OpenAI #Chat: Answer Text Prompt

// Script Example of Business Process Automation
// for 'engine type: 3' ("GraalJS standard mode")
// cf. 'engine type: 2' ("GraalJS Nashorn compatible mode") (renamed from "GraalJS" at 20230526)

//////// START "main()" /////////////////////////////////////////////////////////////////

main();
function main(){ 

////// == Config Retrieving / 工程コンフィグの参照 ==
const strAuthzSetting           = configs.get      ( "AuthzConfU1" );  /// REQUIRED
  engine.log( " AutomatedTask Config: Authz Setting: " + strAuthzSetting );
const strOrgId                  = configs.get      ( "StrConfU2" );    // NotRequired
  engine.log( " AutomatedTask Config: OpenAI-Organization: " + strOrgId );
const strEndUserId              = configs.get      ( "StrConfU3" ) !== "" ?  // NotRequired
                                  configs.get      ( "StrConfU3" ) :
                                  "m" + processInstance.getProcessModelInfoId().toString(); // (default)
  engine.log( " AutomatedTask Config: End User IDs: " + strEndUserId );

const strModel                  = configs.get      ( "StrConfM" ) !== "" ?   // NotRequired
                                  configs.get      ( "StrConfM" ) : "gpt-4-1106-preview"; // (default)
  engine.log( " AutomatedTask Config: OpenAI Model: " + strModel );

const strSystemMsg              = configs.get      ( "StrConfA1" );    // NotRequired
const strUserMsg                = configs.get      ( "StrConfA2" );    // NotRequired
  if( strUserMsg === "" && strSystemMsg    === ""){
    throw new Error( "\n AutomatedTask ConfigError:" +
                     " Config {A1:SystemMsg} or {A2:UserMsg} required \n" );
  }
const strChoises                = configs.get      ( "StrConfB1" );    // NotRequired
const numChoises                = isNaN(parseInt(strChoises,10)) ?
                                  1 : parseInt(strChoises,10);
const strLimit                  = configs.get      ( "StrConfB2" );    // NotRequired
const numLimit                  = isNaN(parseInt(strLimit,10)) ?
                                  4095 : parseInt(strLimit,10);
const strStops                  = configs.get      ( "StrConfB3" );    // NotRequired
const arrStops                  = strStops !== "" ?
                                  strStops.split("\n") : null;

const strPocketResponse1        = configs.getObject( "SelectConfC1" ); /// REQUIRED
const strPocketResponse2        = configs.getObject( "SelectConfC2" ); // NotRequired
const strPocketResponse3        = configs.getObject( "SelectConfC3" ); // NotRequired
const strPocketResponse4        = configs.getObject( "SelectConfC4" ); // NotRequired
const numPocketPromptTokens     = configs.getObject( "SelectConfD1" ); // NotRequired
const numPocketCompletionTokens = configs.getObject( "SelectConfD2" ); // NotRequired
const numPocketTotalTokens      = configs.getObject( "SelectConfD3" ); // NotRequired
const strPocketFinishReasons    = configs.getObject( "SelectConfD4" ); // NotRequired



////// == Data Retrieving / ワークフローデータの参照 ==
// (Nothing. Retrieved via Expression Language in Config Retrieving)


////// == Calculating / 演算 ==

//// OpenAI API > Documentation > API REFERENCE > CHAT
//// https://platform.openai.com/docs/api-reference/chat

/// prepare json
let strJson = {};
    strJson.model = strModel;
    strJson.user  = strEndUserId;
    //  strJson.response_format = {};
    //  strJson.response_format.type = "json_object"; // valid JSON mode
    //  To Make response_content JSON ('json' required in Request Msg)
    strJson.n          = numChoises;
    strJson.max_tokens = numLimit;
    if ( arrStops !== null ){
      strJson.stop = [];
      // Up to 4 sequences where the API will stop generating further tokens.
      const numMaxSeq = 4;
      for ( let i = 0; i < arrStops.length && i < numMaxSeq; i++ ){
        if ( arrStops[i] === "- - -" ){
          strJson.stop.push ( "\n" );
        }else{
          strJson.stop.push ( arrStops[i] );
        }
      }
    }
    strJson.messages = [];
    if ( strSystemMsg !=="" ) {
      let objSystemMsg = {};
          objSystemMsg.role = "system";
          objSystemMsg.content = strSystemMsg;
      strJson.messages.push ( objSystemMsg );
    }
    if ( strUserMsg !=="" ) {
      let objUserMsg = {};
          objUserMsg.role = "user";
          objUserMsg.content = strUserMsg;
      strJson.messages.push ( objUserMsg );
    }

/// prepare request1
let request1Uri = "https://api.openai.com/v1/chat/completions";
let request1 = httpClient.begin(); // HttpRequestWrapper
    request1 = request1.authSetting( strAuthzSetting ); // with "Authorization: Bearer XX"
    if ( strOrgId !== "" ){
      request1 = request1.header( "OpenAI-Organization", strOrgId );
    }
    request1 = request1.body( JSON.stringify( strJson ), "application/json" );

/// try request1
const response1     = request1.post( request1Uri ); // HttpResponseWrapper
engine.log( " AutomatedTask ApiRequest1 Start: " + request1Uri );
const response1Code = response1.getStatusCode() + ""; // JavaNum to string
const response1Body = response1.getResponseAsString();
engine.log( " AutomatedTask ApiResponse1 Status: " + response1Code );
if( response1Code !== "200"){
  throw new Error( "\n AutomatedTask UnexpectedResponseError: " +
                    response1Code + "\n" + response1Body + "\n" );
}


/// parse response1
/* engine.log( response1Body ); // debug
{
  "id": "chatcmpl-8JF1v1NheMIZfeX2AcBjKunKUMN8p",
  "object": "chat.completion",
  "created": 1699596699,
  "model": "gpt-4-1106-preview",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "お客様、この話題につ ... ... "
      },
      "finish_reason": "stop"
    },
    {
      "index": 1,
      "message": {
        "role": "assistant",
        "content": "この度の新聞記事に関し ... ... "
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 1678,
    "completion_tokens": 1401,
    "total_tokens": 3079
  },
  "system_fingerprint": "fp_a24b4d720c"
}
*/
const response1Obj = JSON.parse( response1Body );

let arrFinishReasons = [];
for ( let i = 0; i < response1Obj.choices.length; i++ ){
  arrFinishReasons.push ( response1Obj.choices[i].finish_reason );
}


////// == Data Updating / ワークフローデータへの代入 ==

if( strPocketResponse1 !== null ){
  engine.setData( strPocketResponse1,
                  response1Obj.choices[0]?.message.content ?? ""
                );  // optional chaining - nullish coalescing
}
if( strPocketResponse2 !== null ){
  engine.setData( strPocketResponse2, response1Obj.choices[1]?.message.content ?? "" );
}
if( strPocketResponse3 !== null ){
  engine.setData( strPocketResponse3, response1Obj.choices[2]?.message.content ?? "" );
}
if( strPocketResponse4 !== null ){
  engine.setData( strPocketResponse4, response1Obj.choices[3]?.message.content ?? "" );
}

if( numPocketPromptTokens !== null ){
  engine.setData( numPocketPromptTokens, new java.math.BigDecimal(
                  response1Obj.usage.prompt_tokens ?? 0
                ));
}
if( numPocketCompletionTokens !== null ){
  engine.setData( numPocketCompletionTokens, new java.math.BigDecimal(
                  response1Obj.usage.completion_tokens ?? 0
                ));
}
if( numPocketTotalTokens !== null ){
  engine.setData( numPocketTotalTokens, new java.math.BigDecimal(
                  response1Obj.usage.total_tokens ?? 0
                ));
}
// "??": Nullish coalescing operator (ES11)
// https://developer.mozilla.org/docs/Web/JavaScript/Reference/Operators/Nullish_coalescing

if( strPocketFinishReasons !== null ){
  engine.setData( strPocketFinishReasons, arrFinishReasons.join('\n') );
}

} //////// END "main()" /////////////////////////////////////////////////////////////////


/*
Notes:
- This [Automated Step] obtains the Text Response via OpenAI API (Chat endpoint).
    - Up to 4 Text can be genarated (default: 1)
- If place this [Automated Atep] in the workflow diagram, communication will occur every time a process arrives.
    - Request from the Questetra BPM Suite server to the OpenAI server.
    - Analyzes the response from the OpenAI server and stores the necessary information.
- [HTTP Authz Settings] is required for workflow apps that include this [Automated Step].
    - An API key is required to use OpenAI API. Please obtain an API key in advance.
        - https://platform.openai.com/api-keys
    - Set 'Secret API Key' as communication token. [HTTP Authz Settings] > [Token Fixed Value]
- Model endpoint compatibility (as of Nov 2023)
    - `gpt-4` (dated model releases)
    - `gpt-4-1106-preview`
    - `gpt-4-32k` (dated model releases)
    - `gpt-3.5-turbo` (dated model releases)
    - `gpt-3.5-turbo-16k` (dated model releases)
    - fine-tuned versions of `gpt-3.5-turbo`
    - see: https://platform.openai.com/docs/models/model-endpoint-compatibility (/v1/chat/completions)
- GPT-4 Turbo with 128K context
    - `gpt-4-1106-preview`: a preview of the next generation of GPT-4 (GPT-4 Turbo)
        - the first version of GPT-4 in March 2023 (available to all developers in July 2023)
    - has knowledge of world events up to April 2023.
        - CEO Altman, "We will try to never let it get that out of date again." 
    - has a 128k context window so it can fit the equivalent of more than 300 pages of text in a single prompt.
    - at a 3x cheaper price for input tokens and a 2x cheaper price for output tokens compared to GPT-4.

APPENDIX
- Large amount of allowable response sentences to be generated may exceed the system limit.
    - Generation will be aborted if the allowable amount is reached.
        - `"finish_reason":"length",`
        - If a large number is set for "Number of Responses," the response is more likely to be interrupted.
    - For English, one word or symbol often counts as one token.
        - For average English, it is about 1 token for 4 characters.
        - In Japanese, a single character may be divided into multiple tokens.
        - In the case of average Japanese, a single character is about one token.
    - You can check the approximate number of word tokens at tokenizer.
        - https://platform.openai.com/tokenizer
- Stop Sequences can be set up to four (invalid after the fifth line)
    - If set `\t` or other characters, they will be escaped.
    - To set a newline code (`\n`), set `- - -`. (experimental)
        - The response will always be a single line.
- Note that the workflow app ID is automatically assigned to the `user` parameter of the OpenAI API. (experimental)
    - `processInstance.getProcessModelInfoId()`
- In settings that refer to numeric type data, be careful not to mix in digit separators.
    - The formatting function `#sformat` is useful. (Java String.format)
    - e.g. `#{#sformat("%1.1f", #q_numeric)}` (rounded to one decimal place)
    - "R2272: Output of Strings via EL"
        - https://questetra.zendesk.com/hc/en-us/articles/360024292872-R2272-Output-of-Strings-via-EL-syntax
- If the number of responses is set to more than one, "Finish Reasons" will be in multiple lines.
     - If multiple lines are expected, set the data item to be stored as a multiline string.

Notes-ja:
- この［自動工程］は、OpenAI API （Chat エンドポイント）を通じて、回答文章を取得します。
    - この［自動工程］で生成できる文章は最大4つです。（デフォルト：１）
- この［自動工程］をワークフロー図に配置すれば、案件到達の度に通信が発生します。
    - Questetra BPM Suite サーバから OpenAI サーバに対してリクエストします。
    - OpenAI サーバからのレスポンスを解析し、必要情報を格納します。
- この［自動工程］を含むワークフローアプリには、［HTTP 認証設定］が必要です。
    - OpenAI API の利用には API key が必要です。あらかじめ API Key を取得しておいてください。
        - https://platform.openai.com/api-keys
    - 'Secret API Key' を通信トークンとしてセットします。［HTTP 認証設定］＞［トークン直接指定］
- 対応 MODEL (2023年11月現在)
    - `gpt-4` (dated model releases)
    - `gpt-4-1106-preview`
    - `gpt-4-32k` (dated model releases)
    - `gpt-3.5-turbo` (dated model releases)
    - `gpt-3.5-turbo-16k` (dated model releases)
    - fine-tuned versions of `gpt-3.5-turbo`
    - see: https://platform.openai.com/docs/models/model-endpoint-compatibility (/v1/chat/completions)
- 128K コンテキストを備えた GPT-4 Turbo とは？
    - `gpt-4-1106-preview`: 次世代 GPT-4 (GPT-4 Turbo) のプレビュー版
        - GPT-4 の初版は 2023年3月リリース (2023年7月に全開発者が利用可能)
    - 2023年4月までの出来事に知識がある。
        - アルトマンCEO、「2度と時代遅れにならないようにする」
    - 128k コンテキストで、300ページを超えるテキストに相当する情報を1プロンプトに収納可。
    - GPT-4 と比較して、入力トークン価格が3倍、出力トークンの価格が2倍、安くなった。

APPENDIX-ja
- レスポンス文の生成許容量を大きく設定した場合、システム制限を超える可能性があります。
    - 文章の途中であっても許容量に到達すれば生成が中断されます。
        - `"finish_reason":"length",`
        - "レスポンス数" に大きな数字をセットすると、レスポンス中断の可能性が高くなります。
    - 英語の場合、単語や記号ごとに1トークンになります。
        - 平均的な英語の場合、4文字で1トークン程度になります。
        - 日本語の場合、1文字が複数トークンに分割されることもあります。
        - 平均的な日本語の場合、1文字で1トークン程度になります。
    - ワードトークン数の目安は tokenizer にて確認できます。
        - https://platform.openai.com/tokenizer
- "中断文字列" としてセットできる文節は４つまでです。（5行目以降は無効）
    - "中断文字列" に `\t` 等を設定した場合、エスケープ処理されます。
    - "中断文字列" に改行コード（`\n`）をセットしたい場合、`- - -` をセットします。（試験的）
        - レスポンスは常に一行になります。
- ワークフローアプリIDが、OpenAI API の `user` パラメータに自動的に代入されます。（試験的）
    - `processInstance.getProcessModelInfoId()`
- 数値型データを参照する設定では、桁区切り文字が混入しないように注意してください。
    - フォーマット関数 `#sformat` を使うと便利です。（Java String.format）
    - e.g. `#{#sformat("%1.1f", #q_numeric)}` （小数第一位まで四捨五入）
    - "R2272: 文字列としての出力"
        - https://questetra.zendesk.com/hc/ja/articles/360024292872-R2272
- レスポンス数の設定が複数の場合、"生成終了理由" が複数行になります。
    - 複数行が想定される場合、格納するデータ項目は複数行文字列を設定してください。
*/