historical-dataset-audit.js 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238
  1. #!/usr/bin/env node
  2. const fs = require('fs');
  3. const path = require('path');
  4. function main() {
  5. const args = parseArgs(process.argv.slice(2));
  6. const input = args.input || args.dir || process.env.TIHAO_HISTORY_DATASET || '';
  7. if (!input) throw new Error('Usage: node scripts/historical-dataset-audit.js --input <history-dataset-dir> [--output <dir>] [--min-briefs 5] [--strict]');
  8. const inputPath = path.resolve(input);
  9. const outputDir = path.resolve(args.output || process.env.TIHAO_HISTORY_AUDIT_OUTPUT || path.join(inputPath, 'audit-output'));
  10. const minBriefs = Number(args.minBriefs || process.env.TIHAO_HISTORY_MIN_BRIEFS || 5);
  11. const strict = Boolean(args.strict || process.env.TIHAO_HISTORY_AUDIT_STRICT === 'true');
  12. const records = loadRecords(inputPath);
  13. const summary = auditRecords({ inputPath, records, minBriefs });
  14. fs.mkdirSync(outputDir, { recursive: true });
  15. const jsonPath = path.join(outputDir, 'historical-dataset-audit.json');
  16. const reportPath = path.join(outputDir, 'historical-dataset-audit.md');
  17. fs.writeFileSync(jsonPath, JSON.stringify(summary, null, 2), 'utf8');
  18. fs.writeFileSync(reportPath, withBom(renderReport(summary)), 'utf8');
  19. console.log(JSON.stringify({
  20. input: inputPath,
  21. outputDir,
  22. json: jsonPath,
  23. report: reportPath,
  24. briefCount: summary.briefCount,
  25. readyForLongRun: summary.acceptance.readyForLongRun,
  26. missingCriticalCount: summary.missingCriticalCount
  27. }, null, 2));
  28. if (strict && !summary.acceptance.readyForLongRun) process.exitCode = 2;
  29. }
  30. function loadRecords(inputPath) {
  31. if (!fs.existsSync(inputPath)) throw new Error(`History dataset not found: ${inputPath}`);
  32. const stat = fs.statSync(inputPath);
  33. const files = stat.isDirectory()
  34. ? fs.readdirSync(inputPath).filter(isHistoryRecordFile).map(name => path.join(inputPath, name))
  35. : [inputPath];
  36. return files.sort().map(file => {
  37. try {
  38. return { file, record: JSON.parse(fs.readFileSync(file, 'utf8').replace(/^\uFEFF/, '')) };
  39. } catch (error) {
  40. return { file, record: null, parseError: error.message };
  41. }
  42. });
  43. }
  44. function isHistoryRecordFile(name) {
  45. return name.endsWith('.json') &&
  46. !name.endsWith('-summary.json') &&
  47. name !== 'historical-dataset-audit.json' &&
  48. name !== 'history-dataset-from-csv-summary.json';
  49. }
  50. function auditRecords({ inputPath, records, minBriefs }) {
  51. const items = records.map(item => auditRecord(item));
  52. const validItems = items.filter(item => item.parseOk);
  53. const categories = unique(validItems.map(item => item.category).filter(Boolean));
  54. const briefCount = validItems.length;
  55. const withReference = validItems.filter(item => item.hasReference).length;
  56. const withManualFinalList = validItems.filter(item => item.hasManualFinalList).length;
  57. const withCustomerDecision = validItems.filter(item => item.hasCustomerDecision).length;
  58. const withRejectionReason = validItems.filter(item => item.hasRejectionReason).length;
  59. const withManualSupplementBaseline = validItems.filter(item => item.hasManualSupplementBaseline).length;
  60. const missingCriticalCount = items.filter(item => item.missingCritical.length).length;
  61. const acceptance = {
  62. parseOk: items.every(item => item.parseOk),
  63. minBriefsMet: briefCount >= minBriefs,
  64. allHaveBriefText: validItems.every(item => item.hasBriefText),
  65. allHaveManualFinalList: validItems.every(item => item.hasManualFinalList),
  66. allHaveCustomerDecision: validItems.every(item => item.hasCustomerDecision),
  67. allHaveFeedbackReason: validItems.every(item => item.hasRejectionReason || item.allSelected),
  68. allHaveManualSupplementBaseline: validItems.every(item => item.hasManualSupplementBaseline),
  69. categoryCoverageMet: categories.length >= Math.min(3, minBriefs),
  70. referenceCoverageMeasured: withReference > 0
  71. };
  72. acceptance.readyForLongRun = acceptance.parseOk &&
  73. acceptance.minBriefsMet &&
  74. acceptance.allHaveBriefText &&
  75. acceptance.allHaveManualFinalList &&
  76. acceptance.allHaveCustomerDecision &&
  77. acceptance.allHaveFeedbackReason &&
  78. acceptance.categoryCoverageMet;
  79. acceptance.readyForCustomerEffectProof = acceptance.readyForLongRun &&
  80. acceptance.allHaveManualSupplementBaseline;
  81. return {
  82. inputPath,
  83. generatedAt: new Date().toISOString(),
  84. minBriefs,
  85. briefCount,
  86. categoryCount: categories.length,
  87. categories,
  88. withReference,
  89. withManualFinalList,
  90. withCustomerDecision,
  91. withRejectionReason,
  92. withManualSupplementBaseline,
  93. missingCriticalCount,
  94. acceptance,
  95. items
  96. };
  97. }
  98. function auditRecord({ file, record, parseError }) {
  99. if (!record) {
  100. return {
  101. file,
  102. parseOk: false,
  103. parseError,
  104. missingCritical: ['JSON 无法解析']
  105. };
  106. }
  107. const briefText = record.briefText || record.brief || record.originalBrief || '';
  108. const manualFinalList = array(record.manualFinalList || record.finalCreatorList || record.manualList);
  109. const customerFeedback = array(record.customerFeedback || record.customerDecisions || record.customerSelection);
  110. const manualSupplementBaseline = record.manualSupplementBaseline ?? record.manualSupplementCount ?? record.manualAddedCount ?? record.metrics?.manualSupplementCount;
  111. const hasCustomerDecision = customerFeedback.some(item => /选中|拒绝|通过|未选|淘汰|selected|rejected/i.test(String(item.finalDecision || item.decision || item.status || item['客户选择'] || ''))) ||
  112. manualFinalList.some(item => /选中|拒绝|通过|未选|淘汰|selected|rejected/i.test(String(item.finalDecision || item.customerDecision || item['客户选择'] || '')));
  113. const allSelected = customerFeedback.length > 0 && customerFeedback.every(item => /选中|通过|selected/i.test(String(item.finalDecision || item.decision || item.status || item['客户选择'] || '')));
  114. const hasRejectionReason = customerFeedback.some(item => item.reason || item.feedbackReason || item.rejectReason || item['拒绝原因']) ||
  115. manualFinalList.some(item => item.reason || item.feedbackReason || item.rejectReason || item['拒绝原因']);
  116. const hasReference = array(record.referenceLinks).length > 0 || array(record.referenceAccounts).length > 0 || array(record.referenceVideos).length > 0;
  117. const hasManualSupplementBaseline = Number.isFinite(Number(manualSupplementBaseline)) && Number(manualSupplementBaseline) >= 0;
  118. const missingCritical = [];
  119. if (!record.id && !record.name) missingCritical.push('缺少 id/name');
  120. if (!briefText) missingCritical.push('缺少客户原始 Brief');
  121. if (!manualFinalList.length) missingCritical.push('缺少人工最终提报名单');
  122. if (!hasCustomerDecision) missingCritical.push('缺少客户最终选中/拒绝记录');
  123. if (!hasRejectionReason && !allSelected) missingCritical.push('缺少拒绝原因或商务复核原因');
  124. return {
  125. file,
  126. parseOk: true,
  127. id: record.id || record.name || path.basename(file, '.json'),
  128. name: record.name || record.id || '',
  129. category: record.category || record.input?.category || '',
  130. hasBriefText: Boolean(briefText),
  131. hasReference,
  132. manualFinalListCount: manualFinalList.length,
  133. customerFeedbackCount: customerFeedback.length,
  134. manualSupplementBaseline: hasManualSupplementBaseline ? Number(manualSupplementBaseline) : null,
  135. hasManualFinalList: manualFinalList.length > 0,
  136. hasCustomerDecision,
  137. hasRejectionReason,
  138. hasManualSupplementBaseline,
  139. allSelected,
  140. missingCritical
  141. };
  142. }
  143. function renderReport(summary) {
  144. const lines = [
  145. '# 历史提号数据集验收报告',
  146. '',
  147. `- 输入目录:${summary.inputPath}`,
  148. `- 生成时间:${summary.generatedAt}`,
  149. `- Brief 数:${summary.briefCount}`,
  150. `- 最低要求:${summary.minBriefs}`,
  151. `- 类目数:${summary.categoryCount}`,
  152. `- 是否可用于长期优化长跑:${summary.acceptance.readyForLongRun ? '是' : '否'}`,
  153. `- 是否可用于客户效果证明:${summary.acceptance.readyForCustomerEffectProof ? '是' : '否'}`,
  154. '',
  155. '## 验收门禁',
  156. '',
  157. '| 门禁 | 状态 |',
  158. '| --- | --- |',
  159. `| JSON 可解析 | ${passFail(summary.acceptance.parseOk)} |`,
  160. `| Brief 数达到要求 | ${passFail(summary.acceptance.minBriefsMet)} |`,
  161. `| 每个样本有原始 Brief | ${passFail(summary.acceptance.allHaveBriefText)} |`,
  162. `| 每个样本有人工最终名单 | ${passFail(summary.acceptance.allHaveManualFinalList)} |`,
  163. `| 每个样本有客户选中/拒绝记录 | ${passFail(summary.acceptance.allHaveCustomerDecision)} |`,
  164. `| 拒绝样本有原因 | ${passFail(summary.acceptance.allHaveFeedbackReason)} |`,
  165. `| 每个样本有人工补号量基线 | ${passFail(summary.acceptance.allHaveManualSupplementBaseline)} |`,
  166. `| 类目覆盖足够 | ${passFail(summary.acceptance.categoryCoverageMet)} |`,
  167. '',
  168. '## 样本明细',
  169. '',
  170. '| 样本 | 类目 | 参考账号/视频 | 人工名单 | 客户选择 | 拒绝/复核原因 | 人工补号基线 | 缺口 |',
  171. '| --- | --- | --- | ---: | --- | --- | ---: | --- |',
  172. ...summary.items.map(item => `| ${escapeCell(item.id || path.basename(item.file))} | ${escapeCell(item.category || '')} | ${item.hasReference ? '有' : '无'} | ${item.manualFinalListCount || 0} | ${item.hasCustomerDecision ? '有' : '无'} | ${item.hasRejectionReason || item.allSelected ? '有' : '无'} | ${item.hasManualSupplementBaseline ? item.manualSupplementBaseline : '缺失'} | ${escapeCell((item.missingCritical || []).join(';') || '无')} |`),
  173. '',
  174. '## 使用说明',
  175. '',
  176. '- 只有本报告显示“可用于长期优化长跑=是”,才适合拿来验证客户选中率 30%/50%。',
  177. '- 没有客户最终选择字段时,只能做商务复核通过率,不能宣称客户选中率。',
  178. '- 没有拒绝原因时,无法做失败样本归因,也无法稳定升级团队规则。',
  179. '- 没有人工补号量基线时,不能证明“人工补号量减少 50%”。'
  180. ];
  181. return lines.join('\n');
  182. }
  183. function parseArgs(argv) {
  184. const args = {};
  185. for (let i = 0; i < argv.length; i += 1) {
  186. const raw = argv[i];
  187. if (!raw.startsWith('--')) continue;
  188. const key = raw.slice(2).replace(/-([a-z])/g, (_, char) => char.toUpperCase());
  189. const next = argv[i + 1];
  190. if (!next || next.startsWith('--')) args[key] = true;
  191. else {
  192. args[key] = next;
  193. i += 1;
  194. }
  195. }
  196. return args;
  197. }
  198. function array(value) {
  199. return Array.isArray(value) ? value : [];
  200. }
  201. function unique(values) {
  202. return [...new Set(values.filter(Boolean).map(String))];
  203. }
  204. function passFail(value) {
  205. return value ? '通过' : '未通过';
  206. }
  207. function escapeCell(value) {
  208. return String(value || '').replace(/\|/g, '/').replace(/\n/g, ' ');
  209. }
  210. function withBom(text) {
  211. return `\uFEFF${text}`;
  212. }
  213. if (require.main === module) {
  214. try {
  215. main();
  216. } catch (error) {
  217. console.error(error && error.stack ? error.stack : String(error));
  218. process.exit(1);
  219. }
  220. }
  221. module.exports = {
  222. auditRecords,
  223. loadRecords
  224. };