aiquery2/web/client/src/sections/aiTenderDocument/utils/mermaidSanitizer.js

// mermaidSanitizer.js
export function validateAndFixMermaid(raw) {
   const warnings = [];

   if (!raw || !String(raw).trim()) {
      return { ok: false, fixedSource: "", warnings };
   }

   let text = String(raw)
      // HTML 实体
      .replace(/&/g, "&")
      .replace(/&lt;/g, "<")
      .replace(/&gt;/g, ">")
      .replace(/&quot;/g, '"')
      .replace(/&#39;/g, "'")
      // 零宽字符
      .replace(/[\u200b-\u200f\u2028\u202f\u205f\u2060\ufeff]/g, "")
      // 全角符号
      .replace(/[“”]/g, '"')
      .replace(/[‘’]/g, "'")
      .replace(/\u00a0/g, " ")
      // HTML 换行
      .replace(/<br\s*\/?>/gi, "\n")
      // CRLF
      .replace(/\r\n?/g, "\n")
      .trim();

   // 清理残留的 HTML 标签，但保留 mermaid 语法可能用到的字符
   // 仅移除常见的块级和内联标签，避免误伤 mermaid 节点文本（如 A[<Label>]）
   text = text
      .replace(
         /<\/?(div|p|span|br|pre|code|table|tr|td|th|thead|tbody|ul|ol|li|h[1-6]|img)\b[^>]*>/gi,
         ""
      )
      .trim();

   // 移除可能混入的 Markdown 图片语法
   text = text
      .split("\n")
      .filter((line) => !/^\s*!\[[^\]]*]\([^)]+\)\s*$/.test(line))
      .join("\n")
      .trim();

   // 修复被压成一行的语句，避免多个语句连在一起导致语法错误
   const injectLineBreaks = (value) => {
      let out = String(value || "");
      out = out.replace(/\s+(?=section\b)/gi, "\n");
      out = out.replace(
         /([^\n])\s+(?=[A-Za-z0-9_]+\s*(-->|--|==>|==|<--|<-->)) /g,
         "$1\n"
      );
      out = out.replace(
         /([^\n])\s+(?=[A-Za-z0-9_]+\s*(-->|--|==>|==|<--|<-->))/g,
         "$1\n"
      );
      return out;
   };
   text = injectLineBreaks(text);

   // 去掉代码围栏
   text = text
      .replace(/^\s*```mermaid\s*/i, "")
      .replace(/^\s*```/gm, "")
      .replace(/^\s*~~~mermaid\s*/i, "")
      .replace(/^\s*~~~/gm, "")
      .trim();

   const lines = text.split("\n").filter(Boolean);
   if (!lines.length) {
      return { ok: false, fixedSource: "", warnings };
   }

   // 自动为包含特殊字符（如 |）的节点文本添加引号
   // 处理 [] [[ ]]
   text = text.replace(
      /([A-Za-z0-9_\u4e00-\u9fa5]+)\s*(\[+)([^"\]\n]*\|[^"\]\n]*)(\]+)/g,
      (match, id, open, content, close) => {
         warnings.push(`自动为包含特殊字符的节点添加引号: ${id}`);
         return `${id}${open}"${content}"${close}`;
      }
   );
   // 处理 () (( ))
   text = text.replace(
      /([A-Za-z0-9_\u4e00-\u9fa5]+)\s*(\(+)([^"\)\n]*\|[^"\)\n]*)(\)+)/g,
      (match, id, open, content, close) => {
         warnings.push(`自动为包含特殊字符的节点添加引号: ${id}`);
         return `${id}${open}"${content}"${close}`;
      }
   );
   // 处理 {} {{ }}
   text = text.replace(
      /([A-Za-z0-9_\u4e00-\u9fa5]+)\s*(\{+)([^"\}\n]*\|[^"\}\n]*)(\}+)/g,
      (match, id, open, content, close) => {
         warnings.push(`自动为包含特殊字符的节点添加引号: ${id}`);
         return `${id}${open}"${content}"${close}`;
      }
   );

   // 修复非法边标签
   text = text.replace(
      /(\S+)\s*--\s*([^|\n-]+?)\s*-->\s*(\S+)/g,
      (_, from, label, to) => {
         warnings.push(`修复非法边标签: "${label.trim()}"`);
         return `${from} --|${label.trim()}|--> ${to}`;
      }
   );

   // subgraph 中文名加引号
   text = text.replace(/^\s*subgraph\s+(.+)$/gim, (_, name) => {
      const n = name.trim();
      if (/^["'].*["']$/.test(n)) return `subgraph ${n}`;
      if (/[\u4e00-\u9fa5\s]/.test(n)) {
         warnings.push(`subgraph 名称加引号: ${n}`);
         return `subgraph "${n}"`;
      }
      return `subgraph ${n}`;
   });

   return {
      ok: true,
      fixedSource: text.trim(),
      warnings,
   };
}

const normalizeEscapedNewlines = (raw) => {
   const source = String(raw || "");
   let result = "";
   let bracketDepth = 0;
   let parenthesisDepth = 0;
   let braceDepth = 0;
   let pipeLabel = false;

   for (let index = 0; index < source.length; index += 1) {
      const current = source[index];
      const next = source[index + 1];
      const inLabel =
         bracketDepth > 0 ||
         parenthesisDepth > 0 ||
         braceDepth > 0 ||
         pipeLabel;

      if (current === "\\" && next === "n") {
         result += inLabel ? "<br/>" : "\n";
         index += 1;
         continue;
      }

      if (current === "\n") {
         result += inLabel ? "<br/>" : "\n";
         continue;
      }

      if (current === "[") bracketDepth += 1;
      if (current === "]" && bracketDepth > 0) bracketDepth -= 1;
      if (current === "(") parenthesisDepth += 1;
      if (current === ")" && parenthesisDepth > 0) parenthesisDepth -= 1;
      if (current === "{") braceDepth += 1;
      if (current === "}" && braceDepth > 0) braceDepth -= 1;
      if (
         current === "|" &&
         bracketDepth === 0 &&
         parenthesisDepth === 0 &&
         braceDepth === 0
      ) {
         pipeLabel = !pipeLabel;
      }

      result += current;
   }

   return result;
};

const normalizeMermaidRenderSource = (raw) => {
   let text = String(raw || "")
      .replace(/\r\n?/g, "\n")
      .replace(/<br\s*\/?>/gi, "\n")
      .replace(/[“”]/g, '"')
      .replace(/[‘’]/g, "'")
      .replace(/[：]/g, ":")
      .replace(/\u00a0/g, " ");

   text = normalizeEscapedNewlines(text)
      .replace(/^\s*>\s?/gm, "")
      .replace(/^\s*code\b[:\s]*/gim, "")
      .replace(/^\s*```(?:mermaid)?\s*/gim, "")
      .replace(/^\s*~~~(?:mermaid)?\s*/gim, "")
      .replace(/^\s*subgraph\s+(.+)$/gim, (_, name) => {
         const trimmedName = String(name || "").trim();
         if (!trimmedName || /^["'].*["']$/.test(trimmedName)) {
            return `subgraph ${trimmedName}`.trim();
         }
         if (/[\s\u4e00-\u9fa5]/.test(trimmedName)) {
            return `subgraph "${trimmedName}"`;
         }
         return `subgraph ${trimmedName}`;
      })
      .trim();

   return text;
};

/**
 * 功能：生成 Mermaid 节点视图可依次尝试的渲染源码。
 * 使用场景：兼容接口返回的转义换行、中文 subgraph 和旧 Mermaid 源码。
 *
 * 入参：原始 Mermaid 源码。
 *
 * 返回：去重后的候选源码数组，首项为常规清洗结果。
 *
 * 注意：调用方应依次尝试候选源码，不能在首个候选失败后直接清空预览。
 */
export function getMermaidRenderCandidates(raw) {
   const { fixedSource } = validateAndFixMermaid(raw);
   const candidates = [fixedSource, normalizeMermaidRenderSource(fixedSource)];

   return [...new Set(candidates.map((item) => String(item || "").trim()))].filter(
      Boolean
   );
}