tihao-ai/claude-code-tihao-sourcing/scripts/intake-readiness-audit.js

#!/usr/bin/env node
const fs = require('fs');
const path = require('path');
const { readCsv: readHistoryCsv } = require('./history-dataset-from-csv');
const { readCsv: readVideoCsv } = require('./video-resource-readiness-audit');
const { readCsv: readReviewCsv } = require('./review-metrics');

const HISTORY_REQUIRED = [
  'brief编号',
  '项目名称',
  '类目',
  '客户原始Brief',
  '历史人工补号量基线',
  '参考账号或视频',
  '平台',
  '博主名称',
  '主页链接',
  '人工复核标签',
  '客户选择',
  '拒绝原因'
];

const REVIEW_REQUIRED = [
  'brief编号',
  '策略',
  '排名',
  '平台',
  '博主名称',
  '综合分',
  'brief匹配分',
  '参考风格分',
  '主页证据分',
  '视觉质感分',
  '调性一致分',
  '证据加分',
  '证据风险扣分',
  '推荐理由',
  '风险提示',
  '主页链接',
  '人工复核标签',
  '客户选择',
  '归因类型',
  '反馈原因',
  '本轮人工补号量'
];

const VIDEO_REQUIRED = [
  'brief编号',
  '项目名称',
  '平台',
  '资源角色',
  '博主名称',
  '主页链接',
  '视频链接',
  '封面链接',
  '字幕或ASR文本',
  '帧图链接',
  '标题',
  '发布时间',
  '内容摘要',
  '风格调性标签',
  '画面人设场景信号',
  '风险提示',
  '来源接口或备注',
  '是否真实资源'
];

const FINAL_CUSTOMER_DECISIONS = new Set(['客户选中', '客户拒绝', '已选中', '已拒绝', '选中', '拒绝', '客户通过', '客户淘汰']);
const NEGATIVE_REVIEW_LABELS = new Set(['跑偏', '硬性规则违规', '硬性规则违约', '调性不符', '主页质感不符', '参考账号不像']);

function main() {
  const args = parseArgs(process.argv.slice(2));
  const dataPack = args.dataPack ? path.resolve(args.dataPack) : '';
  const videoPack = args.videoPack ? path.resolve(args.videoPack) : '';
  const historyCsv = path.resolve(args.historyCsv || args.history || (dataPack ? path.join(dataPack, 'history-data-template.csv') : ''));
  const reviewCsv = path.resolve(args.reviewCsv || args.review || (dataPack ? path.join(dataPack, 'manual-review-template.csv') : ''));
  const videoCsv = path.resolve(args.videoCsv || args.video || (videoPack ? path.join(videoPack, 'video-resource-template.csv') : ''));
  if (!historyCsv || !reviewCsv || !videoCsv) {
    throw new Error('Usage: node scripts/intake-readiness-audit.js --data-pack <dir> --video-pack <dir> [--output <dir>] [--strict]');
  }

  const outputDir = path.resolve(args.output || process.env.TIHAO_INTAKE_READINESS_OUTPUT || path.join(path.dirname(historyCsv), 'intake-readiness'));
  const minBriefs = Number(args.minBriefs || process.env.TIHAO_INTAKE_MIN_BRIEFS || 5);
  const summary = buildIntakeReadinessSummary({
    historyCsv,
    reviewCsv,
    videoCsv,
    minBriefs
  });

  fs.mkdirSync(outputDir, { recursive: true });
  const jsonPath = path.join(outputDir, 'intake-readiness-summary.json');
  const reportPath = path.join(outputDir, 'intake-readiness-report.md');
  const repairCsvPath = path.join(outputDir, 'intake-readiness-repair-actions.csv');
  fs.writeFileSync(jsonPath, JSON.stringify(summary, null, 2), 'utf8');
  fs.writeFileSync(reportPath, withBom(renderReport(summary)), 'utf8');
  fs.writeFileSync(repairCsvPath, withBom(renderRepairCsv(summary.repairActions)), 'utf8');

  console.log(JSON.stringify({
    outputDir,
    json: jsonPath,
    report: reportPath,
    readyForHistoryAudit: summary.history.ready,
    readyForVideoResourceAudit: summary.video.ready,
    readyForReviewMetrics: summary.review.readyForReviewMetrics,
    readyForCustomerEffectAudit: summary.review.readyForCustomerEffectAudit,
    overallReady: summary.acceptance.overallReady,
    failureCount: summary.failureCount,
    repairActionCount: summary.repairActions.length
  }, null, 2));

  if (args.strict && !summary.acceptance.overallReady) process.exitCode = 2;
}

function buildIntakeReadinessSummary({ historyCsv, reviewCsv, videoCsv, minBriefs }) {
  const historyRows = safeReadCsv(readHistoryCsv, historyCsv);
  const reviewRows = safeReadCsv(readReviewCsv, reviewCsv);
  const videoRows = safeReadCsv(readVideoCsv, videoCsv);
  const history = auditHistory({ file: historyCsv, rows: historyRows, minBriefs });
  const review = auditReview({ file: reviewCsv, rows: reviewRows });
  const video = auditVideo({ file: videoCsv, rows: videoRows });
  const sections = [history, review, video];
  const issues = sections.flatMap(section => section.issues.map(issue => ({ section: section.id, ...issue })));
  const acceptance = {
    historyReady: history.ready,
    videoReady: video.ready,
    reviewMetricsReady: review.readyForReviewMetrics,
    customerEffectReady: review.readyForCustomerEffectAudit
  };
  acceptance.overallReady = acceptance.historyReady &&
    acceptance.videoReady &&
    acceptance.reviewMetricsReady &&
    acceptance.customerEffectReady;
  const files = { historyCsv, reviewCsv, videoCsv };
  const repairActions = buildRepairActions({ issues, files });
  return {
    generatedAt: new Date().toISOString(),
    minBriefs,
    files,
    acceptance,
    history,
    review,
    video,
    failureCount: issues.length,
    issues,
    repairActions,
    guardrails: [
      'intake readiness 只判断模板是否已被真实材料替换，不替代 history:audit、video:resource-readiness、review:metrics 或 customer-effect:audit。',
      'example.com、示例、候选博主、待客户反馈、人工待补等模板占位内容不得进入真实审计。',
      '没有客户最终选择和本轮人工补号量时，不得声明客户效果证明完成。',
      '没有真实参考视频和真实候选视频时，不得声明视频 A/B 可以开始。'
    ]
  };
}

function auditHistory({ file, rows, minBriefs }) {
  const issues = [];
  const headerOk = headerStartsWith(rows.header, HISTORY_REQUIRED);
  if (!rows.ok) issues.push(issue('read-error', rows.error));
  if (!headerOk) issues.push(issue('header', '历史数据 CSV 表头不符合 data-intake-pack 模板。'));
  const body = rows.body || [];
  const briefIds = unique(body.map(row => value(row['brief编号'])));
  if (briefIds.length < minBriefs) issues.push(issue('min-briefs', `真实历史 Brief 数不足：${briefIds.length}/${minBriefs}。`));
  if (hasPlaceholderRows(body)) issues.push(issue('placeholder', '历史数据表仍包含 example.com、示例、候选或待补等模板占位内容。'));
  if (!body.every(row => value(row['客户原始Brief']).length >= 20)) issues.push(issue('brief-text', '每行必须有足够具体的客户原始 Brief。'));
  if (!body.every(row => toNumberOrNull(row['历史人工补号量基线']) !== null)) issues.push(issue('manual-baseline', '每行必须填写历史人工补号量基线。'));
  if (!body.every(row => isLikelyRealReference(row['参考账号或视频']))) issues.push(issue('reference', '每行必须填写真实参考账号或参考视频链接。'));
  if (!body.every(row => value(row['博主名称']) && isLikelyRealUrl(row['主页链接']))) issues.push(issue('manual-list', '每行必须填写人工最终名单中的博主名称和主页链接。'));
  if (!body.every(row => FINAL_CUSTOMER_DECISIONS.has(value(row['客户选择'])))) issues.push(issue('customer-decision', '每行必须填写客户最终选中或客户拒绝，不能只写待客户反馈。'));
  const rejectedRows = body.filter(row => /拒绝|淘汰|未选/i.test(value(row['客户选择'])));
  if (!rejectedRows.every(row => value(row['拒绝原因']).length >= 2)) issues.push(issue('reject-reason', '客户拒绝样本必须填写拒绝原因。'));
  return {
    id: 'history',
    file,
    ready: issues.length === 0,
    counts: {
      rows: body.length,
      briefCount: briefIds.length,
      rejectedRows: rejectedRows.length
    },
    issues
  };
}

function auditReview({ file, rows }) {
  const issues = [];
  const headerOk = headerStartsWith(rows.header, REVIEW_REQUIRED);
  if (!rows.ok) issues.push(issue('read-error', rows.error));
  if (!headerOk) issues.push(issue('header', '人工复核 CSV 表头不符合 data-intake-pack 模板。'));
  const body = rows.body || [];
  if (!body.length) issues.push(issue('empty', '人工复核表没有候选行。'));
  if (hasPlaceholderRows(body)) issues.push(issue('placeholder', '人工复核表仍包含 example.com、示例、候选或待补等模板占位内容。'));
  if (!body.every(row => value(row['人工复核标签']))) issues.push(issue('review-label', '每行必须填写人工复核标签。'));
  const negativeRows = body.filter(row => NEGATIVE_REVIEW_LABELS.has(value(row['人工复核标签'])));
  if (!negativeRows.every(row => value(row['归因类型']))) issues.push(issue('attribution', '负样本必须填写归因类型。'));
  const hasCustomerDecision = body.some(row => FINAL_CUSTOMER_DECISIONS.has(value(row['客户选择'])));
  if (!hasCustomerDecision) issues.push(issue('customer-decision', '人工复核表至少需要客户最终选中/拒绝记录，才能进入客户效果审计。'));
  const hasCurrentManualSupplement = body.some(row => toNumberOrNull(row['本轮人工补号量']) !== null);
  if (!hasCurrentManualSupplement) issues.push(issue('current-manual-supplement', '缺少本轮人工补号量。'));
  const readyForReviewMetrics = headerOk && body.length > 0 && !hasPlaceholderRows(body) && body.every(row => value(row['人工复核标签'])) && negativeRows.every(row => value(row['归因类型']));
  const readyForCustomerEffectAudit = readyForReviewMetrics && hasCustomerDecision && hasCurrentManualSupplement;
  return {
    id: 'review',
    file,
    readyForReviewMetrics,
    readyForCustomerEffectAudit,
    counts: {
      rows: body.length,
      negativeRows: negativeRows.length,
      rowsWithCustomerDecision: body.filter(row => FINAL_CUSTOMER_DECISIONS.has(value(row['客户选择']))).length,
      rowsWithCurrentManualSupplement: body.filter(row => toNumberOrNull(row['本轮人工补号量']) !== null).length
    },
    issues
  };
}

function auditVideo({ file, rows }) {
  const issues = [];
  const headerOk = headerStartsWith(rows.header, VIDEO_REQUIRED);
  if (!rows.ok) issues.push(issue('read-error', rows.error));
  if (!headerOk) issues.push(issue('header', '视频资源 CSV 表头不符合 video-intake-pack 模板。'));
  const body = rows.body || [];
  if (!body.length) issues.push(issue('empty', '视频资源表没有资源行。'));
  if (hasPlaceholderRows(body)) issues.push(issue('placeholder', '视频资源表仍包含 example.com、示例、候选、待确认或人工待补等模板占位内容。'));
  const realRows = body.filter(row => isYes(row['是否真实资源']));
  const realReferences = realRows.filter(row => value(row['资源角色']) === '参考视频');
  const realCandidates = realRows.filter(row => value(row['资源角色']) === '候选视频');
  if (!realReferences.length) issues.push(issue('real-reference', '缺少真实参考视频资源。'));
  if (!realCandidates.length) issues.push(issue('real-candidate', '缺少真实候选视频资源。'));
  if (!realRows.some(row => isLikelyRealUrl(row['视频链接']))) issues.push(issue('video-url', '缺少真实视频 URL。'));
  if (!realRows.every(row => hasVideoEvidence(row))) issues.push(issue('video-evidence', '真实视频资源必须至少有视频 URL、封面、ASR、帧图或正文证据之一。'));
  return {
    id: 'video',
    file,
    ready: issues.length === 0,
    counts: {
      rows: body.length,
      realRows: realRows.length,
      realReferenceRows: realReferences.length,
      realCandidateRows: realCandidates.length,
      realVideoUrlRows: realRows.filter(row => isLikelyRealUrl(row['视频链接'])).length
    },
    issues
  };
}

function renderReport(summary) {
  const lines = [
    '# 真实材料 Intake Readiness 审计',
    '',
    `- 生成时间：${summary.generatedAt}`,
    `- 是否可进入后续真实审计：${summary.acceptance.overallReady ? '是' : '否'}`,
    `- failureCount：${summary.failureCount}`,
    '',
    '## 文件',
    '',
    `- 历史数据：${summary.files.historyCsv}`,
    `- 人工复核：${summary.files.reviewCsv}`,
    `- 视频资源：${summary.files.videoCsv}`,
    '',
    '## 门禁',
    '',
    '| 门禁 | 状态 |',
    '| --- | --- |',
    `| 历史数据可进入 history:from-csv/history:audit | ${passFail(summary.history.ready)} |`,
    `| 视频资源可进入 video:resource-readiness | ${passFail(summary.video.ready)} |`,
    `| 人工复核可进入 review:metrics | ${passFail(summary.review.readyForReviewMetrics)} |`,
    `| 人工复核可进入 customer-effect:audit | ${passFail(summary.review.readyForCustomerEffectAudit)} |`,
    '',
    '## 统计',
    '',
    '| 区域 | 指标 | 数量 |',
    '| --- | --- | ---: |',
    `| 历史数据 | 行数 | ${summary.history.counts.rows} |`,
    `| 历史数据 | Brief 数 | ${summary.history.counts.briefCount} |`,
    `| 人工复核 | 行数 | ${summary.review.counts.rows} |`,
    `| 人工复核 | 有客户选择行 | ${summary.review.counts.rowsWithCustomerDecision} |`,
    `| 视频资源 | 真实参考视频 | ${summary.video.counts.realReferenceRows} |`,
    `| 视频资源 | 真实候选视频 | ${summary.video.counts.realCandidateRows} |`,
    '',
    '## 问题',
    '',
    summary.issues.length
      ? '| 区域 | 类型 | 说明 |\n| --- | --- | --- |\n' + summary.issues.map(item => `| ${item.section} | ${item.type} | ${escapeCell(item.message)} |`).join('\n')
      : '- 暂无问题。',
    '',
    '## 修复清单',
    '',
    summary.repairActions.length
      ? '| 优先级 | 负责人 | 文件 | 字段 | 修复动作 | 通过标准 |\n| ---: | --- | --- | --- | --- | --- |\n' + summary.repairActions.map(item => `| ${item.priority} | ${item.owner} | ${escapeCell(item.file)} | ${escapeCell(item.field)} | ${escapeCell(item.action)} | ${escapeCell(item.acceptance)} |`).join('\n')
      : '- 暂无修复动作。',
    '',
    '## 后续命令',
    '',
    '```powershell',
    'npm run history:from-csv -- --input <history-data-template.csv> --output <history-dataset目录>',
    'npm run history:audit -- --input <history-dataset目录> --output <历史审计输出目录> --strict',
    'npm run video:resource-readiness -- --input <video-resource-template.csv> --output <视频资源审计输出目录> --strict',
    'npm run review:metrics -- --input <manual-review-template.csv> --output <复核指标输出目录> --strict',
    'npm run customer-effect:audit -- --review-csv <manual-review-template.csv> --history-audit <historical-dataset-audit.json> --current-manual-supplement-count <本轮人工补号量> --output <客户效果输出目录> --strict',
    '```',
    '',
    '## 边界',
    '',
    ...summary.guardrails.map(item => `- ${item}`)
  ];
  return lines.join('\n');
}

function buildRepairActions({ issues, files }) {
  return issues.map((item, index) => {
    const spec = repairSpec(item);
    return {
      priority: index + 1,
      section: item.section,
      type: item.type,
      owner: spec.owner,
      file: spec.file === 'history' ? files.historyCsv : spec.file === 'review' ? files.reviewCsv : files.videoCsv,
      field: spec.field,
      action: spec.action,
      acceptance: spec.acceptance
    };
  });
}

function repairSpec(issueItem) {
  const key = `${issueItem.section}:${issueItem.type}`;
  const specs = {
    'history:min-briefs': ['商务', 'history', 'brief编号', '补齐至少 5 个真实历史 Brief，覆盖不同项目或不同客户需求。', 'historyReady=true，真实历史 Brief 数 >= 5。'],
    'history:placeholder': ['商务', 'history', '全表', '删除 example.com、示例、候选、待补等占位内容，替换为真实客户材料。', '预审不再出现 placeholder。'],
    'history:brief-text': ['商务', 'history', '客户原始Brief', '填写足够具体的客户原始需求，至少覆盖类目、人群、平台、内容方向或禁投项。', '每行客户原始 Brief 长度和信息量达标。'],
    'history:manual-baseline': ['商务', 'history', '历史人工补号量基线', '填写数字，表示历史同类需求人工额外补号数量。', '每行都能解析为非负数字。'],
    'history:reference': ['商务', 'history', '参考账号或视频', '填写真实参考账号主页或参考视频链接，不只写昵称或截图说明。', '每行都有真实可识别 URL。'],
    'history:manual-list': ['商务', 'history', '博主名称/主页链接', '填写人工最终名单中的真实博主名称和主页链接。', '每行博主名称非空且主页链接为真实 URL。'],
    'history:customer-decision': ['商务', 'history', '客户选择', '把待反馈替换为客户最终选中或客户拒绝。', '每行客户选择为客户选中或客户拒绝。'],
    'history:reject-reason': ['商务', 'history', '拒绝原因', '客户拒绝样本填写具体拒绝原因。', '所有拒绝样本拒绝原因非空。'],
    'review:empty': ['商务', 'review', '全表', '把本轮 AI 候选名单粘贴到人工复核模板。', '人工复核表至少有一行候选。'],
    'review:placeholder': ['商务', 'review', '全表', '删除 example.com、示例、候选、待补等占位内容，替换为真实候选和复核结果。', '预审不再出现 placeholder。'],
    'review:review-label': ['商务', 'review', '人工复核标签', '为每个候选填写人工复核标签。', '每行人工复核标签非空。'],
    'review:attribution': ['商务', 'review', '归因类型', '负样本填写需求解析错、隐性规则漏、召回关键词错、主页证据不足、视频证据误判、排序权重错或输出解释错。', '负样本归因覆盖率=100%。'],
    'review:customer-decision': ['商务', 'review', '客户选择', '补客户最终选中或客户拒绝记录。', '至少有客户最终选择，客户效果审计字段齐全。'],
    'review:current-manual-supplement': ['商务', 'review', '本轮人工补号量', '填写本轮 AI 输出后仍需人工补号的数量。', '至少一行本轮人工补号量为非负数字。'],
    'video:empty': ['商务/投放', 'video', '全表', '填写真实参考视频和真实候选视频资源。', '视频资源表至少有参考视频和候选视频。'],
    'video:placeholder': ['商务/投放', 'video', '全表', '删除 example.com、示例、候选、待确认或人工待补等占位内容。', '预审不再出现 placeholder。'],
    'video:real-reference': ['商务/投放', 'video', '资源角色/是否真实资源', '补至少一个真实参考视频，并将是否真实资源标为是。', '真实参考视频行数 >= 1。'],
    'video:real-candidate': ['商务/投放', 'video', '资源角色/是否真实资源', '补至少一个真实候选视频，并将是否真实资源标为是。', '真实候选视频行数 >= 1。'],
    'video:video-url': ['商务/投放', 'video', '视频链接', '补真实视频 URL；只有主页链接或接口 200 不算视频资源。', '真实资源中至少一行有真实视频 URL。'],
    'video:video-evidence': ['商务/投放', 'video', '封面链接/字幕或ASR文本/帧图链接/内容摘要', '为真实视频补封面、ASR、帧图或正文证据。', '真实资源至少有一类可审计证据。']
  };
  const fallback = specs[key] || ['技术/AI', issueItem.section === 'video' ? 'video' : issueItem.section === 'review' ? 'review' : 'history', issueItem.type, issueItem.message, '对应 issue 消失。'];
  return {
    owner: fallback[0],
    file: fallback[1],
    field: fallback[2],
    action: fallback[3],
    acceptance: fallback[4]
  };
}

function renderRepairCsv(actions) {
  const header = ['优先级', '负责人', '区域', '问题类型', '文件', '字段', '修复动作', '通过标准'];
  const rows = actions.map(item => [
    item.priority,
    item.owner,
    item.section,
    item.type,
    item.file,
    item.field,
    item.action,
    item.acceptance
  ]);
  return [header, ...rows].map(row => row.map(csvCell).join(',')).join('\n');
}

function csvCell(value) {
  const text = String(value ?? '');
  return /[",\r\n]/.test(text) ? `"${text.replace(/"/g, '""')}"` : text;
}

function safeReadCsv(reader, file) {
  try {
    if (!file || !fs.existsSync(file)) return { ok: false, header: [], body: [], error: `文件不存在：${file}` };
    const rows = reader(file);
    return { ok: true, header: rows.header || [], body: rows.body || [] };
  } catch (error) {
    return { ok: false, header: [], body: [], error: error.message };
  }
}

function headerStartsWith(header, required) {
  return required.every((key, index) => header[index] === key);
}

function hasPlaceholderRows(rows) {
  return rows.some(row => hasPlaceholderText(Object.values(row).join(' ')));
}

function hasPlaceholderText(text) {
  return /(example\.com|示例|候选博主|候选视频标题|待确认|人工待补|填写客户原始 Brief|暂时只有封面|内部鉴权|密钥)/i.test(String(text || ''));
}

function isLikelyRealReference(input) {
  return isLikelyRealUrl(input) && !hasPlaceholderText(input);
}

function isLikelyRealUrl(input) {
  const text = value(input);
  return /^https?:\/\/\S+/i.test(text) && !/example\.com/i.test(text);
}

function hasVideoEvidence(row) {
  return isLikelyRealUrl(row['视频链接']) ||
    isLikelyRealUrl(row['封面链接']) ||
    isLikelyRealUrl(row['帧图链接']) ||
    value(row['字幕或ASR文本']).length >= 12 ||
    value(row['内容摘要']).length >= 12;
}

function isYes(input) {
  return ['是', 'true', 'yes', '1'].includes(value(input).toLowerCase());
}

function issue(type, message) {
  return { type, message };
}

function value(input) {
  return String(input || '').trim();
}

function toNumberOrNull(input) {
  const text = value(input);
  if (!text) return null;
  const numeric = Number(text);
  return Number.isFinite(numeric) && numeric >= 0 ? numeric : null;
}

function unique(values) {
  return [...new Set(values.map(value).filter(Boolean))];
}

function passFail(value) {
  return value ? '通过' : '未通过';
}

function escapeCell(input) {
  return String(input || '').replace(/\|/g, '/').replace(/\r?\n/g, ' ');
}

function parseArgs(argv) {
  const args = {};
  for (let index = 0; index < argv.length; index += 1) {
    const raw = argv[index];
    if (!raw.startsWith('--')) continue;
    const key = raw.slice(2).replace(/-([a-z])/g, (_, char) => char.toUpperCase());
    const next = argv[index + 1];
    if (!next || next.startsWith('--')) args[key] = true;
    else {
      args[key] = next;
      index += 1;
    }
  }
  return args;
}

function withBom(text) {
  return `\uFEFF${text}`;
}

if (require.main === module) {
  try {
    main();
  } catch (error) {
    console.error(error && error.stack ? error.stack : String(error));
    process.exit(1);
  }
}

module.exports = {
  buildIntakeReadinessSummary,
  renderReport
};