CODEWITHSUNDEEP
jsonazure-form-recognizer
SonOfNye
SonOfNye

Reputation: 75

API call to analyze a custom form

I am trying to make an API call to Microsoft's form recognizer to analyze a form against a custom model and I can't figure out how to do it.

Here is the documentation on the API

https://westus2.dev.cognitive.microsoft.com/docs/services/form-recognizer-api/operations/AnalyzeWithCustomModel

The request body is blank and I don't really know how that ought to be formatted in order to be sent off.

If you look at the POST Train model method, I was able to use that request body to send make that api call work. This indicates that the problem is me and not the API.

I have successfully done this with curl through command line...

curl -X POST "https://formrecognizerbp.cognitiveservices.azure.com/formrecognizer/v1.0-preview/custom/models/[MODEL ID]/analyze" -H "Content-Type: multipart/form-data" -F "form=@\"C:\Temp\Capture1.jpg\";type=image/jpeg" -H "Ocp-Apim-Subscription-Key: [SUBSCRIPTION ID]"

I don't really know/can't figure out how to convert that into a request body similar in format to what the POST Train Model method has.

I keep getting this error because I don't know how to format the request properly.

Internal : Unexpected error Error during Web API HTTP Request HTTP Status Code: 400 HTTP Response Content: {"value":{"error":{"code":"BadRequest","message":"Could not process incoming request: 'Missing content-type boundary.'. Please ensure that it is well-formed."}},"formatters":[],"contentTypes":[],"statusCode":400}

So I guess formatters and contentTypes are the missing pieces but would that just be the file path and the image/jpeg parts of the curl ?

Upvotes: 0

Views: 468

Answers (2)

SonOfNye
SonOfNye

Reputation: 75

The issue was that I did not realize that the curl script was overwritting the multipart/form-data Content-Type with image/jpeg and when I was trying to build out this call differently I was forcing a multipart/form-data Content-Type on what was a jpeg.

Upvotes: 0

Nimisha
Nimisha

Reputation: 36

The /trainCustomModel API expects the data to be present on the Azure Blob storage. The request body to this request needs to contain a valid SAS URL to the training data. Once you successfully create a custom trained model ID, you could use that to analyze the forms. The /AnalyzeWithCustomModel API expects the data to be on your local file storage. Please ensure that you have the replaced the ModelId, API Subscription Key (Note this is not the same as subscription ID) and the local path to the image correctly.

Upvotes: 0

Related Questions