Only copy files not ignored by git for local build

sphinx_polyversion/driver.py

@@ -11,6 +11,7 @@
 from inspect import isawaitable
 from logging import getLogger
 from pathlib import Path
+from subprocess import PIPE, CalledProcessError
 from types import TracebackType
 from typing import (
     TYPE_CHECKING,
@@ -33,6 +34,7 @@
 
 from sphinx_polyversion.builder import Builder, BuildError
 from sphinx_polyversion.environment import Environment
+from sphinx_polyversion.git import get_unignored_files
 from sphinx_polyversion.json import GLOBAL_ENCODER, Encoder, JSONable
 from sphinx_polyversion.utils import shift_path
 
@@ -333,8 +335,32 @@ async def build_local(self) -> None:
         with tempfile.TemporaryDirectory() as tmp:
             path = Path(tmp)
             # copy source files
-            logger.info("Copying source files...")
-            shutil.copytree(self.root, path, symlinks=True, dirs_exist_ok=True)
+            logger.info("Copying source files (except for files ignored by git)...")
+            try:
+                async for file in get_unignored_files(self.root):
+                    source = self.root / file
+                    target = path / file
+                    target.parent.mkdir(parents=True, exist_ok=True)
+                    if source.exists() and not target.exists():
+                        shutil.copy2(source, target, follow_symlinks=False)
+
+                # as .git is not copied, we have to initialize a dummy git repository
+                # in the copied directory (for setuptools-scm)
+                git_init_cmd = ("git", "init", "--initial-branch=dummy")
+                with tempfile.SpooledTemporaryFile(max_size=1024) as f:
+                    process = await asyncio.create_subprocess_exec(
+                        *git_init_cmd, cwd=tmp, stdout=f, stderr=PIPE
+                    )
+                    out, err = await process.communicate()
+                    if process.returncode:
+                        raise CalledProcessError(
+                            process.returncode, " ".join(git_init_cmd), stderr=err
+                        )
+            except CalledProcessError:
+                logger.warning(
+                    "Could not list un-ignored files using git. Copying full working directory..."
+                )
+                shutil.copytree(self.root, path, symlinks=True, dirs_exist_ok=True)
             # setup build environment (e.g. poetry/pip venv)
             async with await self.init_environment(path, rev) as env:
                 # construct metadata to pass to the build process

sphinx_polyversion/git.py

@@ -202,6 +202,7 @@ async def _copy_tree(
     """
     # retrieve commit contents as tar archive
     cmd = ("git", "archive", "--format", "tar", ref)
+    git_init_cmd = ("git", "init", "--initial-branch=dummy")
     with tempfile.SpooledTemporaryFile(max_size=buffer_size) as f:
         process = await asyncio.create_subprocess_exec(
             *cmd, cwd=repo, stdout=f, stderr=PIPE
@@ -213,6 +214,15 @@ async def _copy_tree(
         f.seek(0)
         with tarfile.open(fileobj=f) as tf:
             tf.extractall(str(dest))
+        # initialize dummy git repository in copied directory (required for setuptools-scm)
+        process = await asyncio.create_subprocess_exec(
+            *git_init_cmd, cwd=str(dest), stdout=f, stderr=PIPE
+        )
+        out, err = await process.communicate()
+        if process.returncode:
+            raise CalledProcessError(
+                process.returncode, " ".join(git_init_cmd), stderr=err
+            )
 
 
 async def file_exists(repo: Path, ref: GitRef, file: PurePath) -> bool:
@@ -250,6 +260,34 @@ async def file_exists(repo: Path, ref: GitRef, file: PurePath) -> bool:
     return rc == 0
 
 
+async def get_unignored_files(directory: Path) -> AsyncGenerator[Path, None]:
+    """
+    List all unignored files in the directory.
+
+    Parameters
+    ----------
+    directory : Path
+        Any directory in the repo.
+
+    Returns
+    -------
+    AsyncGenerator[Path, None]
+        The paths to all un-ignored files in the directory (recursive)
+
+    """
+    cmd = (
+        "git",
+        "ls-files",
+        "--cached",
+        "--others",
+        "--exclude-standard",
+    )
+    process = await asyncio.create_subprocess_exec(*cmd, cwd=directory, stdout=PIPE)
+    out, err = await process.communicate()
+    for line in out.decode().splitlines():
+        yield Path(line.strip())
+
+
 # -- VersionProvider API -----------------------------------------------------