Project

General

Profile

1 73 aaronmk
# String manipulation
2
3 1295 aaronmk
import codecs
4 861 aaronmk
import re
5
6 1232 aaronmk
import util
7
8 1622 aaronmk
##### Parsing
9
10 3749 aaronmk
def concat(str0, str1, max_len):
11
    str0, str1 = map(to_raw_str, [str0, str1])
12
    return to_unicode(str0[:max_len-len(str1)]+str1)
13 2932 aaronmk
14 1622 aaronmk
def split(sep, str_):
15
    '''Returns [] if str_ == ""'''
16
    if str_ == '': return []
17
    else: return str_.split(sep)
18
19 2221 aaronmk
def remove_prefix(prefix, str_, removed_ref=None):
20
    if removed_ref == None: removed_ref = [False]
21
22
    removed_ref[0] = str_.startswith(prefix)
23
    if removed_ref[0]: return str_[len(prefix):]
24 1622 aaronmk
    else: return str_
25
26 1680 aaronmk
def remove_prefixes(prefixes, str_):
27
    for prefix in prefixes: str_ = remove_prefix(prefix, str_)
28
    return str_
29
30
def with_prefixes(prefixes, str_): return (p+str_ for p in prefixes)
31
32 2221 aaronmk
def remove_suffix(suffix, str_, removed_ref=None):
33
    if removed_ref == None: removed_ref = [False]
34
35
    removed_ref[0] = str_.endswith(suffix)
36
    if removed_ref[0]: return str_[:-len(suffix)]
37 1622 aaronmk
    else: return str_
38
39 1959 aaronmk
def contains_any(haystack, needles):
40
    for needle in needles:
41
        if haystack.find(needle) >= 0: return True
42
    return False
43
44 1622 aaronmk
def overlaps(str0, str1): return str0.find(str1) >= 0 or str1.find(str0) >= 0
45
46 2761 aaronmk
##### Escaping
47
48
def esc_for_mogrify(query):
49
    '''Escapes a query right before being passed to a mogrifying function.'''
50
    return query.replace('%', '%%')
51
52 1401 aaronmk
##### Unicode
53
54 2882 aaronmk
def to_raw_str(str_):
55
    if isinstance(str_, unicode): str_ = str_.encode('utf_8')
56
    return str_
57
58 1295 aaronmk
unicode_reader = codecs.getreader('utf_8')
59
60 73 aaronmk
def to_unicode(str_):
61
    if isinstance(str_, unicode): return str_
62
    encodings = ['utf_8', 'latin_1']
63
    for encoding in encodings:
64
        try: return unicode(str_, encoding)
65
        except UnicodeDecodeError, e: pass
66
    raise AssertionError(encoding+' is not a catch-all encoding')
67 340 aaronmk
68 3748 aaronmk
def ustr(value):
69 1232 aaronmk
    '''Like built-in str() but converts to unicode object'''
70 3748 aaronmk
    if util.is_str(value): str_ = value # already a string
71
    elif hasattr(value, '__str__'): str_ = value.__str__()
72
    else: str_ = str(value)
73
    return to_unicode(str_)
74 1232 aaronmk
75 3747 aaronmk
def urepr(value):
76
    '''Like built-in repr() but converts to unicode object'''
77
    if hasattr(value, '__repr__'): str_ = value.__repr__()
78
    else: str_ = repr(value)
79
    return to_unicode(str_)
80
81 2502 aaronmk
def repr_no_u(value):
82
    '''Like built-in repr() but removes the "u" in `u'...'`'''
83 3747 aaronmk
    return re.sub(r"^u(?=')", r'', urepr(value))
84 2502 aaronmk
85 1401 aaronmk
##### Line endings
86
87 1622 aaronmk
def extract_line_ending(line):
88
    '''@return tuple (contents, ending)'''
89
    contents = remove_suffix('\r', remove_suffix('\n', line))
90
    return (contents, line[len(contents):])
91 714 aaronmk
92 1622 aaronmk
def remove_line_ending(line): return extract_line_ending(line)[0]
93
94
def ensure_newl(str_): return remove_line_ending(str_)+'\n'
95
96 856 aaronmk
def is_multiline(str_):
97
    newl_idx = str_.find('\n')
98
    return newl_idx >= 0 and newl_idx != len(str_)-1 # has newline before end
99
100 860 aaronmk
def remove_extra_newl(str_):
101 856 aaronmk
    if is_multiline(str_): return str_
102 2480 aaronmk
    else: return str_.rstrip('\n')
103 856 aaronmk
104 714 aaronmk
def std_newl(str_): return str_.replace('\r\n', '\n').replace('\r', '\n')
105
106 3255 aaronmk
def join_lines(lines): return ''.join((l+'\n' for l in lines))
107
108 1401 aaronmk
##### Whitespace
109
110 714 aaronmk
def cleanup(str_): return std_newl(str_.strip())
111 861 aaronmk
112 1364 aaronmk
def single_space(str_): return re.sub(r' {2,}', r' ', str_.strip())
113
114 861 aaronmk
def one_line(str_): return re.sub(r'\n *', r' ', cleanup(str_))
115 1761 aaronmk
116
##### Control characters
117
118
def is_ctrl(char):
119
    '''Whether char is a (non-printable) control character'''
120
    return ord(char) < 32 and not char.isspace()
121
122
def strip_ctrl(str_):
123
    '''Strips (non-printable) control characters'''
124
    return ''.join(filter(lambda c: not is_ctrl(c), str_))
125 2471 aaronmk
126 3172 aaronmk
##### Text
127
128
def first_word(str_): return str_.partition(' ')[0]
129
130 2471 aaronmk
##### Formatting
131
132 3466 aaronmk
def indent(str_, level=1, indent_str='    '):
133
    indent_str *= level
134
    return ('\n'.join((indent_str+l for l in str_.rstrip().split('\n'))))+'\n'
135
136 2477 aaronmk
def as_tt(str_): return '@'+str_+'@'
137
138 2475 aaronmk
def as_code(str_, lang=None, multiline=True):
139 2471 aaronmk
    '''Wraps a string in Redmine tags to syntax-highlight it.'''
140 2480 aaronmk
    str_ = '\n'+str_.rstrip('\n')+'\n'
141 2471 aaronmk
    if lang != None: str_ = '<code class="'+lang+'">'+str_+'</code>'
142 2475 aaronmk
    if multiline: str_ = '<pre>'+str_+'</pre>'
143
    return str_
144 2477 aaronmk
145 2504 aaronmk
def as_inline_table(dict_, key_label='Output', value_label='Input', ustr=ustr):
146 2477 aaronmk
    '''Wraps a dict in Redmine tags to format it as a table.'''
147
    str_ = ''
148 2481 aaronmk
    def row(entry): return (': '.join(entry))+'\n'
149 2477 aaronmk
    str_ += row([key_label, value_label])
150 2481 aaronmk
    for entry in dict_.iteritems(): str_ += row([ustr(v) for v in entry])
151
    return '<pre>\n'+str_+'</pre>'
152 2482 aaronmk
153 2504 aaronmk
def as_table(dict_, key_label='Output', value_label='Input', ustr=ustr):
154 2482 aaronmk
    '''Wraps a dict in Redmine tags to format it as a table.'''
155
    str_ = ''
156
    def row(entry): return ('|'.join(['']+entry+['']))+'\n'# '' for outer border
157
    str_ += row([key_label, value_label])
158
    for entry in dict_.iteritems(): str_ += row([as_tt(ustr(v)) for v in entry])
159
    return '\n'+str_+' ' # space protects last \n so blank line ends table