a2451b9103
Merge in RR/cv-analysis from add-pdf2array-func to master
Squashed commit of the following:
commit 6e6e9a509ede0abf28fb93a2042960efcc9453bd
Author: Julius Unverfehrt <julius.unverfehrt@iqser.com>
Date: Wed Jul 20 09:12:01 2022 +0200
update script with layout parsing, refactor pdf2array
commit 191bc71f58aa5c07b0cadbdb7067cd72c3d8858b
Author: Julius Unverfehrt <julius.unverfehrt@iqser.com>
Date: Wed Jul 20 09:10:06 2022 +0200
update script with layout parsing, refactor pdf2array
commit 25201bbb4151a23784193181272d379232877d2f
Author: Julius Unverfehrt <julius.unverfehrt@iqser.com>
Date: Wed Jul 20 08:33:20 2022 +0200
add pdf2array functionality
41 lines
1.3 KiB
Python
41 lines
1.3 KiB
Python
from functools import partial
|
|
from typing import Iterator, Tuple
|
|
|
|
import fitz
|
|
import numpy as np
|
|
|
|
|
|
def pdf_to_array_and_metadata(pdf: bytes, index=None, dpi=200) -> Iterator[Tuple[np.ndarray, dict]]:
|
|
"""Stream the pages of a PDF as Tuples of page as matrix representation and page metadata.
|
|
Note: If Index is not given or evaluates to None, the whole PDF will be processed.
|
|
"""
|
|
convert_fn = partial(page_to_array_and_metadata, dpi=dpi)
|
|
yield from map(convert_fn, stream_pages(pdf, index))
|
|
|
|
|
|
def page_to_array_and_metadata(page: fitz.Page, dpi):
|
|
metadata = get_page_info(page, dpi)
|
|
pixmap = page.get_pixmap(dpi=dpi)
|
|
array = np.frombuffer(pixmap.samples, dtype=np.uint8).reshape(pixmap.h, pixmap.w, pixmap.n)
|
|
|
|
return array, metadata
|
|
|
|
|
|
def stream_pages(pdf: bytes, index=None) -> Iterator[fitz.Page]:
|
|
with fitz.open(stream=pdf) as pdf_handle:
|
|
if not index:
|
|
yield from pdf_handle
|
|
else:
|
|
for i in index:
|
|
yield pdf_handle[i]
|
|
|
|
|
|
def get_page_info(page, dpi):
|
|
return {
|
|
"index": page.number,
|
|
"rotation": page.rotation,
|
|
"width": page.rect.width, # rotated page width in inches
|
|
"height": page.rect.height, # rotated page height in inches
|
|
"dpi": dpi,
|
|
}
|