gfx/harfbuzz/src/gen-arabic-table.py

   1 #!/usr/bin/env python3
   2
   3 """usage: ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt Blocks.txt
   4
   5 Input files:
   6 * https://unicode.org/Public/UCD/latest/ucd/ArabicShaping.txt
   7 * https://unicode.org/Public/UCD/latest/ucd/UnicodeData.txt
   8 * https://unicode.org/Public/UCD/latest/ucd/Blocks.txt
   9 """
  10
  11 import os.path, sys
  12
  13 if len (sys.argv) != 4:
  14         sys.exit (__doc__)
  15
  16 files = [open (x, encoding='utf-8') for x in sys.argv[1:]]
  17
  18 headers = [[files[0].readline (), files[0].readline ()], [files[2].readline (), files[2].readline ()]]
  19 headers.append (["UnicodeData.txt does not have a header."])
  20 while files[0].readline ().find ('##################') < 0:
  21         pass
  22
  23 blocks = {}
  24 def read_blocks(f):
  25         global blocks
  26         for line in f:
  27
  28                 j = line.find ('#')
  29                 if j >= 0:
  30                         line = line[:j]
  31
  32                 fields = [x.strip () for x in line.split (';')]
  33                 if len (fields) == 1:
  34                         continue
  35
  36                 uu = fields[0].split ('..')
  37                 start = int (uu[0], 16)
  38                 if len (uu) == 1:
  39                         end = start
  40                 else:
  41                         end = int (uu[1], 16)
  42
  43                 t = fields[1]
  44
  45                 for u in range (start, end + 1):
  46                         blocks[u] = t
  47
  48 def print_joining_table(f):
  49
  50         values = {}
  51         for line in f:
  52
  53                 if line[0] == '#':
  54                         continue
  55
  56                 fields = [x.strip () for x in line.split (';')]
  57                 if len (fields) == 1:
  58                         continue
  59
  60                 u = int (fields[0], 16)
  61
  62                 if fields[3] in ["ALAPH", "DALATH RISH"]:
  63                         value = "JOINING_GROUP_" + fields[3].replace(' ', '_')
  64                 else:
  65                         value = "JOINING_TYPE_" + fields[2]
  66                 values[u] = value
  67
  68         short_value = {}
  69         for value in sorted (set ([v for v in values.values ()] + ['JOINING_TYPE_X'])):
  70                 short = ''.join(x[0] for x in value.split('_')[2:])
  71                 assert short not in short_value.values()
  72                 short_value[value] = short
  73
  74         print ()
  75         for value,short in short_value.items():
  76                 print ("#define %s      %s" % (short, value))
  77
  78         uu = sorted(values.keys())
  79         num = len(values)
  80         all_blocks = set([blocks[u] for u in uu])
  81
  82         last = -100000
  83         ranges = []
  84         for u in uu:
  85                 if u - last <= 1+16*5:
  86                         ranges[-1][-1] = u
  87                 else:
  88                         ranges.append([u,u])
  89                 last = u
  90
  91         print ()
  92         print ("static const uint8_t joining_table[] =")
  93         print ("{")
  94         last_block = None
  95         offset = 0
  96         for start,end in ranges:
  97
  98                 print ()
  99                 print ("#define joining_offset_0x%04xu %d" % (start, offset))
 100
 101                 for u in range(start, end+1):
 102
 103                         block = blocks.get(u, last_block)
 104                         value = values.get(u, "JOINING_TYPE_X")
 105
 106                         if block != last_block or u == start:
 107                                 if u != start:
 108                                         print ()
 109                                 if block in all_blocks:
 110                                         print ("\n  /* %s */" % block)
 111                                 else:
 112                                         print ("\n  /* FILLER */")
 113                                 last_block = block
 114                                 if u % 32 != 0:
 115                                         print ()
 116                                         print ("  /* %04X */" % (u//32*32), "  " * (u % 32), end="")
 117
 118                         if u % 32 == 0:
 119                                 print ()
 120                                 print ("  /* %04X */ " % u, end="")
 121                         print ("%s," % short_value[value], end="")
 122                 print ()
 123
 124                 offset += end - start + 1
 125         print ()
 126         occupancy = num * 100. / offset
 127         print ("}; /* Table items: %d; occupancy: %d%% */" % (offset, occupancy))
 128         print ()
 129
 130         page_bits = 12
 131         print ()
 132         print ("static unsigned int")
 133         print ("joining_type (hb_codepoint_t u)")
 134         print ("{")
 135         print ("  switch (u >> %d)" % page_bits)
 136         print ("  {")
 137         pages = set([u>>page_bits for u in [s for s,e in ranges]+[e for s,e in ranges]])
 138         for p in sorted(pages):
 139                 print ("    case 0x%0Xu:" % p)
 140                 for (start,end) in ranges:
 141                         if p not in [start>>page_bits, end>>page_bits]: continue
 142                         offset = "joining_offset_0x%04xu" % start
 143                         print ("      if (hb_in_range<hb_codepoint_t> (u, 0x%04Xu, 0x%04Xu)) return joining_table[u - 0x%04Xu + %s];" % (start, end, start, offset))
 144                 print ("      break;")
 145                 print ("")
 146         print ("    default:")
 147         print ("      break;")
 148         print ("  }")
 149         print ("  return X;")
 150         print ("}")
 151         print ()
 152         for value,short in short_value.items():
 153                 print ("#undef %s" % (short))
 154         print ()
 155
 156 def print_shaping_table(f):
 157
 158         shapes = {}
 159         ligatures = {}
 160         names = {}
 161         for line in f:
 162
 163                 fields = [x.strip () for x in line.split (';')]
 164                 if fields[5][0:1] != '<':
 165                         continue
 166
 167                 items = fields[5].split (' ')
 168                 shape, items = items[0][1:-1], tuple (int (x, 16) for x in items[1:])
 169
 170                 if not shape in ['initial', 'medial', 'isolated', 'final']:
 171                         continue
 172
 173                 c = int (fields[0], 16)
 174                 if len (items) != 1:
 175                         # We only care about lam-alef ligatures
 176                         if len (items) != 2 or items[0] != 0x0644 or items[1] not in [0x0622, 0x0623, 0x0625, 0x0627]:
 177                                 continue
 178
 179                         # Save ligature
 180                         names[c] = fields[1]
 181                         if items not in ligatures:
 182                                 ligatures[items] = {}
 183                         ligatures[items][shape] = c
 184                 else:
 185                         # Save shape
 186                         if items[0] not in names:
 187                                 names[items[0]] = fields[1]
 188                         else:
 189                                 names[items[0]] = os.path.commonprefix ([names[items[0]], fields[1]]).strip ()
 190                         if items[0] not in shapes:
 191                                 shapes[items[0]] = {}
 192                         shapes[items[0]][shape] = c
 193
 194         print ()
 195         print ("static const uint16_t shaping_table[][4] =")
 196         print ("{")
 197
 198         keys = shapes.keys ()
 199         min_u, max_u = min (keys), max (keys)
 200         for u in range (min_u, max_u + 1):
 201                 s = [shapes[u][shape] if u in shapes and shape in shapes[u] else 0
 202                      for shape in  ['initial', 'medial', 'final', 'isolated']]
 203                 value = ', '.join ("0x%04Xu" % c for c in s)
 204                 print ("  {%s}, /* U+%04X %s */" % (value, u, names[u] if u in names else ""))
 205
 206         print ("};")
 207         print ()
 208         print ("#define SHAPING_TABLE_FIRST     0x%04Xu" % min_u)
 209         print ("#define SHAPING_TABLE_LAST      0x%04Xu" % max_u)
 210         print ()
 211
 212         ligas = {}
 213         for pair in ligatures.keys ():
 214                 for shape in ligatures[pair]:
 215                         c = ligatures[pair][shape]
 216                         if shape == 'isolated':
 217                                 liga = (shapes[pair[0]]['initial'], shapes[pair[1]]['final'])
 218                         elif shape == 'final':
 219                                 liga = (shapes[pair[0]]['medial'], shapes[pair[1]]['final'])
 220                         else:
 221                                 raise Exception ("Unexpected shape", shape)
 222                         if liga[0] not in ligas:
 223                                 ligas[liga[0]] = []
 224                         ligas[liga[0]].append ((liga[1], c))
 225         max_i = max (len (ligas[l]) for l in ligas)
 226         print ()
 227         print ("static const struct ligature_set_t {")
 228         print (" uint16_t first;")
 229         print (" struct ligature_pairs_t {")
 230         print ("   uint16_t second;")
 231         print ("   uint16_t ligature;")
 232         print (" } ligatures[%d];" % max_i)
 233         print ("} ligature_table[] =")
 234         print ("{")
 235         for first in sorted (ligas.keys ()):
 236
 237                 print ("  { 0x%04Xu, {" % (first))
 238                 for liga in ligas[first]:
 239                         print ("    { 0x%04Xu, 0x%04Xu }, /* %s */" % (liga[0], liga[1], names[liga[1]]))
 240                 print ("  }},")
 241
 242         print ("};")
 243         print ()
 244
 245
 246
 247 print ("/* == Start of generated table == */")
 248 print ("/*")
 249 print (" * The following table is generated by running:")
 250 print (" *")
 251 print (" *   ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt Blocks.txt")
 252 print (" *")
 253 print (" * on files with these headers:")
 254 print (" *")
 255 for h in headers:
 256         for l in h:
 257                 print (" * %s" % (l.strip()))
 258 print (" */")
 259 print ()
 260 print ("#ifndef HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH")
 261 print ("#define HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH")
 262 print ()
 263
 264 read_blocks (files[2])
 265 print_joining_table (files[0])
 266 print_shaping_table (files[1])
 267
 268 print ()
 269 print ("#endif /* HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH */")
 270 print ()
 271 print ("/* == End of generated table == */")