def test_download_url_without_baseurl_raises_exception(tmp_path):
# arrange
dst_mets = join(tmp_path, 'mets.xml')
copyfile(SRC_METS, dst_mets)
ws1 = Resolver().workspace_from_url(dst_mets)
the_file = _url_to_file(SAMPLE_FILE_URL)
# act
with pytest.raises(Exception) as exc:
ws1.download_file(the_file)
# assert exception message contents
assert "Already tried prepending baseurl '%s'" % str(tmp_path) in str(
exc.value)
def test_workspace_from_url0():
# act
workspace = Resolver().workspace_from_url(METS_HEROLD)
input_files = workspace.mets.find_all_files(fileGrp='OCR-D-IMG')
image_file = input_files[0]
f = workspace.download_file(image_file)
# assert
assert '%s.tif' % f.ID == 'FILE_0001_IMAGE.tif'
assert f.local_filename == 'OCR-D-IMG/FILE_0001_IMAGE.tif'
def test_download_url_with_baseurl(tmp_path):
# arrange
dst_mets = join(tmp_path, 'mets.xml')
copyfile(SRC_METS, dst_mets)
tif_dir = tmp_path / 'OCR-D-IMG'
tif_dir.mkdir()
dst_tif = join(tmp_path, SAMPLE_FILE_URL)
copyfile(join(dirname(SRC_METS), SAMPLE_FILE_URL), dst_tif)
ws1 = Resolver().workspace_from_url(dst_mets,
src_baseurl=dirname(SRC_METS))
the_file = _url_to_file(dst_tif)
# act
# TODO
# semantics changed from .download_url to .download_file
# and from context path 'DEPRECATED' to 'OCR-D-IMG'
f = Path(ws1.download_file(the_file).local_filename)
# assert
assert str(f).endswith(join('OCR-D-IMG', '%s.tif' % SAMPLE_FILE_ID))
assert Path(ws1.directory, f).exists()
