BPOS-TMP-REPO/apps/api/src/modules/report/internal/domain/aggregate.ts

/**
 * 汇报聚合与度量八指标 —— 纯投影(编码规范 §10.2;语义出处 OS 8.1 / 8.3)。
 *
 * 「口径全部挂在**归侧标签与事件流**上——零埋点,统计即查询。」
 * 所以这里的输入只有两样:归侧过的日志行、任务的终态与纠正信号。没有第三种埋点。
 */
import type { AttributionSide, TerminalState } from '@contracts';
import { TERMINAL_STATE_LABEL } from '@contracts';
import type { Window } from './period.ts';
import { withinWindow } from './period.ts';

export type LogFact = {
  readonly taskId: string;
  readonly at: string;
  readonly side: AttributionSide;
  readonly title: string;
  /** 人类工具发卡 —— 打扰率的计量对象(四卡合计) */
  readonly isHumanToolCall: boolean;
  /** 判断次数中由 DMN / 机检确定性求值的 */
  readonly isDeterministicJudgement: boolean;
  readonly isJudgement: boolean;
  /** loop 断点:当下可为与目标所需交集为空 */
  readonly isBreakpoint: boolean;
};

export type TaskFact = {
  readonly taskId: string;
  readonly title: string;
  readonly activityId: string;
  readonly startedAt: string;
  readonly finishedAt?: string | undefined;
  readonly terminal?: TerminalState | undefined;
  /** 提交批次:每次放行提交 = 一个产出批次 */
  readonly submissions: number;
  /** 被纠正批次:驳回重做 / 驳回迭代 / 叫停 / 裁决推翻 */
  readonly corrections: number;
  readonly tokens: number;
  readonly activeMs: number;
  readonly humanCalls: number;
};

/** 度量八指标(OS 8.3)—— 自主率与纠正率必须成对读 */
export type Metrics = {
  readonly autonomy: number;
  readonly correction: number;
  readonly effectiveAutonomy: number;
  readonly interruption: number;
  readonly breakpoint: number;
  readonly dmnCoverage: number;
  /** 单位成本:总消耗 ÷ **放行通过**的任务数 —— 分母只认成功终态 */
  readonly unitCost: { readonly tokens: number; readonly minutes: number; readonly humanCalls: number };
  /** 终态分布:五个终态的占比 */
  readonly terminalMix: Readonly<Record<TerminalState, number>>;
};

const ratio = (num: number, den: number): number => (den === 0 ? 0 : num / den);

export function computeMetrics(logs: readonly LogFact[], tasks: readonly TaskFact[]): Metrics {
  const machine = logs.filter((l) => l.side === 'machine').length;
  const human = logs.filter((l) => l.side === 'human').length;
  // 裁决与等待不进分母
  const autonomy = ratio(machine, machine + human);

  const submissions = tasks.reduce((s, t) => s + t.submissions, 0);
  const corrections = tasks.reduce((s, t) => s + t.corrections, 0);
  const correction = ratio(corrections, submissions);

  const judgements = logs.filter((l) => l.isJudgement).length;
  const deterministic = logs.filter((l) => l.isDeterministicJudgement).length;

  const released = tasks.filter((t) => t.terminal === 'released');
  const terminalCounts = tasks.reduce<Record<string, number>>((acc, t) => {
    if (t.terminal !== undefined) acc[t.terminal] = (acc[t.terminal] ?? 0) + 1;
    return acc;
  }, {});
  const terminalTotal = Object.values(terminalCounts).reduce((s, n) => s + n, 0);

  const terminalMix = (Object.keys(TERMINAL_STATE_LABEL) as TerminalState[]).reduce(
    (acc, state) => ({ ...acc, [state]: ratio(terminalCounts[state] ?? 0, terminalTotal) }),
    {} as Record<TerminalState, number>,
  );

  return {
    autonomy,
    correction,
    effectiveAutonomy: autonomy * (1 - correction),
    interruption: ratio(
      tasks.reduce((s, t) => s + t.humanCalls, 0),
      tasks.length,
    ),
    breakpoint: ratio(logs.filter((l) => l.isBreakpoint).length, logs.length),
    dmnCoverage: ratio(deterministic, judgements),
    unitCost: {
      // 环是否真在产出:分母只认放行通过,否则「跑得多」会被算成「产出多」
      tokens: ratio(
        tasks.reduce((s, t) => s + t.tokens, 0),
        released.length,
      ),
      minutes: ratio(
        tasks.reduce((s, t) => s + t.activeMs, 0) / 60_000,
        released.length,
      ),
      humanCalls: ratio(
        tasks.reduce((s, t) => s + t.humanCalls, 0),
        released.length,
      ),
    },
    terminalMix,
  };
}

export type ReportSection = {
  readonly taskId: string;
  readonly title: string;
  readonly activityId: string;
  readonly terminalLabel?: string | undefined;
  /** 该窗口内这单任务的日志摘句 —— 汇报正文的骨架,零 LLM 即可成篇 */
  readonly lines: readonly string[];
};

export type ReportDraft = {
  readonly sections: readonly ReportSection[];
  readonly metrics: Metrics;
  readonly counts: { readonly touched: number; readonly released: number; readonly logLines: number };
};

/**
 * 汇报草案:按任务分段,段内按时间列日志摘句。
 * 「零手工汇报」的含义是**员工不写**,不是「系统编」——所以草案的每一句都取自任务日志,
 * 叙述性小结(若配了系统用途模型)是锦上添花,不是事实来源。
 */
export function draftReport(
  logs: readonly LogFact[],
  tasks: readonly TaskFact[],
  window: Window,
): ReportDraft {
  const inWindow = logs.filter((l) => withinWindow(l.at, window));
  const touchedIds = new Set(inWindow.map((l) => l.taskId));
  const touched = tasks.filter((t) => touchedIds.has(t.taskId));

  const sections = touched.map<ReportSection>((t) => ({
    taskId: t.taskId,
    title: t.title,
    activityId: t.activityId,
    terminalLabel: t.terminal === undefined ? undefined : TERMINAL_STATE_LABEL[t.terminal],
    lines: inWindow
      .filter((l) => l.taskId === t.taskId)
      .sort((a, b) => (a.at < b.at ? -1 : a.at > b.at ? 1 : 0))
      .map((l) => l.title),
  }));

  return {
    sections,
    metrics: computeMetrics(inWindow, touched),
    counts: {
      touched: touched.length,
      released: touched.filter((t) => t.terminal === 'released').length,
      logLines: inWindow.length,
    },
  };
}