1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/gfx/harfbuzz/src/gen-arabic-table.py Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,197 @@
1.4 +#!/usr/bin/python
1.5 +
1.6 +import sys
1.7 +import os.path
1.8 +
1.9 +if len (sys.argv) != 3:
1.10 + print >>sys.stderr, "usage: ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt"
1.11 + sys.exit (1)
1.12 +
1.13 +files = [file (x) for x in sys.argv[1:]]
1.14 +
1.15 +headers = [[files[0].readline (), files[0].readline ()]]
1.16 +headers.append (["UnicodeData.txt does not have a header."])
1.17 +while files[0].readline ().find ('##################') < 0:
1.18 + pass
1.19 +
1.20 +
1.21 +def print_joining_table(f):
1.22 +
1.23 + print
1.24 + print "static const uint8_t joining_table[] ="
1.25 + print "{"
1.26 +
1.27 + min_u = 0x110000
1.28 + max_u = 0
1.29 + num = 0
1.30 + last = -1
1.31 + block = ''
1.32 + for line in f:
1.33 +
1.34 + if line[0] == '#':
1.35 + if line.find (" characters"):
1.36 + block = line[2:].strip ()
1.37 + continue
1.38 +
1.39 + fields = [x.strip () for x in line.split (';')]
1.40 + if len (fields) == 1:
1.41 + continue
1.42 +
1.43 + u = int (fields[0], 16)
1.44 + if u == 0x200C or u == 0x200D:
1.45 + continue
1.46 + if u < last:
1.47 + raise Exception ("Input data character not sorted", u)
1.48 + min_u = min (min_u, u)
1.49 + max_u = max (max_u, u)
1.50 + num += 1
1.51 +
1.52 + if block:
1.53 + print "\n /* %s */\n" % block
1.54 + block = ''
1.55 +
1.56 + if last != -1:
1.57 + last += 1
1.58 + while last < u:
1.59 + print " JOINING_TYPE_X, /* %04X */" % last
1.60 + last += 1
1.61 + else:
1.62 + last = u
1.63 +
1.64 + if fields[3] in ["ALAPH", "DALATH RISH"]:
1.65 + value = "JOINING_GROUP_" + fields[3].replace(' ', '_')
1.66 + else:
1.67 + value = "JOINING_TYPE_" + fields[2]
1.68 + print " %s, /* %s */" % (value, '; '.join(fields))
1.69 +
1.70 + print
1.71 + print "};"
1.72 + print
1.73 + print "#define JOINING_TABLE_FIRST 0x%04X" % min_u
1.74 + print "#define JOINING_TABLE_LAST 0x%04X" % max_u
1.75 + print
1.76 +
1.77 + occupancy = num * 100 / (max_u - min_u + 1)
1.78 + # Maintain at least 40% occupancy in the table */
1.79 + if occupancy < 40:
1.80 + raise Exception ("Table too sparse, please investigate: ", occupancy)
1.81 +
1.82 +def print_shaping_table(f):
1.83 +
1.84 + shapes = {}
1.85 + ligatures = {}
1.86 + names = {}
1.87 + for line in f:
1.88 +
1.89 + fields = [x.strip () for x in line.split (';')]
1.90 + if fields[5][0:1] != '<':
1.91 + continue
1.92 +
1.93 + items = fields[5].split (' ')
1.94 + shape, items = items[0][1:-1], tuple (int (x, 16) for x in items[1:])
1.95 +
1.96 + if not shape in ['initial', 'medial', 'isolated', 'final']:
1.97 + continue
1.98 +
1.99 + c = int (fields[0], 16)
1.100 + if len (items) != 1:
1.101 + # We only care about lam-alef ligatures
1.102 + if len (items) != 2 or items[0] != 0x0644 or items[1] not in [0x0622, 0x0623, 0x0625, 0x0627]:
1.103 + continue
1.104 +
1.105 + # Save ligature
1.106 + names[c] = fields[1]
1.107 + if items not in ligatures:
1.108 + ligatures[items] = {}
1.109 + ligatures[items][shape] = c
1.110 + pass
1.111 + else:
1.112 + # Save shape
1.113 + if items[0] not in names:
1.114 + names[items[0]] = fields[1]
1.115 + else:
1.116 + names[items[0]] = os.path.commonprefix ([names[items[0]], fields[1]]).strip ()
1.117 + if items[0] not in shapes:
1.118 + shapes[items[0]] = {}
1.119 + shapes[items[0]][shape] = c
1.120 +
1.121 + print
1.122 + print "static const uint16_t shaping_table[][4] ="
1.123 + print "{"
1.124 +
1.125 + keys = shapes.keys ()
1.126 + min_u, max_u = min (keys), max (keys)
1.127 + for u in range (min_u, max_u + 1):
1.128 + s = [shapes[u][shape] if u in shapes and shape in shapes[u] else 0
1.129 + for shape in ['initial', 'medial', 'final', 'isolated']]
1.130 + value = ', '.join ("0x%04X" % c for c in s)
1.131 + print " {%s}, /* U+%04X %s */" % (value, u, names[u] if u in names else "")
1.132 +
1.133 + print "};"
1.134 + print
1.135 + print "#define SHAPING_TABLE_FIRST 0x%04X" % min_u
1.136 + print "#define SHAPING_TABLE_LAST 0x%04X" % max_u
1.137 + print
1.138 +
1.139 + ligas = {}
1.140 + for pair in ligatures.keys ():
1.141 + for shape in ligatures[pair]:
1.142 + c = ligatures[pair][shape]
1.143 + if shape == 'isolated':
1.144 + liga = (shapes[pair[0]]['initial'], shapes[pair[1]]['final'])
1.145 + elif shape == 'final':
1.146 + liga = (shapes[pair[0]]['medial'], shapes[pair[1]]['final'])
1.147 + else:
1.148 + raise Exception ("Unexpected shape", shape)
1.149 + if liga[0] not in ligas:
1.150 + ligas[liga[0]] = []
1.151 + ligas[liga[0]].append ((liga[1], c))
1.152 + max_i = max (len (ligas[l]) for l in ligas)
1.153 + print
1.154 + print "static const struct ligature_set_t {"
1.155 + print " uint16_t first;"
1.156 + print " struct ligature_pairs_t {"
1.157 + print " uint16_t second;"
1.158 + print " uint16_t ligature;"
1.159 + print " } ligatures[%d];" % max_i
1.160 + print "} ligature_table[] ="
1.161 + print "{"
1.162 + keys = ligas.keys ()
1.163 + keys.sort ()
1.164 + for first in keys:
1.165 +
1.166 + print " { 0x%04X, {" % (first)
1.167 + for liga in ligas[first]:
1.168 + print " { 0x%04X, 0x%04X }, /* %s */" % (liga[0], liga[1], names[liga[1]])
1.169 + print " }},"
1.170 +
1.171 + print "};"
1.172 + print
1.173 +
1.174 +
1.175 +
1.176 +print "/* == Start of generated table == */"
1.177 +print "/*"
1.178 +print " * The following table is generated by running:"
1.179 +print " *"
1.180 +print " * ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt"
1.181 +print " *"
1.182 +print " * on files with these headers:"
1.183 +print " *"
1.184 +for h in headers:
1.185 + for l in h:
1.186 + print " * %s" % (l.strip())
1.187 +print " */"
1.188 +print
1.189 +print "#ifndef HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH"
1.190 +print "#define HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH"
1.191 +print
1.192 +
1.193 +print_joining_table (files[0])
1.194 +print_shaping_table (files[1])
1.195 +
1.196 +print
1.197 +print "#endif /* HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH */"
1.198 +print
1.199 +print "/* == End of generated table == */"
1.200 +
