pdf_to_img.py
12.5 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
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
import os
import sys
import fitz
import argparse
from PIL import Image
from io import BytesIO
if sys.version_info[0] < 3:
raise Exception("This program requires at least python3.6")
parser = argparse.ArgumentParser(description='PDF转图片')
parser.add_argument('-i', '--input', help='PDF文件或目录路径，必要参数', required=True)
parser.add_argument('-o', '--output', help='输出图片保存路径，非必要参数，缺省值为PDF文件路径')
parser.add_argument('-e', '--extract', help='默认采用整个页面保存png图片的方式，增加该选项选择提取图片方式转化图片', action="store_true")
args = parser.parse_args()
LOG_BASE = '[pdf to img]'
# 页面保存为png图片参数
ZOOM_X_1 = ZOOM_Y_1 = 1.0
ZOOM_X_2 = ZOOM_Y_2 = 2.0
trans_1 = fitz.Matrix(ZOOM_X_1, ZOOM_X_1).preRotate(0) # zoom factor 1 in each dimension
trans_2 = fitz.Matrix(ZOOM_X_2, ZOOM_X_2).preRotate(0) # zoom factor 2 in each dimension
# 特殊filter处理
ADOBE_FILTER_SET = {'FlateDecode', 'JPXDecode', 'JBIG2Decode'}
# 宽高阈值组合
WH_COUPLE_1 = (500, 500)
WH_COUPLE_2 = (700, 647)
WH_COUPLE_3 = (100, 100)
WH_COUPLE_4 = (100, 300)
WH_COUPLE_5 = (100, 200)
class PDFHandler:
def __init__(self, path, target_path):
self.path = path
self.img_dir_path = os.path.join(target_path, os.path.splitext(os.path.basename(path))[0])
self.xref_set = set()
def get_img_save_path(self, pno, img_index=0, ext='jpeg'):
return os.path.join(self.img_dir_path, 'page_{0}_img_{1}.{2}'.format(pno, img_index, ext))
def page_to_png(self, page):
if page.MediaBoxSize.x > 1500 or page.MediaBoxSize.y > 1500:
pm = page.getPixmap(matrix=trans_1, alpha=False)
else:
pm = page.getPixmap(matrix=trans_2, alpha=False)
img_save_path = self.get_img_save_path(page.number)
pm.writeImage(img_save_path)
# pm.writePNG(img_save_path)
@staticmethod
def getimage(pix):
# RGB
if pix.colorspace.n != 4:
return pix
# GRAY/CMYK
tpix = fitz.Pixmap(fitz.csRGB, pix)
return tpix
def recover_pix(self, doc, xref, smask, colorspace):
if smask != 0:
# we need to reconstruct the alpha channel with the smask
pix1 = fitz.Pixmap(doc, xref)
pix2 = fitz.Pixmap(doc, smask) # 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)
elif colorspace in {'Separation', 'DeviceCMYK'}:
pix = fitz.Pixmap(doc, xref)
tpix = fitz.Pixmap(fitz.csRGB, pix)
return tpix
else:
res = doc.extractImage(xref)
if res is None:
return fitz.Pixmap(doc, xref)
else:
return res
# return doc.extractImage(xref)
@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
def extract_single_image(self, pdf, xref, smask, colorspace, pno, img_index=0):
pix = self.recover_pix(pdf, xref, smask, colorspace)
ext, img_data = self.get_img_data(pix)
if ext == 'jpx':
img_save_path = self.get_img_save_path(pno, img_index=img_index)
jpx_pix = fitz.Pixmap(img_data)
jpx_pix.writeImage(img_save_path)
jpx_pix = None
else:
img_save_path = self.get_img_save_path(pno, img_index=img_index, ext=ext)
with open(img_save_path, "wb") as f:
f.write(img_data)
self.xref_set.add(xref)
@staticmethod
def split_il(il):
broken_il = []
start = 0
length = len(il)
page_to_png = None
for i in range(length):
# 当图片对象含有特殊filter时，特殊处理：整个页面保存为png图片
if il[i][-1] in ADOBE_FILTER_SET:
page_to_png = True
break
else:
for i in range(length):
# 当图片对象够大时，不作碎图合并处理，而是单纯提取
if il[i][2] >= WH_COUPLE_2[0] and il[i][3] >= WH_COUPLE_2[1]:
break
if i == start:
if i == length - 1:
broken_il.append(il[start: length])
continue
elif i == length - 1:
if il[i][2] == il[i - 1][2]:
broken_il.append(il[start: length])
else:
broken_il.append(il[start: i])
broken_il.append(il[i: length])
continue
if il[i][2] != il[i - 1][2]:
broken_il.append(il[start: i])
start = i
elif il[i][3] != il[i - 1][3]:
broken_il.append(il[start: i + 1])
start = i + 1
else:
# 碎图分组结果
return broken_il
return page_to_png
def merge_il(self, pdf, pno, il):
# 尝试碎图合并前的分组
il.sort(key=lambda x: x[0])
broken_il = self.split_il(il)
print('broken_il: {0}'.format(broken_il))
page_to_png = True
# 3.1 当图片对象够大时，不作碎图合并处理，而是单纯提取
if broken_il is None:
page_to_png = False
for img_index, img in enumerate(il):
xref, smask, width, height, _, colorspace, _, _, adobe_filter = img
if width < WH_COUPLE_3[0] or height < WH_COUPLE_3[1]: # 过滤小图(如二维码)
continue
elif xref not in self.xref_set:
self.extract_single_image(pdf, xref, smask, colorspace, pno, img_index)
# 3.2 碎图按照分组合并
elif isinstance(broken_il, list) and len(broken_il) <= 2:
for img_index, img_il in enumerate(broken_il):
# 3.2.1 仅一张碎图，过滤或直接提取
if len(img_il) == 1:
xref, smask, width, height, _, colorspace, _, _, adobe_filter = img_il[0]
# 过滤小图(如二维码)
if width < WH_COUPLE_4[0] or height < WH_COUPLE_4[1] or \
(width < WH_COUPLE_1[0] and height < WH_COUPLE_1[1]):
continue
elif xref not in self.xref_set:
self.extract_single_image(pdf, xref, smask, colorspace, pno, img_index)
page_to_png = False
# 3.2.2 多张碎图，竖向拼接
else:
height_sum = sum([img[3] for img in img_il])
width = img_il[0][2]
# 过滤小图和不常规大图
if width < WH_COUPLE_5[0] or height_sum < WH_COUPLE_5[1] or \
(width > 1000 and height_sum > width * 3):
continue
im_list = []
for img in img_il:
xref, smask, _, height, _, colorspace, _, _, adobe_filter = img
pix = self.recover_pix(pdf, xref, smask, colorspace)
ext, img_data = self.get_img_data(pix)
im = Image.open(BytesIO(img_data))
im_list.append((height, im, ext))
new_img = Image.new(im_list[0][1].mode, (width, height_sum))
h_now = 0
for h, m, _ in im_list:
new_img.paste(m, box=(0, h_now))
h_now += h
img_save_path = self.get_img_save_path(pno, img_index, im_list[0][2])
new_img.save(img_save_path)
page_to_png = False
# 3.3 碎图分组大于2、全过滤、含特殊filter，特殊处理：整个页面保存为png图片
if page_to_png:
page = pdf.loadPage(pno)
self.page_to_png(page)
def extract_image(self, is_extract):
os.makedirs(self.img_dir_path, exist_ok=True)
with fitz.Document(self.path) as pdf:
print('++++++++++' * 5)
print('{0} [start] [pdf_path={1}] [metadata={2}]'.format(LOG_BASE, self.path, pdf.metadata))
for pno in range(pdf.pageCount):
il = pdf.getPageImageList(pno) if is_extract else [] # 获取页面图片对象
# (xref, smask, width, height, bpc, colorspace, alt.colorspace, name, filter, invoker)
print('---------- page: {0} ----------'.format(pno))
print('img_object_list: {0}'.format(il))
# 单纯提取页面图片对象
# for img_index, img in enumerate(il):
# pix = self.recover_pix(pdf, img[0], img[1], img[5])
# ext, img_data = self.get_img_data(pix)
# img_save_path = self.get_img_save_path(pno, img_index, ext)
# with open(img_save_path, "wb") as f:
# f.write(img_data)
# 1.页面图片对象数目为0时，保存整个页面为png图片
if len(il) == 0:
page = pdf.loadPage(pno)
self.page_to_png(page)
# 2.页面图片对象数目为1时：
# 小图(如电子账单的盖章)：保存整个页面为png图片
# 大图：提取图片对象
elif len(il) == 1:
xref, smask, width, height, _, colorspace, _, _, _ = il[0]
# 小图
if width < WH_COUPLE_1[0] and height < WH_COUPLE_1[1]:
page = pdf.loadPage(pno)
self.page_to_png(page)
# 大图
elif xref not in self.xref_set:
self.extract_single_image(pdf, xref, smask, colorspace, pno)
# 3.页面图片对象数目大于1时，特殊处理
else:
self.merge_il(pdf, pno, il)
def extract_image(pdf_path, target_path, is_extract):
pdf_handler = PDFHandler(pdf_path, target_path)
pdf_handler.extract_image(is_extract)
def main():
if not os.path.exists(args.input):
print('PDF文件或目录不存在: {0}'.format(args.input))
return
pdf_path = os.path.realpath(args.input)
# 目录：遍历处理所有pdf文件
if os.path.isdir(pdf_path):
completed_count = 0
failed_list = []
for parent, dirnames, filenames in os.walk(pdf_path):
# 图片保存目录
# target_path = os.path.realpath(args.output) if args.output else parent
target_path = parent.replace(pdf_path, args.output)
for pdf_file in filenames:
if not pdf_file.endswith('pdf') and not pdf_file.endswith('PDF'):
continue
pdf_file_path = os.path.join(parent, pdf_file)
try:
extract_image(pdf_file_path, target_path, args.extract)
except Exception as e:
print('{0} [failed] [err={1}] [pdf_path={2}]'.format(LOG_BASE, e, pdf_file_path))
failed_list.append(pdf_file_path)
else:
print('{0} [completed] [pdf_path={1}]'.format(LOG_BASE, pdf_path))
completed_count += 1
print('{0} [all completed] [completed_count={1}] [failed_count={2}] [failed_pdf_path={3}]'.format(
LOG_BASE, completed_count, len(failed_list), failed_list))
# 文件：处理pdf文件
else:
# 图片保存目录
target_path = os.path.realpath(args.output) if args.output else os.path.dirname(pdf_path)
try:
extract_image(pdf_path, target_path, args.extract)
except Exception as e:
print('{0} [failed] [err={1}] [pdf_path={2}]'.format(LOG_BASE, e, pdf_path))
else:
print('{0} [completed] [pdf_path={1}]'.format(LOG_BASE, pdf_path))
if __name__ == "__main__":
main()