38 lines
1.4 KiB
Python
38 lines
1.4 KiB
Python
from docling.document_converter import (
|
|
DocumentConverter,
|
|
PdfFormatOption,
|
|
PowerpointFormatOption,
|
|
WordFormatOption,
|
|
)
|
|
from docling.datamodel.pipeline_options import PdfPipelineOptions
|
|
from docling.datamodel.base_models import InputFormat
|
|
from utils.path_helpers import patch_python_docx_templates
|
|
|
|
|
|
class DoclingService:
|
|
def __init__(self):
|
|
# Patch python-docx template path resolution before initializing converter
|
|
# This is safe to call in any environment (Docker, development, PyInstaller)
|
|
patch_python_docx_templates()
|
|
|
|
self.pipeline_options = PdfPipelineOptions()
|
|
self.pipeline_options.do_ocr = False
|
|
|
|
self.converter = DocumentConverter(
|
|
allowed_formats=[InputFormat.PPTX, InputFormat.PDF, InputFormat.DOCX],
|
|
format_options={
|
|
InputFormat.DOCX: WordFormatOption(
|
|
pipeline_options=self.pipeline_options,
|
|
),
|
|
InputFormat.PPTX: PowerpointFormatOption(
|
|
pipeline_options=self.pipeline_options,
|
|
),
|
|
InputFormat.PDF: PdfFormatOption(
|
|
pipeline_options=self.pipeline_options,
|
|
),
|
|
},
|
|
)
|
|
|
|
def parse_to_markdown(self, file_path: str) -> str:
|
|
result = self.converter.convert(file_path)
|
|
return result.document.export_to_markdown()
|