Merge pull request #8334 from cr1901/json-no-changes

Suppress modified changes for files which weren't actually modified in JSON output.

src/borg/archiver.py

@@ -1121,11 +1121,22 @@ class Archiver:
     def do_diff(self, args, repository, manifest, key, archive):
         """Diff contents of two archives"""
 
+        def actual_change(j):
+            if j["type"] == "modified":
+                # Added/removed keys will not exist if chunker params differ
+                # between the two archives. Err on the side of caution and assume
+                # a real modification in this case (short-circuiting retrieving
+                # non-existent keys).
+                return not {"added", "removed"} <= j.keys() or not (j["added"] == 0 and j["removed"] == 0)
+            else:
+                # All other change types are indeed changes.
+                return True
+
         def print_json_output(diff, path):
-            print(json.dumps({"path": path, "changes": [j for j, str in diff]}, sort_keys=True, cls=BorgJsonEncoder))
+            print(json.dumps({"path": path, "changes": [j for j, str in diff if actual_change(j)]}, sort_keys=True, cls=BorgJsonEncoder))
 
         def print_text_output(diff, path):
-            print("{:<19} {}".format(' '.join([str for j, str in diff]), path))
+            print("{:<19} {}".format(' '.join([str for j, str in diff if actual_change(j)]), path))
 
         print_output = print_json_output if args.json_lines else print_text_output
 

src/borg/testsuite/__init__.py

@@ -191,6 +191,9 @@ class BaseTestCase(unittest.TestCase):
     def assert_line_exists(self, lines, expected_regexpr):
         assert any(re.search(expected_regexpr, line) for line in lines), f"no match for {expected_regexpr} in {lines}"
 
+    def assert_line_not_exists(self, lines, expected_regexpr):
+        assert not any(re.search(expected_regexpr, line) for line in lines), f"unexpected match for {expected_regexpr} in {lines}"
+
     def _assert_dirs_equal_cmp(self, diff, ignore_flags=False, ignore_xattrs=False, ignore_ns=False):
         self.assert_equal(diff.left_only, [])
         self.assert_equal(diff.right_only, [])

src/borg/testsuite/archiver.py

@@ -23,6 +23,7 @@ from datetime import timedelta
 from hashlib import sha256
 from io import BytesIO, StringIO
 from unittest.mock import patch
+from pathlib import Path
 
 import pytest
 
@@ -4472,6 +4473,7 @@ class DiffArchiverTestCase(ArchiverTestCaseBase):
         self.create_regular_file('file_removed', size=256)
         self.create_regular_file('file_removed2', size=512)
         self.create_regular_file('file_replaced', size=1024)
+        self.create_regular_file('file_touched', size=128)
         os.mkdir('input/dir_replaced_with_file')
         os.chmod('input/dir_replaced_with_file', stat.S_IFDIR | 0o755)
         os.mkdir('input/dir_removed')
@@ -4500,6 +4502,7 @@ class DiffArchiverTestCase(ArchiverTestCaseBase):
         self.create_regular_file('file_replaced', contents=b'0' * 4096)
         os.unlink('input/file_removed')
         os.unlink('input/file_removed2')
+        Path('input/file_touched').touch()
         os.rmdir('input/dir_replaced_with_file')
         self.create_regular_file('dir_replaced_with_file', size=8192)
         os.chmod('input/dir_replaced_with_file', stat.S_IFREG | 0o755)
@@ -4562,6 +4565,15 @@ class DiffArchiverTestCase(ArchiverTestCaseBase):
             change = '0 B' if can_compare_ids else '{:<19}'.format('modified')
             self.assert_line_exists(lines, f"{change}.*input/empty")
 
+            # Do not show a 0 byte change for a file whose contents weren't
+            # modified.
+            self.assert_line_not_exists(lines, '0 B.*input/file_touched')
+            if not content_only:
+                self.assert_line_exists(lines, "[cm]time:.*input/file_touched")
+            else:
+                # And if we're doing content-only, don't show the file at all.
+                assert "input/file_touched" not in output
+
             if are_hardlinks_supported():
                 self.assert_line_exists(lines, f"{change}.*input/hardlink_contents_changed")
             if are_symlinks_supported():
@@ -4610,6 +4622,16 @@ class DiffArchiverTestCase(ArchiverTestCaseBase):
             # File unchanged
             assert not any(get_changes('input/file_unchanged', joutput))
 
+            # Do not show a 0 byte change for a file whose contents weren't
+            # modified.
+            unexpected = {'type': 'modified', 'added': 0, 'removed': 0}
+            assert unexpected not in get_changes('input/file_touched', joutput)
+            if not content_only:
+                assert {"ctime", "mtime"}.issubset({c["type"] for c in get_changes('input/file_touched', joutput)})
+            else:
+                # And if we're doing content-only, don't show the file at all.
+                assert not any(get_changes('input/file_touched', joutput))
+
             # Directory replaced with a regular file
             if 'BORG_TESTS_IGNORE_MODES' not in os.environ and not content_only:
                 assert {'type': 'mode', 'old_mode': 'drwxr-xr-x', 'new_mode': '-rwxr-xr-x'} in \