test

python/cudf/cudf/pandas/scripts/conftest-patch.py

@@ -88,7 +88,7 @@ def pytest_runtest_teardown(item, nextitem):
     ):
         # Write the function call counts to a file
         worker_id = os.getenv("PYTEST_XDIST_WORKER", "master")
-        output_file = f'function_call_counts_{os.path.basename(item.nodeid.split("::")[0])}_{worker_id}.json'
+        output_file = f'{item.nodeid.split("::")[0].replace("/", "__")}_{worker_id}_metrics.json'
         with open(output_file, "w") as f:
             json.dump(dict(function_call_counts), f, indent=4)
         print(f"Function call counts have been written to {output_file}")

python/cudf/cudf/pandas/scripts/run-pandas-tests.sh

@@ -134,7 +134,8 @@ TEST_THAT_CRASH_PYTEST_WORKERS="not test_bitmasks_pyarrow \
 and not test_large_string_pyarrow \
 and not test_interchange_from_corrected_buffer_dtypes \
 and not test_eof_states \
-and not test_array_tz"
+and not test_array_tz \
+and not test_groupby_raises_category"
 
 # TODO: Remove "not db" once a postgres & mysql container is set up on the CI
 PANDAS_CI="1" timeout 600m python -m pytest -p cudf.pandas \
@@ -145,9 +146,13 @@ PANDAS_CI="1" timeout 600m python -m pytest -p cudf.pandas \
     "$@" || [ $? = 1 ]  # Exit success if exit code was 1 (permit test failures but not other errors)
 
 mv *.json ..
+pwd
 ls -al
+pwd
 ls -al tests/
 cd ..
+pwd
 ls -al
+pwd
 ls -al pandas-tests/
 rm -rf pandas-testing/pandas-tests/

python/cudf/cudf/pandas/scripts/summarize-test-results.py

@@ -12,7 +12,9 @@
 """
 
 import argparse
+import glob
 import json
+import os
 
 from rich.console import Console
 from rich.table import Table
@@ -57,6 +59,27 @@ def get_per_module_results(log_file_name):
                 per_module_results[module_name].setdefault(outcome, 0)
                 per_module_results[module_name]["total"] += 1
                 per_module_results[module_name][outcome] += 1
+
+    for key, value in per_module_results.items():
+        processed_name = key.replace("/", "__") + "_*_metrics.json"
+        # Assuming the directory is the same as the module name's directory
+        # directory = os.path.dirname(os.getcwd())
+        pattern = os.path.join(
+            "/nvme/0/pgali/cudf/pandas-testing", processed_name
+        )
+        matching_files = glob.glob(pattern)
+        for file in matching_files:
+            with open(file) as f:
+                function_call_counts = json.load(f)
+            per_module_results[key]["_slow_function_call"] = (
+                per_module_results[key].get("_slow_function_call", 0)
+                + function_call_counts.get("_slow_function_call", 0)
+            )
+            per_module_results[key]["_fast_function_call"] = (
+                per_module_results[key].get("_fast_function_call", 0)
+                + function_call_counts.get("_fast_function_call", 0)
+            )
+            # value["function_call_counts"] = function_call_counts
     return per_module_results
 
 

python/cudf/cudf_pandas_tests/conftest.py

@@ -56,7 +56,7 @@ def pytest_runtest_teardown(item, nextitem):
     ):
         # Write the function call counts to a file
         worker_id = os.getenv("PYTEST_XDIST_WORKER", "master")
-        output_file = f'function_call_counts_{os.path.basename(item.nodeid.split("::")[0])}_{worker_id}.json'
+        output_file = f'{item.nodeid.split("::")[0].replace("/", "__")}_{worker_id}_metrics.json'
         with open(output_file, "w") as f:
             json.dump(dict(function_call_counts), f)
             # for func, count in function_call_counts.items():