feat(camera): add snap/clip capture

This commit is contained in:
Peter Steinberger
2025-12-14 00:48:58 +00:00
parent 2454e67e09
commit a92eb1f33d
19 changed files with 1669 additions and 2 deletions

View File

@@ -0,0 +1,64 @@
import * as fs from "node:fs/promises";
import * as os from "node:os";
import * as path from "node:path";
import { describe, expect, it } from "vitest";
import {
cameraTempPath,
parseCameraClipPayload,
parseCameraSnapPayload,
writeBase64ToFile,
} from "./nodes-camera.js";
describe("nodes camera helpers", () => {
it("parses camera.snap payload", () => {
expect(
parseCameraSnapPayload({
format: "jpg",
base64: "aGk=",
width: 10,
height: 20,
}),
).toEqual({ format: "jpg", base64: "aGk=", width: 10, height: 20 });
});
it("rejects invalid camera.snap payload", () => {
expect(() => parseCameraSnapPayload({ format: "jpg" })).toThrow(
/invalid camera\.snap payload/i,
);
});
it("parses camera.clip payload", () => {
expect(
parseCameraClipPayload({
format: "mp4",
base64: "AAEC",
durationMs: 1234,
hasAudio: true,
}),
).toEqual({
format: "mp4",
base64: "AAEC",
durationMs: 1234,
hasAudio: true,
});
});
it("builds stable temp paths when id provided", () => {
const p = cameraTempPath({
kind: "snap",
facing: "front",
ext: "jpg",
tmpDir: "/tmp",
id: "id1",
});
expect(p).toBe(path.join("/tmp", "clawdis-camera-snap-front-id1.jpg"));
});
it("writes base64 to file", async () => {
const dir = await fs.mkdtemp(path.join(os.tmpdir(), "clawdis-test-"));
const out = path.join(dir, "x.bin");
await writeBase64ToFile(out, "aGk=");
await expect(fs.readFile(out, "utf8")).resolves.toBe("hi");
await fs.rm(dir, { recursive: true, force: true });
});
});

92
src/cli/nodes-camera.ts Normal file
View File

@@ -0,0 +1,92 @@
import { randomUUID } from "node:crypto";
import * as fs from "node:fs/promises";
import * as os from "node:os";
import * as path from "node:path";
export type CameraFacing = "front" | "back";
export type CameraSnapPayload = {
format: string;
base64: string;
width: number;
height: number;
};
export type CameraClipPayload = {
format: string;
base64: string;
durationMs: number;
hasAudio: boolean;
};
function asRecord(value: unknown): Record<string, unknown> {
return typeof value === "object" && value !== null
? (value as Record<string, unknown>)
: {};
}
function asString(value: unknown): string | undefined {
return typeof value === "string" ? value : undefined;
}
function asNumber(value: unknown): number | undefined {
return typeof value === "number" && Number.isFinite(value)
? value
: undefined;
}
function asBoolean(value: unknown): boolean | undefined {
return typeof value === "boolean" ? value : undefined;
}
export function parseCameraSnapPayload(value: unknown): CameraSnapPayload {
const obj = asRecord(value);
const format = asString(obj.format);
const base64 = asString(obj.base64);
const width = asNumber(obj.width);
const height = asNumber(obj.height);
if (!format || !base64 || width === undefined || height === undefined) {
throw new Error("invalid camera.snap payload");
}
return { format, base64, width, height };
}
export function parseCameraClipPayload(value: unknown): CameraClipPayload {
const obj = asRecord(value);
const format = asString(obj.format);
const base64 = asString(obj.base64);
const durationMs = asNumber(obj.durationMs);
const hasAudio = asBoolean(obj.hasAudio);
if (
!format ||
!base64 ||
durationMs === undefined ||
hasAudio === undefined
) {
throw new Error("invalid camera.clip payload");
}
return { format, base64, durationMs, hasAudio };
}
export function cameraTempPath(opts: {
kind: "snap" | "clip";
facing?: CameraFacing;
ext: string;
tmpDir?: string;
id?: string;
}) {
const tmpDir = opts.tmpDir ?? os.tmpdir();
const id = opts.id ?? randomUUID();
const facingPart = opts.facing ? `-${opts.facing}` : "";
const ext = opts.ext.startsWith(".") ? opts.ext : `.${opts.ext}`;
return path.join(
tmpDir,
`clawdis-camera-${opts.kind}${facingPart}-${id}${ext}`,
);
}
export async function writeBase64ToFile(filePath: string, base64: string) {
const buf = Buffer.from(base64, "base64");
await fs.writeFile(filePath, buf);
return { path: filePath, bytes: buf.length };
}

View File

@@ -1,6 +1,13 @@
import type { Command } from "commander";
import { callGateway, randomIdempotencyKey } from "../gateway/call.js";
import { defaultRuntime } from "../runtime.js";
import {
type CameraFacing,
cameraTempPath,
parseCameraClipPayload,
parseCameraSnapPayload,
writeBase64ToFile,
} from "./nodes-camera.js";
type NodesRpcOpts = {
url?: string;
@@ -12,6 +19,11 @@ type NodesRpcOpts = {
params?: string;
invokeTimeout?: string;
idempotencyKey?: string;
facing?: string;
maxWidth?: string;
quality?: string;
duration?: string;
audio?: boolean;
};
type NodeListNode = {
@@ -340,4 +352,203 @@ export function registerNodesCli(program: Command) {
}),
{ timeoutMs: 30_000 },
);
const parseFacing = (value: string): CameraFacing => {
const v = String(value ?? "")
.trim()
.toLowerCase();
if (v === "front" || v === "back") return v;
throw new Error(`invalid facing: ${value} (expected front|back)`);
};
const camera = nodes
.command("camera")
.description("Capture camera media from a paired node");
nodesCallOpts(
camera
.command("snap")
.description("Capture a photo from a node camera (prints MEDIA:<path>)")
.requiredOption("--node <idOrNameOrIp>", "Node id, name, or IP")
.option("--facing <front|back|both>", "Camera facing", "both")
.option("--max-width <px>", "Max width in px (optional)")
.option("--quality <0-1>", "JPEG quality (default 0.9)")
.option(
"--invoke-timeout <ms>",
"Node invoke timeout in ms (default 20000)",
"20000",
)
.action(async (opts: NodesRpcOpts) => {
try {
const nodeId = await resolveNodeId(opts, String(opts.node ?? ""));
const facingOpt = String(opts.facing ?? "both")
.trim()
.toLowerCase();
const facings: CameraFacing[] =
facingOpt === "both"
? ["front", "back"]
: facingOpt === "front" || facingOpt === "back"
? [facingOpt]
: (() => {
throw new Error(
`invalid facing: ${String(opts.facing)} (expected front|back|both)`,
);
})();
const maxWidth = opts.maxWidth
? Number.parseInt(String(opts.maxWidth), 10)
: undefined;
const quality = opts.quality
? Number.parseFloat(String(opts.quality))
: undefined;
const timeoutMs = opts.invokeTimeout
? Number.parseInt(String(opts.invokeTimeout), 10)
: undefined;
const results: Array<{
facing: CameraFacing;
path: string;
width: number;
height: number;
}> = [];
for (const facing of facings) {
const invokeParams: Record<string, unknown> = {
nodeId,
command: "camera.snap",
params: {
facing,
maxWidth: Number.isFinite(maxWidth) ? maxWidth : undefined,
quality: Number.isFinite(quality) ? quality : undefined,
format: "jpg",
},
idempotencyKey: randomIdempotencyKey(),
};
if (typeof timeoutMs === "number" && Number.isFinite(timeoutMs)) {
invokeParams.timeoutMs = timeoutMs;
}
const raw = (await callGatewayCli(
"node.invoke",
opts,
invokeParams,
)) as unknown;
const res =
typeof raw === "object" && raw !== null
? (raw as { payload?: unknown })
: {};
const payload = parseCameraSnapPayload(res.payload);
const filePath = cameraTempPath({
kind: "snap",
facing,
ext: payload.format === "jpeg" ? "jpg" : payload.format,
});
await writeBase64ToFile(filePath, payload.base64);
results.push({
facing,
path: filePath,
width: payload.width,
height: payload.height,
});
}
if (opts.json) {
defaultRuntime.log(JSON.stringify({ files: results }, null, 2));
return;
}
defaultRuntime.log(results.map((r) => `MEDIA:${r.path}`).join("\n"));
} catch (err) {
defaultRuntime.error(`nodes camera snap failed: ${String(err)}`);
defaultRuntime.exit(1);
}
}),
{ timeoutMs: 60_000 },
);
nodesCallOpts(
camera
.command("clip")
.description(
"Capture a short video clip from a node camera (prints MEDIA:<path>)",
)
.requiredOption("--node <idOrNameOrIp>", "Node id, name, or IP")
.option("--facing <front|back>", "Camera facing", "front")
.option("--duration <ms>", "Duration in ms (default 3000)", "3000")
.option("--no-audio", "Disable audio capture")
.option(
"--invoke-timeout <ms>",
"Node invoke timeout in ms (default 45000)",
"45000",
)
.action(async (opts: NodesRpcOpts & { audio?: boolean }) => {
try {
const nodeId = await resolveNodeId(opts, String(opts.node ?? ""));
const facing = parseFacing(String(opts.facing ?? "front"));
const durationMs = Number.parseInt(
String(opts.duration ?? "3000"),
10,
);
const includeAudio = opts.audio !== false;
const timeoutMs = opts.invokeTimeout
? Number.parseInt(String(opts.invokeTimeout), 10)
: undefined;
const invokeParams: Record<string, unknown> = {
nodeId,
command: "camera.clip",
params: {
facing,
durationMs: Number.isFinite(durationMs) ? durationMs : undefined,
includeAudio,
format: "mp4",
},
idempotencyKey: randomIdempotencyKey(),
};
if (typeof timeoutMs === "number" && Number.isFinite(timeoutMs)) {
invokeParams.timeoutMs = timeoutMs;
}
const raw = (await callGatewayCli(
"node.invoke",
opts,
invokeParams,
)) as unknown;
const res =
typeof raw === "object" && raw !== null
? (raw as { payload?: unknown })
: {};
const payload = parseCameraClipPayload(res.payload);
const filePath = cameraTempPath({
kind: "clip",
facing,
ext: payload.format,
});
await writeBase64ToFile(filePath, payload.base64);
if (opts.json) {
defaultRuntime.log(
JSON.stringify(
{
file: {
facing,
path: filePath,
durationMs: payload.durationMs,
hasAudio: payload.hasAudio,
},
},
null,
2,
),
);
return;
}
defaultRuntime.log(`MEDIA:${filePath}`);
} catch (err) {
defaultRuntime.error(`nodes camera clip failed: ${String(err)}`);
defaultRuntime.exit(1);
}
}),
{ timeoutMs: 90_000 },
);
}

View File

@@ -1,3 +1,4 @@
import * as fs from "node:fs/promises";
import { beforeEach, describe, expect, it, vi } from "vitest";
const sendCommand = vi.fn();
@@ -148,4 +149,145 @@ describe("cli program", () => {
);
expect(runtime.log).toHaveBeenCalled();
});
it("runs nodes camera snap and prints two MEDIA paths", async () => {
callGateway
.mockResolvedValueOnce({
ts: Date.now(),
nodes: [
{
nodeId: "ios-node",
displayName: "iOS Node",
remoteIp: "192.168.0.88",
connected: true,
},
],
})
.mockResolvedValueOnce({
ok: true,
nodeId: "ios-node",
command: "camera.snap",
payload: { format: "jpg", base64: "aGk=", width: 1, height: 1 },
})
.mockResolvedValueOnce({
ok: true,
nodeId: "ios-node",
command: "camera.snap",
payload: { format: "jpg", base64: "aGk=", width: 1, height: 1 },
});
const program = buildProgram();
runtime.log.mockClear();
await program.parseAsync(
["nodes", "camera", "snap", "--node", "ios-node"],
{
from: "user",
},
);
expect(callGateway).toHaveBeenNthCalledWith(
2,
expect.objectContaining({
method: "node.invoke",
params: expect.objectContaining({
nodeId: "ios-node",
command: "camera.snap",
timeoutMs: 20000,
idempotencyKey: "idem-test",
params: expect.objectContaining({ facing: "front", format: "jpg" }),
}),
}),
);
expect(callGateway).toHaveBeenNthCalledWith(
3,
expect.objectContaining({
method: "node.invoke",
params: expect.objectContaining({
nodeId: "ios-node",
command: "camera.snap",
timeoutMs: 20000,
idempotencyKey: "idem-test",
params: expect.objectContaining({ facing: "back", format: "jpg" }),
}),
}),
);
const out = String(runtime.log.mock.calls[0]?.[0] ?? "");
const mediaPaths = out
.split("\n")
.filter((l) => l.startsWith("MEDIA:"))
.map((l) => l.replace(/^MEDIA:/, ""))
.filter(Boolean);
expect(mediaPaths).toHaveLength(2);
try {
for (const p of mediaPaths) {
await expect(fs.readFile(p, "utf8")).resolves.toBe("hi");
}
} finally {
await Promise.all(mediaPaths.map((p) => fs.unlink(p).catch(() => {})));
}
});
it("runs nodes camera clip and prints one MEDIA path", async () => {
callGateway
.mockResolvedValueOnce({
ts: Date.now(),
nodes: [
{
nodeId: "ios-node",
displayName: "iOS Node",
remoteIp: "192.168.0.88",
connected: true,
},
],
})
.mockResolvedValueOnce({
ok: true,
nodeId: "ios-node",
command: "camera.clip",
payload: {
format: "mp4",
base64: "aGk=",
durationMs: 3000,
hasAudio: true,
},
});
const program = buildProgram();
runtime.log.mockClear();
await program.parseAsync(
["nodes", "camera", "clip", "--node", "ios-node", "--duration", "3000"],
{ from: "user" },
);
expect(callGateway).toHaveBeenNthCalledWith(
2,
expect.objectContaining({
method: "node.invoke",
params: expect.objectContaining({
nodeId: "ios-node",
command: "camera.clip",
timeoutMs: 45000,
idempotencyKey: "idem-test",
params: expect.objectContaining({
facing: "front",
durationMs: 3000,
includeAudio: true,
format: "mp4",
}),
}),
}),
);
const out = String(runtime.log.mock.calls[0]?.[0] ?? "");
const mediaPath = out.replace(/^MEDIA:/, "").trim();
expect(mediaPath).toMatch(/clawdis-camera-clip-front-.*\.mp4$/);
try {
await expect(fs.readFile(mediaPath, "utf8")).resolves.toBe("hi");
} finally {
await fs.unlink(mediaPath).catch(() => {});
}
});
});