import os
from docling.document_converter import DocumentConverter, PdfFormatOption
from docling.datamodel.pipeline_options import (
    PdfPipelineOptions, RapidOcrOptions,
    TableStructureOptions, TableFormerMode,
)
from docling.datamodel.base_models import InputFormat
from docling.datamodel.accelerator_options import AcceleratorDevice, AcceleratorOptions

os.environ["OMP_NUM_THREADS"] = "4"

pipeline_options = PdfPipelineOptions(
    do_ocr=True,
    ocr_options=RapidOcrOptions(), # CPU-friendly, ~6s/page
    images_scale=2.0, # Higher resolution = better number accuracy
    do_table_structure=True,
    table_structure_options=TableStructureOptions(
        mode=TableFormerMode.ACCURATE, # Most precise for financial tables
        do_cell_matching=False, # Important for scanned pages
    ),
    generate_page_images=False,
    generate_picture_images=False,
    do_code_enrichment=False,
    do_formula_enrichment=False,
    accelerator_options=AcceleratorOptions(num_threads=4, device=AcceleratorDevice.CPU),
)

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

result = converter.convert("laporan_keuangan.pdf")
doc = result.document

# Export for AI agent
md = doc.export_to_markdown(traverse_pictures=True) # traverse_pictures=True is required
doc.save_as_json("output/document.json") # Lossless backup

import fitz # PyMuPDF

BOUNDARY_KEYWORDS = ["INVOICE", "FAKTUR", "LAPORAN KEUANGAN", "BANK STATEMENT", "RECEIPT"]

def detect_boundaries(pdf_path):
    doc = fitz.open(pdf_path)
    documents = []
    current_doc = {"type": "unknown", "start_page": 0}
    for page_num in range(len(doc)):
        text = doc[page_num].get_text().upper()
        for keyword in BOUNDARY_KEYWORDS:
            if keyword in text:
                current_doc["end_page"] = page_num - 1
                if page_num > 0:
                    documents.append(current_doc)
                current_doc = {"type": keyword.lower(), "start_page": page_num}
                break
    current_doc["end_page"] = len(doc) - 1
    documents.append(current_doc)
    doc.close()
    return documents