author | martin.trojer@nokia.com |
Fri, 31 Jul 2009 15:01:17 +0100 | |
changeset 1 | 2fb8b9db1c86 |
permissions | -rw-r--r-- |
1
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
1 |
# |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
2 |
# (re)generate unicode property and type databases |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
3 |
# |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
4 |
# this script converts a unicode 3.2 database file to |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
5 |
# Modules/unicodedata_db.h, Modules/unicodename_db.h, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
6 |
# and Objects/unicodetype_db.h |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
7 |
# |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
8 |
# history: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
9 |
# 2000-09-24 fl created (based on bits and pieces from unidb) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
10 |
# 2000-09-25 fl merged tim's splitbin fixes, separate decomposition table |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
11 |
# 2000-09-25 fl added character type table |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
12 |
# 2000-09-26 fl added LINEBREAK, DECIMAL, and DIGIT flags/fields (2.0) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
13 |
# 2000-11-03 fl expand first/last ranges |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
14 |
# 2001-01-19 fl added character name tables (2.1) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
15 |
# 2001-01-21 fl added decomp compression; dynamic phrasebook threshold |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
16 |
# 2002-09-11 wd use string methods |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
17 |
# 2002-10-18 mvl update to Unicode 3.2 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
18 |
# 2002-10-22 mvl generate NFC tables |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
19 |
# 2002-11-24 mvl expand all ranges, sort names version-independently |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
20 |
# 2002-11-25 mvl add UNIDATA_VERSION |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
21 |
# 2004-05-29 perky add east asian width information |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
22 |
# 2006-03-10 mvl update to Unicode 4.1; add UCD 3.2 delta |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
23 |
# |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
24 |
# written by Fredrik Lundh (fredrik@pythonware.com) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
25 |
# |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
26 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
27 |
import sys |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
28 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
29 |
SCRIPT = sys.argv[0] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
30 |
VERSION = "2.6" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
31 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
32 |
# The Unicode Database |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
33 |
UNIDATA_VERSION = "5.1.0" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
34 |
UNICODE_DATA = "UnicodeData%s.txt" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
35 |
COMPOSITION_EXCLUSIONS = "CompositionExclusions%s.txt" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
36 |
EASTASIAN_WIDTH = "EastAsianWidth%s.txt" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
37 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
38 |
old_versions = ["3.2.0"] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
39 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
40 |
CATEGORY_NAMES = [ "Cn", "Lu", "Ll", "Lt", "Mn", "Mc", "Me", "Nd", |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
41 |
"Nl", "No", "Zs", "Zl", "Zp", "Cc", "Cf", "Cs", "Co", "Cn", "Lm", |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
42 |
"Lo", "Pc", "Pd", "Ps", "Pe", "Pi", "Pf", "Po", "Sm", "Sc", "Sk", |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
43 |
"So" ] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
44 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
45 |
BIDIRECTIONAL_NAMES = [ "", "L", "LRE", "LRO", "R", "AL", "RLE", "RLO", |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
46 |
"PDF", "EN", "ES", "ET", "AN", "CS", "NSM", "BN", "B", "S", "WS", |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
47 |
"ON" ] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
48 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
49 |
EASTASIANWIDTH_NAMES = [ "F", "H", "W", "Na", "A", "N" ] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
50 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
51 |
# note: should match definitions in Objects/unicodectype.c |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
52 |
ALPHA_MASK = 0x01 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
53 |
DECIMAL_MASK = 0x02 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
54 |
DIGIT_MASK = 0x04 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
55 |
LOWER_MASK = 0x08 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
56 |
LINEBREAK_MASK = 0x10 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
57 |
SPACE_MASK = 0x20 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
58 |
TITLE_MASK = 0x40 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
59 |
UPPER_MASK = 0x80 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
60 |
NODELTA_MASK = 0x100 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
61 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
62 |
def maketables(trace=0): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
63 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
64 |
print "--- Reading", UNICODE_DATA % "", "..." |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
65 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
66 |
version = "" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
67 |
unicode = UnicodeData(UNICODE_DATA % version, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
68 |
COMPOSITION_EXCLUSIONS % version, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
69 |
EASTASIAN_WIDTH % version) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
70 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
71 |
print len(filter(None, unicode.table)), "characters" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
72 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
73 |
for version in old_versions: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
74 |
print "--- Reading", UNICODE_DATA % ("-"+version), "..." |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
75 |
old_unicode = UnicodeData(UNICODE_DATA % ("-"+version), |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
76 |
COMPOSITION_EXCLUSIONS % ("-"+version), |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
77 |
EASTASIAN_WIDTH % ("-"+version)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
78 |
print len(filter(None, old_unicode.table)), "characters" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
79 |
merge_old_version(version, unicode, old_unicode) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
80 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
81 |
makeunicodename(unicode, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
82 |
makeunicodedata(unicode, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
83 |
makeunicodetype(unicode, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
84 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
85 |
# -------------------------------------------------------------------- |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
86 |
# unicode character properties |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
87 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
88 |
def makeunicodedata(unicode, trace): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
89 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
90 |
dummy = (0, 0, 0, 0, 0) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
91 |
table = [dummy] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
92 |
cache = {0: dummy} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
93 |
index = [0] * len(unicode.chars) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
94 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
95 |
FILE = "Modules/unicodedata_db.h" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
96 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
97 |
print "--- Preparing", FILE, "..." |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
98 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
99 |
# 1) database properties |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
100 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
101 |
for char in unicode.chars: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
102 |
record = unicode.table[char] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
103 |
if record: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
104 |
# extract database properties |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
105 |
category = CATEGORY_NAMES.index(record[2]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
106 |
combining = int(record[3]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
107 |
bidirectional = BIDIRECTIONAL_NAMES.index(record[4]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
108 |
mirrored = record[9] == "Y" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
109 |
eastasianwidth = EASTASIANWIDTH_NAMES.index(record[15]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
110 |
item = ( |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
111 |
category, combining, bidirectional, mirrored, eastasianwidth |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
112 |
) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
113 |
# add entry to index and item tables |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
114 |
i = cache.get(item) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
115 |
if i is None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
116 |
cache[item] = i = len(table) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
117 |
table.append(item) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
118 |
index[char] = i |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
119 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
120 |
# 2) decomposition data |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
121 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
122 |
decomp_data = [0] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
123 |
decomp_prefix = [""] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
124 |
decomp_index = [0] * len(unicode.chars) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
125 |
decomp_size = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
126 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
127 |
comp_pairs = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
128 |
comp_first = [None] * len(unicode.chars) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
129 |
comp_last = [None] * len(unicode.chars) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
130 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
131 |
for char in unicode.chars: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
132 |
record = unicode.table[char] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
133 |
if record: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
134 |
if record[5]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
135 |
decomp = record[5].split() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
136 |
if len(decomp) > 19: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
137 |
raise Exception, "character %x has a decomposition too large for nfd_nfkd" % char |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
138 |
# prefix |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
139 |
if decomp[0][0] == "<": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
140 |
prefix = decomp.pop(0) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
141 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
142 |
prefix = "" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
143 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
144 |
i = decomp_prefix.index(prefix) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
145 |
except ValueError: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
146 |
i = len(decomp_prefix) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
147 |
decomp_prefix.append(prefix) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
148 |
prefix = i |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
149 |
assert prefix < 256 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
150 |
# content |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
151 |
decomp = [prefix + (len(decomp)<<8)] +\ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
152 |
map(lambda s: int(s, 16), decomp) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
153 |
# Collect NFC pairs |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
154 |
if not prefix and len(decomp) == 3 and \ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
155 |
char not in unicode.exclusions and \ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
156 |
unicode.table[decomp[1]][3] == "0": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
157 |
p, l, r = decomp |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
158 |
comp_first[l] = 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
159 |
comp_last[r] = 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
160 |
comp_pairs.append((l,r,char)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
161 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
162 |
i = decomp_data.index(decomp) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
163 |
except ValueError: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
164 |
i = len(decomp_data) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
165 |
decomp_data.extend(decomp) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
166 |
decomp_size = decomp_size + len(decomp) * 2 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
167 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
168 |
i = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
169 |
decomp_index[char] = i |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
170 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
171 |
f = l = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
172 |
comp_first_ranges = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
173 |
comp_last_ranges = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
174 |
prev_f = prev_l = None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
175 |
for i in unicode.chars: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
176 |
if comp_first[i] is not None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
177 |
comp_first[i] = f |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
178 |
f += 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
179 |
if prev_f is None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
180 |
prev_f = (i,i) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
181 |
elif prev_f[1]+1 == i: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
182 |
prev_f = prev_f[0],i |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
183 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
184 |
comp_first_ranges.append(prev_f) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
185 |
prev_f = (i,i) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
186 |
if comp_last[i] is not None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
187 |
comp_last[i] = l |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
188 |
l += 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
189 |
if prev_l is None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
190 |
prev_l = (i,i) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
191 |
elif prev_l[1]+1 == i: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
192 |
prev_l = prev_l[0],i |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
193 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
194 |
comp_last_ranges.append(prev_l) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
195 |
prev_l = (i,i) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
196 |
comp_first_ranges.append(prev_f) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
197 |
comp_last_ranges.append(prev_l) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
198 |
total_first = f |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
199 |
total_last = l |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
200 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
201 |
comp_data = [0]*(total_first*total_last) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
202 |
for f,l,char in comp_pairs: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
203 |
f = comp_first[f] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
204 |
l = comp_last[l] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
205 |
comp_data[f*total_last+l] = char |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
206 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
207 |
print len(table), "unique properties" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
208 |
print len(decomp_prefix), "unique decomposition prefixes" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
209 |
print len(decomp_data), "unique decomposition entries:", |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
210 |
print decomp_size, "bytes" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
211 |
print total_first, "first characters in NFC" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
212 |
print total_last, "last characters in NFC" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
213 |
print len(comp_pairs), "NFC pairs" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
214 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
215 |
print "--- Writing", FILE, "..." |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
216 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
217 |
fp = open(FILE, "w") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
218 |
print >>fp, "/* this file was generated by %s %s */" % (SCRIPT, VERSION) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
219 |
print >>fp |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
220 |
print >>fp, '#define UNIDATA_VERSION "%s"' % UNIDATA_VERSION |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
221 |
print >>fp, "/* a list of unique database records */" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
222 |
print >>fp, \ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
223 |
"const _PyUnicode_DatabaseRecord _PyUnicode_Database_Records[] = {" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
224 |
for item in table: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
225 |
print >>fp, " {%d, %d, %d, %d, %d}," % item |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
226 |
print >>fp, "};" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
227 |
print >>fp |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
228 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
229 |
print >>fp, "/* Reindexing of NFC first characters. */" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
230 |
print >>fp, "#define TOTAL_FIRST",total_first |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
231 |
print >>fp, "#define TOTAL_LAST",total_last |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
232 |
print >>fp, "struct reindex{int start;short count,index;};" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
233 |
print >>fp, "static struct reindex nfc_first[] = {" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
234 |
for start,end in comp_first_ranges: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
235 |
print >>fp," { %d, %d, %d}," % (start,end-start,comp_first[start]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
236 |
print >>fp," {0,0,0}" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
237 |
print >>fp,"};\n" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
238 |
print >>fp, "static struct reindex nfc_last[] = {" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
239 |
for start,end in comp_last_ranges: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
240 |
print >>fp," { %d, %d, %d}," % (start,end-start,comp_last[start]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
241 |
print >>fp," {0,0,0}" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
242 |
print >>fp,"};\n" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
243 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
244 |
# FIXME: <fl> the following tables could be made static, and |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
245 |
# the support code moved into unicodedatabase.c |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
246 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
247 |
print >>fp, "/* string literals */" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
248 |
print >>fp, "const char *_PyUnicode_CategoryNames[] = {" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
249 |
for name in CATEGORY_NAMES: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
250 |
print >>fp, " \"%s\"," % name |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
251 |
print >>fp, " NULL" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
252 |
print >>fp, "};" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
253 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
254 |
print >>fp, "const char *_PyUnicode_BidirectionalNames[] = {" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
255 |
for name in BIDIRECTIONAL_NAMES: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
256 |
print >>fp, " \"%s\"," % name |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
257 |
print >>fp, " NULL" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
258 |
print >>fp, "};" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
259 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
260 |
print >>fp, "const char *_PyUnicode_EastAsianWidthNames[] = {" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
261 |
for name in EASTASIANWIDTH_NAMES: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
262 |
print >>fp, " \"%s\"," % name |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
263 |
print >>fp, " NULL" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
264 |
print >>fp, "};" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
265 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
266 |
print >>fp, "static const char *decomp_prefix[] = {" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
267 |
for name in decomp_prefix: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
268 |
print >>fp, " \"%s\"," % name |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
269 |
print >>fp, " NULL" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
270 |
print >>fp, "};" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
271 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
272 |
# split record index table |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
273 |
index1, index2, shift = splitbins(index, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
274 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
275 |
print >>fp, "/* index tables for the database records */" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
276 |
print >>fp, "#define SHIFT", shift |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
277 |
Array("index1", index1).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
278 |
Array("index2", index2).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
279 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
280 |
# split decomposition index table |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
281 |
index1, index2, shift = splitbins(decomp_index, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
282 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
283 |
print >>fp, "/* decomposition data */" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
284 |
Array("decomp_data", decomp_data).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
285 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
286 |
print >>fp, "/* index tables for the decomposition data */" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
287 |
print >>fp, "#define DECOMP_SHIFT", shift |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
288 |
Array("decomp_index1", index1).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
289 |
Array("decomp_index2", index2).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
290 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
291 |
index, index2, shift = splitbins(comp_data, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
292 |
print >>fp, "/* NFC pairs */" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
293 |
print >>fp, "#define COMP_SHIFT", shift |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
294 |
Array("comp_index", index).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
295 |
Array("comp_data", index2).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
296 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
297 |
# Generate delta tables for old versions |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
298 |
for version, table, normalization in unicode.changed: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
299 |
cversion = version.replace(".","_") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
300 |
records = [table[0]] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
301 |
cache = {table[0]:0} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
302 |
index = [0] * len(table) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
303 |
for i, record in enumerate(table): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
304 |
try: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
305 |
index[i] = cache[record] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
306 |
except KeyError: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
307 |
index[i] = cache[record] = len(records) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
308 |
records.append(record) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
309 |
index1, index2, shift = splitbins(index, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
310 |
print >>fp, "static const change_record change_records_%s[] = {" % cversion |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
311 |
for record in records: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
312 |
print >>fp, "\t{ %s }," % ", ".join(map(str,record)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
313 |
print >>fp, "};" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
314 |
Array("changes_%s_index" % cversion, index1).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
315 |
Array("changes_%s_data" % cversion, index2).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
316 |
print >>fp, "static const change_record* get_change_%s(Py_UCS4 n)" % cversion |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
317 |
print >>fp, "{" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
318 |
print >>fp, "\tint index;" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
319 |
print >>fp, "\tif (n >= 0x110000) index = 0;" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
320 |
print >>fp, "\telse {" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
321 |
print >>fp, "\t\tindex = changes_%s_index[n>>%d];" % (cversion, shift) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
322 |
print >>fp, "\t\tindex = changes_%s_data[(index<<%d)+(n & %d)];" % \ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
323 |
(cversion, shift, ((1<<shift)-1)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
324 |
print >>fp, "\t}" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
325 |
print >>fp, "\treturn change_records_%s+index;" % cversion |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
326 |
print >>fp, "}\n" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
327 |
print >>fp, "static Py_UCS4 normalization_%s(Py_UCS4 n)" % cversion |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
328 |
print >>fp, "{" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
329 |
print >>fp, "\tswitch(n) {" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
330 |
for k, v in normalization: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
331 |
print >>fp, "\tcase %s: return 0x%s;" % (hex(k), v) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
332 |
print >>fp, "\tdefault: return 0;" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
333 |
print >>fp, "\t}\n}\n" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
334 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
335 |
fp.close() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
336 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
337 |
# -------------------------------------------------------------------- |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
338 |
# unicode character type tables |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
339 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
340 |
def makeunicodetype(unicode, trace): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
341 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
342 |
FILE = "Objects/unicodetype_db.h" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
343 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
344 |
print "--- Preparing", FILE, "..." |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
345 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
346 |
# extract unicode types |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
347 |
dummy = (0, 0, 0, 0, 0, 0) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
348 |
table = [dummy] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
349 |
cache = {0: dummy} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
350 |
index = [0] * len(unicode.chars) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
351 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
352 |
for char in unicode.chars: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
353 |
record = unicode.table[char] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
354 |
if record: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
355 |
# extract database properties |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
356 |
category = record[2] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
357 |
bidirectional = record[4] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
358 |
flags = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
359 |
delta = True |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
360 |
if category in ["Lm", "Lt", "Lu", "Ll", "Lo"]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
361 |
flags |= ALPHA_MASK |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
362 |
if category == "Ll": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
363 |
flags |= LOWER_MASK |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
364 |
if category == "Zl" or bidirectional == "B": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
365 |
flags |= LINEBREAK_MASK |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
366 |
if category == "Zs" or bidirectional in ("WS", "B", "S"): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
367 |
flags |= SPACE_MASK |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
368 |
if category == "Lt": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
369 |
flags |= TITLE_MASK |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
370 |
if category == "Lu": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
371 |
flags |= UPPER_MASK |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
372 |
# use delta predictor for upper/lower/title if it fits |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
373 |
if record[12]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
374 |
upper = int(record[12], 16) - char |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
375 |
if -32768 <= upper <= 32767 and delta: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
376 |
upper = upper & 0xffff |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
377 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
378 |
upper += char |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
379 |
delta = False |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
380 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
381 |
upper = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
382 |
if record[13]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
383 |
lower = int(record[13], 16) - char |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
384 |
if -32768 <= lower <= 32767 and delta: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
385 |
lower = lower & 0xffff |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
386 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
387 |
lower += char |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
388 |
delta = False |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
389 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
390 |
lower = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
391 |
if record[14]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
392 |
title = int(record[14], 16) - char |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
393 |
if -32768 <= lower <= 32767 and delta: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
394 |
title = title & 0xffff |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
395 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
396 |
title += char |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
397 |
delta = False |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
398 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
399 |
title = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
400 |
if not delta: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
401 |
flags |= NODELTA_MASK |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
402 |
# decimal digit, integer digit |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
403 |
decimal = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
404 |
if record[6]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
405 |
flags |= DECIMAL_MASK |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
406 |
decimal = int(record[6]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
407 |
digit = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
408 |
if record[7]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
409 |
flags |= DIGIT_MASK |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
410 |
digit = int(record[7]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
411 |
item = ( |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
412 |
upper, lower, title, decimal, digit, flags |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
413 |
) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
414 |
# add entry to index and item tables |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
415 |
i = cache.get(item) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
416 |
if i is None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
417 |
cache[item] = i = len(table) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
418 |
table.append(item) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
419 |
index[char] = i |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
420 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
421 |
print len(table), "unique character type entries" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
422 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
423 |
print "--- Writing", FILE, "..." |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
424 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
425 |
fp = open(FILE, "w") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
426 |
print >>fp, "/* this file was generated by %s %s */" % (SCRIPT, VERSION) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
427 |
print >>fp |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
428 |
print >>fp, "/* a list of unique character type descriptors */" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
429 |
print >>fp, "const _PyUnicode_TypeRecord _PyUnicode_TypeRecords[] = {" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
430 |
for item in table: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
431 |
print >>fp, " {%d, %d, %d, %d, %d, %d}," % item |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
432 |
print >>fp, "};" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
433 |
print >>fp |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
434 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
435 |
# split decomposition index table |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
436 |
index1, index2, shift = splitbins(index, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
437 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
438 |
print >>fp, "/* type indexes */" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
439 |
print >>fp, "#define SHIFT", shift |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
440 |
Array("index1", index1).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
441 |
Array("index2", index2).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
442 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
443 |
fp.close() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
444 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
445 |
# -------------------------------------------------------------------- |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
446 |
# unicode name database |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
447 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
448 |
def makeunicodename(unicode, trace): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
449 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
450 |
FILE = "Modules/unicodename_db.h" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
451 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
452 |
print "--- Preparing", FILE, "..." |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
453 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
454 |
# collect names |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
455 |
names = [None] * len(unicode.chars) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
456 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
457 |
for char in unicode.chars: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
458 |
record = unicode.table[char] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
459 |
if record: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
460 |
name = record[1].strip() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
461 |
if name and name[0] != "<": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
462 |
names[char] = name + chr(0) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
463 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
464 |
print len(filter(lambda n: n is not None, names)), "distinct names" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
465 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
466 |
# collect unique words from names (note that we differ between |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
467 |
# words inside a sentence, and words ending a sentence. the |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
468 |
# latter includes the trailing null byte. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
469 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
470 |
words = {} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
471 |
n = b = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
472 |
for char in unicode.chars: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
473 |
name = names[char] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
474 |
if name: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
475 |
w = name.split() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
476 |
b = b + len(name) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
477 |
n = n + len(w) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
478 |
for w in w: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
479 |
l = words.get(w) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
480 |
if l: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
481 |
l.append(None) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
482 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
483 |
words[w] = [len(words)] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
484 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
485 |
print n, "words in text;", b, "bytes" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
486 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
487 |
wordlist = words.items() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
488 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
489 |
# sort on falling frequency, then by name |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
490 |
def cmpwords((aword, alist),(bword, blist)): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
491 |
r = -cmp(len(alist),len(blist)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
492 |
if r: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
493 |
return r |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
494 |
return cmp(aword, bword) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
495 |
wordlist.sort(cmpwords) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
496 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
497 |
# figure out how many phrasebook escapes we need |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
498 |
escapes = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
499 |
while escapes * 256 < len(wordlist): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
500 |
escapes = escapes + 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
501 |
print escapes, "escapes" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
502 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
503 |
short = 256 - escapes |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
504 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
505 |
assert short > 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
506 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
507 |
print short, "short indexes in lexicon" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
508 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
509 |
# statistics |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
510 |
n = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
511 |
for i in range(short): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
512 |
n = n + len(wordlist[i][1]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
513 |
print n, "short indexes in phrasebook" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
514 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
515 |
# pick the most commonly used words, and sort the rest on falling |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
516 |
# length (to maximize overlap) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
517 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
518 |
wordlist, wordtail = wordlist[:short], wordlist[short:] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
519 |
wordtail.sort(lambda a, b: len(b[0])-len(a[0])) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
520 |
wordlist.extend(wordtail) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
521 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
522 |
# generate lexicon from words |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
523 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
524 |
lexicon_offset = [0] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
525 |
lexicon = "" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
526 |
words = {} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
527 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
528 |
# build a lexicon string |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
529 |
offset = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
530 |
for w, x in wordlist: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
531 |
# encoding: bit 7 indicates last character in word (chr(128) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
532 |
# indicates the last character in an entire string) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
533 |
ww = w[:-1] + chr(ord(w[-1])+128) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
534 |
# reuse string tails, when possible |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
535 |
o = lexicon.find(ww) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
536 |
if o < 0: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
537 |
o = offset |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
538 |
lexicon = lexicon + ww |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
539 |
offset = offset + len(w) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
540 |
words[w] = len(lexicon_offset) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
541 |
lexicon_offset.append(o) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
542 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
543 |
lexicon = map(ord, lexicon) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
544 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
545 |
# generate phrasebook from names and lexicon |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
546 |
phrasebook = [0] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
547 |
phrasebook_offset = [0] * len(unicode.chars) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
548 |
for char in unicode.chars: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
549 |
name = names[char] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
550 |
if name: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
551 |
w = name.split() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
552 |
phrasebook_offset[char] = len(phrasebook) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
553 |
for w in w: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
554 |
i = words[w] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
555 |
if i < short: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
556 |
phrasebook.append(i) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
557 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
558 |
# store as two bytes |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
559 |
phrasebook.append((i>>8) + short) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
560 |
phrasebook.append(i&255) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
561 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
562 |
assert getsize(phrasebook) == 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
563 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
564 |
# |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
565 |
# unicode name hash table |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
566 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
567 |
# extract names |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
568 |
data = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
569 |
for char in unicode.chars: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
570 |
record = unicode.table[char] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
571 |
if record: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
572 |
name = record[1].strip() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
573 |
if name and name[0] != "<": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
574 |
data.append((name, char)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
575 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
576 |
# the magic number 47 was chosen to minimize the number of |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
577 |
# collisions on the current data set. if you like, change it |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
578 |
# and see what happens... |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
579 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
580 |
codehash = Hash("code", data, 47) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
581 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
582 |
print "--- Writing", FILE, "..." |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
583 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
584 |
fp = open(FILE, "w") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
585 |
print >>fp, "/* this file was generated by %s %s */" % (SCRIPT, VERSION) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
586 |
print >>fp |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
587 |
print >>fp, "#define NAME_MAXLEN", 256 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
588 |
print >>fp |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
589 |
print >>fp, "/* lexicon */" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
590 |
Array("lexicon", lexicon).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
591 |
Array("lexicon_offset", lexicon_offset).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
592 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
593 |
# split decomposition index table |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
594 |
offset1, offset2, shift = splitbins(phrasebook_offset, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
595 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
596 |
print >>fp, "/* code->name phrasebook */" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
597 |
print >>fp, "#define phrasebook_shift", shift |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
598 |
print >>fp, "#define phrasebook_short", short |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
599 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
600 |
Array("phrasebook", phrasebook).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
601 |
Array("phrasebook_offset1", offset1).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
602 |
Array("phrasebook_offset2", offset2).dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
603 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
604 |
print >>fp, "/* name->code dictionary */" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
605 |
codehash.dump(fp, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
606 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
607 |
fp.close() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
608 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
609 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
610 |
def merge_old_version(version, new, old): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
611 |
# Changes to exclusion file not implemented yet |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
612 |
if old.exclusions != new.exclusions: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
613 |
raise NotImplementedError, "exclusions differ" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
614 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
615 |
# In these change records, 0xFF means "no change" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
616 |
bidir_changes = [0xFF]*0x110000 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
617 |
category_changes = [0xFF]*0x110000 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
618 |
decimal_changes = [0xFF]*0x110000 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
619 |
mirrored_changes = [0xFF]*0x110000 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
620 |
# In numeric data, 0 means "no change", |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
621 |
# -1 means "did not have a numeric value |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
622 |
numeric_changes = [0] * 0x110000 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
623 |
# normalization_changes is a list of key-value pairs |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
624 |
normalization_changes = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
625 |
for i in range(0x110000): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
626 |
if new.table[i] is None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
627 |
# Characters unassigned in the new version ought to |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
628 |
# be unassigned in the old one |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
629 |
assert old.table[i] is None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
630 |
continue |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
631 |
# check characters unassigned in the old version |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
632 |
if old.table[i] is None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
633 |
# category 0 is "unassigned" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
634 |
category_changes[i] = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
635 |
continue |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
636 |
# check characters that differ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
637 |
if old.table[i] != new.table[i]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
638 |
for k in range(len(old.table[i])): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
639 |
if old.table[i][k] != new.table[i][k]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
640 |
value = old.table[i][k] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
641 |
if k == 2: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
642 |
#print "CATEGORY",hex(i), old.table[i][k], new.table[i][k] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
643 |
category_changes[i] = CATEGORY_NAMES.index(value) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
644 |
elif k == 4: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
645 |
#print "BIDIR",hex(i), old.table[i][k], new.table[i][k] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
646 |
bidir_changes[i] = BIDIRECTIONAL_NAMES.index(value) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
647 |
elif k == 5: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
648 |
#print "DECOMP",hex(i), old.table[i][k], new.table[i][k] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
649 |
# We assume that all normalization changes are in 1:1 mappings |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
650 |
assert " " not in value |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
651 |
normalization_changes.append((i, value)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
652 |
elif k == 6: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
653 |
#print "DECIMAL",hex(i), old.table[i][k], new.table[i][k] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
654 |
# we only support changes where the old value is a single digit |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
655 |
assert value in "0123456789" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
656 |
decimal_changes[i] = int(value) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
657 |
elif k == 8: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
658 |
# print "NUMERIC",hex(i), `old.table[i][k]`, new.table[i][k] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
659 |
# Since 0 encodes "no change", the old value is better not 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
660 |
assert value != "0" and value != "-1" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
661 |
if not value: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
662 |
numeric_changes[i] = -1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
663 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
664 |
assert re.match("^[0-9]+$", value) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
665 |
numeric_changes[i] = int(value) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
666 |
elif k == 9: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
667 |
if value == 'Y': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
668 |
mirrored_changes[i] = '1' |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
669 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
670 |
mirrored_changes[i] = '0' |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
671 |
elif k == 11: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
672 |
# change to ISO comment, ignore |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
673 |
pass |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
674 |
elif k == 12: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
675 |
# change to simple uppercase mapping; ignore |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
676 |
pass |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
677 |
elif k == 13: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
678 |
# change to simple lowercase mapping; ignore |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
679 |
pass |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
680 |
elif k == 14: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
681 |
# change to simple titlecase mapping; ignore |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
682 |
pass |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
683 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
684 |
class Difference(Exception):pass |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
685 |
raise Difference, (hex(i), k, old.table[i], new.table[i]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
686 |
new.changed.append((version, zip(bidir_changes, category_changes, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
687 |
decimal_changes, mirrored_changes, |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
688 |
numeric_changes), |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
689 |
normalization_changes)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
690 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
691 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
692 |
# -------------------------------------------------------------------- |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
693 |
# the following support code is taken from the unidb utilities |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
694 |
# Copyright (c) 1999-2000 by Secret Labs AB |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
695 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
696 |
# load a unicode-data file from disk |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
697 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
698 |
import sys |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
699 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
700 |
class UnicodeData: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
701 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
702 |
def __init__(self, filename, exclusions, eastasianwidth, expand=1): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
703 |
self.changed = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
704 |
file = open(filename) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
705 |
table = [None] * 0x110000 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
706 |
while 1: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
707 |
s = file.readline() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
708 |
if not s: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
709 |
break |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
710 |
s = s.strip().split(";") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
711 |
char = int(s[0], 16) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
712 |
table[char] = s |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
713 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
714 |
# expand first-last ranges |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
715 |
if expand: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
716 |
field = None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
717 |
for i in range(0, 0x110000): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
718 |
s = table[i] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
719 |
if s: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
720 |
if s[1][-6:] == "First>": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
721 |
s[1] = "" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
722 |
field = s |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
723 |
elif s[1][-5:] == "Last>": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
724 |
s[1] = "" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
725 |
field = None |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
726 |
elif field: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
727 |
f2 = field[:] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
728 |
f2[0] = "%X" % i |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
729 |
table[i] = f2 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
730 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
731 |
# public attributes |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
732 |
self.filename = filename |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
733 |
self.table = table |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
734 |
self.chars = range(0x110000) # unicode 3.2 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
735 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
736 |
file = open(exclusions) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
737 |
self.exclusions = {} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
738 |
for s in file: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
739 |
s = s.strip() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
740 |
if not s: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
741 |
continue |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
742 |
if s[0] == '#': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
743 |
continue |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
744 |
char = int(s.split()[0],16) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
745 |
self.exclusions[char] = 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
746 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
747 |
widths = [None] * 0x110000 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
748 |
for s in open(eastasianwidth): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
749 |
s = s.strip() |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
750 |
if not s: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
751 |
continue |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
752 |
if s[0] == '#': |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
753 |
continue |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
754 |
s = s.split()[0].split(';') |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
755 |
if '..' in s[0]: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
756 |
first, last = [int(c, 16) for c in s[0].split('..')] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
757 |
chars = range(first, last+1) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
758 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
759 |
chars = [int(s[0], 16)] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
760 |
for char in chars: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
761 |
widths[char] = s[1] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
762 |
for i in range(0, 0x110000): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
763 |
if table[i] is not None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
764 |
table[i].append(widths[i]) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
765 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
766 |
def uselatin1(self): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
767 |
# restrict character range to ISO Latin 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
768 |
self.chars = range(256) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
769 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
770 |
# hash table tools |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
771 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
772 |
# this is a straight-forward reimplementation of Python's built-in |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
773 |
# dictionary type, using a static data structure, and a custom string |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
774 |
# hash algorithm. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
775 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
776 |
def myhash(s, magic): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
777 |
h = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
778 |
for c in map(ord, s.upper()): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
779 |
h = (h * magic) + c |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
780 |
ix = h & 0xff000000L |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
781 |
if ix: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
782 |
h = (h ^ ((ix>>24) & 0xff)) & 0x00ffffff |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
783 |
return h |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
784 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
785 |
SIZES = [ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
786 |
(4,3), (8,3), (16,3), (32,5), (64,3), (128,3), (256,29), (512,17), |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
787 |
(1024,9), (2048,5), (4096,83), (8192,27), (16384,43), (32768,3), |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
788 |
(65536,45), (131072,9), (262144,39), (524288,39), (1048576,9), |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
789 |
(2097152,5), (4194304,3), (8388608,33), (16777216,27) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
790 |
] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
791 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
792 |
class Hash: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
793 |
def __init__(self, name, data, magic): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
794 |
# turn a (key, value) list into a static hash table structure |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
795 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
796 |
# determine table size |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
797 |
for size, poly in SIZES: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
798 |
if size > len(data): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
799 |
poly = size + poly |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
800 |
break |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
801 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
802 |
raise AssertionError, "ran out of polynominals" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
803 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
804 |
print size, "slots in hash table" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
805 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
806 |
table = [None] * size |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
807 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
808 |
mask = size-1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
809 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
810 |
n = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
811 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
812 |
hash = myhash |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
813 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
814 |
# initialize hash table |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
815 |
for key, value in data: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
816 |
h = hash(key, magic) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
817 |
i = (~h) & mask |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
818 |
v = table[i] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
819 |
if v is None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
820 |
table[i] = value |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
821 |
continue |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
822 |
incr = (h ^ (h >> 3)) & mask; |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
823 |
if not incr: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
824 |
incr = mask |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
825 |
while 1: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
826 |
n = n + 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
827 |
i = (i + incr) & mask |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
828 |
v = table[i] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
829 |
if v is None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
830 |
table[i] = value |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
831 |
break |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
832 |
incr = incr << 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
833 |
if incr > mask: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
834 |
incr = incr ^ poly |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
835 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
836 |
print n, "collisions" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
837 |
self.collisions = n |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
838 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
839 |
for i in range(len(table)): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
840 |
if table[i] is None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
841 |
table[i] = 0 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
842 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
843 |
self.data = Array(name + "_hash", table) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
844 |
self.magic = magic |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
845 |
self.name = name |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
846 |
self.size = size |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
847 |
self.poly = poly |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
848 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
849 |
def dump(self, file, trace): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
850 |
# write data to file, as a C array |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
851 |
self.data.dump(file, trace) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
852 |
file.write("#define %s_magic %d\n" % (self.name, self.magic)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
853 |
file.write("#define %s_size %d\n" % (self.name, self.size)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
854 |
file.write("#define %s_poly %d\n" % (self.name, self.poly)) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
855 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
856 |
# stuff to deal with arrays of unsigned integers |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
857 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
858 |
class Array: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
859 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
860 |
def __init__(self, name, data): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
861 |
self.name = name |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
862 |
self.data = data |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
863 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
864 |
def dump(self, file, trace=0): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
865 |
# write data to file, as a C array |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
866 |
size = getsize(self.data) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
867 |
if trace: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
868 |
print >>sys.stderr, self.name+":", size*len(self.data), "bytes" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
869 |
file.write("static ") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
870 |
if size == 1: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
871 |
file.write("unsigned char") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
872 |
elif size == 2: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
873 |
file.write("unsigned short") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
874 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
875 |
file.write("unsigned int") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
876 |
file.write(" " + self.name + "[] = {\n") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
877 |
if self.data: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
878 |
s = " " |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
879 |
for item in self.data: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
880 |
i = str(item) + ", " |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
881 |
if len(s) + len(i) > 78: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
882 |
file.write(s + "\n") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
883 |
s = " " + i |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
884 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
885 |
s = s + i |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
886 |
if s.strip(): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
887 |
file.write(s + "\n") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
888 |
file.write("};\n\n") |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
889 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
890 |
def getsize(data): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
891 |
# return smallest possible integer size for the given array |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
892 |
maxdata = max(data) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
893 |
if maxdata < 256: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
894 |
return 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
895 |
elif maxdata < 65536: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
896 |
return 2 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
897 |
else: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
898 |
return 4 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
899 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
900 |
def splitbins(t, trace=0): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
901 |
"""t, trace=0 -> (t1, t2, shift). Split a table to save space. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
902 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
903 |
t is a sequence of ints. This function can be useful to save space if |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
904 |
many of the ints are the same. t1 and t2 are lists of ints, and shift |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
905 |
is an int, chosen to minimize the combined size of t1 and t2 (in C |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
906 |
code), and where for each i in range(len(t)), |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
907 |
t[i] == t2[(t1[i >> shift] << shift) + (i & mask)] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
908 |
where mask is a bitmask isolating the last "shift" bits. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
909 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
910 |
If optional arg trace is non-zero (default zero), progress info |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
911 |
is printed to sys.stderr. The higher the value, the more info |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
912 |
you'll get. |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
913 |
""" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
914 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
915 |
import sys |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
916 |
if trace: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
917 |
def dump(t1, t2, shift, bytes): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
918 |
print >>sys.stderr, "%d+%d bins at shift %d; %d bytes" % ( |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
919 |
len(t1), len(t2), shift, bytes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
920 |
print >>sys.stderr, "Size of original table:", len(t)*getsize(t), \ |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
921 |
"bytes" |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
922 |
n = len(t)-1 # last valid index |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
923 |
maxshift = 0 # the most we can shift n and still have something left |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
924 |
if n > 0: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
925 |
while n >> 1: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
926 |
n >>= 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
927 |
maxshift += 1 |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
928 |
del n |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
929 |
bytes = sys.maxint # smallest total size so far |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
930 |
t = tuple(t) # so slices can be dict keys |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
931 |
for shift in range(maxshift + 1): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
932 |
t1 = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
933 |
t2 = [] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
934 |
size = 2**shift |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
935 |
bincache = {} |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
936 |
for i in range(0, len(t), size): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
937 |
bin = t[i:i+size] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
938 |
index = bincache.get(bin) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
939 |
if index is None: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
940 |
index = len(t2) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
941 |
bincache[bin] = index |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
942 |
t2.extend(bin) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
943 |
t1.append(index >> shift) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
944 |
# determine memory size |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
945 |
b = len(t1)*getsize(t1) + len(t2)*getsize(t2) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
946 |
if trace > 1: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
947 |
dump(t1, t2, shift, b) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
948 |
if b < bytes: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
949 |
best = t1, t2, shift |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
950 |
bytes = b |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
951 |
t1, t2, shift = best |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
952 |
if trace: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
953 |
print >>sys.stderr, "Best:", |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
954 |
dump(t1, t2, shift, bytes) |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
955 |
if __debug__: |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
956 |
# exhaustively verify that the decomposition is correct |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
957 |
mask = ~((~0) << shift) # i.e., low-bit mask of shift bits |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
958 |
for i in xrange(len(t)): |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
959 |
assert t[i] == t2[(t1[i >> shift] << shift) + (i & mask)] |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
960 |
return best |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
961 |
|
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
962 |
if __name__ == "__main__": |
2fb8b9db1c86
Initial QEMU (symbian-qemu-0.9.1-12) import
martin.trojer@nokia.com
parents:
diff
changeset
|
963 |
maketables(1) |