Skip to content
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
40 changes: 39 additions & 1 deletion rig-integrations/rig-bedrock/src/types/assistant_content.rs
Original file line number Diff line number Diff line change
Expand Up @@ -42,11 +42,16 @@ impl TryFrom<AwsConverseOutput> for completion::CompletionResponse<AwsConverseOu
)),
}?;

// Folds cache read/write tokens into `input_tokens`, matching the
// streaming path — Bedrock reports `inputTokens` exclusive of cached
// tokens, while `totalTokens` already includes them.
let usage = value
.0
.usage()
.map(|usage| completion::Usage {
input_tokens: usage.input_tokens as u64,
input_tokens: usage.input_tokens as u64
+ usage.cache_read_input_tokens.unwrap_or_default() as u64
+ usage.cache_write_input_tokens.unwrap_or_default() as u64,
output_tokens: usage.output_tokens as u64,
total_tokens: usage.total_tokens as u64,
})
Expand Down Expand Up @@ -191,6 +196,39 @@ mod tests {
);
}

#[test]
fn aws_converse_output_folds_cache_tokens_into_input() {
let message = aws_bedrock::Message::builder()
.role(aws_bedrock::ConversationRole::Assistant)
.content(aws_bedrock::ContentBlock::Text("txt".into()))
.build()
.unwrap();
let converse_output =
aws_sdk_bedrockruntime::operation::converse::ConverseOutput::builder()
.output(aws_bedrock::ConverseOutput::Message(message))
.stop_reason(aws_bedrock::StopReason::EndTurn)
.build()
.unwrap();
let mut internal: InternalConverseOutput = converse_output.try_into().unwrap();
// Figures shaped like a Converse response with caching active:
// inputTokens excludes cached tokens, totalTokens includes them.
internal.usage = Some(crate::types::converse_output::TokenUsage {
input_tokens: 100,
output_tokens: 50,
total_tokens: 20_150,
cache_read_input_tokens: Some(18_000),
cache_write_input_tokens: Some(2_000),
});

let completion: completion::CompletionResponse<AwsConverseOutput> =
AwsConverseOutput(internal).try_into().unwrap();
assert_eq!(
completion.usage.input_tokens, 20_100,
"cache tokens fold into input"
);
assert_eq!(completion.usage.total_tokens, 20_150);
}

#[test]
fn aws_content_block_to_assistant_content() {
let content_block = aws_bedrock::ContentBlock::Text("text".into());
Expand Down