#!/usr/bin/env python3
"""
Token optimization utilities and configured instances.
Provides centralized access to response formatter and token estimator
with configuration-driven defaults.
"""
from typing import Optional
from .response import ResponseFormatter, default_formatter as _default_formatter
from .tokens import TokenEstimator, token_estimator as _default_token_estimator
def get_response_formatter(
token_warning_threshold: Optional[int] = None,
compact_default_fields: Optional[list] = None
) -> ResponseFormatter:
"""
Get a configured ResponseFormatter instance.
Args:
token_warning_threshold: Override default token warning threshold
compact_default_fields: Override default compact mode fields
Returns:
Configured ResponseFormatter instance
"""
# Import settings lazily to avoid circular imports
try:
from ..config.settings import settings
threshold = token_warning_threshold or settings.token_warning_threshold
fields = compact_default_fields or settings.default_field_selection
except ImportError:
# Fallback to defaults if settings not available
threshold = token_warning_threshold or 4000
fields = compact_default_fields or ["id", "message", "timestamp", "emoji", "agent"]
formatter = ResponseFormatter(token_warning_threshold=threshold)
if fields:
formatter.COMPACT_DEFAULT_FIELDS = fields
return formatter
def get_token_estimator(
model: Optional[str] = None,
daily_limit: Optional[int] = None,
operation_limit: Optional[int] = None
) -> TokenEstimator:
"""
Get a configured TokenEstimator instance.
Args:
model: Override default tokenizer model
daily_limit: Override default daily token limit
operation_limit: Override default operation token limit
Returns:
Configured TokenEstimator instance
"""
# Import settings lazily to avoid circular imports
try:
from ..config.settings import settings
tokenizer_model = model or settings.tokenizer_model
daily = daily_limit or settings.token_daily_limit
operation = operation_limit or settings.token_operation_limit
except ImportError:
# Fallback to defaults if settings not available
tokenizer_model = model or "gpt-4"
daily = daily_limit or 100000
operation = operation_limit or 8000
return TokenEstimator(
model=tokenizer_model,
daily_limit=daily,
operation_limit=operation
)
# Global configured instances (lazy-loaded)
_configured_formatter: Optional[ResponseFormatter] = None
_configured_token_estimator: Optional[TokenEstimator] = None
def get_configured_response_formatter() -> ResponseFormatter:
"""Get the globally configured ResponseFormatter instance."""
global _configured_formatter
if _configured_formatter is None:
_configured_formatter = get_response_formatter()
return _configured_formatter
def get_configured_token_estimator() -> TokenEstimator:
"""Get the globally configured TokenEstimator instance."""
global _configured_token_estimator
if _configured_token_estimator is None:
_configured_token_estimator = get_token_estimator()
return _configured_token_estimator
def reset_configured_instances():
"""Reset global configured instances (useful for testing)."""
global _configured_formatter, _configured_token_estimator
_configured_formatter = None
_configured_token_estimator = None
# Export the configured instances for backward compatibility
configured_formatter = get_configured_response_formatter()
configured_token_estimator = get_configured_token_estimator()
