Revert "SERVER-71548 Avoid deleting files after symbolization"

This reverts commit 9e64ace5f63595b8a6f1609cf24ede6d36693767.

2 files changed, 42 insertions, 82 deletions

diff --git a/buildscripts/resmokelib/testing/symbolizer_service.py b/buildscripts/resmokelib/testing/symbolizer_service.py
index 866a4209239..be4ca070b45 100644
--- a/buildscripts/resmokelib/testing/symbolizer_service.py
+++ b/buildscripts/resmokelib/testing/symbolizer_service.py
@@ -8,7 +8,7 @@ import time
 from datetime import timedelta
 from threading import Lock
 
-from typing import List, Optional, NamedTuple, Set
+from typing import List, Optional, NamedTuple
 
 from buildscripts.resmokelib import config as _config
 from buildscripts.resmokelib.testing.testcases.interface import TestCase
@@ -19,7 +19,6 @@ _lock = Lock()
 
 STACKTRACE_FILE_EXTENSION = ".stacktrace"
 SYMBOLIZE_RETRY_TIMEOUT_SECS = timedelta(minutes=4).total_seconds()
-PROCESSED_FILES_LIST_FILE_PATH = "symbolizer-processed-files.txt"  # noqa
 
 
 class ResmokeSymbolizerConfig(NamedTuple):
@@ -70,8 +69,7 @@ class ResmokeSymbolizer:
         )
         self.symbolizer_service = symbolizer_service if symbolizer_service is not None else SymbolizerService(
         )
-        self.file_service = file_service if file_service is not None else FileService(
-            PROCESSED_FILES_LIST_FILE_PATH)
+        self.file_service = file_service if file_service is not None else FileService()
 
     def symbolize_test_logs(self, test: TestCase,
                             symbolize_retry_timeout: float = SYMBOLIZE_RETRY_TIMEOUT_SECS) -> None:
@@ -109,9 +107,8 @@ class ResmokeSymbolizer:
                 if time.perf_counter() - start_time > symbolize_retry_timeout:
                     break
 
-            # To avoid performing the same actions on these files again, we mark them as processed
-            self.file_service.add_to_processed_files(files)
-            self.file_service.write_processed_files(PROCESSED_FILES_LIST_FILE_PATH)
+            # To avoid performing the same actions on these files again, we remove them
+            self.file_service.remove_all(files)
 
             test.logger.info("\nEND Symbolization \nSymbolization process completed. ")
 
@@ -165,9 +162,8 @@ class ResmokeSymbolizer:
 
         files = self.file_service.find_all_children_recursively(dir_path)
         files = self.file_service.filter_by_extension(files, STACKTRACE_FILE_EXTENSION)
-        files = self.file_service.filter_out_empty_files(files)
+        self.file_service.remove_empty(files)
         files = self.file_service.filter_out_non_files(files)
-        files = self.file_service.filter_out_already_processed_files(files)
 
         return files
 
@@ -175,51 +171,6 @@ class ResmokeSymbolizer:
 class FileService:
     """A service for working with files."""
 
-    def __init__(self, processed_files_list_path: str = PROCESSED_FILES_LIST_FILE_PATH):
-        """Initialize FileService instance."""
-        self._processed_files = self.load_processed_files(processed_files_list_path)
-
-    @staticmethod
-    def load_processed_files(file_path: str) -> Set[str]:
-        """
-        Load processed files info from a file.
-
-        :param: path to a file where we store processed files info.
-        """
-        if os.path.exists(file_path):
-            with open(file_path, "r") as file:
-                return {line for line in set(file.readlines()) if line}
-        return set()
-
-    def add_to_processed_files(self, files: List[str]) -> None:
-        """
-        Bulk add to collection of processed files.
-
-        :param files: files to add to processed files collection
-        :return: None
-        """
-        for file in files:
-            self._processed_files.add(file)
-
-    def write_processed_files(self, file_path: str) -> None:
-        """
-        Write processed files info to a file.
-
-        :param file_path: path to a file where we store processed files info
-        :return: None
-        """
-        with open(file_path, "w") as file:
-            file.write("\n".join(self._processed_files))
-
-    def is_processed(self, file: str) -> bool:
-        """
-        Check if file is already processed or not.
-
-        :param file: file path
-        :return: whether the file is already processed or not
-        """
-        return file in self._processed_files
-
     @staticmethod
     def find_all_children_recursively(dir_path: str) -> List[str]:
         """
@@ -254,24 +205,25 @@ class FileService:
         """
         return [f for f in files if os.path.isfile(f)]
 
-    def filter_out_already_processed_files(self, files: List[str]):
+    @staticmethod
+    def remove_empty(files: List[str]) -> None:
         """
-        Filter out already processed files.
+        Delete files that are empty.
 
-        :param files: list of file paths
-        :return: non-processed files
+        :param files: list of paths
         """
-        return [f for f in files if not self.is_processed(f)]
+        for file in [f for f in files if os.stat(f).st_size == 0]:
+            os.remove(file)
 
     @staticmethod
-    def filter_out_empty_files(files: List[str]) -> List[str]:
+    def remove_all(files: List[str]) -> None:
         """
-        Filter our files that are empty.
+        Delete all files.
 
         :param files: list of paths
-        :return: Non-empty files
         """
-        return [f for f in files if not os.stat(f).st_size == 0]
+        for file in files:
+            os.remove(file)
 
     @staticmethod
     def check_path_exists(path: str) -> bool:
diff --git a/buildscripts/tests/resmokelib/testing/test_symbolizer_service.py b/buildscripts/tests/resmokelib/testing/test_symbolizer_service.py
index f2bf4ecabff..805182442ed 100644
--- a/buildscripts/tests/resmokelib/testing/test_symbolizer_service.py
+++ b/buildscripts/tests/resmokelib/testing/test_symbolizer_service.py
@@ -36,6 +36,7 @@ class TestResmokeSymbolizer(unittest.TestCase):
         self.assertEqual(self.symbolizer_service_mock.run_symbolizer_script.call_count, 5)
         for i, call in enumerate(self.symbolizer_service_mock.run_symbolizer_script.call_arg_list):
             self.assertEqual(call.args[0], f"file{i}.stacktrace")
+        self.file_service_mock.remove_all.assert_called_once_with(stacktrace_files)
 
     def test_symbolize_test_logs_hit_timeout(self):
         stacktrace_files = [f"file{i}.stacktrace" for i in range(5)]
@@ -46,6 +47,7 @@ class TestResmokeSymbolizer(unittest.TestCase):
         self.assertEqual(self.symbolizer_service_mock.run_symbolizer_script.call_count, 1)
         for i, call in enumerate(self.symbolizer_service_mock.run_symbolizer_script.call_arg_list):
             self.assertEqual(call.args[0], f"file{i}.stacktrace")
+        self.file_service_mock.remove_all.assert_called_once_with(stacktrace_files)
 
     def test_symbolize_test_logs_should_not_symbolize(self):
         self.config_mock.is_windows.return_value = True
@@ -189,10 +191,11 @@ class TestFileService(unittest.TestCase):
                 with open(file, "w") as fstream:
                     fstream.write("stacktrace")
 
-            self.assertEqual(
-                set(self.file_service.filter_out_empty_files(abs_file_paths)), set(abs_file_paths))
+            self.file_service.remove_empty(abs_file_paths)
+            for file in abs_file_paths:
+                self.assertTrue(os.path.exists(file))
 
-    def test_filter_out_empty_files_if_partly_empty(self):
+    def test_remove_empty_files_if_partly_empty(self):
         with TemporaryDirectory() as tmpdir:
             abs_dir_paths = [os.path.join(tmpdir, d) for d in self.relative_dir_paths]
             abs_file_paths = [os.path.join(tmpdir, f) for f in self.relative_file_paths]
@@ -205,14 +208,14 @@ class TestFileService(unittest.TestCase):
                 fstream.write("stacktrace")
             Path(abs_file_paths[3]).touch()
 
-            filtered = self.file_service.filter_out_empty_files(abs_file_paths)
+            self.file_service.remove_empty(abs_file_paths)
 
-            self.assertTrue(abs_file_paths[0] in filtered)
-            self.assertFalse(abs_file_paths[1] in filtered)
-            self.assertTrue(abs_file_paths[2] in filtered)
-            self.assertFalse(abs_file_paths[3] in filtered)
+            self.assertTrue(os.path.exists(abs_file_paths[0]))
+            self.assertFalse(os.path.exists(abs_file_paths[1]))
+            self.assertTrue(os.path.exists(abs_file_paths[2]))
+            self.assertFalse(os.path.exists(abs_file_paths[3]))
 
-    def test_filter_out_empty_files_if_all_empty(self):
+    def test_remove_empty_files_if_all_empty(self):
         with TemporaryDirectory() as tmpdir:
             abs_dir_paths = [os.path.join(tmpdir, d) for d in self.relative_dir_paths]
             abs_file_paths = [os.path.join(tmpdir, f) for f in self.relative_file_paths]
@@ -221,15 +224,20 @@ class TestFileService(unittest.TestCase):
             for file in abs_file_paths:
                 Path(file).touch()
 
-            filtered = self.file_service.filter_out_empty_files(abs_file_paths)
+            self.file_service.remove_empty(abs_file_paths)
             for file in abs_file_paths:
-                self.assertFalse(file in filtered)
+                self.assertFalse(os.path.exists(file))
 
-    def test_filter_out_already_processed_files(self):
-        processed_files = ["processed-file.stacktrace"]
-        files = [
-            "file.stacktrace", "other-file.stacktrace", "another-file.stacktrace", *processed_files
-        ]
-        self.file_service.add_to_processed_files(processed_files)
-        filtered = self.file_service.filter_out_already_processed_files(files)
-        self.assertTrue(all(file not in processed_files for file in filtered))
+    def test_remove_all_files(self):
+        with TemporaryDirectory() as tmpdir:
+            abs_dir_paths = [os.path.join(tmpdir, d) for d in self.relative_dir_paths]
+            abs_file_paths = [os.path.join(tmpdir, f) for f in self.relative_file_paths]
+            for dir_ in abs_dir_paths:
+                Path(dir_).mkdir(parents=True)
+            for file in abs_file_paths:
+                with open(file, "w") as fstream:
+                    fstream.write("stacktrace")
+
+            self.file_service.remove_all(abs_file_paths)
+            for file in abs_file_paths:
+                self.assertFalse(os.path.exists(file))