1#!/usr/bin/python
2
3import sys
4
5if len (sys.argv) != 4:
6	print >>sys.stderr, "usage: ./gen-indic-table.py IndicSyllabicCategory.txt IndicMatraCategory.txt Blocks.txt"
7	sys.exit (1)
8
9BLACKLISTED_BLOCKS = ["Thai", "Lao", "Tibetan"]
10
11files = [file (x) for x in sys.argv[1:]]
12
13headers = [[f.readline () for i in range (2)] for f in files]
14
15data = [{} for f in files]
16values = [{} for f in files]
17for 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:
42defaults = ('Other', 'Not_Applicable', 'No_Block')
43for i,v in enumerate (defaults):
44	values[i][v] = values[i].get (v, 0) + 1
45combined = {}
46for 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
53combined = {k:v for k,v in combined.items() if v[2] not in BLACKLISTED_BLOCKS}
54data = combined
55del combined
56num = len (data)
57
58for 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
63singles = {}
64for u in [0x00A0, 0x25CC]:
65	singles[u] = data[u]
66	del data[u]
67
68print "/* == Start of generated table == */"
69print "/*"
70print " * The following table is generated by running:"
71print " *"
72print " *   ./gen-indic-table.py IndicSyllabicCategory.txt IndicMatraCategory.txt Blocks.txt"
73print " *"
74print " * on files with these headers:"
75print " *"
76for h in headers:
77	for l in h:
78		print " * %s" % (l.strip())
79print " */"
80print
81print '#include "hb-ot-shape-complex-indic-private.hh"'
82print
83
84# Shorten values
85short = [{
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	"Consonant_Prefixed":	'CPrf',
95	"Other":		'x',
96},{
97	"Not_Applicable":	'x',
98}]
99all_shorts = [{},{}]
100
101# Add some of the values, to make them more readable, and to avoid duplicates
102
103
104for i in range (2):
105	for v,s in short[i].items ():
106		all_shorts[i][s] = v
107
108what = ["INDIC_SYLLABIC_CATEGORY", "INDIC_MATRA_CATEGORY"]
109what_short = ["ISC", "IMC"]
110for i in range (2):
111	print
112	vv = values[i].keys ()
113	vv.sort ()
114	for v in vv:
115		v_no_and = v.replace ('_And_', '_')
116		if v in short[i]:
117			s = short[i][v]
118		else:
119			s = ''.join ([c for c in v_no_and if ord ('A') <= ord (c) <= ord ('Z')])
120			if s in all_shorts[i]:
121				raise Exception ("Duplicate short value alias", v, all_shorts[i][s])
122			all_shorts[i][s] = v
123			short[i][v] = s
124		print "#define %s_%s	%s_%s	%s/* %3d chars; %s */" % \
125			(what_short[i], s, what[i], v.upper (), \
126			'	'* ((48-1 - len (what[i]) - 1 - len (v)) / 8), \
127			values[i][v], v)
128print
129print "#define _(S,M) INDIC_COMBINE_CATEGORIES (ISC_##S, IMC_##M)"
130print
131print
132
133total = 0
134used = 0
135last_block = None
136def print_block (block, start, end, data):
137	global total, used, last_block
138	if block and block != last_block:
139		print
140		print
141		print "  /* %s */" % block
142	num = 0
143	assert start % 8 == 0
144	assert (end+1) % 8 == 0
145	for u in range (start, end+1):
146		if u % 8 == 0:
147			print
148			print "  /* %04X */" % u,
149		if u in data:
150			num += 1
151		d = data.get (u, defaults)
152		sys.stdout.write ("%9s" % ("_(%s,%s)," % (short[0][d[0]], short[1][d[1]])))
153
154	total += end - start + 1
155	used += num
156	if block:
157		last_block = block
158
159uu = data.keys ()
160uu.sort ()
161
162last = -100000
163num = 0
164offset = 0
165starts = []
166ends = []
167print "static const INDIC_TABLE_ELEMENT_TYPE indic_table[] = {"
168for u in uu:
169	if u <= last:
170		continue
171	block = data[u][2]
172
173	start = u//8*8
174	end = start+1
175	while end in uu and block == data[end][2]:
176		end += 1
177	end = (end-1)//8*8 + 7
178
179	if start != last + 1:
180		if start - last <= 1+16*3:
181			print_block (None, last+1, start-1, data)
182			last = start-1
183		else:
184			if last >= 0:
185				ends.append (last + 1)
186				offset += ends[-1] - starts[-1]
187			print
188			print
189			print "#define indic_offset_0x%04xu %d" % (start, offset)
190			starts.append (start)
191
192	print_block (block, start, end, data)
193	last = end
194ends.append (last + 1)
195offset += ends[-1] - starts[-1]
196print
197print
198occupancy = used * 100. / total
199page_bits = 12
200print "}; /* Table items: %d; occupancy: %d%% */" % (offset, occupancy)
201print
202print "INDIC_TABLE_ELEMENT_TYPE"
203print "hb_indic_get_categories (hb_codepoint_t u)"
204print "{"
205print "  switch (u >> %d)" % page_bits
206print "  {"
207pages = set([u>>page_bits for u in starts+ends+singles.keys()])
208for p in sorted(pages):
209	print "    case 0x%0Xu:" % p
210	for (start,end) in zip (starts, ends):
211		if p not in [start>>page_bits, end>>page_bits]: continue
212		offset = "indic_offset_0x%04xu" % start
213		print "      if (hb_in_range (u, 0x%04Xu, 0x%04Xu)) return indic_table[u - 0x%04Xu + %s];" % (start, end-1, start, offset)
214	for u,d in singles.items ():
215		if p != u>>page_bits: continue
216		print "      if (unlikely (u == 0x%04Xu)) return _(%s,%s);" % (u, short[0][d[0]], short[1][d[1]])
217	print "      break;"
218	print ""
219print "    default:"
220print "      break;"
221print "  }"
222print "  return _(x,x);"
223print "}"
224print
225print "#undef _"
226for i in range (2):
227	print
228	vv = values[i].keys ()
229	vv.sort ()
230	for v in vv:
231		print "#undef %s_%s" % \
232			(what_short[i], short[i][v])
233print
234print "/* == End of generated table == */"
235
236# Maintain at least 30% occupancy in the table */
237if occupancy < 30:
238	raise Exception ("Table too sparse, please investigate: ", occupancy)
239