token_usage#

Utilities for normalizing and aggregating token usage across providers.

Functions

parse_copilot_otel_usage(otel_path)

Parse Copilot OTEL JSONL usage and aggregate token counts over lines.

zero_token_usage()

Classes

TokenUsage(input_tokens, output_tokens, ...)

class kernelfoundry.algorithm.utils.token_usage.TokenUsage(input_tokens, output_tokens, cached_input_tokens)[source]#
input_tokens: int#

Alias for field number 0

output_tokens: int#

Alias for field number 1

cached_input_tokens: int#

Alias for field number 2

__add__(other: dict[str, Any] | TokenUsage) → TokenUsage[source]#

Add token usage from dict or another TokenUsage.

__radd__(other: dict[str, Any]) → TokenUsage[source]#

Support reverse addition (dict + TokenUsage).

kernelfoundry.algorithm.utils.token_usage.zero_token_usage() → TokenUsage[source]#
kernelfoundry.algorithm.utils.token_usage.parse_copilot_otel_usage(otel_path: str | Path) → TokenUsage[source]#

Parse Copilot OTEL JSONL usage and aggregate token counts over lines. Returns zero usage when the file is absent (i.e. copilot did not start)