Home | History | Annotate | Download | only in src 
      1 #!/usr/bin/python
      2 
      3 import sys
      4 
      5 if len (sys.argv) != 4:
      6 	print >>sys.stderr, "usage: ./gen-indic-table.py IndicSyllabicCategory.txt IndicMatraCategory.txt Blocks.txt"
      7 	sys.exit (1)
      8 
      9 BLACKLISTED_BLOCKS = ["Thai", "Lao", "Tibetan"]
     10 
     11 files = [file (x) for x in sys.argv[1:]]
     12 
     13 headers = [[f.readline () for i in range (2)] for f in files]
     14 
     15 data = [{} for f in files]
     16 values = [{} for f in files]
     17 for i, f in enumerate (files):
     18 	for line in f:
     19 
     20 		j = line.find ('#')
     21 		if j >= 0:
     22 			line = line[:j]
     23 
     24 		fields = [x.strip () for x in line.split (';')]
     25 		if len (fields) == 1:
     26 			continue
     27 
     28 		uu = fields[0].split ('..')
     29 		start = int (uu[0], 16)
     30 		if len (uu) == 1:
     31 			end = start
     32 		else:
     33 			end = int (uu[1], 16)
     34 
     35 		t = fields[1]
     36 
     37 		for u in range (start, end + 1):
     38 			data[i][u] = t
     39 		values[i][t] = values[i].get (t, 0) + end - start + 1
     40 
     41 # Merge data into one dict:
     42 defaults = ('Other', 'Not_Applicable', 'No_Block')
     43 for i,v in enumerate (defaults):
     44 	values[i][v] = values[i].get (v, 0) + 1
     45 combined = {}
     46 for i,d in enumerate (data):
     47 	for u,v in d.items ():
     48 		if i == 2 and not u in combined:
     49 			continue
     50 		if not u in combined:
     51 			combined[u] = list (defaults)
     52 		combined[u][i] = v
     53 combined = {k:v for k,v in combined.items() if v[2] not in BLACKLISTED_BLOCKS}
     54 data = combined
     55 del combined
     56 num = len (data)
     57 
     58 for u in [0x17CD, 0x17CE, 0x17CF, 0x17D0, 0x17D3]:
     59 	if data[u][0] == 'Other':
     60 		data[u][0] = "Vowel_Dependent"
     61 
     62 # Move the outliers NO-BREAK SPACE and DOTTED CIRCLE out
     63 singles = {}
     64 for u in [0x00A0, 0x25CC]:
     65 	singles[u] = data[u]
     66 	del data[u]
     67 
     68 print "/* == Start of generated table == */"
     69 print "/*"
     70 print " * The following table is generated by running:"
     71 print " *"
     72 print " *   ./gen-indic-table.py IndicSyllabicCategory.txt IndicMatraCategory.txt Blocks.txt"
     73 print " *"
     74 print " * on files with these headers:"
     75 print " *"
     76 for h in headers:
     77 	for l in h:
     78 		print " * %s" % (l.strip())
     79 print " */"
     80 print
     81 print '#include "hb-ot-shape-complex-indic-private.hh"'
     82 print
     83 
     84 # Shorten values
     85 short = [{
     86 	"Bindu":		'Bi',
     87 	"Cantillation_Mark":	'Ca',
     88 	"Joiner":		'ZWJ',
     89 	"Non_Joiner":		'ZWNJ',
     90 	"Number":		'Nd',
     91 	"Visarga":		'Vs',
     92 	"Vowel":		'Vo',
     93 	"Vowel_Dependent":	'M',
     94 	"Other":		'x',
     95 },{
     96 	"Not_Applicable":	'x',
     97 }]
     98 all_shorts = [{},{}]
     99 
    100 # Add some of the values, to make them more readable, and to avoid duplicates
    101 
    102 
    103 for i in range (2):
    104 	for v,s in short[i].items ():
    105 		all_shorts[i][s] = v
    106 
    107 what = ["INDIC_SYLLABIC_CATEGORY", "INDIC_MATRA_CATEGORY"]
    108 what_short = ["ISC", "IMC"]
    109 for i in range (2):
    110 	print
    111 	vv = values[i].keys ()
    112 	vv.sort ()
    113 	for v in vv:
    114 		v_no_and = v.replace ('_And_', '_')
    115 		if v in short[i]:
    116 			s = short[i][v]
    117 		else:
    118 			s = ''.join ([c for c in v_no_and if ord ('A') <= ord (c) <= ord ('Z')])
    119 			if s in all_shorts[i]:
    120 				raise Exception ("Duplicate short value alias", v, all_shorts[i][s])
    121 			all_shorts[i][s] = v
    122 			short[i][v] = s
    123 		print "#define %s_%s	%s_%s	%s/* %3d chars; %s */" % \
    124 			(what_short[i], s, what[i], v.upper (), \
    125 			'	'* ((48-1 - len (what[i]) - 1 - len (v)) / 8), \
    126 			values[i][v], v)
    127 print
    128 print "#define _(S,M) INDIC_COMBINE_CATEGORIES (ISC_##S, IMC_##M)"
    129 print
    130 print
    131 
    132 total = 0
    133 used = 0
    134 last_block = None
    135 def print_block (block, start, end, data):
    136 	global total, used, last_block
    137 	if block and block != last_block:
    138 		print
    139 		print
    140 		print "  /* %s */" % block
    141 	num = 0
    142 	assert start % 8 == 0
    143 	assert (end+1) % 8 == 0
    144 	for u in range (start, end+1):
    145 		if u % 8 == 0:
    146 			print
    147 			print "  /* %04X */" % u,
    148 		if u in data:
    149 			num += 1
    150 		d = data.get (u, defaults)
    151 		sys.stdout.write ("%9s" % ("_(%s,%s)," % (short[0][d[0]], short[1][d[1]])))
    152 
    153 	total += end - start + 1
    154 	used += num
    155 	if block:
    156 		last_block = block
    157 
    158 uu = data.keys ()
    159 uu.sort ()
    160 
    161 last = -100000
    162 num = 0
    163 offset = 0
    164 starts = []
    165 ends = []
    166 print "static const INDIC_TABLE_ELEMENT_TYPE indic_table[] = {"
    167 for u in uu:
    168 	if u <= last:
    169 		continue
    170 	block = data[u][2]
    171 
    172 	start = u//8*8
    173 	end = start+1
    174 	while end in uu and block == data[end][2]:
    175 		end += 1
    176 	end = (end-1)//8*8 + 7
    177 
    178 	if start != last + 1:
    179 		if start - last <= 1+16*3:
    180 			print_block (None, last+1, start-1, data)
    181 			last = start-1
    182 		else:
    183 			if last >= 0:
    184 				ends.append (last + 1)
    185 				offset += ends[-1] - starts[-1]
    186 			print
    187 			print
    188 			print "#define indic_offset_0x%04xu %d" % (start, offset)
    189 			starts.append (start)
    190 
    191 	print_block (block, start, end, data)
    192 	last = end
    193 ends.append (last + 1)
    194 offset += ends[-1] - starts[-1]
    195 print
    196 print
    197 occupancy = used * 100. / total
    198 page_bits = 12
    199 print "}; /* Table items: %d; occupancy: %d%% */" % (offset, occupancy)
    200 print
    201 print "INDIC_TABLE_ELEMENT_TYPE"
    202 print "hb_indic_get_categories (hb_codepoint_t u)"
    203 print "{"
    204 print "  switch (u >> %d)" % page_bits
    205 print "  {"
    206 pages = set([u>>page_bits for u in starts+ends+singles.keys()])
    207 for p in sorted(pages):
    208 	print "    case 0x%0Xu:" % p
    209 	for (start,end) in zip (starts, ends):
    210 		if p not in [start>>page_bits, end>>page_bits]: continue
    211 		offset = "indic_offset_0x%04xu" % start
    212 		print "      if (hb_in_range (u, 0x%04Xu, 0x%04Xu)) return indic_table[u - 0x%04Xu + %s];" % (start, end, start, offset)
    213 	for u,d in singles.items ():
    214 		if p != u>>page_bits: continue
    215 		print "      if (unlikely (u == 0x%04Xu)) return _(%s,%s);" % (u, short[0][d[0]], short[1][d[1]])
    216 	print "      break;"
    217 	print ""
    218 print "    default:"
    219 print "      break;"
    220 print "  }"
    221 print "  return _(x,x);"
    222 print "}"
    223 print
    224 print "#undef _"
    225 for i in range (2):
    226 	print
    227 	vv = values[i].keys ()
    228 	vv.sort ()
    229 	for v in vv:
    230 		print "#undef %s_%s" % \
    231 			(what_short[i], short[i][v])
    232 print
    233 print "/* == End of generated table == */"
    234 
    235 # Maintain at least 30% occupancy in the table */
    236 if occupancy < 30:
    237 	raise Exception ("Table too sparse, please investigate: ", occupancy)
    238