services/report_service.py

import json
import logging
import os
import asyncio
import tempfile
from typing import List, Dict, Optional, Any
import openai

from core.config import settings
from core.prompts import get_report_prompt, get_report_suggestion_prompt
from services.s3_service import s3_service

logger = logging.getLogger(__name__)

class ReportService:
    def __init__(self):
        self.openai_client = openai.OpenAI(api_key=settings.OPENAI_API_KEY)

    async def generate_format_suggestions(
        self,
        file_key: Optional[str] = None,
        text_input: Optional[str] = None,
        language: str = "Japanese"
    ) -> List[Dict[str, str]]:
        """
        Generates 4 AI-suggested report formats based on the content.
        Uses asyncio.to_thread for all blocking I/O operations.
        """
        try:
            system_prompt = get_report_suggestion_prompt(language)

            if file_key:
                # Download PDF from S3 (non-blocking)
                tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".pdf")
                tmp_path = tmp.name
                tmp.close()

                try:
                    await asyncio.to_thread(
                        s3_service.s3_client.download_file,
                        settings.AWS_S3_BUCKET,
                        file_key,
                        tmp_path
                    )
                    
                    # Upload to OpenAI (non-blocking)
                    def upload_to_openai():
                        with open(tmp_path, "rb") as f:
                            return self.openai_client.files.create(
                                file=f,
                                purpose="assistants"
                            )
                    
                    uploaded_file = await asyncio.to_thread(upload_to_openai)
                    
                    messages = [
                        {"role": "system", "content": system_prompt},
                        {
                            "role": "user",
                            "content": [
                                {
                                    "type": "file",
                                    "file": {"file_id": uploaded_file.id}
                                }
                            ]
                        }
                    ]
                    
                    # Call OpenAI (non-blocking)
                    response = await asyncio.to_thread(
                        self.openai_client.chat.completions.create,
                        model="gpt-4o-mini",
                        messages=messages,
                        response_format={"type": "json_object"},
                        temperature=0.7
                    )
                    
                    # Clean up OpenAI file (non-blocking)
                    await asyncio.to_thread(
                        self.openai_client.files.delete,
                        uploaded_file.id
                    )
                    
                    raw_content = response.choices[0].message.content
                    
                finally:
                    if os.path.exists(tmp_path):
                        await asyncio.to_thread(os.remove, tmp_path)
            
            elif text_input:
                messages = [
                    {"role": "system", "content": system_prompt},
                    {"role": "user", "content": f"Analyze this content:\n\n{text_input}"}
                ]
                
                # Call OpenAI (non-blocking)
                response = await asyncio.to_thread(
                    self.openai_client.chat.completions.create,
                    model="gpt-4o-mini",
                    messages=messages,
                    response_format={"type": "json_object"},
                    temperature=0.7
                )
                raw_content = response.choices[0].message.content
            
            else:
                raise ValueError("Either file_key or text_input must be provided")

            data = json.loads(raw_content)
            return data.get("suggestions", [])

        except Exception as e:
            logger.error(f"Format suggestion failed: {e}")
            return []

    async def generate_report(
        self,
        file_key: Optional[str] = None,
        text_input: Optional[str] = None,
        format_key: str = "briefing_doc",
        custom_prompt: Optional[str] = None,
        language: str = "Japanese"
    ) -> str:
        """
        Generates a full report based on the selected format.
        Uses asyncio.to_thread for all blocking I/O operations.
        """
        try:
            base_prompt = get_report_prompt(format_key, custom_prompt or "", language)
            
            # Language styling instruction
            if language == "Japanese":
                system_prompt = (
                    "あなたは日本語でレポートを作成するAIアシスタントです。すべての回答は日本語で書いてください。\n\n"
                    f"{base_prompt}\n\n"
                    "重要: レポート全体を日本語で書いてください。回答はマークダウン形式で、適切な見出し、箇跨書き、構造を使用して読みやすくフォーマットしてください。"
                )
            else:
                system_prompt = (
                    "You are an AI assistant that creates reports in English. Write all responses in English.\n\n"
                    f"{base_prompt}\n\n"
                    "IMPORTANT: Write the entire report in English. Please format your response in markdown with proper headings, bullet points, and structure for easy reading."
                )

            if file_key:
                # Download PDF from S3 (non-blocking)
                tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".pdf")
                tmp_path = tmp.name
                tmp.close()

                try:
                    await asyncio.to_thread(
                        s3_service.s3_client.download_file,
                        settings.AWS_S3_BUCKET,
                        file_key,
                        tmp_path
                    )
                    
                    # Upload to OpenAI (non-blocking)
                    def upload_to_openai():
                        with open(tmp_path, "rb") as f:
                            return self.openai_client.files.create(
                                file=f,
                                purpose="assistants"
                            )
                    
                    uploaded_file = await asyncio.to_thread(upload_to_openai)
                    
                    messages = [
                        {"role": "system", "content": system_prompt},
                        {
                            "role": "user",
                            "content": [
                                {
                                    "type": "file",
                                    "file": {"file_id": uploaded_file.id}
                                }
                            ]
                        }
                    ]
                    
                    # Call OpenAI (non-blocking)
                    response = await asyncio.to_thread(
                        self.openai_client.chat.completions.create,
                        model="gpt-4o-mini",
                        messages=messages,
                        temperature=0.7
                    )
                    
                    # Clean up OpenAI (non-blocking)
                    await asyncio.to_thread(
                        self.openai_client.files.delete,
                        uploaded_file.id
                    )
                    
                    return response.choices[0].message.content
                    
                finally:
                    if os.path.exists(tmp_path):
                        await asyncio.to_thread(os.remove, tmp_path)
            
            elif text_input:
                messages = [
                    {"role": "system", "content": system_prompt},
                    {"role": "user", "content": f"Please analyze the following content and generate a report based on it:\n\n{text_input}"}
                ]
                
                # Call OpenAI (non-blocking)
                response = await asyncio.to_thread(
                    self.openai_client.chat.completions.create,
                    model="gpt-4o-mini",
                    messages=messages,
                    temperature=0.7
                )
                return response.choices[0].message.content
            
            else:
                raise ValueError("Either file_key or text_input must be provided")

        except Exception as e:
            logger.error(f"Report generation failed: {e}")
            raise

report_service = ReportService()