|
1
|
|
|
|
|
2
|
|
|
class StateMachine(object): |
|
3
|
|
|
''' |
|
4
|
|
|
Abstract simple character-driven state-machine with result transformations. |
|
5
|
|
|
|
|
6
|
|
|
Useful for implementig simple parsers. |
|
7
|
|
|
''' |
|
8
|
|
|
jumps = {} # state machine jumps |
|
9
|
|
|
start = '' # character which started current state |
|
10
|
|
|
current = '' # initial and current state |
|
11
|
|
|
pending = '' # buffer of current state data |
|
12
|
|
|
|
|
13
|
|
|
def transform(self, data, current, start, partial=False): |
|
14
|
|
|
method = getattr(self, 'transform_%s' % current, None) |
|
15
|
|
|
return method(data, current, start, partial=False) if method else data |
|
16
|
|
|
|
|
17
|
|
|
def look(self, value, current, start): |
|
18
|
|
|
offset = len(start) |
|
19
|
|
|
for mark, next in self.jumps[current].items(): |
|
20
|
|
|
index = value.find(mark, offset) |
|
21
|
|
|
if -1 != index: |
|
22
|
|
|
yield index, mark, next |
|
23
|
|
|
yield len(value), '', None # failing is also an option |
|
24
|
|
|
|
|
25
|
|
|
def finalize(self): |
|
26
|
|
|
if self.pending: |
|
27
|
|
|
yield self.transform(self.pending, self.current, self.start, True) |
|
28
|
|
|
self.pending = '' |
|
29
|
|
|
self.current = '' |
|
30
|
|
|
|
|
31
|
|
|
def __iter__(self): |
|
32
|
|
|
while True: |
|
33
|
|
|
index, mark, next = min( |
|
34
|
|
|
self.look(self.pending, self.current, self.start), |
|
35
|
|
|
key=lambda x: (x[0], -len(x[1])) |
|
36
|
|
|
) |
|
37
|
|
|
if next is None: |
|
38
|
|
|
break |
|
39
|
|
|
data = self.transform( |
|
40
|
|
|
self.pending[:index], |
|
41
|
|
|
self.current, |
|
42
|
|
|
self.start |
|
43
|
|
|
) |
|
44
|
|
|
self.start = mark |
|
45
|
|
|
self.current = next |
|
46
|
|
|
self.pending = self.pending[index:] |
|
47
|
|
|
yield data |
|
48
|
|
|
|
|
49
|
|
|
|
|
50
|
|
|
class BlobTransform(StateMachine): |
|
51
|
|
|
jumps = { # state machine jumps |
|
52
|
|
|
'escape': { |
|
53
|
|
|
'': 'literal', |
|
54
|
|
|
}, |
|
55
|
|
|
'lit': { |
|
56
|
|
|
'\\': 'escape', |
|
57
|
|
|
} |
|
58
|
|
|
} |
|
59
|
|
|
|
ergoithz /
browsepy
