import { readFileSync } from "fs";

export async function jsonToAgent(filePath, agentPromptFile='agent.txt') {


    console.log('processing file:', filePath);

    let content;
    let simple_content;
    const fileRes = readFileSync(filePath);
    if (fileRes) {
        const filecontent = fileRes.toString();
        
        // simplify content
        let jsonData;
        try {
            jsonData = JSON.parse(filecontent);
        } catch (err) {
            console.error("Failed to parse JSON file:", err);
            return;
        }
        content = jsonData;

        let simplify = { input: [] };
        simplify.input.push(jsonData.thread.text);
        jsonData.replies.forEach(element => {
            simplify.input.push(element.text);
        });

        simple_content = JSON.stringify(simplify);
        // console.log(simple_content);

    } else {
        console.error("Failed to load JSON file");
        return;
    }

    // return;
    console.log('sending to agent...', content.replies.length + 1);

    // 使用 OpenAI Chat Completions API 搭配 Structured Outputs (json_schema)
    const response = await fetch('https://api.openai.com/v1/responses', {
        method: 'POST',
        headers: {
            'Content-Type': 'application/json',
            'Authorization': `Bearer ${process.env.OPENAI_API_KEY}`
        },
        body: JSON.stringify({
            // model: "gpt-4.1-mini",
            // input: [
            //     { role: "system", content: system_prompt.replaceAll("{{N}}", content.replies.length + 1) },
            //     { role: "user", content: simple_content }
            // ],
            prompt:{
                id: "pmpt_69607acd9c10819696e7a25fd66cdd6b016fc7e2ef2c3ca5",
                // version: "4",
                variables:{
                    count: (content.replies.length + 1).toString(),
                    content: simple_content
                }
            },
            store: false,
            text: {
                format:{
                    type: "json_schema",
                    name: "conversation_analysis",
                    strict: true,
                    schema: {
                        type: "object",
                        properties: {
                            output: {
                                type: "array",
                                description: "包含對話中所有留言的分析陣列",
                                items: {
                                    type: "object",
                                    properties: {
                                        summry: { 
                                            type: "string", 
                                            description: "簡短精煉的全域對話主題（20字以內）。注意：此欄位在所有 array items 中必須完全一致。" 
                                        },
                                        keywords: { 
                                            type: "array", 
                                            items: { type: "string" },
                                            description: "從留言中提取的關鍵字（事件、概念、情緒等）"
                                        },
                                        number: { 
                                            type: "integer", 
                                            description: "目前留言的序號" 
                                        },
                                        total: { 
                                            type: "integer", 
                                            description: "留言總數 count" 
                                        },
                                        content: { 
                                            type: "string", 
                                            description: "原始留言內容" 
                                        },
                                        teaser:{
                                            type: "string",
                                            description: "10字內具備共鳴的核心情緒"
                                        }                                       
                                    },
                                    required: ["summry", "keywords", "number", "total", "content", "teaser"],
                                    additionalProperties: false
                                }
                            }
                        },
                        required: ["output"],
                        additionalProperties: false
                    }
                }
            }
        })
    });

   
    const data = await response.json();

    
    if (data.error) {
        console.error("API Error:", data.error);
        return null;
    }



     try {
        // 從 choices[0].message.content 中解析 AI 回傳的 JSON 字串
        let output_content = data.output.find(item => item.type === "message");
        if (!output_content) {
            console.error("No message content found in agent response:", data);
            return data;
        }
        // console.log("Raw agent content:", output_content);
        // parse JSON
        let output = JSON.parse(output_content.content[0].text);
        // console.log("Agent output:", output);

        // add metadata info
        output.output= output.output.map((item, index) => {

            let metadata= index==0? content.thread : content.replies[index -1];

            return {
                ...item,
                metadata: metadata
            };
        });
        

        return output;
    } catch (err) {
        console.error("Failed to parse agent response as JSON:", err);

        return data;
    }
}

export async function parseTest(filepath){

    const fileRes = readFileSync(filepath);
    if (!fileRes) {
        console.error("Failed to load JSON file");
        return;
    }
    const filecontent = fileRes.toString();

    let data;
    try {
        data = JSON.parse(filecontent);
    } catch (err) {
        console.error("Failed to parse JSON file:", err);
        return;
    }
    

    try {
        // 從 choices[0].message.content 中解析 AI 回傳的 JSON 字串
        let output_content = data.output.find(item => item.type === "message");
        if (!output_content) {
            console.error("No message content found in agent response:", data);
            return data;
        }
        // console.log("Raw agent content:", output_content);
        // parse JSON
        let output = JSON.parse(output_content.content[0].text);
        // console.log("Agent output:", output);

        // add metadata info
        output.output= output.output.map((item, index) => {

            let metadata= index==0? content.thread : content.replies[index -1];

            return {
                ...item,
                metadata: metadata
            };
        });
        

        return output;
    } catch (err) {
        console.error("Failed to parse agent response as JSON:", err);

        return data;
    }
}