How to call a workflow that requires 2 strings and 1 file as inputs by uploading files and values simultaneously via the "HTTP Request" node in a Dify conversation flow, or are there any other solutions?General Discussion

[wuyufu0]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
• [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:)
• Please do not modify this template :) and fill in all the required fields.

Content

How to call a workflow that requires 2 strings and 1 file as inputs by uploading files and values simultaneously via the "HTTP Request" node in a Dify conversation flow, or are there any other solutions?

[dosubot[bot]]

Hi @wuyufu0! I'm Dosu and I’m helping the dify team.

You can upload 2 strings and 1 file simultaneously in a Dify workflow using the HTTP Request node by setting the method to POST, adding an Authorization header, and choosing "form-data" as the body type. In the body, add a field of type "file" for your file, and fields of type "text" for your string values (for example, a "data" field containing JSON or separate fields for each string) cite.

Note that files larger than 800KB are not supported in the HTTP Request node; for larger files, use a Python code node instead cite.

Alternatively, you can first upload the file to Dify (using the /v1/files/upload endpoint), get the file ID, and then call your workflow API (like /v1/workflows/run) with the file ID and string parameters in a JSON payload via a POST request with Content-Type: application/json cite.

If you need to handle multiple files, use an iteration node or code node to process each file individually, as the HTTP Request node only supports single file uploads cite.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}    

[wuyufu0]

@dosu
In Method 1 you mentioned, no matter whether I specify the Content-Type as multipart/form-data (I have tried both uppercase and lowercase), the output always shows application/json, which results in a 415 error.
In Method 2, since I couldn't find the ID returned by the HTTP request in the HTTP node, I wrote a Python script to retrieve the ID. However, the result returned a 400 error, and I don't know the reason.

[dosubot[bot]]

Here's what's happening:

415 error: The /v1/workflows/run endpoint only accepts application/json with file IDs, not multipart/form-data. If you try to send files and strings together as form-data, you'll get a 415 error because the API expects a JSON payload, not raw file uploads. The correct format is a JSON body with an inputs object containing your string values and the file ID, not the file itself (source).

Content-Type issues: If you manually set Content-Type: multipart/form-data, the HTTP client (like httpx, used by Dify) can't add the required boundary, which causes 400/415 errors. You should let the client set this header automatically when uploading files (source, source).

400 error: When uploading a file, the /files/upload endpoint only accepts a single file, not an array. If your request sends multiple files or the wrong structure, you'll get a 400 error (source). Also, if you try to call /v1/workflows/run with a malformed JSON payload (missing inputs or wrong file reference), you'll get a 400.

How to do it:

1. Upload your file first using /v1/files/upload (POST, form-data, field name file).
2. Extract the file ID from the upload response (use a code node if you can't do this directly in the HTTP node).
3. Call /v1/workflows/run with a POST request, Content-Type: application/json, and a body like:

   {
     "inputs": {
       "year": "2024",
       "month": "04",
       "knowledge_query": "your_file_id_here"
     }
   }

   Replace "your_file_id_here" with the actual file ID from step 2.

Common pitfalls:

• Don't manually set Content-Type: multipart/form-data when uploading files; let the client handle it (source).
• Don't try to upload files directly to /v1/workflows/run; always use file IDs in JSON.
• Make sure your JSON payload matches what the workflow expects.

If you need to pass the file ID between nodes, use a code node to extract it from the upload response and pass it as a variable to the workflow call.

Let me know if you need a concrete example or help with node configuration!

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}    

[wuyufu0]

In the first method you mentioned, regardless of whether I set the Content-Type to multipart/form-data (I've tried both uppercase and lowercase versions), the Content-Type in the output is always application/json, which results in a 415 error.
For the second method, since I couldn't find the ID returned by the HTTP request in the HTTP node, I wrote a Python script to obtain this ID. However, the result returned a 400 error, and I don't understand what's causing this, even though I've ensured the variables are correct.@dosu