[arabic-table] Read Blocks.txt and shuffle code around
No functional change.
diff --git a/src/gen-arabic-table.py b/src/gen-arabic-table.py
index da5a4fc..534f560 100755
--- a/src/gen-arabic-table.py
+++ b/src/gen-arabic-table.py
@@ -3,34 +3,48 @@
import sys
import os.path
-if len (sys.argv) != 3:
- print >>sys.stderr, "usage: ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt"
+if len (sys.argv) != 4:
+ print >>sys.stderr, "usage: ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt Blocks.txt"
sys.exit (1)
files = [file (x) for x in sys.argv[1:]]
-headers = [[files[0].readline (), files[0].readline ()]]
+headers = [[files[0].readline (), files[0].readline ()], [files[2].readline (), files[2].readline ()]]
headers.append (["UnicodeData.txt does not have a header."])
while files[0].readline ().find ('##################') < 0:
pass
+blocks = {}
+def read_blocks(f):
+ global blocks
+ for line in f:
+
+ j = line.find ('#')
+ if j >= 0:
+ line = line[:j]
+
+ fields = [x.strip () for x in line.split (';')]
+ if len (fields) == 1:
+ continue
+
+ uu = fields[0].split ('..')
+ start = int (uu[0], 16)
+ if len (uu) == 1:
+ end = start
+ else:
+ end = int (uu[1], 16)
+
+ t = fields[1]
+
+ for u in range (start, end + 1):
+ blocks[u] = t
def print_joining_table(f):
- print
- print "static const uint8_t joining_table[] ="
- print "{"
-
- min_u = 0x110000
- max_u = 0
- num = 0
- last = -1
- block = ''
+ values = {}
for line in f:
if line[0] == '#':
- if line.find (" characters"):
- block = line[2:].strip ()
continue
fields = [x.strip () for x in line.split (';')]
@@ -40,30 +54,29 @@
u = int (fields[0], 16)
if u == 0x200C or u == 0x200D:
continue
- if u < last:
- raise Exception ("Input data character not sorted", u)
- min_u = min (min_u, u)
- max_u = max (max_u, u)
- num += 1
-
- if block:
- print "\n /* %s */\n" % block
- block = ''
-
- if last != -1:
- last += 1
- while last < u:
- print " JOINING_TYPE_X, /* %04X */" % last
- last += 1
- else:
- last = u
if fields[3] in ["ALAPH", "DALATH RISH"]:
value = "JOINING_GROUP_" + fields[3].replace(' ', '_')
else:
value = "JOINING_TYPE_" + fields[2]
- print " %s, /* %s */" % (value, '; '.join(fields))
+ values[u] = (value, fields)
+ keys = values.keys()
+ min_u = min(keys)
+ max_u = max(keys)
+ num = len(values)
+ print
+ print "static const uint8_t joining_table[] ="
+ print "{"
+ last_block = None
+ for u in range(min_u, max_u+1):
+ value,fields = values.get(u, ("JOINING_TYPE_X", ["%04X"%u]))
+ block = blocks.get(u, last_block)
+ if block != last_block:
+ print "\n /* %s */\n" % block
+ last_block = block
+
+ print " %s, /* %s */" % (value, '; '.join(fields))
print
print "};"
print
@@ -174,7 +187,7 @@
print "/*"
print " * The following table is generated by running:"
print " *"
-print " * ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt"
+print " * ./gen-arabic-table.py ArabicShaping.txt UnicodeData.txt Blocks.txt"
print " *"
print " * on files with these headers:"
print " *"
@@ -187,6 +200,7 @@
print "#define HB_OT_SHAPE_COMPLEX_ARABIC_TABLE_HH"
print
+read_blocks (files[2])
print_joining_table (files[0])
print_shaping_table (files[1])