temp save, completely untested code

Author: sebastianMindee

mindee/error/__init__.py

@@ -7,3 +7,5 @@
     MindeeHTTPServerError,
     handle_error,
 )
+from mindee.error.mindee_image_error import MindeeImageError
+from mindee.error.mindee_pdf_error import MindeePDFError

mindee/error/mindee_image_error.py

@@ -0,0 +1,2 @@
+class MindeeImageError(RuntimeError):
+    """An exception relating to errors during image operations."""

mindee/error/mindee_pdf_error.py

@@ -0,0 +1,2 @@
+class MindeePDFError(RuntimeError):
+    """An exception relating to errors during PDF operations."""

mindee/extraction/__init__.py

@@ -1,6 +1,6 @@
 from mindee.extraction.common.extracted_image import ExtractedImage
 from mindee.extraction.common.image_extractor import (
-    attach_image_as_new_file,
+    attach_images_as_new_file,
     extract_multiple_images_from_source,
 )
 from mindee.extraction.multi_receipts_extractor import multi_receipts_extractor

mindee/extraction/common/__init__.py

@@ -1,5 +1,5 @@
 from mindee.extraction.common.extracted_image import ExtractedImage
 from mindee.extraction.common.image_extractor import (
-    attach_image_as_new_file,
+    attach_images_as_new_file,
     extract_multiple_images_from_source,
 )

mindee/extraction/common/image_extractor.py

@@ -11,35 +11,34 @@
 from mindee.input.sources import BytesInput, LocalInputSource
 
 
-def attach_image_as_new_file(  # type: ignore
-    input_buffer: BinaryIO,
+def attach_images_as_new_file(  # type: ignore
+    input_buffer_list: List[BinaryIO],
 ) -> pdfium.PdfDocument:
     """
-    Attaches an image as a new page in a PdfDocument object.
+    Attaches a list of images as new pages in a PdfDocument object.
 
-    :param input_buffer: Input buffer.
+    :param input_buffer_list: List of images, represented as buffers.
     :return: A PdfDocument handle.
     """
-    # Create a new page in the PdfDocument
-    input_buffer.seek(0)
-    image = Image.open(input_buffer)
-    image.convert("RGB")
-    image_buffer = io.BytesIO()
-    image.save(image_buffer, format="JPEG")
-
     pdf = pdfium.PdfDocument.new()
-
-    image_pdf = pdfium.PdfImage.new(pdf)
-    image_pdf.load_jpeg(image_buffer)
-    width, height = image_pdf.get_size()
-
-    matrix = pdfium.PdfMatrix().scale(width, height)
-    image_pdf.set_matrix(matrix)
-
-    page = pdf.new_page(width, height)
-    page.insert_obj(image_pdf)
-    page.gen_content()
-    image.close()
+    for input_buffer in input_buffer_list:
+        input_buffer.seek(0)
+        image = Image.open(input_buffer)
+        image.convert("RGB")
+        image_buffer = io.BytesIO()
+        image.save(image_buffer, format="JPEG")
+
+        image_pdf = pdfium.PdfImage.new(pdf)
+        image_pdf.load_jpeg(image_buffer)
+        width, height = image_pdf.get_size()
+
+        matrix = pdfium.PdfMatrix().scale(width, height)
+        image_pdf.set_matrix(matrix)
+
+        page = pdf.new_page(width, height)
+        page.insert_obj(image_pdf)
+        page.gen_content()
+        image.close()
     return pdf
 
 
@@ -160,4 +159,4 @@ def load_pdf_doc(input_file: LocalInputSource) -> pdfium.PdfDocument:  # type: i
         input_file.file_object.seek(0)
         return pdfium.PdfDocument(input_file.file_object)
 
-    return attach_image_as_new_file(input_file.file_object)
+    return attach_images_as_new_file([input_file.file_object])

mindee/image_operations/__init__.py

@@ -0,0 +1,33 @@
+import io
+from typing import Union
+
+from PIL import Image
+
+
+def compress_image(
+    image_buffer: bytes,
+    quality: int = 85,
+    max_width: Union[int, float, None] = None,
+    max_height: Union[int, float, None] = None,
+) -> bytes:
+    """
+    Compresses an image with the given parameters.
+
+    :param image_buffer: Buffer representation of an image.
+    :param quality: Quality to apply to the image (JPEG compression).
+    :param max_width: Maximum bound for the width.
+    :param max_height: Maximum bound for the height.
+    :return:
+    """
+    with Image.open(io.BytesIO(image_buffer)) as img:
+        original_width, original_height = img.size
+        max_width = max_width or original_width
+        max_height = max_height or original_height
+        if max_width or max_height:
+            img.thumbnail((int(max_width), int(max_height)), Image.Resampling.LANCZOS)
+
+        output_buffer = io.BytesIO()
+        img.save(output_buffer, format="JPEG", quality=quality, optimize=True)
+
+        compressed_image = output_buffer.getvalue()
+    return compressed_image

mindee/image_operations/image_compressor.py

@@ -0,0 +1,30 @@
+from dataclasses import dataclass
+from typing import Tuple
+
+
+@dataclass
+class PDFCharData:
+    """Data class representing character data."""
+
+    char: str
+    """The character."""
+    left: int
+    """Left bound."""
+    right: int
+    """Right bound."""
+    top: int
+    """Top bound."""
+    bottom: int
+    """Bottom bound."""
+    font_name: str
+    """The font name."""
+    font_size: int
+    """The font size in pt."""
+    font_weight: int
+    """The font weight."""
+    font_flags: int
+    """The font flags."""
+    font_stroke_color: Tuple[int, int, int, int]
+    """RGBA representation of the font's stroke color."""
+    font_fill_color: Tuple[int, int, int, int]
+    """RGBA representation of the font's fill color."""