feat: Use docker api to pull files from container

docker_agent_sandbox/sandbox.py

@@ -172,6 +172,42 @@ class DockerSandbox:
     # File I/O
     # ------------------------------------------------------------------
 
+    def _resolve_path(self, path: str) -> Path:
+        """Resolve *path* against working_dir if relative."""
+        p = Path(path)
+        if not p.is_absolute() and self._working_dir is not None:
+            p = Path(self._working_dir) / p
+        return p
+
+    def read_file(self, path: str) -> bytes:
+        """
+        Read the file at *path* from the container using ``get_archive``.
+
+        Returns the raw file bytes. Raises ``FileNotFoundError`` if the path
+        does not exist and ``RuntimeError`` if the container is not running.
+        """
+        if self._container is None:
+            raise RuntimeError("Sandbox container is not running.")
+
+        p = self._resolve_path(path)
+
+        try:
+            stream, _ = self._container.get_archive(str(p))
+        except docker.errors.NotFound:
+            raise FileNotFoundError(f"No such file in container: {path!r}") from None
+
+        buf = io.BytesIO()
+        for chunk in stream:
+            buf.write(chunk)
+        buf.seek(0)
+
+        with tarfile.open(fileobj=buf) as tar:
+            member = tar.getmembers()[0]
+            f = tar.extractfile(member)
+            if f is None:
+                raise IsADirectoryError(f"{path!r} is a directory, not a file")
+            return f.read()
+
     def write_file(self, path: str, content: str) -> None:
         """
         Write *content* to *path* inside the container using ``put_archive``.
@@ -182,9 +218,7 @@ class DockerSandbox:
         if self._container is None:
             raise RuntimeError("Sandbox container is not running.")
 
-        p = Path(path)
-        if not p.is_absolute() and self._working_dir is not None:
-            p = Path(self._working_dir) / p
+        p = self._resolve_path(path)
         encoded = content.encode("utf-8")
 
         self._container.exec_run(["mkdir", "-p", str(p.parent)])

docker_agent_sandbox/tools/edit_file.py

@@ -2,7 +2,6 @@
 
 from __future__ import annotations
 
-from shlex import quote
 from typing import TYPE_CHECKING
 
 from langchain_core.tools import BaseTool, tool
@@ -36,11 +35,21 @@ def make_edit_file_tool(sandbox: "DockerSandbox") -> BaseTool:
 
         Returns a confirmation with the number of lines affected, or an error.
         """
-        logger.debug("Editing file inside sandbox: {!r}", path)
-        exit_code, content = sandbox.exec(f"cat -- {quote(path)}")
-        if exit_code != 0:
-            return f"[ERROR reading {path!r} for edit] {content.strip()}"
+        _MAX_EDIT_BYTES = 1_000_000  # 1 MB
 
+        logger.debug("Editing file inside sandbox: {!r}", path)
+        try:
+            data = sandbox.read_file(path)
+        except (FileNotFoundError, IsADirectoryError, RuntimeError) as exc:
+            return f"[ERROR reading {path!r} for edit] {exc}"
+
+        if len(data) > _MAX_EDIT_BYTES:
+            return (
+                f"[ERROR] {path!r} is {len(data)} bytes; edit_file only supports files "
+                f"up to {_MAX_EDIT_BYTES} bytes."
+            )
+
+        content = data.decode("utf-8", errors="replace")
         count = content.count(old_str)
         if count == 0:
             return (

docker_agent_sandbox/tools/read_file.py

@@ -2,7 +2,6 @@
 
 from __future__ import annotations
 
-from shlex import quote
 from typing import TYPE_CHECKING
 
 from langchain_core.tools import BaseTool, tool
@@ -25,36 +24,29 @@ def make_read_file_tool(sandbox: "DockerSandbox") -> BaseTool:
 
         *offset* is the number of bytes to skip from the start of the file.
         *length* is the maximum number of bytes to return.  If the file is
-        longer than ``offset + length``, the
-        output is trimmed and a summary line is appended showing how many
-        bytes were omitted.
+        longer than ``offset + length``, the output is trimmed and a summary
+        line is appended showing how many bytes were omitted.
 
         Returns the (possibly trimmed) file contents as text, or an error message.
         """
         logger.debug(
             "Reading file inside sandbox: {} offset={} length={}", path, offset, length
         )
-        exit_code, wc_out = sandbox.exec(f"wc -c -- {quote(path)}")
-        if exit_code != 0:
-            return f"[ERROR reading {path!r}] {wc_out.strip()}"
         try:
-            total = int(wc_out.split()[0])
-        except (ValueError, IndexError):
-            return f"[ERROR parsing file size for {path!r}] {wc_out.strip()}"
+            data = sandbox.read_file(path)
+        except (FileNotFoundError, IsADirectoryError, RuntimeError) as exc:
+            return f"[ERROR reading {path!r}] {exc}"
 
-        exit_code, chunk = sandbox.exec(
-            f"dd if={quote(path)} iflag=skip_bytes,count_bytes"
-            f" skip={offset} count={length} 2>/dev/null"
-        )
-        if exit_code != 0:
-            return f"[ERROR reading {path!r}] {chunk.strip()}"
+        total = len(data)
+        chunk = data[offset : offset + length]
+        text = chunk.decode("utf-8", errors="replace")
 
         suffix = ""
         if offset + length < total:
             remaining = total - (offset + length)
             suffix = f"\n[... {remaining} more bytes not shown (total {total} bytes). Use offset/length to read further.]"
         elif offset > 0 or total > length:
-            suffix = f"\n[File total: {total} bytes, showing {len(chunk)} chars from offset {offset}.]"
-        return chunk + suffix
+            suffix = f"\n[File total: {total} bytes, showing {len(chunk)} bytes from offset {offset}.]"
+        return text + suffix
 
     return read_file