def test_check_files_exist_and_md5(self):
"""test _check_files_exist_and_md5"""
file1 = "tmp.spreadsheet_validator.file1"
file2 = "tmp.spreadsheet_validator.file2"
with open(file1, "w") as f1, open(file2, "w") as f2:
print("foo", file=f1)
print("bar", file=f2)
reads_file_1_md5 = utils.md5(file1)
dict_list = [{
"reads_file_1": file1,
"reads_file_1_md5": reads_file_1_md5,
"reads_file_2": file2,
"reads_file_2_md5": utils.md5(file2),
}]
self.assertEqual(
[],
spreadsheet_validator.SpreadsheetValidator.
_check_files_exist_and_md5(dict_list, os.getcwd(), md5_threads=1),
)
self.assertEqual(
[],
spreadsheet_validator.SpreadsheetValidator.
_check_files_exist_and_md5(dict_list, os.getcwd(), md5_threads=2),
)
dict_list[0]["reads_file_1_md5"] = "12345"
expected = ["md5_error\t" + file1 + "\t12345\t" + reads_file_1_md5]
self.assertEqual(
expected,
spreadsheet_validator.SpreadsheetValidator.
_check_files_exist_and_md5(dict_list, os.getcwd(), md5_threads=1),
)
self.assertEqual(
[],
spreadsheet_validator.SpreadsheetValidator.
_check_files_exist_and_md5(dict_list, os.getcwd(), md5_threads=0),
)
os.unlink(file1)
expected = ["File_not_found\t" + file1 + "\t2"]
self.assertEqual(
expected,
spreadsheet_validator.SpreadsheetValidator.
_check_files_exist_and_md5(dict_list, os.getcwd()),
)
os.unlink(file2)
expected.append("File_not_found\t" + file2 + "\t2")
self.assertEqual(
expected,
spreadsheet_validator.SpreadsheetValidator.
_check_files_exist_and_md5(dict_list, os.getcwd()),
)
def test_check_files_exist_and_md5(self):
'''test _check_files_exist_and_md5'''
file1 = 'tmp.spreadsheet_validator.file1'
file2 = 'tmp.spreadsheet_validator.file2'
with open(file1, 'w') as f1, open(file2, 'w') as f2:
print('foo', file=f1)
print('bar', file=f2)
reads_file_1_md5 = utils.md5(file1)
dict_list = [{
'reads_file_1': file1,
'reads_file_1_md5': reads_file_1_md5,
'reads_file_2': file2,
'reads_file_2_md5': utils.md5(file2),
}]
self.assertEqual([],
spreadsheet_validator.SpreadsheetValidator.
_check_files_exist_and_md5(dict_list,
os.getcwd(),
md5_threads=1))
self.assertEqual([],
spreadsheet_validator.SpreadsheetValidator.
_check_files_exist_and_md5(dict_list,
os.getcwd(),
md5_threads=2))
dict_list[0]['reads_file_1_md5'] = '12345'
expected = ['md5_error\t' + file1 + '\t12345\t' + reads_file_1_md5]
self.assertEqual(
expected,
spreadsheet_validator.SpreadsheetValidator.
_check_files_exist_and_md5(dict_list, os.getcwd(), md5_threads=1))
self.assertEqual([],
spreadsheet_validator.SpreadsheetValidator.
_check_files_exist_and_md5(dict_list,
os.getcwd(),
md5_threads=0))
os.unlink(file1)
expected = ['File_not_found\t' + file1 + '\t2']
self.assertEqual(
expected,
spreadsheet_validator.SpreadsheetValidator.
_check_files_exist_and_md5(dict_list, os.getcwd()))
os.unlink(file2)
expected.append('File_not_found\t' + file2 + '\t2')
self.assertEqual(
expected,
spreadsheet_validator.SpreadsheetValidator.
_check_files_exist_and_md5(dict_list, os.getcwd()))
def _copy_reads_file(cls, old_name, new_name, expected_md5):
old_name_md5 = utils.md5(old_name)
if old_name_md5 != expected_md5:
raise Error(
"MD5 given by submitter "
+ expected_md5
+ " does not match calculated MD5 "
+ old_name_md5
)
utils.rsync_and_md5(old_name, new_name, expected_md5)
def test_copy_reads_file(self):
'''test _copy_reads_file'''
original_reads_file = os.path.join(data_dir, 'copy_reads_file.fq')
correct_md5 = '1bf5fb5b641e793ba40285779a059132'
tmp_reads_file = 'tmp.read_pair_importer.copy_reads_file.out'
if os.path.exists(tmp_reads_file):
os.unlink(tmp_reads_file)
read_pair_importer.ReadPairImporter._copy_reads_file(original_reads_file, tmp_reads_file, correct_md5)
self.assertTrue(os.path.exists(tmp_reads_file))
self.assertEqual(correct_md5, utils.md5(tmp_reads_file))
with self.assertRaises(read_pair_importer.Error):
read_pair_importer.ReadPairImporter._copy_reads_file(original_reads_file, tmp_reads_file, 'wrong md5')
os.unlink(tmp_reads_file)
def test_copy_reads_file(self):
"""test _copy_reads_file"""
original_reads_file = os.path.join(data_dir, "copy_reads_file.fq")
correct_md5 = "1bf5fb5b641e793ba40285779a059132"
tmp_reads_file = "tmp.read_pair_importer.copy_reads_file.out"
if os.path.exists(tmp_reads_file):
os.unlink(tmp_reads_file)
read_pair_importer.ReadPairImporter._copy_reads_file(
original_reads_file, tmp_reads_file, correct_md5)
self.assertTrue(os.path.exists(tmp_reads_file))
self.assertEqual(correct_md5, utils.md5(tmp_reads_file))
with self.assertRaises(read_pair_importer.Error):
read_pair_importer.ReadPairImporter._copy_reads_file(
original_reads_file, tmp_reads_file, "wrong md5")
os.unlink(tmp_reads_file)
def _write_md5_file(infile):
print('Start calculate md5 of', infile)
with open(infile + '.md5', 'w') as f:
print(utils.md5(infile), os.path.basename(infile), sep=' ', file=f)
print('End calculate md5 of', infile)
def _check_md5(filename, expect_md5):
got_md5 = utils.md5(filename)
if expect_md5 == got_md5:
return (filename, None, None)
else:
return (filename, expect_md5, got_md5)
def test_md5(self):
'''test md5'''
filename = os.path.join(data_dir, 'md5.txt')
expected = 'f738cdb99c0c6a670aa1e5742066b0cf'
got = utils.md5(filename)
self.assertEqual(expected, got)
def _write_md5_file(infile):
print("Start calculate md5 of", infile)
with open(infile + ".md5", "w") as f:
print(utils.md5(infile), os.path.basename(infile), sep=" ", file=f)
print("End calculate md5 of", infile)
def test_md5(self):
"""test md5"""
filename = os.path.join(data_dir, "md5.txt")
expected = "f738cdb99c0c6a670aa1e5742066b0cf"
got = utils.md5(filename)
self.assertEqual(expected, got)