✨ add support for image and pdf compression (#297)

Author: sebastianMindee

examples/auto_invoice_splitter_extraction_example.py

@@ -1,6 +1,6 @@
 from mindee import Client
 from mindee.extraction.pdf_extractor import PdfExtractor
-from mindee.input.sources import PathInput
+from mindee.input.sources.path_input import PathInput
 from mindee.product.invoice.invoice_v4 import InvoiceV4
 from mindee.product.invoice_splitter.invoice_splitter_v1 import InvoiceSplitterV1
 

mindee/client.py

@@ -7,14 +7,12 @@
 from mindee.input import WorkflowOptions
 from mindee.input.local_response import LocalResponse
 from mindee.input.page_options import PageOptions
-from mindee.input.sources import (
-    Base64Input,
-    BytesInput,
-    FileInput,
-    LocalInputSource,
-    PathInput,
-    UrlInputSource,
-)
+from mindee.input.sources.base_64_input import Base64Input
+from mindee.input.sources.bytes_input import BytesInput
+from mindee.input.sources.file_input import FileInput
+from mindee.input.sources.local_input_source import LocalInputSource
+from mindee.input.sources.path_input import PathInput
+from mindee.input.sources.url_input_source import UrlInputSource
 from mindee.logger import logger
 from mindee.mindee_http.endpoint import CustomEndpoint, Endpoint
 from mindee.mindee_http.mindee_api import MindeeApi

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/common/extracted_image.py

@@ -5,7 +5,8 @@
 from PIL import Image
 
 from mindee.error.mindee_error import MindeeError
-from mindee.input.sources import FileInput, LocalInputSource
+from mindee.input.sources.file_input import FileInput
+from mindee.input.sources.local_input_source import LocalInputSource
 from mindee.logger import logger
 
 

mindee/extraction/common/image_extractor.py

@@ -8,7 +8,8 @@
 from mindee.extraction.common.extracted_image import ExtractedImage
 from mindee.geometry.point import Point
 from mindee.geometry.polygon import get_min_max_x, get_min_max_y
-from mindee.input.sources import BytesInput, LocalInputSource
+from mindee.input.sources.bytes_input import BytesInput
+from mindee.input.sources.local_input_source import LocalInputSource
 
 
 def attach_image_as_new_file(  # type: ignore
@@ -158,6 +159,6 @@ def load_pdf_doc(input_file: LocalInputSource) -> pdfium.PdfDocument:  # type: i
     """
     if input_file.is_pdf():
         input_file.file_object.seek(0)
-        return pdfium.PdfDocument(input_file.file_object)
+        return pdfium.PdfDocument(input_file.file_object.read())
 
     return attach_image_as_new_file(input_file.file_object)

mindee/extraction/multi_receipts_extractor/multi_receipts_extractor.py

@@ -5,7 +5,7 @@
 from mindee.extraction.common.image_extractor import (
     extract_multiple_images_from_source,
 )
-from mindee.input.sources import LocalInputSource
+from mindee.input.sources.local_input_source import LocalInputSource
 from mindee.parsing.common.inference import Inference
 
 

mindee/extraction/pdf_extractor/extracted_pdf.py

@@ -4,7 +4,7 @@
 import pypdfium2 as pdfium
 
 from mindee.error.mindee_error import MindeeError
-from mindee.input.sources import BytesInput
+from mindee.input.sources.bytes_input import BytesInput
 
 
 class ExtractedPdf:

mindee/extraction/pdf_extractor/pdf_extractor.py

@@ -7,7 +7,7 @@
 
 from mindee.error.mindee_error import MindeeError
 from mindee.extraction.pdf_extractor.extracted_pdf import ExtractedPdf
-from mindee.input.sources import LocalInputSource
+from mindee.input.sources.local_input_source import LocalInputSource
 from mindee.product.invoice_splitter.invoice_splitter_v1_page_group import (
     InvoiceSplitterV1PageGroup,
 )

mindee/image_operations/__init__.py

@@ -0,0 +1 @@
+from mindee.image_operations.image_compressor import compress_image

mindee/image_operations/image_compressor.py

@@ -0,0 +1,35 @@
+import io
+from typing import BinaryIO, Union
+
+from PIL import Image
+
+
+def compress_image(
+    image_buffer: Union[BinaryIO, 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, also accepts BinaryIO.
+    :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:
+    """
+    if isinstance(image_buffer, bytes):
+        image_buffer = io.BytesIO(image_buffer)
+    with Image.open(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/input/__init__.py

@@ -3,7 +3,8 @@
 from mindee.input.sources.base_64_input import Base64Input
 from mindee.input.sources.bytes_input import BytesInput
 from mindee.input.sources.file_input import FileInput
-from mindee.input.sources.local_input_source import InputType, LocalInputSource
+from mindee.input.sources.input_type import InputType
+from mindee.input.sources.local_input_source import LocalInputSource
 from mindee.input.sources.path_input import PathInput
 from mindee.input.sources.url_input_source import UrlInputSource
 from mindee.input.workflow_options import WorkflowOptions

mindee/input/sources/__init__.py

@@ -1,6 +1,7 @@
 from mindee.input.sources.base_64_input import Base64Input
 from mindee.input.sources.bytes_input import BytesInput
 from mindee.input.sources.file_input import FileInput
-from mindee.input.sources.local_input_source import InputType, LocalInputSource
+from mindee.input.sources.input_type import InputType
+from mindee.input.sources.local_input_source import LocalInputSource
 from mindee.input.sources.path_input import PathInput
 from mindee.input.sources.url_input_source import UrlInputSource

mindee/input/sources/base_64_input.py

@@ -1,7 +1,8 @@
 import base64
 import io
 
-from mindee.input.sources.local_input_source import InputType, LocalInputSource
+from mindee.input.sources.input_type import InputType
+from mindee.input.sources.local_input_source import LocalInputSource
 
 
 class Base64Input(LocalInputSource):

mindee/input/sources/bytes_input.py

@@ -1,6 +1,7 @@
 import io
 
-from mindee.input.sources.local_input_source import InputType, LocalInputSource
+from mindee.input.sources.input_type import InputType
+from mindee.input.sources.local_input_source import LocalInputSource
 
 
 class BytesInput(LocalInputSource):

mindee/input/sources/file_input.py

@@ -1,7 +1,8 @@
 import os
 from typing import BinaryIO
 
-from mindee.input.sources.local_input_source import InputType, LocalInputSource
+from mindee.input.sources.input_type import InputType
+from mindee.input.sources.local_input_source import LocalInputSource
 
 
 class FileInput(LocalInputSource):

mindee/input/sources/input_type.py

@@ -0,0 +1,11 @@
+from enum import Enum
+
+
+class InputType(Enum):
+    """The input type, for internal use."""
+
+    FILE = "file"
+    BASE64 = "base64"
+    BYTES = "bytes"
+    PATH = "path"
+    URL = "url"

mindee/input/sources/local_input_source.py

@@ -1,15 +1,18 @@
 import io
 import mimetypes
 import tempfile
-from enum import Enum
 from typing import BinaryIO, Optional, Sequence, Tuple
 
 import pypdfium2 as pdfium
 
 from mindee.error.mimetype_error import MimeTypeError
 from mindee.error.mindee_error import MindeeError, MindeeSourceError
+from mindee.image_operations.image_compressor import compress_image
 from mindee.input.page_options import KEEP_ONLY, REMOVE
+from mindee.input.sources.input_type import InputType
 from mindee.logger import logger
+from mindee.pdf.pdf_compressor import compress_pdf
+from mindee.pdf.pdf_utils import has_source_text
 
 mimetypes.add_type("image/heic", ".heic")
 mimetypes.add_type("image/heic", ".heif")
@@ -25,16 +28,6 @@
 ]
 
 
-class InputType(Enum):
-    """The input type, for internal use."""
-
-    FILE = "file"
-    BASE64 = "base64"
-    BYTES = "bytes"
-    PATH = "path"
-    URL = "url"
-
-
 class LocalInputSource:
     """Base class for all input sources coming from the local machine."""
 
@@ -202,3 +195,43 @@ def read_contents(self, close_file: bool) -> Tuple[str, bytes]:
     def close(self) -> None:
         """Close the file object."""
         self.file_object.close()
+
+    def has_source_text(self) -> bool:
+        """
+        If the file is a PDF, checks if it has source text.
+
+        :return: True if the file is a PDF and has source text. False otherwise.
+        """
+        if not self.is_pdf():
+            return False
+        return has_source_text(self.file_object.read())
+
+    def compress(
+        self,
+        quality: int = 85,
+        max_width: Optional[int] = None,
+        max_height: Optional[int] = None,
+        force_source_text: bool = False,
+        disable_source_text: bool = True,
+    ) -> None:
+        """
+        Compresses the file object, either as a PDF or an image.
+
+        :param quality: Quality of the compression. For images, this is the JPEG quality.
+            For PDFs, this affects image quality within the PDF.
+        :param max_width: Maximum width for image resizing. Ignored for PDFs.
+        :param max_height: Maximum height for image resizing. Ignored for PDFs.
+        :param force_source_text: For PDFs, whether to force compression even if source text is present.
+        :param disable_source_text: For PDFs, whether to disable source text during compression.
+        """
+        new_file_bytes: bytes
+        if self.is_pdf():
+            new_file_bytes = compress_pdf(
+                self.file_object, quality, force_source_text, disable_source_text
+            )
+        else:
+            new_file_bytes = compress_image(
+                self.file_object, quality, max_width, max_height
+            )
+
+        self.file_object = io.BytesIO(new_file_bytes)

mindee/input/sources/path_input.py

@@ -2,7 +2,8 @@
 from pathlib import Path
 from typing import Union
 
-from mindee.input.sources.local_input_source import InputType, LocalInputSource
+from mindee.input.sources.input_type import InputType
+from mindee.input.sources.local_input_source import LocalInputSource
 
 
 class PathInput(LocalInputSource):

mindee/input/sources/url_input_source.py

@@ -10,7 +10,7 @@
 
 from mindee.error.mindee_error import MindeeSourceError
 from mindee.input.sources.bytes_input import BytesInput
-from mindee.input.sources.local_input_source import InputType
+from mindee.input.sources.input_type import InputType
 from mindee.logger import logger
 
 

mindee/mindee_http/endpoint.py

@@ -4,7 +4,8 @@
 import requests
 from requests import Response
 
-from mindee.input.sources import LocalInputSource, UrlInputSource
+from mindee.input.sources.local_input_source import LocalInputSource
+from mindee.input.sources.url_input_source import UrlInputSource
 from mindee.mindee_http.base_endpoint import BaseEndpoint
 from mindee.mindee_http.mindee_api import MindeeApi
 from mindee.parsing.common.string_dict import StringDict

mindee/pdf/__init__.py

@@ -0,0 +1,7 @@
+from mindee.pdf.pdf_char_data import PDFCharData
+from mindee.pdf.pdf_compressor import compress_pdf
+from mindee.pdf.pdf_utils import (
+    extract_text_from_pdf,
+    has_source_text,
+    lerp,
+)

mindee/pdf/pdf_char_data.py

@@ -0,0 +1,32 @@
+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: float
+    """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."""
+    page_id: int
+    """ID of the page the character was found on."""