/[pcre]/code/trunk/maint/MultiStage2.py
ViewVC logotype

Contents of /code/trunk/maint/MultiStage2.py

Parent Directory Parent Directory | Revision Log Revision Log


Revision 349 - (show annotations)
Wed Jul 2 18:42:11 2008 UTC (12 years, 4 months ago) by ph10
File MIME type: text/x-python
File size: 8840 byte(s)
Error occurred while calculating annotation data.
Preliminary patch for upgraded Unicode Property support.
1 #! /usr/bin/python
2
3 # Multistage table builder
4 # (c) Peter Kankowski, 2008
5
6 import re
7 import string
8 import sys
9
10 MAX_UNICODE = 0x110000
11 NOTACHAR = 0xffffffff
12
13 # Parse a line of CaseFolding.txt, Scripts.txt, and DerivedGeneralCategory.txt file
14 def make_get_names(enum):
15 return lambda chardata: enum.index(chardata[1])
16
17 def get_case_folding_value(chardata):
18 if chardata[1] != 'C' and chardata[1] != 'S':
19 return 0
20 return int(chardata[2], 16) - int(chardata[0], 16)
21
22 def get_other_case(chardata):
23 if chardata[12] != '':
24 return int(chardata[12], 16) - int(chardata[0], 16)
25 if chardata[13] != '':
26 return int(chardata[13], 16) - int(chardata[0], 16)
27 return 0
28
29 # Read the whole table in memory
30 def read_table(file_name, get_value, default_value):
31 file = open(file_name, 'r')
32 table = [default_value] * MAX_UNICODE
33 for line in file:
34 line = re.sub(r'#.*', '', line)
35 chardata = map(string.strip, line.split(';'))
36 if len(chardata) <= 1:
37 continue
38 value = get_value(chardata)
39
40 m = re.match(r'([0-9a-fA-F]+)(\.\.([0-9a-fA-F]+))?$', chardata[0])
41 char = int(m.group(1), 16)
42 #PH last = char if m.group(3) is None else int(m.group(3), 16)
43 if m.group(3) is None:
44 last = char
45 else:
46 last = int(m.group(3), 16)
47 for i in range(char, last + 1):
48 table[i] = value
49 file.close()
50 return table
51
52 # Get the smallest possible C language type for the values
53 def get_type_size(table):
54 type_size = [("uschar", 1), ("pcre_uint16", 2), ("pcre_uint32", 4),
55 ("signed char", 1), ("pcre_int16", 2), ("pcre_int32", 4)]
56 limits = [(0, 255), (0, 65535), (0, 4294967295),
57 (-128, 127), (-32768, 32767), (-2147483648, 2147483647)]
58 minval = min(table)
59 maxval = max(table)
60 for num, (minlimit, maxlimit) in enumerate(limits):
61 if minlimit <= minval and maxval <= maxlimit:
62 return type_size[num]
63 else:
64 raise OverflowError, "Too large to fit into C types"
65
66 def get_tables_size(*tables):
67 total_size = 0
68 for table in tables:
69 type, size = get_type_size(table)
70 total_size += size * len(table)
71 return total_size
72
73 # Compress the table into the two stages
74 def compress_table(table, block_size):
75 blocks = {} # Dictionary for finding identical blocks
76 stage1 = [] # Stage 1 table contains block numbers (indices into stage 2 table)
77 stage2 = [] # Stage 2 table contains the blocks with property values
78 table = tuple(table)
79 for i in range(0, len(table), block_size):
80 block = table[i:i+block_size]
81 start = blocks.get(block)
82 if start is None:
83 # Allocate a new block
84 start = len(stage2) / block_size
85 stage2 += block
86 blocks[block] = start
87 stage1.append(start)
88
89 return stage1, stage2
90
91 # Print a table
92 def print_table(table, table_name, block_size = None):
93 type, size = get_type_size(table)
94 ELEMS_PER_LINE = 16
95
96 s = "const %s %s[] = { /* %d bytes" % (type, table_name, size * len(table))
97 if block_size:
98 s += ", block = %d" % block_size
99 print s + " */"
100 table = tuple(table)
101 if block_size is None:
102 fmt = "%3d," * ELEMS_PER_LINE + " /* U+%04X */"
103 mult = MAX_UNICODE / len(table)
104 for i in range(0, len(table), ELEMS_PER_LINE):
105 print fmt % (table[i:i+ELEMS_PER_LINE] + (i * mult,))
106 else:
107 #PH fmt = "%3d," * (ELEMS_PER_LINE if block_size > ELEMS_PER_LINE else block_size) + "\n"
108 if block_size > ELEMS_PER_LINE:
109 fmt = "%3d," * ELEMS_PER_LINE + "\n"
110 else:
111 fmt = "%3d," * block_size + "\n"
112 if block_size > ELEMS_PER_LINE:
113 fmt = fmt * (block_size / ELEMS_PER_LINE)
114 for i in range(0, len(table), block_size):
115 print ("/* block %d */\n" + fmt) % ((i / block_size,) + table[i:i+block_size])
116 print "};\n"
117
118 # Extract the unique combinations of properties into records
119 def combine_tables(*tables):
120 records = {}
121 index = []
122 for t in zip(*tables):
123 i = records.get(t)
124 if i is None:
125 i = records[t] = len(records)
126 index.append(i)
127 return index, records
128
129 def print_records(records):
130 print 'const ucd_record ucd_records[] = { /* %d bytes */' % (len(records) * 4)
131 records = zip(records.keys(), records.values())
132 records.sort(None, lambda x: x[1])
133 for i, record in enumerate(records):
134 print (' {' + '%6d, ' * len(record[0]) + '}, /* %3d */') % (record[0] + (i,))
135 print '};\n'
136
137 script_names = ['Arabic', 'Armenian', 'Bengali', 'Bopomofo', 'Braille', 'Buginese', 'Buhid', 'Canadian_Aboriginal', \
138 'Cherokee', 'Common', 'Coptic', 'Cypriot', 'Cyrillic', 'Deseret', 'Devanagari', 'Ethiopic', 'Georgian', \
139 'Glagolitic', 'Gothic', 'Greek', 'Gujarati', 'Gurmukhi', 'Han', 'Hangul', 'Hanunoo', 'Hebrew', 'Hiragana', \
140 'Inherited', 'Kannada', 'Katakana', 'Kharoshthi', 'Khmer', 'Lao', 'Latin', 'Limbu', 'Linear_B', 'Malayalam', \
141 'Mongolian', 'Myanmar', 'New_Tai_Lue', 'Ogham', 'Old_Italic', 'Old_Persian', 'Oriya', 'Osmanya', 'Runic', \
142 'Shavian', 'Sinhala', 'Syloti_Nagri', 'Syriac', 'Tagalog', 'Tagbanwa', 'Tai_Le', 'Tamil', 'Telugu', 'Thaana', \
143 'Thai', 'Tibetan', 'Tifinagh', 'Ugaritic', 'Yi', \
144 'Balinese', 'Cuneiform', 'Nko', 'Phags_Pa', 'Phoenician']
145
146 category_names = ['Cc', 'Cf', 'Cn', 'Co', 'Cs', 'Ll', 'Lm', 'Lo', 'Lt', 'Lu',
147 'Mc', 'Me', 'Mn', 'Nd', 'Nl', 'No', 'Pc', 'Pd', 'Pe', 'Pf', 'Pi', 'Po', 'Ps',
148 'Sc', 'Sk', 'Sm', 'So', 'Zl', 'Zp', 'Zs' ]
149
150
151 script = read_table('Unicode.tables/Scripts.txt', make_get_names(script_names), script_names.index('Common'))
152 category = read_table('Unicode.tables/DerivedGeneralCategory.txt', make_get_names(category_names), category_names.index('Cn'))
153 other_case = read_table('Unicode.tables/UnicodeData.txt', get_other_case, 0)
154 # case_fold = read_table('CaseFolding.txt', get_case_folding_value, 0)
155
156 table, records = combine_tables(script, category, other_case)
157
158 # Find the optimum block size for the two-stage table
159 min_size = sys.maxint
160 for block_size in [2 ** i for i in range(5,10)]:
161 size = len(records) * 4
162 stage1, stage2 = compress_table(table, block_size)
163 size += get_tables_size(stage1, stage2)
164 #print "/* block size %5d => %5d bytes */" % (block_size, size)
165 if size < min_size:
166 min_size = size
167 min_stage1, min_stage2 = stage1, stage2
168 min_block_size = block_size
169
170 print "#ifdef HAVE_CONFIG_H"
171 print "#include \"config.h\""
172 print "#endif"
173 print "#include \"pcre_internal.h\""
174 print
175 print "/* Unicode character database. */"
176 print "/* This file was autogenerated by MultiStage2.py script. */"
177 print "/* Total size: %d bytes, block size: %d. */" % (min_size, min_block_size)
178 print_records(records)
179 print_table(min_stage1, 'ucd_stage1')
180 print_table(min_stage2, 'ucd_stage2', min_block_size)
181 print "#if UCD_BLOCK_SIZE != %d" % min_block_size
182 print "#error Please correct UCD_BLOCK_SIZE in pcre_internal.h"
183 print "#endif"
184
185 """
186
187 # Three-stage tables:
188
189 # Find the optimum block size for 3-stage table
190 min_size = sys.maxint
191 for stage3_block in [2 ** i for i in range(2,6)]:
192 stage_i, stage3 = compress_table(table, stage3_block)
193 for stage2_block in [2 ** i for i in range(5,10)]:
194 size = len(records) * 4
195 stage1, stage2 = compress_table(stage_i, stage2_block)
196 size += get_tables_size(stage1, stage2, stage3)
197 # print "/* %5d / %3d => %5d bytes */" % (stage2_block, stage3_block, size)
198 if size < min_size:
199 min_size = size
200 min_stage1, min_stage2, min_stage3 = stage1, stage2, stage3
201 min_stage2_block, min_stage3_block = stage2_block, stage3_block
202
203 print "/* Total size: %d bytes" % min_size */
204 print_records(records)
205 print_table(min_stage1, 'ucd_stage1')
206 print_table(min_stage2, 'ucd_stage2', min_stage2_block)
207 print_table(min_stage3, 'ucd_stage3', min_stage3_block)
208
209 """

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.5