michael@0: #!/usr/bin/python
michael@0: 
michael@0: import sys
michael@0: import os.path
michael@0: 
michael@0: if len (sys.argv) != 3:
michael@0: 	print >>sys.stderr, "usage: ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt"
michael@0: 	sys.exit (1)
michael@0: 
michael@0: files = [file (x) for x in sys.argv[1:]]
michael@0: 
michael@0: headers = [[files[0].readline (), files[0].readline ()]]
michael@0: headers.append (["UnicodeData.txt does not have a header."])
michael@0: while files[0].readline ().find ('##################') < 0:
michael@0: 	pass
michael@0: 
michael@0: 
michael@0: def print_joining_table(f):
michael@0: 
michael@0: 	print
michael@0: 	print "static const uint8_t joining_table[] ="
michael@0: 	print "{"
michael@0: 
michael@0: 	min_u = 0x110000
michael@0: 	max_u = 0
michael@0: 	num = 0
michael@0: 	last = -1
michael@0: 	block = ''
michael@0: 	for line in f:
michael@0: 
michael@0: 		if line[0] == '#':
michael@0: 			if line.find (" characters"):
michael@0: 				block = line[2:].strip ()
michael@0: 			continue
michael@0: 
michael@0: 		fields = [x.strip () for x in line.split (';')]
michael@0: 		if len (fields) == 1:
michael@0: 			continue
michael@0: 
michael@0: 		u = int (fields[0], 16)
michael@0: 		if u == 0x200C or u == 0x200D:
michael@0: 			continue
michael@0: 		if u < last:
michael@0: 			raise Exception ("Input data character not sorted", u)
michael@0: 		min_u = min (min_u, u)
michael@0: 		max_u = max (max_u, u)
michael@0: 		num += 1
michael@0: 
michael@0: 		if block:
michael@0: 			print "\n  /* %s */\n" % block
michael@0: 			block = ''
michael@0: 
michael@0: 		if last != -1:
michael@0: 			last += 1
michael@0: 			while last < u:
michael@0: 				print "  JOINING_TYPE_X, /* %04X */" % last
michael@0: 				last += 1
michael@0: 		else:
michael@0: 			last = u
michael@0: 
michael@0: 		if fields[3] in ["ALAPH", "DALATH RISH"]:
michael@0: 			value = "JOINING_GROUP_" + fields[3].replace(' ', '_')
michael@0: 		else:
michael@0: 			value = "JOINING_TYPE_" + fields[2]
michael@0: 		print "  %s, /* %s */" % (value, '; '.join(fields))
michael@0: 
michael@0: 	print
michael@0: 	print "};"
michael@0: 	print
michael@0: 	print "#define JOINING_TABLE_FIRST	0x%04X" % min_u
michael@0: 	print "#define JOINING_TABLE_LAST	0x%04X" % max_u
michael@0: 	print
michael@0: 
michael@0: 	occupancy = num * 100 / (max_u - min_u + 1)
michael@0: 	# Maintain at least 40% occupancy in the table */
michael@0: 	if occupancy < 40:
michael@0: 		raise Exception ("Table too sparse, please investigate: ", occupancy)
michael@0: 
michael@0: def print_shaping_table(f):
michael@0: 
michael@0: 	shapes = {}
michael@0: 	ligatures = {}
michael@0: 	names = {}
michael@0: 	for line in f:
michael@0: 
michael@0: 		fields = [x.strip () for x in line.split (';')]
michael@0: 		if fields[5][0:1] != '<':
michael@0: 			continue
michael@0: 
michael@0: 		items = fields[5].split (' ')
michael@0: 		shape, items = items[0][1:-1], tuple (int (x, 16) for x in items[1:])
michael@0: 
michael@0: 		if not shape in ['initial', 'medial', 'isolated', 'final']:
michael@0: 			continue
michael@0: 
michael@0: 		c = int (fields[0], 16)
michael@0: 		if len (items) != 1:
michael@0: 			# We only care about lam-alef ligatures
michael@0: 			if len (items) != 2 or items[0] != 0x0644 or items[1] not in [0x0622, 0x0623, 0x0625, 0x0627]:
michael@0: 				continue
michael@0: 
michael@0: 			# Save ligature
michael@0: 			names[c] = fields[1]
michael@0: 			if items not in ligatures:
michael@0: 				ligatures[items] = {}
michael@0: 			ligatures[items][shape] = c
michael@0: 			pass
michael@0: 		else:
michael@0: 			# Save shape
michael@0: 			if items[0] not in names:
michael@0: 				names[items[0]] = fields[1]
michael@0: 			else:
michael@0: 				names[items[0]] = os.path.commonprefix ([names[items[0]], fields[1]]).strip ()
michael@0: 			if items[0] not in shapes:
michael@0: 				shapes[items[0]] = {}
michael@0: 			shapes[items[0]][shape] = c
michael@0: 
michael@0: 	print
michael@0: 	print "static const uint16_t shaping_table[][4] ="
michael@0: 	print "{"
michael@0: 
michael@0: 	keys = shapes.keys ()
michael@0: 	min_u, max_u = min (keys), max (keys)
michael@0: 	for u in range (min_u, max_u + 1):
michael@0: 		s = [shapes[u][shape] if u in shapes and shape in shapes[u] else 0
michael@0: 		     for shape in  ['initial', 'medial', 'final', 'isolated']]
michael@0: 		value = ', '.join ("0x%04X" % c for c in s)
michael@0: 		print "  {%s}, /* U+%04X %s */" % (value, u, names[u] if u in names else "")
michael@0: 
michael@0: 	print "};"
michael@0: 	print
michael@0: 	print "#define SHAPING_TABLE_FIRST	0x%04X" % min_u
michael@0: 	print "#define SHAPING_TABLE_LAST	0x%04X" % max_u
michael@0: 	print
michael@0: 
michael@0: 	ligas = {}
michael@0: 	for pair in ligatures.keys ():
michael@0: 		for shape in ligatures[pair]:
michael@0: 			c = ligatures[pair][shape]
michael@0: 			if shape == 'isolated':
michael@0: 				liga = (shapes[pair[0]]['initial'], shapes[pair[1]]['final'])
michael@0: 			elif shape == 'final':
michael@0: 				liga = (shapes[pair[0]]['medial'], shapes[pair[1]]['final'])
michael@0: 			else:
michael@0: 				raise Exception ("Unexpected shape", shape)
michael@0: 			if liga[0] not in ligas:
michael@0: 				ligas[liga[0]] = []
michael@0: 			ligas[liga[0]].append ((liga[1], c))
michael@0: 	max_i = max (len (ligas[l]) for l in ligas)
michael@0: 	print
michael@0: 	print "static const struct ligature_set_t {"
michael@0: 	print " uint16_t first;"
michael@0: 	print " struct ligature_pairs_t {"
michael@0: 	print "   uint16_t second;"
michael@0: 	print "   uint16_t ligature;"
michael@0: 	print " } ligatures[%d];" % max_i
michael@0: 	print "} ligature_table[] ="
michael@0: 	print "{"
michael@0: 	keys = ligas.keys ()
michael@0: 	keys.sort ()
michael@0: 	for first in keys:
michael@0: 
michael@0: 		print "  { 0x%04X, {" % (first)
michael@0: 		for liga in ligas[first]:
michael@0: 			print "    { 0x%04X, 0x%04X }, /* %s */" % (liga[0], liga[1], names[liga[1]])
michael@0: 		print "  }},"
michael@0: 
michael@0: 	print "};"
michael@0: 	print
michael@0: 
michael@0: 
michael@0: 
michael@0: print "/* == Start of generated table == */"
michael@0: print "/*"
michael@0: print " * The following table is generated by running:"
michael@0: print " *"
michael@0: print " *   ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt"
michael@0: print " *"
michael@0: print " * on files with these headers:"
michael@0: print " *"
michael@0: for h in headers:
michael@0: 	for l in h:
michael@0: 		print " * %s" % (l.strip())
michael@0: print " */"
michael@0: print
michael@0: print "#ifndef HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH"
michael@0: print "#define HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH"
michael@0: print
michael@0: 
michael@0: print_joining_table (files[0])
michael@0: print_shaping_table (files[1])
michael@0: 
michael@0: print
michael@0: print "#endif /* HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH */"
michael@0: print
michael@0: print "/* == End of generated table == */"
michael@0: