utils/date_parser.py

"""
Natural language date/time parsing utilities.
Converts human-readable date expressions to structured datetime objects.
Handles timezone conversions and validation for calendar events.
"""

import re
import logging
from datetime import datetime, timedelta
from typing import Optional, Tuple, Dict
import pytz
import dateparser

logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)


def parse_datetime(
    text: str,
    reference_time: Optional[datetime] = None,
    timezone: str = 'UTC'
) -> Optional[datetime]:
    """
    Parse natural language datetime expression.
    
    Args:
        text: Natural language datetime (e.g., "tomorrow at 3pm", "next Monday")
        reference_time: Reference datetime (defaults to now)
        timezone: Timezone string (e.g., 'UTC', 'America/New_York', 'Europe/London')
        
    Returns:
        Parsed datetime object or None if parsing failed
    
    Examples:
        >>> parse_datetime("tomorrow at 3pm")
        datetime(2026, 1, 17, 15, 0)
        >>> parse_datetime("next Monday at 10:30am")
        datetime(2026, 1, 20, 10, 30)
        >>> parse_datetime("in 2 hours")
        datetime(2026, 1, 16, 18, 0)  # If current time is 16:00
        >>> parse_datetime("December 25th 2025 at 9am")
        datetime(2025, 12, 25, 9, 0)
    """
    if not text or not text.strip():
        logger.warning("Empty text provided for datetime parsing")
        return None
    
    if reference_time is None:
        reference_time = datetime.now()
    
    try:
        # Get timezone object
        try:
            tz = pytz.timezone(timezone)
        except pytz.UnknownTimeZoneError:
            logger.warning(f"Unknown timezone '{timezone}', falling back to UTC")
            tz = pytz.UTC
        
        # Use dateparser library for robust parsing
        parsed = dateparser.parse(
            text,
            settings={
                'RELATIVE_BASE': reference_time,
                'TIMEZONE': timezone,
                'RETURN_AS_TIMEZONE_AWARE': True,
                'PREFER_DATES_FROM': 'future',
                'PREFER_DAY_OF_MONTH': 'first'
            }
        )
        
        if parsed is None:
            logger.warning(f"Failed to parse datetime: '{text}'")
            return None
        
        # Ensure timezone awareness
        if parsed.tzinfo is None:
            parsed = tz.localize(parsed)
        
        # Convert to target timezone if needed
        if parsed.tzinfo != tz:
            parsed = parsed.astimezone(tz)
        
        logger.info(f"Parsed '{text}' -> {parsed.strftime('%Y-%m-%d %H:%M:%S %Z')}")
        return parsed
        
    except Exception as e:
        logger.error(f"Error parsing datetime '{text}': {e}")
        return None


def parse_duration(text: str) -> Optional[timedelta]:
    """
    Parse natural language duration expression.
    
    Args:
        text: Duration text (e.g., "30 minutes", "1 hour", "2 hours 30 minutes")
        
    Returns:
        timedelta object or None if parsing failed
    
    Examples:
        >>> parse_duration("30 minutes")
        timedelta(minutes=30)
        >>> parse_duration("1 hour")
        timedelta(hours=1)
        >>> parse_duration("2 hours 30 minutes")
        timedelta(hours=2, minutes=30)
        >>> parse_duration("90 minutes")
        timedelta(minutes=90)
    """
    if not text or not text.strip():
        return None
    
    text = text.lower().strip()
    
    try:
        total_minutes = 0
        
        # Pattern for hours
        hour_patterns = [
            r'(\d+(?:\.\d+)?)\s*(?:hour|hr|h)s?',
            r'(\d+(?:\.\d+)?)\s*h\b'
        ]
        
        for pattern in hour_patterns:
            matches = re.findall(pattern, text)
            for match in matches:
                total_minutes += float(match) * 60
        
        # Pattern for minutes
        minute_patterns = [
            r'(\d+(?:\.\d+)?)\s*(?:minute|min|m)s?',
            r'(\d+(?:\.\d+)?)\s*m\b'
        ]
        
        for pattern in minute_patterns:
            matches = re.findall(pattern, text)
            for match in matches:
                total_minutes += float(match)
        
        # Pattern for days
        day_pattern = r'(\d+(?:\.\d+)?)\s*(?:day|d)s?'
        day_matches = re.findall(day_pattern, text)
        for match in day_matches:
            total_minutes += float(match) * 24 * 60
        
        if total_minutes > 0:
            duration = timedelta(minutes=total_minutes)
            logger.info(f"Parsed duration '{text}' -> {duration}")
            return duration
        
        # Try parsing as simple number (assume minutes)
        number_match = re.search(r'^\s*(\d+(?:\.\d+)?)\s*$', text)
        if number_match:
            minutes = float(number_match.group(1))
            duration = timedelta(minutes=minutes)
            logger.info(f"Parsed duration '{text}' as {minutes} minutes -> {duration}")
            return duration
        
        logger.warning(f"Could not parse duration: '{text}'")
        return None
        
    except Exception as e:
        logger.error(f"Error parsing duration '{text}': {e}")
        return None


def get_default_duration() -> timedelta:
    """
    Get default meeting duration.
    
    Returns:
        timedelta of 1 hour
    """
    return timedelta(hours=1)


def validate_future_datetime(dt: datetime, allow_past: bool = False) -> bool:
    """
    Validate that datetime is in the future.
    
    Args:
        dt: Datetime to validate
        allow_past: If True, allow past datetimes
        
    Returns:
        True if valid, False otherwise
    """
    if dt is None:
        return False
    
    if allow_past:
        return True
    
    now = datetime.now(dt.tzinfo) if dt.tzinfo else datetime.now()
    
    if dt <= now:
        logger.warning(f"DateTime {dt} is in the past (now: {now})")
        return False
    
    return True


def format_for_google_calendar(dt: datetime) -> str:
    """
    Format datetime for Google Calendar API.
    
    Args:
        dt: Datetime object
        
    Returns:
        ISO 8601 formatted string
    
    Example:
        >>> dt = datetime(2026, 1, 17, 15, 0, tzinfo=pytz.UTC)
        >>> format_for_google_calendar(dt)
        '2026-01-17T15:00:00Z'
    """
    if dt is None:
        raise ValueError("Cannot format None datetime")
    
    # Ensure timezone awareness
    if dt.tzinfo is None:
        dt = pytz.UTC.localize(dt)
    
    # Convert to UTC for Google Calendar
    dt_utc = dt.astimezone(pytz.UTC)
    
    # Format as ISO 8601
    formatted = dt_utc.strftime('%Y-%m-%dT%H:%M:%S') + 'Z'
    
    return formatted


def parse_datetime_range(
    text: str,
    reference_time: Optional[datetime] = None,
    timezone: str = 'UTC',
    default_duration: Optional[timedelta] = None
) -> Optional[Tuple[datetime, datetime]]:
    """
    Parse a datetime range from text.
    
    Handles formats like:
    - "tomorrow at 3pm for 2 hours"
    - "next Monday 10am to 11:30am"
    - "December 25th at 9am" (uses default duration)
    
    Args:
        text: Natural language datetime range
        reference_time: Reference datetime
        timezone: Timezone string
        default_duration: Default duration if not specified (defaults to 1 hour)
        
    Returns:
        Tuple of (start_datetime, end_datetime) or None
    """
    if default_duration is None:
        default_duration = get_default_duration()
    
    try:
        # Check for duration pattern (e.g., "tomorrow at 3pm for 2 hours")
        duration_pattern = r'(.+?)\s+(?:for|lasting)\s+(.+?)$'
        duration_match = re.search(duration_pattern, text, re.IGNORECASE)
        
        if duration_match:
            datetime_text = duration_match.group(1)
            duration_text = duration_match.group(2)
            
            start_dt = parse_datetime(datetime_text, reference_time, timezone)
            duration = parse_duration(duration_text)
            
            if start_dt and duration:
                end_dt = start_dt + duration
                return (start_dt, end_dt)
        
        # Check for "from...to..." pattern
        range_pattern = r'(?:from\s+)?(.+?)\s+(?:to|until|-)\s+(.+?)$'
        range_match = re.search(range_pattern, text, re.IGNORECASE)
        
        if range_match:
            start_text = range_match.group(1)
            end_text = range_match.group(2)
            
            start_dt = parse_datetime(start_text, reference_time, timezone)
            
            # Try parsing end as full datetime
            end_dt = parse_datetime(end_text, reference_time, timezone)
            
            # If end_dt is before start_dt, it might be just a time on the same day
            if end_dt and start_dt and end_dt < start_dt:
                # Extract time from end_text and apply to start date
                time_pattern = r'(\d{1,2})(?::(\d{2}))?\s*(am|pm)?'
                time_match = re.search(time_pattern, end_text, re.IGNORECASE)
                
                if time_match:
                    hour = int(time_match.group(1))
                    minute = int(time_match.group(2)) if time_match.group(2) else 0
                    period = time_match.group(3)
                    
                    if period and period.lower() == 'pm' and hour != 12:
                        hour += 12
                    elif period and period.lower() == 'am' and hour == 12:
                        hour = 0
                    
                    end_dt = start_dt.replace(hour=hour, minute=minute, second=0, microsecond=0)
            
            if start_dt and end_dt:
                return (start_dt, end_dt)
        
        # Single datetime with default duration
        start_dt = parse_datetime(text, reference_time, timezone)
        
        if start_dt:
            end_dt = start_dt + default_duration
            return (start_dt, end_dt)
        
        return None
        
    except Exception as e:
        logger.error(f"Error parsing datetime range '{text}': {e}")
        return None


def create_event_times(
    start_text: str,
    duration_text: Optional[str] = None,
    end_text: Optional[str] = None,
    timezone: str = 'UTC',
    validate_future: bool = True
) -> Optional[Dict[str, str]]:
    """
    Create properly formatted event times for calendar services.
    
    Args:
        start_text: Natural language start time
        duration_text: Optional duration text
        end_text: Optional end time text (if not using duration)
        timezone: Timezone string
        validate_future: Whether to validate that start is in the future
        
    Returns:
        Dictionary with 'start' and 'end' ISO formatted strings, or None
    
    Example:
        >>> create_event_times("tomorrow at 3pm", "2 hours")
        {'start': '2026-01-17T15:00:00Z', 'end': '2026-01-17T17:00:00Z'}
    """
    try:
        # Parse start datetime
        start_dt = parse_datetime(start_text, timezone=timezone)
        
        if not start_dt:
            logger.error(f"Failed to parse start time: '{start_text}'")
            return None
        
        # Validate future if required
        if validate_future and not validate_future_datetime(start_dt):
            logger.error(f"Start time is in the past: {start_dt}")
            return None
        
        # Determine end datetime
        end_dt = None
        
        if end_text:
            # Parse explicit end time
            end_dt = parse_datetime(end_text, timezone=timezone)
            
            # If end is before start, assume same day
            if end_dt and end_dt < start_dt:
                # Try to interpret as time on same day
                time_pattern = r'(\d{1,2})(?::(\d{2}))?\s*(am|pm)?'
                time_match = re.search(time_pattern, end_text, re.IGNORECASE)
                
                if time_match:
                    hour = int(time_match.group(1))
                    minute = int(time_match.group(2)) if time_match.group(2) else 0
                    period = time_match.group(3)
                    
                    if period and period.lower() == 'pm' and hour != 12:
                        hour += 12
                    elif period and period.lower() == 'am' and hour == 12:
                        hour = 0
                    
                    end_dt = start_dt.replace(hour=hour, minute=minute, second=0, microsecond=0)
        
        elif duration_text:
            # Parse duration and calculate end
            duration = parse_duration(duration_text)
            if duration:
                end_dt = start_dt + duration
            else:
                logger.warning(f"Failed to parse duration: '{duration_text}', using default")
                end_dt = start_dt + get_default_duration()
        else:
            # Use default duration
            end_dt = start_dt + get_default_duration()
        
        if not end_dt:
            logger.error("Failed to determine end time")
            return None
        
        if end_dt <= start_dt:
            logger.error(f"End time ({end_dt}) must be after start time ({start_dt})")
            return None
        
        # Format for Google Calendar API
        result = {
            'start': format_for_google_calendar(start_dt),
            'end': format_for_google_calendar(end_dt),
            'start_datetime': start_dt,
            'end_datetime': end_dt
        }
        
        logger.info(f"Created event times: {result['start']} to {result['end']}")
        return result
        
    except Exception as e:
        logger.error(f"Error creating event times: {e}")
        return None


# Timezone helpers
def get_user_timezone() -> str:
    """
    Get user's local timezone.
    
    Returns:
        Timezone string (e.g., 'America/New_York')
    """
    try:
        import tzlocal
        local_tz = tzlocal.get_localzone()
        return str(local_tz)
    except Exception:
        logger.warning("Could not determine local timezone, using UTC")
        return 'UTC'


def list_common_timezones() -> list:
    """Get list of common timezones."""
    return [
        'UTC',
        'America/New_York',
        'America/Chicago',
        'America/Denver',
        'America/Los_Angeles',
        'America/Toronto',
        'Europe/London',
        'Europe/Paris',
        'Europe/Berlin',
        'Asia/Tokyo',
        'Asia/Shanghai',
        'Asia/Hong_Kong',
        'Asia/Singapore',
        'Australia/Sydney',
        'Pacific/Auckland'
    ]


if __name__ == "__main__":
    # Test examples
    print("=== Date Parser Tests ===\n")
    
    test_cases = [
        "tomorrow at 3pm",
        "next Monday at 10:30am",
        "in 2 hours",
        "December 25th 2025 at 9am",
        "January 20th at 2pm",
    ]
    
    print("DateTime Parsing:")
    for text in test_cases:
        result = parse_datetime(text)
        print(f"  '{text}' -> {result}")
    
    print("\nDuration Parsing:")
    duration_cases = [
        "30 minutes",
        "1 hour",
        "2 hours 30 minutes",
        "90 minutes",
        "1.5 hours"
    ]
    
    for text in duration_cases:
        result = parse_duration(text)
        print(f"  '{text}' -> {result}")
    
    print("\nEvent Time Creation:")
    event_result = create_event_times(
        "tomorrow at 3pm",
        duration_text="2 hours",
        timezone="America/New_York"
    )
    if event_result:
        print(f"  Start: {event_result['start']}")
        print(f"  End: {event_result['end']}")
    
    print(f"\nCurrent timezone: {get_user_timezone()}")