|
1 | 1 | """Tests for psm_utils.io.tsv.""" |
2 | 2 |
|
3 | | -import hashlib |
4 | 3 | import os |
5 | 4 |
|
6 | 5 | from psm_utils.io.parquet import ParquetReader, ParquetWriter |
|
32 | 31 | ] |
33 | 32 |
|
34 | 33 |
|
35 | | -def compute_checksum(filename): |
36 | | - hash_func = hashlib.sha256() |
37 | | - with open(filename, "rb") as f: |
38 | | - for chunk in iter(lambda: f.read(4096), b""): |
39 | | - hash_func.update(chunk) |
40 | | - return hash_func.hexdigest() |
41 | | - |
42 | | - |
43 | 34 | class TestParquetWriter: |
44 | | - expected_checksum = "1e5ee7afc5d4131bce8f1d0908136b8c559303abb7bbd7d052afa111d5e64f0c" |
45 | | - |
46 | 35 | def test_write_psm(self): |
47 | 36 | with ParquetWriter("test.pq") as writer: |
48 | 37 | for test_case in test_cases: |
49 | 38 | writer.write_psm(PSM(**test_case)) |
50 | | - actual_checksum = compute_checksum("test.pq") |
51 | | - assert actual_checksum == self.expected_checksum, "Checksums do not match" |
| 39 | + |
| 40 | + with ParquetReader("test.pq") as reader: |
| 41 | + for i, psm in enumerate(reader): |
| 42 | + assert psm == PSM(**test_cases[i]) |
| 43 | + |
52 | 44 | os.remove("test.pq") |
53 | 45 |
|
54 | 46 | def test_write_file(self): |
55 | 47 | with ParquetWriter("test.pq") as writer: |
56 | 48 | writer.write_file(PSMList(psm_list=[PSM(**t) for t in test_cases])) |
57 | | - actual_checksum = compute_checksum("test.pq") |
58 | | - assert actual_checksum == self.expected_checksum, "Checksums do not match" |
59 | | - # os.remove("test.pq") |
| 49 | + |
| 50 | + with ParquetReader("test.pq") as reader: |
| 51 | + for i, psm in enumerate(reader): |
| 52 | + assert psm == PSM(**test_cases[i]) |
| 53 | + |
| 54 | + os.remove("test.pq") |
60 | 55 |
|
61 | 56 |
|
62 | 57 | class TestParquetReader: |
63 | 58 | def test_iter(self): |
64 | | - # Write test cases to file |
65 | | - ParquetWriter("test.pq").write_file(PSMList(psm_list=[PSM(**t) for t in test_cases])) |
66 | | - |
67 | 59 | # Read test cases from file |
68 | | - for i, psm in enumerate(ParquetReader("test.pq")): |
69 | | - assert psm == PSM(**test_cases[i]) |
70 | | - |
71 | | - os.remove("test.pq") |
| 60 | + with ParquetReader("tests/test_data/test.pq") as reader: |
| 61 | + for i, psm in enumerate(reader): |
| 62 | + assert psm == PSM(**test_cases[i]) |
0 commit comments