claude-self-reflect 3.2.4 → 3.3.0

package/mcp-server/src/temporal_utils.py

@@ -0,0 +1,384 @@
+"""
+Temporal utilities for Claude Self Reflect.
+Handles session detection, time parsing, and temporal query helpers.
+"""
+
+import re
+from datetime import datetime, timedelta, timezone
+from typing import List, Dict, Any, Optional, Tuple, Union
+from functools import lru_cache
+import logging
+from dataclasses import dataclass
+from collections import defaultdict
+
+logger = logging.getLogger(__name__)
+
+@dataclass
+class WorkSession:
+    """Represents a work session - a group of related conversations."""
+    session_id: str
+    start_time: datetime
+    end_time: datetime
+    conversation_ids: List[str]
+    project: str
+    duration_minutes: int
+    main_topics: List[str]
+    files_touched: List[str]
+    message_count: int
+    
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for JSON serialization."""
+        return {
+            'session_id': self.session_id,
+            'start_time': self.start_time.isoformat(),
+            'end_time': self.end_time.isoformat(),
+            'conversation_ids': self.conversation_ids,
+            'project': self.project,
+            'duration_minutes': self.duration_minutes,
+            'main_topics': self.main_topics,
+            'files_touched': self.files_touched,
+            'message_count': self.message_count
+        }
+
+
+class SessionDetector:
+    """Detects and groups conversations into work sessions."""
+    
+    def __init__(self, 
+                 time_gap_minutes: int = 30,
+                 min_session_chunks: int = 1,
+                 merge_similar_topics: bool = True):
+        """
+        Initialize session detector.
+        
+        Args:
+            time_gap_minutes: Minutes of inactivity to split sessions
+            min_session_chunks: Minimum chunks to constitute a session
+            merge_similar_topics: Whether to merge adjacent similar topics
+        """
+        self.time_gap = timedelta(minutes=time_gap_minutes)
+        self.min_chunks = min_session_chunks
+        self.merge_similar = merge_similar_topics
+    
+    def detect_sessions(self, chunks: List[Dict[str, Any]]) -> List[WorkSession]:
+        """
+        Group conversation chunks into work sessions.
+        
+        Args:
+            chunks: List of conversation chunks with metadata
+            
+        Returns:
+            List of WorkSession objects
+        """
+        if not chunks:
+            return []
+        
+        # Sort by timestamp
+        sorted_chunks = sorted(chunks, key=lambda x: self._parse_timestamp(x.get('timestamp')))
+        
+        sessions = []
+        current_session_chunks = []
+        
+        for chunk in sorted_chunks:
+            chunk_time = self._parse_timestamp(chunk.get('timestamp'))
+            
+            if not current_session_chunks:
+                current_session_chunks.append(chunk)
+                continue
+            
+            last_time = self._parse_timestamp(current_session_chunks[-1].get('timestamp'))
+            time_gap = chunk_time - last_time
+            
+            # Check if we should start a new session
+            if time_gap > self.time_gap or chunk.get('project') != current_session_chunks[-1].get('project'):
+                # Finalize current session
+                if len(current_session_chunks) >= self.min_chunks:
+                    session = self._create_session(current_session_chunks)
+                    if session:
+                        sessions.append(session)
+                current_session_chunks = [chunk]
+            else:
+                current_session_chunks.append(chunk)
+        
+        # Don't forget the last session
+        if len(current_session_chunks) >= self.min_chunks:
+            session = self._create_session(current_session_chunks)
+            if session:
+                sessions.append(session)
+        
+        return sessions
+    
+    def _create_session(self, chunks: List[Dict[str, Any]]) -> Optional[WorkSession]:
+        """Create a WorkSession from chunks."""
+        if not chunks:
+            return None
+        
+        start_time = self._parse_timestamp(chunks[0].get('timestamp'))
+        end_time = self._parse_timestamp(chunks[-1].get('timestamp'))
+        
+        # Aggregate metadata
+        conversation_ids = list(set(c.get('conversation_id') for c in chunks if c.get('conversation_id')))
+        files = []
+        topics = []
+        message_count = 0
+        
+        for chunk in chunks:
+            if chunk.get('files_analyzed'):
+                files.extend(chunk['files_analyzed'])
+            if chunk.get('concepts'):
+                topics.extend(chunk['concepts'])
+            message_count += chunk.get('message_count', 1)
+        
+        # Deduplicate and limit
+        files = list(set(files))[:20]
+        
+        # Get most common topics
+        topic_counts = defaultdict(int)
+        for topic in topics:
+            topic_counts[topic] += 1
+        main_topics = sorted(topic_counts.keys(), key=lambda x: topic_counts[x], reverse=True)[:10]
+        
+        duration_minutes = int((end_time - start_time).total_seconds() / 60)
+        
+        # Generate session ID from start time and project
+        project = chunks[0].get('project', 'unknown')
+        session_id = f"{project}_{start_time.strftime('%Y%m%d_%H%M%S')}"
+        
+        return WorkSession(
+            session_id=session_id,
+            start_time=start_time,
+            end_time=end_time,
+            conversation_ids=conversation_ids,
+            project=project,
+            duration_minutes=duration_minutes,
+            main_topics=main_topics,
+            files_touched=files,
+            message_count=message_count
+        )
+    
+    def _parse_timestamp(self, timestamp_str: str) -> datetime:
+        """Parse timestamp string to datetime."""
+        if not timestamp_str:
+            return datetime.now(timezone.utc)
+        
+        # Handle ISO format with Z suffix
+        if timestamp_str.endswith('Z'):
+            timestamp_str = timestamp_str[:-1] + '+00:00'
+        
+        try:
+            dt = datetime.fromisoformat(timestamp_str)
+            if dt.tzinfo is None:
+                dt = dt.replace(tzinfo=timezone.utc)
+            return dt
+        except (ValueError, AttributeError):
+            logger.warning(f"Failed to parse timestamp: {timestamp_str}")
+            return datetime.now(timezone.utc)
+
+
+class TemporalParser:
+    """Parses natural language time expressions."""
+    
+    def __init__(self):
+        """Initialize the temporal parser."""
+        self.relative_patterns = {
+            'today': (0, 0),
+            'yesterday': (-1, -1),
+            'tomorrow': (1, 1),
+            'this week': (-7, 0),
+            'last week': (-14, -7),
+            'this month': (-30, 0),
+            'last month': (-60, -30),
+            'past week': (-7, 0),
+            'past month': (-30, 0),
+            'past year': (-365, 0),
+        }
+    
+    def parse_time_expression(self, 
+                             expr: str, 
+                             base_time: Optional[datetime] = None) -> Tuple[datetime, datetime]:
+        """
+        Parse natural language time expression into datetime range.
+        
+        Args:
+            expr: Natural language time expression
+            base_time: Base time for relative calculations (default: now)
+            
+        Returns:
+            Tuple of (start_datetime, end_datetime)
+        """
+        if not base_time:
+            base_time = datetime.now(timezone.utc)
+        
+        expr_lower = expr.lower().strip()
+        
+        # Check for ISO timestamp
+        if self._looks_like_iso(expr):
+            try:
+                dt = datetime.fromisoformat(expr.replace('Z', '+00:00'))
+                if dt.tzinfo is None:
+                    dt = dt.replace(tzinfo=timezone.utc)
+                return (dt, dt)
+            except ValueError:
+                pass
+        
+        # Check for relative patterns
+        for pattern, (start_days, end_days) in self.relative_patterns.items():
+            if pattern in expr_lower:
+                start = base_time + timedelta(days=start_days)
+                end = base_time + timedelta(days=end_days) if end_days != 0 else base_time
+                
+                # Adjust to day boundaries
+                start = start.replace(hour=0, minute=0, second=0, microsecond=0)
+                if end_days == 0:
+                    end = base_time
+                else:
+                    end = end.replace(hour=23, minute=59, second=59, microsecond=999999)
+                
+                return (start, end)
+        
+        # Check for "N days/hours/minutes ago"
+        ago_match = re.match(r'(\d+)\s*(day|hour|minute)s?\s*ago', expr_lower)
+        if ago_match:
+            amount = int(ago_match.group(1))
+            unit = ago_match.group(2)
+            
+            if unit == 'day':
+                delta = timedelta(days=amount)
+            elif unit == 'hour':
+                delta = timedelta(hours=amount)
+            else:  # minute
+                delta = timedelta(minutes=amount)
+            
+            target_time = base_time - delta
+            return (target_time, target_time)
+        
+        # Check for "last N days/hours"
+        last_match = re.match(r'(?:last|past)\s*(\d+)\s*(day|hour|minute)s?', expr_lower)
+        if last_match:
+            amount = int(last_match.group(1))
+            unit = last_match.group(2)
+            
+            if unit == 'day':
+                delta = timedelta(days=amount)
+            elif unit == 'hour':
+                delta = timedelta(hours=amount)
+            else:  # minute
+                delta = timedelta(minutes=amount)
+            
+            start = base_time - delta
+            return (start, base_time)
+        
+        # Check for "since X"
+        if expr_lower.startswith('since '):
+            remaining = expr[6:].strip()
+            start, _ = self.parse_time_expression(remaining, base_time)
+            return (start, base_time)
+        
+        # Default: treat as "today"
+        logger.warning(f"Could not parse time expression '{expr}', defaulting to today")
+        start = base_time.replace(hour=0, minute=0, second=0, microsecond=0)
+        return (start, base_time)
+    
+    @lru_cache(maxsize=128)
+    def _looks_like_iso(self, expr: str) -> bool:
+        """Check if string looks like ISO timestamp."""
+        iso_pattern = r'^\d{4}-\d{2}-\d{2}'
+        return bool(re.match(iso_pattern, expr))
+    
+    def format_relative_time(self, timestamp: Union[str, datetime]) -> str:
+        """
+        Format timestamp as relative time string.
+        
+        Args:
+            timestamp: Timestamp to format
+            
+        Returns:
+            Relative time string like "2 hours ago", "yesterday"
+        """
+        if isinstance(timestamp, str):
+            if timestamp.endswith('Z'):
+                timestamp = timestamp[:-1] + '+00:00'
+            try:
+                dt = datetime.fromisoformat(timestamp)
+            except ValueError:
+                return timestamp
+        else:
+            dt = timestamp
+        
+        if dt.tzinfo is None:
+            dt = dt.replace(tzinfo=timezone.utc)
+        
+        now = datetime.now(timezone.utc)
+        delta = now - dt
+        
+        # Format based on time difference
+        if delta.total_seconds() < 60:
+            return "just now"
+        elif delta.total_seconds() < 3600:
+            minutes = int(delta.total_seconds() / 60)
+            return f"{minutes} minute{'s' if minutes != 1 else ''} ago"
+        elif delta.total_seconds() < 86400:
+            hours = int(delta.total_seconds() / 3600)
+            return f"{hours} hour{'s' if hours != 1 else ''} ago"
+        elif delta.days == 1:
+            return "yesterday"
+        elif delta.days < 7:
+            return f"{delta.days} days ago"
+        elif delta.days < 30:
+            weeks = delta.days // 7
+            return f"{weeks} week{'s' if weeks != 1 else ''} ago"
+        elif delta.days < 365:
+            months = delta.days // 30
+            return f"{months} month{'s' if months != 1 else ''} ago"
+        else:
+            years = delta.days // 365
+            return f"{years} year{'s' if years != 1 else ''} ago"
+
+
+def group_by_time_period(chunks: List[Dict[str, Any]], 
+                         granularity: str = 'day') -> Dict[str, List[Dict[str, Any]]]:
+    """
+    Group chunks by time period.
+    
+    Args:
+        chunks: List of conversation chunks
+        granularity: 'hour', 'day', 'week', or 'month'
+        
+    Returns:
+        Dictionary mapping time period keys to chunks
+    """
+    grouped = defaultdict(list)
+    
+    for chunk in chunks:
+        timestamp_str = chunk.get('timestamp')
+        if not timestamp_str:
+            continue
+        
+        if timestamp_str.endswith('Z'):
+            timestamp_str = timestamp_str[:-1] + '+00:00'
+        
+        try:
+            dt = datetime.fromisoformat(timestamp_str)
+        except ValueError:
+            continue
+        
+        if dt.tzinfo is None:
+            dt = dt.replace(tzinfo=timezone.utc)
+        
+        # Generate period key based on granularity
+        if granularity == 'hour':
+            key = dt.strftime('%Y-%m-%d %H:00')
+        elif granularity == 'day':
+            key = dt.strftime('%Y-%m-%d')
+        elif granularity == 'week':
+            # Get start of week
+            week_start = dt - timedelta(days=dt.weekday())
+            key = week_start.strftime('%Y-W%V')
+        elif granularity == 'month':
+            key = dt.strftime('%Y-%m')
+        else:
+            key = dt.strftime('%Y-%m-%d')
+        
+        grouped[key].append(chunk)
+    
+    return dict(grouped)

package/mcp-server/src/utils.py

@@ -1,84 +1,167 @@
-"""Shared utilities for claude-self-reflect MCP server and scripts."""
+"""Utility functions and ProjectResolver for Claude Self-Reflect MCP server."""
 
+import os
+import hashlib
+import re
 from pathlib import Path
+from typing import Optional, List, Tuple
+from config import logger, CLAUDE_PROJECTS_PATH
 
-
-def normalize_project_name(project_path: str, _depth: int = 0) -> str:
-    """
-    Normalize project name for consistent hashing across import/search.
+class ProjectResolver:
+    """Resolves project names and paths for Claude conversations."""
     
-    Handles various path formats:
-    - Claude logs format: -Users-kyle-Code-claude-self-reflect -> claude-self-reflect
-    - File paths in Claude logs: /path/to/-Users-kyle-Code-claude-self-reflect/file.jsonl -> claude-self-reflect  
-    - Regular file paths: /path/to/project/file.txt -> project
-    - Regular paths: /path/to/project -> project
-    - Already normalized: project -> project
-    - Docker mount paths: /logs/-Users-name-projects-project -> project
+    @staticmethod
+    def get_current_project() -> Optional[str]:
+        """Get the current project name from working directory."""
+        cwd = os.getcwd()
+        
+        # Check if we're in a known project directory
+        if '/projects/' in cwd or '/repos/' in cwd or '/code/' in cwd:
+            # Extract project name from path
+            parts = cwd.split('/')
+            for i, part in enumerate(parts):
+                if part in ['projects', 'repos', 'code'] and i + 1 < len(parts):
+                    return parts[i + 1]
+        
+        # Fall back to last directory name
+        return Path(cwd).name
     
-    Args:
-        project_path: Project path or name in any format
-        _depth: Internal recursion depth counter (do not use)
+    @staticmethod
+    def normalize_project_name(project_name: str) -> str:
+        """Normalize project name for consistent matching."""
+        # Remove common prefixes/suffixes
+        name = project_name
+        for prefix in ['/Users/', '/home/', 'projects/', 'repos/', 'code/']:
+            if name.startswith(prefix):
+                name = name[len(prefix):]
+        
+        # Convert path separators to underscores
+        name = name.replace('/', '_').replace('-', '_')
         
-    Returns:
-        Normalized project name suitable for consistent hashing
-    """
-    if not project_path:
-        return ""
+        # Remove trailing underscores
+        name = name.rstrip('_')
+        
+        return name
     
-    # Prevent infinite recursion on malformed inputs
-    if _depth > 10:
-        return Path(project_path).name
+    @staticmethod
+    def get_project_hash(project_name: str) -> str:
+        """Get hash for project name (used in collection naming)."""
+        normalized = ProjectResolver.normalize_project_name(project_name)
+        return hashlib.md5(normalized.encode()).hexdigest()[:8]
     
-    # Remove trailing slashes
-    project_path = project_path.rstrip('/')
+    @staticmethod
+    def find_project_collections(
+        all_collections: List[str], 
+        project_name: str
+    ) -> List[str]:
+        """Find collections belonging to a specific project."""
+        normalized = ProjectResolver.normalize_project_name(project_name)
+        project_hash = ProjectResolver.get_project_hash(project_name)
+        
+        matching = []
+        for collection in all_collections:
+            # Check if collection matches project hash
+            if collection.startswith(f"conv_{project_hash}_"):
+                matching.append(collection)
+            # Also check for project name in collection
+            elif normalized in collection.replace('-', '_'):
+                matching.append(collection)
+        
+        return matching
     
-    # Handle Claude logs format (starts with dash)
-    if project_path.startswith('-'):
-        # For paths like -Users-kyle-Code-claude-self-reflect
-        # We want to extract the actual project name which may contain dashes
-        # Strategy: Find common parent directories and extract what comes after
+    @staticmethod
+    def extract_project_from_collection(collection_name: str) -> str:
+        """Extract project name from collection name."""
+        # Remove conv_ prefix and suffixes
+        name = collection_name
+        if name.startswith('conv_'):
+            name = name[5:]
+        
+        # Remove hash prefix if present
+        if '_' in name and len(name.split('_')[0]) == 8:
+            # Likely a hash, remove it
+            parts = name.split('_', 1)
+            if len(parts) > 1:
+                name = parts[1]
         
-        # Remove leading dash and convert back to path-like format
-        path_str = project_path[1:].replace('-', '/')
-        path_parts = Path(path_str).parts
+        # Remove embedding type suffix
+        for suffix in ['_voyage', '_local']:
+            if name.endswith(suffix):
+                name = name[:-len(suffix)]
         
-        # Look for common project parent directories
-        project_parents = {'projects', 'code', 'Code', 'repos', 'repositories', 
-                          'dev', 'Development', 'work', 'src', 'github'}
+        return name
+
+def parse_natural_language_time(time_str: str) -> Tuple[Optional[str], Optional[str]]:
+    """Parse natural language time strings into ISO timestamps."""
+    from datetime import datetime, timedelta, timezone
+    
+    now = datetime.now(timezone.utc)
+    time_str_lower = time_str.lower().strip()
+    
+    # Handle relative times
+    if 'yesterday' in time_str_lower:
+        start = (now - timedelta(days=1)).replace(hour=0, minute=0, second=0)
+        end = start + timedelta(days=1)
+        return start.isoformat(), end.isoformat()
+    
+    if 'today' in time_str_lower:
+        start = now.replace(hour=0, minute=0, second=0)
+        return start.isoformat(), now.isoformat()
+    
+    # Parse "last X" patterns
+    last_match = re.match(r'last (\d+) (hour|day|week|month)', time_str_lower)
+    if last_match:
+        amount = int(last_match.group(1))
+        unit = last_match.group(2)
         
-        # Find the project name after a known parent directory
-        for i, part in enumerate(path_parts):
-            if part.lower() in project_parents and i + 1 < len(path_parts):
-                # Everything after the parent directory is the project name
-                # Join remaining parts with dash if project name has multiple components
-                remaining = path_parts[i + 1:]
-                return '-'.join(remaining)
+        if unit == 'hour':
+            delta = timedelta(hours=amount)
+        elif unit == 'day':
+            delta = timedelta(days=amount)
+        elif unit == 'week':
+            delta = timedelta(weeks=amount)
+        elif unit == 'month':
+            delta = timedelta(days=amount * 30)
+        else:
+            delta = timedelta(days=7)
         
-        # Fallback: just use the last component
-        return path_parts[-1] if path_parts else project_path
+        start = now - delta
+        return start.isoformat(), now.isoformat()
     
-    # Check if this is a file path that contains a Claude logs directory
-    # Pattern: /path/to/-Users-...-projects-..../filename
-    path_obj = Path(project_path)
+    # Parse "past X" patterns
+    past_match = re.match(r'past (\d+) (hour|day|week|month)', time_str_lower)
+    if past_match:
+        amount = int(past_match.group(1))
+        unit = past_match.group(2)
+        
+        if unit == 'hour':
+            delta = timedelta(hours=amount)
+        elif unit == 'day':
+            delta = timedelta(days=amount)
+        elif unit == 'week':
+            delta = timedelta(weeks=amount)
+        elif unit == 'month':
+            delta = timedelta(days=amount * 30)
+        else:
+            delta = timedelta(days=7)
+        
+        start = now - delta
+        return start.isoformat(), now.isoformat()
     
-    # Check if this is a Docker mount path specifically
-    # e.g., /logs/-Users-ramakrishnanannaswamy-projects-claude-self-reflect
-    if str(path_obj).startswith("/logs/") and path_obj.name.startswith("-"):
-        # Process this directory name recursively (Docker case only)
-        return normalize_project_name(path_obj.name, _depth + 1)
+    # Default to last week
+    if 'week' in time_str_lower:
+        start = now - timedelta(days=7)
+        return start.isoformat(), now.isoformat()
     
-    # Look for a parent directory that starts with dash (Claude logs format)
-    for parent in path_obj.parents:
-        parent_name = parent.name
-        if parent_name.startswith("-"):
-            # Found a Claude logs directory, process it
-            return normalize_project_name(parent_name, _depth + 1)
+    # Default to last 24 hours
+    start = now - timedelta(days=1)
+    return start.isoformat(), now.isoformat()
+
+def escape_xml(text: str, attr: bool = False) -> str:
+    """Escape text for XML output."""
+    from xml.sax.saxutils import escape
     
-    # Handle regular paths - if it's a file, get the parent directory
-    # Otherwise use the directory/project name itself
-    if path_obj.suffix:  # It's a file (has an extension)
-        # Use the parent directory name
-        return path_obj.parent.name
-    else:
-        # Use the directory name itself
-        return path_obj.name
+    if attr:
+        # For attributes, also escape quotes
+        return escape(text, {'"': '&quot;'})
+    return escape(text)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "claude-self-reflect",
-  "version": "3.2.4",
+  "version": "3.3.0",
   "description": "Give Claude perfect memory of all your conversations - Installation wizard for Python MCP server",
   "keywords": [
     "claude",
@@ -41,9 +41,18 @@
     "mcp-server/run-mcp-clean.sh",
     "mcp-server/run-mcp-docker.sh",
     "scripts/import-*.py",
+    "scripts/streaming-*.py",
+    "scripts/add-timestamp-indexes.py",
+    "scripts/debug-*.py",
+    "scripts/check-collections.py",
+    "scripts/status.py",
+    "scripts/utils.py",
     "scripts/importer/**/*.py",
+    "scripts/delta-metadata-update.py",
     "scripts/delta-metadata-update-safe.py",
     "scripts/force-metadata-recovery.py",
+    "scripts/precompact-hook.sh",
+    "scripts/import-latest.py",
     "shared/**/*.py",
     ".claude/agents/*.md",
     "config/qdrant-config.yaml",