import shutil

import pymupdf as fitz
import pytest
from docx import Document
from pptx import Presentation
from pptx.util import Inches

from wisegrade.parsing import _ocr_engine, parse_document


def test_pdf_pages_and_images(tmp_path):
    pdf = fitz.open()
    pdf.new_page().insert_text((72, 72), "A sample engineering report with sufficiently long searchable text.")
    path = tmp_path / "report.pdf"
    pdf.save(path)
    parsed = parse_document(path, "final_report", tmp_path / "assets", ocr=False)
    assert parsed["pages"][0]["page_number"] == 1
    assert "engineering" in parsed["pages"][0]["text"]
    assert (tmp_path / "assets" / "page-1.png").exists()


def test_scanned_pdf_requires_offline_ocr_package(tmp_path):
    with pytest.raises(RuntimeError, match="离线OCR模型目录"):
        _ocr_engine(str(tmp_path / "missing-ocr"))


@pytest.mark.skipif(not shutil.which("libreoffice"), reason="LibreOffice is installed in the deployment image")
def test_docx_and_pptx_render(tmp_path):
    word = Document()
    word.add_heading("Automation experiment", 1)
    word.add_paragraph("The system measured repeatable engineering results.")
    table = word.add_table(rows=1, cols=2)
    table.cell(0, 0).text, table.cell(0, 1).text = "metric", "accuracy"
    docx = tmp_path / "stage.docx"
    word.save(docx)
    parsed = parse_document(docx, "stage2", tmp_path / "docx-assets", ocr=False)
    assert parsed["headings"] and parsed["tables"]
    assert parsed["pages"][0]["rendered_image_path"]

    deck = Presentation()
    slide = deck.slides.add_slide(deck.slide_layouts[5])
    slide.shapes.title.text = "Robot presentation"
    box = slide.shapes.add_textbox(Inches(1), Inches(2), Inches(4), Inches(1))
    box.text = "Results and analysis"
    pptx = tmp_path / "slides.pptx"
    deck.save(pptx)
    slides = parse_document(pptx, "presentation", tmp_path / "pptx-assets", ocr=False)["pages"]
    assert len(slides) == 1 and "Robot presentation" in slides[0]["text"]
    assert slides[0]["rendered_image_path"]