update: improve test selection performance

improve code for test selection to prevent multiple database calls

Author: guotin

pytest_rts/tests/test_git.py

@@ -13,7 +13,7 @@
  [
  (
  FAKE_DIFF_1,
- [
+ {
  23,
  29,
  30,
@@ -42,10 +42,10 @@
  219,
  228,
  238,
- ],
+ },
  ),
- (FAKE_DIFF_2, [83, 240]),
- (FAKE_DIFF_3, [16, 24]),
+ (FAKE_DIFF_2, {83, 240}),
+ (FAKE_DIFF_3, {16, 24}),
  ],
 )
 def test_get_changed_lines(diff: str, real_changed_lines: List[int]) -> None:

pytest_rts/utils/common.py

@@ -50,15 +50,21 @@ def get_tests_from_changes(
 
  tests: Set[str] = set()
  for changed_file in file_diffs:
- changed_lines = get_changed_lines(file_diffs[changed_file])
- for line_number in coverage_data.contexts_by_lineno(changed_file):
- if line_number in changed_lines:
- tests.update(
- strip_pytest_cov_testname(testname)
- for testname in coverage_data.contexts_by_lineno(changed_file)[
- line_number
- ]
- )
+
+ contexts = coverage_data.contexts_by_lineno(changed_file)
+ if not contexts:
+ continue
+
+ changed_lines_with_tests = get_changed_lines(
+ file_diffs[changed_file]
+ ).intersection(contexts.keys())
+
+ tests.update(
+ strip_pytest_cov_testname(testname)
+ for line in changed_lines_with_tests
+ for testname in contexts[line]
+ )
+
  return tests
 
 

pytest_rts/utils/git.py

@@ -1,7 +1,7 @@
 """This module contains code for git operations"""
 import re
 import subprocess
-from typing import Dict, List
+from typing import Dict, List, Set
 
 from pydriller import GitRepository
 
@@ -23,7 +23,7 @@ def get_file_diff_dict_current(files: List[str]) -> Dict[str, str]:
  return {file_path: get_file_diff_data_current(file_path) for file_path in files}
 
 
-def get_changed_lines(diff: str) -> List[int]:
+def get_changed_lines(diff: str) -> Set[int]:
  """Parse changed lines from git diff -U0 output.
  - Change data according to git diff output:
  - @@ -old0,old1 +new0,new1 @@
@@ -32,7 +32,7 @@ def get_changed_lines(diff: str) -> List[int]:
  """
  regex = r"[@][@]\s+[-][0-9]+(?:,[0-9]+)?\s+[+][0-9]+(?:,[0-9]+)?\s+[@][@]"
  line_changes = re.findall(regex, diff)
- changed_lines = []
+ changed_lines: Set[int] = set()
  for change in line_changes:
  changed_line = change.strip("@").split()
 
@@ -45,9 +45,9 @@ def get_changed_lines(diff: str) -> List[int]:
  old[0] = old[0].strip("-")
 
  if int(old[1]) == 0:
- changed_lines.append(int(old[0]))
+ changed_lines.add(int(old[0]))
  else:
- changed_lines.extend(range(int(old[0]), int(old[0]) + int(old[1])))
+ changed_lines.update(range(int(old[0]), int(old[0]) + int(old[1])))
 
  return changed_lines