docagent-cli 0.0.35__py3-none-any.whl

docagent_cli/file_ops.py

@@ -0,0 +1,473 @@
+"""Helpers for tracking file operations and computing diffs for CLI display."""
+
+from __future__ import annotations
+
+import difflib
+import logging
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import TYPE_CHECKING, Any, Literal
+
+logger = logging.getLogger(__name__)
+
+if TYPE_CHECKING:
+    from deepagents.backends.protocol import BackendProtocol
+
+FileOpStatus = Literal["pending", "success", "error"]
+
+
+@dataclass
+class ApprovalPreview:
+    """Data used to render HITL previews."""
+
+    title: str
+    details: list[str]
+    diff: str | None = None
+    diff_title: str | None = None
+    error: str | None = None
+
+
+def _safe_read(path: Path) -> str | None:
+    """Read file content, returning None on failure.
+
+    Returns:
+        File content as string, or None if reading fails.
+    """
+    try:
+        return path.read_text(encoding="utf-8")
+    except (OSError, UnicodeDecodeError) as e:
+        logger.debug("Failed to read file %s: %s", path, e)
+        return None
+
+
+def _count_lines(text: str) -> int:
+    """Count lines in text, treating empty strings as zero lines.
+
+    Returns:
+        Number of lines in the text.
+    """
+    if not text:
+        return 0
+    return len(text.splitlines())
+
+
+def compute_unified_diff(
+    before: str,
+    after: str,
+    display_path: str,
+    *,
+    max_lines: int | None = 800,
+    context_lines: int = 3,
+) -> str | None:
+    """Compute a unified diff between before and after content.
+
+    Args:
+        before: Original content
+        after: New content
+        display_path: Path for display in diff headers
+        max_lines: Maximum number of diff lines (None for unlimited)
+        context_lines: Number of context lines around changes (default 3)
+
+    Returns:
+        Unified diff string or None if no changes
+    """
+    before_lines = before.splitlines()
+    after_lines = after.splitlines()
+    diff_lines = list(
+        difflib.unified_diff(
+            before_lines,
+            after_lines,
+            fromfile=f"{display_path} (before)",
+            tofile=f"{display_path} (after)",
+            lineterm="",
+            n=context_lines,
+        )
+    )
+    if not diff_lines:
+        return None
+    if max_lines is not None and len(diff_lines) > max_lines:
+        truncated = diff_lines[: max_lines - 1]
+        truncated.append("...")
+        return "\n".join(truncated)
+    return "\n".join(diff_lines)
+
+
+@dataclass
+class FileOpMetrics:
+    """Line and byte level metrics for a file operation."""
+
+    lines_read: int = 0
+    start_line: int | None = None
+    end_line: int | None = None
+    lines_written: int = 0
+    lines_added: int = 0
+    lines_removed: int = 0
+    bytes_written: int = 0
+
+
+@dataclass
+class FileOperationRecord:
+    """Track a single filesystem tool call."""
+
+    tool_name: str
+    display_path: str
+    physical_path: Path | None
+    tool_call_id: str | None
+    args: dict[str, Any] = field(default_factory=dict)
+    status: FileOpStatus = "pending"
+    error: str | None = None
+    metrics: FileOpMetrics = field(default_factory=FileOpMetrics)
+    diff: str | None = None
+    before_content: str | None = None
+    after_content: str | None = None
+    read_output: str | None = None
+    hitl_approved: bool = False
+
+
+def resolve_physical_path(
+    path_str: str | None, assistant_id: str | None
+) -> Path | None:
+    """Convert a virtual/relative path to a physical filesystem path.
+
+    Returns:
+        Resolved physical Path, or None if path is empty or resolution fails.
+    """
+    if not path_str:
+        return None
+    try:
+        if assistant_id and path_str.startswith("/memories/"):
+            from docagent_cli.config import settings
+
+            agent_dir = settings.get_agent_dir(assistant_id)
+            suffix = path_str.removeprefix("/memories/").lstrip("/")
+            return (agent_dir / suffix).resolve()
+        path = Path(path_str)
+        if path.is_absolute():
+            return path
+        return (Path.cwd() / path).resolve()
+    except (OSError, ValueError):
+        return None
+
+
+def format_display_path(path_str: str | None) -> str:
+    """Format a path for display.
+
+    Returns:
+        Formatted path string suitable for display.
+    """
+    if not path_str:
+        return "(unknown)"
+    try:
+        path = Path(path_str)
+        if path.is_absolute():
+            return path.name or str(path)
+        return str(path)
+    except (OSError, ValueError):
+        return str(path_str)
+
+
+def build_approval_preview(
+    tool_name: str,
+    args: dict[str, Any],
+    assistant_id: str | None,
+) -> ApprovalPreview | None:
+    """Collect summary info and diff for HITL approvals.
+
+    Returns:
+        ApprovalPreview with diff and details, or None if tool not supported.
+    """
+    path_str = str(args.get("file_path") or args.get("path") or "")
+    display_path = format_display_path(path_str)
+    physical_path = resolve_physical_path(path_str, assistant_id)
+
+    if tool_name == "write_file":
+        content = str(args.get("content", ""))
+        before = (
+            _safe_read(physical_path)
+            if physical_path and physical_path.exists()
+            else ""
+        )
+        after = content
+        diff = compute_unified_diff(before or "", after, display_path, max_lines=100)
+        additions = 0
+        if diff:
+            additions = sum(
+                1
+                for line in diff.splitlines()
+                if line.startswith("+") and not line.startswith("+++")
+            )
+        total_lines = _count_lines(after)
+        details = [
+            f"File: {path_str}",
+            "Action: Create new file"
+            + (" (overwrites existing content)" if before else ""),
+            f"Lines to write: {additions or total_lines}",
+        ]
+        return ApprovalPreview(
+            title=f"Write {display_path}",
+            details=details,
+            diff=diff,
+            diff_title=f"Diff {display_path}",
+        )
+
+    if tool_name == "edit_file":
+        if physical_path is None:
+            return ApprovalPreview(
+                title=f"Update {display_path}",
+                details=[f"File: {path_str}", "Action: Replace text"],
+                error="Unable to resolve file path.",
+            )
+        before = _safe_read(physical_path)
+        if before is None:
+            return ApprovalPreview(
+                title=f"Update {display_path}",
+                details=[f"File: {path_str}", "Action: Replace text"],
+                error="Unable to read current file contents.",
+            )
+        old_string = str(args.get("old_string", ""))
+        new_string = str(args.get("new_string", ""))
+        replace_all = bool(args.get("replace_all"))
+        from deepagents.backends.utils import perform_string_replacement
+
+        replacement = perform_string_replacement(
+            before, old_string, new_string, replace_all
+        )
+        if isinstance(replacement, str):
+            return ApprovalPreview(
+                title=f"Update {display_path}",
+                details=[f"File: {path_str}", "Action: Replace text"],
+                error=replacement,
+            )
+        after, occurrences = replacement
+        diff = compute_unified_diff(before, after, display_path, max_lines=None)
+        additions = 0
+        deletions = 0
+        if diff:
+            additions = sum(
+                1
+                for line in diff.splitlines()
+                if line.startswith("+") and not line.startswith("+++")
+            )
+            deletions = sum(
+                1
+                for line in diff.splitlines()
+                if line.startswith("-") and not line.startswith("---")
+            )
+        action = "all occurrences" if replace_all else "single occurrence"
+        details = [
+            f"File: {path_str}",
+            f"Action: Replace text ({action})",
+            f"Occurrences matched: {occurrences}",
+            f"Lines changed: +{additions} / -{deletions}",
+        ]
+        return ApprovalPreview(
+            title=f"Update {display_path}",
+            details=details,
+            diff=diff,
+            diff_title=f"Diff {display_path}",
+        )
+
+    return None
+
+
+class FileOpTracker:
+    """Collect file operation metrics during a CLI interaction."""
+
+    def __init__(
+        self, *, assistant_id: str | None, backend: BackendProtocol | None = None
+    ) -> None:
+        """Initialize the tracker."""
+        self.assistant_id = assistant_id
+        self.backend = backend
+        self.active: dict[str | None, FileOperationRecord] = {}
+        self.completed: list[FileOperationRecord] = []
+
+    def start_operation(
+        self, tool_name: str, args: dict[str, Any], tool_call_id: str | None
+    ) -> None:
+        """Begin tracking a file operation.
+
+        Creates a record for the operation and, for write/edit operations,
+        captures the file's content before modification.
+        """
+        if tool_name not in {"read_file", "write_file", "edit_file"}:
+            return
+        path_str = str(args.get("file_path") or args.get("path") or "")
+        display_path = format_display_path(path_str)
+        record = FileOperationRecord(
+            tool_name=tool_name,
+            display_path=display_path,
+            physical_path=resolve_physical_path(path_str, self.assistant_id),
+            tool_call_id=tool_call_id,
+            args=args,
+        )
+        if tool_name in {"write_file", "edit_file"}:
+            if self.backend and path_str:
+                try:
+                    responses = self.backend.download_files([path_str])
+                    if (
+                        responses
+                        and responses[0].content is not None
+                        and responses[0].error is None
+                    ):
+                        record.before_content = responses[0].content.decode("utf-8")
+                    else:
+                        record.before_content = ""
+                except (OSError, UnicodeDecodeError, AttributeError) as e:
+                    logger.debug(
+                        "Failed to read before_content for %s: %s", path_str, e
+                    )
+                    record.before_content = ""
+            elif record.physical_path:
+                record.before_content = _safe_read(record.physical_path) or ""
+        self.active[tool_call_id] = record
+
+    def complete_with_message(self, tool_message: Any) -> FileOperationRecord | None:  # noqa: ANN401  # Tool message type is dynamic
+        """Complete a file operation with the tool message result.
+
+        Returns:
+            The completed FileOperationRecord, or None if no matching operation.
+        """
+        tool_call_id = getattr(tool_message, "tool_call_id", None)
+        record = self.active.get(tool_call_id)
+        if record is None:
+            return None
+
+        content = tool_message.content
+        if isinstance(content, list):
+            # Some tool messages may return list segments; join them for analysis.
+            joined = []
+            for item in content:
+                if isinstance(item, str):
+                    joined.append(item)
+                else:
+                    joined.append(str(item))
+            content_text = "\n".join(joined)
+        else:
+            content_text = str(content) if content is not None else ""
+
+        if getattr(
+            tool_message, "status", "success"
+        ) != "success" or content_text.lower().startswith("error"):
+            record.status = "error"
+            record.error = content_text
+            self._finalize(record)
+            return record
+
+        record.status = "success"
+
+        if record.tool_name == "read_file":
+            record.read_output = content_text
+            lines = _count_lines(content_text)
+            record.metrics.lines_read = lines
+            offset = record.args.get("offset")
+            limit = record.args.get("limit")
+            if isinstance(offset, int):
+                if offset > lines:
+                    offset = 0
+                record.metrics.start_line = offset + 1
+                if lines:
+                    record.metrics.end_line = offset + lines
+            elif lines:
+                record.metrics.start_line = 1
+                record.metrics.end_line = lines
+            if isinstance(limit, int) and lines > limit:
+                record.metrics.end_line = (record.metrics.start_line or 1) + limit - 1
+        else:
+            # For write/edit operations, read back from backend (or local filesystem)
+            self._populate_after_content(record)
+            if record.after_content is None:
+                record.status = "error"
+                record.error = "Could not read updated file content."
+                self._finalize(record)
+                return record
+            record.metrics.lines_written = _count_lines(record.after_content)
+            before_lines = _count_lines(record.before_content or "")
+            diff = compute_unified_diff(
+                record.before_content or "",
+                record.after_content,
+                record.display_path,
+                max_lines=100,
+            )
+            record.diff = diff
+            if diff:
+                additions = sum(
+                    1
+                    for line in diff.splitlines()
+                    if line.startswith("+") and not line.startswith("+++")
+                )
+                deletions = sum(
+                    1
+                    for line in diff.splitlines()
+                    if line.startswith("-") and not line.startswith("---")
+                )
+                record.metrics.lines_added = additions
+                record.metrics.lines_removed = deletions
+            elif record.tool_name == "write_file" and not (record.before_content or ""):
+                record.metrics.lines_added = record.metrics.lines_written
+            record.metrics.bytes_written = len(record.after_content.encode("utf-8"))
+            if (
+                record.diff is None
+                and (record.before_content or "") != record.after_content
+            ):
+                record.diff = compute_unified_diff(
+                    record.before_content or "",
+                    record.after_content,
+                    record.display_path,
+                    max_lines=100,
+                )
+            if record.diff is None and before_lines != record.metrics.lines_written:
+                record.metrics.lines_added = max(
+                    record.metrics.lines_written - before_lines, 0
+                )
+
+        self._finalize(record)
+        return record
+
+    def mark_hitl_approved(self, tool_name: str, args: dict[str, Any]) -> None:
+        """Mark operations matching tool_name and file_path as HIL-approved."""
+        file_path = args.get("file_path") or args.get("path")
+        if not file_path:
+            return
+
+        # Mark all active records that match
+        for record in self.active.values():
+            if record.tool_name == tool_name:
+                record_path = record.args.get("file_path") or record.args.get("path")
+                if record_path == file_path:
+                    record.hitl_approved = True
+
+    def _populate_after_content(self, record: FileOperationRecord) -> None:
+        # Use backend if available (works for any BackendProtocol implementation)
+        if self.backend:
+            try:
+                file_path = record.args.get("file_path") or record.args.get("path")
+                if file_path:
+                    responses = self.backend.download_files([file_path])
+                    if (
+                        responses
+                        and responses[0].content is not None
+                        and responses[0].error is None
+                    ):
+                        record.after_content = responses[0].content.decode("utf-8")
+                    else:
+                        record.after_content = None
+                else:
+                    record.after_content = None
+            except (OSError, UnicodeDecodeError, AttributeError) as e:
+                logger.debug(
+                    "Failed to read after_content for %s: %s",
+                    record.args.get("file_path") or record.args.get("path"),
+                    e,
+                )
+                record.after_content = None
+        else:
+            # Fallback: direct filesystem read when no backend provided
+            if record.physical_path is None:
+                record.after_content = None
+                return
+            record.after_content = _safe_read(record.physical_path)
+
+    def _finalize(self, record: FileOperationRecord) -> None:
+        self.completed.append(record)
+        self.active.pop(record.tool_call_id, None)

docagent_cli/formatting.py

@@ -0,0 +1,28 @@
+"""Lightweight text-formatting helpers.
+
+Keep this module free of heavy dependencies so it can be imported anywhere
+in the CLI without pulling in large frameworks.
+"""
+
+from __future__ import annotations
+
+
+def format_duration(seconds: float) -> str:
+    """Format a duration in seconds into a human-readable string.
+
+    Args:
+        seconds: Duration in seconds.
+
+    Returns:
+        Formatted string like `"5s"`, `"2.3s"`, `"5m 12s"`, or `"1h 23m 4s"`.
+    """
+    rounded = round(seconds, 1)
+    if rounded < 60:  # noqa: PLR2004
+        if rounded % 1 == 0:
+            return f"{int(rounded)}s"
+        return f"{rounded:.1f}s"
+    minutes, secs = divmod(int(rounded), 60)
+    if minutes < 60:  # noqa: PLR2004
+        return f"{minutes}m {secs}s"
+    hours, minutes = divmod(minutes, 60)
+    return f"{hours}h {minutes}m {secs}s"

docagent_cli/hooks.py

@@ -0,0 +1,206 @@
+"""Lightweight hook dispatch for external tool integration.
+
+Loads hook configuration from `~/.docagent/hooks.json` and fires matching
+commands with JSON payloads on stdin. Subprocess work is offloaded to a
+background thread so the caller's event loop is never stalled. Failures are
+logged but never bubble up to the caller.
+
+Config format (`~/.docagent/hooks.json`):
+
+```json
+{"hooks": [{"command": ["bash", "adapter.sh"], "events": ["session.start"]}]}
+```
+
+If `events` is omitted or empty the hook receives **all** events.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import json
+import logging
+import subprocess  # noqa: S404
+from concurrent.futures import ThreadPoolExecutor
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+_hooks_config: list[dict[str, Any]] | None = None
+"""Cached config — loaded lazily on first dispatch."""
+
+_background_tasks: set[asyncio.Task[None]] = set()
+"""Strong references to fire-and-forget tasks to prevent GC."""
+
+
+def _load_hooks() -> list[dict[str, Any]]:
+    """Load and cache hook definitions from the config file.
+
+    Returns:
+        An empty list when the file is missing or malformed so that normal
+            execution is never interrupted.
+    """
+    global _hooks_config  # noqa: PLW0603
+    if _hooks_config is not None:
+        return _hooks_config
+
+    from docagent_cli.model_config import DEFAULT_CONFIG_DIR
+
+    hooks_path = DEFAULT_CONFIG_DIR / "hooks.json"
+
+    if not hooks_path.is_file():
+        _hooks_config = []
+        return _hooks_config
+
+    try:
+        data = json.loads(hooks_path.read_text())
+        if not isinstance(data, dict):
+            logger.warning(
+                "Hooks config at %s must be a JSON object, got %s",
+                hooks_path,
+                type(data).__name__,
+            )
+            _hooks_config = []
+            return _hooks_config
+        hooks = data.get("hooks", [])
+        if not isinstance(hooks, list):
+            logger.warning(
+                "Hooks config 'hooks' key at %s must be a list, got %s",
+                hooks_path,
+                type(hooks).__name__,
+            )
+            _hooks_config = []
+            return _hooks_config
+        _hooks_config = hooks
+    except (json.JSONDecodeError, OSError) as exc:
+        logger.warning("Failed to load hooks config from %s: %s", hooks_path, exc)
+        _hooks_config = []
+
+    return _hooks_config
+
+
+def _run_single_hook(command: list[str], event: str, payload_bytes: bytes) -> None:
+    """Execute a single hook command, writing the JSON payload to its stdin.
+
+    Uses `subprocess.run` which automatically kills the child process on
+    timeout, preventing zombie/orphan process leaks.
+
+    Args:
+        command: The command and arguments to run.
+        event: Event name (for logging).
+        payload_bytes: JSON payload to write to the command's stdin.
+    """
+    try:
+        subprocess.run(  # noqa: S603
+            command,
+            input=payload_bytes,
+            stdout=subprocess.DEVNULL,
+            stderr=subprocess.DEVNULL,
+            start_new_session=True,
+            timeout=5,
+            check=False,
+        )
+    except subprocess.TimeoutExpired:
+        logger.warning("Hook command timed out (>5s) for event %s: %s", event, command)
+    except (FileNotFoundError, PermissionError) as exc:
+        logger.warning("Hook command failed for event %s: %s — %s", event, command, exc)
+    except Exception:
+        logger.debug(
+            "Hook dispatch failed for event %s: %s",
+            event,
+            command,
+            exc_info=True,
+        )
+
+
+def _dispatch_hook_sync(
+    event: str, payload_bytes: bytes, hooks: list[dict[str, Any]]
+) -> None:
+    """Dispatch matching hooks, running them concurrently via a thread pool.
+
+    Iterates over all configured hooks, skipping those whose event filter
+    does not match or whose `command` is missing/invalid. Matching hooks are
+    executed concurrently with a 5-second timeout per command. Errors are caught
+    per-hook and logged without propagating.
+
+    Args:
+        event: Dotted event name (e.g. `'session.start'`).
+        payload_bytes: JSON payload to write to each command's stdin.
+        hooks: List of hook definition dicts from the config file.
+    """
+    matching: list[list[str]] = []
+    for hook in hooks:
+        command = hook.get("command")
+        if not isinstance(command, list) or not command:
+            continue
+
+        events = hook.get("events")
+        # Empty/missing events list means "subscribe to everything".
+        if events and event not in events:
+            continue
+
+        matching.append(command)
+
+    if not matching:
+        return
+
+    if len(matching) == 1:
+        _run_single_hook(matching[0], event, payload_bytes)
+        return
+
+    with ThreadPoolExecutor(max_workers=len(matching)) as pool:
+        futures = [
+            pool.submit(_run_single_hook, cmd, event, payload_bytes) for cmd in matching
+        ]
+        for future in futures:
+            future.result()
+
+
+async def dispatch_hook(event: str, payload: dict[str, Any]) -> None:
+    """Fire matching hook commands with `payload` serialized as JSON on stdin.
+
+    The `event` name is automatically injected into the payload under the
+    `"event"` key so callers don't need to duplicate it.
+
+    The blocking subprocess work is offloaded to a thread so the caller's
+    event loop is never stalled. Matching hooks run concurrently, each with
+    a 5-second timeout. Errors are logged and never propagated.
+
+    Args:
+        event: Dotted event name (e.g. `'session.start'`).
+        payload: Arbitrary JSON-serializable dict sent on the command's stdin.
+    """
+    try:
+        hooks = _load_hooks()
+        if not hooks:
+            return
+
+        payload_bytes = json.dumps({"event": event, **payload}).encode()
+        await asyncio.to_thread(_dispatch_hook_sync, event, payload_bytes, hooks)
+    except Exception:
+        logger.warning(
+            "Unexpected error in dispatch_hook for event %s",
+            event,
+            exc_info=True,
+        )
+
+
+def dispatch_hook_fire_and_forget(event: str, payload: dict[str, Any]) -> None:
+    """Schedule `dispatch_hook` as a background task with a strong reference.
+
+    Use this instead of bare `create_task(dispatch_hook(...))` to prevent the
+    task from being garbage collected before completion.
+
+    Safe to call from sync code as long as an event loop is running.
+
+    Args:
+        event: Dotted event name (e.g. `'session.start'`).
+        payload: Arbitrary JSON-serializable dict sent on the command's stdin.
+    """
+    try:
+        loop = asyncio.get_running_loop()
+    except RuntimeError:
+        logger.debug("No running event loop; skipping hook for %s", event)
+        return
+    task = loop.create_task(dispatch_hook(event, payload))
+    _background_tasks.add(task)
+    task.add_done_callback(_background_tasks.discard)