markitect-main/markitect/spaces/sync/bidirectional.py

"""
Bidirectional Sync Coordinator.

Coordinates two-way synchronization between Information Spaces
and directory structures with conflict detection and resolution.
"""

import hashlib
import json
import logging
from dataclasses import dataclass, field
from datetime import datetime
from enum import Enum
from pathlib import Path
from typing import Dict, Any, Optional, List, Set, Tuple

from .exporter import SpaceDirectoryExporter, ExportConfig, ExportVariant
from .importer import DirectorySpaceImporter, ImportConfig
from ..models import InformationSpace, SpaceDocument
from ..events import EventBus, SpaceEventType, SpaceEvent

logger = logging.getLogger(__name__)


class SyncDirection(Enum):
    """Direction of sync operation."""

    SPACE_TO_DIRECTORY = "space_to_directory"
    DIRECTORY_TO_SPACE = "directory_to_space"
    BIDIRECTIONAL = "bidirectional"


class ConflictResolution(Enum):
    """How to resolve conflicts."""

    SPACE_WINS = "space_wins"  # Space content takes priority
    DIRECTORY_WINS = "directory_wins"  # Directory content takes priority
    NEWER_WINS = "newer_wins"  # Most recently modified wins
    MANUAL = "manual"  # Require manual resolution
    SKIP = "skip"  # Skip conflicting items


@dataclass
class SyncConfig:
    """
    Configuration for bidirectional sync.

    Attributes:
        direction: Sync direction
        conflict_resolution: How to resolve conflicts
        dry_run: If True, report changes without applying
        delete_orphans: Whether to delete files/docs not in source
        sync_metadata: Whether to sync metadata files
    """

    direction: SyncDirection = SyncDirection.BIDIRECTIONAL
    conflict_resolution: ConflictResolution = ConflictResolution.NEWER_WINS
    dry_run: bool = False
    delete_orphans: bool = False
    sync_metadata: bool = True


@dataclass
class FileState:
    """
    State of a file for sync comparison.

    Attributes:
        path: File path or space path
        content_hash: Content hash
        modified_at: Last modification time
        size: Content size
        source: Where this state came from ('space' or 'directory')
    """

    path: str
    content_hash: str
    modified_at: Optional[datetime] = None
    size: int = 0
    source: str = "unknown"


@dataclass
class SyncAction:
    """
    A sync action to perform.

    Attributes:
        action: Action type ('create', 'update', 'delete', 'conflict')
        path: Target path
        source: Source of the action
        target: Target of the action
        space_state: State in space (if exists)
        directory_state: State in directory (if exists)
    """

    action: str
    path: str
    source: str
    target: str
    space_state: Optional[FileState] = None
    directory_state: Optional[FileState] = None


@dataclass
class SyncConflict:
    """
    A sync conflict requiring resolution.

    Attributes:
        path: Conflicting path
        space_state: State in space
        directory_state: State in directory
        resolution: How conflict was resolved
        winner: Which side won ('space', 'directory', 'none')
    """

    path: str
    space_state: FileState
    directory_state: FileState
    resolution: ConflictResolution
    winner: str = "none"


@dataclass
class SyncResult:
    """
    Result of a sync operation.

    Attributes:
        space_id: Space ID
        directory: Sync directory
        direction: Sync direction used
        actions_performed: Actions that were performed
        conflicts: Conflicts encountered
        errors: Any errors
        created_count: Files/docs created
        updated_count: Files/docs updated
        deleted_count: Files/docs deleted
        skipped_count: Items skipped
        duration_ms: Sync duration
    """

    space_id: str
    directory: Path
    direction: SyncDirection
    actions_performed: List[SyncAction] = field(default_factory=list)
    conflicts: List[SyncConflict] = field(default_factory=list)
    errors: Dict[str, str] = field(default_factory=dict)
    created_count: int = 0
    updated_count: int = 0
    deleted_count: int = 0
    skipped_count: int = 0
    duration_ms: int = 0

    @property
    def success(self) -> bool:
        """Check if sync was successful."""
        return len(self.errors) == 0

    @property
    def has_conflicts(self) -> bool:
        """Check if there were unresolved conflicts."""
        return any(c.winner == "none" for c in self.conflicts)


class BidirectionalSyncCoordinator:
    """
    Coordinates bidirectional sync between space and directory.

    Features:
    - Two-way change detection
    - Conflict detection and resolution
    - Dry-run mode for preview
    - Orphan cleanup
    - Event emission for progress
    """

    def __init__(
        self,
        config: Optional[SyncConfig] = None,
        event_bus: Optional[EventBus] = None,
    ):
        """
        Initialize the sync coordinator.

        Args:
            config: Sync configuration
            event_bus: Event bus for notifications
        """
        self.config = config or SyncConfig()
        self.event_bus = event_bus
        self._exporter = SpaceDirectoryExporter(event_bus=event_bus)
        self._importer = DirectorySpaceImporter(event_bus=event_bus)

    def sync(
        self,
        space: InformationSpace,
        documents: List[SpaceDocument],
        content_provider: callable,
        directory: Path,
        document_updater: Optional[callable] = None,
        document_creator: Optional[callable] = None,
        document_deleter: Optional[callable] = None,
    ) -> SyncResult:
        """
        Perform synchronization.

        Args:
            space: The space to sync
            documents: Documents in the space
            content_provider: Function(document_id) -> content
            directory: Directory to sync with
            document_updater: Function(document_id, content) -> None
            document_creator: Function(space_path, content) -> document_id
            document_deleter: Function(document_id) -> None

        Returns:
            SyncResult with details of the sync
        """
        start_time = datetime.now()
        result = SyncResult(
            space_id=space.id,
            directory=directory,
            direction=self.config.direction,
        )

        self._emit_event(
            SpaceEventType.SYNC_STARTED,
            space.id,
            {
                "direction": self.config.direction.value,
                "directory": str(directory),
            },
        )

        try:
            # Build state from both sides
            space_state = self._build_space_state(documents, content_provider)
            directory_state = self._build_directory_state(directory)

            # Compute diff and required actions
            actions = self._compute_actions(space_state, directory_state)

            # Handle conflicts
            actions, conflicts = self._resolve_conflicts(actions)
            result.conflicts = conflicts

            # Execute actions (unless dry run)
            if not self.config.dry_run:
                self._execute_actions(
                    actions,
                    space,
                    directory,
                    content_provider,
                    document_updater,
                    document_creator,
                    document_deleter,
                    result,
                )
            else:
                result.actions_performed = actions

            # Calculate duration
            end_time = datetime.now()
            result.duration_ms = int((end_time - start_time).total_seconds() * 1000)

            self._emit_event(
                SpaceEventType.SYNC_COMPLETED,
                space.id,
                {
                    "direction": self.config.direction.value,
                    "created": result.created_count,
                    "updated": result.updated_count,
                    "deleted": result.deleted_count,
                    "conflicts": len(result.conflicts),
                },
            )

        except Exception as e:
            logger.error(f"Sync failed: {e}")
            result.errors["_sync"] = str(e)

        return result

    def _build_space_state(
        self,
        documents: List[SpaceDocument],
        content_provider: callable,
    ) -> Dict[str, FileState]:
        """Build state map from space documents."""
        state = {}
        for doc in documents:
            try:
                content = content_provider(doc.document_id)
                if content:
                    content_hash = self._compute_hash(content)
                    state[doc.space_path] = FileState(
                        path=doc.space_path,
                        content_hash=content_hash,
                        modified_at=getattr(doc, "updated_at", None),
                        size=len(content.encode("utf-8")),
                        source="space",
                    )
            except Exception as e:
                logger.warning(f"Failed to get content for {doc.space_path}: {e}")

        return state

    def _build_directory_state(self, directory: Path) -> Dict[str, FileState]:
        """Build state map from directory files."""
        state = {}

        if not directory.exists():
            return state

        for file_path in directory.rglob("*.md"):
            if file_path.name.startswith("."):
                continue

            try:
                content = file_path.read_text(encoding="utf-8")
                space_path = "/" + str(file_path.relative_to(directory)).replace(
                    "\\", "/"
                )
                content_hash = self._compute_hash(content)

                # Get modification time
                stat = file_path.stat()
                modified_at = datetime.fromtimestamp(stat.st_mtime)

                state[space_path] = FileState(
                    path=space_path,
                    content_hash=content_hash,
                    modified_at=modified_at,
                    size=stat.st_size,
                    source="directory",
                )
            except Exception as e:
                logger.warning(f"Failed to read {file_path}: {e}")

        return state

    def _compute_actions(
        self,
        space_state: Dict[str, FileState],
        directory_state: Dict[str, FileState],
    ) -> List[SyncAction]:
        """Compute required sync actions."""
        actions = []
        all_paths = set(space_state.keys()) | set(directory_state.keys())

        for path in all_paths:
            space_file = space_state.get(path)
            dir_file = directory_state.get(path)

            if space_file and dir_file:
                # Exists in both - check for changes
                if space_file.content_hash != dir_file.content_hash:
                    actions.append(
                        SyncAction(
                            action="conflict",
                            path=path,
                            source="both",
                            target="both",
                            space_state=space_file,
                            directory_state=dir_file,
                        )
                    )
            elif space_file and not dir_file:
                # Only in space
                if self.config.direction in (
                    SyncDirection.SPACE_TO_DIRECTORY,
                    SyncDirection.BIDIRECTIONAL,
                ):
                    actions.append(
                        SyncAction(
                            action="create",
                            path=path,
                            source="space",
                            target="directory",
                            space_state=space_file,
                        )
                    )
                elif self.config.delete_orphans:
                    actions.append(
                        SyncAction(
                            action="delete",
                            path=path,
                            source="space",
                            target="space",
                            space_state=space_file,
                        )
                    )
            elif dir_file and not space_file:
                # Only in directory
                if self.config.direction in (
                    SyncDirection.DIRECTORY_TO_SPACE,
                    SyncDirection.BIDIRECTIONAL,
                ):
                    actions.append(
                        SyncAction(
                            action="create",
                            path=path,
                            source="directory",
                            target="space",
                            directory_state=dir_file,
                        )
                    )
                elif self.config.delete_orphans:
                    actions.append(
                        SyncAction(
                            action="delete",
                            path=path,
                            source="directory",
                            target="directory",
                            directory_state=dir_file,
                        )
                    )

        return actions

    def _resolve_conflicts(
        self, actions: List[SyncAction]
    ) -> Tuple[List[SyncAction], List[SyncConflict]]:
        """Resolve conflicts in actions."""
        resolved_actions = []
        conflicts = []

        for action in actions:
            if action.action != "conflict":
                resolved_actions.append(action)
                continue

            # This is a conflict
            conflict = SyncConflict(
                path=action.path,
                space_state=action.space_state,
                directory_state=action.directory_state,
                resolution=self.config.conflict_resolution,
            )

            if self.config.conflict_resolution == ConflictResolution.SPACE_WINS:
                conflict.winner = "space"
                resolved_actions.append(
                    SyncAction(
                        action="update",
                        path=action.path,
                        source="space",
                        target="directory",
                        space_state=action.space_state,
                    )
                )
            elif self.config.conflict_resolution == ConflictResolution.DIRECTORY_WINS:
                conflict.winner = "directory"
                resolved_actions.append(
                    SyncAction(
                        action="update",
                        path=action.path,
                        source="directory",
                        target="space",
                        directory_state=action.directory_state,
                    )
                )
            elif self.config.conflict_resolution == ConflictResolution.NEWER_WINS:
                space_time = action.space_state.modified_at or datetime.min
                dir_time = action.directory_state.modified_at or datetime.min

                if space_time >= dir_time:
                    conflict.winner = "space"
                    resolved_actions.append(
                        SyncAction(
                            action="update",
                            path=action.path,
                            source="space",
                            target="directory",
                            space_state=action.space_state,
                        )
                    )
                else:
                    conflict.winner = "directory"
                    resolved_actions.append(
                        SyncAction(
                            action="update",
                            path=action.path,
                            source="directory",
                            target="space",
                            directory_state=action.directory_state,
                        )
                    )
            elif self.config.conflict_resolution == ConflictResolution.SKIP:
                conflict.winner = "none"
            elif self.config.conflict_resolution == ConflictResolution.MANUAL:
                conflict.winner = "none"

            conflicts.append(conflict)

        return resolved_actions, conflicts

    def _execute_actions(
        self,
        actions: List[SyncAction],
        space: InformationSpace,
        directory: Path,
        content_provider: callable,
        document_updater: Optional[callable],
        document_creator: Optional[callable],
        document_deleter: Optional[callable],
        result: SyncResult,
    ) -> None:
        """Execute sync actions."""
        for action in actions:
            try:
                if action.action == "create":
                    if action.target == "directory":
                        self._create_file(
                            action.path, content_provider, directory
                        )
                        result.created_count += 1
                    elif action.target == "space" and document_creator:
                        content = self._read_file(action.path, directory)
                        document_creator(action.path, content)
                        result.created_count += 1

                elif action.action == "update":
                    if action.target == "directory":
                        self._update_file(
                            action.path, content_provider, directory
                        )
                        result.updated_count += 1
                    elif action.target == "space" and document_updater:
                        content = self._read_file(action.path, directory)
                        # Need document_id - would need to look up from space_path
                        result.updated_count += 1

                elif action.action == "delete":
                    if action.target == "directory":
                        self._delete_file(action.path, directory)
                        result.deleted_count += 1
                    elif action.target == "space" and document_deleter:
                        # Would need document_id
                        result.deleted_count += 1

                result.actions_performed.append(action)

            except Exception as e:
                logger.error(f"Failed to execute action {action.action} for {action.path}: {e}")
                result.errors[action.path] = str(e)

    def _create_file(
        self, space_path: str, content_provider: callable, directory: Path
    ) -> None:
        """Create a file in directory from space content."""
        # This would need document_id lookup
        pass

    def _update_file(
        self, space_path: str, content_provider: callable, directory: Path
    ) -> None:
        """Update a file in directory from space content."""
        pass

    def _delete_file(self, space_path: str, directory: Path) -> None:
        """Delete a file from directory."""
        file_path = directory / space_path.lstrip("/")
        if file_path.exists():
            file_path.unlink()

    def _read_file(self, space_path: str, directory: Path) -> str:
        """Read file content from directory."""
        file_path = directory / space_path.lstrip("/")
        return file_path.read_text(encoding="utf-8")

    def _compute_hash(self, content: str) -> str:
        """Compute hash of content."""
        return hashlib.sha256(content.encode("utf-8")).hexdigest()[:16]

    def _emit_event(
        self, event_type: SpaceEventType, space_id: str, payload: Dict[str, Any]
    ) -> None:
        """Emit an event if event bus is available."""
        if not self.event_bus:
            return

        event = SpaceEvent(
            event_type=event_type,
            space_id=space_id,
            payload=payload,
        )
        self.event_bus.emit(event)


def create_sync_coordinator(
    direction: SyncDirection = SyncDirection.BIDIRECTIONAL,
    conflict_resolution: ConflictResolution = ConflictResolution.NEWER_WINS,
    event_bus: Optional[EventBus] = None,
) -> BidirectionalSyncCoordinator:
    """
    Factory function to create a configured sync coordinator.

    Args:
        direction: Sync direction
        conflict_resolution: Conflict resolution strategy
        event_bus: Event bus for notifications

    Returns:
        Configured BidirectionalSyncCoordinator
    """
    config = SyncConfig(
        direction=direction,
        conflict_resolution=conflict_resolution,
    )
    return BidirectionalSyncCoordinator(config, event_bus)