|
Packit Service |
5bcba8 |
#!/usr/bin/python
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
import sys
|
|
Packit Service |
5bcba8 |
import os.path
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
if len (sys.argv) != 4:
|
|
Packit Service |
5bcba8 |
print >>sys.stderr, "usage: ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt Blocks.txt"
|
|
Packit Service |
5bcba8 |
sys.exit (1)
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
files = [file (x) for x in sys.argv[1:]]
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
headers = [[files[0].readline (), files[0].readline ()], [files[2].readline (), files[2].readline ()]]
|
|
Packit Service |
5bcba8 |
headers.append (["UnicodeData.txt does not have a header."])
|
|
Packit Service |
5bcba8 |
while files[0].readline ().find ('##################') < 0:
|
|
Packit Service |
5bcba8 |
pass
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
blocks = {}
|
|
Packit Service |
5bcba8 |
def read_blocks(f):
|
|
Packit Service |
5bcba8 |
global blocks
|
|
Packit Service |
5bcba8 |
for line in f:
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
j = line.find ('#')
|
|
Packit Service |
5bcba8 |
if j >= 0:
|
|
Packit Service |
5bcba8 |
line = line[:j]
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
fields = [x.strip () for x in line.split (';')]
|
|
Packit Service |
5bcba8 |
if len (fields) == 1:
|
|
Packit Service |
5bcba8 |
continue
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
uu = fields[0].split ('..')
|
|
Packit Service |
5bcba8 |
start = int (uu[0], 16)
|
|
Packit Service |
5bcba8 |
if len (uu) == 1:
|
|
Packit Service |
5bcba8 |
end = start
|
|
Packit Service |
5bcba8 |
else:
|
|
Packit Service |
5bcba8 |
end = int (uu[1], 16)
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
t = fields[1]
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
for u in range (start, end + 1):
|
|
Packit Service |
5bcba8 |
blocks[u] = t
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
def print_joining_table(f):
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
values = {}
|
|
Packit Service |
5bcba8 |
for line in f:
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
if line[0] == '#':
|
|
Packit Service |
5bcba8 |
continue
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
fields = [x.strip () for x in line.split (';')]
|
|
Packit Service |
5bcba8 |
if len (fields) == 1:
|
|
Packit Service |
5bcba8 |
continue
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
u = int (fields[0], 16)
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
if fields[3] in ["ALAPH", "DALATH RISH"]:
|
|
Packit Service |
5bcba8 |
value = "JOINING_GROUP_" + fields[3].replace(' ', '_')
|
|
Packit Service |
5bcba8 |
else:
|
|
Packit Service |
5bcba8 |
value = "JOINING_TYPE_" + fields[2]
|
|
Packit Service |
5bcba8 |
values[u] = value
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
short_value = {}
|
|
Packit Service |
5bcba8 |
for value in set([v for v in values.values()] + ['JOINING_TYPE_X']):
|
|
Packit Service |
5bcba8 |
short = ''.join(x[0] for x in value.split('_')[2:])
|
|
Packit Service |
5bcba8 |
assert short not in short_value.values()
|
|
Packit Service |
5bcba8 |
short_value[value] = short
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
for value,short in short_value.items():
|
|
Packit Service |
5bcba8 |
print "#define %s %s" % (short, value)
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
uu = sorted(values.keys())
|
|
Packit Service |
5bcba8 |
num = len(values)
|
|
Packit Service |
5bcba8 |
all_blocks = set([blocks[u] for u in uu])
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
last = -100000
|
|
Packit Service |
5bcba8 |
ranges = []
|
|
Packit Service |
5bcba8 |
for u in uu:
|
|
Packit Service |
5bcba8 |
if u - last <= 1+16*5:
|
|
Packit Service |
5bcba8 |
ranges[-1][-1] = u
|
|
Packit Service |
5bcba8 |
else:
|
|
Packit Service |
5bcba8 |
ranges.append([u,u])
|
|
Packit Service |
5bcba8 |
last = u
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
print "static const uint8_t joining_table[] ="
|
|
Packit Service |
5bcba8 |
print "{"
|
|
Packit Service |
5bcba8 |
last_block = None
|
|
Packit Service |
5bcba8 |
offset = 0
|
|
Packit Service |
5bcba8 |
for start,end in ranges:
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
print "#define joining_offset_0x%04xu %d" % (start, offset)
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
for u in range(start, end+1):
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
block = blocks.get(u, last_block)
|
|
Packit Service |
5bcba8 |
value = values.get(u, "JOINING_TYPE_X")
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
if block != last_block or u == start:
|
|
Packit Service |
5bcba8 |
if u != start:
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
if block in all_blocks:
|
|
Packit Service |
5bcba8 |
print "\n /* %s */" % block
|
|
Packit Service |
5bcba8 |
else:
|
|
Packit Service |
5bcba8 |
print "\n /* FILLER */"
|
|
Packit Service |
5bcba8 |
last_block = block
|
|
Packit Service |
5bcba8 |
if u % 32 != 0:
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
print " /* %04X */" % (u//32*32), " " * (u % 32),
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
if u % 32 == 0:
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
print " /* %04X */ " % u,
|
|
Packit Service |
5bcba8 |
sys.stdout.write("%s," % short_value[value])
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
offset += end - start + 1
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
occupancy = num * 100. / offset
|
|
Packit Service |
5bcba8 |
print "}; /* Table items: %d; occupancy: %d%% */" % (offset, occupancy)
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
page_bits = 12;
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
print "static unsigned int"
|
|
Packit Service |
5bcba8 |
print "joining_type (hb_codepoint_t u)"
|
|
Packit Service |
5bcba8 |
print "{"
|
|
Packit Service |
5bcba8 |
print " switch (u >> %d)" % page_bits
|
|
Packit Service |
5bcba8 |
print " {"
|
|
Packit Service |
5bcba8 |
pages = set([u>>page_bits for u in [s for s,e in ranges]+[e for s,e in ranges]])
|
|
Packit Service |
5bcba8 |
for p in sorted(pages):
|
|
Packit Service |
5bcba8 |
print " case 0x%0Xu:" % p
|
|
Packit Service |
5bcba8 |
for (start,end) in ranges:
|
|
Packit Service |
5bcba8 |
if p not in [start>>page_bits, end>>page_bits]: continue
|
|
Packit Service |
5bcba8 |
offset = "joining_offset_0x%04xu" % start
|
|
Packit Service |
5bcba8 |
print " if (hb_in_range<hb_codepoint_t> (u, 0x%04Xu, 0x%04Xu)) return joining_table[u - 0x%04Xu + %s];" % (start, end, start, offset)
|
|
Packit Service |
5bcba8 |
print " break;"
|
|
Packit Service |
5bcba8 |
print ""
|
|
Packit Service |
5bcba8 |
print " default:"
|
|
Packit Service |
5bcba8 |
print " break;"
|
|
Packit Service |
5bcba8 |
print " }"
|
|
Packit Service |
5bcba8 |
print " return X;"
|
|
Packit Service |
5bcba8 |
print "}"
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
for value,short in short_value.items():
|
|
Packit Service |
5bcba8 |
print "#undef %s" % (short)
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
def print_shaping_table(f):
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
shapes = {}
|
|
Packit Service |
5bcba8 |
ligatures = {}
|
|
Packit Service |
5bcba8 |
names = {}
|
|
Packit Service |
5bcba8 |
for line in f:
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
fields = [x.strip () for x in line.split (';')]
|
|
Packit Service |
5bcba8 |
if fields[5][0:1] != '<':
|
|
Packit Service |
5bcba8 |
continue
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
items = fields[5].split (' ')
|
|
Packit Service |
5bcba8 |
shape, items = items[0][1:-1], tuple (int (x, 16) for x in items[1:])
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
if not shape in ['initial', 'medial', 'isolated', 'final']:
|
|
Packit Service |
5bcba8 |
continue
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
c = int (fields[0], 16)
|
|
Packit Service |
5bcba8 |
if len (items) != 1:
|
|
Packit Service |
5bcba8 |
# We only care about lam-alef ligatures
|
|
Packit Service |
5bcba8 |
if len (items) != 2 or items[0] != 0x0644 or items[1] not in [0x0622, 0x0623, 0x0625, 0x0627]:
|
|
Packit Service |
5bcba8 |
continue
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
# Save ligature
|
|
Packit Service |
5bcba8 |
names[c] = fields[1]
|
|
Packit Service |
5bcba8 |
if items not in ligatures:
|
|
Packit Service |
5bcba8 |
ligatures[items] = {}
|
|
Packit Service |
5bcba8 |
ligatures[items][shape] = c
|
|
Packit Service |
5bcba8 |
pass
|
|
Packit Service |
5bcba8 |
else:
|
|
Packit Service |
5bcba8 |
# Save shape
|
|
Packit Service |
5bcba8 |
if items[0] not in names:
|
|
Packit Service |
5bcba8 |
names[items[0]] = fields[1]
|
|
Packit Service |
5bcba8 |
else:
|
|
Packit Service |
5bcba8 |
names[items[0]] = os.path.commonprefix ([names[items[0]], fields[1]]).strip ()
|
|
Packit Service |
5bcba8 |
if items[0] not in shapes:
|
|
Packit Service |
5bcba8 |
shapes[items[0]] = {}
|
|
Packit Service |
5bcba8 |
shapes[items[0]][shape] = c
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
print "static const uint16_t shaping_table[][4] ="
|
|
Packit Service |
5bcba8 |
print "{"
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
keys = shapes.keys ()
|
|
Packit Service |
5bcba8 |
min_u, max_u = min (keys), max (keys)
|
|
Packit Service |
5bcba8 |
for u in range (min_u, max_u + 1):
|
|
Packit Service |
5bcba8 |
s = [shapes[u][shape] if u in shapes and shape in shapes[u] else 0
|
|
Packit Service |
5bcba8 |
for shape in ['initial', 'medial', 'final', 'isolated']]
|
|
Packit Service |
5bcba8 |
value = ', '.join ("0x%04Xu" % c for c in s)
|
|
Packit Service |
5bcba8 |
print " {%s}, /* U+%04X %s */" % (value, u, names[u] if u in names else "")
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
print "};"
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
print "#define SHAPING_TABLE_FIRST 0x%04Xu" % min_u
|
|
Packit Service |
5bcba8 |
print "#define SHAPING_TABLE_LAST 0x%04Xu" % max_u
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
ligas = {}
|
|
Packit Service |
5bcba8 |
for pair in ligatures.keys ():
|
|
Packit Service |
5bcba8 |
for shape in ligatures[pair]:
|
|
Packit Service |
5bcba8 |
c = ligatures[pair][shape]
|
|
Packit Service |
5bcba8 |
if shape == 'isolated':
|
|
Packit Service |
5bcba8 |
liga = (shapes[pair[0]]['initial'], shapes[pair[1]]['final'])
|
|
Packit Service |
5bcba8 |
elif shape == 'final':
|
|
Packit Service |
5bcba8 |
liga = (shapes[pair[0]]['medial'], shapes[pair[1]]['final'])
|
|
Packit Service |
5bcba8 |
else:
|
|
Packit Service |
5bcba8 |
raise Exception ("Unexpected shape", shape)
|
|
Packit Service |
5bcba8 |
if liga[0] not in ligas:
|
|
Packit Service |
5bcba8 |
ligas[liga[0]] = []
|
|
Packit Service |
5bcba8 |
ligas[liga[0]].append ((liga[1], c))
|
|
Packit Service |
5bcba8 |
max_i = max (len (ligas[l]) for l in ligas)
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
print "static const struct ligature_set_t {"
|
|
Packit Service |
5bcba8 |
print " uint16_t first;"
|
|
Packit Service |
5bcba8 |
print " struct ligature_pairs_t {"
|
|
Packit Service |
5bcba8 |
print " uint16_t second;"
|
|
Packit Service |
5bcba8 |
print " uint16_t ligature;"
|
|
Packit Service |
5bcba8 |
print " } ligatures[%d];" % max_i
|
|
Packit Service |
5bcba8 |
print "} ligature_table[] ="
|
|
Packit Service |
5bcba8 |
print "{"
|
|
Packit Service |
5bcba8 |
keys = ligas.keys ()
|
|
Packit Service |
5bcba8 |
keys.sort ()
|
|
Packit Service |
5bcba8 |
for first in keys:
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
print " { 0x%04Xu, {" % (first)
|
|
Packit Service |
5bcba8 |
for liga in ligas[first]:
|
|
Packit Service |
5bcba8 |
print " { 0x%04Xu, 0x%04Xu }, /* %s */" % (liga[0], liga[1], names[liga[1]])
|
|
Packit Service |
5bcba8 |
print " }},"
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
print "};"
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
print "/* == Start of generated table == */"
|
|
Packit Service |
5bcba8 |
print "/*"
|
|
Packit Service |
5bcba8 |
print " * The following table is generated by running:"
|
|
Packit Service |
5bcba8 |
print " *"
|
|
Packit Service |
5bcba8 |
print " * ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt Blocks.txt"
|
|
Packit Service |
5bcba8 |
print " *"
|
|
Packit Service |
5bcba8 |
print " * on files with these headers:"
|
|
Packit Service |
5bcba8 |
print " *"
|
|
Packit Service |
5bcba8 |
for h in headers:
|
|
Packit Service |
5bcba8 |
for l in h:
|
|
Packit Service |
5bcba8 |
print " * %s" % (l.strip())
|
|
Packit Service |
5bcba8 |
print " */"
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
print "#ifndef HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH"
|
|
Packit Service |
5bcba8 |
print "#define HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH"
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
read_blocks (files[2])
|
|
Packit Service |
5bcba8 |
print_joining_table (files[0])
|
|
Packit Service |
5bcba8 |
print_shaping_table (files[1])
|
|
Packit Service |
5bcba8 |
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
print "#endif /* HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH */"
|
|
Packit Service |
5bcba8 |
print
|
|
Packit Service |
5bcba8 |
print "/* == End of generated table == */"
|
|
Packit Service |
5bcba8 |
|