	return {
+		name,
+		version: SHERPA_VERSION,
+		license: "Apache-2.0",
+		tarball: npmTarball(name, SHERPA_VERSION),
+		size,
+		integrity,
+		unpackedSize,
+		fileCount,
+		requiredFiles: ["package.json", "sherpa-onnx.node"],
+	};
+}
+
+// Each platform package bundles sherpa-onnx.node, the sherpa-onnx C/C++ API
+// libraries and onnxruntime, all built for the named architecture.
+export const SHERPA_ONNX_PLATFORM_PACKAGES: Readonly<Record<VoiceTarget, VoiceNpmPackage>> = {
+	"darwin-x64": sherpaPlatform("darwin-x64", 11191481, "sha512-7BLRpjM6w4f9W46/nmkmq8lEKUayhebvcpslCVQ+6QN2uReYlZEMDZlSpXMjme+hUFrPfRz8P3UNq8ep/4d19g==", 37678686, 7),
+	"darwin-arm64": sherpaPlatform("darwin-arm64", 10047754, "sha512-FPNgJMgnWVl/KhRTIhG3KL3A4Om63Rn4YKXc9/uHY7SzLcvqLJLc/h7UBWJwduXvv7K18t5NpxHR6XgXn4sjWw==", 34089738, 7),
+	"linux-x64": sherpaPlatform("linux-x64", 11089653, "sha512-6plnhjagsSeTntCgnlag86hWbs/uZE9Crms1LgOb68/1nKsIQjMd+WG519m+aPwT6TrsBOiEMzrx41t8sL5L5g==", 33381117, 7),
+	"linux-arm64": sherpaPlatform("linux-arm64", 13910679, "sha512-Tlg7a70b/Wge3OF8IgTHF9jhSVCsLyKQKhwc4BsJ5A+dL/SrFtGBjzuHp4XeLhiiOT7afCxX5PdSn/D4c8Lnuw==", 40306856, 7),
+	"win32-x64": sherpaPlatform("win32-x64", 8894875, "sha512-oZF1c9VPOKtMwn83Bboc5XSWL+76BRoyB3eUuVnCknBKxwSULZU2Foia9VHWzU+n4I12rPsP6z6H9Rp1hD9o8g==", 23453246, 8),
+};
+
+// node-cpal has no dependencies or install scripts and needs Node >= 22. Its
+// tarball ships every prebuilt binary; the default backend loads
+// bin/<platform>-<arch>/index.node through facade.js. The five binary headers were
+// inspected (not loaded) and each is built for its own architecture. Published
+// from saeta-eth/node-cpal tag v1.0.0, commit 20e2ac8d1922de7c4af50431e2ddc21a421cf764.
+// 0.1.1 is not used: its darwin-x64 and linux-arm64 binaries were built for the
+// other architecture.
+const NODE_CPAL_VERSION = "1.0.0";
+
+export const NODE_CPAL: VoiceNpmPackage = {
+	name: "node-cpal",
+	version: NODE_CPAL_VERSION,
+	license: "ISC",
+	tarball: npmTarball("node-cpal", NODE_CPAL_VERSION),
+	size: 7723259,
+	integrity: "sha512-d71jIzKDjw1HepqaJQMPd+/k3ZLqYZWhLAv6a8aafIGnYoA1QWlShA5iB5CgijIJtQxq9PdQAk99JQKspqlnTA==",
+	unpackedSize: 20363834,
+	fileCount: 40,
+	requiredFiles: ["package.json", "index.js", "facade.js", "cpal-values.js"],
+};
+
+// The oldest glibc the pinned Linux native binaries accept.
+export const VOICE_LINUX_MIN_GLIBC = "2.34";
+
+export function voiceRuntimeManifest(target: VoiceTarget): VoiceRuntimeManifest {
+	const packages = [
+		SHERPA_ONNX_NODE,
+		SHERPA_ONNX_PLATFORM_PACKAGES[target],
+		{ ...NODE_CPAL, requiredFiles: [...NODE_CPAL.requiredFiles, `bin/${target}/index.node`] },
+	];
+	// The id names both native versions, so a runtime installed with other pins
+	// lives in another directory and is never reported as this one.
+	// npm serves these tarballs directly; no redirect is expected or followed.
+	return { id: `sherpa-onnx-${SHERPA_VERSION}-node-cpal-${NODE_CPAL_VERSION}-${target}`, target, packages, redirects: [] };
+}
+
+const PARAKEET_REPO = "csukuangfj/sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8";
+const PARAKEET_REVISION = "2bda32ec70b097a55adaa07d9a7173915b43cc78";
+
+// The int8 sherpa-onnx export of NVIDIA Parakeet TDT 0.6B v3 at a fixed commit.
+// Observed redirects: tokens.txt answers 307 with a same-origin resolve-cache path,
+// the LFS/Xet files answer 302 to us.aws.cdn.hf.co/xet-bridge-us/. Other CDN
+// regions were not observed, so they are refused rather than guessed.
+export const PARAKEET_MODEL_MANIFEST: VoiceModelManifest = {
+	id: `parakeet-tdt-0.6b-v3-int8-${PARAKEET_REVISION.slice(0, 12)}`,
+	revision: PARAKEET_REVISION,
+	baseUrl: `https://huggingface.co/${PARAKEET_REPO}/resolve/${PARAKEET_REVISION}`,
+	files: [
+		{ name: "encoder.int8.onnx", size: 652184281, sha256: "acfc2b4456377e15d04f0243af540b7fe7c992f8d898d751cf134c3a55fd2247" },
+		{ name: "decoder.int8.onnx", size: 11845275, sha256: "179e50c43d1a9de79c8a24149a2f9bac6eb5981823f2a2ed88d655b24248db4e" },
+		{ name: "joiner.int8.onnx", size: 6355277, sha256: "3164c13fc2821009440d20fcb5fdc78bff28b4db2f8d0f0b329101719c0948b3" },
+		{ name: "tokens.txt", size: 93939, sha256: "d58544679ea4bc6ac563d1f545eb7d474bd6cfa467f0a6e2c1dc1c7d37e3c35d" },
+	],
+	redirects: [
+		{ origin: "https://huggingface.co", pathPrefix: `/api/resolve-cache/models/${PARAKEET_REPO}/${PARAKEET_REVISION}/` },
+		{ origin: "https://us.aws.cdn.hf.co", pathPrefix: "/xet-bridge-us/" },
+	],
+};
+
+export interface VoiceNotice {
+	component: string;
+	license: string;
+	source: string;
+	attribution?: string;
+	// What is known to be missing from the distributed artifact, stated plainly.
+	unresolved?: string;
+}
+
+export const VOICE_NOTICES: readonly VoiceNotice[] = [
+	{
+		component: "parakeet-tdt-0.6b-v3",
+		license: "CC-BY-4.0",
+		source: "https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3",
+		attribution: "Parakeet TDT 0.6B v3 by NVIDIA, licensed under CC-BY-4.0; int8 sherpa-onnx export by csukuangfj.",
+		unresolved: `The ${PARAKEET_REPO} export has no license file or model card license of its own; the license is taken from the upstream NVIDIA model.`,
+	},
+	{
+		component: "sherpa-onnx-node",
+		license: "Apache-2.0",
+		source: "https://github.com/k2-fsa/sherpa-onnx",
+		unresolved: "The sherpa-onnx npm tarballs ship no LICENSE or NOTICE file, and the platform packages bundle onnxruntime (MIT) without its license text.",
+	},
+	{
+		component: "node-cpal",
+		license: "ISC",
+		source: "https://github.com/saeta-eth/node-cpal",
+		unresolved: "The node-cpal@1.0.0 tarball ships no LICENSE or NOTICE file, and its prebuilt binaries link Rust crates whose notices are not included.",
+	},
+];