Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 1 | #!/usr/bin/env python |
| 2 | |
| 3 | from __future__ import print_function, division, absolute_import |
Behdad Esfahbod | 3eb936f | 2010-10-05 18:36:58 -0400 | [diff] [blame] | 4 | |
Ebrahim Byagowi | 80395f1 | 2018-03-29 22:00:41 +0430 | [diff] [blame] | 5 | import io, os.path, sys |
Behdad Esfahbod | 3eb936f | 2010-10-05 18:36:58 -0400 | [diff] [blame] | 6 | |
Behdad Esfahbod | 3f5327a | 2014-06-20 16:17:42 -0400 | [diff] [blame] | 7 | if len (sys.argv) != 4: |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 8 | print ("usage: ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt Blocks.txt", file=sys.stderr) |
Behdad Esfahbod | 697a65c | 2011-06-01 20:52:00 -0400 | [diff] [blame] | 9 | sys.exit (1) |
| 10 | |
Ebrahim Byagowi | 80395f1 | 2018-03-29 22:00:41 +0430 | [diff] [blame] | 11 | files = [io.open (x, encoding='utf-8') for x in sys.argv[1:]] |
Behdad Esfahbod | 697a65c | 2011-06-01 20:52:00 -0400 | [diff] [blame] | 12 | |
Behdad Esfahbod | 3f5327a | 2014-06-20 16:17:42 -0400 | [diff] [blame] | 13 | headers = [[files[0].readline (), files[0].readline ()], [files[2].readline (), files[2].readline ()]] |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 14 | headers.append (["UnicodeData.txt does not have a header."]) |
| 15 | while files[0].readline ().find ('##################') < 0: |
Behdad Esfahbod | 88e7f37 | 2010-12-21 14:18:24 -0500 | [diff] [blame] | 16 | pass |
Behdad Esfahbod | 3eb936f | 2010-10-05 18:36:58 -0400 | [diff] [blame] | 17 | |
Behdad Esfahbod | 3f5327a | 2014-06-20 16:17:42 -0400 | [diff] [blame] | 18 | blocks = {} |
| 19 | def read_blocks(f): |
| 20 | global blocks |
| 21 | for line in f: |
| 22 | |
| 23 | j = line.find ('#') |
| 24 | if j >= 0: |
| 25 | line = line[:j] |
| 26 | |
| 27 | fields = [x.strip () for x in line.split (';')] |
| 28 | if len (fields) == 1: |
| 29 | continue |
| 30 | |
| 31 | uu = fields[0].split ('..') |
| 32 | start = int (uu[0], 16) |
| 33 | if len (uu) == 1: |
| 34 | end = start |
| 35 | else: |
| 36 | end = int (uu[1], 16) |
| 37 | |
| 38 | t = fields[1] |
| 39 | |
| 40 | for u in range (start, end + 1): |
| 41 | blocks[u] = t |
Behdad Esfahbod | 14d7841 | 2010-11-17 16:52:58 -0500 | [diff] [blame] | 42 | |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 43 | def print_joining_table(f): |
| 44 | |
Behdad Esfahbod | 3f5327a | 2014-06-20 16:17:42 -0400 | [diff] [blame] | 45 | values = {} |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 46 | for line in f: |
| 47 | |
| 48 | if line[0] == '#': |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 49 | continue |
| 50 | |
| 51 | fields = [x.strip () for x in line.split (';')] |
| 52 | if len (fields) == 1: |
| 53 | continue |
| 54 | |
| 55 | u = int (fields[0], 16) |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 56 | |
| 57 | if fields[3] in ["ALAPH", "DALATH RISH"]: |
| 58 | value = "JOINING_GROUP_" + fields[3].replace(' ', '_') |
| 59 | else: |
| 60 | value = "JOINING_TYPE_" + fields[2] |
Behdad Esfahbod | f886707 | 2014-06-20 16:30:10 -0400 | [diff] [blame] | 61 | values[u] = value |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 62 | |
Behdad Esfahbod | 200dfe3 | 2014-06-20 16:20:59 -0400 | [diff] [blame] | 63 | short_value = {} |
Behdad Esfahbod | f886707 | 2014-06-20 16:30:10 -0400 | [diff] [blame] | 64 | for value in set([v for v in values.values()] + ['JOINING_TYPE_X']): |
Behdad Esfahbod | 200dfe3 | 2014-06-20 16:20:59 -0400 | [diff] [blame] | 65 | short = ''.join(x[0] for x in value.split('_')[2:]) |
| 66 | assert short not in short_value.values() |
| 67 | short_value[value] = short |
| 68 | |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 69 | print () |
Behdad Esfahbod | 200dfe3 | 2014-06-20 16:20:59 -0400 | [diff] [blame] | 70 | for value,short in short_value.items(): |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 71 | print ("#define %s %s" % (short, value)) |
Behdad Esfahbod | 200dfe3 | 2014-06-20 16:20:59 -0400 | [diff] [blame] | 72 | |
Behdad Esfahbod | b900fa2 | 2014-06-20 17:59:43 -0400 | [diff] [blame] | 73 | uu = sorted(values.keys()) |
Behdad Esfahbod | 3f5327a | 2014-06-20 16:17:42 -0400 | [diff] [blame] | 74 | num = len(values) |
Behdad Esfahbod | 2390d9b | 2014-06-21 14:07:00 -0600 | [diff] [blame] | 75 | all_blocks = set([blocks[u] for u in uu]) |
Behdad Esfahbod | b900fa2 | 2014-06-20 17:59:43 -0400 | [diff] [blame] | 76 | |
Behdad Esfahbod | dcee838 | 2014-06-22 11:29:59 -0600 | [diff] [blame] | 77 | last = -100000 |
Behdad Esfahbod | b900fa2 | 2014-06-20 17:59:43 -0400 | [diff] [blame] | 78 | ranges = [] |
| 79 | for u in uu: |
Behdad Esfahbod | 2390d9b | 2014-06-21 14:07:00 -0600 | [diff] [blame] | 80 | if u - last <= 1+16*5: |
Behdad Esfahbod | b900fa2 | 2014-06-20 17:59:43 -0400 | [diff] [blame] | 81 | ranges[-1][-1] = u |
| 82 | else: |
| 83 | ranges.append([u,u]) |
| 84 | last = u |
| 85 | |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 86 | print () |
| 87 | print ("static const uint8_t joining_table[] =") |
| 88 | print ("{") |
Behdad Esfahbod | 3f5327a | 2014-06-20 16:17:42 -0400 | [diff] [blame] | 89 | last_block = None |
Behdad Esfahbod | b900fa2 | 2014-06-20 17:59:43 -0400 | [diff] [blame] | 90 | offset = 0 |
| 91 | for start,end in ranges: |
Behdad Esfahbod | f886707 | 2014-06-20 16:30:10 -0400 | [diff] [blame] | 92 | |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 93 | print () |
| 94 | print ("#define joining_offset_0x%04xu %d" % (start, offset)) |
Behdad Esfahbod | f886707 | 2014-06-20 16:30:10 -0400 | [diff] [blame] | 95 | |
Behdad Esfahbod | b900fa2 | 2014-06-20 17:59:43 -0400 | [diff] [blame] | 96 | for u in range(start, end+1): |
| 97 | |
| 98 | block = blocks.get(u, last_block) |
| 99 | value = values.get(u, "JOINING_TYPE_X") |
| 100 | |
| 101 | if block != last_block or u == start: |
| 102 | if u != start: |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 103 | print () |
Behdad Esfahbod | 2390d9b | 2014-06-21 14:07:00 -0600 | [diff] [blame] | 104 | if block in all_blocks: |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 105 | print ("\n /* %s */" % block) |
Behdad Esfahbod | 2390d9b | 2014-06-21 14:07:00 -0600 | [diff] [blame] | 106 | else: |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 107 | print ("\n /* FILLER */") |
Behdad Esfahbod | b900fa2 | 2014-06-20 17:59:43 -0400 | [diff] [blame] | 108 | last_block = block |
| 109 | if u % 32 != 0: |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 110 | print () |
| 111 | print (" /* %04X */" % (u//32*32), " " * (u % 32), end="") |
Behdad Esfahbod | b900fa2 | 2014-06-20 17:59:43 -0400 | [diff] [blame] | 112 | |
| 113 | if u % 32 == 0: |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 114 | print () |
| 115 | print (" /* %04X */ " % u, end="") |
| 116 | print ("%s," % short_value[value], end="") |
| 117 | print () |
Behdad Esfahbod | 3f5327a | 2014-06-20 16:17:42 -0400 | [diff] [blame] | 118 | |
Behdad Esfahbod | b900fa2 | 2014-06-20 17:59:43 -0400 | [diff] [blame] | 119 | offset += end - start + 1 |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 120 | print () |
Behdad Esfahbod | b900fa2 | 2014-06-20 17:59:43 -0400 | [diff] [blame] | 121 | occupancy = num * 100. / offset |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 122 | print ("}; /* Table items: %d; occupancy: %d%% */" % (offset, occupancy)) |
| 123 | print () |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 124 | |
Behdad Esfahbod | 2390d9b | 2014-06-21 14:07:00 -0600 | [diff] [blame] | 125 | page_bits = 12; |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 126 | print () |
| 127 | print ("static unsigned int") |
| 128 | print ("joining_type (hb_codepoint_t u)") |
| 129 | print ("{") |
| 130 | print (" switch (u >> %d)" % page_bits) |
| 131 | print (" {") |
Behdad Esfahbod | b900fa2 | 2014-06-20 17:59:43 -0400 | [diff] [blame] | 132 | pages = set([u>>page_bits for u in [s for s,e in ranges]+[e for s,e in ranges]]) |
| 133 | for p in sorted(pages): |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 134 | print (" case 0x%0Xu:" % p) |
Behdad Esfahbod | b900fa2 | 2014-06-20 17:59:43 -0400 | [diff] [blame] | 135 | for (start,end) in ranges: |
| 136 | if p not in [start>>page_bits, end>>page_bits]: continue |
Behdad Esfahbod | c09a607 | 2014-07-11 15:05:36 -0400 | [diff] [blame] | 137 | offset = "joining_offset_0x%04xu" % start |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 138 | print (" if (hb_in_range<hb_codepoint_t> (u, 0x%04Xu, 0x%04Xu)) return joining_table[u - 0x%04Xu + %s];" % (start, end, start, offset)) |
| 139 | print (" break;") |
| 140 | print ("") |
| 141 | print (" default:") |
| 142 | print (" break;") |
| 143 | print (" }") |
| 144 | print (" return X;") |
| 145 | print ("}") |
| 146 | print () |
Behdad Esfahbod | 200dfe3 | 2014-06-20 16:20:59 -0400 | [diff] [blame] | 147 | for value,short in short_value.items(): |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 148 | print ("#undef %s" % (short)) |
| 149 | print () |
Behdad Esfahbod | 200dfe3 | 2014-06-20 16:20:59 -0400 | [diff] [blame] | 150 | |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 151 | def print_shaping_table(f): |
| 152 | |
| 153 | shapes = {} |
| 154 | ligatures = {} |
| 155 | names = {} |
| 156 | for line in f: |
| 157 | |
| 158 | fields = [x.strip () for x in line.split (';')] |
| 159 | if fields[5][0:1] != '<': |
| 160 | continue |
| 161 | |
| 162 | items = fields[5].split (' ') |
| 163 | shape, items = items[0][1:-1], tuple (int (x, 16) for x in items[1:]) |
| 164 | |
| 165 | if not shape in ['initial', 'medial', 'isolated', 'final']: |
| 166 | continue |
| 167 | |
| 168 | c = int (fields[0], 16) |
| 169 | if len (items) != 1: |
| 170 | # We only care about lam-alef ligatures |
| 171 | if len (items) != 2 or items[0] != 0x0644 or items[1] not in [0x0622, 0x0623, 0x0625, 0x0627]: |
| 172 | continue |
| 173 | |
| 174 | # Save ligature |
| 175 | names[c] = fields[1] |
| 176 | if items not in ligatures: |
| 177 | ligatures[items] = {} |
| 178 | ligatures[items][shape] = c |
| 179 | pass |
| 180 | else: |
| 181 | # Save shape |
| 182 | if items[0] not in names: |
| 183 | names[items[0]] = fields[1] |
| 184 | else: |
| 185 | names[items[0]] = os.path.commonprefix ([names[items[0]], fields[1]]).strip () |
| 186 | if items[0] not in shapes: |
| 187 | shapes[items[0]] = {} |
| 188 | shapes[items[0]][shape] = c |
| 189 | |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 190 | print () |
| 191 | print ("static const uint16_t shaping_table[][4] =") |
| 192 | print ("{") |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 193 | |
| 194 | keys = shapes.keys () |
| 195 | min_u, max_u = min (keys), max (keys) |
| 196 | for u in range (min_u, max_u + 1): |
Behdad Esfahbod | 07cfbe2 | 2012-09-06 01:16:39 -0400 | [diff] [blame] | 197 | s = [shapes[u][shape] if u in shapes and shape in shapes[u] else 0 |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 198 | for shape in ['initial', 'medial', 'final', 'isolated']] |
Behdad Esfahbod | c09a607 | 2014-07-11 15:05:36 -0400 | [diff] [blame] | 199 | value = ', '.join ("0x%04Xu" % c for c in s) |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 200 | print (" {%s}, /* U+%04X %s */" % (value, u, names[u] if u in names else "")) |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 201 | |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 202 | print ("};") |
| 203 | print () |
| 204 | print ("#define SHAPING_TABLE_FIRST 0x%04Xu" % min_u) |
| 205 | print ("#define SHAPING_TABLE_LAST 0x%04Xu" % max_u) |
| 206 | print () |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 207 | |
Behdad Esfahbod | 939c010 | 2012-04-10 17:20:05 -0400 | [diff] [blame] | 208 | ligas = {} |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 209 | for pair in ligatures.keys (): |
| 210 | for shape in ligatures[pair]: |
| 211 | c = ligatures[pair][shape] |
| 212 | if shape == 'isolated': |
| 213 | liga = (shapes[pair[0]]['initial'], shapes[pair[1]]['final']) |
| 214 | elif shape == 'final': |
| 215 | liga = (shapes[pair[0]]['medial'], shapes[pair[1]]['final']) |
| 216 | else: |
| 217 | raise Exception ("Unexpected shape", shape) |
Behdad Esfahbod | 939c010 | 2012-04-10 17:20:05 -0400 | [diff] [blame] | 218 | if liga[0] not in ligas: |
| 219 | ligas[liga[0]] = [] |
| 220 | ligas[liga[0]].append ((liga[1], c)) |
| 221 | max_i = max (len (ligas[l]) for l in ligas) |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 222 | print () |
| 223 | print ("static const struct ligature_set_t {") |
| 224 | print (" uint16_t first;") |
| 225 | print (" struct ligature_pairs_t {") |
| 226 | print (" uint16_t second;") |
| 227 | print (" uint16_t ligature;") |
| 228 | print (" } ligatures[%d];" % max_i) |
| 229 | print ("} ligature_table[] =") |
| 230 | print ("{") |
Ebrahim Byagowi | 26e0cbd | 2018-03-29 21:22:47 +0430 | [diff] [blame] | 231 | for first in sorted (ligas.keys ()): |
Behdad Esfahbod | 939c010 | 2012-04-10 17:20:05 -0400 | [diff] [blame] | 232 | |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 233 | print (" { 0x%04Xu, {" % (first)) |
Behdad Esfahbod | 939c010 | 2012-04-10 17:20:05 -0400 | [diff] [blame] | 234 | for liga in ligas[first]: |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 235 | print (" { 0x%04Xu, 0x%04Xu }, /* %s */" % (liga[0], liga[1], names[liga[1]])) |
| 236 | print (" }},") |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 237 | |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 238 | print ("};") |
| 239 | print () |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 240 | |
| 241 | |
| 242 | |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 243 | print ("/* == Start of generated table == */") |
| 244 | print ("/*") |
| 245 | print (" * The following table is generated by running:") |
| 246 | print (" *") |
| 247 | print (" * ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt Blocks.txt") |
| 248 | print (" *") |
| 249 | print (" * on files with these headers:") |
| 250 | print (" *") |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 251 | for h in headers: |
| 252 | for l in h: |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 253 | print (" * %s" % (l.strip())) |
| 254 | print (" */") |
| 255 | print () |
| 256 | print ("#ifndef HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH") |
| 257 | print ("#define HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH") |
| 258 | print () |
Behdad Esfahbod | 14d7841 | 2010-11-17 16:52:58 -0500 | [diff] [blame] | 259 | |
Behdad Esfahbod | 3f5327a | 2014-06-20 16:17:42 -0400 | [diff] [blame] | 260 | read_blocks (files[2]) |
Behdad Esfahbod | ae4a2b9 | 2012-04-10 16:25:08 -0400 | [diff] [blame] | 261 | print_joining_table (files[0]) |
| 262 | print_shaping_table (files[1]) |
Behdad Esfahbod | d606daa | 2011-09-20 14:34:06 -0400 | [diff] [blame] | 263 | |
Ebrahim Byagowi | cab2c2c | 2018-03-29 12:48:47 +0430 | [diff] [blame] | 264 | print () |
| 265 | print ("#endif /* HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH */") |
| 266 | print () |
| 267 | print ("/* == End of generated table == */") |