def is_sorted__via_lt(iterable, *, key=None, __lt__=None):
it = map_if(key, iterable)
del iterable
if __lt__ is None:
__lt__ = operator.lt
return all(not __lt__(b, a) for a, b in zip_me2(it))
def is_sorted__via_le(iterable, *, key=None, __le__=None):
it = map_if(key, iterable)
del iterable
if __le__ is None:
__le__ = operator.le
return all(starmap(__le__, zip_me2(it)))
def verify_arbitrary_LCP(self, suffices, LCP):
if not suffices:
return not LCP
L = len(suffices)
if not len(LCP) == L - 1: return False
LCP_def = [len_lcp_f(s1, s2) for s1, s2 in zip_me2(suffices)]
return LCP == LCP_def
def make_isuffix2is_L(self, string):
# -> [Bool]
assert string
L = len(string)
isuffix2is_L = [None] * L
# the first succ char == the last char == the last L char
isuffix2is_L[-1] = is_succ_L = True
for prev_isuffix, (succ, prev) in zip(reversed(range(L - 1)),
zip_me2(reversed(string))):
if prev < succ:
# prev is S
r = False
elif prev > succ:
# prev is L
r = True
else:
r = is_succ_L
is_succ_L = isuffix2is_L[prev_isuffix] = r
return isuffix2is_L
def is_strict_sorted(iterable, *, key=None):
#bug: return is_sorted__via_lt(iterable, key=key)
it = map_if(key, iterable)
del iterable
__lt__ = operator.lt
return all(starmap(__lt__, zip_me2(it)))
def make_big_string(self, alphabet_size, string, charL2beginSA,
charS2endSA):
'''return (big_alphabet_size, big_string
, isuffix2is_L :: ISuffix -> Bool
, big_isuffix2LMS_isuffix
#, sorted_LMS_substring_isuffices
, pseudo_sorted_LMS_substring_isuffices
, isuffix2is_LMS
, ichar2big_ichar :: [(None|BigISuffix)]
, to_LMS_substring_suffix
#, to_pseudo_LMS_substring_suffix
)
NOTE:
sorted_LMS_substring_isuffices
* exclude_1size_suffices except string[-1:]
* include leading L suffices
'''
L = len(string)
isuffix2is_L = self.make_isuffix2is_L(string)
isuffix2is_L = isuffix2is_L.__getitem__
# isuffix2is_L done
def isuffix2is_LMS(isuffix):
return (not isuffix2is_L(isuffix)
and (isuffix == 0 or isuffix2is_L(isuffix - 1)))
LMS_isuffices = list(filter(isuffix2is_LMS, range(L - 1)))
# may be empty
big_ichar2LMS_isuffix = big_isuffix2LMS_isuffix = LMS_isuffices
# big_isuffix2LMS_isuffix done
ichar2big_ichar = [None] * L # may be all be None finally
for big_ichar, (begin,
end) in enumerate(zip_me2(chain(LMS_isuffices, [L]))):
for i in range(begin, end):
ichar2big_ichar[i] = big_ichar
else:
big_ichar = begin = end = None
del big_ichar, begin, end
if self.testing:
print(string)
print(ichar2big_ichar)
assert self.verify_ichar2big_ichar(string, isuffix2is_L,
isuffix2is_LMS, ichar2big_ichar)
# NOTE: assume 1size LMS_isuffix followed with '$'
# so, ichar=L-1 isnot more special than LMS_ichars/1size LMS_isuffices
'''bug:
sorted_LMS_substring_isuffices =\
self.make_sorted_LMS_substring_isuffices(
alphabet_size, string, table
, isuffix2is_L, LMS_isuffices
, charL2beginSA, charS2endSA)
# sorted_LMS_substring_isuffices done
if self.testing:
assert self.verify_sorted_LMS_substring_isuffices(
to_LMS_substring_suffix, sorted_LMS_substring_isuffices)
sorted_LMS_whole_substring_isuffices =\
list(filter(isuffix2is_LMS, sorted_LMS_substring_isuffices))
gs = groupby(sorted_LMS_whole_substring_isuffices
, key=to_LMS_substring_suffix)
'''
pseudo_sorted_LMS_substring_isuffices =\
self.make_pseudo_sorted_LMS_substring_isuffices(
alphabet_size, string
, isuffix2is_L, LMS_isuffices
, charL2beginSA, charS2endSA)
def to_pseudo_LMS_substring_suffix(isuffix):
'to_LMS_substring_suffix append m/M'
suffix = to_LMS_substring_suffix(isuffix)
end_isuffix_ex = isuffix + len(suffix)
mM = -1 if end_isuffix_ex == L else alphabet_size
#pseudo_suffix = suffix + [mM] # type(suffix) may be input array type
pseudo_suffix = [*suffix, mM]
return pseudo_suffix
def to_LMS_substring_suffix(isuffix):
'''* include leading Ls
* exclude_1size_suffices except string[-1:]
'''
may_big_ichar = ichar2big_ichar[isuffix]
if may_big_ichar is None:
next_big_ichar = 0
else:
big_ichar = may_big_ichar
next_big_ichar = big_ichar + 1
del big_ichar
if next_big_ichar == len(LMS_isuffices):
return string[isuffix:]
next_LMS_isuffix = LMS_isuffices[next_big_ichar]
return string[isuffix:next_LMS_isuffix + 1]
if self.testing:
try:
assert self.verify_pseudo_sorted_LMS_substring_isuffices(
to_pseudo_LMS_substring_suffix,
pseudo_sorted_LMS_substring_isuffices)
except:
print(string)
print(pseudo_sorted_LMS_substring_isuffices)
print(*map(to_pseudo_LMS_substring_suffix,
pseudo_sorted_LMS_substring_isuffices))
raise
pseudo_sorted_LMS_whole_substring_isuffices =\
list(filter(isuffix2is_LMS, pseudo_sorted_LMS_substring_isuffices))
# O(L)
gs = groupby(pseudo_sorted_LMS_whole_substring_isuffices,
key=to_pseudo_LMS_substring_suffix)
big_string = [None] * len(LMS_isuffices)
big_char = -1
for big_char, (_, g) in enumerate(gs):
for LMS_isuffix in g:
big_ichar = ichar2big_ichar[LMS_isuffix]
big_string[big_ichar] = big_char
assert None not in big_string
big_alphabet_size = big_char + 1
if self.testing:
assert self.verify_big_string(ichar2big_ichar,
to_pseudo_LMS_substring_suffix,
big_ichar2LMS_isuffix,
big_alphabet_size, big_string)
return (
big_alphabet_size,
big_string,
isuffix2is_L,
big_isuffix2LMS_isuffix
#, sorted_LMS_substring_isuffices
,
pseudo_sorted_LMS_substring_isuffices,
isuffix2is_LMS,
ichar2big_ichar,
to_LMS_substring_suffix
#, to_pseudo_LMS_substring_suffix
)