src/gen-arabic-joining-table.py

#!/usr/bin/python

import sys

header = sys.stdin.readline(), sys.stdin.readline()
dic = dict()
for line in sys.stdin:
	if line[:1] != '0':
		continue

	fields = [x.strip() for x in line.split(';')]
	u = int(fields[0], 16)

	if u < 0x0600 or (u > 0x07FF and u != 0x200C and u != 0x200D):
		raise Exception ("Ooops, unexpected unicode character: ", fields)
	dic[u] = fields

print "  /*"
print "   * The following table is generated by running:"
print "   *"
print "   *   ./gen-arabic-joining-table.py < ArabicShaping.txt"
print "   *"
print "   * on the ArabicShaping.txt file with the header:"
print "   *"
for line in header:
	print "   * %s" % (line.strip())
print "   */"
print "  /* == Start of generated table == */"
for i in range(0x0600, 0x0800):
	if i not in dic:
		print "  JOINING_TYPE_X, /* %04X */" % i
	else:
		entry = dic[i]
		if entry[3] in ["ALAPH", "DALATH RISH"]:
			value = "JOINING_GROUP_" + entry[3].replace(' ', '_')
		else:
			value = "JOINING_TYPE_" + entry[2]
		print "  %s, /* %s */" % (value, '; '.join(entry))
print "  /* == End of generated table == */"