Merge pull request #118 from alexander-densley/main

add density and height to python as well

Author: annapo23

py_zerox/pyzerox/core/zerox.py

@@ -7,6 +7,7 @@
 import aiofiles
 import aiofiles.os as async_os
 import asyncio
+from ..constants import PDFConversionDefaultOptions
 
 # Package Imports
 from ..processor import (
@@ -26,6 +27,8 @@ async def zerox(
     cleanup: bool = True,
     concurrency: int = 10,
     file_path: Optional[str] = "",
+    image_density: int = PDFConversionDefaultOptions.DPI,
+    image_height: tuple[Optional[int], int] = PDFConversionDefaultOptions.SIZE,
     maintain_format: bool = False,
     model: str = "gpt-4o-mini",
     output_dir: Optional[str] = None,
@@ -130,7 +133,7 @@ async def zerox(
                                                  **subset_pdf_create_kwargs)
 
         # Convert the file to a series of images, below function returns a list of image paths in page order
-        images = await convert_pdf_to_images(local_path=local_path, temp_dir=temp_directory)
+        images = await convert_pdf_to_images(image_density=image_density, image_height=image_height, local_path=local_path, temp_dir=temp_directory)
 
         if maintain_format:
             for image in images:

py_zerox/pyzerox/processor/pdf.py

@@ -11,14 +11,14 @@
 from ..models import litellmmodel
 
 
-async def convert_pdf_to_images(local_path: str, temp_dir: str) -> List[str]:
+async def convert_pdf_to_images(image_density: int, image_height: tuple[Optional[int], int], local_path: str, temp_dir: str) -> List[str]:
     """Converts a PDF file to a series of images in the temp_dir. Returns a list of image paths in page order."""
     options = {
         "pdf_path": local_path,
         "output_folder": temp_dir,
-        "dpi": PDFConversionDefaultOptions.DPI,
+        "dpi": image_density,
         "fmt": PDFConversionDefaultOptions.FORMAT,
-        "size": PDFConversionDefaultOptions.SIZE,
+        "size": image_height,
         "thread_count": PDFConversionDefaultOptions.THREAD_COUNT,
         "use_pdftocairo": PDFConversionDefaultOptions.USE_PDFTOCAIRO,
         "paths_only": True,