big-rag

import * as fs from "fs";

/**
 * Parse BSL (Business Script Language) files for 1C:Enterprise
 * BSL files contain source code in Russian (1C language)
 */
export async function parseBSL(filePath: string): Promise<string> {
  try {
    // BSL files are text files, read as UTF-8
    // If UTF-8 fails, the error will be caught and handled
    const content = await fs.promises.readFile(filePath, "utf-8");
    
    // BSL files should preserve line breaks for code readability
    // Remove excessive whitespace but keep structure
    const cleaned = content
      .replace(/\r\n?/g, "\n")  // Normalize line endings
      .replace(/[ \t]+/g, " ")  // Collapse horizontal whitespace
      .replace(/\n{3,}/g, "\n\n")  // Keep paragraph separation
      .trim();
    
    return cleaned;
  } catch (error) {
    console.error(`Error parsing BSL file ${filePath}:`, error);
    return "";
  }
}