1#!/usr/bin/env python 2 3from __future__ import print_function, division, absolute_import 4 5import io, os.path, sys 6 7if len (sys.argv) != 4: 8 print ("usage: ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt Blocks.txt", file=sys.stderr) 9 sys.exit (1) 10 11files = [io.open (x, encoding='utf-8') for x in sys.argv[1:]] 12 13headers = [[files[0].readline (), files[0].readline ()], [files[2].readline (), files[2].readline ()]] 14headers.append (["UnicodeData.txt does not have a header."]) 15while files[0].readline ().find ('##################') < 0: 16 pass 17 18blocks = {} 19def 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 42 43def print_joining_table(f): 44 45 values = {} 46 for line in f: 47 48 if line[0] == '#': 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) 56 57 if fields[3] in ["ALAPH", "DALATH RISH"]: 58 value = "JOINING_GROUP_" + fields[3].replace(' ', '_') 59 else: 60 value = "JOINING_TYPE_" + fields[2] 61 values[u] = value 62 63 short_value = {} 64 for value in set([v for v in values.values()] + ['JOINING_TYPE_X']): 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 69 print () 70 for value,short in short_value.items(): 71 print ("#define %s %s" % (short, value)) 72 73 uu = sorted(values.keys()) 74 num = len(values) 75 all_blocks = set([blocks[u] for u in uu]) 76 77 last = -100000 78 ranges = [] 79 for u in uu: 80 if u - last <= 1+16*5: 81 ranges[-1][-1] = u 82 else: 83 ranges.append([u,u]) 84 last = u 85 86 print () 87 print ("static const uint8_t joining_table[] =") 88 print ("{") 89 last_block = None 90 offset = 0 91 for start,end in ranges: 92 93 print () 94 print ("#define joining_offset_0x%04xu %d" % (start, offset)) 95 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: 103 print () 104 if block in all_blocks: 105 print ("\n /* %s */" % block) 106 else: 107 print ("\n /* FILLER */") 108 last_block = block 109 if u % 32 != 0: 110 print () 111 print (" /* %04X */" % (u//32*32), " " * (u % 32), end="") 112 113 if u % 32 == 0: 114 print () 115 print (" /* %04X */ " % u, end="") 116 print ("%s," % short_value[value], end="") 117 print () 118 119 offset += end - start + 1 120 print () 121 occupancy = num * 100. / offset 122 print ("}; /* Table items: %d; occupancy: %d%% */" % (offset, occupancy)) 123 print () 124 125 page_bits = 12; 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 (" {") 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): 134 print (" case 0x%0Xu:" % p) 135 for (start,end) in ranges: 136 if p not in [start>>page_bits, end>>page_bits]: continue 137 offset = "joining_offset_0x%04xu" % start 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 () 147 for value,short in short_value.items(): 148 print ("#undef %s" % (short)) 149 print () 150 151def 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 189 190 print () 191 print ("static const uint16_t shaping_table[][4] =") 192 print ("{") 193 194 keys = shapes.keys () 195 min_u, max_u = min (keys), max (keys) 196 for u in range (min_u, max_u + 1): 197 s = [shapes[u][shape] if u in shapes and shape in shapes[u] else 0 198 for shape in ['initial', 'medial', 'final', 'isolated']] 199 value = ', '.join ("0x%04Xu" % c for c in s) 200 print (" {%s}, /* U+%04X %s */" % (value, u, names[u] if u in names else "")) 201 202 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 () 207 208 ligas = {} 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) 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) 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 ("{") 231 for first in sorted (ligas.keys ()): 232 233 print (" { 0x%04Xu, {" % (first)) 234 for liga in ligas[first]: 235 print (" { 0x%04Xu, 0x%04Xu }, /* %s */" % (liga[0], liga[1], names[liga[1]])) 236 print (" }},") 237 238 print ("};") 239 print () 240 241 242 243print ("/* == Start of generated table == */") 244print ("/*") 245print (" * The following table is generated by running:") 246print (" *") 247print (" * ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt Blocks.txt") 248print (" *") 249print (" * on files with these headers:") 250print (" *") 251for h in headers: 252 for l in h: 253 print (" * %s" % (l.strip())) 254print (" */") 255print () 256print ("#ifndef HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH") 257print ("#define HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH") 258print () 259 260read_blocks (files[2]) 261print_joining_table (files[0]) 262print_shaping_table (files[1]) 263 264print () 265print ("#endif /* HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH */") 266print () 267print ("/* == End of generated table == */") 268