22f99965ecb4567cce7f389349a822477b0c4eaa,tests/test_basic.py,,test_looks_scanned,#,35

Before Change

    CaseStudy_ACS.pdf contains a transparent image overlaying the entire page.
    This overlaying transparent image fools TreeExtractor into thinking it is scanned.
    
    output = pdftotree.parse("tests/input/CaseStudy_ACS.pdf", favor_figures="True")
    assert output.count("ocrx_word") == 1  // single appearance in ocr-capabilities
    output = pdftotree.parse("tests/input/CaseStudy_ACS.pdf", favor_figures="False")
    assert output.count("ocrx_word") >= 1000

After Change

    output = pdftotree.parse("tests/input/CaseStudy_ACS.pdf")
    soup = BeautifulSoup(output)
    assert len(soup.find_all(class_="ocrx_word")) >= 1000
    assert len(soup.find_all("figure")) == 3

    // Adapted from https://github.com/ocropus/hocr-tools/blob/v1.3.0/hocr-check
    def get_prop(node: Tag, name: str) -> Optional[str]:
        title = node.get("title")

Instances