Mercurial > repos > shellac > guppy_basecaller
annotate env/lib/python3.7/site-packages/boltons/strutils.py @ 2:6af9afd405e9 draft
"planemo upload commit 0a63dd5f4d38a1f6944587f52a8cd79874177fc1"
author | shellac |
---|---|
date | Thu, 14 May 2020 14:56:58 -0400 |
parents | 26e78fe6e8c4 |
children |
rev | line source |
---|---|
0
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1 # -*- coding: utf-8 -*- |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
2 """So much practical programming involves string manipulation, which |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
3 Python readily accommodates. Still, there are dozens of basic and |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
4 common capabilities missing from the standard library, several of them |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
5 provided by ``strutils``. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
6 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
7 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
8 from __future__ import print_function |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
9 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
10 import re |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
11 import sys |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
12 import uuid |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
13 import zlib |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
14 import string |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
15 import unicodedata |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
16 import collections |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
17 from gzip import GzipFile |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
18 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
19 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
20 from cStringIO import cStringIO as StringIO |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
21 except ImportError: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
22 from io import BytesIO as StringIO |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
23 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
24 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
25 from collections.abc import Mapping |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
26 except ImportError: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
27 from collections import Mapping |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
28 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
29 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
30 unicode, str, bytes, basestring = unicode, str, str, basestring |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
31 from HTMLParser import HTMLParser |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
32 import htmlentitydefs |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
33 except NameError: # basestring not defined in Python 3 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
34 unicode, str, bytes, basestring = str, bytes, bytes, (str, bytes) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
35 unichr = chr |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
36 from html.parser import HTMLParser |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
37 from html import entities as htmlentitydefs |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
38 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
39 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
40 __all__ = ['camel2under', 'under2camel', 'slugify', 'split_punct_ws', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
41 'unit_len', 'ordinalize', 'cardinalize', 'pluralize', 'singularize', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
42 'asciify', 'is_ascii', 'is_uuid', 'html2text', 'strip_ansi', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
43 'bytes2human', 'find_hashtags', 'a10n', 'gzip_bytes', 'gunzip_bytes', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
44 'iter_splitlines', 'indent', 'escape_shell_args', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
45 'args2cmd', 'args2sh', 'parse_int_list', 'format_int_list', 'unwrap_text'] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
46 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
47 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
48 _punct_ws_str = string.punctuation + string.whitespace |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
49 _punct_re = re.compile('[' + _punct_ws_str + ']+') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
50 _camel2under_re = re.compile('((?<=[a-z0-9])[A-Z]|(?!^)[A-Z](?=[a-z]))') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
51 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
52 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
53 def camel2under(camel_string): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
54 """Converts a camelcased string to underscores. Useful for turning a |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
55 class name into a function name. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
56 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
57 >>> camel2under('BasicParseTest') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
58 'basic_parse_test' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
59 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
60 return _camel2under_re.sub(r'_\1', camel_string).lower() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
61 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
62 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
63 def under2camel(under_string): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
64 """Converts an underscored string to camelcased. Useful for turning a |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
65 function name into a class name. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
66 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
67 >>> under2camel('complex_tokenizer') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
68 'ComplexTokenizer' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
69 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
70 return ''.join(w.capitalize() or '_' for w in under_string.split('_')) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
71 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
72 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
73 def slugify(text, delim='_', lower=True, ascii=False): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
74 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
75 A basic function that turns text full of scary characters |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
76 (i.e., punctuation and whitespace), into a relatively safe |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
77 lowercased string separated only by the delimiter specified |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
78 by *delim*, which defaults to ``_``. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
79 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
80 The *ascii* convenience flag will :func:`asciify` the slug if |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
81 you require ascii-only slugs. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
82 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
83 >>> slugify('First post! Hi!!!!~1 ') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
84 'first_post_hi_1' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
85 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
86 >>> slugify("Kurt Gödel's pretty cool.", ascii=True) == \ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
87 b'kurt_goedel_s_pretty_cool' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
88 True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
89 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
90 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
91 ret = delim.join(split_punct_ws(text)) or delim if text else '' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
92 if ascii: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
93 ret = asciify(ret) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
94 if lower: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
95 ret = ret.lower() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
96 return ret |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
97 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
98 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
99 def split_punct_ws(text): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
100 """While :meth:`str.split` will split on whitespace, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
101 :func:`split_punct_ws` will split on punctuation and |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
102 whitespace. This used internally by :func:`slugify`, above. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
103 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
104 >>> split_punct_ws('First post! Hi!!!!~1 ') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
105 ['First', 'post', 'Hi', '1'] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
106 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
107 return [w for w in _punct_re.split(text) if w] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
108 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
109 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
110 def unit_len(sized_iterable, unit_noun='item'): # TODO: len_units()/unitize()? |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
111 """Returns a plain-English description of an iterable's |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
112 :func:`len()`, conditionally pluralized with :func:`cardinalize`, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
113 detailed below. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
114 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
115 >>> print(unit_len(range(10), 'number')) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
116 10 numbers |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
117 >>> print(unit_len('aeiou', 'vowel')) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
118 5 vowels |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
119 >>> print(unit_len([], 'worry')) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
120 No worries |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
121 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
122 count = len(sized_iterable) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
123 units = cardinalize(unit_noun, count) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
124 if count: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
125 return u'%s %s' % (count, units) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
126 return u'No %s' % (units,) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
127 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
128 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
129 _ORDINAL_MAP = {'1': 'st', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
130 '2': 'nd', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
131 '3': 'rd'} # 'th' is the default |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
132 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
133 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
134 def ordinalize(number, ext_only=False): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
135 """Turns *number* into its cardinal form, i.e., 1st, 2nd, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
136 3rd, 4th, etc. If the last character isn't a digit, it returns the |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
137 string value unchanged. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
138 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
139 Args: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
140 number (int or str): Number to be cardinalized. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
141 ext_only (bool): Whether to return only the suffix. Default ``False``. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
142 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
143 >>> print(ordinalize(1)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
144 1st |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
145 >>> print(ordinalize(3694839230)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
146 3694839230th |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
147 >>> print(ordinalize('hi')) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
148 hi |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
149 >>> print(ordinalize(1515)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
150 1515th |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
151 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
152 numstr, ext = unicode(number), '' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
153 if numstr and numstr[-1] in string.digits: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
154 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
155 # first check for teens |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
156 if numstr[-2] == '1': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
157 ext = 'th' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
158 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
159 # all other cases |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
160 ext = _ORDINAL_MAP.get(numstr[-1], 'th') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
161 except IndexError: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
162 # single digit numbers (will reach here based on [-2] above) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
163 ext = _ORDINAL_MAP.get(numstr[-1], 'th') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
164 if ext_only: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
165 return ext |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
166 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
167 return numstr + ext |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
168 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
169 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
170 def cardinalize(unit_noun, count): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
171 """Conditionally pluralizes a singular word *unit_noun* if |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
172 *count* is not one, preserving case when possible. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
173 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
174 >>> vowels = 'aeiou' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
175 >>> print(len(vowels), cardinalize('vowel', len(vowels))) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
176 5 vowels |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
177 >>> print(3, cardinalize('Wish', 3)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
178 3 Wishes |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
179 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
180 if count == 1: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
181 return unit_noun |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
182 return pluralize(unit_noun) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
183 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
184 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
185 def singularize(word): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
186 """Semi-intelligently converts an English plural *word* to its |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
187 singular form, preserving case pattern. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
188 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
189 >>> singularize('chances') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
190 'chance' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
191 >>> singularize('Activities') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
192 'Activity' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
193 >>> singularize('Glasses') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
194 'Glass' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
195 >>> singularize('FEET') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
196 'FOOT' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
197 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
198 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
199 orig_word, word = word, word.strip().lower() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
200 if not word or word in _IRR_S2P: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
201 return orig_word |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
202 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
203 irr_singular = _IRR_P2S.get(word) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
204 if irr_singular: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
205 singular = irr_singular |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
206 elif not word.endswith('s'): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
207 return orig_word |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
208 elif len(word) == 2: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
209 singular = word[:-1] # or just return word? |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
210 elif word.endswith('ies') and word[-4:-3] not in 'aeiou': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
211 singular = word[:-3] + 'y' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
212 elif word.endswith('es') and word[-3] == 's': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
213 singular = word[:-2] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
214 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
215 singular = word[:-1] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
216 return _match_case(orig_word, singular) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
217 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
218 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
219 def pluralize(word): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
220 """Semi-intelligently converts an English *word* from singular form to |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
221 plural, preserving case pattern. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
222 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
223 >>> pluralize('friend') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
224 'friends' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
225 >>> pluralize('enemy') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
226 'enemies' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
227 >>> pluralize('Sheep') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
228 'Sheep' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
229 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
230 orig_word, word = word, word.strip().lower() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
231 if not word or word in _IRR_P2S: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
232 return orig_word |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
233 irr_plural = _IRR_S2P.get(word) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
234 if irr_plural: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
235 plural = irr_plural |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
236 elif word.endswith('y') and word[-2:-1] not in 'aeiou': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
237 plural = word[:-1] + 'ies' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
238 elif word[-1] == 's' or word.endswith('ch') or word.endswith('sh'): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
239 plural = word if word.endswith('es') else word + 'es' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
240 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
241 plural = word + 's' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
242 return _match_case(orig_word, plural) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
243 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
244 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
245 def _match_case(master, disciple): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
246 if not master.strip(): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
247 return disciple |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
248 if master.lower() == master: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
249 return disciple.lower() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
250 elif master.upper() == master: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
251 return disciple.upper() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
252 elif master.title() == master: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
253 return disciple.title() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
254 return disciple |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
255 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
256 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
257 # Singular to plural map of irregular pluralizations |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
258 _IRR_S2P = {'addendum': 'addenda', 'alga': 'algae', 'alumna': 'alumnae', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
259 'alumnus': 'alumni', 'analysis': 'analyses', 'antenna': 'antennae', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
260 'appendix': 'appendices', 'axis': 'axes', 'bacillus': 'bacilli', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
261 'bacterium': 'bacteria', 'basis': 'bases', 'beau': 'beaux', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
262 'bison': 'bison', 'bureau': 'bureaus', 'cactus': 'cacti', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
263 'calf': 'calves', 'child': 'children', 'corps': 'corps', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
264 'corpus': 'corpora', 'crisis': 'crises', 'criterion': 'criteria', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
265 'curriculum': 'curricula', 'datum': 'data', 'deer': 'deer', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
266 'diagnosis': 'diagnoses', 'die': 'dice', 'dwarf': 'dwarves', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
267 'echo': 'echoes', 'elf': 'elves', 'ellipsis': 'ellipses', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
268 'embargo': 'embargoes', 'emphasis': 'emphases', 'erratum': 'errata', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
269 'fireman': 'firemen', 'fish': 'fish', 'focus': 'foci', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
270 'foot': 'feet', 'formula': 'formulae', 'formula': 'formulas', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
271 'fungus': 'fungi', 'genus': 'genera', 'goose': 'geese', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
272 'half': 'halves', 'hero': 'heroes', 'hippopotamus': 'hippopotami', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
273 'hoof': 'hooves', 'hypothesis': 'hypotheses', 'index': 'indices', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
274 'knife': 'knives', 'leaf': 'leaves', 'life': 'lives', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
275 'loaf': 'loaves', 'louse': 'lice', 'man': 'men', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
276 'matrix': 'matrices', 'means': 'means', 'medium': 'media', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
277 'memorandum': 'memoranda', 'millennium': 'milennia', 'moose': 'moose', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
278 'mosquito': 'mosquitoes', 'mouse': 'mice', 'nebula': 'nebulae', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
279 'neurosis': 'neuroses', 'nucleus': 'nuclei', 'oasis': 'oases', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
280 'octopus': 'octopi', 'offspring': 'offspring', 'ovum': 'ova', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
281 'ox': 'oxen', 'paralysis': 'paralyses', 'parenthesis': 'parentheses', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
282 'person': 'people', 'phenomenon': 'phenomena', 'potato': 'potatoes', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
283 'radius': 'radii', 'scarf': 'scarves', 'scissors': 'scissors', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
284 'self': 'selves', 'sense': 'senses', 'series': 'series', 'sheep': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
285 'sheep', 'shelf': 'shelves', 'species': 'species', 'stimulus': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
286 'stimuli', 'stratum': 'strata', 'syllabus': 'syllabi', 'symposium': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
287 'symposia', 'synopsis': 'synopses', 'synthesis': 'syntheses', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
288 'tableau': 'tableaux', 'that': 'those', 'thesis': 'theses', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
289 'thief': 'thieves', 'this': 'these', 'tomato': 'tomatoes', 'tooth': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
290 'teeth', 'torpedo': 'torpedoes', 'vertebra': 'vertebrae', 'veto': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
291 'vetoes', 'vita': 'vitae', 'watch': 'watches', 'wife': 'wives', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
292 'wolf': 'wolves', 'woman': 'women'} |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
293 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
294 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
295 # Reverse index of the above |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
296 _IRR_P2S = dict([(v, k) for k, v in _IRR_S2P.items()]) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
297 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
298 HASHTAG_RE = re.compile(r"(?:^|\s)[##]{1}(\w+)", re.UNICODE) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
299 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
300 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
301 def find_hashtags(string): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
302 """Finds and returns all hashtags in a string, with the hashmark |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
303 removed. Supports full-width hashmarks for Asian languages and |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
304 does not false-positive on URL anchors. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
305 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
306 >>> find_hashtags('#atag http://asite/#ananchor') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
307 ['atag'] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
308 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
309 ``find_hashtags`` also works with unicode hashtags. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
310 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
311 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
312 # the following works, doctest just struggles with it |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
313 # >>> find_hashtags(u"can't get enough of that dignity chicken #肯德基 woo") |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
314 # [u'\u80af\u5fb7\u57fa'] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
315 return HASHTAG_RE.findall(string) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
316 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
317 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
318 def a10n(string): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
319 """That thing where "internationalization" becomes "i18n", what's it |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
320 called? Abbreviation? Oh wait, no: ``a10n``. (It's actually a form |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
321 of `numeronym`_.) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
322 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
323 >>> a10n('abbreviation') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
324 'a10n' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
325 >>> a10n('internationalization') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
326 'i18n' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
327 >>> a10n('') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
328 '' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
329 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
330 .. _numeronym: http://en.wikipedia.org/wiki/Numeronym |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
331 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
332 if len(string) < 3: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
333 return string |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
334 return '%s%s%s' % (string[0], len(string[1:-1]), string[-1]) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
335 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
336 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
337 ANSI_ESCAPE_BEGIN = '\x1b[' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
338 ANSI_TERMINATORS = ('H', 'f', 'A', 'B', 'C', 'D', 'R', 's', 'u', 'J', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
339 'K', 'h', 'l', 'p', 'm') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
340 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
341 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
342 def strip_ansi(text): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
343 """Strips ANSI escape codes from *text*. Useful for the occasional |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
344 time when a log or redirected output accidentally captures console |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
345 color codes and the like. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
346 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
347 >>> strip_ansi('\x1b[0m\x1b[1;36mart\x1b[46;34m\xdc') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
348 'art' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
349 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
350 The test above is an excerpt from ANSI art on |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
351 `sixteencolors.net`_. This function does not interpret or render |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
352 ANSI art, but you can do so with `ansi2img`_ or `escapes.js`_. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
353 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
354 .. _sixteencolors.net: http://sixteencolors.net |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
355 .. _ansi2img: http://www.bedroomlan.org/projects/ansi2img |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
356 .. _escapes.js: https://github.com/atdt/escapes.js |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
357 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
358 # TODO: move to cliutils.py |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
359 nansi, keep, i, text_len = [], True, 0, len(text) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
360 while i < text_len: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
361 if not keep and text[i] in ANSI_TERMINATORS: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
362 keep = True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
363 elif keep: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
364 keep_end_i = text.find(ANSI_ESCAPE_BEGIN, i) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
365 if keep_end_i < 0: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
366 break |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
367 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
368 nansi.append(text[i:keep_end_i]) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
369 i, keep = keep_end_i, False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
370 i += 1 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
371 if not nansi: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
372 return text |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
373 return type(text)().join(nansi) # attempted unicode + str support |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
374 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
375 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
376 def asciify(text, ignore=False): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
377 """Converts a unicode or bytestring, *text*, into a bytestring with |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
378 just ascii characters. Performs basic deaccenting for all you |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
379 Europhiles out there. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
380 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
381 Also, a gentle reminder that this is a **utility**, primarily meant |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
382 for slugification. Whenever possible, make your application work |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
383 **with** unicode, not against it. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
384 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
385 Args: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
386 text (str or unicode): The string to be asciified. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
387 ignore (bool): Configures final encoding to ignore remaining |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
388 unasciified unicode instead of replacing it. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
389 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
390 >>> asciify('Beyoncé') == b'Beyonce' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
391 True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
392 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
393 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
394 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
395 return text.encode('ascii') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
396 except UnicodeDecodeError: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
397 # this usually means you passed in a non-unicode string |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
398 text = text.decode('utf-8') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
399 return text.encode('ascii') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
400 except UnicodeEncodeError: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
401 mode = 'replace' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
402 if ignore: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
403 mode = 'ignore' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
404 transd = unicodedata.normalize('NFKD', text.translate(DEACCENT_MAP)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
405 ret = transd.encode('ascii', mode) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
406 return ret |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
407 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
408 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
409 def is_ascii(text): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
410 """Check if a unicode or bytestring, *text*, is composed of ascii |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
411 characters only. Raises :exc:`ValueError` if argument is not text. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
412 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
413 Args: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
414 text (str or unicode): The string to be checked. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
415 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
416 >>> is_ascii('Beyoncé') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
417 False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
418 >>> is_ascii('Beyonce') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
419 True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
420 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
421 if isinstance(text, unicode): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
422 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
423 text.encode('ascii') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
424 except UnicodeEncodeError: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
425 return False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
426 elif isinstance(text, bytes): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
427 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
428 text.decode('ascii') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
429 except UnicodeDecodeError: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
430 return False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
431 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
432 raise ValueError('expected text or bytes, not %r' % type(text)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
433 return True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
434 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
435 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
436 class DeaccenterDict(dict): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
437 "A small caching dictionary for deaccenting." |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
438 def __missing__(self, key): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
439 ch = self.get(key) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
440 if ch is not None: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
441 return ch |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
442 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
443 de = unicodedata.decomposition(unichr(key)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
444 p1, _, p2 = de.rpartition(' ') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
445 if int(p2, 16) == 0x308: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
446 ch = self.get(key) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
447 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
448 ch = int(p1, 16) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
449 except (IndexError, ValueError): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
450 ch = self.get(key, key) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
451 self[key] = ch |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
452 return ch |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
453 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
454 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
455 from collections import defaultdict |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
456 except ImportError: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
457 # no defaultdict means that __missing__ isn't supported in |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
458 # this version of python, so we define __getitem__ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
459 def __getitem__(self, key): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
460 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
461 return super(DeaccenterDict, self).__getitem__(key) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
462 except KeyError: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
463 return self.__missing__(key) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
464 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
465 del defaultdict |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
466 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
467 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
468 # http://chmullig.com/2009/12/python-unicode-ascii-ifier/ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
469 # For something more complete, investigate the unidecode |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
470 # or isounidecode packages, which are capable of performing |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
471 # crude transliteration. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
472 _BASE_DEACCENT_MAP = { |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
473 0xc6: u"AE", # Æ LATIN CAPITAL LETTER AE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
474 0xd0: u"D", # Ð LATIN CAPITAL LETTER ETH |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
475 0xd8: u"OE", # Ø LATIN CAPITAL LETTER O WITH STROKE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
476 0xde: u"Th", # Þ LATIN CAPITAL LETTER THORN |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
477 0xc4: u'Ae', # Ä LATIN CAPITAL LETTER A WITH DIAERESIS |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
478 0xd6: u'Oe', # Ö LATIN CAPITAL LETTER O WITH DIAERESIS |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
479 0xdc: u'Ue', # Ü LATIN CAPITAL LETTER U WITH DIAERESIS |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
480 0xc0: u"A", # À LATIN CAPITAL LETTER A WITH GRAVE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
481 0xc1: u"A", # Á LATIN CAPITAL LETTER A WITH ACUTE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
482 0xc3: u"A", # Ã LATIN CAPITAL LETTER A WITH TILDE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
483 0xc7: u"C", # Ç LATIN CAPITAL LETTER C WITH CEDILLA |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
484 0xc8: u"E", # È LATIN CAPITAL LETTER E WITH GRAVE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
485 0xc9: u"E", # É LATIN CAPITAL LETTER E WITH ACUTE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
486 0xca: u"E", # Ê LATIN CAPITAL LETTER E WITH CIRCUMFLEX |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
487 0xcc: u"I", # Ì LATIN CAPITAL LETTER I WITH GRAVE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
488 0xcd: u"I", # Í LATIN CAPITAL LETTER I WITH ACUTE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
489 0xd2: u"O", # Ò LATIN CAPITAL LETTER O WITH GRAVE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
490 0xd3: u"O", # Ó LATIN CAPITAL LETTER O WITH ACUTE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
491 0xd5: u"O", # Õ LATIN CAPITAL LETTER O WITH TILDE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
492 0xd9: u"U", # Ù LATIN CAPITAL LETTER U WITH GRAVE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
493 0xda: u"U", # Ú LATIN CAPITAL LETTER U WITH ACUTE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
494 0xdf: u"ss", # ß LATIN SMALL LETTER SHARP S |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
495 0xe6: u"ae", # æ LATIN SMALL LETTER AE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
496 0xf0: u"d", # ð LATIN SMALL LETTER ETH |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
497 0xf8: u"oe", # ø LATIN SMALL LETTER O WITH STROKE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
498 0xfe: u"th", # þ LATIN SMALL LETTER THORN, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
499 0xe4: u'ae', # ä LATIN SMALL LETTER A WITH DIAERESIS |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
500 0xf6: u'oe', # ö LATIN SMALL LETTER O WITH DIAERESIS |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
501 0xfc: u'ue', # ü LATIN SMALL LETTER U WITH DIAERESIS |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
502 0xe0: u"a", # à LATIN SMALL LETTER A WITH GRAVE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
503 0xe1: u"a", # á LATIN SMALL LETTER A WITH ACUTE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
504 0xe3: u"a", # ã LATIN SMALL LETTER A WITH TILDE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
505 0xe7: u"c", # ç LATIN SMALL LETTER C WITH CEDILLA |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
506 0xe8: u"e", # è LATIN SMALL LETTER E WITH GRAVE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
507 0xe9: u"e", # é LATIN SMALL LETTER E WITH ACUTE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
508 0xea: u"e", # ê LATIN SMALL LETTER E WITH CIRCUMFLEX |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
509 0xec: u"i", # ì LATIN SMALL LETTER I WITH GRAVE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
510 0xed: u"i", # í LATIN SMALL LETTER I WITH ACUTE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
511 0xf2: u"o", # ò LATIN SMALL LETTER O WITH GRAVE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
512 0xf3: u"o", # ó LATIN SMALL LETTER O WITH ACUTE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
513 0xf5: u"o", # õ LATIN SMALL LETTER O WITH TILDE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
514 0xf9: u"u", # ù LATIN SMALL LETTER U WITH GRAVE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
515 0xfa: u"u", # ú LATIN SMALL LETTER U WITH ACUTE |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
516 0x2018: u"'", # ‘ LEFT SINGLE QUOTATION MARK |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
517 0x2019: u"'", # ’ RIGHT SINGLE QUOTATION MARK |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
518 0x201c: u'"', # “ LEFT DOUBLE QUOTATION MARK |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
519 0x201d: u'"', # ” RIGHT DOUBLE QUOTATION MARK |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
520 } |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
521 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
522 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
523 DEACCENT_MAP = DeaccenterDict(_BASE_DEACCENT_MAP) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
524 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
525 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
526 _SIZE_SYMBOLS = ('B', 'K', 'M', 'G', 'T', 'P', 'E', 'Z', 'Y') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
527 _SIZE_BOUNDS = [(1024 ** i, sym) for i, sym in enumerate(_SIZE_SYMBOLS)] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
528 _SIZE_RANGES = list(zip(_SIZE_BOUNDS, _SIZE_BOUNDS[1:])) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
529 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
530 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
531 def bytes2human(nbytes, ndigits=0): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
532 """Turns an integer value of *nbytes* into a human readable format. Set |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
533 *ndigits* to control how many digits after the decimal point |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
534 should be shown (default ``0``). |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
535 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
536 >>> bytes2human(128991) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
537 '126K' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
538 >>> bytes2human(100001221) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
539 '95M' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
540 >>> bytes2human(0, 2) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
541 '0.00B' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
542 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
543 abs_bytes = abs(nbytes) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
544 for (size, symbol), (next_size, next_symbol) in _SIZE_RANGES: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
545 if abs_bytes <= next_size: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
546 break |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
547 hnbytes = float(nbytes) / size |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
548 return '{hnbytes:.{ndigits}f}{symbol}'.format(hnbytes=hnbytes, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
549 ndigits=ndigits, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
550 symbol=symbol) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
551 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
552 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
553 class HTMLTextExtractor(HTMLParser): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
554 def __init__(self): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
555 self.reset() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
556 self.strict = False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
557 self.convert_charrefs = True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
558 self.result = [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
559 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
560 def handle_data(self, d): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
561 self.result.append(d) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
562 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
563 def handle_charref(self, number): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
564 if number[0] == u'x' or number[0] == u'X': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
565 codepoint = int(number[1:], 16) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
566 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
567 codepoint = int(number) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
568 self.result.append(unichr(codepoint)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
569 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
570 def handle_entityref(self, name): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
571 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
572 codepoint = htmlentitydefs.name2codepoint[name] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
573 except KeyError: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
574 self.result.append(u'&' + name + u';') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
575 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
576 self.result.append(unichr(codepoint)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
577 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
578 def get_text(self): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
579 return u''.join(self.result) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
580 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
581 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
582 def html2text(html): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
583 """Strips tags from HTML text, returning markup-free text. Also, does |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
584 a best effort replacement of entities like " " |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
585 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
586 >>> r = html2text(u'<a href="#">Test &<em>(\u0394ημώ)</em></a>') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
587 >>> r == u'Test &(\u0394\u03b7\u03bc\u03ce)' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
588 True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
589 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
590 # based on answers to http://stackoverflow.com/questions/753052/ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
591 s = HTMLTextExtractor() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
592 s.feed(html) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
593 return s.get_text() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
594 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
595 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
596 _EMPTY_GZIP_BYTES = b'\x1f\x8b\x08\x089\xf3\xb9U\x00\x03empty\x00\x03\x00\x00\x00\x00\x00\x00\x00\x00\x00' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
597 _NON_EMPTY_GZIP_BYTES = b'\x1f\x8b\x08\x08\xbc\xf7\xb9U\x00\x03not_empty\x00K\xaa,I-N\xcc\xc8\xafT\xe4\x02\x00\xf3nb\xbf\x0b\x00\x00\x00' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
598 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
599 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
600 def gunzip_bytes(bytestring): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
601 """The :mod:`gzip` module is great if you have a file or file-like |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
602 object, but what if you just have bytes. StringIO is one |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
603 possibility, but it's often faster, easier, and simpler to just |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
604 use this one-liner. Use this tried-and-true utility function to |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
605 decompress gzip from bytes. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
606 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
607 >>> gunzip_bytes(_EMPTY_GZIP_BYTES) == b'' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
608 True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
609 >>> gunzip_bytes(_NON_EMPTY_GZIP_BYTES).rstrip() == b'bytesahoy!' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
610 True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
611 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
612 return zlib.decompress(bytestring, 16 + zlib.MAX_WBITS) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
613 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
614 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
615 def gzip_bytes(bytestring, level=6): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
616 """Turn some bytes into some compressed bytes. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
617 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
618 >>> len(gzip_bytes(b'a' * 10000)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
619 46 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
620 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
621 Args: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
622 bytestring (bytes): Bytes to be compressed |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
623 level (int): An integer, 1-9, controlling the |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
624 speed/compression. 1 is fastest, least compressed, 9 is |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
625 slowest, but most compressed. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
626 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
627 Note that all levels of gzip are pretty fast these days, though |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
628 it's not really a competitor in compression, at any level. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
629 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
630 out = StringIO() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
631 f = GzipFile(fileobj=out, mode='wb', compresslevel=level) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
632 f.write(bytestring) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
633 f.close() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
634 return out.getvalue() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
635 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
636 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
637 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
638 _line_ending_re = re.compile(r'(\r\n|\n|\x0b|\f|\r|\x85|\x2028|\x2029)', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
639 re.UNICODE) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
640 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
641 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
642 def iter_splitlines(text): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
643 r"""Like :meth:`str.splitlines`, but returns an iterator of lines |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
644 instead of a list. Also similar to :meth:`file.next`, as that also |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
645 lazily reads and yields lines from a file. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
646 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
647 This function works with a variety of line endings, but as always, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
648 be careful when mixing line endings within a file. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
649 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
650 >>> list(iter_splitlines('\nhi\nbye\n')) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
651 ['', 'hi', 'bye', ''] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
652 >>> list(iter_splitlines('\r\nhi\rbye\r\n')) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
653 ['', 'hi', 'bye', ''] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
654 >>> list(iter_splitlines('')) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
655 [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
656 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
657 prev_end, len_text = 0, len(text) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
658 # print('last: %r' % last_idx) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
659 # start, end = None, None |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
660 for match in _line_ending_re.finditer(text): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
661 start, end = match.start(1), match.end(1) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
662 # print(start, end) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
663 if prev_end <= start: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
664 yield text[prev_end:start] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
665 if end == len_text: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
666 yield '' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
667 prev_end = end |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
668 tail = text[prev_end:] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
669 if tail: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
670 yield tail |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
671 return |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
672 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
673 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
674 def indent(text, margin, newline='\n', key=bool): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
675 """The missing counterpart to the built-in :func:`textwrap.dedent`. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
676 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
677 Args: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
678 text (str): The text to indent. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
679 margin (str): The string to prepend to each line. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
680 newline (str): The newline used to rejoin the lines (default: ``\\n``) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
681 key (callable): Called on each line to determine whether to |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
682 indent it. Default: :class:`bool`, to ensure that empty lines do |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
683 not get whitespace added. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
684 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
685 indented_lines = [(margin + line if key(line) else line) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
686 for line in iter_splitlines(text)] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
687 return newline.join(indented_lines) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
688 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
689 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
690 def is_uuid(obj, version=4): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
691 """Check the argument is either a valid UUID object or string. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
692 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
693 Args: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
694 obj (object): The test target. Strings and UUID objects supported. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
695 version (int): The target UUID version, set to 0 to skip version check. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
696 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
697 >>> is_uuid('e682ccca-5a4c-4ef2-9711-73f9ad1e15ea') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
698 True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
699 >>> is_uuid('0221f0d9-d4b9-11e5-a478-10ddb1c2feb9') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
700 False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
701 >>> is_uuid('0221f0d9-d4b9-11e5-a478-10ddb1c2feb9', version=1) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
702 True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
703 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
704 if not isinstance(obj, uuid.UUID): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
705 try: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
706 obj = uuid.UUID(obj) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
707 except (TypeError, ValueError, AttributeError): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
708 return False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
709 if version and obj.version != int(version): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
710 return False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
711 return True |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
712 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
713 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
714 def escape_shell_args(args, sep=' ', style=None): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
715 """Returns an escaped version of each string in *args*, according to |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
716 *style*. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
717 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
718 Args: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
719 args (list): A list of arguments to escape and join together |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
720 sep (str): The separator used to join the escaped arguments. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
721 style (str): The style of escaping to use. Can be one of |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
722 ``cmd`` or ``sh``, geared toward Windows and Linux/BSD/etc., |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
723 respectively. If *style* is ``None``, then it is picked |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
724 according to the system platform. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
725 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
726 See :func:`args2cmd` and :func:`args2sh` for details and example |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
727 output for each style. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
728 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
729 if not style: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
730 style = 'cmd' if sys.platform == 'win32' else 'sh' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
731 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
732 if style == 'sh': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
733 return args2sh(args, sep=sep) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
734 elif style == 'cmd': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
735 return args2cmd(args, sep=sep) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
736 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
737 raise ValueError("style expected one of 'cmd' or 'sh', not %r" % style) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
738 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
739 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
740 _find_sh_unsafe = re.compile(r'[^a-zA-Z0-9_@%+=:,./-]').search |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
741 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
742 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
743 def args2sh(args, sep=' '): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
744 """Return a shell-escaped string version of *args*, separated by |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
745 *sep*, based on the rules of sh, bash, and other shells in the |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
746 Linux/BSD/MacOS ecosystem. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
747 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
748 >>> print(args2sh(['aa', '[bb]', "cc'cc", 'dd"dd'])) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
749 aa '[bb]' 'cc'"'"'cc' 'dd"dd' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
750 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
751 As you can see, arguments with no special characters are not |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
752 escaped, arguments with special characters are quoted with single |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
753 quotes, and single quotes themselves are quoted with double |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
754 quotes. Double quotes are handled like any other special |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
755 character. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
756 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
757 Based on code from the :mod:`pipes`/:mod:`shlex` modules. Also |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
758 note that :mod:`shlex` and :mod:`argparse` have functions to split |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
759 and parse strings escaped in this manner. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
760 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
761 ret_list = [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
762 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
763 for arg in args: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
764 if not arg: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
765 ret_list.append("''") |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
766 continue |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
767 if _find_sh_unsafe(arg) is None: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
768 ret_list.append(arg) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
769 continue |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
770 # use single quotes, and put single quotes into double quotes |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
771 # the string $'b is then quoted as '$'"'"'b' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
772 ret_list.append("'" + arg.replace("'", "'\"'\"'") + "'") |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
773 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
774 return ' '.join(ret_list) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
775 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
776 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
777 def args2cmd(args, sep=' '): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
778 r"""Return a shell-escaped string version of *args*, separated by |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
779 *sep*, using the same rules as the Microsoft C runtime. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
780 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
781 >>> print(args2cmd(['aa', '[bb]', "cc'cc", 'dd"dd'])) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
782 aa [bb] cc'cc dd\"dd |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
783 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
784 As you can see, escaping is through backslashing and not quoting, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
785 and double quotes are the only special character. See the comment |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
786 in the code for more details. Based on internal code from the |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
787 :mod:`subprocess` module. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
788 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
789 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
790 # technique description from subprocess below |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
791 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
792 1) Arguments are delimited by white space, which is either a |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
793 space or a tab. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
794 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
795 2) A string surrounded by double quotation marks is |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
796 interpreted as a single argument, regardless of white space |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
797 contained within. A quoted string can be embedded in an |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
798 argument. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
799 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
800 3) A double quotation mark preceded by a backslash is |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
801 interpreted as a literal double quotation mark. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
802 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
803 4) Backslashes are interpreted literally, unless they |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
804 immediately precede a double quotation mark. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
805 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
806 5) If backslashes immediately precede a double quotation mark, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
807 every pair of backslashes is interpreted as a literal |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
808 backslash. If the number of backslashes is odd, the last |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
809 backslash escapes the next double quotation mark as |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
810 described in rule 3. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
811 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
812 See http://msdn.microsoft.com/en-us/library/17w5ykft.aspx |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
813 or search http://msdn.microsoft.com for |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
814 "Parsing C++ Command-Line Arguments" |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
815 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
816 result = [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
817 needquote = False |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
818 for arg in args: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
819 bs_buf = [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
820 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
821 # Add a space to separate this argument from the others |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
822 if result: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
823 result.append(' ') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
824 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
825 needquote = (" " in arg) or ("\t" in arg) or not arg |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
826 if needquote: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
827 result.append('"') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
828 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
829 for c in arg: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
830 if c == '\\': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
831 # Don't know if we need to double yet. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
832 bs_buf.append(c) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
833 elif c == '"': |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
834 # Double backslashes. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
835 result.append('\\' * len(bs_buf)*2) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
836 bs_buf = [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
837 result.append('\\"') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
838 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
839 # Normal char |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
840 if bs_buf: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
841 result.extend(bs_buf) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
842 bs_buf = [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
843 result.append(c) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
844 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
845 # Add remaining backslashes, if any. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
846 if bs_buf: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
847 result.extend(bs_buf) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
848 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
849 if needquote: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
850 result.extend(bs_buf) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
851 result.append('"') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
852 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
853 return ''.join(result) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
854 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
855 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
856 def parse_int_list(range_string, delim=',', range_delim='-'): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
857 """Returns a sorted list of positive integers based on |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
858 *range_string*. Reverse of :func:`format_int_list`. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
859 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
860 Args: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
861 range_string (str): String of comma separated positive |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
862 integers or ranges (e.g. '1,2,4-6,8'). Typical of a custom |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
863 page range string used in printer dialogs. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
864 delim (char): Defaults to ','. Separates integers and |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
865 contiguous ranges of integers. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
866 range_delim (char): Defaults to '-'. Indicates a contiguous |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
867 range of integers. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
868 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
869 >>> parse_int_list('1,3,5-8,10-11,15') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
870 [1, 3, 5, 6, 7, 8, 10, 11, 15] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
871 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
872 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
873 output = [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
874 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
875 for x in range_string.strip().split(delim): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
876 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
877 # Range |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
878 if range_delim in x: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
879 range_limits = list(map(int, x.split(range_delim))) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
880 output += list(range(min(range_limits), max(range_limits)+1)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
881 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
882 # Empty String |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
883 elif not x: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
884 continue |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
885 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
886 # Integer |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
887 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
888 output.append(int(x)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
889 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
890 return sorted(output) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
891 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
892 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
893 def format_int_list(int_list, delim=',', range_delim='-', delim_space=False): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
894 """Returns a sorted range string from a list of positive integers |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
895 (*int_list*). Contiguous ranges of integers are collapsed to min |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
896 and max values. Reverse of :func:`parse_int_list`. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
897 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
898 Args: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
899 int_list (list): List of positive integers to be converted |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
900 into a range string (e.g. [1,2,4,5,6,8]). |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
901 delim (char): Defaults to ','. Separates integers and |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
902 contiguous ranges of integers. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
903 range_delim (char): Defaults to '-'. Indicates a contiguous |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
904 range of integers. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
905 delim_space (bool): Defaults to ``False``. If ``True``, adds a |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
906 space after all *delim* characters. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
907 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
908 >>> format_int_list([1,3,5,6,7,8,10,11,15]) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
909 '1,3,5-8,10-11,15' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
910 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
911 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
912 output = [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
913 contig_range = collections.deque() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
914 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
915 for x in sorted(int_list): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
916 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
917 # Handle current (and first) value. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
918 if len(contig_range) < 1: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
919 contig_range.append(x) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
920 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
921 # Handle current value, given multiple previous values are contiguous. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
922 elif len(contig_range) > 1: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
923 delta = x - contig_range[-1] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
924 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
925 # Current value is contiguous. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
926 if delta == 1: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
927 contig_range.append(x) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
928 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
929 # Current value is non-contiguous. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
930 elif delta > 1: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
931 range_substr = '{0:d}{1}{2:d}'.format(min(contig_range), |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
932 range_delim, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
933 max(contig_range)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
934 output.append(range_substr) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
935 contig_range.clear() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
936 contig_range.append(x) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
937 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
938 # Current value repeated. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
939 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
940 continue |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
941 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
942 # Handle current value, given no previous contiguous integers |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
943 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
944 delta = x - contig_range[0] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
945 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
946 # Current value is contiguous. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
947 if delta == 1: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
948 contig_range.append(x) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
949 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
950 # Current value is non-contiguous. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
951 elif delta > 1: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
952 output.append('{0:d}'.format(contig_range.popleft())) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
953 contig_range.append(x) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
954 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
955 # Current value repeated. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
956 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
957 continue |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
958 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
959 # Handle the last value. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
960 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
961 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
962 # Last value is non-contiguous. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
963 if len(contig_range) == 1: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
964 output.append('{0:d}'.format(contig_range.popleft())) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
965 contig_range.clear() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
966 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
967 # Last value is part of contiguous range. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
968 elif len(contig_range) > 1: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
969 range_substr = '{0:d}{1}{2:d}'.format(min(contig_range), |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
970 range_delim, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
971 max(contig_range)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
972 output.append(range_substr) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
973 contig_range.clear() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
974 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
975 if delim_space: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
976 output_str = (delim+' ').join(output) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
977 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
978 output_str = delim.join(output) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
979 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
980 return output_str |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
981 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
982 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
983 class MultiReplace(object): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
984 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
985 MultiReplace is a tool for doing multiple find/replace actions in one pass. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
986 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
987 Given a mapping of values to be replaced it allows for all of the matching |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
988 values to be replaced in a single pass which can save a lot of performance |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
989 on very large strings. In addition to simple replace, it also allows for |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
990 replacing based on regular expressions. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
991 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
992 Keyword Arguments: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
993 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
994 :type regex: bool |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
995 :param regex: Treat search keys as regular expressions [Default: False] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
996 :type flags: int |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
997 :param flags: flags to pass to the regex engine during compile |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
998 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
999 Dictionary Usage:: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1000 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1001 from lrmslib import stringutils |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1002 s = stringutils.MultiReplace({ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1003 'foo': 'zoo', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1004 'cat': 'hat', |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1005 'bat': 'kraken' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1006 }) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1007 new = s.sub('The foo bar cat ate a bat') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1008 new == 'The zoo bar hat ate a kraken' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1009 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1010 Iterable Usage:: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1011 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1012 from lrmslib import stringutils |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1013 s = stringutils.MultiReplace([ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1014 ('foo', 'zoo'), |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1015 ('cat', 'hat'), |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1016 ('bat', 'kraken)' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1017 ]) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1018 new = s.sub('The foo bar cat ate a bat') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1019 new == 'The zoo bar hat ate a kraken' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1020 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1021 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1022 The constructor can be passed a dictionary or other mapping as well as |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1023 an iterable of tuples. If given an iterable, the substitution will be run |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1024 in the order the replacement values are specified in the iterable. This is |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1025 also true if it is given an OrderedDict. If given a dictionary then the |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1026 order will be non-deterministic:: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1027 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1028 >>> 'foo bar baz'.replace('foo', 'baz').replace('baz', 'bar') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1029 'bar bar bar' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1030 >>> m = MultiReplace({'foo': 'baz', 'baz': 'bar'}) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1031 >>> m.sub('foo bar baz') |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1032 'baz bar bar' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1033 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1034 This is because the order of replacement can matter if you're inserting |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1035 something that might be replaced by a later substitution. Pay attention and |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1036 if you need to rely on order then consider using a list of tuples instead |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1037 of a dictionary. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1038 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1039 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1040 def __init__(self, sub_map, **kwargs): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1041 """Compile any regular expressions that have been passed.""" |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1042 options = { |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1043 'regex': False, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1044 'flags': 0, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1045 } |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1046 options.update(kwargs) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1047 self.group_map = {} |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1048 regex_values = [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1049 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1050 if isinstance(sub_map, Mapping): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1051 sub_map = sub_map.items() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1052 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1053 for idx, vals in enumerate(sub_map): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1054 group_name = 'group{0}'.format(idx) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1055 if isinstance(vals[0], basestring): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1056 # If we're not treating input strings like a regex, escape it |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1057 if not options['regex']: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1058 exp = re.escape(vals[0]) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1059 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1060 exp = vals[0] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1061 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1062 exp = vals[0].pattern |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1063 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1064 regex_values.append('(?P<{0}>{1})'.format( |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1065 group_name, |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1066 exp |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1067 )) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1068 self.group_map[group_name] = vals[1] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1069 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1070 self.combined_pattern = re.compile( |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1071 '|'.join(regex_values), |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1072 flags=options['flags'] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1073 ) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1074 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1075 def _get_value(self, match): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1076 """Given a match object find replacement value.""" |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1077 group_dict = match.groupdict() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1078 key = [x for x in group_dict if group_dict[x]][0] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1079 return self.group_map[key] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1080 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1081 def sub(self, text): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1082 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1083 Run substitutions on the input text. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1084 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1085 Given an input string, run all substitutions given in the |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1086 constructor. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1087 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1088 return self.combined_pattern.sub(self._get_value, text) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1089 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1090 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1091 def multi_replace(text, sub_map, **kwargs): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1092 """Shortcut function to invoke multi-replace in a single command.""" |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1093 m = MultiReplace(sub_map, **kwargs) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1094 return m.sub(text) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1095 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1096 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1097 def unwrap_text(text, ending='\n\n'): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1098 r""" |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1099 Unwrap text, the natural complement to :func:`textwrap.wrap`. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1100 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1101 >>> text = "Short \n lines \nwrapped\nsmall.\n\nAnother\nparagraph." |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1102 >>> unwrap_text(text) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1103 'Short lines wrapped small.\n\nAnother paragraph.' |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1104 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1105 Args: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1106 text: A string to unwrap. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1107 ending (str): The string to join all unwrapped paragraphs |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1108 by. Pass ``None`` to get the list. Defaults to '\n\n' for |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1109 compatibility with Markdown and RST. |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1110 |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1111 """ |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1112 all_grafs = [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1113 cur_graf = [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1114 for line in text.splitlines(): |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1115 line = line.strip() |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1116 if line: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1117 cur_graf.append(line) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1118 else: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1119 all_grafs.append(' '.join(cur_graf)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1120 cur_graf = [] |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1121 if cur_graf: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1122 all_grafs.append(' '.join(cur_graf)) |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1123 if ending is None: |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1124 return all_grafs |
26e78fe6e8c4
"planemo upload commit c699937486c35866861690329de38ec1a5d9f783"
shellac
parents:
diff
changeset
|
1125 return ending.join(all_grafs) |