python-oxlib/oxlib/text.py
2008-07-06 12:21:55 +02:00

244 lines
7.7 KiB
Python
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# -*- coding: utf-8 -*-
# vi:si:et:sw=4:sts=4:ts=4
# GPL written 2008 by j@pad.ma
import re
def findRe(string, regexp):
result = re.compile(regexp, re.DOTALL).findall(string)
if result:
return result[0].strip()
return ''
def findString(string, string0='', string1 = ''):
"""Return the string between string0 and string1.
If string0 or string1 is left out, begining or end of string is used.
>>> findString('i am not there', string1=' not there')
'i am'
>>> findString('i am not there', 'i am ', ' there')
'not'
>>> findString('i am not there', 'i am not t')
'here'
"""
if string0:
string0 = re.escape(string0)
else:
string0 = '^'
if string1:
string1 = re.escape(string1)
else:
string1 = '$'
return findRe(string, string0 + '(.*?)' + string1)
def removeSpecialCharacters(text):
"""
Removes special characters inserted by Word.
"""
text = text.replace(u'\u2013', '-')
text = text.replace(u'\u2026O', "'")
text = text.replace(u'\u2019', "'")
text = text.replace(u'', "'")
text = text.replace(u'', "'")
text = text.replace(u'', "-")
return text
def wrap(text, width):
"""
A word-wrap function that preserves existing line breaks and most spaces in
the text. Expects that existing line breaks are posix newlines (\n).
See http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/148061
"""
return reduce(lambda line, word, width=width: '%s%s%s' %
(line,
' \n'[(len(line[line.rfind('\n')+1:])
+ len(word.split('\n',1)[0]
) >= width)],
word),
text.split(' ')
)
def wrapString(string, length, separator, balance):
if balance:
# balance lines: test if same number of lines
# can be achieved with a shorter line length
lines = wrapString(string, length, separator, False).split(separator)
if len(lines) > 1:
while length > max(map(lambda x : len(x), string.split(' '))):
length -= 1
if len(wrapString(string, length, separator, False).split(separator)) > len(lines):
length += 1
break
words = string.split(' ')
lines = ['']
for word in words:
if len(lines[len(lines) - 1] + word + ' ') <= length + 1:
# word fits in current line
lines[len(lines) - 1] += word + ' ';
else:
if len(word) <= length:
# word fits in next line
lines.append(word + ' ')
else:
# word is longer than line
position = length - len(lines[len(lines) - 1])
lines[len(lines) - 1] += word[0:position]
for i in range(position, len(word), length):
lines.append(word[i:i+length]);
lines[len(lines) - 1] += ' '
return separator.join(lines).strip()
def truncateString(s, num):
"""Truncates a string after a certain number of chacters, but ends with a word
>>> truncateString('Truncates a string after a certain number of chacters, but ends with a word', 23)
'Truncates a string...'
>>> truncateString('Truncates a string', 23)
'Truncates a string'
"""
length = int(num)
if len(s) <= length:
return s
words = s.split()
ts = ""
while words and len(ts) + len(words[0]) < length:
ts += " " + words.pop(0)
if words:
ts += "..."
return ts.strip()
def trimString(string, num):
"""Truncates a string after a certain number of chacters, adding ... at -10 characters
>>> trimString('Truncates a string after a certain number of chacters', 23)
'Truncates ...f chacters'
>>> trimString('Truncates a string', 23)
'Truncates a string'
"""
if len(string) > num:
string = string[:num - 13] + '...' + string[-10:]
return string
def truncateWords(s, num):
"Truncates a string after a certain number of words."
length = int(num)
words = s.split()
if len(words) > length:
words = words[:length]
if not words[-1].endswith('...'):
words.append('...')
return ' '.join(words)
def getValidFilename(s):
"""
Returns the given string converted to a string that can be used for a clean
filename. Specifically, leading and trailing spaces are removed;
all non-filename-safe characters are removed.
>>> getValidFilename("john's portrait in 2004.jpg")
'john_s_portrait_in_2004.jpg'
"""
s = s.strip()
s = s.replace(' ', '_')
s = re.sub(r'[^-A-Za-z0-9_.\[\]\ ]', '_', s)
s = s.replace('__', '_').replace('__', '_')
return s
def getTextList(list_, last_word='or'):
"""
>>> getTextList(['a', 'b', 'c', 'd'])
'a, b, c or d'
>>> getTextList(['a', 'b', 'c'], 'and')
'a, b and c'
>>> getTextList(['a', 'b'], 'and')
'a and b'
>>> getTextList(['a'])
'a'
>>> getTextList([])
''
"""
if len(list_) == 0: return ''
if len(list_) == 1: return list_[0]
return '%s %s %s' % (', '.join([str(i) for i in list_][:-1]), last_word, list_[-1])
def getListText(text, last_word='or'):
"""
>>> getListText('a, b, c or d')
['a', 'b', 'c', 'd']
>>> getListText('a, b and c', 'and')
['a', 'b', 'c']
>>> getListText('a and b', 'and')
['a', 'b']
>>> getListText('a')
['a']
>>> getListText('')
[]
"""
list_ = []
if text:
list_ = text.split(', ')
if list_:
i=len(list_)-1
last = list_[i].split(last_word)
if len(last) == 2:
list_[i] = last[0].strip()
list_.append(last[1].strip())
return list_
def normalizeNewlines(text):
return re.sub(r'\r\n|\r|\n', '\n', text)
def recapitalize(text):
"Recapitalizes text, placing caps after end-of-sentence punctuation."
#capwords = ()
text = text.lower()
capsRE = re.compile(r'(?:^|(?<=[\.\?\!] ))([a-z])')
text = capsRE.sub(lambda x: x.group(1).upper(), text)
#for capword in capwords:
# capwordRE = re.compile(r'\b%s\b' % capword, re.I)
# text = capwordRE.sub(capword, text)
return text
def phone2numeric(phone):
"Converts a phone number with letters into its numeric equivalent."
letters = re.compile(r'[A-PR-Y]', re.I)
char2number = lambda m: {'a': '2', 'c': '2', 'b': '2', 'e': '3',
'd': '3', 'g': '4', 'f': '3', 'i': '4', 'h': '4', 'k': '5',
'j': '5', 'm': '6', 'l': '5', 'o': '6', 'n': '6', 'p': '7',
's': '7', 'r': '7', 'u': '8', 't': '8', 'w': '9', 'v': '8',
'y': '9', 'x': '9'}.get(m.group(0).lower())
return letters.sub(char2number, phone)
def compressString(s):
import cStringIO, gzip
zbuf = cStringIO.StringIO()
zfile = gzip.GzipFile(mode='wb', compresslevel=6, fileobj=zbuf)
zfile.write(s)
zfile.close()
return zbuf.getvalue()
smart_split_re = re.compile('("(?:[^"\\\\]*(?:\\\\.[^"\\\\]*)*)"|\'(?:[^\'\\\\]*(?:\\\\.[^\'\\\\]*)*)\'|[^\\s]+)')
def smartSplit(text):
"""
Generator that splits a string by spaces, leaving quoted phrases together.
Supports both single and double quotes, and supports escaping quotes with
backslashes. In the output, strings will keep their initial and trailing
quote marks.
>>> list(smartSplit('This is "a person\\'s" test.'))
['This', 'is', '"a person\\'s"', 'test.']
"""
for bit in smart_split_re.finditer(text):
bit = bit.group(0)
if bit[0] == '"':
yield '"' + bit[1:-1].replace('\\"', '"').replace('\\\\', '\\') + '"'
elif bit[0] == "'":
yield "'" + bit[1:-1].replace("\\'", "'").replace("\\\\", "\\") + "'"
else:
yield bit