get_metrics.py

import argparse
import subprocess
from typing import Iterable, List, Tuple

_TEAM_AUTHORS = frozenset([
    'Brendan Higgins', 'David Gow', 'Heidi Fahim', 'Felix Guo',
    'Avinash Kondareddy', 'Daniel Latypov'
])

# Manual list of files to exclude in `_test_names`.
_NOT_TESTS=frozenset(['lib/kunit/test.c'])

def _outside_contributors_and_patches() -> Tuple[int, int]:
  cmd = ['git', 'log', '--format=%an', '--', 'include/kunit', 'lib/kunit', 'tools/testing/kunit']
  contributors = set()
  patches = 0
  with subprocess.Popen(cmd, stdout=subprocess.PIPE, text=True) as p:
    for author in p.stdout:
      if author.strip() not in _TEAM_AUTHORS:
        contributors.add(author)
        patches += 1
  return len(contributors), patches

def _is_test_file(filename: str) -> bool:
  if filename.startswith('Documentation/') or filename in _NOT_TESTS:
    return False
  return 'test' in filename or filename.endswith('_kunit.c')

def _test_names() -> List[str]:
  cmd = ['git', 'grep', '-l',  '-e', '#include <kunit/test.h']
  with subprocess.Popen(cmd, stdout=subprocess.PIPE, text=True) as p:
    return list(filter(_is_test_file, (line.strip() for line in p.stdout)))

def _test_authors(test_names: List[str]) -> List[str]:
  cmd = ['git', '--no-pager', 'shortlog', '--no-merges', '-n', '-s', '--'] + test_names
  return subprocess.check_output(cmd, text=True).strip().split('\n')

def _num_test_cases() -> int:
  cmd = ['git', 'grep', 'KUNIT_CASE']
  count = 0
  with subprocess.Popen(cmd, stdout=subprocess.PIPE, text=True) as p:
    for line in p.stdout:
      line = line.strip()
      if line.startswith('Documentation/') or line.startswith('include/kunit/test.h'):
        continue
      count += 1
    return count


parser = argparse.ArgumentParser(description='Optional app description')
parser.add_argument('-v', '--verbose', action='store_true')

args = parser.parse_args()

num_test_cases = _num_test_cases()
test_names = _test_names()
contributors, patches = _outside_contributors_and_patches()

print(f'Number of test cases: {num_test_cases}')
print(f'Number of patches: {patches}')
print(f'Unique contributors: {contributors}')
print(f'Number of total tests: {len(test_names)}')

test_authors = _test_authors(test_names)  # this step takes ~15s as of Aug 2022
print(f'Number of test authors: {len(test_authors)}')

if args.verbose:
  print(f'All tests: ' + '\n'.join(test_names))
  print(f'Test authors: ' + '\n'.join(test_authors))