mirror of https://github.com/python/cpython
372 lines
10 KiB
Python
372 lines
10 KiB
Python
"""A collection of string operations (most are no longer used in Python 1.6).
|
|
|
|
Warning: most of the code you see here isn't normally used nowadays. With
|
|
Python 1.6, many of these functions are implemented as methods on the
|
|
standard string object. They used to be implemented by a built-in module
|
|
called strop, but strop is now obsolete itself.
|
|
|
|
Public module variables:
|
|
|
|
whitespace -- a string containing all characters considered whitespace
|
|
lowercase -- a string containing all characters considered lowercase letters
|
|
uppercase -- a string containing all characters considered uppercase letters
|
|
letters -- a string containing all characters considered letters
|
|
digits -- a string containing all characters considered decimal digits
|
|
hexdigits -- a string containing all characters considered hexadecimal digits
|
|
octdigits -- a string containing all characters considered octal digits
|
|
|
|
"""
|
|
|
|
# Some strings for ctype-style character classification
|
|
whitespace = ' \t\n\r\v\f'
|
|
lowercase = 'abcdefghijklmnopqrstuvwxyz'
|
|
uppercase = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
|
|
letters = lowercase + uppercase
|
|
digits = '0123456789'
|
|
hexdigits = digits + 'abcdef' + 'ABCDEF'
|
|
octdigits = '01234567'
|
|
|
|
# Case conversion helpers
|
|
_idmap = ''
|
|
for i in range(256): _idmap = _idmap + chr(i)
|
|
del i
|
|
|
|
# Backward compatible names for exceptions
|
|
index_error = ValueError
|
|
atoi_error = ValueError
|
|
atof_error = ValueError
|
|
atol_error = ValueError
|
|
|
|
# convert UPPER CASE letters to lower case
|
|
def lower(s):
|
|
"""lower(s) -> string
|
|
|
|
Return a copy of the string s converted to lowercase.
|
|
|
|
"""
|
|
return s.lower()
|
|
|
|
# Convert lower case letters to UPPER CASE
|
|
def upper(s):
|
|
"""upper(s) -> string
|
|
|
|
Return a copy of the string s converted to uppercase.
|
|
|
|
"""
|
|
return s.upper()
|
|
|
|
# Swap lower case letters and UPPER CASE
|
|
def swapcase(s):
|
|
"""swapcase(s) -> string
|
|
|
|
Return a copy of the string s with upper case characters
|
|
converted to lowercase and vice versa.
|
|
|
|
"""
|
|
return s.swapcase()
|
|
|
|
# Strip leading and trailing tabs and spaces
|
|
def strip(s):
|
|
"""strip(s) -> string
|
|
|
|
Return a copy of the string s with leading and trailing
|
|
whitespace removed.
|
|
|
|
"""
|
|
return s.strip()
|
|
|
|
# Strip leading tabs and spaces
|
|
def lstrip(s):
|
|
"""lstrip(s) -> string
|
|
|
|
Return a copy of the string s with leading whitespace removed.
|
|
|
|
"""
|
|
return s.lstrip()
|
|
|
|
# Strip trailing tabs and spaces
|
|
def rstrip(s):
|
|
"""rstrip(s) -> string
|
|
|
|
Return a copy of the string s with trailing whitespace
|
|
removed.
|
|
|
|
"""
|
|
return s.rstrip()
|
|
|
|
|
|
# Split a string into a list of space/tab-separated words
|
|
# NB: split(s) is NOT the same as splitfields(s, ' ')!
|
|
def split(s, sep=None, maxsplit=-1):
|
|
"""split(s [,sep [,maxsplit]]) -> list of strings
|
|
|
|
Return a list of the words in the string s, using sep as the
|
|
delimiter string. If maxsplit is given, splits into at most
|
|
maxsplit words. If sep is not specified, any whitespace string
|
|
is a separator.
|
|
|
|
(split and splitfields are synonymous)
|
|
|
|
"""
|
|
return s.split(sep, maxsplit)
|
|
splitfields = split
|
|
|
|
# Join fields with optional separator
|
|
def join(words, sep = ' '):
|
|
"""join(list [,sep]) -> string
|
|
|
|
Return a string composed of the words in list, with
|
|
intervening occurences of sep. The default separator is a
|
|
single space.
|
|
|
|
(joinfields and join are synonymous)
|
|
|
|
"""
|
|
return sep.join(words)
|
|
joinfields = join
|
|
|
|
# Find substring, raise exception if not found
|
|
def index(s, *args):
|
|
"""index(s, sub [,start [,end]]) -> int
|
|
|
|
Like find but raises ValueError when the substring is not found.
|
|
|
|
"""
|
|
return s.index(*args)
|
|
|
|
# Find last substring, raise exception if not found
|
|
def rindex(s, *args):
|
|
"""rindex(s, sub [,start [,end]]) -> int
|
|
|
|
Like rfind but raises ValueError when the substring is not found.
|
|
|
|
"""
|
|
return s.rindex(*args)
|
|
|
|
# Count non-overlapping occurrences of substring
|
|
def count(s, *args):
|
|
"""count(s, sub[, start[,end]]) -> int
|
|
|
|
Return the number of occurrences of substring sub in string
|
|
s[start:end]. Optional arguments start and end are
|
|
interpreted as in slice notation.
|
|
|
|
"""
|
|
return s.count(*args)
|
|
|
|
# Find substring, return -1 if not found
|
|
def find(s, *args):
|
|
"""find(s, sub [,start [,end]]) -> in
|
|
|
|
Return the lowest index in s where substring sub is found,
|
|
such that sub is contained within s[start,end]. Optional
|
|
arguments start and end are interpreted as in slice notation.
|
|
|
|
Return -1 on failure.
|
|
|
|
"""
|
|
return s.find(*args)
|
|
|
|
# Find last substring, return -1 if not found
|
|
def rfind(s, *args):
|
|
"""rfind(s, sub [,start [,end]]) -> int
|
|
|
|
Return the highest index in s where substring sub is found,
|
|
such that sub is contained within s[start,end]. Optional
|
|
arguments start and end are interpreted as in slice notation.
|
|
|
|
Return -1 on failure.
|
|
|
|
"""
|
|
return s.rfind(*args)
|
|
|
|
# for a bit of speed
|
|
_float = float
|
|
_int = int
|
|
_long = long
|
|
_StringType = type('')
|
|
|
|
# Convert string to float
|
|
def atof(s):
|
|
"""atof(s) -> float
|
|
|
|
Return the floating point number represented by the string s.
|
|
|
|
"""
|
|
return _float(s)
|
|
|
|
|
|
# Convert string to integer
|
|
def atoi(s , base=10):
|
|
"""atoi(s [,base]) -> int
|
|
|
|
Return the integer represented by the string s in the given
|
|
base, which defaults to 10. The string s must consist of one
|
|
or more digits, possibly preceded by a sign. If base is 0, it
|
|
is chosen from the leading characters of s, 0 for octal, 0x or
|
|
0X for hexadecimal. If base is 16, a preceding 0x or 0X is
|
|
accepted.
|
|
|
|
"""
|
|
return _int(s, base)
|
|
|
|
|
|
# Convert string to long integer
|
|
def atol(s, base=10):
|
|
"""atol(s [,base]) -> long
|
|
|
|
Return the long integer represented by the string s in the
|
|
given base, which defaults to 10. The string s must consist
|
|
of one or more digits, possibly preceded by a sign. If base
|
|
is 0, it is chosen from the leading characters of s, 0 for
|
|
octal, 0x or 0X for hexadecimal. If base is 16, a preceding
|
|
0x or 0X is accepted. A trailing L or l is not accepted,
|
|
unless base is 0.
|
|
|
|
"""
|
|
return _long(s, base)
|
|
|
|
|
|
# Left-justify a string
|
|
def ljust(s, width):
|
|
"""ljust(s, width) -> string
|
|
|
|
Return a left-justified version of s, in a field of the
|
|
specified width, padded with spaces as needed. The string is
|
|
never truncated.
|
|
|
|
"""
|
|
return s.ljust(width)
|
|
|
|
# Right-justify a string
|
|
def rjust(s, width):
|
|
"""rjust(s, width) -> string
|
|
|
|
Return a right-justified version of s, in a field of the
|
|
specified width, padded with spaces as needed. The string is
|
|
never truncated.
|
|
|
|
"""
|
|
return s.rjust(width)
|
|
|
|
# Center a string
|
|
def center(s, width):
|
|
"""center(s, width) -> string
|
|
|
|
Return a center version of s, in a field of the specified
|
|
width. padded with spaces as needed. The string is never
|
|
truncated.
|
|
|
|
"""
|
|
return s.center(width)
|
|
|
|
# Zero-fill a number, e.g., (12, 3) --> '012' and (-3, 3) --> '-03'
|
|
# Decadent feature: the argument may be a string or a number
|
|
# (Use of this is deprecated; it should be a string as with ljust c.s.)
|
|
def zfill(x, width):
|
|
"""zfill(x, width) -> string
|
|
|
|
Pad a numeric string x with zeros on the left, to fill a field
|
|
of the specified width. The string x is never truncated.
|
|
|
|
"""
|
|
if type(x) == type(''): s = x
|
|
else: s = `x`
|
|
n = len(s)
|
|
if n >= width: return s
|
|
sign = ''
|
|
if s[0] in ('-', '+'):
|
|
sign, s = s[0], s[1:]
|
|
return sign + '0'*(width-n) + s
|
|
|
|
# Expand tabs in a string.
|
|
# Doesn't take non-printing chars into account, but does understand \n.
|
|
def expandtabs(s, tabsize=8):
|
|
"""expandtabs(s [,tabsize]) -> string
|
|
|
|
Return a copy of the string s with all tab characters replaced
|
|
by the appropriate number of spaces, depending on the current
|
|
column, and the tabsize (default 8).
|
|
|
|
"""
|
|
return s.expandtabs(tabsize)
|
|
|
|
# Character translation through look-up table.
|
|
def translate(s, table, deletions=""):
|
|
"""translate(s,table [,deletechars]) -> string
|
|
|
|
Return a copy of the string s, where all characters occurring
|
|
in the optional argument deletechars are removed, and the
|
|
remaining characters have been mapped through the given
|
|
translation table, which must be a string of length 256.
|
|
|
|
"""
|
|
return s.translate(table, deletions)
|
|
|
|
# Capitalize a string, e.g. "aBc dEf" -> "Abc def".
|
|
def capitalize(s):
|
|
"""capitalize(s) -> string
|
|
|
|
Return a copy of the string s with only its first character
|
|
capitalized.
|
|
|
|
"""
|
|
return s.capitalize()
|
|
|
|
# Capitalize the words in a string, e.g. " aBc dEf " -> "Abc Def".
|
|
# See also regsub.capwords().
|
|
def capwords(s, sep=None):
|
|
"""capwords(s, [sep]) -> string
|
|
|
|
Split the argument into words using split, capitalize each
|
|
word using capitalize, and join the capitalized words using
|
|
join. Note that this replaces runs of whitespace characters by
|
|
a single space.
|
|
|
|
"""
|
|
return join(map(capitalize, s.split(sep)), sep or ' ')
|
|
|
|
# Construct a translation string
|
|
_idmapL = None
|
|
def maketrans(fromstr, tostr):
|
|
"""maketrans(frm, to) -> string
|
|
|
|
Return a translation table (a string of 256 bytes long)
|
|
suitable for use in string.translate. The strings frm and to
|
|
must be of the same length.
|
|
|
|
"""
|
|
if len(fromstr) != len(tostr):
|
|
raise ValueError, "maketrans arguments must have same length"
|
|
global _idmapL
|
|
if not _idmapL:
|
|
_idmapL = map(None, _idmap)
|
|
L = _idmapL[:]
|
|
fromstr = map(ord, fromstr)
|
|
for i in range(len(fromstr)):
|
|
L[fromstr[i]] = tostr[i]
|
|
return joinfields(L, "")
|
|
|
|
# Substring replacement (global)
|
|
def replace(s, old, new, maxsplit=-1):
|
|
"""replace (str, old, new[, maxsplit]) -> string
|
|
|
|
Return a copy of string str with all occurrences of substring
|
|
old replaced by new. If the optional argument maxsplit is
|
|
given, only the first maxsplit occurrences are replaced.
|
|
|
|
"""
|
|
return s.replace(old, new, maxsplit)
|
|
|
|
|
|
# Try importing optional built-in module "strop" -- if it exists,
|
|
# it redefines some string operations that are 100-1000 times faster.
|
|
# It also defines values for whitespace, lowercase and uppercase
|
|
# that match <ctype.h>'s definitions.
|
|
|
|
try:
|
|
from strop import maketrans, lowercase, uppercase, whitespace
|
|
letters = lowercase + uppercase
|
|
except ImportError:
|
|
pass # Use the original versions
|