# pyright: ignore
#!/usr/bin/env python3
"""
Test regular pipeline only (no VLM) to see if markdown export works.
"""

from pathlib import Path
from docling.document_converter import DocumentConverter, PdfFormatOption
from docling.datamodel.base_models import InputFormat
from docling.datamodel.pipeline_options import PdfPipelineOptions, TableFormerMode

# Test file
pdf_path = Path("output/debug_pages/TMUS_992_Q423_page_017.pdf")

print(f"Testing REGULAR pipeline only on: {pdf_path}")
print("=" * 50)

# Regular pipeline configuration
regular_options = PdfPipelineOptions()
regular_options.do_table_structure = True
regular_options.table_structure_options.do_cell_matching = True
regular_options.table_structure_options.mode = TableFormerMode.ACCURATE
regular_options.do_ocr = True
regular_options.images_scale = 2.0

format_options = PdfFormatOption(pipeline_options=regular_options)

converter = DocumentConverter(
    format_options={InputFormat.PDF: format_options}
)

print("Converting with REGULAR pipeline...")
result = converter.convert(str(pdf_path))

print("
Conversion complete!")
print(f"Document has {len(result.document.pages)} pages")
print(f"Document has {len(result.document.tables)} tables")

# Export to markdown
markdown = result.document.export_to_markdown()
print(f"
Markdown length: {len(markdown)} characters")

# Save output
output_path = Path("test_regular_output.md")
with open(output_path, "w") as f:
    f.write(markdown)
print(f"Output saved to: {output_path}")

# Show preview
print("
First 2000 chars of markdown:")
print(markdown[:2000])