"""SSE parser: keep-alive comments, fragmented tool_calls, usage chunk, finish_reason."""

import json

import httpx
import pytest
import respx

from app.core.config import Settings
from app.llm.openrouter import LLMClient

CHUNKS = [
    ": OPENROUTER PROCESSING",
    "",
    'data: {"id":"gen-1","model":"anthropic/claude-sonnet-4.6","choices":[{"index":0,"delta":{"content":"Bonjour"},"finish_reason":null}]}',
    'data: {"id":"gen-1","model":"anthropic/claude-sonnet-4.6","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"id":"call_1","type":"function","function":{"name":"financial_calc","arguments":"{\\"function\\": \\"fv_l"}}]},"finish_reason":null}]}',
    ": OPENROUTER PROCESSING",
    'data: {"id":"gen-1","model":"anthropic/claude-sonnet-4.6","choices":[{"index":0,"delta":{"tool_calls":[{"index":0,"function":{"arguments":"ump\\", \\"params\\": {\\"rate\\": 0.05}}"}}]},"finish_reason":null}]}',
    'data: {"id":"gen-1","model":"anthropic/claude-sonnet-4.6","choices":[{"index":0,"delta":{},"finish_reason":"tool_calls"}]}',
    'data: {"id":"gen-1","model":"anthropic/claude-sonnet-4.6","choices":[],"usage":{"prompt_tokens":120,"completion_tokens":30,"cost":0.00123}}',
    "data: [DONE]",
]


@pytest.mark.asyncio
async def test_parse_fragmented_tool_calls() -> None:
    settings = Settings(OPENROUTER_API_KEY="k", OPENROUTER_BASE_URL="https://or.test/api/v1")
    client = LLMClient(settings)
    body = "\n".join(CHUNKS) + "\n"
    with respx.mock(base_url="https://or.test/api/v1") as mock:
        mock.post("/chat/completions").mock(
            return_value=httpx.Response(200, content=body.encode(),
                                        headers={"content-type": "text/event-stream"}))
        events = [ev async for ev in client.stream_chat(
            [{"role": "user", "content": "hi"}], None, ["anthropic/claude-sonnet-4.6", "openai/gpt-5.5"])]
        sent = json.loads(mock.calls[0].request.content)
        assert sent["models"] == ["anthropic/claude-sonnet-4.6", "openai/gpt-5.5"]
        assert sent["usage"] == {"include": True}
        assert mock.calls[0].request.headers["X-Title"] == "UQO-Chat"
    types = [e.type for e in events]
    assert types[0] == "text_delta" and events[0].data["delta"] == "Bonjour"
    end = next(e for e in events if e.type == "tool_call_end")
    assert end.data["name"] == "financial_calc"
    assert json.loads(end.data["arguments"]) == {"function": "fv_lump", "params": {"rate": 0.05}}
    usage = next(e for e in events if e.type == "usage")
    assert usage.data["input_tokens"] == 120 and usage.data["cost_usd"] == 0.00123
    done = events[-1]
    assert done.type == "done" and done.data["finish_reason"] == "tool_calls"
    await client.aclose()


@pytest.mark.asyncio
async def test_retry_then_error() -> None:
    settings = Settings(OPENROUTER_API_KEY="k", OPENROUTER_BASE_URL="https://or.test/api/v1")
    client = LLMClient(settings)
    with respx.mock(base_url="https://or.test/api/v1") as mock:
        mock.post("/chat/completions").mock(return_value=httpx.Response(500, text="boom"))
        with pytest.raises(Exception):  # noqa: B017
            _ = [ev async for ev in client.stream_chat([{"role": "user", "content": "x"}], None,
                                                        ["m"])]
        assert len(mock.calls) == 3
    await client.aclose()