improve portability of reproducible tarballs by replacing external tar command with tarfile module

easybuild/tools/filetools.py

@@ -2772,15 +2772,32 @@ def get_source_tarball_from_git(filename, target_dir, git_config):
 
     if keep_git_dir:
         # create archive of git repo including .git directory
-        tar_cmd = ['tar', 'cfvz', archive_path, repo_name]
+        tar_cmd = f"tar cfvz {archive_path} {repo_name}"
     else:
         # create reproducible archive
-        # see https://reproducible-builds.org/docs/archives/
-        tar_cmd = [
+        tar_cmd = reproducible_archive_cmd(repo_name, archive_path)
+
+    run_shell_cmd(tar_cmd, work_dir=tmpdir, hidden=True, verbose_dry_run=True)
+
+    # cleanup (repo_name dir does not exist in dry run mode)
+    remove(tmpdir)
+
+    return archive_path
+
+
+def reproducible_archive_cmd(dir_name, archive_name):
+    """
+    Return string with command to make reproducible archive from a given directory
+    see https://reproducible-builds.org/docs/archives/
+    """
+    try:
+        cmd_pipe = [
+            # stop on failure of any command in the pipe
+            'set', '-eo pipefail', ';',
             # print names of all files and folders excluding .git directory
-            'find', repo_name, '-name ".git"', '-prune', '-o', '-print0',
-            # reset access and modification timestamps to epoch 0 (equivalent to --mtime in GNU tar)
-            '-exec', 'touch', '--date=@0', '{}', r'\;',
+            'find', str(dir_name), '-name ".git"', '-prune', '-o', '-print0',
+            # reset access and modification timestamps to epoch 0
+            '-exec', 'touch', '--date=1970-01-01T00:00:00.00Z', '{}', r'\;',
             # reset file permissions of cloned repo (equivalent to --mode in GNU tar)
             '-exec', 'chmod', '"go+u,go-w"', '{}', r'\;', '|',
             # sort file list (equivalent to --sort in GNU tar)
@@ -2789,14 +2806,12 @@ def get_source_tarball_from_git(filename, target_dir, git_config):
             'tar', '--create', '--no-recursion', '--owner=0', '--group=0', '--numeric-owner',
             '--format=gnu', '--null', '--files-from', '-', '|',
             # compress tarball with gzip without original file name and timestamp
-            'gzip', '--no-name', '>', archive_path
+            'gzip', '--no-name', '>', str(archive_name)
         ]
-    run_shell_cmd(' '.join(tar_cmd), work_dir=tmpdir, hidden=True, verbose_dry_run=True)
+    except TypeError as err:
+        raise EasyBuildError("reproducible_archive_cmd: wrong directory or archive name given") from err
 
-    # cleanup (repo_name dir does not exist in dry run mode)
-    remove(tmpdir)
-
-    return archive_path
+    return " ".join(cmd_pipe)
 
 
 def move_file(path, target_path, force_in_dry_run=False):

test/framework/filetools.py

@@ -2939,16 +2939,11 @@ def run_check():
             'git_repo': '[email protected]:easybuilders/testrepository.git',
             'test_prefix': self.test_prefix,
         }
-        reprod_tar_cmd_pattern = (
-            r' running shell command "find {} -name \".git\" -prune -o -print0 -exec touch -t 197001010100 {{}} \; |'
-            r' LC_ALL=C sort --zero-terminated | tar --create --no-recursion --owner=0 --group=0 --numeric-owner'
-            r' --format=gnu --null --files-from - | gzip --no-name > %(test_prefix)s/target/test.tar.gz'
-        )
 
         expected = '\n'.join([
             r'  running shell command "git clone --depth 1 --branch tag_for_tests %(git_repo)s"',
             r"  \(in .*/tmp.*\)",
-            reprod_tar_cmd_pattern.format("testrepository"),
+            ft.reproducible_archive_cmd("testrepository", "%(test_prefix)s/target/test.tar.gz"),
             r"  \(in .*/tmp.*\)",
         ]) % string_args
         run_check()
@@ -2957,7 +2952,7 @@ def run_check():
         expected = '\n'.join([
             r'  running shell command "git clone --depth 1 --branch tag_for_tests %(git_repo)s test123"',
             r"  \(in .*/tmp.*\)",
-            reprod_tar_cmd_pattern.format("test123"),
+            ft.reproducible_archive_cmd("test123", "%(test_prefix)s/target/test.tar.gz"),
             r"  \(in .*/tmp.*\)",
         ]) % string_args
         run_check()
@@ -2967,7 +2962,7 @@ def run_check():
         expected = '\n'.join([
             r'  running shell command "git clone --depth 1 --branch tag_for_tests --recursive %(git_repo)s"',
             r"  \(in .*/tmp.*\)",
-            reprod_tar_cmd_pattern.format("testrepository"),
+            ft.reproducible_archive_cmd("testrepository", "%(test_prefix)s/target/test.tar.gz"),
             r"  \(in .*/tmp.*\)",
         ]) % string_args
         run_check()
@@ -2977,7 +2972,7 @@ def run_check():
             '  running shell command "git clone --depth 1 --branch tag_for_tests --recursive'
             + ' --recurse-submodules=\'!vcflib\' --recurse-submodules=\'!sdsl-lite\' %(git_repo)s"',
             r"  \(in .*/tmp.*\)",
-            reprod_tar_cmd_pattern.format("testrepository"),
+            ft.reproducible_archive_cmd("testrepository", "%(test_prefix)s/target/test.tar.gz"),
             r"  \(in .*/tmp.*\)",
         ]) % string_args
         run_check()
@@ -2991,7 +2986,7 @@ def run_check():
             + ' clone --depth 1 --branch tag_for_tests --recursive'
             + ' --recurse-submodules=\'!vcflib\' --recurse-submodules=\'!sdsl-lite\' %(git_repo)s"',
             r"  \(in .*/tmp.*\)",
-            reprod_tar_cmd_pattern.format("testrepository"),
+            ft.reproducible_archive_cmd("testrepository", "%(test_prefix)s/target/test.tar.gz"),
             r"  \(in .*/tmp.*\)",
         ]) % string_args
         run_check()
@@ -3015,7 +3010,7 @@ def run_check():
             r"  \(in .*/tmp.*\)",
             r'  running shell command "git checkout 8456f86 && git submodule update --init --recursive"',
             r"  \(in testrepository\)",
-            reprod_tar_cmd_pattern.format("testrepository"),
+            ft.reproducible_archive_cmd("testrepository", "%(test_prefix)s/target/test.tar.gz"),
             r"  \(in .*/tmp.*\)",
         ]) % string_args
         run_check()
@@ -3026,7 +3021,7 @@ def run_check():
             r"  \(in .*/tmp.*\)",
             r'  running shell command "git checkout 8456f86"',
             r"  \(in testrepository\)",
-            reprod_tar_cmd_pattern.format("testrepository"),
+            ft.reproducible_archive_cmd("testrepository", "%(test_prefix)s/target/test.tar.gz"),
             r"  \(in .*/tmp.*\)",
         ]) % string_args
         run_check()
@@ -3038,7 +3033,7 @@ def run_check():
             r"  \(in /.*\)",
             r'  running shell command "git checkout 8456f86"',
             r"  \(in /.*/testrepository\)",
-            reprod_tar_cmd_pattern.format("testrepository"),
+            ft.reproducible_archive_cmd("testrepository", "%(test_prefix)s/target/test.tar.gz"),
             r"  \(in /.*\)",
         ]) % string_args
         run_check()