🪢 Open source agent evals & observability: Trace, evaluate, and improve LLM applications with one open platform.
TypeScript analyticsautogenevaluationlangchainlarge-language-modelsllama-indexllmllm-evaluationllm-observabilityllmopsmonitoringobservabilityopen-sourceopenaiplaygroundprompt-engineeringprompt-managementself-hostedycombinator
/**
* Does an observation answer the trace panel's search box?
*
* ONE definition, because two views read it now: the Tree's flat result list
* (TraceSearchList) and the Timeline, which keeps its bars and dims the ones
* that miss. A second copy is a second chance for the two to disagree about
* what a match is — and the Timeline states a count, so a disagreement would be
* visible as a number that does not match the list beside it.
*
* The rule is the one the result list has always used: a case-insensitive
* substring of the observation's type, name or id. Nothing is searched that the
* trace panel does not already show.
*/
export type SearchableNode = {
id: string;
name: string;
type: string;
};
export function matchesSearchQuery(
node: SearchableNode,
query: string,
): boolean {
// Trimmed, because a trailing space is a keystroke on the way to the next
// word and not a thing anybody means to search for. The result list already
// treated an all-whitespace query as empty; this extends the same reading to
// the padding around a real one.
const needle = query.trim().toLowerCase();
// An empty query is not a match-everything — it is no search at all, and the
// callers decide what to show instead.
if (!needle) return false;
return (
node.type.toLowerCase().includes(needle) ||
node.name.toLowerCase().includes(needle) ||
node.id.toLowerCase().includes(needle)
);
}