import fs from "node:fs";
import path from "node:path";

export const DOCX_REAL_WORLD_CORPUS = Object.freeze([
  {
    id: "changqing-tender-requirements",
    label: "长庆油田智能健康技术服务招标技术要求",
    markdownPath: "tmp/长庆油田智能健康技术服务招标技术要求.md",
    requiredFeatures: ["inline-code", "table-math-alignment"]
  },
  {
    id: "health-data-schema",
    label: "运动健康类和营养饮食数据结构升级",
    markdownPath: "tmp/运动健康类和营养饮食数据数据结构升级.md",
    requiredFeatures: ["json-code-indentation"],
    codeIndentationProbe: {
      rootLine: "{",
      nestedLineIncludes: '"heart_rate_daily"',
      minimumIndentPt: 4
    }
  },
  {
    id: "m4n-table-design",
    label: "M4N 全量表设计模块化讲解",
    markdownPath: "tmp/数据中台项目周报_2026_W30_附件_M4N全量表设计模块化讲解.md",
    requiredFeatures: ["mermaid", "long-document"]
  }
]);

export function resolveDocxRealWorldCorpus(repositoryDirectory) {
  return DOCX_REAL_WORLD_CORPUS.map((entry) => ({
    ...entry,
    absoluteMarkdownPath: path.resolve(
      repositoryDirectory,
      entry.markdownPath
    ),
    byteLength: fs.existsSync(path.resolve(repositoryDirectory, entry.markdownPath))
      ? fs.statSync(path.resolve(repositoryDirectory, entry.markdownPath)).size
      : 0
  }));
}

export function sortDocxRealWorldCorpusBySize(repositoryDirectory) {
  return resolveDocxRealWorldCorpus(repositoryDirectory).sort(
    (first, second) =>
      first.byteLength - second.byteLength ||
      first.id.localeCompare(second.id, "en")
  );
}

export function validateDocxRealWorldCorpus(repositoryDirectory) {
  const corpus = sortDocxRealWorldCorpusBySize(repositoryDirectory);
  const failures = [];
  for (const entry of corpus) {
    if (!fs.existsSync(entry.absoluteMarkdownPath)) {
      failures.push(`${entry.id} 缺少 Markdown：${entry.markdownPath}`);
      continue;
    }
    const source = fs.readFileSync(entry.absoluteMarkdownPath, "utf8");
    if (!source.trim()) {
      failures.push(`${entry.id} Markdown 为空`);
    }
    if (
      entry.requiredFeatures.includes("json-code-indentation") &&
      !/```json\s*[\s\S]*?\n[\t ]+\S/gu.test(source)
    ) {
      failures.push(`${entry.id} 缺少带缩进的 JSON 围栏探针`);
    }
    if (
      entry.requiredFeatures.includes("inline-code") &&
      !/(^|[^`])`[^`\n]+`([^`]|$)/mu.test(source)
    ) {
      failures.push(`${entry.id} 缺少行内代码探针`);
    }
    if (
      entry.requiredFeatures.includes("table-math-alignment") &&
      !/^\|.*\$[^$]+\$.*\|/mu.test(source)
    ) {
      failures.push(`${entry.id} 缺少表格公式对齐探针`);
    }
  }
  return { corpus, failures };
}

export function summarizeDocxReleaseGateMatrix({
  baselineCaseCount,
  corpusCaseCounts
}) {
  const corpusCaseCount = Object.values(corpusCaseCounts).reduce(
    (sum, value) => sum + value,
    0
  );
  return {
    baselineCaseCount,
    corpusDocumentCount: Object.keys(corpusCaseCounts).length,
    corpusCaseCounts,
    corpusCaseCount,
    totalCaseCount: baselineCaseCount + corpusCaseCount
  };
}