plomlompom.com Git - misc/blob - 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 spine_limit = 1 * points_per_cm
  12
  13 parser = argparse.ArgumentParser(description="build print-ready book PDF")
  14 parser.add_argument("-i", "--input_file", action="append", required=True, help="input PDF file")
  15 parser.add_argument("-o", "--output_file", required=True, help="output PDF file")
  16 parser.add_argument("-p", "--page_range", action="append", help="page range, e.g., '3-end'")
  17 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")
  18 parser.add_argument("-s", "--symmetry", action="store_true", help="alternate horizontal crops between odd and even pages")
  19 parser.add_argument("-r", "--rotate", dest="rotate", type=int, action="append", help="rotate page of number by 90° (usable multiple times on same page!)")
  20 parser.add_argument("-n", "--nup4", action='store_true', help="puts 4 input pages onto 1 output page")
  21 parser.add_argument("-a", "--analyze", action="store_true", help="in --nup4, print lines identifying spine, page borders")
  22 parser.add_argument("-m", "--margin", type=float, default=0.43, help="print margin for --nup4 in cm (default 0.43)")
  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 new_page_num = 0
  40 for i, input_file in enumerate(args.input_file):
  41     file = open(input_file, 'rb')
  42     opened_files += [file]
  43     reader = pypdf.PdfReader(file)
  44     range_string = None
  45     if args.page_range and len(args.page_range) > i:
  46         range_string = args.page_range[i]
  47     start_page, end_page = parse_page_range(range_string, reader.pages)
  48     for old_page_num in range(start_page, end_page):
  49         new_page_num += 1
  50         page = reader.pages[old_page_num]
  51         pages_to_add += [page]
  52         print("-i, -p: read in %s page number %d as new page %d" % (input_file, old_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, crop symmetry
  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       if args.symmetry:
  88           print("-c, -t: to pages %d to %d applying crops: left %dcm, bottom %dcm, right %dcm, top %dcm (but alternating left and right crop between even and odd pages)" % (start_page + 1, end_page, crop_left, crop_bottom, crop_right, crop_top))
  89       else:
  90           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))
  91       cropped_width  = a4_width - crop_left - crop_right
  92       cropped_height = a4_height - crop_bottom - crop_top
  93       zoom = 1
  94       zoom_horizontal = a4_width / (a4_width - crop_left - crop_right)
  95       zoom_vertical = a4_height / (a4_height - crop_bottom - crop_top)
  96       if (zoom_horizontal > 1 and zoom_vertical < 1) or (zoom_horizontal < 1 and zoom_vertical > 1):
  97           print("Error: opposing zooms.")
  98           exit(1)
  99       elif zoom_horizontal + zoom_vertical > 2:
 100           zoom = min(zoom_horizontal, zoom_vertical)
 101       else:
 102           zoom = max(zoom_horizontal, zoom_vertical)
 103       for page_num in range(start_page, end_page):
 104           if args.symmetry and page_num % 2:
 105               crops_at_page[page_num] = (crop_right, crop_bottom, crop_left, crop_top)
 106           else:
 107               crops_at_page[page_num] = (crop_left, crop_bottom, crop_right, crop_top)
 108           zoom_at_page[page_num] = zoom
 109
 110 writer = pypdf.PdfWriter()
 111 if not args.nup4:
 112     # single-page output
 113     print("building 1-input-page-per-output-page book")
 114     odd_page = True
 115     for i, page in enumerate(pages_to_add):
 116         crop_left, crop_bottom, crop_right, crop_top = crops_at_page[i]
 117         zoom = zoom_at_page[i]
 118         page.add_transformation(pypdf.Transformation().translate(tx=-crop_left, ty=-crop_bottom))
 119         page.add_transformation(pypdf.Transformation().scale(zoom, zoom))
 120         cropped_width  = a4_width - crop_left - crop_right
 121         cropped_height = a4_height - crop_bottom - crop_top
 122         page.mediabox.right = cropped_width * zoom
 123         page.mediabox.top = cropped_height * zoom
 124         writer.add_page(page)
 125         odd_page = not odd_page
 126
 127 else:
 128     print("-n: building 4-input-pages-per-output-page book")
 129     n_pages_per_axis = 2
 130     printable_margin = args.margin * points_per_cm
 131     printable_scale = (a4_width - 2*printable_margin)/a4_width
 132     half_width = a4_width / n_pages_per_axis
 133     half_height = a4_height / n_pages_per_axis
 134     section_scale_factor = 1 / n_pages_per_axis
 135     spine_part_of_page = (spine_limit / half_width) / printable_scale
 136     bonus_shrink_factor = 1 - spine_part_of_page
 137     new_page_order = []
 138     new_i_order = []
 139     eight_pack = []
 140     mod_to_8 = len(pages_to_add) % 8
 141     if mod_to_8 > 0:
 142         for _ in range(8 - mod_to_8):
 143             new_page = pypdf.PageObject.create_blank_page(width=a4_width, height=a4_height)
 144             pages_to_add += [new_page]
 145     i = 0
 146     n_eights = 0
 147     for page in pages_to_add:
 148         if i == 0:
 149             eight_pack = []
 150         eight_pack += [page]
 151         i += 1
 152         if i == 8:
 153             i = 0
 154             new_i_order += [8 * n_eights + 3,
 155                             8 * n_eights + 0,
 156                             8 * n_eights + 7,
 157                             8 * n_eights + 4,
 158                             8 * n_eights + 1,
 159                             8 * n_eights + 2,
 160                             8 * n_eights + 5,
 161                             8 * n_eights + 6]
 162             n_eights += 1
 163             new_page_order += [eight_pack[3]]  # page front, upper left
 164             new_page_order += [eight_pack[0]]  # page front, upper right
 165             new_page_order += [eight_pack[7]]  # page front, lower left
 166             new_page_order += [eight_pack[4]]  # page front, lower right
 167             new_page_order += [eight_pack[1]]  # page back, upper left
 168             new_page_order += [eight_pack[2]]  # page back, upper right
 169             new_page_order += [eight_pack[5]]  # page back, lower left
 170             new_page_order += [eight_pack[6]]  # page back, lower right
 171     i = 0
 172     page_count = 0
 173     front_page = True
 174     for j, page in enumerate(new_page_order):
 175         if i == 0:
 176             new_page = pypdf.PageObject.create_blank_page(width=a4_width, height=a4_height)
 177
 178         # in-section transformations: align pages on top, left-hand pages to left, right-hand to right
 179         new_i = new_i_order[j]
 180         crop_left, crop_bottom, crop_right, crop_top = crops_at_page[new_i]
 181         zoom = zoom_at_page[new_i]
 182         page.add_transformation(pypdf.Transformation().translate(ty=(a4_height / zoom - (a4_height - crop_top))))
 183         if i == 0 or i == 2:
 184             page.add_transformation(pypdf.Transformation().translate(tx=-crop_left))
 185         elif i == 1 or i == 3:
 186             page.add_transformation(pypdf.Transformation().translate(tx=(a4_width / zoom - (a4_width - crop_right))))
 187         page.add_transformation(pypdf.Transformation().scale(zoom * bonus_shrink_factor, zoom * bonus_shrink_factor))
 188         if i == 2 or i == 3:
 189             page.add_transformation(pypdf.Transformation().translate(ty=-2*printable_margin/printable_scale))
 190
 191         # outer section transformations
 192         page.add_transformation(pypdf.Transformation().translate(ty=(1-bonus_shrink_factor)*a4_height))
 193         if i == 0 or i == 1:
 194             y_section = a4_height
 195             page.mediabox.bottom = half_height
 196             page.mediabox.top    = a4_height
 197         if i == 2 or i == 3:
 198             y_section = 0
 199             page.mediabox.bottom = 0
 200             page.mediabox.top  =   half_height
 201         if i == 0 or i == 2:
 202             x_section = 0
 203             page.mediabox.left   = 0
 204             page.mediabox.right  = half_width
 205         if i == 1 or i == 3:
 206             page.add_transformation(pypdf.Transformation().translate(tx=(1-bonus_shrink_factor)*a4_width))
 207             x_section = a4_width
 208             page.mediabox.left   = half_width
 209             page.mediabox.right  = a4_width
 210         page.add_transformation(pypdf.Transformation().translate(tx=x_section, ty=y_section))
 211         page.add_transformation(pypdf.Transformation().scale(section_scale_factor, section_scale_factor))
 212         new_page.merge_page(page)
 213         page_count += 1
 214         print("merged page number", page_count)
 215         i += 1
 216         if i > 3:
 217             from reportlab.pdfgen import canvas
 218             if args.analyze:
 219                 # borders
 220                 packet = io.BytesIO()
 221                 c = canvas.Canvas(packet, pagesize=A4)
 222                 c.setLineWidth(0.1)
 223                 c.line(0, a4_height, a4_width, a4_height)
 224                 c.line(0, half_height, a4_width, half_height)
 225                 c.line(0, 0, a4_width, 0)
 226                 c.line(0, a4_height, 0, 0)
 227                 c.line(half_width, a4_height, half_width, 0)
 228                 c.line(a4_width, a4_height, a4_width, 0)
 229                 c.save()
 230                 new_pdf = pypdf.PdfReader(packet)
 231                 new_page.merge_page(new_pdf.pages[0])
 232             printable_offset_x = printable_margin
 233             printable_offset_y = printable_margin * a4_height / a4_width
 234             new_page.add_transformation(pypdf.Transformation().scale(printable_scale, printable_scale))
 235             new_page.add_transformation(pypdf.Transformation().translate(tx=printable_offset_x, ty=printable_offset_y))
 236             x_left_spine_limit = half_width * bonus_shrink_factor
 237             x_right_spine_limit = a4_width - x_left_spine_limit
 238             if args.analyze or front_page:
 239                 packet = io.BytesIO()
 240                 c = canvas.Canvas(packet, pagesize=A4)
 241             if args.analyze:
 242                 # # spine lines
 243                 c.setLineWidth(0.1)
 244                 c.line(x_left_spine_limit, a4_height, x_left_spine_limit, 0)
 245                 c.line(x_right_spine_limit, a4_height, x_right_spine_limit, 0)
 246             if front_page:
 247                 c.setLineWidth(0.2)
 248
 249                 start_up_left_left_x = x_left_spine_limit - 0.5 * cut_width
 250                 start_up_left_right_x = x_left_spine_limit + 0.5 * cut_width
 251                 middle_point_up_left_y = half_height + middle_point_depth
 252                 end_point_up_left_y = half_height + cut_depth
 253                 c.line(start_up_left_right_x, half_height, x_left_spine_limit, end_point_up_left_y)
 254                 c.line(x_left_spine_limit, end_point_up_left_y, x_left_spine_limit, middle_point_up_left_y)
 255                 c.line(x_left_spine_limit, middle_point_up_left_y, start_up_left_left_x, half_height)
 256
 257                 start_down_right_left_x = x_right_spine_limit - 0.5 * cut_width
 258                 start_down_right_right_x = x_right_spine_limit + 0.5 * cut_width
 259                 middle_point_down_right_y = half_height - middle_point_depth
 260                 end_point_down_right_y = half_height - cut_depth
 261                 c.line(start_down_right_left_x, half_height, x_right_spine_limit, end_point_down_right_y)
 262                 c.line(x_right_spine_limit, end_point_down_right_y, x_right_spine_limit, middle_point_down_right_y)
 263                 c.line(x_right_spine_limit, middle_point_down_right_y, start_down_right_right_x, half_height)
 264
 265             if args.analyze or front_page:
 266                 c.save()
 267                 new_pdf = pypdf.PdfReader(packet)
 268                 new_page.merge_page(new_pdf.pages[0])
 269             writer.add_page(new_page)
 270             i = 0
 271             front_page = not front_page
 272
 273 # write and close
 274 for file in opened_files:
 275     file.close()
 276 with open(args.output_file, 'wb') as output_file:
 277     writer.write(output_file)