def test_try_to_get_a_hidden_pdf_at_root_of_cpd(mock_findcpd, mock_findsibl,
mock_tryhidden,
mock_writehidden):
# if binary already on disk.
scrapealias = scrape_cDM.ScrapeAlias('_', '_')
scrapealias.alias_dir = 'fake/filepath'
scrapealias.find_sibling_files = mock_findsibl
scrapealias.try_getting_hidden_pdf = mock_tryhidden
scrapealias.write_hidden_pdf_if_a_binary = mock_writehidden
mock_findcpd.return_value = ('imag1', 'img')
mock_findsibl.return_value = ['imag1.img', 'imag2.img', 'imag3.img']
scrapealias.try_to_get_a_hidden_pdf_at_root_of_cpd('fakefile')
mock_findcpd.assert_called_with('fake/filepath/Cpd', 'fakefile')
mock_findsibl.assert_called_with('fakefile')
assert not mock_tryhidden.called
assert not mock_writehidden.called
# if binary not already on disk
scrapealias = scrape_cDM.ScrapeAlias('_', '_')
scrapealias.alias_dir = 'fake/filepath'
scrapealias.find_sibling_files = mock_findsibl
scrapealias.try_getting_hidden_pdf = mock_tryhidden
scrapealias.write_hidden_pdf_if_a_binary = mock_writehidden
mock_findcpd.return_value = ('imag1', 'other')
mock_findsibl.return_value = ['imag1.img', 'imag2.img', 'imag3.img']
mock_tryhidden.return_value = 'imag_binary'
scrapealias.try_to_get_a_hidden_pdf_at_root_of_cpd('fakefile')
mock_findcpd.assert_called_with('fake/filepath/Cpd', 'fakefile')
mock_findsibl.assert_called_with('fakefile')
mock_tryhidden.assert_called_with('imag1', 'other')
mock_writehidden.assert_called_with('imag_binary', 'fake/filepath/Cpd',
'imag1', 'other')
def test_write_hidden_pdf_if_a_binary__xml_received_instead_of_binary_failure(
mock_API, binary_decodes_fixture):
# binary = """<?xml version="1.0" encoding="utf-8"?><cpd><type>Document-PDF</type><page><pagetitle>Page 1</pagetitle><pagefile>3605.pdfpage</pagefile><pageptr>3604</pageptr></page></cpd>""".encode('utf-8')
filepath, pointer, filetype = 'imag_filepath', 'imag_pointer', 'imag_filetype'
scrapealias = scrape_cDM.ScrapeAlias('_', '_')
assert scrapealias.write_hidden_pdf_if_a_binary(
binary_decodes_fixture, filepath, pointer, filetype) is False
def test_find_sibling_files():
scrapealias = scrape_cDM.ScrapeAlias('_', '_')
scrapealias.tree_snapshot = [['imag_a', ['imagd_1'], ['a1', 'a2', 'a3']],
[['imag_b'], [], ['b1', 'b2']],
[['imag_c'], [], ['c1', 'c2', 'c3', 'c4']]]
assert set(scrapealias.find_sibling_files('a1')) == {'a1', 'a2', 'a3'}
assert set(scrapealias.find_sibling_files('b2')) == {'b1', 'b2'}
assert set(
scrapealias.find_sibling_files('c3')) == {'c1', 'c2', 'c3', 'c4'}
def test_try_getting_hidden_pdf(mock_API):
scrapealias = scrape_cDM.ScrapeAlias('imag_path', 'imag_alias')
import urllib
mock_API.retrieve_binary.return_value = 'actual_imag_binary'
assert scrapealias.try_getting_hidden_pdf(
'imag_pointer', 'imag_filetype') == 'actual_imag_binary'
mock_API.retrieve_binary.side_effect = urllib.error.HTTPError(
'imag', b"", 42, 43, 'imag_exception occurredd')
assert scrapealias.try_getting_hidden_pdf('imag_pointer',
'imag_filetype') is False
def test_calculate_chunks(mock_count_root_objects):
scrapealias = scrape_cDM.ScrapeAlias('_', '_')
for total, chunksize, chunks in (
(899, 100, 9),
(900, 100, 10),
(999, 1000, 1),
(1, 1, 2),
):
scrapealias.count_root_objects.return_value = total
assert scrapealias.calculate_chunks(chunksize) == chunks
mock_count_root_objects.assert_called_with()
def test_main_loop(mock_docpd, mock_doroot, mock_docoll):
scrapealias = scrape_cDM.ScrapeAlias('_', '_')
scrapealias.do_collection_level_metadata = mock_docoll
scrapealias.do_root_level_objects = mock_doroot
scrapealias.do_compound_objects = mock_docpd
assert not mock_docoll.called
assert not mock_doroot.called
assert not mock_docpd.called
scrapealias.main()
mock_docoll.assert_called_with()
mock_doroot.assert_called_with()
mock_docpd.assert_called_with()
def test_parse_children_of_cpd(mock_arepointers, mock_ETparse,
ETparse_fixture):
scrapealias = scrape_cDM.ScrapeAlias('_', '_')
scrapealias.alias_dir = 'imag_dir'
scrapealias.are_child_pointers_pdfpages = mock_arepointers
scrape_cDM.ET.parse = mock_ETparse
mock_ETparse.return_value = ETparse_fixture
mock_ETparse.findall = ETparse_fixture
mock_arepointers.return_value = True
assert scrapealias.parse_children_of_cpd('imag_parent') is False
mock_ETparse.assert_called_with('imag_dir/Cpd/imag_parent_cpd.xml')
mock_ETparse.findall.assert_called_with('imag_elem', 'imag_parent_cpd.xml')
mock_arepointers.return_value = False
assert scrapealias.parse_children_of_cpd(
'imag_parent') == 'imag_children_pointers_list'
def test_are_child_pointers_pdfpages(mock_try_to_get, mock_has_pdfpage):
scrapealias = scrape_cDM.ScrapeAlias('_', '_')
scrapealias.try_to_get_a_hidden_pdf_at_root_of_cpd = mock_try_to_get
# scrape_cDM.has_pdfpage_elems = mock_has_pdfpage
mock_has_pdfpage.return_value = False
assert scrapealias.are_child_pointers_pdfpages('imag_list',
'imag_filename') is False
mock_has_pdfpage.assert_called_with('imag_list')
assert not mock_try_to_get.called
mock_has_pdfpage.return_value = True
mock_try_to_get.return_value = False
assert scrapealias.are_child_pointers_pdfpages('imag_list',
'imag_filename') is True
mock_has_pdfpage.assert_called_with('imag_list')
mock_try_to_get.assert_called_with('imag_filename')
mock_has_pdfpage.return_value = True
mock_try_to_get.return_value = True
assert scrapealias.are_child_pointers_pdfpages('imag_list',
'imag_filename') is True
mock_has_pdfpage.assert_called_with('imag_list')
mock_try_to_get.assert_called_with('imag_filename')
def test_do_collection_level_metadata(mock_API, mock_os):
mock_os.makedirs.return_value = True
mock_API.retrieve_collection_total_recs.return_value = 'total_recs'
mock_API.retrieve_collection_metadata.return_value = 'coll_metadata'
mock_API.retrieve_collection_fields_json.return_value = 'fields_json'
mock_API.retrieve_collecion_fields_xml.return_value = 'fields_xml'
scrapealias = scrape_cDM.ScrapeAlias('imag_path', 'imag_alias')
scrapealias.alias_dir = 'imag_filepath'
mock_os.listdir.return_value = ('Collection_TotalRecs.xml',
'Collection_Metadata.xml',
'Collection_Fields.json',
'Collection_Fields.xml')
scrapealias.do_collection_level_metadata()
assert not mock_API.retrieve_collection_total_recs.called
assert not mock_API.retrieve_collection_metadata.called
assert not mock_API.retrieve_collection_fields_json.called
assert not mock_API.retrieve_collection_fields_xml.called
mock_os.listdir.return_value = ('')
scrapealias.do_collection_level_metadata()
mock_API.retrieve_collection_total_recs.assert_called_with('imag_alias')
mock_API.retrieve_collection_metadata.assert_called_with('imag_alias')
mock_API.retrieve_collection_fields_json.assert_called_with('imag_alias')
mock_API.retrieve_collection_fields_xml.assert_called_with('imag_alias')
fake_json_fields_call = mock_API.retrieve_collection_fields_json(
'imag_alias')
fake_xml_fields_call = mock_API.retrieve_collection_fields_xml(
'imag_alias')
assert mock_API.write_xml_to_file.call_count == 3
assert mock_API.write_json_to_file.call_count == 1
mock_API.write_json_to_file.assert_called_with(fake_json_fields_call,
'imag_filepath',
'Collection_Fields')
mock_API.write_xml_to_file.assert_called_with(fake_xml_fields_call,
'imag_filepath',
'Collection_Fields')
def test_count_root_objects(mock_ET, total_recs_etree_fixture):
mock_ET.parse.return_value = total_recs_etree_fixture
scrapealias = scrape_cDM.ScrapeAlias('_', '_')
scrapealias.alias_dir = 'imag_alias_dir'
assert scrapealias.count_root_objects() == 20
mock_ET.parse.assert_called_with('imag_alias_dir/Collection_TotalRecs.xml')
