Skip to content

Commit

Permalink
Merge pull request #76 from Snailed/parsing-of-logs-fail
Browse files Browse the repository at this point in the history 
Fix listing log files when amount of std and output logs mismatch
  • Loading branch information
@Snailed
Snailed authored Aug 6, 2024
2 parents 050cc45 + 7e83af8 commit a547f9e
Show file tree
Hide file tree
Showing 2 changed files with 102 additions and 76 deletions.
8 changes: 4 additions & 4 deletions carbontracker/parser.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -319,15 +319,15 @@ def get_all_logs(log_dir):
std_logs = sorted(list(filter(std_re.match, files)))
if len(output_logs) != len(std_logs):
# Try to remove the files with no matching output/std logs
op_fn = [f.split("_")[0] for f in output_logs]
std_fn = [f.split("_")[0] for f in std_logs]
op_fn = [f.split("_carbontracker")[0] for f in output_logs]
std_fn = [f.split("_carbontracker")[0] for f in std_logs]
if len(std_logs) > len(output_logs):
missing_logs = list(set(std_fn) - set(op_fn))
[std_logs.remove(f + "_carbontracker.log") for f in missing_logs]
else:
missing_logs = list(set(op_fn) - set(std_fn))
[output_logs.remove(f + "carbontracker_output.log") for f in missing_logs]
### Even after removel if then there is a mismatch, then throw the error
[output_logs.remove(f + "_carbontracker_output.log") for f in missing_logs]
### Even after removal if then there is a mismatch, then throw the error
if len(output_logs) != len(std_logs):
raise exceptions.MismatchedLogFilesError(
f"Found {len(output_logs)} output logs and {len(std_logs)} "
Expand Down
170 changes: 98 additions & 72 deletions tests/test_parser.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -25,25 +25,27 @@ def test_get_all_logs(self, mock_getsize, mock_isfile, mock_listdir):
log_dir = "/path/to/logs"

self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log1.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker_output.log"),
contents="output_log1 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log2.log"),
os.path.join(log_dir, "32487_2024-06-26T141608Z_carbontracker_output.log"),
contents="output_log2 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log1.log"), contents="std_log1 content"
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker.log"),
contents="std_log1 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log2.log"), contents="std_log2 content"
os.path.join(log_dir, "32487_2024-06-26T141608Z_carbontracker.log"),
contents="std_log2 content",
)

mock_listdir.return_value = [
"carbontracker_output_log1.log",
"carbontracker_output_log2.log",
"carbontracker_log1.log",
"carbontracker_log2.log",
"10151_2024-03-26T105926Z_carbontracker_output.log",
"32487_2024-06-26T141608Z_carbontracker_output.log",
"10151_2024-03-26T105926Z_carbontracker.log",
"32487_2024-06-26T141608Z_carbontracker.log",
]

mock_isfile.side_effect = lambda path: path.endswith(".log")
Expand All @@ -52,12 +54,12 @@ def test_get_all_logs(self, mock_getsize, mock_isfile, mock_listdir):
output_logs, std_logs = parser.get_all_logs(log_dir)

expected_output_logs = [
os.path.join(log_dir, "carbontracker_output_log1.log"),
os.path.join(log_dir, "carbontracker_output_log2.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker_output.log"),
os.path.join(log_dir, "32487_2024-06-26T141608Z_carbontracker_output.log"),
]
expected_std_logs = [
os.path.join(log_dir, "carbontracker_log1.log"),
os.path.join(log_dir, "carbontracker_log2.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker.log"),
os.path.join(log_dir, "32487_2024-06-26T141608Z_carbontracker.log"),
]

self.assertCountEqual(output_logs, expected_output_logs)
Expand Down Expand Up @@ -118,25 +120,27 @@ def test_get_most_recent_logs(self, mock_getmtime, mock_isfile, mock_listdir):
log_dir = "/path/to/logs"

self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log1.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker_output.log"),
contents="output_log1 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log2.log"),
os.path.join(log_dir, "32487_2024-06-26T141608Z_carbontracker_output.log"),
contents="output_log2 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log1.log"), contents="std_log1 content"
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker.log"),
contents="std_log1 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log2.log"), contents="std_log2 content"
os.path.join(log_dir, "32487_2024-06-26T141608Z_carbontracker.log"),
contents="std_log2 content",
)

mock_listdir.return_value = [
"carbontracker_output_log1.log",
"carbontracker_output_log2.log",
"carbontracker_log1.log",
"carbontracker_log2.log",
"10151_2024-03-26T105926Z_carbontracker_output.log",
"32487_2024-06-26T141608Z_carbontracker_output.log",
"10151_2024-03-26T105926Z_carbontracker.log",
"32487_2024-06-26T141608Z_carbontracker.log",
]

mock_isfile.side_effect = lambda path: path.endswith(".log")
Expand All @@ -149,8 +153,12 @@ def test_get_most_recent_logs(self, mock_getmtime, mock_isfile, mock_listdir):

std_log, output_log = parser.get_most_recent_logs(log_dir)

expected_std_log = os.path.join(log_dir, "carbontracker_log2.log")
expected_output_log = os.path.join(log_dir, "carbontracker_output_log2.log")
expected_std_log = os.path.join(
log_dir, "32487_2024-06-26T141608Z_carbontracker.log"
)
expected_output_log = os.path.join(
log_dir, "32487_2024-06-26T141608Z_carbontracker_output.log"
)

self.assertEqual(std_log, expected_std_log)
self.assertEqual(output_log, expected_output_log)
Expand Down Expand Up @@ -205,16 +213,17 @@ def test_parse_all_logs(self, mock_isfile, mock_listdir, mock_open):
log_dir = "/path/to/logs"

self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log1.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker_output.log"),
contents="output_log1 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log1.log"), contents="std_log1 content"
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker.log"),
contents="std_log1 content",
)

mock_listdir.return_value = [
"carbontracker_output_log1.log",
"carbontracker_log1.log",
"10151_2024-03-26T105926Z_carbontracker_output.log",
"10151_2024-03-26T105926Z_carbontracker.log",
]
mock_isfile.side_effect = lambda path: path.endswith(".log")
mock_open.return_value.read.return_value = "content"
Expand All @@ -224,11 +233,11 @@ def test_parse_all_logs(self, mock_isfile, mock_listdir, mock_open):
self.assertEqual(len(logs), 1)
self.assertEqual(
logs[0]["output_filename"],
os.path.join(log_dir, "carbontracker_output_log1.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker_output.log"),
)
self.assertEqual(
logs[0]["standard_filename"],
os.path.join(log_dir, "carbontracker_log1.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker.log"),
)

@mock.patch("builtins.open", new_callable=mock.mock_open)
Expand All @@ -245,11 +254,12 @@ def test_parse_logs(self, mock_get_devices, mock_isfile, mock_listdir, mock_open
)

self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log1.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker_output.log"),
contents="output_log1 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log1.log"), contents=std_log_data
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker.log"),
contents=std_log_data,
)

mock_isfile.side_effect = lambda path: path.endswith(".log")
Expand All @@ -261,8 +271,8 @@ def test_parse_logs(self, mock_get_devices, mock_isfile, mock_listdir, mock_open

components = parser.parse_logs(
log_dir,
os.path.join(log_dir, "carbontracker_log1.log"),
os.path.join(log_dir, "carbontracker_output_log1.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker_output.log"),
)

self.assertIn("gpu", components)
Expand All @@ -287,48 +297,54 @@ def test_get_all_logs_mismatched_files(

# Create three matching pairs of log files
self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log1.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker_output.log"),
contents="output_log1 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log1.log"), contents="std_log1 content"
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker.log"),
contents="std_log1 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log2.log"),
os.path.join(log_dir, "32487_2024-06-26T141608Z_carbontracker_output.log"),
contents="output_log2 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log2.log"), contents="std_log2 content"
os.path.join(log_dir, "32487_2024-06-26T141608Z_carbontracker.log"),
contents="std_log2 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log3.log"),
os.path.join(log_dir, "40793_2024-03-26T131535Z_carbontracker_output.log"),
contents="output_log3 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log3.log"), contents="std_log3 content"
os.path.join(log_dir, "40793_2024-03-26T131535Z_carbontracker.log"),
contents="std_log3 content",
)

# Add extra unmatched output log file
self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log4.log"),
os.path.join(log_dir, "9803_2024-03-26T105836Z_carbontracker_output.log"),
contents="output_log4 content",
)

mock_listdir.return_value = [
"carbontracker_output_log1.log",
"carbontracker_output_log2.log",
"carbontracker_output_log3.log",
"carbontracker_output_log4.log",
"carbontracker_log1.log",
"carbontracker_log2.log",
"carbontracker_log3.log",
"10151_2024-03-26T105926Z_carbontracker_output.log",
"32487_2024-06-26T141608Z_carbontracker_output.log",
"40793_2024-03-26T131535Z_carbontracker_output.log",
"9803_2024-03-26T105836Z_carbontracker_output.log",
"10151_2024-03-26T105926Z_carbontracker.log",
"32487_2024-06-26T141608Z_carbontracker.log",
"40793_2024-03-26T131535Z_carbontracker.log",
]

mock_isfile.side_effect = lambda path: path.endswith(".log")
mock_getsize.return_value = 100

with self.assertRaises(exceptions.MismatchedLogFilesError):
parser.get_all_logs(log_dir)
expected = [os.path.join(log_dir, f) for f in mock_listdir.return_value]
self.assertTupleEqual(
parser.get_all_logs(log_dir),
(expected[:3], expected[4:]),
)

@mock.patch("os.listdir")
@mock.patch("os.path.isfile")
Expand All @@ -340,47 +356,53 @@ def test_get_all_logs_mismatched_files_extra_std_log(

# Create three matching pairs of log files
self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log1.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker_output.log"),
contents="output_log1 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log1.log"), contents="std_log1 content"
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker.log"),
contents="std_log1 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log2.log"),
os.path.join(log_dir, "32487_2024-06-26T141608Z_carbontracker_output.log"),
contents="output_log2 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log2.log"), contents="std_log2 content"
os.path.join(log_dir, "32487_2024-06-26T141608Z_carbontracker.log"),
contents="std_log2 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log3.log"),
os.path.join(log_dir, "40793_2024-03-26T131535Z_carbontracker_output.log"),
contents="output_log3 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log3.log"), contents="std_log3 content"
os.path.join(log_dir, "40793_2024-03-26T131535Z_carbontracker.log"),
contents="std_log3 content",
)

# Add extra unmatched std log file
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log4.log"), contents="std_log4 content"
os.path.join(log_dir, "9803_2024-03-26T105836Z_carbontracker.log"),
contents="std_log4 content",
)

mock_listdir.return_value = [
"carbontracker_output_log1.log",
"carbontracker_output_log2.log",
"carbontracker_output_log3.log",
"carbontracker_log1.log",
"carbontracker_log2.log",
"carbontracker_log3.log",
"carbontracker_log4.log",
"10151_2024-03-26T105926Z_carbontracker_output.log",
"32487_2024-06-26T141608Z_carbontracker_output.log",
"40793_2024-03-26T131535Z_carbontracker_output.log",
"10151_2024-03-26T105926Z_carbontracker.log",
"32487_2024-06-26T141608Z_carbontracker.log",
"40793_2024-03-26T131535Z_carbontracker.log",
"9803_2024-03-26T105836Z_carbontracker.log",
]

mock_isfile.side_effect = lambda path: path.endswith(".log")
mock_getsize.return_value = 100

with self.assertRaises(exceptions.MismatchedLogFilesError):
parser.get_all_logs(log_dir)
expected = [os.path.join(log_dir, f) for f in mock_listdir.return_value]
self.assertTupleEqual(
parser.get_all_logs(log_dir), (expected[:3], expected[3:6])
)

@mock.patch("builtins.open", new_callable=mock.mock_open)
def test_get_consumption_no_equivalents(self, mock_open):
Expand Down Expand Up @@ -553,16 +575,17 @@ def test_aggregate_consumption_actual(self, mock_isfile, mock_listdir, mock_open
)

self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log1.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker_output.log"),
contents=output_log_content,
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log1.log"), contents="std_log1 content"
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker.log"),
contents="std_log1 content",
)

mock_listdir.return_value = [
"carbontracker_output_log1.log",
"carbontracker_log1.log",
"10151_2024-03-26T105926Z_carbontracker_output.log",
"10151_2024-03-26T105926Z_carbontracker.log",
]
mock_isfile.side_effect = lambda path: path.endswith(".log")
mock_open.return_value.read.return_value = output_log_content
Expand Down Expand Up @@ -946,11 +969,12 @@ def test_parse_epoch_mismatch(
)

self.fs.create_file(
os.path.join(log_dir, "carbontracker_output_log1.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker_output.log"),
contents="output_log1 content",
)
self.fs.create_file(
os.path.join(log_dir, "carbontracker_log1.log"), contents=std_log_data
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker.log"),
contents=std_log_data,
)

mock_isfile.side_effect = lambda path: path.endswith(".log")
Expand All @@ -963,6 +987,8 @@ def test_parse_epoch_mismatch(
with self.assertRaises(exceptions.MismatchedEpochsError):
components = parser.parse_logs(
log_dir,
os.path.join(log_dir, "carbontracker_log1.log"),
os.path.join(log_dir, "carbontracker_output_log1.log"),
os.path.join(log_dir, "10151_2024-03-26T105926Z_carbontracker.log"),
os.path.join(
log_dir, "10151_2024-03-26T105926Z_carbontracker_output.log"
),
)

0 comments on commit a547f9e

Please sign in to comment.