Applying SF patch #949329 on behalf of Raymond Hettinger.

This commit is contained in:
Armin Rigo 2004-05-19 19:10:18 +00:00
parent 16581c8614
commit ba91b9fdda
2 changed files with 34 additions and 34 deletions

View File

@ -5,7 +5,7 @@ There are two kinds of tables: sets, for which a member test is provided,
and mappings, for which a mapping function is provided.
"""
import unicodedata, sets
import unicodedata
assert unicodedata.unidata_version == '3.2.0'
@ -16,7 +16,7 @@ def in_table_a1(code):
return (c & 0xFFFF) not in (0xFFFE, 0xFFFF)
b1_set = sets.Set([173, 847, 6150, 6155, 6156, 6157, 8203, 8204, 8205, 8288, 65279] + range(65024,65040))
b1_set = set([173, 847, 6150, 6155, 6156, 6157, 8203, 8204, 8205, 8288, 65279] + range(65024,65040))
def in_table_b1(code):
return ord(code) in b1_set
@ -217,7 +217,7 @@ def in_table_c11_c12(code):
def in_table_c21(code):
return ord(code) < 128 and unicodedata.category(code) == "Cc"
c22_specials = sets.Set([1757, 1807, 6158, 8204, 8205, 8232, 8233, 65279] + range(8288,8292) + range(8298,8304) + range(65529,65533) + range(119155,119163))
c22_specials = set([1757, 1807, 6158, 8204, 8205, 8232, 8233, 65279] + range(8288,8292) + range(8298,8304) + range(65529,65533) + range(119155,119163))
def in_table_c22(code):
c = ord(code)
if c < 128: return False
@ -244,22 +244,22 @@ def in_table_c5(code):
return unicodedata.category(code) == "Cs"
c6_set = sets.Set(range(65529,65534))
c6_set = set(range(65529,65534))
def in_table_c6(code):
return ord(code) in c6_set
c7_set = sets.Set(range(12272,12284))
c7_set = set(range(12272,12284))
def in_table_c7(code):
return ord(code) in c7_set
c8_set = sets.Set([832, 833, 8206, 8207] + range(8234,8239) + range(8298,8304))
c8_set = set([832, 833, 8206, 8207] + range(8234,8239) + range(8298,8304))
def in_table_c8(code):
return ord(code) in c8_set
c9_set = sets.Set([917505] + range(917536,917632))
c9_set = set([917505] + range(917536,917632))
def in_table_c9(code):
return ord(code) in c9_set
@ -270,3 +270,4 @@ def in_table_d1(code):
def in_table_d2(code):
return unicodedata.bidirectional(code) == "L"

View File

@ -1,5 +1,4 @@
import re, unicodedata, sys, sets
from sets import Set
import re, unicodedata, sys
if sys.maxunicode == 65535:
raise RuntimeError, "need UCS-4 Python"
@ -40,10 +39,10 @@ def compact_set(l):
single.append(prev)
tuple = " + ".join(["range(%d,%d)" % t for t in tuple])
if not single:
return "sets.Set(%s)" % tuple
return "set(%s)" % tuple
if not tuple:
return "sets.Set(%s)" % repr(single)
return "sets.Set(%s + %s)" % (repr(single),tuple)
return "set(%s)" % repr(single)
return "set(%s + %s)" % (repr(single),tuple)
############## Read the tables in the RFC #######################
@ -114,7 +113,7 @@ There are two kinds of tables: sets, for which a member test is provided,
and mappings, for which a mapping function is provided.
\"\"\"
import unicodedata, sets
import unicodedata
"""
print "assert unicodedata.unidata_version == %s" % repr(unicodedata.unidata_version)
@ -124,14 +123,14 @@ print "assert unicodedata.unidata_version == %s" % repr(unicodedata.unidata_vers
name, table = tables[0]
del tables[0]
assert name == "A.1"
table = Set(table.keys())
Cn = Set(gen_category(["Cn"]))
table = set(table.keys())
Cn = set(gen_category(["Cn"]))
# FDD0..FDEF are process internal codes
Cn -= Set(range(0xFDD0, 0xFDF0))
Cn -= set(range(0xFDD0, 0xFDF0))
# not a character
Cn -= Set(range(0xFFFE, 0x110000, 0x10000))
Cn -= Set(range(0xFFFF, 0x110000, 0x10000))
Cn -= set(range(0xFFFE, 0x110000, 0x10000))
Cn -= set(range(0xFFFF, 0x110000, 0x10000))
# assert table == Cn
@ -251,8 +250,8 @@ name, table = tables[0]
del tables[0]
assert name == "C.1.2"
# table = Set(table.keys())
# Zs = Set(gen_category(["Zs"])) - Set([0x20])
# table = set(table.keys())
# Zs = set(gen_category(["Zs"])) - set([0x20])
# assert Zs == table
print """
@ -268,9 +267,9 @@ name, table_c21 = tables[0]
del tables[0]
assert name == "C.2.1"
Cc = Set(gen_category(["Cc"]))
Cc_ascii = Cc & Set(range(128))
table_c21 = Set(table_c21.keys())
Cc = set(gen_category(["Cc"]))
Cc_ascii = Cc & set(range(128))
table_c21 = set(table_c21.keys())
assert Cc_ascii == table_c21
print """
@ -285,7 +284,7 @@ del tables[0]
assert name == "C.2.2"
Cc_nonascii = Cc - Cc_ascii
table_c22 = Set(table_c22.keys())
table_c22 = set(table_c22.keys())
assert len(Cc_nonascii - table_c22) == 0
specials = list(table_c22 - Cc_nonascii)
@ -308,8 +307,8 @@ name, table = tables[0]
del tables[0]
assert name == "C.3"
Co = Set(gen_category(["Co"]))
assert Set(table.keys()) == Co
Co = set(gen_category(["Co"]))
assert set(table.keys()) == Co
print """
def in_table_c3(code):
@ -322,10 +321,10 @@ name, table = tables[0]
del tables[0]
assert name == "C.4"
nonchar = Set(range(0xFDD0,0xFDF0) +
nonchar = set(range(0xFDD0,0xFDF0) +
range(0xFFFE,0x110000,0x10000) +
range(0xFFFF,0x110000,0x10000))
table = Set(table.keys())
table = set(table.keys())
assert table == nonchar
print """
@ -341,8 +340,8 @@ name, table = tables[0]
del tables[0]
assert name == "C.5"
Cs = Set(gen_category(["Cs"]))
assert Set(table.keys()) == Cs
Cs = set(gen_category(["Cs"]))
assert set(table.keys()) == Cs
print """
def in_table_c5(code):
@ -410,8 +409,8 @@ name, table = tables[0]
del tables[0]
assert name == "D.1"
RandAL = Set(gen_bidirectional(["R","AL"]))
assert Set(table.keys()) == RandAL
RandAL = set(gen_bidirectional(["R","AL"]))
assert set(table.keys()) == RandAL
print """
def in_table_d1(code):
@ -423,8 +422,8 @@ name, table = tables[0]
del tables[0]
assert name == "D.2"
L = Set(gen_bidirectional(["L"]))
assert Set(table.keys()) == L
L = set(gen_bidirectional(["L"]))
assert set(table.keys()) == L
print """
def in_table_d2(code):