Only copy files not ignored by git for local build

sphinx_polyversion/driver.py

@@ -334,7 +334,7 @@ async def build_local(self) -> None:
             path = Path(tmp)
             # copy source files
             logger.info("Copying source files...")
-            shutil.copytree(self.root, path, symlinks=True, dirs_exist_ok=True)
+            await self.vcs.checkout_local(self.root, path)
             # 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

@@ -5,6 +5,7 @@
 import asyncio
 import enum
 import re
+import shutil
 import tarfile
 import tempfile
 from asyncio.subprocess import PIPE
@@ -210,6 +211,7 @@ async def _copy_tree(
 
     """
     # retrieve commit contents as tar archive
+    # NOTE: doesn't support git submodules
     cmd = ("git", "archive", "--format", "tar", ref)
     with tempfile.SpooledTemporaryFile(max_size=buffer_size) as f:
         process = await asyncio.create_subprocess_exec(
@@ -259,6 +261,39 @@ 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.
+
+    This uses git to retrieve all tracked and untracked files but excludes
+    files ignored e.g. by `.gitignore`.
+
+    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)
+    while line := await process.stdout.readline():  # type: ignore[union-attr]
+        yield Path(line.strip().decode())
+    await process.wait()
+    if process.returncode:
+        raise CalledProcessError(process.returncode, " ".join(cmd))
+
+
 # -- VersionProvider API -----------------------------------------------------
 
 
@@ -401,6 +436,10 @@ class Git(VersionProvider[GitRef]):
     """
     Provide versions from git repository.
 
+    .. warning::
+
+        Currently git submodules aren't supported. Feel free to open an issue!
+
     Parameters
     ----------
     branch_regex : str | re.Pattern
@@ -509,6 +548,32 @@ async def checkout(self, root: Path, dest: Path, revision: GitRef) -> None:
         """
         await _copy_tree(root, revision.obj, dest, self.buffer_size)
 
+    async def checkout_local(self, root: Path, dest: Path) -> None:
+        """
+        Create copy of the local working directory at the given path.
+
+        Parameters
+        ----------
+        root : Path
+            The root path of the project.
+        dest : Path
+            The destination to extract the revision to.
+
+        """
+        try:
+            # NOTE: doesn't support git submodules
+            async for file in _get_unignored_files(root):
+                source = root / file
+                target = dest / file
+                target.parent.mkdir(parents=True, exist_ok=True)
+                assert source.exists()
+                shutil.copy2(source, target, follow_symlinks=False)
+        except CalledProcessError:
+            logger.warning(
+                "Could not list un-ignored files using git. Copying full working directory..."
+            )
+            shutil.copytree(root, dest, symlinks=True, dirs_exist_ok=True)
+
     async def predicate(self, root: Path, ref: GitRef) -> bool:
         """
         Check whether a revision should be build.

sphinx_polyversion/vcs.py

@@ -48,6 +48,20 @@ async def checkout(self, root: Path, dest: Path, revision: RT) -> None:
 
         """
 
+    @abstractmethod
+    async def checkout_local(self, root: Path, dest: Path) -> None:
+        """
+        Create copy of the local working directory at the given path.
+
+        Parameters
+        ----------
+        root : Path
+            The root path of the project.
+        dest : Path
+            The destination to extract the revision to.
+
+        """
+
     @abstractmethod
     async def retrieve(self, root: Path) -> Iterable[RT]:
         """