"""Fetch example images for the README via the mcgibs MCP server.

Uses the FastMCP in-process Client to call tools through the MCP protocol,
exactly as any MCP client would. Images are saved to docs/images/.
"""

import asyncio
import base64
import sys
from pathlib import Path

from fastmcp import Client

# Wire up the server module so the client is initialized
import mcgibs.server as server_module
from mcgibs.client import GIBSClient
from mcgibs.server import mcp

OUT = Path(__file__).parent.parent / "docs" / "images"


async def save_imagery(client: Client, filename: str, tool: str, params: dict) -> None:
    """Call an imagery tool and save the returned image to disk."""
    print(f"  Fetching {filename}...", file=sys.stderr)
    result = await client.call_tool(tool, params)

    # Find the ImageContent in the response
    for content in result.content:
        if content.type == "image":
            img_bytes = base64.b64decode(content.data)
            path = OUT / filename
            path.write_bytes(img_bytes)
            size_kb = len(img_bytes) / 1024
            print(f"  Saved {path} ({size_kb:.0f} KB)", file=sys.stderr)
            return

    # If we got text only (error case), print it
    for content in result.content:
        if content.type == "text":
            print(f"  Warning: {content.text[:200]}", file=sys.stderr)


async def main():
    OUT.mkdir(parents=True, exist_ok=True)

    # Initialize GIBS client (normally done by middleware on first connection)
    print("Initializing GIBS client...", file=sys.stderr)
    gibs = GIBSClient()
    await gibs.initialize()
    server_module._client = gibs
    print(f"Ready — {len(gibs.layer_index)} layers loaded", file=sys.stderr)

    try:
        async with Client(mcp) as client:
            # 1. Amazon true color — natural satellite view
            await save_imagery(client, "amazon-true-color.jpg", "get_imagery", {
                "layer_id": "MODIS_Terra_CorrectedReflectance_TrueColor",
                "date": "2025-08-01",
                "place": "Amazon Rainforest",
                "width": 1024,
                "height": 768,
                "format": "jpeg",
            })

            # 2. Global surface air temperature — scientific data product
            await save_imagery(client, "temperature-global.jpg", "get_imagery", {
                "layer_id": "AIRS_L3_Surface_Air_Temperature_Daily_Day",
                "date": "2025-07-15",
                "bbox": [-180, -90, 180, 90],
                "width": 1024,
                "height": 512,
                "format": "jpeg",
            })

            # 3. Arctic sea ice comparison — March (max) vs September (min)
            await save_imagery(
                client, "arctic-ice-comparison.jpg", "compare_dates", {
                    "layer_id": "AMSRU2_Sea_Ice_Concentration_12km",
                    "date_before": "2025-03-01",
                    "date_after": "2025-09-01",
                    "bbox": [-180, 60, 180, 90],
                },
            )

            # 4. Surface temperature colormap legend
            await save_imagery(
                client, "temperature-legend.png", "get_legend", {
                    "layer_id": "AIRS_L3_Surface_Air_Temperature_Daily_Day",
                    "orientation": "horizontal",
                },
            )

            # 5. Colormap explanation (text, not image)
            print("  Fetching temperature colormap explanation...", file=sys.stderr)
            result = await client.call_tool(
                "explain_layer_colormap",
                {"layer_id": "AIRS_L3_Surface_Air_Temperature_Daily_Day"},
            )
            explanation = result.content[0].text
            (OUT / "temperature-colormap.txt").write_text(explanation)
            print(f"  Saved colormap explanation", file=sys.stderr)

    finally:
        await gibs.close()
        server_module._client = None

    print("Done!", file=sys.stderr)


asyncio.run(main())