WebExtrator Task Query API Integration Guide

POST https://api.acedata.cloud/webextrator/tasks

The WebExtrator Task Query API is used to query the results of historical render / extract tasks. Common usages include:

Callback to check the complete envelope after an asynchronous task is completed (besides callback_url push or active polling).
Audit what has been submitted — task records store both the original request and the final response.
Batch refill — pull multiple records at once by id or trace_id.

Task records are retained in Redis for 7 days.

The task query interface is free (not counted towards Credits usage).

¶ Authentication

Authorization: Bearer YOUR_API_KEY
Content-Type:  application/json

You can only query tasks under your own AceDataCloud account.

¶ Request Parameters

The request body is a discriminative union based on action, with two types of actions:

¶ `action: "retrieve"` — Single Query

Field	Type	Required	Description
`action`	const	✅	Fixed `"retrieve"`.
`id`	string	One of	Task ID (appears in the `task_id` field of each render/extract envelope).
`trace_id`	string	One of	Call chain ID (the `trace_id` field of the envelope).

Either id or trace_id must be provided.

¶ `action: "retrieve_batch"` — Batch Query

Field	Type	Required	Description
`action`	const	✅	Fixed `"retrieve_batch"`.
`ids`	string[]	One of	List of task IDs.
`trace_ids`	string[]	One of	List of call chain IDs.
`offset`	number	❌	Pagination offset (default 0).
`limit`	number	❌	Page size, 1–100 (default 50).

Either ids or trace_ids must be provided.

¶ Single Response

{
  "task": {
    "id": "550e8400-e29b-41d4-a716-446655440000",
    "trace_id": "550e8400-e29b-41d4-a716-446655440001",
    "type": "extract",
    "created_at": 1777717800.05,
    "started_at": 1777717800.123,
    "finished_at": 1777717802.535,
    "elapsed": 2.412,
    "request": {
      "url": "https://en.wikipedia.org/wiki/Diffbot",
      "expected_type": "article"
    },
    "response": {
      "success": true,
      "data": { /* complete extract envelope */ }
    }
  }
}

If not found, it returns { "task": null } (HTTP 200, not 404).

The timing fields of the task object are described as follows.

created_at, task creation time, Unix timestamp (seconds, float).
started_at, task execution start time, Unix timestamp (seconds, float). It is null when the task has not started.
finished_at, task completion time, Unix timestamp (seconds, float). It is null when the task is not completed.
elapsed, task execution duration, in seconds (float, rounded to 3 decimal places). It is null when the task is not completed.

¶ Batch Response

{
  "tasks": [
    { /* same structure as single .task */ },
    { /* ... */ }
  ],
  "offset": 0,
  "limit":  50
}

Non-existent IDs will not cause an error, they will simply be missing from tasks.

¶ Examples

¶ Query Single by task_id

curl -X POST https://api.acedata.cloud/webextrator/tasks \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "action": "retrieve",
    "id": "550e8400-e29b-41d4-a716-446655440000"
  }'

¶ Query Single by trace_id

curl -X POST https://api.acedata.cloud/webextrator/tasks \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "action": "retrieve",
    "trace_id": "550e8400-e29b-41d4-a716-446655440001"
  }'

¶ Batch Query

curl -X POST https://api.acedata.cloud/webextrator/tasks \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "action": "retrieve_batch",
    "ids": [
      "550e8400-e29b-41d4-a716-446655440000",
      "550e8400-e29b-41d4-a716-446655440002"
    ],
    "limit": 50
  }'

¶ Python (requests) — Polling Until Completion

import os, time, requests

API_KEY = os.environ["ACEDATA_API_KEY"]
BASE = "https://api.acedata.cloud"

# 1) Submit asynchronous extraction
queue = requests.post(
    f"{BASE}/webextrator/extract",
    headers={"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"},
    json={"url": "https://example.com", "mode": "async"},
).json()

job_id = queue["jobId"]

# 2) Use Tasks API to poll until the task is completed
while True:
    r = requests.post(
        f"{BASE}/webextrator/tasks",
        headers={"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"},
        json={"action": "retrieve", "id": job_id},
    ).json()
    task = r.get("task")
    if task and task.get("finished_at"):
        print("Elapsed time", task["elapsed"], "seconds")
        print(task["response"]["data"]["title"])
        break
    time.sleep(2)

¶ Node.js (fetch) — Pull Complete Envelope After Receiving Callback

// In your callback_url handler:
app.post('/hooks/webextrator', async (req, res) => {
  res.status(200).end();              // Quick ack first

  const taskId = req.body?.task_id;
  if (!taskId) return;

  const fetchRes = await fetch('https://api.acedata.cloud/webextrator/tasks', {
    method: 'POST',
    headers: {
      Authorization: `Bearer ${process.env.ACEDATA_API_KEY}`,
      'Content-Type': 'application/json',
    },
    body: JSON.stringify({ action: 'retrieve', id: taskId }),
  });
  const { task } = await fetchRes.json();
  console.log('Complete envelope:', task.response.data);
});

¶ Error Responses

HTTP	`error.code`	Meaning
400	`bad_request`	Validation failed (missing `action`, both `id` and `trace_id` provided, etc.).
401	`unauthorized`	Missing or invalid `Authorization: Bearer …`.

{ "error": { "code": "bad_request", "message": "..." } }

¶ Tips and Pitfalls

If you can customize trace_id, then do so. Upload it in the original render/extract request ?trace_id=… (QueryString), aligning it with your own business ID (workflow run id, etc.), and then you can query tasks using the business ID. If not provided, the server will automatically generate a UUID.
Retention period is 7 days. Tasks older than this will return task: null — if long-term archiving is needed, please store it in your own database.
Task queries are free. You can query as many times as you want; the fees for the original render/extract calls have already been paid.
Prefer using asynchronous + callbacks instead of polling. If business allows, pass callback_url in the original request, allowing the platform to push the envelope to you, which is more efficient than polling every 2 seconds.