"""
System Prompt Configuration Module
=====================================
Manage global system prompts, prepend to samples,
and preview formatted chat JSON.
"""

from dataclasses import dataclass
from typing import List, Dict, Any, Optional
import json
import pandas as pd


@dataclass
class SystemPromptConfig:
    """Configuration for system prompt handling."""
    system_prompt: str = "You are a helpful AI assistant."
    prepend_to_all: bool = True


def build_chat_json(
    instruction: str,
    output: str,
    system_prompt: str = "",
    context: str = "",
) -> Dict[str, Any]:
    """
    Build a single chat-format JSON entry.
    Returns {"messages": [{"role": ..., "content": ...}, ...]}.
    """
    messages = []

    if system_prompt:
        messages.append({"role": "system", "content": system_prompt})

    user_content = instruction
    if context:
        user_content += f"\n\nContext: {context}"

    messages.append({"role": "user", "content": user_content})
    messages.append({"role": "assistant", "content": output})

    return {"messages": messages}


def preview_formatted(
    df: pd.DataFrame,
    system_prompt: str,
    instruction_col: str,
    output_col: str,
    input_col: Optional[str] = None,
    n: int = 3,
) -> List[Dict[str, Any]]:
    """
    Generate a preview of n formatted chat-JSON samples.
    """
    previews = []
    for i, (_, row) in enumerate(df.head(n).iterrows()):
        instruction = str(row.get(instruction_col, ''))
        output = str(row.get(output_col, ''))
        context = str(row.get(input_col, '')) if input_col and input_col in df.columns else ''
        previews.append(
            build_chat_json(instruction, output, system_prompt, context)
        )
    return previews


def preview_formatted_json(
    df: pd.DataFrame,
    system_prompt: str,
    instruction_col: str,
    output_col: str,
    input_col: Optional[str] = None,
    n: int = 3,
) -> str:
    """Return a pretty-printed JSON string of n sample entries."""
    samples = preview_formatted(
        df, system_prompt, instruction_col, output_col, input_col, n
    )
    return json.dumps(samples, indent=2, ensure_ascii=False)