feat: Transferable (#852)

In #676 for
copying to/from `DockerContainer`, it was suggested that we should
clarify the interface & usage a bit before proceeding. This PR aims to
push that conversation forward with some test cases illustrating
proposed usages. Although here I created a `Transferable` object, in
most cases there's no need for the caller to explicitly construct a
Transferable, just pass the `bytes | Path`

**Proposed use cases:**

1. Copy into a container by passing a `TransferSpec` into the
initializer:
```
DockerContainer(... transferrables=[TransferSpec(source, destination_in_container), ...)
```

2. Copy into the container via the builder pattern, prior to starting
the container:
```
DockerContainer(...)
     .with_copy_into_container(b"some_bytes", destination_in_container)
     .with_copy_into_container(some_path, destination_in_container)
 ```

3. Copy into the container at runtime:
```
with DockerContainer(...) as container:
container.copy_into_container(b"some_bytes", destination_in_container)
    container.copy_into_container(some_path, destination_in_container)
```

---------

Co-authored-by: Roy Moore <roy@moore.co.il>

Author: rhoban13

core/README.rst

@@ -16,6 +16,8 @@ Testcontainers Core
 
 .. autoclass:: testcontainers.core.wait_strategies.WaitStrategy
 
+.. autoclass:: testcontainers.core.transferable.Transferable
+
 .. raw:: html
 
     <hr>

core/testcontainers/core/container.py

@@ -1,5 +1,8 @@
 import contextlib
+import io
+import pathlib
 import sys
+import tarfile
 from os import PathLike
 from socket import socket
 from types import TracebackType
@@ -18,6 +21,7 @@
 from testcontainers.core.exceptions import ContainerConnectException, ContainerStartException
 from testcontainers.core.labels import LABEL_SESSION_ID, SESSION_ID
 from testcontainers.core.network import Network
+from testcontainers.core.transferable import Transferable, TransferSpec
 from testcontainers.core.utils import is_arm, setup_logger
 from testcontainers.core.wait_strategies import LogMessageWaitStrategy
 from testcontainers.core.waiting_utils import WaitStrategy
@@ -69,6 +73,7 @@ def __init__(
         network: Optional[Network] = None,
         network_aliases: Optional[list[str]] = None,
         _wait_strategy: Optional[WaitStrategy] = None,
+        transferables: Optional[list[TransferSpec]] = None,
         **kwargs: Any,
     ) -> None:
         self.env = env or {}
@@ -100,6 +105,11 @@ def __init__(
         self._kwargs = kwargs
         self._wait_strategy: Optional[WaitStrategy] = _wait_strategy
 
+        self._transferable_specs: list[TransferSpec] = []
+        if transferables:
+            for t in transferables:
+                self.with_copy_into_container(*t)
+
     def with_env(self, key: str, value: str) -> Self:
         self.env[key] = value
         return self
@@ -208,6 +218,10 @@ def start(self) -> Self:
             self._wait_strategy.wait_until_ready(self)
 
         logger.info("Container started: %s", self._container.short_id)
+
+        for t in self._transferable_specs:
+            self._transfer_into_container(*t)
+
         return self
 
     def stop(self, force: bool = True, delete_volume: bool = True) -> None:
@@ -318,6 +332,68 @@ def _configure(self) -> None:
         # placeholder if subclasses want to define this and use the default start method
         pass
 
+    def with_copy_into_container(
+        self, transferable: Transferable, destination_in_container: str, mode: int = 0o644
+    ) -> Self:
+        self._transferable_specs.append((transferable, destination_in_container, mode))
+        return self
+
+    def copy_into_container(self, transferable: Transferable, destination_in_container: str, mode: int = 0o644) -> None:
+        return self._transfer_into_container(transferable, destination_in_container, mode)
+
+    def _transfer_into_container(self, transferable: Transferable, destination_in_container: str, mode: int) -> None:
+        if isinstance(transferable, bytes):
+            self._transfer_file_content_into_container(transferable, destination_in_container, mode)
+        elif isinstance(transferable, pathlib.Path):
+            if transferable.is_file():
+                self._transfer_file_content_into_container(transferable.read_bytes(), destination_in_container, mode)
+            elif transferable.is_dir():
+                self._transfer_directory_into_container(transferable, destination_in_container, mode)
+            else:
+                raise TypeError(f"Path {transferable} is neither a file nor directory")
+        else:
+            raise TypeError("source must be bytes or PathLike")
+
+    def _transfer_file_content_into_container(
+        self, file_content: bytes, destination_in_container: str, mode: int
+    ) -> None:
+        fileobj = io.BytesIO()
+        with tarfile.open(fileobj=fileobj, mode="w") as tar:
+            tarinfo = tarfile.TarInfo(name=destination_in_container)
+            tarinfo.size = len(file_content)
+            tarinfo.mode = mode
+            tar.addfile(tarinfo, io.BytesIO(file_content))
+        fileobj.seek(0)
+        assert self._container is not None
+        rv = self._container.put_archive(path="/", data=fileobj.getvalue())
+        assert rv is True
+
+    def _transfer_directory_into_container(
+        self, source_directory: pathlib.Path, destination_in_container: str, mode: int
+    ) -> None:
+        assert self._container is not None
+        result = self._container.exec_run(["mkdir", "-p", destination_in_container])
+        assert result.exit_code == 0
+
+        fileobj = io.BytesIO()
+        with tarfile.open(fileobj=fileobj, mode="w") as tar:
+            tar.add(source_directory, arcname=source_directory.name)
+        fileobj.seek(0)
+        rv = self._container.put_archive(path=destination_in_container, data=fileobj.getvalue())
+        assert rv is True
+
+    def copy_from_container(self, source_in_container: str, destination_on_host: pathlib.Path) -> None:
+        assert self._container is not None
+        tar_stream, _ = self._container.get_archive(source_in_container)
+
+        for chunk in tar_stream:
+            with tarfile.open(fileobj=io.BytesIO(chunk)) as tar:
+                for member in tar.getmembers():
+                    with open(destination_on_host, "wb") as f:
+                        fileobj = tar.extractfile(member)
+                        assert fileobj is not None
+                        f.write(fileobj.read())
+
 
 class Reaper:
     """

core/testcontainers/core/transferable.py

@@ -0,0 +1,6 @@
+import pathlib
+from typing import Union
+
+Transferable = Union[bytes, pathlib.Path]
+
+TransferSpec = Union[tuple[Transferable, str], tuple[Transferable, str, int]]

core/tests/test_core.py

@@ -1,7 +1,9 @@
 import tempfile
 from pathlib import Path
 
+import pytest
 from testcontainers.core.container import DockerContainer
+from testcontainers.core.transferable import Transferable, TransferSpec
 
 
 def test_garbage_collection_is_defensive():
@@ -46,3 +48,109 @@ def test_docker_container_with_env_file():
             assert "ADMIN_EMAIL=admin@example.org" in output
             assert "ROOT_URL=example.org/app" in output
             print(output)
+
+
+@pytest.fixture(name="transferable", params=(bytes, Path))
+def copy_sources_fixture(request, tmp_path: Path):
+    """
+    Provide source argument for tests of copy_into_container
+    """
+    raw_data = b"hello world"
+    if request.param is bytes:
+        return raw_data
+    elif request.param is Path:
+        my_file = tmp_path / "my_file"
+        my_file.write_bytes(raw_data)
+        return my_file
+    pytest.fail("Invalid type")
+
+
+def test_copy_into_container_at_runtime(transferable: Transferable):
+    # Given
+    destination_in_container = "/tmp/my_file"
+
+    with DockerContainer("bash", command="sleep infinity") as container:
+        # When
+        container.copy_into_container(transferable, destination_in_container)
+        result = container.exec(f"cat {destination_in_container}")
+
+    # Then
+    assert result.exit_code == 0
+    assert result.output == b"hello world"
+
+
+def test_copy_into_container_at_startup(transferable: Transferable):
+    # Given
+    destination_in_container = "/tmp/my_file"
+
+    container = DockerContainer("bash", command="sleep infinity")
+    container.with_copy_into_container(transferable, destination_in_container)
+
+    with container:
+        # When
+        result = container.exec(f"cat {destination_in_container}")
+
+    # Then
+    assert result.exit_code == 0
+    assert result.output == b"hello world"
+
+
+def test_copy_into_container_via_initializer(transferable: Transferable):
+    # Given
+    destination_in_container = "/tmp/my_file"
+    transferables: list[TransferSpec] = [(transferable, destination_in_container, 0o644)]
+
+    with DockerContainer("bash", command="sleep infinity", transferables=transferables) as container:
+        # When
+        result = container.exec(f"cat {destination_in_container}")
+
+    # Then
+    assert result.exit_code == 0
+    assert result.output == b"hello world"
+
+
+def test_copy_file_from_container(tmp_path: Path):
+    # Given
+    file_in_container = "/tmp/foo.txt"
+    destination_on_host = tmp_path / "foo.txt"
+    assert not destination_on_host.is_file()
+
+    with DockerContainer("bash", command="sleep infinity") as container:
+        result = container.exec(f'bash -c "echo -n hello world > {file_in_container}"')
+        assert result.exit_code == 0
+
+        # When
+        container.copy_from_container(file_in_container, destination_on_host)
+
+    # Then
+    assert destination_on_host.is_file()
+    assert destination_on_host.read_text() == "hello world"
+
+
+def test_copy_directory_into_container(tmp_path: Path):
+    # Given
+    source_dir = tmp_path / "my_directory"
+    source_dir.mkdir()
+    my_file = source_dir / "my_file"
+    my_file.write_bytes(b"hello world")
+
+    destination_in_container = "/tmp/my_destination_directory"
+
+    with DockerContainer("bash", command="sleep infinity") as container:
+        # When
+        container.copy_into_container(source_dir, destination_in_container)
+        result = container.exec(f"ls {destination_in_container}")
+
+        # Then - my_directory exists
+        assert result.exit_code == 0
+        assert result.output == b"my_directory\n"
+
+        # Then - my_file is in directory
+        result = container.exec(f"ls {destination_in_container}/my_directory")
+        assert result.exit_code == 0
+        assert result.output == b"my_file\n"
+
+        # Then - my_file contents are correct
+        result = container.exec(f"cat {destination_in_container}/my_directory/my_file")
+        assert result.exit_code == 0
+        assert result.output == b"hello world"