Let mk_hyb_file.py replace SS in .chr.txt files with . am: fcb08dbdee am: 676a881f8f
am: 3bfee08b16
Change-Id: Ic360ebf8bc82e7ff9fd6d76d84a0257fda9ed136
This commit is contained in:
@@ -35,6 +35,10 @@ import getopt
|
|||||||
|
|
||||||
VERBOSE = False
|
VERBOSE = False
|
||||||
|
|
||||||
|
# U+00DF is LATIN SMALL LETTER SHARP S
|
||||||
|
# U+1E9E is LATIN CAPITAL LETTER SHARP S
|
||||||
|
SHARP_S_TO_DOUBLE = u'\u00dfSS'
|
||||||
|
SHARP_S_TO_CAPITAL = u'\u00df\u1e9e'
|
||||||
|
|
||||||
if sys.version_info[0] >= 3:
|
if sys.version_info[0] >= 3:
|
||||||
def unichr(x):
|
def unichr(x):
|
||||||
@@ -283,8 +287,12 @@ def load_chr(fn):
|
|||||||
for i, l in enumerate(f):
|
for i, l in enumerate(f):
|
||||||
l = l.strip()
|
l = l.strip()
|
||||||
if len(l) > 2:
|
if len(l) > 2:
|
||||||
# lowercase maps to multi-character uppercase sequence, ignore uppercase for now
|
if l == SHARP_S_TO_DOUBLE:
|
||||||
l = l[:1]
|
# replace with lowercasing from capital letter sharp s
|
||||||
|
l = SHARP_S_TO_CAPITAL
|
||||||
|
else:
|
||||||
|
# lowercase maps to multi-character uppercase sequence, ignore uppercase for now
|
||||||
|
l = l[:1]
|
||||||
else:
|
else:
|
||||||
assert len(l) == 2, 'expected 2 chars in chr'
|
assert len(l) == 2, 'expected 2 chars in chr'
|
||||||
for c in l:
|
for c in l:
|
||||||
@@ -419,6 +427,9 @@ def verify_file_sorted(lines, fn):
|
|||||||
file_lines = [l.strip() for l in io.open(fn, encoding='UTF-8')]
|
file_lines = [l.strip() for l in io.open(fn, encoding='UTF-8')]
|
||||||
line_set = set(lines)
|
line_set = set(lines)
|
||||||
file_set = set(file_lines)
|
file_set = set(file_lines)
|
||||||
|
if SHARP_S_TO_DOUBLE in file_set:
|
||||||
|
# ignore difference of double capital letter s and capital letter sharp s
|
||||||
|
file_set.symmetric_difference_update([SHARP_S_TO_DOUBLE, SHARP_S_TO_CAPITAL])
|
||||||
if line_set == file_set:
|
if line_set == file_set:
|
||||||
return True
|
return True
|
||||||
for line in line_set - file_set:
|
for line in line_set - file_set:
|
||||||
|
|||||||
Reference in New Issue
Block a user