KnowledgeBase/server/src/common/json-utils.ts

/**
 * Safely parses JSON from a string, handling markdown code blocks and leading/trailing text.
 */
export function safeParseJson<T = any>(text: string): T | null {
  if (!text) return null;

  let jsonStr = text.trim();

  // 1. Try to extract JSON from markdown code blocks if they exist
  // Matches ```json ... ``` or just ``` ... ```
  const codeBlockRegex = /```(?:json)?\s*([\s\S]*?)\s*```/i;
  const match = jsonStr.match(codeBlockRegex);

  if (match && match[1]) {
    jsonStr = match[1].trim();
  } else {
    // 2. If no markdown block, try to find the start and end of JSON characters
    // This handles cases where the AI adds introductory or concluding text outside the block
    const firstOpenBrace = jsonStr.indexOf('{');
    const firstOpenBracket = jsonStr.indexOf('[');
    
    let startIndex = -1;
    if (firstOpenBrace !== -1 && (firstOpenBracket === -1 || firstOpenBrace < firstOpenBracket)) {
      startIndex = firstOpenBrace;
    } else if (firstOpenBracket !== -1) {
      startIndex = firstOpenBracket;
    }

    if (startIndex !== -1) {
      const lastCloseBrace = jsonStr.lastIndexOf('}');
      const lastCloseBracket = jsonStr.lastIndexOf(']');
      const endIndex = Math.max(lastCloseBrace, lastCloseBracket);
      
      if (endIndex !== -1 && endIndex > startIndex) {
        jsonStr = jsonStr.substring(startIndex, endIndex + 1);
      }
    }
  }

  try {
    return JSON.parse(jsonStr) as T;
  } catch (error) {
    console.error('[safeParseJson] Failed to parse JSON:', error);
    console.error('[safeParseJson] Original text:', text);
    console.error('[safeParseJson] Extracted string:', jsonStr);
    return null;
  }
}