tests/test_interpretation.py

import pytest

from poetic.interpretation import build_interpret_contents, parse_interpretation


def test_includes_poem_and_correction():
    c = build_interpret_contents("床前明月光", "要更温暖的色调")
    assert "床前明月光" in c
    assert "要更温暖的色调" in c


def test_omits_correction_line_when_none_given():
    c = build_interpret_contents("床前明月光")
    assert "更正" not in c
    assert "correction" not in c.lower()


def test_parses_guaranteed_json_into_both_fields():
    out = parse_interpretation('{"understandingZh":"一幅月夜图","briefEn":"A moonlit night"}')
    assert out.understanding_zh == "一幅月夜图"
    assert out.brief_en == "A moonlit night"


def test_raises_clear_error_on_malformed_json():
    with pytest.raises(ValueError, match="[Ii]nterpretation"):
        parse_interpretation("not json")


def test_raises_when_fields_missing():
    with pytest.raises(ValueError, match="[Ii]nterpretation"):
        parse_interpretation('{"understandingZh":"只有一半"}')


# The local/remote LLMs (MiniCPM, Qwen) have no enforced JSON mode like Vertex AI's
# response schema, so the parser must repair the two common wrappings.

def test_strips_markdown_code_fences():
    text = '```json\n{"understandingZh":"月夜","briefEn":"moonlit"}\n```'
    out = parse_interpretation(text)
    assert out.understanding_zh == "月夜"
    assert out.brief_en == "moonlit"


def test_extracts_json_object_from_surrounding_prose():
    text = '好的，以下是结果：{"understandingZh":"月夜","briefEn":"moonlit"} 希望您喜欢。'
    out = parse_interpretation(text)
    assert out.brief_en == "moonlit"


def test_repairs_json_escaped_json():
    # Verbatim failure mode from MiniCPM4.1 on ZeroGPU: the whole object is
    # emitted with escaped quotes and newlines, as if nested inside a string.
    text = '{\\n  \\"understandingZh\\": \\"筝声悠扬，渐远渐行。\\",\\n  \\"briefEn\\": \\"A painting of a distant scene.\\"}'
    out = parse_interpretation(text)
    assert out.understanding_zh == "筝声悠扬，渐远渐行。"
    assert out.brief_en == "A painting of a distant scene."


def test_rescues_truncated_unterminated_json():
    # Verbatim from ZeroGPU: escaped JSON that stops before the closing brace.
    text = (
        '{\\n  \\"understandingZh\\": \\"这首诗描绘了筝弦的悠扬与狂风的动荡。\\",'
        '\\n  \\"briefEn\\": \\"a painting only — featuring a distant journey,'
        ' the scene filled with depth and movement.\\"'
    )
    out = parse_interpretation(text)
    assert out.understanding_zh == "这首诗描绘了筝弦的悠扬与狂风的动荡。"
    assert out.brief_en.startswith("a painting only")
    assert out.brief_en.endswith("movement.")


def test_tolerates_literal_newlines_inside_string_values():
    # Seen live from MiniCPM on ZeroGPU: valid-looking JSON with real newline
    # characters inside the string values (invalid in strict JSON).
    text = '{\n  "understandingZh": "第一句。\n第二句。",\n  "briefEn": "line one\nline two"\n}'
    out = parse_interpretation(text)
    assert "第二句" in out.understanding_zh
    assert "line two" in out.brief_en