Add fallback for find_files on git/hg archives

setup.cfg

@@ -50,6 +50,9 @@ setuptools.finalize_distribution_options =
 setuptools_scm.files_command =
     .hg = setuptools_scm.file_finder_hg:hg_find_files
     .git = setuptools_scm.file_finder_git:git_find_files
+setuptools_scm.files_command_fallback =
+    .hg_archival.txt = setuptools_scm.file_finder_hg:hg_archive_find_files
+    .git_archival.txt = setuptools_scm.file_finder_git:git_archive_find_files
 setuptools_scm.local_scheme =
     node-and-date = setuptools_scm.version:get_local_node_and_date
     node-and-timestamp = setuptools_scm.version:get_local_node_and_timestamp

src/setuptools_scm/file_finder.py

@@ -11,7 +11,10 @@
 
 
 def scm_find_files(
-    path: _t.PathT, scm_files: set[str], scm_dirs: set[str]
+    path: _t.PathT,
+    scm_files: set[str],
+    scm_dirs: set[str],
+    force_all_files: bool = False,
 ) -> list[str]:
     """ setuptools compatible file finder that follows symlinks
 
@@ -20,6 +23,7 @@ def scm_find_files(
       (including symlinks to directories)
     - scm_dirs: set of scm controlled directories
       (including directories containing no scm controlled files)
+    - force_all_files: ignore ``scm_files`` and ``scm_dirs`` and list everything.
 
     scm_files and scm_dirs must be absolute with symlinks resolved (realpath),
     with normalized case (normcase)
@@ -38,7 +42,7 @@ def _link_not_in_scm(n: str) -> bool:
             fn = os.path.join(realdirpath, os.path.normcase(n))
             return os.path.islink(fn) and fn not in scm_files
 
-        if realdirpath not in scm_dirs:
+        if not force_all_files and realdirpath not in scm_dirs:
             # directory not in scm, don't walk it's content
             dirnames[:] = []
             continue
@@ -54,13 +58,16 @@ def _link_not_in_scm(n: str) -> bool:
             # symlink loop protection
             dirnames[:] = []
             continue
-        dirnames[:] = [dn for dn in dirnames if not _link_not_in_scm(dn)]
+        dirnames[:] = [
+            dn for dn in dirnames if force_all_files or not _link_not_in_scm(dn)
+        ]
         for filename in filenames:
-            if _link_not_in_scm(filename):
+            if not force_all_files and _link_not_in_scm(filename):
                 continue
             # dirpath + filename with symlinks preserved
             fullfilename = os.path.join(dirpath, filename)
-            if os.path.normcase(os.path.realpath(fullfilename)) in scm_files:
+            is_tracked = os.path.normcase(os.path.realpath(fullfilename)) in scm_files
+            if force_all_files or is_tracked:
                 res.append(os.path.join(path, os.path.relpath(fullfilename, realpath)))
         seen.add(realdirpath)
     return res

src/setuptools_scm/file_finder_git.py

@@ -9,6 +9,7 @@
 
 from .file_finder import is_toplevel_acceptable
 from .file_finder import scm_find_files
+from .utils import data_from_mime
 from .utils import do_ex
 from .utils import trace
 
@@ -101,3 +102,20 @@ def git_find_files(path: _t.PathT = "") -> list[str]:
         trace("toplevel mismatch", toplevel, fullpath)
     git_files, git_dirs = _git_ls_files_and_dirs(toplevel)
     return scm_find_files(path, git_files, git_dirs)
+
+
+def git_archive_find_files(path: _t.PathT = "") -> list[str]:
+    # This function assumes that ``path`` is obtained from a git archive
+    # and therefore all the files that should be ignored were already removed.
+    archival = os.path.join(path, ".git_archival.txt")
+    if not os.path.exists(archival):
+        return []
+
+    data = data_from_mime(archival)
+
+    if "$Format" in data.get("node", ""):
+        # Substitutions have not been performed, so not a reliable archive
+        return []
+
+    trace("git archive detected - fallback to listing all files")
+    return scm_find_files(path, set(), set(), force_all_files=True)

src/setuptools_scm/file_finder_hg.py

@@ -2,10 +2,16 @@
 
 import os
 import subprocess
+from typing import TYPE_CHECKING
 
 from .file_finder import is_toplevel_acceptable
 from .file_finder import scm_find_files
+from .utils import data_from_mime
 from .utils import do_ex
+from .utils import trace
+
+if TYPE_CHECKING:
+    from . import _types as _t
 
 
 def _hg_toplevel(path: str) -> str | None:
@@ -49,3 +55,20 @@ def hg_find_files(path: str = "") -> list[str]:
     assert toplevel is not None
     hg_files, hg_dirs = _hg_ls_files_and_dirs(toplevel)
     return scm_find_files(path, hg_files, hg_dirs)
+
+
+def hg_archive_find_files(path: _t.PathT = "") -> list[str]:
+    # This function assumes that ``path`` is obtained from a mercurial archive
+    # and therefore all the files that should be ignored were already removed.
+    archival = os.path.join(path, ".hg_archival.txt")
+    if not os.path.exists(archival):
+        return []
+
+    data = data_from_mime(archival)
+
+    if "node" not in data:
+        # Ensure file is valid
+        return []
+
+    trace("hg archive detected - fallback to listing all files")
+    return scm_find_files(path, set(), set(), force_all_files=True)

src/setuptools_scm/integration.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import itertools
 import os
 import warnings
 from typing import Any
@@ -91,7 +92,10 @@ def version_keyword(
 
 
 def find_files(path: _t.PathT = "") -> list[str]:
-    for ep in iter_entry_points("setuptools_scm.files_command"):
+    for ep in itertools.chain(
+        iter_entry_points("setuptools_scm.files_command"),
+        iter_entry_points("setuptools_scm.files_command_fallback"),
+    ):
         command = ep.load()
         if isinstance(command, str):
             # this technique is deprecated

testing/test_file_finder.py

@@ -201,3 +201,33 @@ def test_symlink_not_in_scm_while_target_is(inwd: WorkDir) -> None:
 @pytest.mark.skip_commit
 def test_not_commited(inwd: WorkDir) -> None:
     assert find_files() == []
+
+
+def test_unexpanded_git_archival(wd: WorkDir, monkeypatch: pytest.MonkeyPatch) -> None:
+    # When substitutions in `.git_archival.txt` are not expanded, files should
+    # not be automatically listed.
+    monkeypatch.chdir(wd.cwd)
+    (wd.cwd / ".git_archival.txt").write_text("node: $Format:%H$", encoding="utf-8")
+    (wd.cwd / "file1.txt").touch()
+    assert find_files() == []
+
+
+@pytest.mark.parametrize("archive_file", (".git_archival.txt", ".hg_archival.txt"))
+def test_archive(
+    wd: WorkDir, monkeypatch: pytest.MonkeyPatch, archive_file: str
+) -> None:
+    # When substitutions in `.git_archival.txt` are not expanded, files should
+    # not be automatically listed.
+    monkeypatch.chdir(wd.cwd)
+    sha = "a1bda3d984d1a40d7b00ae1d0869354d6d503001"
+    (wd.cwd / archive_file).write_text(f"node: {sha}", encoding="utf-8")
+    (wd.cwd / "data").mkdir()
+    (wd.cwd / "data" / "datafile").touch()
+
+    datalink = wd.cwd / "data" / "datalink"
+    if sys.platform != "win32":
+        datalink.symlink_to("data/datafile")
+    else:
+        os.link("data/datafile", datalink)
+
+    assert set(find_files()) == _sep({archive_file, "data/datafile", "data/datalink"})