1#!/usr/bin/python
2
3import sys
4import os.path
5
6if len (sys.argv) != 3:
7	print >>sys.stderr, "usage: ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt"
8	sys.exit (1)
9
10files = [file (x) for x in sys.argv[1:]]
11
12headers = [[files[0].readline (), files[0].readline ()]]
13headers.append (["UnicodeData.txt does not have a header."])
14while files[0].readline ().find ('##################') < 0:
15	pass
16
17
18def print_joining_table(f):
19
20	print
21	print "static const uint8_t joining_table[] ="
22	print "{"
23
24	min_u = 0x110000
25	max_u = 0
26	num = 0
27	last = -1
28	block = ''
29	for line in f:
30
31		if line[0] == '#':
32			if line.find (" characters"):
33				block = line[2:].strip ()
34			continue
35
36		fields = [x.strip () for x in line.split (';')]
37		if len (fields) == 1:
38			continue
39
40		u = int (fields[0], 16)
41		if u == 0x200C or u == 0x200D:
42			continue
43		if u < last:
44			raise Exception ("Input data character not sorted", u)
45		min_u = min (min_u, u)
46		max_u = max (max_u, u)
47		num += 1
48
49		if block:
50			print "\n  /* %s */\n" % block
51			block = ''
52
53		if last != -1:
54			last += 1
55			while last < u:
56				print "  JOINING_TYPE_X, /* %04X */" % last
57				last += 1
58		else:
59			last = u
60
61		if fields[3] in ["ALAPH", "DALATH RISH"]:
62			value = "JOINING_GROUP_" + fields[3].replace(' ', '_')
63		else:
64			value = "JOINING_TYPE_" + fields[2]
65		print "  %s, /* %s */" % (value, '; '.join(fields))
66
67	print
68	print "};"
69	print
70	print "#define JOINING_TABLE_FIRST	0x%04X" % min_u
71	print "#define JOINING_TABLE_LAST	0x%04X" % max_u
72	print
73
74	occupancy = num * 100 / (max_u - min_u + 1)
75	# Maintain at least 40% occupancy in the table */
76	if occupancy < 40:
77		raise Exception ("Table too sparse, please investigate: ", occupancy)
78
79def print_shaping_table(f):
80
81	shapes = {}
82	ligatures = {}
83	names = {}
84	for line in f:
85
86		fields = [x.strip () for x in line.split (';')]
87		if fields[5][0:1] != '<':
88			continue
89
90		items = fields[5].split (' ')
91		shape, items = items[0][1:-1], tuple (int (x, 16) for x in items[1:])
92
93		if not shape in ['initial', 'medial', 'isolated', 'final']:
94			continue
95
96		c = int (fields[0], 16)
97		if len (items) != 1:
98			# We only care about lam-alef ligatures
99			if len (items) != 2 or items[0] != 0x0644 or items[1] not in [0x0622, 0x0623, 0x0625, 0x0627]:
100				continue
101
102			# Save ligature
103			names[c] = fields[1]
104			if items not in ligatures:
105				ligatures[items] = {}
106			ligatures[items][shape] = c
107			pass
108		else:
109			# Save shape
110			if items[0] not in names:
111				names[items[0]] = fields[1]
112			else:
113				names[items[0]] = os.path.commonprefix ([names[items[0]], fields[1]]).strip ()
114			if items[0] not in shapes:
115				shapes[items[0]] = {}
116			shapes[items[0]][shape] = c
117
118	print
119	print "static const uint16_t shaping_table[][4] ="
120	print "{"
121
122	keys = shapes.keys ()
123	min_u, max_u = min (keys), max (keys)
124	for u in range (min_u, max_u + 1):
125		s = [shapes[u][shape] if u in shapes and shape in shapes[u] else 0
126		     for shape in  ['initial', 'medial', 'final', 'isolated']]
127		value = ', '.join ("0x%04X" % c for c in s)
128		print "  {%s}, /* U+%04X %s */" % (value, u, names[u] if u in names else "")
129
130	print "};"
131	print
132	print "#define SHAPING_TABLE_FIRST	0x%04X" % min_u
133	print "#define SHAPING_TABLE_LAST	0x%04X" % max_u
134	print
135
136	ligas = {}
137	for pair in ligatures.keys ():
138		for shape in ligatures[pair]:
139			c = ligatures[pair][shape]
140			if shape == 'isolated':
141				liga = (shapes[pair[0]]['initial'], shapes[pair[1]]['final'])
142			elif shape == 'final':
143				liga = (shapes[pair[0]]['medial'], shapes[pair[1]]['final'])
144			else:
145				raise Exception ("Unexpected shape", shape)
146			if liga[0] not in ligas:
147				ligas[liga[0]] = []
148			ligas[liga[0]].append ((liga[1], c))
149	max_i = max (len (ligas[l]) for l in ligas)
150	print
151	print "static const struct ligature_set_t {"
152	print " uint16_t first;"
153	print " struct ligature_pairs_t {"
154	print "   uint16_t second;"
155	print "   uint16_t ligature;"
156	print " } ligatures[%d];" % max_i
157	print "} ligature_table[] ="
158	print "{"
159	keys = ligas.keys ()
160	keys.sort ()
161	for first in keys:
162
163		print "  { 0x%04X, {" % (first)
164		for liga in ligas[first]:
165			print "    { 0x%04X, 0x%04X }, /* %s */" % (liga[0], liga[1], names[liga[1]])
166		print "  }},"
167
168	print "};"
169	print
170
171
172
173print "/* == Start of generated table == */"
174print "/*"
175print " * The following table is generated by running:"
176print " *"
177print " *   ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt"
178print " *"
179print " * on files with these headers:"
180print " *"
181for h in headers:
182	for l in h:
183		print " * %s" % (l.strip())
184print " */"
185print
186print "#ifndef HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH"
187print "#define HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH"
188print
189
190print_joining_table (files[0])
191print_shaping_table (files[1])
192
193print
194print "#endif /* HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH */"
195print
196print "/* == End of generated table == */"
197
198