Merge pull request #989 from mlco2/bug/csv_ordering

benoit-cty · web-flow · commit d0ff5e47eb5c · 2025-11-29T09:52:17.000+01:00
Sort CSV headers before comparing so that we don't spuriously create backup output files.
diff --git a/codecarbon/output_methods/file.py b/codecarbon/output_methods/file.py
@@ -68,10 +68,10 @@ def has_valid_headers(self, data: EmissionsData) -> bool:
                 # No entries
                 return True
             dict_from_csv = dict(csv_entries_list[0])
-            list_of_column_names = list(dict_from_csv.keys())
-            return list(data.values.keys()) == list_of_column_names
+            list_of_column_names = sorted(dict_from_csv.keys())
+            return sorted(data.values.keys()) == list_of_column_names
 
-    def out(self, total: EmissionsData, _: EmissionsData):
+    def out(self, total: EmissionsData, _):
         """
         Save the emissions data from a whole run to a CSV file.
 
diff --git a/codecarbon/output_methods/http.py b/codecarbon/output_methods/http.py
@@ -19,7 +19,7 @@ class HTTPOutput(BaseOutput):
     def __init__(self, endpoint_url: str):
         self.endpoint_url: str = endpoint_url
 
-    def out(self, total: EmissionsData, delta: EmissionsData):
+    def out(self, total: EmissionsData, _: EmissionsData):
         try:
             payload = dataclasses.asdict(total)
             payload["user"] = getpass.getuser()
@@ -56,14 +56,14 @@ def __init__(
         )
         self.run_id = self.api.run_id
 
-    def live_out(self, total: EmissionsData, delta: EmissionsData):
+    def live_out(self, _, delta: EmissionsData):
         # Called at regular intervals
         try:
             self.api.add_emission(dataclasses.asdict(delta))
         except Exception as e:
             logger.error(e, exc_info=True)
 
-    def out(self, total: EmissionsData, delta: EmissionsData):
+    def out(self, _, delta: EmissionsData):
         # Called on exit
         try:
             self.api.add_emission(dataclasses.asdict(delta))
diff --git a/codecarbon/output_methods/metrics/logfire.py b/codecarbon/output_methods/metrics/logfire.py
@@ -59,7 +59,7 @@ def __init__(self):
             "codecarbon_ram_energy", unit="(kWh)", description="Energy used per RAM"
         )
 
-    def out(self, total: EmissionsData, delta: EmissionsData):
+    def out(self, _, delta: EmissionsData):
         try:
             self.duration.add(delta.duration)
             self.emissions.add(delta.emissions)
@@ -75,5 +75,5 @@ def out(self, total: EmissionsData, delta: EmissionsData):
         except Exception as e:
             logger.error(e, exc_info=True)
 
-    def live_out(self, total: EmissionsData, delta: EmissionsData):
-        self.out(total, delta)
+    def live_out(self, _: EmissionsData, delta: EmissionsData):
+        self.out(None, delta)
diff --git a/pyproject.toml b/pyproject.toml
@@ -90,6 +90,7 @@ dev = [
     "requests",
     "requests-mock",
     "responses",
+    "logfire>=1.0.1",
 ]
 doc = [
   "sphinx",
diff --git a/tests/output_methods/test_file.py b/tests/output_methods/test_file.py
@@ -2,7 +2,7 @@
 import shutil
 import tempfile
 import unittest
-from unittest.mock import MagicMock, patch
+from unittest.mock import patch
 
 import pandas as pd
 
@@ -66,7 +66,7 @@ def test_file_output_initialization_invalid_dir(self):
 
     def test_has_valid_headers_success(self):
         file_output = FileOutput("test.csv", self.temp_dir)
-        file_output.out(self.emissions_data, MagicMock())
+        file_output.out(self.emissions_data, None)
 
         self.assertTrue(file_output.has_valid_headers(self.emissions_data))
 
@@ -77,9 +77,19 @@ def test_has_valid_headers_success_with_empty_file(self):
 
         self.assertTrue(file_output.has_valid_headers(self.emissions_data))
 
+    def test_has_valid_headers_different_order_success(self):
+        file_output = FileOutput("test.csv", self.temp_dir)
+        file_output.out(self.emissions_data, None)
+
+        df = pd.read_csv(os.path.join(self.temp_dir, "test.csv"))
+        df = df[list(reversed(df.columns))]
+        df.to_csv(os.path.join(self.temp_dir, "test.csv"), index=False)
+
+        self.assertTrue(file_output.has_valid_headers(self.emissions_data))
+
     def test_has_valid_headers_failure(self):
         file_output = FileOutput("test.csv", self.temp_dir)
-        file_output.out(self.emissions_data, MagicMock())
+        file_output.out(self.emissions_data, None)
 
         df = pd.read_csv(os.path.join(self.temp_dir, "test.csv"))
         df.rename(columns={"wue": "new_header"}, inplace=True)
@@ -90,10 +100,10 @@ def test_has_valid_headers_failure(self):
     @patch("codecarbon.output_methods.file.FileOutput.has_valid_headers")
     def test_file_output_out_file_exists_invalid_headers(self, mock_has_valid_headers):
         file_output = FileOutput("test.csv", self.temp_dir, on_csv_write="append")
-        file_output.out(self.emissions_data, MagicMock())
+        file_output.out(self.emissions_data, None)
 
         mock_has_valid_headers.return_value = False
-        file_output.out(self.emissions_data, MagicMock())
+        file_output.out(self.emissions_data, None)
 
         df = pd.read_csv(os.path.join(self.temp_dir, "test.csv.bak"))
         self.assertEqual(len(df), 1)
@@ -102,63 +112,67 @@ def test_file_output_out_file_exists_invalid_headers(self, mock_has_valid_header
 
     def test_file_output_out_update_no_file_exists(self):
         file_output = FileOutput("test.csv", self.temp_dir, on_csv_write="update")
-        file_output.out(self.emissions_data, MagicMock())
+        file_output.out(self.emissions_data, None)
 
         df = pd.read_csv(os.path.join(self.temp_dir, "test.csv"))
         self.assertEqual(len(df), 1)
 
     def test_file_output_out_append_no_file_exists(self):
         file_output = FileOutput("test.csv", self.temp_dir, on_csv_write="append")
-        file_output.out(self.emissions_data, MagicMock())
+        file_output.out(self.emissions_data, None)
 
         df = pd.read_csv(os.path.join(self.temp_dir, "test.csv"))
         self.assertEqual(len(df), 1)
 
     def test_file_output_out_append_file_exists(self):
         file_output = FileOutput("test.csv", self.temp_dir, on_csv_write="append")
-        file_output.out(self.emissions_data, MagicMock())
-        file_output.out(self.emissions_data, MagicMock())
+        file_output.out(self.emissions_data, None)
+        file_output.out(self.emissions_data, None)
 
         df = pd.read_csv(os.path.join(self.temp_dir, "test.csv"))
         self.assertEqual(len(df), 2)
 
     def test_file_output_out_update_file_exists_no_matching_row(self):
         file_output = FileOutput("test.csv", self.temp_dir, on_csv_write="update")
-        file_output.out(self.emissions_data, MagicMock())
+        file_output.out(self.emissions_data, None)
 
         updated_emissions_data = self.emissions_data
         updated_emissions_data.run_id = "new_test_run_id"
-        file_output.out(updated_emissions_data, MagicMock())
+        file_output.out(updated_emissions_data, None)
 
         df = pd.read_csv(os.path.join(self.temp_dir, "test.csv"))
         self.assertEqual(len(df), 2)
 
     def test_file_output_out_update_file_exists_multiple_matching_rows(self):
         file_output = FileOutput("test.csv", self.temp_dir, on_csv_write="update")
-        file_output.out(self.emissions_data, MagicMock())
+        file_output.out(self.emissions_data, None)
 
         # Manually add a duplicate row to simulate the condition
         df = pd.read_csv(os.path.join(self.temp_dir, "test.csv"))
         df = pd.concat([df, df])
         df.to_csv(os.path.join(self.temp_dir, "test.csv"), index=False)
 
-        file_output.out(self.emissions_data, MagicMock())
+        file_output.out(self.emissions_data, None)
 
         df = pd.read_csv(os.path.join(self.temp_dir, "test.csv"))
         self.assertEqual(len(df), 3)
 
     def test_file_output_out_update_file_exists_one_matchingrows(self):
         file_output = FileOutput("test.csv", self.temp_dir, on_csv_write="update")
-        file_output.out(self.emissions_data, MagicMock())
+        file_output.out(self.emissions_data, None)
         df = pd.read_csv(os.path.join(self.temp_dir, "test.csv"))
         self.assertEqual(df["cpu_power"].iloc[0], 20)
 
         new_data = self.emissions_data
         new_data.cpu_power = 2
-        file_output.out(new_data, MagicMock())
+        file_output.out(new_data, None)
         df = pd.read_csv(os.path.join(self.temp_dir, "test.csv"))
         self.assertEqual(df["cpu_power"].iloc[0], 2)
 
+    # def test_file_output_out_consistent_column_ordering(self):
+    #     file_output = FileOutput("test.csv", self.temp_dir, on_csv_write="append")
+    #     file_output.out(self.emissions_data, None)
+
     def test_file_output_task_out(self):
         task_emissions_data = [
             TaskEmissionsData(
diff --git a/tests/output_methods/test_http.py b/tests/output_methods/test_http.py
@@ -0,0 +1,183 @@
+import unittest
+from unittest.mock import MagicMock, patch
+
+from codecarbon.output_methods.emissions_data import EmissionsData
+from codecarbon.output_methods.http import CodeCarbonAPIOutput, HTTPOutput
+
+
+class TestHTTPOutput(unittest.TestCase):
+    def setUp(self):
+        self.emissions_data = EmissionsData(
+            timestamp="2023-01-01T00:00:00",
+            project_name="test_project",
+            run_id="test_run_id",
+            experiment_id="test_experiment_id",
+            duration=10,
+            emissions=0.5,
+            emissions_rate=0.05,
+            cpu_power=20,
+            gpu_power=30,
+            ram_power=5,
+            cpu_energy=200,
+            gpu_energy=300,
+            ram_energy=50,
+            energy_consumed=550,
+            water_consumed=0.1,
+            country_name="Testland",
+            country_iso_code="TS",
+            region="Test Region",
+            cloud_provider="Test Cloud",
+            cloud_region="test-cloud-1",
+            os="TestOS",
+            python_version="3.8",
+            codecarbon_version="2.0",
+            cpu_count=4,
+            cpu_model="Test CPU",
+            gpu_count=1,
+            gpu_model="Test GPU",
+            longitude=0,
+            latitude=0,
+            ram_total_size=16,
+            tracking_mode="machine",
+            on_cloud="true",
+            pue=1.5,
+            wue=0.5,
+        )
+        self.url = "http://test.com/emissions"
+        self.http_output = HTTPOutput(endpoint_url=self.url)
+
+    @patch(
+        "codecarbon.output_methods.http.requests.post",
+        return_value=MagicMock(status_code=201),
+    )
+    def test_http_output_post_success(self, mock_post):
+        self.http_output.out(self.emissions_data, self.emissions_data)
+
+        mock_post.assert_called_once()
+        self.assertEqual(mock_post.call_args[0][0], self.url)
+
+    @patch("codecarbon.output_methods.http.logger.warning")
+    @patch(
+        "codecarbon.output_methods.http.requests.post",
+        return_value=MagicMock(status_code=418),
+    )
+    def test_http_output_post_unexpected_status(self, mock_post, mock_logger):
+        self.http_output.out(self.emissions_data, self.emissions_data)
+
+        mock_post.assert_called_once()
+        mock_logger.assert_called_once()
+
+    @patch("codecarbon.output_methods.http.logger.error")
+    @patch(
+        "codecarbon.output_methods.http.requests.post",
+        side_effect=Exception("Test exception"),
+    )
+    def test_http_output_post_exception(self, mock_post, mock_logger):
+        self.http_output.out(self.emissions_data, self.emissions_data)
+        mock_post.assert_called_once()
+        mock_logger.assert_called_once()
+
+
+class TestCodeCarbonAPIOutput(unittest.TestCase):
+    def setUp(self):
+        self.emissions_data = EmissionsData(
+            timestamp="2023-01-01T00:00:00",
+            project_name="test_project",
+            run_id="test_run_id",
+            experiment_id="test_experiment_id",
+            duration=10,
+            emissions=0.5,
+            emissions_rate=0.05,
+            cpu_power=20,
+            gpu_power=30,
+            ram_power=5,
+            cpu_energy=200,
+            gpu_energy=300,
+            ram_energy=50,
+            energy_consumed=550,
+            water_consumed=0.1,
+            country_name="Testland",
+            country_iso_code="TS",
+            region="Test Region",
+            cloud_provider="Test Cloud",
+            cloud_region="test-cloud-1",
+            os="TestOS",
+            python_version="3.8",
+            codecarbon_version="2.0",
+            cpu_count=4,
+            cpu_model="Test CPU",
+            gpu_count=1,
+            gpu_model="Test GPU",
+            longitude=0,
+            latitude=0,
+            ram_total_size=16,
+            tracking_mode="machine",
+            on_cloud="true",
+            pue=1.5,
+            wue=0.5,
+        )
+        self.url = "http://test.com/emissions"
+        self.experiment_id = (
+            None  # Set to None so that ApiClient won't attempt a run on initialisation
+        )
+        self.api_key = "test_key"
+
+        self.add_emission_patcher = patch(
+            "codecarbon.output_methods.http.ApiClient.add_emission"
+        )
+        self.mock_add_emission = self.add_emission_patcher.start()
+        self.addCleanup(self.add_emission_patcher.stop)
+
+    def test_codecarbon_api_output_initialization(self):
+        CodeCarbonAPIOutput(
+            endpoint_url=self.url,
+            experiment_id=self.experiment_id,
+            api_key=self.api_key,
+            conf=None,
+        )
+
+    def test_codecarbon_api_live_out(self):
+        api_output = CodeCarbonAPIOutput(
+            endpoint_url=self.url,
+            experiment_id=self.experiment_id,
+            api_key=self.api_key,
+            conf=None,
+        )
+
+        api_output.live_out(None, self.emissions_data)
+        self.mock_add_emission.assert_called_once()
+
+    @patch("codecarbon.output_methods.http.logger.error")
+    def test_codecarbon_live_out_api_call_failure(self, mock_logger):
+        self.mock_add_emission.side_effect = Exception("Test exception")
+        api_output = CodeCarbonAPIOutput(
+            endpoint_url=self.url,
+            experiment_id=self.experiment_id,
+            api_key=self.api_key,
+            conf=None,
+        )
+        api_output.live_out(None, self.emissions_data)
+        mock_logger.assert_called_once()
+
+    def test_codecarbon_api_out(self):
+        api_output = CodeCarbonAPIOutput(
+            endpoint_url=self.url,
+            experiment_id=self.experiment_id,
+            api_key=self.api_key,
+            conf=None,
+        )
+
+        api_output.out(None, self.emissions_data)
+        self.mock_add_emission.assert_called_once()
+
+    @patch("codecarbon.output_methods.http.logger.error")
+    def test_codecarbon_out_api_call_failure(self, mock_logger):
+        self.mock_add_emission.side_effect = Exception("Test exception")
+        api_output = CodeCarbonAPIOutput(
+            endpoint_url=self.url,
+            experiment_id=self.experiment_id,
+            api_key=self.api_key,
+            conf=None,
+        )
+        api_output.out(None, self.emissions_data)
+        mock_logger.assert_called_once()
diff --git a/tests/output_methods/test_logfire.py b/tests/output_methods/test_logfire.py
diff --git a/uv.lock b/uv.lock

Original file line number	Diff line number	Diff line change
`@@ -90,6 +90,7 @@ dev = [`
`90`	`90`	`"requests",`
`91`	`91`	`"requests-mock",`
`92`	`92`	`"responses",`
	`93`	`+ "logfire>=1.0.1",`
`93`	`94`	`]`
`94`	`95`	`doc = [`
`95`	`96`	`"sphinx",`