tihao-ai/claude-code-tihao-sourcing/scripts/local-seed-material-index.js

#!/usr/bin/env node
const fs = require('fs');
const path = require('path');
const zlib = require('zlib');

const ROOT = path.resolve(__dirname, '..');
const DEFAULT_OUTPUT = path.join(ROOT, 'outputs', 'local-seed-material-index-latest');

function main() {
  const args = parseArgs(process.argv.slice(2));
  const workspaceRoot = path.resolve(args.workspaceRoot || path.join(ROOT, '..'));
  const outputsDir = path.resolve(args.outputs || path.join(ROOT, 'outputs'));
  const outputDir = path.resolve(args.output || DEFAULT_OUTPUT);
  const summary = buildLocalSeedMaterialIndex({ root: ROOT, workspaceRoot, outputsDir });

  fs.mkdirSync(outputDir, { recursive: true });
  const summaryPath = path.join(outputDir, 'local-seed-material-index-summary.json');
  const reportPath = path.join(outputDir, 'local-seed-material-index.md');
  const csvPath = path.join(outputDir, 'local-seed-material-index.csv');
  fs.writeFileSync(summaryPath, JSON.stringify(summary, null, 2), 'utf8');
  fs.writeFileSync(reportPath, withBom(renderReport(summary)), 'utf8');
  fs.writeFileSync(csvPath, withBom(renderCsv(summary)), 'utf8');

  console.log(JSON.stringify({
    outputDir,
    summary: summaryPath,
    report: reportPath,
    csv: csvPath,
    materialCount: summary.materialCount,
    existingMaterialCount: summary.existingMaterialCount,
    canCloseProofGap: summary.canCloseProofGap
  }, null, 2));

  if (args.strict && summary.directCustomerProof !== false) process.exitCode = 1;
}

function buildLocalSeedMaterialIndex({ root, workspaceRoot, outputsDir }) {
  const specs = [
    {
      id: 'dha-brief-workbook',
      title: 'DHA Brief 原始工作簿',
      file: path.join(workspaceRoot, 'dha_brief.xlsx'),
      role: 'brief_seed',
      usableFor: ['补真实 Brief 模板的需求字段', '识别 DHA/母婴/小红书账号类型线索', '作为候选池召回种子'],
      missingForProof: [
        '至少 5 个真实历史 Brief',
        '人工最终提报名单',
        '客户最终选中/拒绝记录',
        '拒绝原因或商务复核标签',
        '历史人工补号量基线'
      ]
    },
    {
      id: 'dna-brief-workbook-legacy-name',
      title: 'DNA Brief 旧命名工作簿',
      file: path.join(workspaceRoot, 'dna_brief.xlsx'),
      role: 'brief_seed_legacy_name',
      usableFor: ['兼容旧交接命名，确认是否另有 Brief 种子材料'],
      missingForProof: ['若文件不存在，不影响当前 DHA 种子材料索引']
    },
    {
      id: 'dha-poc-recommendation-workbook',
      title: 'AI 提号 DHA 小红书推荐名单 PoC',
      file: path.join(workspaceRoot, 'output', 'dha-tihao-poc', 'AI提号-DHA小红书推荐名单-PoC.xlsx'),
      role: 'candidate_seed',
      usableFor: ['候选池种子', '商务复核展示样例', '参考账号分析口径校准'],
      missingForProof: [
        '客户选择/拒绝结果',
        '负样本归因',
        '本轮人工补号量',
        '可审计的生成策略和 live 采集证明'
      ]
    },
    {
      id: 'dha-poc-social-reference-workbook',
      title: 'AI 提号 DHA 社媒参考分析版',
      file: path.join(workspaceRoot, 'output', 'dha-tihao-poc', 'AI提号-DHA小红书推荐名单-社媒参考分析版.xlsx'),
      role: 'reference_analysis_seed',
      usableFor: ['参考账号基准分析', '候选池补充', '视频/主页证据字段设计参考'],
      missingForProof: [
        '真实候选视频资源',
        '真实视频 A/B live proof',
        '商务复核结果',
        '客户最终选择'
      ]
    }
  ];

  const materials = specs.map(spec => inspectMaterial({ root, workspaceRoot, spec }));
  const pipelineArtifacts = inspectPipelineArtifacts({ root, outputsDir });
  const liveAutomationCandidates = inspectLiveAutomationCandidates({ root, outputsDir });
  const existingMaterialCount = materials.filter(item => item.exists).length;
  const workbookCount = materials.filter(item => item.exists && item.extension === '.xlsx').length;
  const promisingLiveAggregateCount = liveAutomationCandidates.filter(item => item.liveAutomationPass).length;

  return {
    generatedAt: new Date().toISOString(),
    root,
    workspaceRoot,
    outputsDir,
    proofLevel: 'not_business_proof',
    directCustomerProof: false,
    canCloseProofGap: false,
    complete: false,
    materialCount: materials.length,
    existingMaterialCount,
    workbookCount,
    promisingLiveAggregateCount,
    materials,
    pipelineArtifacts,
    liveAutomationCandidates,
    guardrails: [
      '本索引只把本地已有 Brief/PoC/历史输出做成可发现的种子材料台账，不关闭 proof-gap。',
      'DHA Brief 和 PoC 推荐表可用于补真实模板、召回候选和校准口径，但不能证明客户效果。',
      '历史 pipeline 失败产物、模板数据、currentManualSupplementCount=0 或候选名单本身不能证明人工补号量下降。',
      'live overnight aggregate 即使是真实自动化证据，也必须被 proof-gap:closure 在当前审计上下文中接受后，才可关闭 live/provider 缺口。',
      '不得在本索引或后续文档中写入 sessionToken、Authorization、模型 token 或 npm token。'
    ],
    nextActions: [
      {
        owner: '商务',
        title: '把 DHA Brief 种子转写进统一历史数据模板',
        action: '只把真实客户 Brief 字段、参考账号/视频线索和账号类型要求转写进 outputs/data-intake-pack-latest/history-data-template.csv；不足 5 个 Brief 时继续保持未完成。',
        acceptance: 'intake:readiness historyReady=true，且 history:audit readyForCustomerEffectProof=true。'
      },
      {
        owner: '商务/投放',
        title: '从 PoC 候选里补真实候选视频资源',
        action: '为候选账号补至少一个真实候选视频 URL、封面、ASR/字幕、帧图或正文证据，并写入 video-resource-template.csv。',
        acceptance: 'video:resource-readiness readyForVideoAbPreflight=true，realCandidateRows>=1。'
      },
      {
        owner: '技术/AI',
        title: '只把本索引用作补证入口，不作为完成证明',
        action: '补齐真实材料后重跑 intake:readiness、video:resource-readiness、proof-gap:closure、optimization:completion 和 round:refresh。',
        acceptance: 'proof-gap-closure.openCount=0 且 optimization-completion.readyForClaim=true 后，才进入长期目标完成判断。'
      }
    ]
  };
}

function inspectMaterial({ root, workspaceRoot, spec }) {
  const exists = fs.existsSync(spec.file);
  const extension = path.extname(spec.file).toLowerCase();
  const base = {
    id: spec.id,
    title: spec.title,
    role: spec.role,
    path: relPath(root, spec.file),
    workspacePath: relPath(workspaceRoot, spec.file),
    exists,
    extension,
    proofLevel: 'not_business_proof',
    directCustomerProof: false,
    canCloseProofGap: false,
    usableFor: spec.usableFor,
    missingForProof: spec.missingForProof
  };
  if (!exists) {
    return {
      ...base,
      status: 'missing',
      summary: '本地未发现该文件；只记录旧交接线索。'
    };
  }
  const stat = fs.statSync(spec.file);
  const workbook = extension === '.xlsx' ? readWorkbookSummary(spec.file) : null;
  const allText = workbook ? workbook.searchText : '';
  const signals = detectSignals(allText);
  return {
    ...base,
    status: 'available_seed',
    size: stat.size,
    updatedAt: stat.mtime.toISOString(),
    workbook: workbook ? {
      sheetCount: workbook.sheets.length,
      sheets: workbook.sheets.map(sheet => ({
        name: sheet.name,
        rowCount: sheet.rowCount,
        columnCount: sheet.columnCount,
        headers: sheet.headers,
        sampleRows: sheet.sampleRows
      })),
      secretLikeHitCount: workbook.secretLikeHitCount,
      parseWarnings: workbook.parseWarnings
    } : null,
    signals,
    summary: renderMaterialSummary({ spec, signals, workbook })
  };
}

function inspectPipelineArtifacts({ root, outputsDir }) {
  const pipelineDir = path.join(outputsDir, 'optimization-pipeline-latest');
  const pipeline = readJson(path.join(pipelineDir, 'optimization-pipeline-summary.json'));
  const history = readJson(path.join(pipelineDir, 'history-audit', 'historical-dataset-audit.json'));
  const customer = readJson(path.join(pipelineDir, 'customer-effect-audit', 'customer-effect-summary.json'));
  return {
    proofLevel: 'not_business_proof',
    directCustomerProof: false,
    canCloseProofGap: false,
    pipeline: pipeline ? {
      path: relPath(root, path.join(pipelineDir, 'optimization-pipeline-summary.json')),
      readyForClaim: Boolean(pipeline.readyForClaim),
      passCount: Number(pipeline.counts?.pass || 0),
      failCount: Number(pipeline.counts?.fail || 0),
      currentManualSupplementCount: pipeline.inputs?.currentManualSupplementCount ?? null,
      warning: '旧 pipeline 只能说明编排链路曾运行；包含模板/示例输入或 currentManualSupplementCount=0 时不能证明人工补号量下降。'
    } : null,
    historyAudit: history ? {
      path: relPath(root, path.join(pipelineDir, 'history-audit', 'historical-dataset-audit.json')),
      briefCount: Number(history.briefCount || 0),
      readyForCustomerEffectProof: Boolean(history.acceptance?.readyForCustomerEffectProof),
      readyForLongRun: Boolean(history.acceptance?.readyForLongRun),
      categoryCoverageMet: Boolean(history.acceptance?.categoryCoverageMet),
      warning: '历史审计未达到 readyForCustomerEffectProof=true 前，不能支撑客户效果证明。'
    } : null,
    customerEffect: customer ? {
      path: relPath(root, path.join(pipelineDir, 'customer-effect-audit', 'customer-effect-summary.json')),
      overallPass: Boolean(customer.acceptance?.overallPass),
      customerSelectedRateMeasured: Boolean(customer.acceptance?.customerSelectedRateMeasured),
      referenceCustomerSelectedRateMeasured: Boolean(customer.acceptance?.referenceCustomerSelectedRateMeasured),
      historyReadyForCustomerEffectProof: Boolean(customer.acceptance?.historyReadyForCustomerEffectProof),
      missingEvidence: customer.missingEvidence || [],
      warning: '客户效果审计 overallPass=false 或选择率未测量时，不能证明提号效果达标。'
    } : null
  };
}

function inspectLiveAutomationCandidates({ root, outputsDir }) {
  if (!fs.existsSync(outputsDir)) return [];
  return fs.readdirSync(outputsDir, { withFileTypes: true })
    .filter(entry => entry.isDirectory() && entry.name.startsWith('overnight-quality-'))
    .map(entry => {
      const dir = path.join(outputsDir, entry.name);
      const aggregatePath = path.join(dir, 'aggregate-summary.json');
      const aggregate = readJson(aggregatePath);
      if (!aggregate) return null;
      const failedGateCount = Number(aggregate.acceptance?.failedGateCount ?? aggregate.failedGateCount ?? 0);
      const liveAutomationPass = Boolean(aggregate.manifest?.liveEnabled) &&
        aggregate.acceptance?.overallPass === true &&
        Number(aggregate.failureCount || 0) === 0 &&
        failedGateCount === 0;
      return {
        id: entry.name,
        path: relPath(root, aggregatePath),
        updatedAt: fs.statSync(dir).mtime.toISOString(),
        liveEnabled: Boolean(aggregate.manifest?.liveEnabled),
        overallPass: Boolean(aggregate.acceptance?.overallPass),
        failureCount: Number(aggregate.failureCount || 0),
        failedGateCount,
        runCount: Number(aggregate.runCount || 0),
        fixtureCount: Number(aggregate.fixtureCount || 0),
        variantCount: Number(aggregate.variantCount || 0),
        liveAutomationPass,
        proofLevel: liveAutomationPass ? 'real_evidence_for_live_automation' : 'not_business_proof',
        directCustomerProof: false,
        canCloseCustomerEffectProof: false,
        canCloseProofGapNow: false,
        missingForClosure: liveAutomationPass
          ? ['需要 proof-gap:closure 在当前审计上下文中明确接受该 aggregate', '仍不能替代客户选择和人工补号量证明']
          : ['liveEnabled=true', 'overallPass=true', 'failureCount=0', 'failedGateCount=0']
      };
    })
    .filter(Boolean)
    .sort((a, b) => String(b.updatedAt).localeCompare(String(a.updatedAt)))
    .slice(0, 8);
}

function readWorkbookSummary(file) {
  const parseWarnings = [];
  try {
    const zip = readZipEntries(fs.readFileSync(file));
    const sharedStrings = parseSharedStrings(readZipText(zip, 'xl/sharedStrings.xml') || '');
    const workbookXml = readZipText(zip, 'xl/workbook.xml') || '';
    const relsXml = readZipText(zip, 'xl/_rels/workbook.xml.rels') || '';
    const sheets = parseWorkbookSheets(workbookXml, relsXml);
    if (sheets.length === 0) {
      Object.keys(zip)
        .filter(name => /^xl\/worksheets\/sheet\d+\.xml$/.test(name))
        .sort()
        .forEach((name, index) => sheets.push({ name: `Sheet${index + 1}`, target: name }));
    }
    const sheetSummaries = sheets.map(sheet => parseSheet({
      name: sheet.name,
      xml: readZipText(zip, sheet.target) || '',
      sharedStrings
    }));
    const searchText = sheetSummaries
      .flatMap(sheet => [sheet.name, ...sheet.allValues])
      .join('\n');
    return {
      sheets: sheetSummaries,
      searchText,
      secretLikeHitCount: countSecretLike(searchText),
      parseWarnings
    };
  } catch (error) {
    return {
      sheets: [],
      searchText: '',
      secretLikeHitCount: 0,
      parseWarnings: [`xlsx 解析失败：${error.message}`]
    };
  }
}

function readZipEntries(buffer) {
  const eocdOffset = findEndOfCentralDirectory(buffer);
  if (eocdOffset < 0) throw new Error('未找到 ZIP central directory');
  const totalEntries = buffer.readUInt16LE(eocdOffset + 10);
  const centralDirectoryOffset = buffer.readUInt32LE(eocdOffset + 16);
  const entries = {};
  let offset = centralDirectoryOffset;
  for (let index = 0; index < totalEntries; index += 1) {
    if (buffer.readUInt32LE(offset) !== 0x02014b50) throw new Error('ZIP central directory 损坏');
    const method = buffer.readUInt16LE(offset + 10);
    const compressedSize = buffer.readUInt32LE(offset + 20);
    const fileNameLength = buffer.readUInt16LE(offset + 28);
    const extraLength = buffer.readUInt16LE(offset + 30);
    const commentLength = buffer.readUInt16LE(offset + 32);
    const localHeaderOffset = buffer.readUInt32LE(offset + 42);
    const name = buffer.slice(offset + 46, offset + 46 + fileNameLength).toString('utf8');
    const localNameLength = buffer.readUInt16LE(localHeaderOffset + 26);
    const localExtraLength = buffer.readUInt16LE(localHeaderOffset + 28);
    const dataStart = localHeaderOffset + 30 + localNameLength + localExtraLength;
    const compressed = buffer.slice(dataStart, dataStart + compressedSize);
    let data;
    if (method === 0) data = compressed;
    else if (method === 8) data = zlib.inflateRawSync(compressed);
    else data = Buffer.alloc(0);
    entries[name] = data;
    offset += 46 + fileNameLength + extraLength + commentLength;
  }
  return entries;
}

function findEndOfCentralDirectory(buffer) {
  const min = Math.max(0, buffer.length - 66000);
  for (let offset = buffer.length - 22; offset >= min; offset -= 1) {
    if (buffer.readUInt32LE(offset) === 0x06054b50) return offset;
  }
  return -1;
}

function readZipText(entries, name) {
  const item = entries[name];
  return item ? item.toString('utf8') : '';
}

function parseSharedStrings(xml) {
  if (!xml) return [];
  return matchAll(xml, /<(?:\w+:)?si\b[\s\S]*?<\/(?:\w+:)?si>/g).map(item => {
    const textParts = [];
    const regex = /<(?:\w+:)?t\b[^>]*>([\s\S]*?)<\/(?:\w+:)?t>/g;
    let match;
    while ((match = regex.exec(item)) !== null) textParts.push(decodeXml(match[1]));
    return textParts.join('');
  });
}

function parseWorkbookSheets(workbookXml, relsXml) {
  const relMap = {};
  for (const rel of matchAll(relsXml, /<Relationship\b([^>]*?)\/>/g)) {
    const attrs = parseAttrs(rel);
    if (attrs.Id && attrs.Target) {
      relMap[attrs.Id] = normalizeWorkbookTarget(attrs.Target);
    }
  }
  return matchAll(workbookXml, /<(?:\w+:)?sheet\b([^>]*?)\/>/g).map((tag, index) => {
    const attrs = parseAttrs(tag);
    const rid = attrs['r:id'] || attrs.id || '';
    return {
      name: attrs.name || `Sheet${index + 1}`,
      target: relMap[rid] || `xl/worksheets/sheet${index + 1}.xml`
    };
  });
}

function normalizeWorkbookTarget(target) {
  const value = String(target || '').replace(/\\/g, '/');
  if (value.startsWith('/')) return path.posix.normalize(value.slice(1));
  return path.posix.normalize(path.posix.join('xl', value));
}

function parseSheet({ name, xml, sharedStrings }) {
  const rows = [];
  let maxColumn = 0;
  for (const rowXml of matchAll(xml, /<(?:\w+:)?row\b[^>]*>([\s\S]*?)<\/(?:\w+:)?row>/g)) {
    const values = [];
    for (const cellXml of matchAll(rowXml, /<(?:\w+:)?c\b([^>]*?)>([\s\S]*?)<\/(?:\w+:)?c>/g)) {
      const attrs = parseAttrs(cellXml);
      const body = cellXml;
      const colIndex = columnIndex((attrs.r || '').replace(/\d+/g, ''));
      const value = parseCellValue(body, attrs, sharedStrings);
      if (value !== '') {
        values[colIndex >= 0 ? colIndex : values.length] = sanitizeCell(value);
        maxColumn = Math.max(maxColumn, (colIndex >= 0 ? colIndex : values.length - 1) + 1);
      }
    }
    if (values.some(value => String(value || '').trim())) rows.push(values);
  }
  const headers = (rows[0] || []).map(value => sanitizeCell(value)).filter(Boolean).slice(0, 20);
  const sampleRows = rows.slice(1, 6).map(row => row.map(value => sanitizeCell(value)).slice(0, 12));
  const allValues = rows.flat().map(value => String(value || '')).filter(Boolean);
  return {
    name,
    rowCount: rows.length,
    columnCount: maxColumn,
    headers,
    sampleRows,
    allValues
  };
}

function parseCellValue(body, attrs, sharedStrings) {
  if (attrs.t === 'inlineStr') {
    const inline = /<(?:\w+:)?is\b[\s\S]*?<(?:\w+:)?t\b[^>]*>([\s\S]*?)<\/(?:\w+:)?t>[\s\S]*?<\/(?:\w+:)?is>/.exec(body);
    return inline ? decodeXml(inline[1]) : '';
  }
  const value = /<(?:\w+:)?v\b[^>]*>([\s\S]*?)<\/(?:\w+:)?v>/.exec(body);
  if (!value) return '';
  const raw = decodeXml(value[1]);
  if (attrs.t === 's') return sharedStrings[Number(raw)] || '';
  if (attrs.t === 'b') return raw === '1' ? 'TRUE' : 'FALSE';
  return raw;
}

function detectSignals(text) {
  const urls = text.match(/https?:\/\/[^\s'"<>]+/g) || [];
  return {
    mentionsDha: /DHA/i.test(text),
    mentionsXiaohongshu: /小红书|xiaohongshu/i.test(text),
    mentionsReference: /参考|reference/i.test(text),
    mentionsCustomerDecision: /客户.*(选中|拒绝|通过)|最终选择|customer/i.test(text),
    mentionsManualSupplement: /人工补号|补号量|manual supplement/i.test(text),
    urlCount: urls.length,
    sampleUrls: unique(urls.map(sanitizePublicUrl)).slice(0, 5)
  };
}

function renderMaterialSummary({ spec, signals, workbook }) {
  const parts = [];
  if (signals.mentionsDha) parts.push('包含 DHA 线索');
  if (signals.mentionsXiaohongshu) parts.push('包含小红书线索');
  if (signals.mentionsReference) parts.push('包含参考账号/参考内容线索');
  if (signals.urlCount > 0) parts.push(`发现 URL ${signals.urlCount} 个`);
  if (workbook?.sheets?.length) parts.push(`工作表 ${workbook.sheets.length} 个`);
  return parts.length ? parts.join('；') : `${spec.title} 可作为本地种子材料，但不能直接证明业务效果。`;
}

function renderReport(summary) {
  return [
    '# 本地种子材料索引',
    '',
    `- 生成时间：${summary.generatedAt}`,
    `- proofLevel：${summary.proofLevel}`,
    `- directCustomerProof：${summary.directCustomerProof}`,
    `- canCloseProofGap：${summary.canCloseProofGap}`,
    `- 本地材料：${summary.existingMaterialCount}/${summary.materialCount}`,
    `- 可疑 live 自动化候选：${summary.promisingLiveAggregateCount}`,
    '',
    '## 边界',
    '',
    ...summary.guardrails.map(item => `- ${item}`),
    '',
    '## 本地材料',
    '',
    '| ID | 状态 | 路径 | 可用于 | 缺失证明 | 摘要 |',
    '| --- | --- | --- | --- | --- | --- |',
    ...summary.materials.map(item => tableRow([
      item.id,
      item.status,
      item.path,
      item.usableFor.join('；'),
      item.missingForProof.join('；'),
      item.summary || ''
    ])),
    '',
    '## 工作簿概览',
    '',
    ...summary.materials.filter(item => item.workbook).flatMap(item => [
      `### ${item.title}`,
      '',
      `- 文件：${item.path}`,
      `- secretLikeHitCount：${item.workbook.secretLikeHitCount}`,
      '',
      '| Sheet | 行数 | 列数 | 表头 |',
      '| --- | ---: | ---: | --- |',
      ...item.workbook.sheets.map(sheet => tableRow([
        sheet.name,
        sheet.rowCount,
        sheet.columnCount,
        sheet.headers.join(' / ')
      ])),
      ''
    ]),
    '## 旧 Pipeline 产物边界',
    '',
    `- pipeline.readyForClaim：${summary.pipelineArtifacts.pipeline?.readyForClaim ?? 'missing'}`,
    `- pipeline.failCount：${summary.pipelineArtifacts.pipeline?.failCount ?? 'missing'}`,
    `- history.readyForCustomerEffectProof：${summary.pipelineArtifacts.historyAudit?.readyForCustomerEffectProof ?? 'missing'}`,
    `- customerEffect.overallPass：${summary.pipelineArtifacts.customerEffect?.overallPass ?? 'missing'}`,
    '',
    '## Live 自动化候选',
    '',
    '| 输出 | liveEnabled | overallPass | failureCount | failedGateCount | proofLevel | 当前可关闭缺口 |',
    '| --- | ---: | ---: | ---: | ---: | --- | --- |',
    ...summary.liveAutomationCandidates.map(item => tableRow([
      item.path,
      item.liveEnabled,
      item.overallPass,
      item.failureCount,
      item.failedGateCount,
      item.proofLevel,
      item.canCloseProofGapNow
    ])),
    '',
    '## 下一步',
    '',
    '| 负责人 | 动作 | 验收 |',
    '| --- | --- | --- |',
    ...summary.nextActions.map(item => tableRow([item.owner, `${item.title}：${item.action}`, item.acceptance])),
    ''
  ].join('\n');
}

function renderCsv(summary) {
  const rows = [
    ['类型', 'ID', '状态', '路径', 'proofLevel', 'directCustomerProof', 'canCloseProofGap', '可用于', '缺失证明', '摘要'],
    ...summary.materials.map(item => [
      'local-material',
      item.id,
      item.status,
      item.path,
      item.proofLevel,
      item.directCustomerProof,
      item.canCloseProofGap,
      item.usableFor.join('；'),
      item.missingForProof.join('；'),
      item.summary || ''
    ]),
    ...summary.liveAutomationCandidates.map(item => [
      'live-automation-candidate',
      item.id,
      item.liveAutomationPass ? 'pass' : 'not_passed',
      item.path,
      item.proofLevel,
      item.directCustomerProof,
      item.canCloseProofGapNow,
      'live/provider 自动化证据候选',
      item.missingForClosure.join('；'),
      `runCount=${item.runCount}`
    ])
  ];
  return rows.map(row => row.map(csvCell).join(',')).join('\n');
}

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

function parseAttrs(tag) {
  const attrs = {};
  for (const match of matchAll(tag, /([:\w-]+)="([^"]*)"/g)) {
    const [, key, value] = /([:\w-]+)="([^"]*)"/.exec(match) || [];
    if (key) attrs[key] = decodeXml(value || '');
  }
  return attrs;
}

function matchAll(text, regex) {
  return Array.from(String(text || '').matchAll(regex)).map(match => match[0]);
}

function columnIndex(column) {
  if (!column) return -1;
  let value = 0;
  for (const char of column.toUpperCase()) value = value * 26 + (char.charCodeAt(0) - 64);
  return value - 1;
}

function decodeXml(text) {
  return String(text || '')
    .replace(/&lt;/g, '<')
    .replace(/&gt;/g, '>')
    .replace(/&amp;/g, '&')
    .replace(/&quot;/g, '"')
    .replace(/&apos;/g, "'");
}

function sanitizeCell(value) {
  const text = String(value || '').replace(/\s+/g, ' ').trim();
  const withoutTracking = sanitizeUrlsInText(text);
  if (isSecretLike(withoutTracking)) return '[REDACTED_SECRET_LIKE_VALUE]';
  return withoutTracking.slice(0, 240);
}

function sanitizeUrlsInText(text) {
  return String(text || '').replace(/https?:\/\/[^\s'"<>]+/g, match => sanitizePublicUrl(match));
}

function sanitizePublicUrl(value) {
  const text = String(value || '').trim();
  if (!text) return '';
  if (isSecretLike(text) && !/^https?:\/\//i.test(text)) return '[REDACTED_SECRET_LIKE_VALUE]';
  try {
    const parsed = new URL(text);
    parsed.search = '';
    parsed.hash = '';
    return parsed.toString();
  } catch (_) {
    return text
      .replace(/([?&])(?:xsec_token|access_token|refresh_token|session_token|sessionToken|api_key|apikey|token)=[^&#\s]+/gi, '$1')
      .replace(/[?&]+$/, '');
  }
}

function isSecretLike(text) {
  return /Authorization\s*[:=]\s*Bearer/i.test(text) ||
    /sessionToken\s*[:=]/i.test(text) ||
    /(?:xsec_token|access_token|refresh_token|session_token|api_key|apikey|token)=[^&#\s]+/i.test(text) ||
    /sk-[A-Za-z0-9_-]{10,}/.test(text) ||
    /npm_[A-Za-z0-9]{10,}/.test(text);
}

function countSecretLike(text) {
  return String(text || '').split(/\s+/).filter(isSecretLike).length;
}

function unique(values) {
  return Array.from(new Set(values.filter(Boolean)));
}

function readJson(file) {
  try {
    if (!fs.existsSync(file)) return null;
    return JSON.parse(fs.readFileSync(file, 'utf8'));
  } catch (_) {
    return null;
  }
}

function relPath(root, file) {
  const relative = path.relative(root, file).replace(/\\/g, '/');
  return relative || '.';
}

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

function tableRow(values) {
  return `| ${values.map(value => escapeCell(value)).join(' | ')} |`;
}

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

function csvCell(value) {
  return `"${String(value ?? '').replace(/"/g, '""')}"`;
}

if (require.main === module) main();

module.exports = {
  buildLocalSeedMaterialIndex,
  readWorkbookSummary,
  sanitizeCell,
  sanitizePublicUrl,
  sanitizeUrlsInText,
  isSecretLike
};