def test_predict_pdf_works(predictor, test_pdf): # FIXME strange outputs, with '\n's in the dict predictions, metadata = predictor.predict_pdf(test_pdf) predictions = [p for p in predictions][0] assert predictions["class"] == "formula" assert dict(**predictions["probabilities"]) == {'formula': 1.0, 'other': 1.576210689757148e-17, 'signature': 5.364939675629517e-24, 'logo': 8.815339177836606e-25} metadata = list(metadata) metadata = dict(**metadata[0]) metadata.pop("document_filename") # temp filename cannot be tested assert metadata == {'px_width': 389.0, 'px_height': 389.0, 'width': 194.49999000000003, 'height': 194.49998999999997, 'x1': 320.861, 'x2': 515.36099, 'y1': 347.699, 'y2': 542.19899, 'page_width': 595.2800000000001, 'page_height': 841.89, 'page_rotation': 0, 'page_idx': 1, 'n_pages': 3}