/**
 * Shared low-level invocation of python/docling_parser.py.
 * Used by PdfParser (as one link in its docling → pymupdf → pdf-parse fallback
 * chain) and by DocxParser (as its only strategy — PyMuPDF/pdf-parse cannot
 * read DOCX, so there is no fallback to chain for that format).
 */

import path from 'path';
import type { ParsedDocument, ParserOptions } from '../../types';
import { executePythonScript } from '../../utils/pythonRunner';
import { ragDebug } from '../../utils/ragLogger.js';

export async function parseWithDoclingScript(
  filePath: string,
  format: 'pdf' | 'docx',
  logLabel: string,
  options?: ParserOptions
): Promise<ParsedDocument> {
  ragDebug(logLabel, 'Using docling for professional-grade parsing...');

  const scriptPath = path.join(process.cwd(), 'python', 'docling_parser.py');
  const extractTables = options?.extractTables !== false ? 'true' : 'false';
  const args = [filePath, extractTables];
  if (options?.imageOutputDir) args.push(options.imageOutputDir);

  const result = await executePythonScript<{
    success: boolean;
    content: string;
    metadata: {
      format: string;
      strategy: string;
      title?: string;
      page_count: number;
      has_tables: boolean;
      char_count: number;
      image_count: number | null;
    };
    error?: string;
  }>(scriptPath, args, {
    timeout: 120000,
    onStatus: options?.onStatus,
  });

  if (!result.success || !result.data?.success) {
    throw new Error(result.error || result.data?.error || 'Docling parsing failed');
  }

  const data = result.data;

  return {
    content: data.content,
    metadata: {
      format,
      strategy: 'docling',
      title: data.metadata.title,
      pageCount: data.metadata.page_count,
      hasTables: data.metadata.has_tables,
      imageCount: data.metadata.image_count ?? undefined,
    },
  };
}

/**
 * Shared low-level invocation of python/docling_parser.py.
 * Used by PdfParser (as one link in its docling → pymupdf → pdf-parse fallback
 * chain) and by DocxParser (as its only strategy — PyMuPDF/pdf-parse cannot
 * read DOCX, so there is no fallback to chain for that format).
 */

import path from 'path';
import type { ParsedDocument, ParserOptions } from '../../types';
import { executePythonScript } from '../../utils/pythonRunner';
import { ragDebug } from '../../utils/ragLogger.js';

export async function parseWithDoclingScript(
  filePath: string,
  format: 'pdf' | 'docx',
  logLabel: string,
  options?: ParserOptions
): Promise<ParsedDocument> {
  ragDebug(logLabel, 'Using docling for professional-grade parsing...');

  const scriptPath = path.join(process.cwd(), 'python', 'docling_parser.py');
  const extractTables = options?.extractTables !== false ? 'true' : 'false';
  const args = [filePath, extractTables];
  if (options?.imageOutputDir) args.push(options.imageOutputDir);

  const result = await executePythonScript<{
    success: boolean;
    content: string;
    metadata: {
      format: string;
      strategy: string;
      title?: string;
      page_count: number;
      has_tables: boolean;
      char_count: number;
      image_count: number | null;
    };
    error?: string;
  }>(scriptPath, args, {
    timeout: 120000,
    onStatus: options?.onStatus,
  });

  if (!result.success || !result.data?.success) {
    throw new Error(result.error || result.data?.error || 'Docling parsing failed');
  }

  const data = result.data;

  return {
    content: data.content,
    metadata: {
      format,
      strategy: 'docling',
      title: data.metadata.title,
      pageCount: data.metadata.page_count,
      hasTables: data.metadata.has_tables,
      imageCount: data.metadata.image_count ?? undefined,
    },
  };
}