pr-reviewer/tests/test_reviewer.py

"""Tests for app.reviewer: _safe_json, render_markdown, and review()."""

from __future__ import annotations

import json

import pytest

from app import reviewer as reviewer_mod
from app.reviewer import Reviewer, _safe_json

from .conftest import FakeResponse, patch_httpx


# --------------------------------------------------------------------------- #
# _safe_json
# --------------------------------------------------------------------------- #


def test_safe_json_plain() -> None:
    assert _safe_json('{"a": 1, "b": "x"}') == {"a": 1, "b": "x"}


def test_safe_json_json_fenced() -> None:
    text = '```json\n{"summary": "hi", "block": []}\n```'
    assert _safe_json(text) == {"summary": "hi", "block": []}


def test_safe_json_bare_fenced() -> None:
    text = '```\n{"k": true}\n```'
    assert _safe_json(text) == {"k": True}


def test_safe_json_prose_wrapped() -> None:
    text = 'Sure, here is the review:\n{"summary": "ok"}\nHope that helps!'
    assert _safe_json(text) == {"summary": "ok"}


def test_safe_json_malformed_raises() -> None:
    with pytest.raises(json.JSONDecodeError):
        _safe_json("this is not json at all")


# --------------------------------------------------------------------------- #
# render_markdown
# --------------------------------------------------------------------------- #


def _render(cfg, pr, review):
    return Reviewer(cfg).render_markdown(pr, review)


def test_render_mention_applied_for_configured_author(make_cfg) -> None:
    cfg = make_cfg(MENTION_AUTHORS=["openswe"])
    pr = {"author": "openswe"}
    review = {
        "summary": "Looks good.",
        "block": [],
        "fix": [],
        "nit": [],
        "question": [],
    }
    out = _render(cfg, pr, review)
    assert out.startswith("**Summary:** @openswe Looks good.")


def test_render_mention_case_insensitive(make_cfg) -> None:
    cfg = make_cfg(MENTION_AUTHORS=["openswe"])
    pr = {"author": "OpenSWE"}
    review = {"summary": "Nice."}
    out = _render(cfg, pr, review)
    # Mention preserves the actual author casing, but matching is case-insensitive.
    assert "@OpenSWE Nice." in out


def test_render_no_mention_for_other_author(make_cfg) -> None:
    cfg = make_cfg(MENTION_AUTHORS=["openswe"])
    pr = {"author": "octocat"}
    review = {"summary": "Fine."}
    out = _render(cfg, pr, review)
    assert "@octocat" not in out
    assert out.startswith("**Summary:** Fine.")


def test_render_empty_categories_omitted(make_cfg) -> None:
    cfg = make_cfg()
    pr = {"author": "octocat"}
    review = {
        "summary": "s",
        "block": ["real block"],
        "fix": [],  # omitted
        "nit": ["", "  "],  # all blank -> omitted
        "question": ["a question"],
    }
    out = _render(cfg, pr, review)
    assert "## BLOCK" in out
    assert "## FIX" not in out
    assert "## NIT" not in out
    assert "## QUESTION" in out


def test_render_section_order(make_cfg) -> None:
    cfg = make_cfg()
    pr = {"author": "octocat"}
    review = {
        "summary": "s",
        "block": ["b"],
        "fix": ["f"],
        "nit": ["n"],
        "question": ["q"],
        "overall": "closing",
    }
    out = _render(cfg, pr, review)
    order = [out.index(h) for h in ("## BLOCK", "## FIX", "## NIT", "## QUESTION")]
    assert order == sorted(order)
    # Overall renders last.
    assert out.index("## Overall") > out.index("## QUESTION")
    assert "closing" in out


def test_render_summary_line_present(make_cfg) -> None:
    cfg = make_cfg()
    out = _render(cfg, {"author": "x"}, {"summary": "top line"})
    assert out.splitlines()[0] == "**Summary:** top line"


# --------------------------------------------------------------------------- #
# review()
# --------------------------------------------------------------------------- #


def _fireworks_response(payload_obj: dict) -> FakeResponse:
    content = json.dumps(payload_obj)
    return FakeResponse(
        200,
        json_data={"choices": [{"message": {"content": content}}]},
    )


def test_review_parses_response_and_builds_payload(
    monkeypatch, make_cfg, sample_pr
) -> None:
    cfg = make_cfg()
    review_obj = {
        "summary": "Solid change.",
        "block": [],
        "fix": ["`x.py:1` handle None"],
        "nit": [],
        "question": [],
        "overall": "LGTM with a fix",
        "recommended_event": "COMMENT",
    }
    calls = patch_httpx(monkeypatch, reviewer_mod, _fireworks_response(review_obj))

    result = Reviewer(cfg).review(sample_pr, "diff --git a/x b/x\n+hello\n")

    assert result["summary"] == "Solid change."
    assert result["recommended_event"] == "COMMENT"
    # render_markdown output is attached.
    assert "_body_markdown" in result
    assert result["_body_markdown"].startswith("**Summary:**")

    # One POST to chat/completions with the expected payload.
    assert len(calls) == 1
    call = calls[0]
    assert call["method"] == "POST"
    assert call["url"].endswith("/chat/completions")
    body = call["json"]
    assert body["model"] == cfg.FIREWORKS_MODEL
    assert body["temperature"] == cfg.FIREWORKS_TEMPERATURE
    assert body["max_tokens"] == cfg.FIREWORKS_MAX_TOKENS
    assert body["messages"][0]["role"] == "system"
    assert body["messages"][1]["role"] == "user"


def test_review_truncates_large_diff(monkeypatch, make_cfg, sample_pr) -> None:
    cfg = make_cfg(MAX_DIFF_BYTES=500)
    review_obj = {"summary": "ok"}
    calls = patch_httpx(monkeypatch, reviewer_mod, _fireworks_response(review_obj))

    # Use a marker char that appears nowhere else in the prompt scaffolding.
    big_diff = "Z" * 5000
    Reviewer(cfg).review(sample_pr, big_diff)

    user_content = calls[0]["json"]["messages"][1]["content"]
    assert "[diff truncated for length]" in user_content
    # The huge original diff must not be sent in full.
    assert "Z" * 5000 not in user_content
    # Only up to MAX_DIFF_BYTES of the diff body survives.
    assert user_content.count("Z") == cfg.MAX_DIFF_BYTES