blob: 1c45ab3b865bf8e8cb2b3d3c65438d3709168b4f [file] [log] [blame]
# module 'string' -- A collection of string operations
# XXX Some of these operations are incredibly slow and should be built in
# Some strings for ctype-style character classification
whitespace = ' \t\n'
lowercase = 'abcdefghijklmnopqrstuvwxyz'
uppercase = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
letters = lowercase + uppercase
digits = '0123456789'
hexdigits = digits + 'abcdef' + 'ABCDEF'
octdigits = '01234567'
# Case conversion helpers
_caseswap = {}
for i in range(26):
_caseswap[lowercase[i]] = uppercase[i]
_caseswap[uppercase[i]] = lowercase[i]
del i
# convert UPPER CASE letters to lower case
def lower(s):
res = ''
for c in s:
if 'A' <= c <= 'Z': c = _caseswap[c]
res = res + c
return res
# Convert lower case letters to UPPER CASE
def upper(s):
res = ''
for c in s:
if 'a' <= c <= 'z': c = _caseswap[c]
res = res + c
return res
# Swap lower case letters and UPPER CASE
def swapcase(s):
res = ''
for c in s:
if 'a' <= c <= 'z' or 'A' <= c <= 'Z': c = _caseswap[c]
res = res + c
return res
# Strip leading and trailing tabs and spaces
def strip(s):
i, j = 0, len(s)
while i < j and s[i] in whitespace: i = i+1
while i < j and s[j-1] in whitespace: j = j-1
return s[i:j]
# Split a string into a list of space/tab-separated words
# NB: split(s) is NOT the same as splitfields(s, ' ')!
def split(s):
res = []
i, n = 0, len(s)
while i < n:
while i < n and s[i] in whitespace: i = i+1
if i = n: break
j = i
while j < n and s[j] not in whitespace: j = j+1
res.append(s[i:j])
i = j
return res
# Split a list into fields separated by a given string
# NB: splitfields(s, ' ') is NOT the same as split(s)!
def splitfields(s, sep):
res = []
ns = len(s)
nsep = len(sep)
i = j = 0
while j+nsep <= ns:
if s[j:j+nsep] = sep:
res.append(s[i:j])
i = j = j + nsep
else:
j = j + 1
res.append(s[i:])
return res
# Join words with spaces between them
def join(words):
res = ''
for w in words:
res = res + (' ' + w)
return res[1:]
# Join fields with separator
def joinfields(words, sep):
res = ''
for w in words:
res = res + (sep + w)
return res[len(sep):]
# Find substring
index_error = 'substring not found in string.index'
def index(s, sub):
n = len(sub)
for i in range(len(s) + 1 - n):
if sub = s[i:i+n]: return i
raise index_error, (s, sub)
# Convert string to integer
atoi_error = 'non-numeric argument to string.atoi'
def atoi(str):
s = str
if s[:1] in '+-': s = s[1:]
if not s: raise atoi_error, str
for c in s:
if c not in digits: raise atoi_error, str
return eval(str)
# Left-justify a string
def ljust(s, width):
n = width - len(s)
if n <= 0: return s
return s + ' '*n
# Right-justify a string
def rjust(s, width):
n = width - len(s)
if n <= 0: return s
return ' '*n + s
# Center a string
def center(s, width):
n = width - len(s)
if n <= 0: return s
half = n/2
if n%2 and width%2:
# This ensures that center(center(s, i), j) = center(s, j)
half = half+1
return ' '*half + s + ' '*(n-half)
# 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):
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