claude-mpm 4.21.3__py3-none-any.whl → 5.0.2__py3-none-any.whl

claude_mpm/services/agents/sources/__init__.py

@@ -0,0 +1,13 @@
+"""Git source sync services for agent templates.
+
+This module provides services for syncing agent templates from remote
+Git repositories, with ETag-based caching for efficient updates and
+SQLite-based state tracking.
+"""
+
+from claude_mpm.services.agents.sources.agent_sync_state import AgentSyncState
+from claude_mpm.services.agents.sources.git_source_sync_service import (
+    GitSourceSyncService,
+)
+
+__all__ = ["AgentSyncState", "GitSourceSyncService"]

claude_mpm/services/agents/sources/agent_sync_state.py

@@ -0,0 +1,516 @@
+"""SQLite-based state tracking for agent sync operations.
+
+This service manages sync state for agent files from Git sources, providing:
+- Per-file SHA-256 content hash tracking
+- Source metadata (URLs, last commit SHA, ETags)
+- Complete sync history audit trail
+- Change detection for efficient incremental updates
+
+Database Location: ~/.config/claude-mpm/agent_sync.db
+Thread Safety: Uses connection-per-operation pattern (safe for single-threaded use)
+Performance: Optimized with indexes; expected <10ms per operation
+"""
+
+import logging
+import sqlite3
+from contextlib import contextmanager
+from datetime import datetime, timedelta, timezone
+from pathlib import Path
+from typing import Any, Dict, List, Optional
+
+logger = logging.getLogger(__name__)
+
+
+class AgentSyncStateError(Exception):
+    """Base exception for sync state errors."""
+
+
+class DatabaseError(AgentSyncStateError):
+    """Database operation errors."""
+
+
+class AgentSyncState:
+    """Service for tracking agent sync state in SQLite database.
+
+    Responsibilities:
+    - Manage SQLite connection lifecycle
+    - Track per-file content hashes (SHA-256)
+    - Record sync history with timestamps
+    - Query file change status
+    - Provide migration utilities
+
+    Design Decision: Connection-per-operation pattern
+
+    Rationale: Simplifies connection management and prevents connection leaks.
+    SQLite allows concurrent reads but serializes writes, so connection pooling
+    provides minimal benefit for our read-heavy workload.
+
+    Trade-offs:
+    - Simplicity: No pool management overhead
+    - Safety: Automatic cleanup via context manager
+    - Performance: Negligible overhead (<1ms per connection)
+
+    Extension Points: Can add connection pooling if multi-threaded access needed.
+    """
+
+    # Schema version for migrations
+    SCHEMA_VERSION = 1
+
+    def __init__(self, db_path: Optional[Path] = None):
+        """Initialize sync state service.
+
+        Args:
+            db_path: Path to SQLite database (defaults to ~/.config/claude-mpm/agent_sync.db)
+        """
+        if db_path:
+            self.db_path = Path(db_path)
+        else:
+            # Default location: ~/.config/claude-mpm/agent_sync.db
+            config_dir = Path.home() / ".config" / "claude-mpm"
+            config_dir.mkdir(parents=True, exist_ok=True)
+            self.db_path = config_dir / "agent_sync.db"
+
+        # Initialize database
+        self._initialize_database()
+
+    @contextmanager
+    def _get_connection(self):
+        """Context manager for database connections.
+
+        Yields:
+            sqlite3.Connection with foreign keys enabled and row factory
+
+        Error Handling:
+        - Exception during transaction: Rolls back automatically
+        - Connection errors: Propagates to caller
+        - Cleanup: Always closes connection
+        """
+        conn = sqlite3.connect(str(self.db_path))
+        conn.row_factory = sqlite3.Row  # Enable dict-like access
+        conn.execute("PRAGMA foreign_keys = ON")  # Enable FK constraints
+        try:
+            yield conn
+            conn.commit()
+        except Exception:
+            conn.rollback()
+            raise
+        finally:
+            conn.close()
+
+    def _initialize_database(self):
+        """Initialize database schema if not exists.
+
+        Creates all tables, indexes, and metadata on first run.
+        Verifies schema version on subsequent runs.
+        """
+        with self._get_connection() as conn:
+            # Check if database exists and has schema
+            cursor = conn.execute(
+                "SELECT name FROM sqlite_master WHERE type='table' AND name='sources'"
+            )
+            if cursor.fetchone() is None:
+                # Database is new, create schema
+                self._create_schema(conn)
+                logger.info(f"Initialized sync state database: {self.db_path}")
+            else:
+                # Verify schema version
+                version = self._get_schema_version(conn)
+                if version != self.SCHEMA_VERSION:
+                    logger.warning(
+                        f"Schema version mismatch: expected {self.SCHEMA_VERSION}, found {version}"
+                    )
+                    # TODO: Implement migration in future ticket
+
+    def _create_schema(self, conn: sqlite3.Connection):
+        """Create database schema with all tables and indexes.
+
+        Schema Design:
+        - sources: Track Git repositories or file sources
+        - agent_files: Track individual files and content hashes
+        - sync_history: Audit trail of all sync operations
+        - schema_metadata: Schema versioning for migrations
+        """
+        conn.executescript(
+            """
+            -- Sources table: Track Git repositories or file sources
+            CREATE TABLE sources (
+                id TEXT PRIMARY KEY,                    -- Source identifier (e.g., "github-remote", "local-project")
+                url TEXT NOT NULL,                      -- Source URL or file path
+                last_sha TEXT,                          -- Last synced commit SHA (Git sources only)
+                last_sync_time TEXT,                    -- ISO 8601 timestamp of last sync
+                etag TEXT,                              -- HTTP ETag for GitHub raw URLs
+                enabled INTEGER DEFAULT 1,              -- 0=disabled, 1=enabled
+                created_at TEXT DEFAULT CURRENT_TIMESTAMP,
+                updated_at TEXT DEFAULT CURRENT_TIMESTAMP
+            );
+
+            -- Agent files table: Track individual files and content hashes
+            CREATE TABLE agent_files (
+                source_id TEXT NOT NULL,                -- FK to sources.id
+                file_path TEXT NOT NULL,                -- Relative path (e.g., "research.md")
+                content_sha TEXT NOT NULL,              -- SHA-256 hash of file content
+                local_path TEXT,                        -- Absolute path to cached file
+                synced_at TEXT NOT NULL,                -- ISO 8601 timestamp when file was synced
+                file_size INTEGER,                      -- File size in bytes
+                PRIMARY KEY (source_id, file_path),
+                FOREIGN KEY (source_id) REFERENCES sources(id) ON DELETE CASCADE
+            );
+
+            -- Sync history table: Audit trail of all sync operations
+            CREATE TABLE sync_history (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                source_id TEXT NOT NULL,                -- FK to sources.id
+                sync_time TEXT NOT NULL,                -- ISO 8601 timestamp
+                status TEXT NOT NULL,                   -- 'success', 'partial', 'error'
+                files_synced INTEGER DEFAULT 0,         -- Number of files downloaded
+                files_cached INTEGER DEFAULT 0,         -- Number of cache hits
+                files_failed INTEGER DEFAULT 0,         -- Number of failed downloads
+                error_message TEXT,                     -- Error details if status='error'
+                duration_ms INTEGER,                    -- Sync duration in milliseconds
+                FOREIGN KEY (source_id) REFERENCES sources(id) ON DELETE CASCADE
+            );
+
+            -- Performance indexes
+            CREATE INDEX idx_agent_files_source ON agent_files(source_id);
+            CREATE INDEX idx_agent_files_path ON agent_files(file_path);
+            CREATE INDEX idx_sync_history_source_time ON sync_history(source_id, sync_time DESC);
+            CREATE INDEX idx_sync_history_status ON sync_history(status);
+
+            -- Metadata table for schema versioning
+            CREATE TABLE schema_metadata (
+                key TEXT PRIMARY KEY,
+                value TEXT NOT NULL
+            );
+
+            INSERT INTO schema_metadata (key, value) VALUES ('version', '1');
+            INSERT INTO schema_metadata (key, value) VALUES ('created_at', datetime('now'));
+        """
+        )
+
+    def _get_schema_version(self, conn: sqlite3.Connection) -> int:
+        """Get current schema version.
+
+        Args:
+            conn: Database connection
+
+        Returns:
+            Schema version number (0 if schema_metadata table doesn't exist)
+        """
+        try:
+            cursor = conn.execute(
+                "SELECT value FROM schema_metadata WHERE key = 'version'"
+            )
+            row = cursor.fetchone()
+            return int(row[0]) if row else 0
+        except sqlite3.OperationalError:
+            return 0
+
+    # ==============================================================================
+    # SOURCE MANAGEMENT
+    # ==============================================================================
+
+    def register_source(self, source_id: str, url: str, enabled: bool = True) -> None:
+        """Register or update a sync source.
+
+        Args:
+            source_id: Unique source identifier (e.g., "github-remote")
+            url: Source URL or file path
+            enabled: Whether source is enabled for sync
+
+        Example:
+            sync_state.register_source(
+                source_id="github-remote",
+                url="https://raw.githubusercontent.com/bobmatnyc/claude-mpm-agents/main",
+                enabled=True
+            )
+        """
+        with self._get_connection() as conn:
+            conn.execute(
+                """
+                INSERT INTO sources (id, url, enabled, updated_at)
+                VALUES (?, ?, ?, ?)
+                ON CONFLICT(id) DO UPDATE SET
+                    url = excluded.url,
+                    enabled = excluded.enabled,
+                    updated_at = excluded.updated_at
+                """,
+                (source_id, url, int(enabled), datetime.now(timezone.utc).isoformat()),
+            )
+        logger.debug(f"Registered source: {source_id} -> {url}")
+
+    def update_source_sync_metadata(
+        self,
+        source_id: str,
+        last_sha: Optional[str] = None,
+        etag: Optional[str] = None,
+    ) -> None:
+        """Update source sync metadata (commit SHA, ETag).
+
+        Args:
+            source_id: Source identifier
+            last_sha: Latest commit SHA (Git sources)
+            etag: HTTP ETag (GitHub raw URLs)
+
+        Example:
+            sync_state.update_source_sync_metadata(
+                source_id="github-remote",
+                last_sha="abc123def456",
+                etag='"W/abc123"'
+            )
+        """
+        with self._get_connection() as conn:
+            conn.execute(
+                """
+                UPDATE sources
+                SET last_sha = ?, etag = ?, last_sync_time = ?, updated_at = ?
+                WHERE id = ?
+                """,
+                (
+                    last_sha,
+                    etag,
+                    datetime.now(timezone.utc).isoformat(),
+                    datetime.now(timezone.utc).isoformat(),
+                    source_id,
+                ),
+            )
+        logger.debug(f"Updated source metadata: {source_id}")
+
+    def get_source_info(self, source_id: str) -> Optional[Dict[str, Any]]:
+        """Get source metadata.
+
+        Args:
+            source_id: Source identifier
+
+        Returns:
+            Source metadata dict or None if not found
+
+        Example:
+            >>> info = sync_state.get_source_info("github-remote")
+            >>> print(info["url"])
+            https://raw.githubusercontent.com/...
+        """
+        with self._get_connection() as conn:
+            cursor = conn.execute("SELECT * FROM sources WHERE id = ?", (source_id,))
+            row = cursor.fetchone()
+            return dict(row) if row else None
+
+    def get_all_sources(self, enabled_only: bool = False) -> List[Dict[str, Any]]:
+        """Get all registered sources.
+
+        Args:
+            enabled_only: Only return enabled sources
+
+        Returns:
+            List of source metadata dicts
+        """
+        with self._get_connection() as conn:
+            query = "SELECT * FROM sources"
+            if enabled_only:
+                query += " WHERE enabled = 1"
+            cursor = conn.execute(query)
+            return [dict(row) for row in cursor.fetchall()]
+
+    # ==============================================================================
+    # FILE TRACKING
+    # ==============================================================================
+
+    def track_file(
+        self,
+        source_id: str,
+        file_path: str,
+        content_sha: str,
+        local_path: Optional[str] = None,
+        file_size: Optional[int] = None,
+    ) -> None:
+        """Track agent file with content hash.
+
+        Args:
+            source_id: Source identifier
+            file_path: Relative file path (e.g., "research.md")
+            content_sha: SHA-256 hash of file content
+            local_path: Absolute path to cached file
+            file_size: File size in bytes
+
+        Example:
+            sync_state.track_file(
+                source_id="github-remote",
+                file_path="research.md",
+                content_sha="abc123def456...",
+                local_path="/home/user/.claude-mpm/cache/research.md",
+                file_size=2048
+            )
+        """
+        with self._get_connection() as conn:
+            conn.execute(
+                """
+                INSERT INTO agent_files (source_id, file_path, content_sha, local_path, synced_at, file_size)
+                VALUES (?, ?, ?, ?, ?, ?)
+                ON CONFLICT(source_id, file_path) DO UPDATE SET
+                    content_sha = excluded.content_sha,
+                    local_path = excluded.local_path,
+                    synced_at = excluded.synced_at,
+                    file_size = excluded.file_size
+                """,
+                (
+                    source_id,
+                    file_path,
+                    content_sha,
+                    local_path,
+                    datetime.now(timezone.utc).isoformat(),
+                    file_size,
+                ),
+            )
+        logger.debug(f"Tracked file: {source_id}/{file_path} -> {content_sha[:8]}...")
+
+    def get_file_hash(self, source_id: str, file_path: str) -> Optional[str]:
+        """Get stored content hash for file.
+
+        Args:
+            source_id: Source identifier
+            file_path: Relative file path
+
+        Returns:
+            SHA-256 hash or None if not tracked
+        """
+        with self._get_connection() as conn:
+            cursor = conn.execute(
+                "SELECT content_sha FROM agent_files WHERE source_id = ? AND file_path = ?",
+                (source_id, file_path),
+            )
+            row = cursor.fetchone()
+            return row["content_sha"] if row else None
+
+    def has_file_changed(
+        self, source_id: str, file_path: str, current_sha: str
+    ) -> bool:
+        """Check if file content has changed.
+
+        Args:
+            source_id: Source identifier
+            file_path: Relative file path
+            current_sha: Current SHA-256 hash
+
+        Returns:
+            True if changed or not tracked, False if unchanged
+
+        Example:
+            >>> changed = sync_state.has_file_changed(
+            ...     "github-remote", "research.md", "abc123..."
+            ... )
+            >>> if changed:
+            ...     print("File needs update")
+        """
+        stored_sha = self.get_file_hash(source_id, file_path)
+        if stored_sha is None:
+            return True  # Not tracked = changed
+        return stored_sha != current_sha
+
+    # ==============================================================================
+    # SYNC HISTORY
+    # ==============================================================================
+
+    def record_sync_result(
+        self,
+        source_id: str,
+        status: str,
+        files_synced: int = 0,
+        files_cached: int = 0,
+        files_failed: int = 0,
+        error_message: Optional[str] = None,
+        duration_ms: Optional[int] = None,
+    ) -> int:
+        """Record sync operation result.
+
+        Args:
+            source_id: Source identifier
+            status: 'success', 'partial', or 'error'
+            files_synced: Number of files downloaded
+            files_cached: Number of cache hits
+            files_failed: Number of failed downloads
+            error_message: Error details if status='error'
+            duration_ms: Sync duration in milliseconds
+
+        Returns:
+            Sync history record ID
+
+        Example:
+            record_id = sync_state.record_sync_result(
+                source_id="github-remote",
+                status="success",
+                files_synced=5,
+                files_cached=3,
+                duration_ms=1500
+            )
+        """
+        with self._get_connection() as conn:
+            cursor = conn.execute(
+                """
+                INSERT INTO sync_history (
+                    source_id, sync_time, status, files_synced, files_cached,
+                    files_failed, error_message, duration_ms
+                )
+                VALUES (?, ?, ?, ?, ?, ?, ?, ?)
+                """,
+                (
+                    source_id,
+                    datetime.now(timezone.utc).isoformat(),
+                    status,
+                    files_synced,
+                    files_cached,
+                    files_failed,
+                    error_message,
+                    duration_ms,
+                ),
+            )
+            return cursor.lastrowid
+
+    def get_sync_history(self, source_id: str, limit: int = 10) -> List[Dict[str, Any]]:
+        """Get recent sync history for source.
+
+        Args:
+            source_id: Source identifier
+            limit: Maximum number of records
+
+        Returns:
+            List of sync history records (most recent first)
+
+        Example:
+            >>> history = sync_state.get_sync_history("github-remote", limit=5)
+            >>> for record in history:
+            ...     print(f"{record['sync_time']}: {record['status']}")
+        """
+        with self._get_connection() as conn:
+            cursor = conn.execute(
+                """
+                SELECT * FROM sync_history
+                WHERE source_id = ?
+                ORDER BY sync_time DESC
+                LIMIT ?
+                """,
+                (source_id, limit),
+            )
+            return [dict(row) for row in cursor.fetchall()]
+
+    def cleanup_old_history(self, days: int = 30) -> int:
+        """Remove sync history older than specified days.
+
+        Args:
+            days: Number of days to retain
+
+        Returns:
+            Number of records deleted
+
+        Example:
+            deleted = sync_state.cleanup_old_history(days=30)
+            print(f"Cleaned up {deleted} old sync records")
+        """
+        cutoff = (datetime.now(timezone.utc) - timedelta(days=days)).isoformat()
+
+        with self._get_connection() as conn:
+            cursor = conn.execute(
+                "DELETE FROM sync_history WHERE sync_time < ?", (cutoff,)
+            )
+            deleted = cursor.rowcount
+
+        logger.info(f"Cleaned up {deleted} sync history records older than {days} days")
+        return deleted