home · contact · privacy
In Bookmaker, document code and its execution better.
[misc] / bookmaker.py
1 #!/usr/bin/env python3
2 import pypdf
3 import argparse
4 import io
5 from reportlab.lib.pagesizes import A4
6 a4_width, a4_height = A4
7 points_per_cm = 10 * 72 / 25.4
8 cut_depth = 1.95 * points_per_cm
9 cut_width = 1.05 * points_per_cm
10 middle_point_depth = 0.4 * points_per_cm
11
12 parser = argparse.ArgumentParser(description="build print-ready book PDF")
13 parser.add_argument("-i", "--input_file", action="append", required=True, help="input PDF file")
14 parser.add_argument("-o", "--output_file", required=True, help="output PDF file")
15 parser.add_argument("-p", "--page_range", action="append", help="page range, e.g., '3-end'")
16 parser.add_argument("-c", "--crop_range", action="append", help="cm crops left, bottom, right, top – e.g., '10,10,10,10'; prefix with ':'-delimited page range to limit effect")
17 parser.add_argument("-t", "--symmetry", action="store_true", help="alternate horizontal crops between odd and even pages")
18 parser.add_argument("-r", "--rotate", dest="rotate", type=int, action="append", help="rotate page of number by 90° (usable multiple times on same page!)")
19 parser.add_argument("-n", "--nup4", action='store_true', help="puts 4 input pages onto 1 output page")
20 parser.add_argument("-a", "--analyze", action="store_true", help="in --nup4, print lines identifying spine, page borders")
21 parser.add_argument("-m", "--margin", type=float, default=0.43, help="print margin for --nup4 in cm (default 0.43)")
22 parser.add_argument("-s", "--spine", type=float, default=1, help="on --nup4, cm width of margin hidden in spine (default 1)")
23 args = parser.parse_args()
24
25
26 # select pages from input files
27 def parse_page_range(range_string, pages):
28     start_page = 0
29     end_page = len(pages)
30     if range_string:
31         start, end = range_string.split('-')
32         if not (len(start) == 0 or start == "start"):
33             start_page = int(start) - 1
34         if not (len(end) == 0 or end == "end"):
35             end_page = int(end)
36     return start_page, end_page
37 pages_to_add = []
38 opened_files = []
39 for i, input_file in enumerate(args.input_file):
40     file = open(input_file, 'rb')
41     opened_files += [file]
42     reader = pypdf.PdfReader(file)
43     range_string = None
44     if args.page_range and len(args.page_range) > i:
45         range_string = args.page_range[i]
46     start_page, end_page = parse_page_range(range_string, reader.pages)
47     new_page_num = 0
48     for old_page_num in range(start_page, end_page):
49         new_page_num += 1
50         page = reader.pages[new_page_num]
51         pages_to_add += [page]
52         print("-i, -p: read in %s page number %d as new page %d" % (input_file, new_page_num+1, new_page_num))
53
54 # rotate page canvas
55 if args.rotate:
56     for rotate in args.rotate:
57         page = pages_to_add[rotate - 1]
58         page.add_transformation(pypdf.Transformation().translate(tx=-a4_width/2, ty=-a4_height/2))
59         page.add_transformation(pypdf.Transformation().rotate(-90))
60         page.add_transformation(pypdf.Transformation().translate(tx=a4_width/2, ty=a4_height/2))
61         print("-r: rotating (by 90°) page", rotate)
62
63 # normalize all pages to portrait A4
64 for page in pages_to_add:
65     if "/Rotate" in page:
66         page.rotate(360 - page["/Rotate"])
67     page.mediabox.left = 0
68     page.mediabox.bottom = 0
69     page.mediabox.top = a4_height
70     page.mediabox.right = a4_width
71     page.cropbox = page.mediabox
72
73 # determine page crops, zooms
74 crops_at_page = [(0,0,0,0)]*len(pages_to_add)
75 zoom_at_page = [1]*len(pages_to_add)
76 if args.crop_range:
77   for crop_range in args.crop_range:
78       initial_split = crop_range.split(':')
79       if len(initial_split) > 1:
80           page_range = initial_split[0]
81           crops = initial_split[1]
82       else:
83           page_range = None
84           crops = initial_split[0]
85       start_page, end_page = parse_page_range(page_range, pages_to_add)
86       crop_left, crop_bottom, crop_right, crop_top = [float(x) * points_per_cm for x in  crops.split(',')]
87       print("-c: to pages %d to %d applying crops: left %dcm, bottom %dcm, right %dcm, top %dcm" % (start_page + 1, end_page, crop_left, crop_bottom, crop_right, crop_top))
88       cropped_width  = a4_width - crop_left - crop_right
89       cropped_height = a4_height - crop_bottom - crop_top
90       zoom = 1
91       zoom_horizontal = a4_width / (a4_width - crop_left - crop_right)
92       zoom_vertical = a4_height / (a4_height - crop_bottom - crop_top)
93       if (zoom_horizontal > 1 and zoom_vertical < 1) or (zoom_horizontal < 1 and zoom_vertical > 1):
94           print("Error: opposing zooms.")
95           exit(1)
96       elif zoom_horizontal + zoom_vertical > 2:
97           zoom = min(zoom_horizontal, zoom_vertical)
98       else:
99           zoom = max(zoom_horizontal, zoom_vertical)
100       for page_num in range(start_page, end_page):
101           crops_at_page[page_num] = (crop_left, crop_bottom, crop_right, crop_top)
102           zoom_at_page[page_num] = zoom
103
104 writer = pypdf.PdfWriter()
105 if not args.nup4:
106     odd_page = True
107     for i, page in enumerate(pages_to_add):
108         crop_left, crop_bottom, crop_right, crop_top = crops_at_page[i]
109         zoom = zoom_at_page[i]
110         if args.symmetry and odd_page:
111             page.add_transformation(pypdf.Transformation().translate(tx=-crop_right, ty=-crop_bottom))
112         else:
113             page.add_transformation(pypdf.Transformation().translate(tx=-crop_left, ty=-crop_bottom))
114         page.add_transformation(pypdf.Transformation().scale(zoom, zoom))
115         cropped_width  = a4_width - crop_left - crop_right
116         cropped_height = a4_height - crop_bottom - crop_top
117         page.mediabox.right = cropped_width * zoom
118         page.mediabox.top = cropped_height * zoom
119         writer.add_page(page)
120         odd_page = not odd_page
121
122 else:
123     n_pages_per_axis = 2
124     spine_limit = args.spine * points_per_cm
125     printable_margin = args.margin * points_per_cm
126     printable_scale = (a4_width - 2*printable_margin)/a4_width
127     half_width = a4_width / n_pages_per_axis
128     half_height = a4_height / n_pages_per_axis
129     section_scale_factor = 1 / n_pages_per_axis
130     spine_part_of_page = (spine_limit / half_width) / printable_scale
131     bonus_shrink_factor = 1 - spine_part_of_page
132     new_page_order = []
133     new_i_order = []
134     eight_pack = []
135     mod_to_8 = len(pages_to_add) % 8
136     if mod_to_8 > 0:
137         for _ in range(8 - mod_to_8):
138             new_page = pypdf.PageObject.create_blank_page(width=a4_width, height=a4_height)
139             pages_to_add += [new_page]
140     i = 0
141     n_eights = 0
142     for page in pages_to_add:
143         if i == 0:
144             eight_pack = []
145         eight_pack += [page]
146         i += 1
147         if i == 8:
148             i = 0
149             new_i_order += [8 * n_eights + 3,
150                             8 * n_eights + 0,
151                             8 * n_eights + 7,
152                             8 * n_eights + 4,
153                             8 * n_eights + 1,
154                             8 * n_eights + 2,
155                             8 * n_eights + 5,
156                             8 * n_eights + 6]
157             n_eights += 1
158             new_page_order += [eight_pack[3]]  # page front, upper left
159             new_page_order += [eight_pack[0]]  # page front, upper right
160             new_page_order += [eight_pack[7]]  # page front, lower left
161             new_page_order += [eight_pack[4]]  # page front, lower right
162             new_page_order += [eight_pack[1]]  # page back, upper left
163             new_page_order += [eight_pack[2]]  # page back, upper right
164             new_page_order += [eight_pack[5]]  # page back, lower left
165             new_page_order += [eight_pack[6]]  # page back, lower right
166     i = 0
167     page_count = 0
168     front_page = True
169     for j, page in enumerate(new_page_order):
170         if i == 0:
171             new_page = pypdf.PageObject.create_blank_page(width=a4_width, height=a4_height)
172
173         # in-section transformations: align pages on top, left-hand pages to left, right-hand to right
174         new_i = new_i_order[j]
175         crop_left, crop_bottom, crop_right, crop_top = crops_at_page[new_i]
176         zoom = zoom_at_page[new_i]
177         page.add_transformation(pypdf.Transformation().translate(ty=(a4_height / zoom - (a4_height - crop_top))))
178         if i == 0 or i == 2:
179             if args.symmetry:
180                 page.add_transformation(pypdf.Transformation().translate(tx=-crop_left))
181             else:
182                 page.add_transformation(pypdf.Transformation().translate(tx=-crop_right))
183         elif i == 1 or i == 3:
184             if args.symmetry:
185                 page.add_transformation(pypdf.Transformation().translate(tx=(a4_width / zoom - (a4_width - crop_left))))
186             else:
187                 page.add_transformation(pypdf.Transformation().translate(tx=(a4_width / zoom - (a4_width - crop_right))))
188         page.add_transformation(pypdf.Transformation().scale(zoom * bonus_shrink_factor, zoom * bonus_shrink_factor))
189         if i == 2 or i == 3:
190             page.add_transformation(pypdf.Transformation().translate(ty=-2*printable_margin/printable_scale))
191
192         # outer section transformations
193         page.add_transformation(pypdf.Transformation().translate(ty=(1-bonus_shrink_factor)*a4_height))
194         if i == 0 or i == 1:
195             y_section = a4_height
196             page.mediabox.bottom = half_height
197             page.mediabox.top    = a4_height
198         if i == 2 or i == 3:
199             y_section = 0
200             page.mediabox.bottom = 0
201             page.mediabox.top  =   half_height
202         if i == 0 or i == 2:
203             x_section = 0
204             page.mediabox.left   = 0
205             page.mediabox.right  = half_width
206         if i == 1 or i == 3:
207             page.add_transformation(pypdf.Transformation().translate(tx=(1-bonus_shrink_factor)*a4_width))
208             x_section = a4_width
209             page.mediabox.left   = half_width
210             page.mediabox.right  = a4_width
211         page.add_transformation(pypdf.Transformation().translate(tx=x_section, ty=y_section))
212         page.add_transformation(pypdf.Transformation().scale(section_scale_factor, section_scale_factor))
213         new_page.merge_page(page)
214         page_count += 1
215         print("merged page number", page_count)
216         i += 1
217         if i > 3:
218             from reportlab.pdfgen import canvas
219             if args.analyze:
220                 # borders
221                 packet = io.BytesIO()
222                 c = canvas.Canvas(packet, pagesize=A4)
223                 c.setLineWidth(0.1)
224                 c.line(0, a4_height, a4_width, a4_height)
225                 c.line(0, half_height, a4_width, half_height)
226                 c.line(0, 0, a4_width, 0)
227                 c.line(0, a4_height, 0, 0)
228                 c.line(half_width, a4_height, half_width, 0)
229                 c.line(a4_width, a4_height, a4_width, 0)
230                 c.save()
231                 new_pdf = pypdf.PdfReader(packet)
232                 new_page.merge_page(new_pdf.pages[0])
233             printable_offset_x = printable_margin
234             printable_offset_y = printable_margin * a4_height / a4_width
235             new_page.add_transformation(pypdf.Transformation().scale(printable_scale, printable_scale))
236             new_page.add_transformation(pypdf.Transformation().translate(tx=printable_offset_x, ty=printable_offset_y))
237             x_left_spine_limit = half_width * bonus_shrink_factor
238             x_right_spine_limit = a4_width - x_left_spine_limit
239             if args.analyze or front_page:
240                 packet = io.BytesIO()
241                 c = canvas.Canvas(packet, pagesize=A4)
242             if args.analyze:
243                 # # spine lines
244                 c.setLineWidth(0.1)
245                 c.line(x_left_spine_limit, a4_height, x_left_spine_limit, 0)
246                 c.line(x_right_spine_limit, a4_height, x_right_spine_limit, 0)
247             if front_page:
248                 c.setLineWidth(0.2)
249
250                 start_up_left_left_x = x_left_spine_limit - 0.5 * cut_width
251                 start_up_left_right_x = x_left_spine_limit + 0.5 * cut_width
252                 middle_point_up_left_y = half_height + middle_point_depth
253                 end_point_up_left_y = half_height + cut_depth
254                 c.line(start_up_left_right_x, half_height, x_left_spine_limit, end_point_up_left_y)
255                 c.line(x_left_spine_limit, end_point_up_left_y, x_left_spine_limit, middle_point_up_left_y)
256                 c.line(x_left_spine_limit, middle_point_up_left_y, start_up_left_left_x, half_height)
257
258                 start_down_right_left_x = x_right_spine_limit - 0.5 * cut_width
259                 start_down_right_right_x = x_right_spine_limit + 0.5 * cut_width
260                 middle_point_down_right_y = half_height - middle_point_depth
261                 end_point_down_right_y = half_height - cut_depth
262                 c.line(start_down_right_left_x, half_height, x_right_spine_limit, end_point_down_right_y)
263                 c.line(x_right_spine_limit, end_point_down_right_y, x_right_spine_limit, middle_point_down_right_y)
264                 c.line(x_right_spine_limit, middle_point_down_right_y, start_down_right_right_x, half_height)
265
266             if args.analyze or front_page:
267                 c.save()
268                 new_pdf = pypdf.PdfReader(packet)
269                 new_page.merge_page(new_pdf.pages[0])
270             writer.add_page(new_page)
271             i = 0
272             front_page = not front_page
273
274 # write and close
275 for file in opened_files:
276     file.close()
277 with open(args.output_file, 'wb') as output_file:
278     writer.write(output_file)