72 lines
2.3 KiB
TypeScript
72 lines
2.3 KiB
TypeScript
|
/**
|
||
|
|
* Context-window usage for a specific session.
|
||
|
|
*
|
||
|
|
* `useSessionUIStore.getContextUsage` cannot serve this panel. It reads
|
||
|
|
* `getSyncMessages(sessionId)` with **no directory**, which resolves to the
|
||
|
|
* *current* directory's child store, and it keys off the store's own
|
||
|
|
* `currentSessionId`. A session held by another directory — a worktree, or any
|
||
|
|
* moment right after a directory switch — therefore reads as "no messages" and
|
||
|
|
* the readout silently disappears while the header still shows a value.
|
||
|
|
*
|
||
|
|
* This computes the same quantity from messages the caller has already
|
||
|
|
* subscribed to for a known session and directory, so there is no hidden
|
||
|
|
* global read to race with.
|
||
|
|
*/
|
||
|
|
|
||
|
|
type MessageTokens = {
|
||
|
|
input?: number;
|
||
|
|
output?: number;
|
||
|
|
reasoning?: number;
|
||
|
|
cache?: { read?: number; write?: number };
|
||
|
|
};
|
||
|
|
|
||
|
|
type MessageLike = {
|
||
|
|
id?: string;
|
||
|
|
role?: string;
|
||
|
|
tokens?: MessageTokens;
|
||
|
|
};
|
||
|
|
|
||
|
|
type WorkStatusContextUsage = {
|
||
|
|
totalTokens: number;
|
||
|
|
/** Context limit actually used for the ratio, after the default fallback. */
|
||
|
|
limit: number;
|
||
|
|
/** Unrounded, so the panel and the header cannot disagree by a rounding step. */
|
||
|
|
percent: number;
|
||
|
|
};
|
||
|
|
|
||
|
|
/** The store's own fallback when a model exposes no context limit. */
|
||
|
|
export const DEFAULT_CONTEXT_LIMIT = 200_000;
|
||
|
|
|
||
|
|
const sumTokens = (tokens: MessageTokens): number => (
|
||
|
|
(tokens.input ?? 0)
|
||
|
|
+ (tokens.output ?? 0)
|
||
|
|
+ (tokens.reasoning ?? 0)
|
||
|
|
+ (tokens.cache?.read ?? 0)
|
||
|
|
+ (tokens.cache?.write ?? 0)
|
||
|
|
);
|
||
|
|
|
||
|
|
/**
|
||
|
|
* Usage from the newest assistant message that reported a non-zero token count.
|
||
|
|
* Each assistant turn reports the whole window it saw, so the latest one is the
|
||
|
|
* current fill — not a sum across turns.
|
||
|
|
*/
|
||
|
|
export const computeContextUsage = (
|
||
|
|
messages: readonly MessageLike[],
|
||
|
|
contextLimit: number,
|
||
|
|
): WorkStatusContextUsage | null => {
|
||
|
|
if (messages.length === 0) return null;
|
||
|
|
|
||
|
|
for (let index = messages.length - 1; index >= 0; index -= 1) {
|
||
|
|
const message = messages[index];
|
||
|
|
if (message?.role !== 'assistant' || !message.tokens) continue;
|
||
|
|
|
||
|
|
const totalTokens = sumTokens(message.tokens);
|
||
|
|
if (totalTokens <= 0) continue;
|
||
|
|
|
||
|
|
const limit = contextLimit > 0 ? contextLimit : DEFAULT_CONTEXT_LIMIT;
|
||
|
|
return { totalTokens, limit, percent: (totalTokens / limit) * 100 };
|
||
|
|
}
|
||
|
|
|
||
|
|
return null;
|
||
|
|
};
|