Add end to end test

Author: msg555

dirdiff/cli.py

@@ -22,7 +22,7 @@
 }
 
 
-def parse_args():
+def parse_args(args=None) -> argparse.Namespace:
     parser = argparse.ArgumentParser(
         description="Computes the directory difference `upper = merged - lower`"
     )
@@ -113,7 +113,7 @@ def parse_args():
             const=True,
             help="Attempt to chown files to preserve ownership, only applies to file-based output",
         )
-    return parser.parse_args()
+    return parser.parse_args(args=args)
 
 
 def setup_logging(verbose: int) -> None:
@@ -185,8 +185,8 @@ def _get_backend(
     return OutputBackendFile(output, preserve_owners=preserve_owners)
 
 
-def main() -> int:
-    args = parse_args()
+def main(args=None) -> int:
+    args = parse_args(args=args)
     setup_logging(1 + args.verbose - args.quiet)
 
     with ExitStack() as stack:
@@ -203,7 +203,7 @@ def main() -> int:
                     args.output_type,
                     args.output,
                     args.force,
-                    args.preserve_owners,
+                    getattr(args, "preserve_owners", False),
                 )
             )
 

dirdiff/output_tar.py

@@ -17,7 +17,15 @@ def __init__(self, tf: TarFile, *, archive_root=".") -> None:
         self.archive_root = archive_root
 
     def _get_tar_info(self, name: str, st: StatInfo) -> TarInfo:
-        ti = TarInfo(os.path.join(self.archive_root, name))
+        if name in (".", os.path.sep):
+            arch_name = self.archive_root
+        elif name.startswith(f".{os.path.sep}"):
+            arch_name = os.path.join(self.archive_root, name[2:])
+        elif name.startswith(os.path.sep):
+            arch_name = os.path.join(self.archive_root, name[1:])
+        else:
+            arch_name = os.path.join(self.archive_root, name)
+        ti = TarInfo(arch_name)
         ti.mode = stat.S_IMODE(st.mode)
         ti.mtime = st.mtime
         ti.uid = st.uid

tests/cases/generic/diff.tgz

@@ -0,0 +1,68 @@
+import hashlib
+import os
+import tarfile
+import tempfile
+
+from dirdiff.cli import main
+
+
+def tar_summarize(tf: tarfile.TarFile):
+    result = {}
+    while ti := tf.next():
+        assert ti.name == "." or ti.name.startswith(f".{os.path.sep}")
+        assert ti.name not in result
+        parent = os.path.dirname(ti.name)
+
+        if parent:
+            assert parent in result
+
+        content_hash = ""
+        if ti.isfile():
+            hsh = hashlib.sha256()
+            f = tf.extractfile(ti)
+            assert f is not None
+            with f:
+                while data := f.read(2**16):
+                    hsh.update(data)
+            content_hash = hsh.hexdigest()
+
+        result[ti.name] = dict(
+            uid=ti.uid,
+            gid=ti.gid,
+            mode=ti.mode,
+            size=ti.size,
+            mtime=ti.mtime,
+            typ=ti.type,
+            content_hash=content_hash,
+        )
+    return result
+
+
+def run_test(name: str) -> None:
+    test_path = os.path.join(os.path.dirname(__file__), "cases", name)
+    with tempfile.NamedTemporaryFile(delete=False) as tmpf:
+        tmp_name = tmpf.name
+
+    try:
+        args = [
+            os.path.join(test_path, "merged.tgz"),
+            os.path.join(test_path, "lower.tgz"),
+            "--output",
+            tmp_name,
+        ]
+        result = main(args=args)
+        assert result == 0
+
+        with tarfile.open(tmp_name, mode="r|") as tf:
+            summary = tar_summarize(tf)
+    finally:
+        os.remove(tmp_name)
+
+    with tarfile.open(os.path.join(test_path, "diff.tgz"), mode="r|gz") as tf:
+        expected_summary = tar_summarize(tf)
+
+    assert summary == expected_summary
+
+
+def test_generic():
+    run_test("generic")