feat(task): replace built-in task adaptors with a sandboxed JS plugin system (#7076)

This commit is contained in:
Calcium-Ion
2026-08-29 18:51:57 +08:00
committed by GitHub
parent 7037ac15bd
commit eb48396d5f
336 changed files with 52333 additions and 6369 deletions
+501
View File
@@ -0,0 +1,501 @@
export const meta = {
apiVersion: 1,
key: "doubao",
name: "Doubao Video",
icon: "Doubao.Color",
description: {
en: "Volcengine Doubao Seedance video generation (text-to-video, image-to-video, and video-to-video)",
zh: "火山引擎豆包 Seedance 视频生成(文生视频、图生视频、视频生视频)",
},
version: "1.0.0",
author: { name: "QuantumNous" },
channelTypes: [54, 45], // VolcEngine-type channels serve Ark video models with the same wire format
models: [
"doubao-seedance-1-0-pro-250528",
"doubao-seedance-1-0-lite-t2v",
"doubao-seedance-1-0-lite-i2v",
"doubao-seedance-1-5-pro-251215",
"doubao-seedance-2-0-260128",
"doubao-seedance-2-0-fast-260128",
"doubao-seedance-2-0-mini-260615",
"doubao-seedance-2-5-260628",
],
fetchMode: "per_task",
usageSchema: {
tokens: {
type: "number",
unit: "token",
description: {
en: "Upstream billing tokens (estimated at submit, actual on completion).",
zh: "上游计费 token(提交时预估,完成后按实际值)。",
},
},
resolution: {
enum: ["480p", "720p", "1080p", "4k"],
description: {
en: "Output video resolution; Seedance token unit price varies by resolution tier.",
zh: "输出视频分辨率;Seedance token 单价随分辨率档位变化。",
},
},
video_input: {
enum: ["none", "video"],
description: {
en: "Whether the request includes reference video input; Seedance prices video-to-video tokens at a lower unit rate.",
zh: "请求是否包含参考视频输入;Seedance 对视频生视频 token 按更低单价计费。",
},
},
},
// Official Ark formula tokens = (input + output seconds) × W × H × 24 / 1024,
// 16:9 max-pixel sizes, cross-checked against Volcengine price examples.
usageExamples: [
{ label: "480p · 5s", facts: { tokens: 48038, resolution: "480p", video_input: "none" } },
{ label: "720p · 5s", facts: { tokens: 108000, resolution: "720p", video_input: "none" } },
{ label: "1080p · 5s", facts: { tokens: 243000, resolution: "1080p", video_input: "none" } },
{ label: "4k · 5s", facts: { tokens: 972000, resolution: "4k", video_input: "none" } },
{ label: "720p · 10s", facts: { tokens: 216000, resolution: "720p", video_input: "none" } },
{ label: "720p · 5s (+4s 输入视频)", facts: { tokens: 194400, resolution: "720p", video_input: "video" } },
],
routes: [
{ method: "POST", path: "/doubao/api/v3/contents/generations/tasks", type: "submit", decode: "createTask", render: "taskCreated" },
{ method: "GET", path: "/doubao/api/v3/contents/generations/tasks/:task_id", type: "query", render: "taskStatus" },
],
protocols: [{ name: "openai_responses", supports: ["stream", "sync", "background"] }, "openai_video"],
};
function trimmed(value) {
return String(value || "").trim();
}
function draftTaskIds(content) {
const ids = [];
if (!Array.isArray(content)) return ids;
for (const item of content) {
if (!item || typeof item !== "object" || Array.isArray(item)) continue;
if (item.type !== "draft_task") continue;
const draft = item.draft_task;
if (!draft || typeof draft !== "object" || Array.isArray(draft)) continue;
const id = trimmed(draft.id);
if (id) ids.push(id);
}
return ids;
}
function rewriteDraftTaskContent(content, originTasks) {
if (!Array.isArray(content)) return content;
return content.map(function (item) {
if (!item || typeof item !== "object" || Array.isArray(item) || item.type !== "draft_task") return item;
const draft = item.draft_task;
if (!draft || typeof draft !== "object" || Array.isArray(draft) || !trimmed(draft.id)) return item;
const publicId = trimmed(draft.id);
let upstream = "";
if (Array.isArray(originTasks)) {
for (const task of originTasks) {
if (task && task.taskId === publicId) {
upstream = trimmed(task.upstreamTaskId);
break;
}
}
}
if (!upstream) throw new Error("origin task is unavailable");
return Object.assign({}, item, { draft_task: Object.assign({}, draft, { id: upstream }) });
});
}
function normalizeResolution(value) {
const raw = trimmed(value).toLowerCase();
if (["480p", "720p", "1080p", "4k"].includes(raw)) return raw;
const parts = raw.replace("*", "x").split("x");
if (parts.length !== 2) return "720p";
const max = Math.max(Number(parts[0]), Number(parts[1]));
if (max >= 3840) return "4k";
if (max >= 1920) return "1080p";
if (max >= 1280) return "720p";
return "480p";
}
function hasVideo(content) {
return Array.isArray(content) && content.some((item) => item && (item.type === "video_url" || Object.prototype.hasOwnProperty.call(item, "video_url")));
}
// Max-pixel 16:9 dimensions per resolution tier. Used when ratio is absent or
// adaptive so the submit-time estimate overestimates rather than underestimates.
// Official Ark formula: tokens = seconds × width × height × 24 / 1024.
// Video input duration is omitted; extractUsageOnComplete overlays the real bill.
function resolutionMaxPixels(resolution) {
if (resolution === "480p") return [854, 480];
if (resolution === "1080p") return [1920, 1080];
if (resolution === "4k") return [3840, 2160];
return [1280, 720];
}
function estimateTokens(seconds, resolution) {
const dims = resolutionMaxPixels(resolution);
return (seconds * dims[0] * dims[1] * 24) / 1024;
}
function videoInputRatio(model, resolution, content) {
const video = hasVideo(content);
const res = trimmed(resolution).toLowerCase();
if (model === "doubao-seedance-2-5-260628") {
if (res === "1080p") return video ? 7.0 / 10.7 : 11.7 / 10.7;
return video ? 42 / 70 : 1;
}
if (model === "doubao-seedance-2-0-260128") {
if (res === "1080p") return video ? 31 / 46 : 51 / 46;
if (res === "4k") return video ? 16 / 46 : 26 / 46;
return video ? 28 / 46 : 1;
}
if (model === "doubao-seedance-2-0-fast-260128") return video ? 22 / 37 : 1;
if (model === "doubao-seedance-2-0-mini-260615") return video ? 14 / 23 : 1;
return 1;
}
function responsesInput(req) {
const texts = [],
images = [];
const input = req.input;
if (typeof input === "string") texts.push(input);
else if (Array.isArray(input)) {
for (const item of input) {
if (typeof item === "string") {
texts.push(item);
continue;
}
if (!item || typeof item !== "object" || Array.isArray(item)) continue;
const content = item.content === undefined ? [item] : Array.isArray(item.content) ? item.content : [item.content];
for (const part of content) {
if (typeof part === "string") {
texts.push(part);
continue;
}
if (!part || typeof part !== "object" || Array.isArray(part)) continue;
if (["input_text", "text"].includes(part.type) && typeof part.text === "string") texts.push(part.text);
if (["input_image", "image_url"].includes(part.type)) {
let image = part.image_url;
if (image && typeof image === "object") image = image.url;
if (trimmed(image)) images.push(trimmed(image));
}
}
}
}
return {
prompt: texts
.filter(function (text) {
return trimmed(text);
})
.join("\n"),
images: images,
};
}
function responsesVideoText(ctx) {
const artifact = ctx && ctx.artifacts && ctx.artifacts.video;
const url = trimmed(artifact && artifact.url);
if (!url) throw new Error("video artifact is unavailable");
const escaped = url.replace(/&/g, "&amp;").replace(/"/g, "&quot;").replace(/</g, "&lt;").replace(/>/g, "&gt;");
return '<video controls src="' + escaped + '"></video>';
}
export const native = {
createTask: function (ctx) {
if (!ctx.body || ctx.body.kind !== "json") throw new Error("JSON body required");
const body = ctx.body.value;
if (!body || typeof body !== "object" || Array.isArray(body)) throw new Error("request body must be an object");
const model = trimmed(body.model);
if (!model) throw new Error("model is required");
if (body.content !== undefined && !Array.isArray(body.content)) throw new Error("content must be an array");
const content = Array.isArray(body.content) ? body.content : [];
const texts = [];
let hasReference = false;
for (const item of content) {
if (!item || typeof item !== "object" || Array.isArray(item)) continue;
if (item.type === "text" && typeof item.text === "string") texts.push(item.text);
else hasReference = true;
}
if (!texts.length && !hasReference) throw new Error("content is required");
const requestBody = {
model: model,
prompt: texts
.filter(function (text) {
return trimmed(text);
})
.join("\n"),
metadata: body,
};
const seconds = Number(body.duration);
if (Number.isFinite(seconds) && seconds > 0) requestBody.seconds = seconds;
const intent = { kind: "submit", model: model, action: hasReference ? "image_to_video" : "text_to_video", requestBody: requestBody };
const originTaskIds = draftTaskIds(content);
if (originTaskIds.length) intent.originTaskIds = originTaskIds;
return intent;
},
taskCreated: function (ctx, task) {
const data = task.data && typeof task.data === "object" && !Array.isArray(task.data) ? task.data : {};
return Object.assign({}, data, { id: task.task_id });
},
taskStatus: function (ctx, task) {
if (task.data && typeof task.data === "object" && !Array.isArray(task.data)) return Object.assign({}, task.data, { id: task.task_id });
const statusMap = { NOT_START: "queued", SUBMITTED: "queued", QUEUED: "queued", IN_PROGRESS: "running", SUCCESS: "succeeded", FAILURE: "failed" };
const output = { id: task.task_id, status: statusMap[task.status] || "queued" };
if (task.fail_reason) output.error = { message: task.fail_reason };
return output;
},
error: function (ctx, error) {
return { error: { code: error.code, message: error.message } };
},
};
export function buildSubmitRequest(ctx) {
const req = ctx.requestBody;
const metadata = req.metadata || {};
const body = Object.assign({ model: req.model || "", content: [] }, metadata);
const imageContent = [];
const images = Array.isArray(req.images) ? req.images : [];
for (const url of images) imageContent.push({ type: "image_url", image_url: { url: url } });
const metadataContent = Array.isArray(body.content) ? body.content : [];
body.content = imageContent.concat(metadataContent).filter((item) => item && item.type !== "text");
const hasReference = body.content.length > 0;
if (trimmed(req.prompt) || !hasReference) body.content.push({ type: "text", text: req.prompt || "" });
if (Array.isArray(body.content)) body.content = rewriteDraftTaskContent(body.content, ctx.originTasks);
const seconds = Number.parseInt(req.seconds || "", 10);
if (seconds > 0) body.duration = seconds;
body.model = ctx.upstreamModel || body.model;
return {
url: ctx.baseUrl + "/api/v3/contents/generations/tasks",
method: "POST",
headers: { "Content-Type": "application/json", Accept: "application/json", Authorization: "Bearer " + ctx.apiKey },
body: body,
action: hasReference ? "image_to_video" : "text_to_video",
rewriteModel: body.model,
};
}
export function parseSubmitResponse(ctx, resp) {
if (!resp.body || !resp.body.id) throw new Error("task_id is empty");
return { taskId: resp.body.id, taskData: resp.body };
}
export function extractUsage(ctx) {
const req = ctx.requestBody || {};
const metadata = req.metadata || {};
if (ctx.usagePurpose === "billing_ratios") {
const ratio = videoInputRatio(ctx.model, metadata.resolution, metadata.content);
return ratio === 1 ? null : { video_input_ratio: ratio };
}
let seconds = Number(req.seconds || req.duration || metadata.duration || 0);
if (!Number.isFinite(seconds) || seconds <= 0) {
const frames = Number(metadata.frames);
seconds = Number.isFinite(frames) && frames > 0 ? Math.floor(frames / 24) : 15;
}
if (seconds <= 0) seconds = 5;
seconds = Math.min(seconds, 3600);
const rawResolution = metadata.resolution || req.size;
const raw = trimmed(rawResolution).toLowerCase();
const recognized = ["480p", "720p", "1080p", "4k"].includes(raw) || raw.replace("*", "x").split("x").length === 2;
const resolution = recognized ? normalizeResolution(rawResolution) : "1080p";
return {
tokens: estimateTokens(seconds, resolution),
resolution: resolution,
video_input: hasVideo(metadata.content) ? "video" : "none",
};
}
export function buildQueryRequest(ctx) {
return {
url: ctx.baseUrl + "/api/v3/contents/generations/tasks/" + ctx.taskId,
method: "GET",
headers: { Accept: "application/json", "Content-Type": "application/json", Authorization: "Bearer " + ctx.apiKey },
};
}
export function parseTaskResult(ctx, body) {
if (body.status === "pending" || body.status === "queued") return { status: "QUEUED", progress: "10%" };
if (body.status === "processing" || body.status === "running") return { status: "IN_PROGRESS", progress: "50%" };
if (body.status === "succeeded") {
const result = { status: "SUCCESS", progress: "100%", url: body.content && body.content.video_url ? body.content.video_url : "" };
const usage = body.usage || {};
const completionTokens = Number(usage.completion_tokens || 0);
const totalTokens = Number(usage.total_tokens || 0);
if (Number.isFinite(completionTokens) && completionTokens > 0) result.completionTokens = completionTokens;
if (Number.isFinite(totalTokens) && totalTokens > 0) result.totalTokens = totalTokens;
return result;
}
if (body.status === "failed" || body.status === "expired" || body.status === "cancelled") {
const reason = body.error && body.error.message ? body.error.message : body.status;
return { status: "FAILURE", progress: "100%", reason: reason };
}
return { status: "IN_PROGRESS", progress: "30%" };
}
function artifactData(ctx) {
const data = (ctx && ctx.data) || {};
if (data.data && typeof data.data === "object" && data.data.task_id && Object.prototype.hasOwnProperty.call(data.data, "data")) return data.data.data || {};
return data;
}
export function listArtifacts(task) {
if (task.status !== "SUCCESS") return [];
const content = artifactData(task).content || {};
const artifacts = [];
if (trimmed(content.video_url)) artifacts.push({ key: "video", type: "video" });
if (trimmed(content.last_frame_url)) artifacts.push({ key: "last_frame", type: "image", mimeType: "image/png" });
return artifacts;
}
export function buildContentRequest(ctx) {
const content = artifactData(ctx).content || {};
const urls = { video: content.video_url, last_frame: content.last_frame_url };
const url = trimmed(urls[ctx.artifactKey]);
if (!url) throw new Error("artifact_not_found");
return { url: url, method: ctx.clientRequest.method, credentialless: true };
}
export function extractUsageOnComplete(task, taskResult, body) {
if (!body || body.status !== "succeeded") return {};
const facts = {};
const usage = body.usage || {};
let tokens = Number(usage.completion_tokens);
if (!Number.isFinite(tokens) || tokens <= 0) tokens = Number(usage.total_tokens);
if (Number.isFinite(tokens) && tokens > 0) facts.tokens = tokens;
const content = body.content || {};
const resolution = trimmed(content.resolution || body.resolution).toLowerCase();
if (["480p", "720p", "1080p", "4k"].includes(resolution)) facts.resolution = resolution;
return facts;
}
export const protocols = {
openai_responses: {
decodeRequest: function (ctx) {
if (!ctx.body || ctx.body.kind !== "json") throw new Error("JSON body required");
const req = ctx.body.value;
if (!req || typeof req !== "object" || Array.isArray(req)) throw new Error("request body must be an object");
const model = trimmed(req.model);
if (!model) throw new Error("model is required");
if (req.input !== undefined && typeof req.input !== "string" && !Array.isArray(req.input)) throw new Error("input must be a string or array");
if (req.images !== undefined && !Array.isArray(req.images)) throw new Error("images must be an array");
if (req.metadata !== undefined && (!req.metadata || typeof req.metadata !== "object" || Array.isArray(req.metadata)))
throw new Error("metadata must be an object");
const input = responsesInput(req);
const prompt = input.prompt || trimmed(req.prompt);
const images = [];
for (const image of [req.image, req.input_reference].concat(req.images || [], input.images)) {
if (trimmed(image) && !images.includes(trimmed(image))) images.push(trimmed(image));
}
if (!prompt && images.length === 0) throw new Error("input is required");
const metadata = Object.assign({}, req.metadata || {});
if (Object.prototype.hasOwnProperty.call(req, "resolution")) metadata.resolution = req.resolution;
else if (req.size && !metadata.resolution) metadata.resolution = normalizeResolution(req.size);
const requestBody = { model: model, prompt: prompt, metadata: metadata };
if (images.length) requestBody.images = images;
if (Object.prototype.hasOwnProperty.call(req, "seconds")) requestBody.seconds = req.seconds;
else if (Object.prototype.hasOwnProperty.call(req, "duration")) requestBody.seconds = req.duration;
if (Object.prototype.hasOwnProperty.call(req, "size")) requestBody.size = req.size;
const intent = { kind: "submit", model: model, action: images.length ? "image_to_video" : "text_to_video", requestBody: requestBody };
const originTaskIds = draftTaskIds(metadata.content);
if (originTaskIds.length) intent.originTaskIds = originTaskIds;
return intent;
},
renderEvents: function (ctx, task, previousState) {
const status = String(task.status || "UNKNOWN").toUpperCase();
const value = Number(String(task.progress || "").replace("%", ""));
const progress = Number.isFinite(value) && value >= 0 && value <= 100 ? value : null;
const state = { status: status, progress: progress };
if (status === "SUCCESS") {
const text = responsesVideoText(ctx);
const events = previousState && previousState.status === status ? [] : [{ type: "output", data: text }];
return { events: events, state: state, done: true };
}
if (status === "FAILURE")
return { events: [{ type: "error", code: "task_failed", message: task.fail_reason || "task failed" }], state: state, done: true };
if (previousState && previousState.status === status && previousState.progress === progress) return { events: [], state: state, done: false };
const event = { type: "progress", message: status.toLowerCase() };
if (progress !== null) event.progress = progress;
return { events: [event], state: state, done: false };
},
renderFinal: function (ctx, _task) {
return {
output: [
{
type: "message",
status: "completed",
role: "assistant",
content: [{ type: "output_text", text: responsesVideoText(ctx), annotations: [], logprobs: [] }],
},
],
metadata: { vendor: "doubao" },
};
},
},
};
const legacyRenderers = {
openai_video: function (task) {
const data = task.data || {};
const statusMap = { NOT_START: "queued", SUBMITTED: "queued", QUEUED: "queued", IN_PROGRESS: "in_progress", SUCCESS: "completed", FAILURE: "failed" };
const output = {
id: task.task_id,
object: "video",
model: task.properties ? task.properties.origin_model_name || "" : "",
status: statusMap[task.status] || "unknown",
progress: Number(String(task.progress || "0").replace("%", "")),
created_at: task.created_at,
completed_at: task.updated_at,
};
if (data.status === "failed") output.error = { message: data.error ? data.error.message || "" : "", code: data.error ? data.error.code || "" : "" };
return output;
},
};
protocols.openai_video = {
decodeRequest: function (ctx) {
if (!ctx.body || (ctx.body.kind !== "json" && ctx.body.kind !== "multipart")) throw new Error("JSON or multipart body required");
if (ctx.body.kind === "json") {
if (!ctx.body.value || Array.isArray(ctx.body.value)) throw new Error("JSON object required");
const req = ctx.body.value;
const seconds = req.seconds === undefined ? req.duration : req.seconds;
if (seconds !== undefined && (!Number.isFinite(Number(seconds)) || Number(seconds) <= 0 || Number(seconds) > 3600))
throw new Error("seconds must be between 1 and 3600");
return {
kind: "submit",
model: ctx.model,
action: req.input_reference || req.image ? "image_to_video" : "text_to_video",
requestBody: Object.assign({}, req, { model: ctx.model }),
};
}
const first = function (name) {
const values = (ctx.body.fields || {})[name] || [];
if (values.length > 1) throw new Error(name + " must be provided once");
return values[0];
};
const req = {};
const fields = ctx.body.fields || {};
for (const name of Object.keys(fields)) {
req[name] = first(name);
}
if (req.metadata !== undefined) {
let parsed;
try {
parsed = JSON.parse(req.metadata);
} catch (e) {
throw new Error("metadata must be a JSON object string");
}
if (!parsed || typeof parsed !== "object" || Array.isArray(parsed)) throw new Error("metadata must be a JSON object string");
req.metadata = parsed;
}
if ((ctx.body.files || []).length) throw new Error("Doubao requires image and video references to be URLs inside metadata.content");
if (req.seconds !== undefined) req.seconds = Number(req.seconds);
else if (req.duration !== undefined) req.seconds = Number(req.duration);
const seconds = req.seconds === undefined ? req.duration : req.seconds;
if (seconds !== undefined && (!Number.isFinite(Number(seconds)) || Number(seconds) <= 0 || Number(seconds) > 3600))
throw new Error("seconds must be between 1 and 3600");
return {
kind: "submit",
model: ctx.model,
action: req.input_reference || req.image ? "image_to_video" : "text_to_video",
requestBody: Object.assign({}, req, { model: ctx.model }),
};
},
render: function (ctx, task) {
return legacyRenderers.openai_video(task);
},
};