Merge pull request #2336 from Keerthana-64/ref-args-exact-name

fix: read reference arguments by exact name in revision APIs

Author: Byron

doc/gix-backend.md

@@ -374,7 +374,7 @@ and predate the current fallback; they do not establish native coverage.
 | Worktree inventory, `worktree list` | Main and linked worktree metadata, including bare main repositories and locks | Prunable entries; GIX-14 |
 | `Repo.merge_base`, `Repo.is_ancestor` | Native graph queries for two revisions | Octopus/fork-point and other options |
 | `Commit.count` | Reachable commit count, skip/limit/first-parent | Path filters and other revision options |
-| `Commit.iter_items`, `Repo.iter_commits` | First-parent walks and a single tip | General history ordering; GIX-8 |
+| `Commit.iter_items`, `Repo.iter_commits` | First-parent walks and a single tip | Reflog walks, general history ordering; GIX-8 |
 | Index entry reads, `ls_files` | Stage/mode/OID/path and exposed index flags | Custom and sparse indexes; GIX-3/4 |
 | `IndexFile.version`, `update_index` query | Native index version | Actual `update-index` mutations |
 | Index editing and `IndexFile.write` | Opaque byte snapshots; atomic locked publication | Private-file edits always use Git: arbitrary native index loading and strict pathname validation are missing; GIX-3/4/13/26 |
@@ -392,6 +392,25 @@ and predate the current fallback; they do not establish native coverage.
 | `Repo.untracked_files` | Native directory walk | Extra status options and non-root command directories |
 | `Repo.ignored` | Native excludes with tracked-file suppression | Symlink/submodule traversal, unsupported path normalization, and unavailable path metadata |
 
+Revision APIs preserve a reference object's validated full name. The shared
+`Git._check_revision()` first uses the exact reference reader to verify existence
+and the symbolic-reference chain; it passes the full name rather than the resolved
+object ID. Strings retain Git's name guessing, and commit objects retain their IDs.
+This prevents same-named tags from replacing branches without losing the ref
+identity needed by reflog walks, fork-point lookups and remote archives.
+`merge_base()` validates nested list/tuple operands before string conversion.
+`exact-refs.sh` covers these paths with shadowing tags and missing references in
+SHA-1/SHA-256 and files/reftable repositories on both backends.
+
+These name-sensitive operations retain CLI fallback. GixPython 0.3.0's
+`Repository.rev_walk()` has no Git-compatible reflog-walk mode,
+`Repository.merge_base()` exposes graph merge bases rather than reflog-based
+fork-point selection, and no upload-archive client API is exposed. The regressions
+compare Git 2.54.0's reflog history, old fork point after an upstream rewind,
+and default upload-archive server behavior, which rejects literal object IDs.
+Native replacements must expose matching reflog walking and fork-point selection
+and the upload-archive protocol; no Python implementation substitutes for them.
+
 Reference reads require an exact name, unlike revision APIs that intentionally
 apply Git's name guessing. `reference_info()` uses
 `Repository.try_find_reference()` and verifies the returned name for `HEAD` and

doc/source/changes.rst

@@ -57,7 +57,14 @@ API changes
 * Reference reads require the exact name. A missing branch no longer resolves to
   a similarly named tag, so an unborn branch's initial commit has no parent.
   A tag named ``REBASE_HEAD`` no longer reports a rebase in progress. Revision
-  APIs such as ``Repo.commit(rev)`` retain Git's name guessing.
+  APIs such as ``Repo.commit(rev)`` retain Git's name guessing for strings.
+  A reference object passed as a revision, for example to ``Repo.commit()``,
+  ``Repo.iter_commits()``, ``Repo.merge_base()``, ``IndexFile.from_tree()`` or
+  ``HEAD.reset()``, is validated by its exact name and passed using its full ref
+  name, so a tag named like a branch or remote-tracking branch no longer takes
+  its place. The ref name also preserves reflog walks, fork-point lookups and
+  remote archives. A missing reference raises ``ValueError``, as its ``object``
+  property does.
 * Packed tag names containing non-UTF-8 bytes are represented losslessly as
   surrogate-escaped strings. ``safe_encode()`` recovers their original bytes,
   and Gix reference validation and resolution receive those bytes directly.

git/cmd.py

@@ -1023,6 +1023,24 @@ def _check_operand(value: Any, label: str = "operand") -> str:
             raise UnsafeOptionError(f"Invalid {label}: {value!r}")
         return value
 
+    @staticmethod
+    def _check_revision(value: Any, label: str = "revision") -> str:
+        """Validate a revision, preserving a reference object's full name.
+
+        A reference converts to its short name, for which Git's revision lookup tries
+        other references, such as a same-named tag, before ``refs/heads/`` and
+        ``refs/remotes/``. Verify the exact reference exists before passing its full
+        name, retaining its identity for reflog walks and remote archives.
+        """
+        from git.refs.symbolic import SymbolicReference
+
+        if isinstance(value, SymbolicReference):
+            value.dereference_recursive(value.repo, value.path)
+            value = value.path
+        if isinstance(value, os.PathLike):
+            value = os.fsdecode(value)
+        return Git._check_operand(value, label)
+
     def _require_version(self) -> None:
         key = None
         if self._version_info_token is not self._refresh_token:

git/diff.py

@@ -261,8 +261,10 @@ def diff(
                 clusterable_short_options="46abceflmnpqrstuvwzBCDMNRW",
             )
 
-        if isinstance(other, str):
-            other = self.repo.rev_parse(Git._check_operand(other, "revision")).hexsha
+        from git.refs.symbolic import SymbolicReference
+
+        if isinstance(other, (str, SymbolicReference)):
+            other = self.repo.rev_parse(Git._check_revision(other)).hexsha
         native = _backend.tree_diff(self.repo, self, other, paths, create_patch, kwargs)
         if native is not NotImplemented:
             return native

git/index/base.py

@@ -280,8 +280,7 @@ def _tree_oid(self, tree: Treeish) -> str:
             tree = tree[0]
         if isinstance(tree, bytes):
             tree = tree.hex() if len(tree) == self.repo._oid_size else tree.decode("ascii")
-        value = str(tree)
-        Git._check_operand(value, "tree revision")
+        value = Git._check_revision(tree, "tree revision")
         return self.repo.git._call_process_safe("rev_parse", "--verify", "--end-of-options", value + "^{tree}")
 
     def _index_path(self) -> PathLike:

git/objects/base.py

@@ -9,6 +9,7 @@
 
 import gitdb.typ as dbtyp
 
+from git.cmd import Git
 from git.compat import force_text
 from git.exc import WorkTreeRepositoryUnsupported
 from git.util import LazyMixin, bin_to_hex, join_path_native, stream_copy
@@ -138,7 +139,7 @@ def new(cls, repo: "Repo", id: Union[str, "Reference"]) -> AnyGitObject:
             This cannot be a ``__new__`` method as it would always call :meth:`__init__`
             with the input id which is not necessarily a binsha.
         """
-        return repo.rev_parse(str(id))
+        return repo.rev_parse(Git._check_revision(id))
 
     @classmethod
     def new_from_sha(cls, repo: "Repo", sha1: bytes) -> AnyGitObject:

git/objects/commit.py

@@ -465,8 +465,8 @@ def iter_items(
             args_list.extend(paths_tup)
         # END if paths
 
-        Git._check_operand(str(rev), "revision")
-        native = _backend.history(repo.git, str(rev), paths, kwargs)
+        rev = Git._check_revision(rev)
+        native = _backend.history(repo.git, rev, paths, kwargs)
         if native is not NotImplemented:
             return (cls(repo, hex_to_bin(oid)) for oid in native)
         proc = repo.git._call_process_safe("rev_list", rev, args_list, as_process=True, **kwargs)

git/refs/head.py

@@ -121,7 +121,7 @@ def reset(
         # END working tree handling
 
         try:
-            oid = self.repo.rev_parse(Git._check_operand(commit, "revision")).hexsha
+            oid = self.repo.rev_parse(Git._check_revision(commit)).hexsha
             self.repo.git._call_process_safe("reset", mode, oid, "--", paths, **kwargs)
         except GitCommandError as e:
             # git nowadays may use 1 as status to indicate there are still unstaged

git/refs/symbolic.py

@@ -575,7 +575,7 @@ def _create(
     ) -> T_References:
         full_path = cls.to_full_path(path)
         cls._get_validated_ref_path(repo, full_path)
-        target = repo.rev_parse(str(reference)) if resolve else reference
+        target = repo.rev_parse(Git._check_revision(reference)) if resolve else reference
         ref = cls(repo, full_path)
         if force:
             ref.set_reference(target, logmsg)

git/refs/tag.py

@@ -170,7 +170,7 @@ def create(
 
         name = Git._check_operand(path, "tag")
         cls._get_validated_ref_path(repo, cls.to_full_path(name))
-        reference = Git._check_operand(reference, "revision")
+        reference = Git._check_revision(reference)
         oid = repo.rev_parse(reference).hexsha
         options = [] if allow_unsafe_options else ["--no-sign"]
         repo.git._call_process_safe("tag", *options, "--", name, oid, **kwargs)