Mercurial > p > roundup > code

comparison roundup/anypy/strings.py @ 6048:81990ac0b013

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
flak8: move import; whitespace changes; ignore undefined global 'unicode' in py3
author John Rouillard <rouilj@ieee.org>
date Sat, 11 Jan 2020 16:38:38 -0500
parents bb7865241f8a
children 82f870433b18
comparison
equal deleted inserted replaced
6047:142553f58694 6048:81990ac0b013
3 # encoded using UTF-8, and the Python 2 unicode type is only used in a 3 # encoded using UTF-8, and the Python 2 unicode type is only used in a
4 # few places, generally for interacting with external modules 4 # few places, generally for interacting with external modules
5 # requiring that type to be used. 5 # requiring that type to be used.
6 6
7 import sys 7 import sys
8 import io
9
8 _py3 = sys.version_info[0] > 2 10 _py3 = sys.version_info[0] > 2
9 11
10 import io
11 if _py3: 12 if _py3:
12 StringIO = io.StringIO 13 StringIO = io.StringIO
13 else: 14 else:
14 StringIO = io.BytesIO 15 StringIO = io.BytesIO
16
15 17
16 def b2s(b): 18 def b2s(b):
17 """Convert a UTF-8 encoded bytes object to the internal string format.""" 19 """Convert a UTF-8 encoded bytes object to the internal string format."""
18 if _py3: 20 if _py3:
19 return b.decode('utf-8') 21 return b.decode('utf-8')
20 else: 22 else:
21 return b 23 return b
22 24
25
23 def s2b(s): 26 def s2b(s):
24 """Convert a string object to UTF-8 encoded bytes.""" 27 """Convert a string object to UTF-8 encoded bytes."""
25 if _py3: 28 if _py3:
26 return s.encode('utf-8') 29 return s.encode('utf-8')
27 else: 30 else:
28 return s 31 return s
29 32
33
30 def bs2b(s): 34 def bs2b(s):
31 """Convert a string object or UTF-8 encoded bytes to UTF-8 encoded bytes.""" 35 """Convert a string object or UTF-8 encoded bytes to UTF-8 encoded bytes.
36 """
32 if _py3: 37 if _py3:
33 if isinstance(s, bytes): 38 if isinstance(s, bytes):
34 return s 39 return s
35 else: 40 else:
36 return s.encode('utf-8') 41 return s.encode('utf-8')
37 else: 42 else:
38 return s 43 return s
39 44
45
40 def s2u(s, errors='strict'): 46 def s2u(s, errors='strict'):
41 """Convert a string object to a Unicode string.""" 47 """Convert a string object to a Unicode string."""
42 if _py3: 48 if _py3:
43 return s 49 return s
44 else: 50 else:
45 return unicode(s, 'utf-8', errors) 51 return unicode(s, 'utf-8', errors) # noqa: 821
52
46 53
47 def u2s(u): 54 def u2s(u):
48 """Convert a Unicode string to the internal string format.""" 55 """Convert a Unicode string to the internal string format."""
49 if _py3: 56 if _py3:
50 return u 57 return u
51 else: 58 else:
52 return u.encode('utf-8') 59 return u.encode('utf-8')
53 60
61
54 def us2u(s, errors='strict'): 62 def us2u(s, errors='strict'):
55 """Convert a string or Unicode string to a Unicode string.""" 63 """Convert a string or Unicode string to a Unicode string."""
56 if _py3: 64 if _py3:
57 return s 65 return s
58 else: 66 else:
59 if isinstance(s, unicode): 67 if isinstance(s, unicode): # noqa: 821
60 return s 68 return s
61 else: 69 else:
62 return unicode(s, 'utf-8', errors) 70 return unicode(s, 'utf-8', errors) # noqa: 821
71
63 72
64 def us2s(u): 73 def us2s(u):
65 """Convert a string or Unicode string to the internal string format.""" 74 """Convert a string or Unicode string to the internal string format."""
66 if _py3: 75 if _py3:
67 return u 76 return u
68 else: 77 else:
69 if isinstance(u, unicode): 78 if isinstance(u, unicode): # noqa: 821
70 return u.encode('utf-8') 79 return u.encode('utf-8')
71 else: 80 else:
72 return u 81 return u
82
73 83
74 def uany2s(u): 84 def uany2s(u):
75 """Convert a Unicode string or other object to the internal string format. 85 """Convert a Unicode string or other object to the internal string format.
76 86
77 Objects that are not Unicode strings are passed to str().""" 87 Objects that are not Unicode strings are passed to str()."""
78 if _py3: 88 if _py3:
79 return str(u) 89 return str(u)
80 else: 90 else:
81 if isinstance(u, unicode): 91 if isinstance(u, unicode): # noqa: 821
82 return u.encode('utf-8') 92 return u.encode('utf-8')
83 else: 93 else:
84 return str(u) 94 return str(u)
95
85 96
86 def is_us(s): 97 def is_us(s):
87 """Return whether an object is a string or Unicode string.""" 98 """Return whether an object is a string or Unicode string."""
88 if _py3: 99 if _py3:
89 return isinstance(s, str) 100 return isinstance(s, str)
90 else: 101 else:
91 return isinstance(s, str) or isinstance(s, unicode) 102 return isinstance(s, str) or isinstance(s, unicode) # noqa: 821
103
92 104
93 def uchr(c): 105 def uchr(c):
94 """Return the Unicode string containing the given character.""" 106 """Return the Unicode string containing the given character."""
95 if _py3: 107 if _py3:
96 return chr(c) 108 return chr(c)
97 else: 109 else:
98 return unichr(c) 110 return unichr(c) # noqa: 821
99 111
100 # CSV files used for export and import represent strings in the style 112 # CSV files used for export and import represent strings in the style
101 # used by repr in Python 2; this means that each byte of the UTF-8 113 # used by repr in Python 2; this means that each byte of the UTF-8
102 # representation is represented by a \x escape if not a printable 114 # representation is represented by a \x escape if not a printable
103 # ASCII character. When such a representation is interpreted by eval 115 # ASCII character. When such a representation is interpreted by eval
106 # as ISO-8859-1 produces the correct byte-string which must then be 118 # as ISO-8859-1 produces the correct byte-string which must then be
107 # decoded as UTF-8 to produce the correct Unicode string. The same 119 # decoded as UTF-8 to produce the correct Unicode string. The same
108 # representations are also used for journal storage in RDBMS 120 # representations are also used for journal storage in RDBMS
109 # databases, so that the database can be compatible between Python 2 121 # databases, so that the database can be compatible between Python 2
110 # and Python 3. 122 # and Python 3.
123
111 124
112 def repr_export(v): 125 def repr_export(v):
113 """Return a Python-2-style representation of a value for export to CSV.""" 126 """Return a Python-2-style representation of a value for export to CSV."""
114 if _py3: 127 if _py3:
115 if isinstance(v, str): 128 if isinstance(v, str):
122 return '{%s}' % ', '.join(repr_vals) 135 return '{%s}' % ', '.join(repr_vals)
123 else: 136 else:
124 return repr(v) 137 return repr(v)
125 else: 138 else:
126 return repr(v) 139 return repr(v)
140
127 141
128 def eval_import(s): 142 def eval_import(s):
129 """Evaluate a Python-2-style value imported from a CSV file.""" 143 """Evaluate a Python-2-style value imported from a CSV file."""
130 if _py3: 144 if _py3:
131 v = eval(s) 145 v = eval(s)
Roundup Issue Tracker: http://roundup-tracker.org/