Skip to content

Commit 6e7e216

Browse files
restore function that was needlessly edited
1 parent 940a574 commit 6e7e216

File tree

3 files changed

+26
-23
lines changed

3 files changed

+26
-23
lines changed

mindee/extraction/__init__.py

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,6 @@
11
from mindee.extraction.common.extracted_image import ExtractedImage
22
from mindee.extraction.common.image_extractor import (
3+
attach_image_as_new_file,
34
extract_multiple_images_from_source,
45
)
56
from mindee.extraction.multi_receipts_extractor import multi_receipts_extractor
Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,5 @@
11
from mindee.extraction.common.extracted_image import ExtractedImage
22
from mindee.extraction.common.image_extractor import (
3+
attach_image_as_new_file,
34
extract_multiple_images_from_source,
45
)

mindee/extraction/common/image_extractor.py

Lines changed: 24 additions & 23 deletions
Original file line numberDiff line numberDiff line change
@@ -129,35 +129,36 @@ def load_pdf_doc(input_file: LocalInputSource) -> pdfium.PdfDocument: # type: i
129129
input_file.file_object.seek(0)
130130
return pdfium.PdfDocument(input_file.file_object.read())
131131

132-
return attach_images_as_new_file([input_file.file_object])
132+
return attach_image_as_new_file(input_file.file_object)
133133

134134

135-
def attach_images_as_new_file( # type: ignore
136-
input_buffer_list: List[BinaryIO],
135+
def attach_image_as_new_file( # type: ignore
136+
input_buffer: BinaryIO,
137137
) -> pdfium.PdfDocument:
138138
"""
139-
Attaches a list of images as new pages in a PdfDocument object.
139+
Attaches an image as a new page in a PdfDocument object.
140140
141-
:param input_buffer_list: List of images, represented as buffers.
141+
:param input_buffer: Input buffer.
142142
:return: A PdfDocument handle.
143143
"""
144+
# Create a new page in the PdfDocument
145+
input_buffer.seek(0)
146+
image = Image.open(input_buffer)
147+
image.convert("RGB")
148+
image_buffer = io.BytesIO()
149+
image.save(image_buffer, format="JPEG")
150+
144151
pdf = pdfium.PdfDocument.new()
145-
for input_buffer in input_buffer_list:
146-
input_buffer.seek(0)
147-
image = Image.open(input_buffer)
148-
image.convert("RGB")
149-
image_buffer = io.BytesIO()
150-
image.save(image_buffer, format="JPEG")
151-
152-
image_pdf = pdfium.PdfImage.new(pdf)
153-
image_pdf.load_jpeg(image_buffer)
154-
width, height = image_pdf.get_size()
155-
156-
matrix = pdfium.PdfMatrix().scale(width, height)
157-
image_pdf.set_matrix(matrix)
158-
159-
page = pdf.new_page(width, height)
160-
page.insert_obj(image_pdf)
161-
page.gen_content()
162-
image.close()
152+
153+
image_pdf = pdfium.PdfImage.new(pdf)
154+
image_pdf.load_jpeg(image_buffer)
155+
width, height = image_pdf.get_size()
156+
157+
matrix = pdfium.PdfMatrix().scale(width, height)
158+
image_pdf.set_matrix(matrix)
159+
160+
page = pdf.new_page(width, height)
161+
page.insert_obj(image_pdf)
162+
page.gen_content()
163+
image.close()
163164
return pdf

0 commit comments

Comments
 (0)