If you are a Coze VIP, you can directly download the workflow and import it

https://wwaod.lanzn.com/b011lwl8ij
Password:3qnp

Taking sora2 as an example, from inputting text and uploading images to finally getting the url, a simple loop workflow (the overall diagram is below)

1. Start

Add input variables according to the prompt and image, according to the parameters required by the interface. If more parameters need to be input, you can also add them
企业微信截图_1767773636845.png

2. Request model (add http node)

Configure the http request according to the documentation
1.The first red box configures the request method and url
2.The second red box configures the request headers
3.The third place configures the request body. Please choose the format according to the corresponding request body in the documentation. In this example, json is used, where the fields marked with start are parameters obtained from the previous input. You can directly enter {} in the editor to select them
企业微信截图_17677740268061.png

3. Process the request result to get taskId (add code node)

As shown in the figure, when inputting, directly get the output body of the previous step as the variable value
Use code to parse the body and get the id in it (in this example, JavaScript is selected for the code, as follows)

async function main({ params }: Args): Promise<Output> {
    // 1. Get the body string from params.input
    // Note: It is assumed here that params.input is the entire HTTP response object
    const bodyStr = params.input;

    // 2. Parse the string body into a JSON object
    // Because the value of body is a string "{\"id\":...}", JSON.parse is needed
    let taskId = "";
    try {
        const bodyObj = JSON.parse(bodyStr);
        taskId = bodyObj.id;
    } catch (e) {
        // Prevent runtime errors caused by parsing failure
        console.log("JSON parsing failed:", e);
    }

    // 3. Build the output object and return taskId
    const ret = {
        "taskId": taskId
    };

    return ret;
}

Please make the variable name in the final output position consistent with the parameter name in ret in the code
企业微信截图_17677742429398.png

4. Loop request query interface to get url (add loop node)

1. Set up the loop (in this example, an infinite loop is used and the loop is exited according to the judgment condition). Set it as needed

The intermediate variable is the taskId output by the previous code node
Outputs url and status
企业微信截图_17677745518897.png

2. Add an http node in the loop to request task query

The parameters and request body should be based on the query task of the model used in the documentation
企业微信截图_17677753812075.png

3. Add a code node to get url and task status

As shown in the figure, the input gets the output body of the query task. The code is as follows

async function main({ params }: Args): Promise<Output> {
    // 1. Get the body string from params.input
    const bodyStr = params.input;

    // 2. Parse the string body into a JSON object
    let url = "";
    let status = "";
    try {
        const bodyObj = JSON.parse(bodyStr);
        url = bodyObj.video_url;
        status = bodyObj.status;
    } catch (e) {
        // Prevent runtime errors caused by parsing failure
        console.log("JSON parsing failed:", e);
    }

    // 3. Build the output object and return taskId
    const ret = {
        "url": url,
        "status": status
    };

    return ret;
}

4.Add a judgment node to determine whether the query obtained a url

In this example, the judgment node sets two branches. The first branch checks whether the url output by the previous code node is empty
The second branch checks whether the status is failed

5.Execute according to the judgment result

Add a node to terminate the loop. When either the first judgment of the judgment node succeeds or the second judgment succeeds, terminate the loop
Add a timer node and a continue loop node. When both judgments of the judgment node are false, execute the timer first (to prevent request failure caused by concurrent requests, it is recommended to set it to 1500ms), and after the timer connect to continue loop

Five, process the result obtained by the loop (add a code node)
Because the result obtained after the loop is an array, processing is needed to get only the final output result (code as follows)
The input is the output result of the loop

async function main({ params }: Args): Promise<Output> {

    // Get input
    const urls = params.urls;
    const statuss = params.statuss;

    // 3. Build the output object and return taskId
    const ret = {
        "url": urls[urls?.length -1],
        "status": statuss[statuss?.length -1]
    };

    return ret;
}

Outputs the processed url and status
企业微信截图_17677761996893.png

6. Determine whether the url has a value to determine whether video generation succeeded or failed

7. Finally output the result

As shown in the figure, output the success and failure results and then end. You can output text or variables according to your needs, or put the output result in the end node for output
企业微信截图_17677765862529.png

8. Achieved effect

Coze workflow simple configuration from input to getting url

If you are a Coze VIP, you can directly download the workflow and import it#

Taking sora2 as an example, from inputting text and uploading images to finally getting the url, a simple loop workflow (the overall diagram is below)#

1. Start#

2. Request model (add http node)#

3. Process the request result to get taskId (add code node)#

4. Loop request query interface to get url (add loop node)#

1. Set up the loop (in this example, an infinite loop is used and the loop is exited according to the judgment condition). Set it as needed#

2. Add an http node in the loop to request task query#

3. Add a code node to get url and task status#

4.Add a judgment node to determine whether the query obtained a url#

5.Execute according to the judgment result#

6. Determine whether the url has a value to determine whether video generation succeeded or failed#

7. Finally output the result#

#

8. Achieved effect#