Function fetchCandidateSessions

scripts/export-deepseek-session-traces.ts:236–375

(args: Args)

}

async function fetchCandidateSessions(args: Args): Promise<CandidateRow[]> {
  const targetStructs = TARGETS.flatMap((target) =>
    target.modelIds.map((modelId) => ({
      agent_id: target.agentId,
      model_id: modelId,
    })),
  )

  const datePredicate =
    args.startDate || args.endDate
      ? `
        ${args.startDate ? 'AND m.finished_at >= TIMESTAMP(@startDate)' : ''}
        ${args.endDate ? 'AND m.finished_at < TIMESTAMP(@endDate)' : ''}
      `
      : `
        AND m.finished_at >= TIMESTAMP_SUB(
          TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL @beforeDays DAY),
          INTERVAL @lookbackDays DAY
        )
        AND m.finished_at < TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL @beforeDays DAY)
      `

  const isEvenTotalSample = args.totalSessions && args.sampleMode === 'even'
  const samplingCtes = isEvenTotalSample
    ? `
    ranked_by_time AS (
      SELECT
        *,
        ROW_NUMBER() OVER (ORDER BY last_finished_at DESC, max_message_count DESC) AS time_rank,
        COUNT(*) OVER () AS total_count
      FROM session_summary
    ),
    bucketed AS (
      SELECT
        *,
        CAST(FLOOR((time_rank - 1) * @sessionLimit / total_count) AS INT64) AS sample_bucket
      FROM ranked_by_time
    ),
    ranked AS (
      SELECT
        *,
        ROW_NUMBER() OVER (
          PARTITION BY sample_bucket
          ORDER BY RAND()
        ) AS target_rank
      FROM bucketed
    )
      `
    : `
    ranked AS (
      SELECT
        *,
        ${
          args.totalSessions
            ? `ROW_NUMBER() OVER (
              ORDER BY ${
                args.sampleMode === 'random'
                  ? 'RAND()'