1
|
# I/O
|
2
|
|
3
|
import operator
|
4
|
import timeouts
|
5
|
|
6
|
##### Exceptions
|
7
|
|
8
|
class InputStreamsOnlyException(Exception):
|
9
|
def __init__(self): Exception.__init__(self, 'Only input streams allowed')
|
10
|
|
11
|
##### Wrapped streams
|
12
|
|
13
|
class WrapStream:
|
14
|
'''Forwards close() to the underlying stream'''
|
15
|
def __init__(self, stream): self.stream = stream
|
16
|
|
17
|
def close(self): self.stream.close()
|
18
|
|
19
|
##### Line iteration
|
20
|
|
21
|
class StreamIter(WrapStream):
|
22
|
'''Iterates over the lines in a stream.
|
23
|
Unlike stream.__iter__(), doesn't wait for EOF to start returning lines.
|
24
|
Offers curr() method.
|
25
|
'''
|
26
|
def __init__(self, stream):
|
27
|
WrapStream.__init__(self, stream)
|
28
|
self.line = None
|
29
|
|
30
|
def __iter__(self): return self
|
31
|
|
32
|
def curr(self):
|
33
|
if self.line == None: self.line = self.readline()
|
34
|
if self.line == '': raise StopIteration
|
35
|
return self.line
|
36
|
|
37
|
def next(self):
|
38
|
line = self.curr()
|
39
|
self.line = None
|
40
|
return line
|
41
|
|
42
|
# Define as a separate method so it can be overridden
|
43
|
def readline(self): return self.stream.readline()
|
44
|
|
45
|
def copy(from_, to):
|
46
|
for line in StreamIter(from_): to.write(line)
|
47
|
|
48
|
def consume(in_):
|
49
|
for line in StreamIter(in_): pass
|
50
|
|
51
|
def read_all(stream): return reduce(operator.add, StreamIter(stream), '')
|
52
|
|
53
|
##### Timeouts
|
54
|
|
55
|
class TimeoutInputStream(WrapStream):
|
56
|
'''@param timeout sec'''
|
57
|
def __init__(self, stream, timeout):
|
58
|
WrapStream.__init__(self, stream)
|
59
|
self.timeout = timeout
|
60
|
|
61
|
def readline(self):
|
62
|
return timeouts.run(lambda: self.stream.readline(), self.timeout)
|
63
|
|
64
|
def write(self, str_): raise InputStreamsOnlyException()
|
65
|
|
66
|
##### Filtered/traced streams
|
67
|
|
68
|
class FilterStream(StreamIter):
|
69
|
'''Wraps a stream, filtering each string read or written'''
|
70
|
def __init__(self, filter_, stream):
|
71
|
StreamIter.__init__(self, stream)
|
72
|
self.filter = filter_
|
73
|
|
74
|
def readline(self): return self.filter(StreamIter.readline(self))
|
75
|
|
76
|
def read(self, n): return self.readline() # forward all reads to readline()
|
77
|
|
78
|
def write(self, str_): return self.stream.write(self.filter(str_))
|
79
|
|
80
|
class TracedStream(FilterStream):
|
81
|
'''Wraps a stream, running a trace function on each string read or written
|
82
|
'''
|
83
|
def __init__(self, trace, stream):
|
84
|
def filter_(str_):
|
85
|
trace(str_)
|
86
|
return str_
|
87
|
FilterStream.__init__(self, filter_, stream)
|
88
|
|
89
|
class LineCountStream(TracedStream):
|
90
|
'''Wraps a unidirectional stream, making the current line number available.
|
91
|
Lines start counting from 1.'''
|
92
|
def __init__(self, stream):
|
93
|
self.line_num = 1
|
94
|
def trace(str_): self.line_num += str_.count('\n')
|
95
|
TracedStream.__init__(self, trace, stream)
|
96
|
|
97
|
class LineCountInputStream(TracedStream):
|
98
|
'''Wraps an input stream, making the current line number available.
|
99
|
Lines start counting from 1.
|
100
|
This is faster than LineCountStream for input streams, because all reading
|
101
|
is done through readline() so newlines don't need to be explicitly counted.
|
102
|
'''
|
103
|
def __init__(self, stream):
|
104
|
self.line_num = 1
|
105
|
def trace(str_):
|
106
|
if str_ != '': self.line_num += 1 # EOF is not a line
|
107
|
TracedStream.__init__(self, trace, stream)
|
108
|
|
109
|
def write(self, str_): raise InputStreamsOnlyException()
|
110
|
|
111
|
class ProgressInputStream(TracedStream):
|
112
|
'''Wraps an input stream, reporting the # lines read every n lines and after
|
113
|
the last line is read.
|
114
|
@param log The output stream for progress messages
|
115
|
'''
|
116
|
def __init__(self, stream, log, msg='Read %d line(s)', n=100):
|
117
|
self.eof = False
|
118
|
|
119
|
def trace(str_):
|
120
|
msg_ = msg # may be modified, so can't have same name as outer var
|
121
|
if str_ == None: # closed
|
122
|
if self.eof: return # not closed prematurely
|
123
|
str_ = '' # closed prematurely, so signal EOF
|
124
|
msg_ += ' (not all input read)'
|
125
|
|
126
|
self.eof = eof = str_ == ''
|
127
|
if eof: line_ending = '\n'
|
128
|
else: line_ending = '\r'
|
129
|
|
130
|
line_ct = self.stream.line_num - 1 # make it 0-based
|
131
|
if eof or line_ct % n == 0: log.write((msg_ % line_ct)+line_ending)
|
132
|
self.trace = trace
|
133
|
|
134
|
TracedStream.__init__(self, trace, LineCountInputStream(stream))
|
135
|
|
136
|
def close(self):
|
137
|
self.trace(None) # signal closed
|
138
|
TracedStream.close(self)
|
139
|
|
140
|
class CaptureStream(TracedStream):
|
141
|
'''Wraps a stream, capturing matching text.
|
142
|
Matches can be retrieved from self.matches.'''
|
143
|
def __init__(self, stream, start_str, end_str):
|
144
|
self.recording = False
|
145
|
self.matches = []
|
146
|
|
147
|
def trace(str_):
|
148
|
start_idx = 0
|
149
|
if not self.recording:
|
150
|
start_idx = str_.find(start_str)
|
151
|
if start_idx >= 0:
|
152
|
self.recording = True
|
153
|
self.matches.append('')
|
154
|
if self.recording:
|
155
|
end_idx = str_.find(end_str)
|
156
|
if end_idx >= 0:
|
157
|
self.recording = False
|
158
|
end_idx += len(end_str)
|
159
|
else: end_idx = len(str_)
|
160
|
|
161
|
self.matches[-1] += str_[start_idx:end_idx]
|
162
|
|
163
|
TracedStream.__init__(self, trace, stream)
|