pdf_to_img.py
11.8 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
import os
import fitz
import signal
from PIL import Image
from io import BytesIO
from django.core.management import BaseCommand
from common.mixins import LoggerMixin
class Command(BaseCommand, LoggerMixin):
def __init__(self):
super().__init__()
self.log_base = '[pdf to img]'
# 处理文件开关
self.switch = True
# pdf页面转图片
self.zoom_x = 2.0
self.zoom_y = 2.0
self.trans = fitz.Matrix(self.zoom_x, self.zoom_y).preRotate(0) # zoom factor 2 in each dimension
# 优雅退出信号：15
signal.signal(signal.SIGTERM, self.signal_handler)
def signal_handler(self, sig, frame):
self.switch = False # 停止处理文件
@staticmethod
def getimage(pix):
if pix.colorspace.n != 4:
return pix
tpix = fitz.Pixmap(fitz.csRGB, pix)
return tpix
def recoverpix(self, doc, item):
x = item[0] # xref of PDF image
s = item[1] # xref of its /SMask
is_rgb = True if item[5] == 'DeviceRGB' else False
# RGB
if is_rgb:
if s == 0:
return doc.extractImage(x)
# we need to reconstruct the alpha channel with the smask
pix1 = fitz.Pixmap(doc, x)
pix2 = fitz.Pixmap(doc, s) # create pixmap of the /SMask entry
# sanity check
if not (pix1.irect == pix2.irect and pix1.alpha == pix2.alpha == 0 and pix2.n == 1):
pix2 = None
return self.getimage(pix1)
pix = fitz.Pixmap(pix1) # copy of pix1, alpha channel added
pix.setAlpha(pix2.samples) # treat pix2.samples as alpha value
pix1 = pix2 = None # free temp pixmaps
return self.getimage(pix)
# GRAY/CMYK
pix1 = fitz.Pixmap(doc, x)
pix = fitz.Pixmap(pix1) # copy of pix1, alpha channel added
if s != 0:
pix2 = fitz.Pixmap(doc, s) # create pixmap of the /SMask entry
# sanity check
if not (pix1.irect == pix2.irect and pix1.alpha == pix2.alpha == 0 and pix2.n == 1):
pix2 = None
return self.getimage(pix1)
pix.setAlpha(pix2.samples) # treat pix2.samples as alpha value
pix1 = pix2 = None # free temp pixmaps
pix = fitz.Pixmap(fitz.csRGB, pix) # GRAY/CMYK to RGB
return self.getimage(pix)
@staticmethod
def get_img_data(pix):
if type(pix) is dict: # we got a raw image
ext = pix["ext"]
img_data = pix["image"]
else: # we got a pixmap
ext = 'png'
img_data = pix.getPNGData()
return ext, img_data
@staticmethod
def split_il(il):
small_img_il_list = []
big_img_il_list = []
start = 0
index = 0
length = len(il)
for i in range(length):
if il[i][2] >= 700 and il[i][3] >= 647:
if start < i:
small_img_il_list.append((il[start: i], index))
index += 1
else:
start += 1
big_img_il_list.append((il[i], index))
index += 1
continue
if i == start:
if i == length - 1:
small_img_il_list.append((il[start: length], index))
continue
elif i == length - 1:
if il[i][2] == il[i - 1][2]:
small_img_il_list.append((il[start: length], index))
else:
small_img_il_list.append((il[start: i], index))
small_img_il_list.append((il[i: length], index+1))
continue
if il[i][2] != il[i - 1][2]:
small_img_il_list.append((il[start: i], index))
index += 1
start = i
elif il[i][3] != il[i - 1][3] and il[i][2] < 1200:
small_img_il_list.append((il[start: i + 1], index))
index += 1
start = i + 1
return small_img_il_list, big_img_il_list
def handle(self, *args, **kwargs):
pdf_dir = '/Users/clay/Desktop/问题PDF'
img_dir = '/Users/clay/Desktop/问题PDF'
for d in os.listdir(pdf_dir):
# if d in ['.DS_Store', 'CH-B008003736.pdf', 'CH-B006317088.pdf', 'CH-B008487476.pdf', 'CH-B006337608.pdf',
# 'CH-B006391612.pdf', 'CH-B006536124.pdf', 'CH-B006526652.pdf', 'CH-B009003592.pdf']:
# continue
# if d != 'CH-B006393152.PDF':
# if d != 'CH-B006526652.pdf':
if d != 'CH-B008487944.pdf':
continue
pdf_path = os.path.join(pdf_dir, d)
if os.path.isfile(pdf_path):
img_save_path = os.path.join(img_dir, d[:-4])
# if os.path.exists(img_save_path):
# continue
os.makedirs(img_save_path, exist_ok=True)
with fitz.Document(pdf_path) as pdf:
self.cronjob_log.info('{0} [pdf_path={1}] [metadata={2}]'.format(
self.log_base, pdf_path, pdf.metadata))
xref_set = set()
for pno in range(pdf.pageCount):
print('---------------------------------------')
il = pdf.getPageImageList(pno)
# (xref, smask, width, height, bpc, colorspace, alt.colorspace, name, filter, invoker)
print(il)
# for img_index, img in enumerate(il):
# pix = self.recoverpix(pdf, img)
# ext, img_data = self.get_img_data(pix)
# save_path = os.path.join(img_save_path, 'page_{0}_img_{1}.{2}'.format(
# pno, img_index, ext))
# with open(save_path, "wb") as f:
# f.write(img_data)
if len(il) == 0:
page = pdf.loadPage(pno)
pm = page.getPixmap(matrix=self.trans, alpha=False)
save_path = os.path.join(img_save_path, 'page_{0}_img_0.png'.format(page.number))
pm.writePNG(save_path)
elif len(il) == 1:
width = il[0][2]
height = il[0][3]
colorspace = il[0][5]
adobe_filter = il[0][-1]
if colorspace == '' or adobe_filter in ['', '']:
continue
# 小图
if width < 500 and height < 500:
page = pdf.loadPage(pno)
pm = page.getPixmap(matrix=self.trans, alpha=False)
save_path = os.path.join(img_save_path, 'page_{0}_img_0.png'.format(page.number))
pm.writePNG(save_path)
# 大图
elif il[0][0] not in xref_set:
pix = self.recoverpix(pdf, il[0])
ext, img_data = self.get_img_data(pix)
save_path = os.path.join(img_save_path, 'page_{0}_img_0.{1}'.format(pno, ext))
with open(save_path, "wb") as f:
f.write(img_data)
xref_set.add(il[0][0])
else:
il.sort(key=lambda x: x[0])
small_img_il_list, big_img_il_list = self.split_il(il)
print(small_img_il_list)
print(big_img_il_list)
print('+++++++++++++++++++++++++++++++++++')
if len(small_img_il_list) > 2: # 单页无规律小图过多时，使用页面转图片
page = pdf.loadPage(pno)
pm = page.getPixmap(matrix=self.trans, alpha=False)
save_path = os.path.join(img_save_path, 'page_{0}_img_0.png'.format(page.number))
pm.writePNG(save_path)
else: # 提取图片
for img_il, img_index in big_img_il_list:
if img_il[0] in xref_set:
continue
pix = self.recoverpix(pdf, img_il)
ext, img_data = self.get_img_data(pix)
save_path = os.path.join(img_save_path, 'page_{0}_img_{1}.{2}'.format(
pno, img_index, ext))
with open(save_path, "wb") as f:
f.write(img_data)
xref_set.add(img_il[0])
for img_il, img_index in small_img_il_list:
# 小图
if len(img_il) == 1 and img_il[0][2] < 500 and img_il[0][3] < 500:
page = pdf.loadPage(pno)
pm = page.getPixmap(matrix=self.trans, alpha=False)
save_path = os.path.join(img_save_path,
'page_{0}_img_0.png'.format(page.number))
pm.writePNG(save_path)
elif len(img_il) == 1 and img_il[0][0] not in xref_set: # 当只有一张图片时， 简化处理
pix = self.recoverpix(pdf, img_il[0])
ext, img_data = self.get_img_data(pix)
save_path = os.path.join(img_save_path, 'page_{0}_img_{1}.{2}'.format(
pno, img_index, ext))
with open(save_path, "wb") as f:
f.write(img_data)
xref_set.add(img_il[0][0])
else: # 多张图片，竖向拼接
height_sum = 0
im_list = []
width = img_il[0][2]
for img in img_il:
# xref = img[0]
# if xref in xref_list:
# continue
height = img[3]
pix = self.recoverpix(pdf, img)
ext, img_data = self.get_img_data(pix)
# xref_list.append(xref)
im = Image.open(BytesIO(img_data))
im_list.append((height, im, ext))
height_sum += height
save_path = os.path.join(img_save_path, 'page_{0}_img_{1}.{2}'.format(
pno, img_index, im_list[0][2]))
res = Image.new(im_list[0][1].mode, (width, height_sum))
h_now = 0
for h, m, _ in im_list:
res.paste(m, box=(0, h_now))
h_now += h
res.save(save_path)