Support list in response body

[mingshl]

Description

When an array as a string in parameter, the string in list are not escaped.

This PR will escape the string when it's a json array.

for example, before this change:

POST /_plugins/_ml/connectors/_create
{
  "name": "Amazon Bedrock Connector: Claude Instant V1",
  "version": "1",
  "description": "The connector to bedrock Claude model",
  "protocol": "aws_sigv4",
  "parameters": {
    "max_tokens_to_sample": "8000",
    "service_name": "bedrock",
    "temperature": "1.0E-4",
    "response_filter": "$.completion",
    "region": "us-west-2",
    "anthropic_version": "bedrock-2023-05-31",
    "inputs":"please summerize the documents"
  },
  "credential": {
    "access_key": " ",
        "secret_key": " ",
        "session_token": "" },
  "actions": [
    {
      "action_type": "PREDICT",
      "method": "POST",
      "url": "https://bedrock-runtime.us-west-2.amazonaws.com/model/anthropic.claude-instant-v1/invoke",
      "headers": {
        "x-amz-content-sha256": "required",
        "content-type": "application/json"
      },
      "request_body":  "{\"prompt\":\"${parameters.prompt}\",\"max_tokens_to_sample\":300,\"temperature\":0.5,\"top_k\":250,\"top_p\":1,\"stop_sequences\":[\"\\n\\nHuman:\"]}"
    }
  ]
}


POST /_plugins/_ml/models/_register
{
  "name": "Bedrock agent model with prompt",
  "function_name": "remote",
  "description": "test model",
  "connector_id": "0CwoD5EB6KAJXDLxYDvd"}
  
POST /_plugins/_ml/models/2SwoD5EB6KAJXDLxezto/_deploy

## success with string
POST /_plugins/_ml/models/2SwoD5EB6KAJXDLxezto/_predict  
{
  "parameters": {
    "prompt":"\n\nHuman: You are a professional data analysist. You will always answer question based on the given context first. If the answer is not directly shown in the context, you will analyze the data and find the answer. If you don't know the answer, just say I don't know. Context: ${parameters.context}. \n\n Human: please summarize the documents \n\n Assistant:",
    "context":"Dr. Eric Goldberg is a fantastic doctor who has correctly diagnosed every issue that my wife and I have had. Unlike many of my past doctors, Dr. Goldberg is very accessible and we have been able to schedule appointments with him and his staff very quickly. We are happy to have him in the neighborhood and look forward to being his patients for many years to come."
  }
}


##failed with list 

POST /_plugins/_ml/models/2SwoD5EB6KAJXDLxezto/_predict  
{
  "parameters": {
    "prompt":"\n\nHuman: You are a professional data analysist. You will always answer question based on the given context first. If the answer is not directly shown in the context, you will analyze the data and find the answer. If you don't know the answer, just say I don't know. Context: ${parameters.context}. \n\n Human: please summarize the documents \n\n Assistant:",
    "context":["Dr. Eric Goldberg is a fantastic doctor who has correctly diagnosed every issue that my wife and I have had. Unlike many of my past doctors, Dr. Goldberg is very accessible and we have been able to schedule appointments with him and his staff very quickly. We are happy to have him in the neighborhood and look forward to being his patients for many years to come."]
  }
}

Related Issues

#2839

Check List

• New functionality includes testing.
• New functionality has been documented.
• API changes companion pull request created.
• Commits are signed per the DCO using --signoff.
• Public documentation issue/PR created.

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[dhrubo-os]

I assume failing tests are related to this change?

[mingshl]

match by regex expression is not a good way and it only solves the type casting from jsonstring(list) into a jsonstring(escape list). I will rethink about a better way to do this

[mingshl]

came up with a better solution to try fix the invalid payload by adding escape,

in this fix, we don't worry if the parameter is a list or a map, we try help to fix the invalid payload by adding escape,

if adding escape helps, return the payload after adding escape,
else, throw the same illegal argument exception, and indicate the payload is still invalid.

please help review @Zhangxunmt @dhrubo-os

[dhrubo-os]

Test is failing

[mingshl]

Details

• What went wrong:
  Execution failed for task ':opensearch-ml-plugin:integTest'.

Could not resolve all files for configuration ':opensearch-ml-plugin:opensearch_distro_extracted_testclusters-opensearch-ml-plugin-integTest-0-3.0.0-SNAPSHOT-'.

Could not resolve opensearch-distribution-snapshot:opensearch:3.0.0-SNAPSHOT.

@dhrubo-os any ideas for the linux test failure? I saw the window tests passed all the way through. This failure is not related to the code changes in this PR.

[mingshl]

all tests passed, merging now.

[ylwu-amzn]

Why not use existing escape method? This escape method will replace "\n" in "This is a \new book" , which is not necessary. It may confuse user why same sentence will generate different embeddings after this PR

[mingshl]

it won't be a break change, because before this change, if it's an invalid payload, it will throw exception directly, it won't generate embedding at all.

[ylwu-amzn]

If payload is wrong will escape all parameters again. Escape good parameters again may generate wrong result. For example, the original prompt is "please replace all \n of the context with --", if we escape again , it becomes "please replace all \\n of the context with --"

[mingshl]

if a string is already partial escaped, meaning the string escaped for the next line symbol but doesn't escape for double quote, this change will still not fix the payload. Please let me know if there is better way to identify partial escape cases.