def invert_image(image_path):
from PIL import Image
from fitz import Pixmap
pix = Pixmap(image_path)
pix.invertIRect(pix.irect)
buffer = BytesIO(pix.getImageData())
del pix
return Image.open(buffer)
def to_raw_dict(cls, image:fitz.Pixmap, bbox:fitz.Rect):
'''Store Pixmap to raw dict.'''
return {
'type': BlockType.IMAGE.value,
'bbox': tuple(bbox),
'ext': 'png',
'width': image.width,
'height': image.height,
'image': image.getPNGData()
}
def _pixmap_to_cv_image(pixmap: fitz.Pixmap):
'''Convert fitz Pixmap to opencv image.
Args:
pixmap (fitz.Pixmap): PyMuPDF Pixmap.
'''
import cv2 as cv
import numpy as np
img_byte = pixmap.tobytes()
return cv.imdecode(np.frombuffer(img_byte, np.uint8), cv.IMREAD_COLOR)
def manga_to_PDF(dir_path, one_file=True, width=None, height=None):
if dir_path[-1] == '\\' or dir_path[-1] == '/':
dir_path = dir_path[0:-1]
if not os.path.isdir(dir_path):
raise ValueError('传入的路径并非文件夹')
from fitz import Document, Pixmap, Rect
from glob import _iglob as glob
if one_file:
title = os.path.basename(dir_path)
with Document() as doc:
for file_path in glob(os.path.join(dir_path, "*", "*.jpg"), False,
False):
pixmap = Pixmap(file_path)
if width and height:
pixmap = Pixmap(pixmap, width, height, None)
elif width:
pixmap = Pixmap(pixmap, width,
int(pixmap.height / pixmap.width * width),
None)
elif height:
pixmap = Pixmap(pixmap,
int(pixmap.width / pixmap.height * height),
height, None)
rect = Rect(0, 0, pixmap.width, pixmap.height)
page = doc.newPage(width=pixmap.width, height=pixmap.height)
page.insertImage(rect, pixmap=pixmap)
doc.save(os.path.join(dir_path, title + ".pdf"), deflate=True)
else:
for chap in glob(os.path.join(dir_path, "*"), False, True):
title = os.path.basename(chap)
with Document() as doc:
for file_path in glob(os.path.join(chap, "*.jpg"), False,
False):
pixmap = Pixmap(file_path)
if width and height:
pixmap = Pixmap(pixmap, width, height, None)
elif width:
pixmap = Pixmap(
pixmap, width,
int(pixmap.height / pixmap.width * width), None)
elif height:
pixmap = Pixmap(
pixmap, int(pixmap.width / pixmap.height * height),
height, None)
rect = Rect(0, 0, pixmap.width, pixmap.height)
page = doc.newPage(width=pixmap.width,
height=pixmap.height)
page.insertImage(rect, pixmap=pixmap)
doc.save(os.path.join(dir_path, title + ".pdf"), deflate=True)
def get_images(page, imblocks, page_index):
im_list = []
i = 0
pagepix = page.getPixmap(page_index)
pw = pagepix.width
ph = pagepix.height
# print("PH: {}\nPW: {}\n\n".format(ph, pw))
for block in imblocks:
try:
ending = block["ext"]
pix = Pixmap(block["image"])
x0 = block["bbox"][0]
y0 = block["bbox"][1]
x1 = block["bbox"][2]
y1 = block["bbox"][3]
w = pix.width
h = pix.height
filename = "static/images/image_{}-{}.{}".format(
page_index, i, ending)
if is_relevant_image(pw, ph, pix, x0, y0, x1, y1):
if pix.n < 5: # this is GRAY or RGB
pix.writePNG(filename)
else: # CMYK: convert to RGB first
pix1 = fitz.Pixmap(fitz.csRGB, pix)
pix1.writePNG(filename)
pix1 = None
pix = None
if h > w:
im_list.append((filename, "H"))
else:
im_list.append((filename, "W"))
i += 1
except:
pass
finally:
pass
return im_list
def _to_raw_dict(image: fitz.Pixmap, bbox: fitz.Rect):
'''Store Pixmap ``image`` to raw dict.
Args:
image (fitz.Pixmap): Pixmap to store.
bbox (fitz.Rect): Boundary box the pixmap.
Returns:
dict: Raw dict of the pixmap.
'''
return {
'type': BlockType.IMAGE.value,
'bbox': tuple(bbox),
'width': image.width,
'height': image.height,
'image': image.tobytes()
}
def _to_raw_dict(cls, image: fitz.Pixmap, bbox: fitz.Rect):
"""Store Pixmap ``image`` to raw dict.
Args:
image (fitz.Pixmap): Pixmap to store.
bbox (fitz.Rect): Boundary box the pixmap.
Returns:
dict: Raw dict of the pixmap.
"""
return {
'type': BlockType.IMAGE.value,
'bbox': tuple(bbox),
'ext': 'png',
'width': image.width,
'height': image.height,
'image': image.getPNGData()
}