|
1
|
|
|
|
|
2
|
|
|
|
|
3
|
|
|
class StateMachine(object): |
|
4
|
|
|
''' |
|
5
|
|
|
Abstract simple character-driven state-machine for string transforms. |
|
6
|
|
|
|
|
7
|
|
|
Useful for implementig simple transpilations, compressors and so on. |
|
8
|
|
|
|
|
9
|
|
|
Important: when implementing this class, you must set the :attr:`current` |
|
10
|
|
|
attribute to a key defined in :attr:`jumps` dict. |
|
11
|
|
|
''' |
|
12
|
|
|
jumps = {} # state machine jumps |
|
13
|
|
|
start = '' # character which started current state |
|
14
|
|
|
current = '' # initial and current state |
|
15
|
|
|
pending = '' # buffer of current state data |
|
16
|
|
|
streaming = False # stream mode toggle |
|
17
|
|
|
|
|
18
|
|
|
def __init__(self, data=''): |
|
19
|
|
|
self.pending = data |
|
20
|
|
|
|
|
21
|
|
|
def transform(self, data, mark, next): |
|
22
|
|
|
method = getattr(self, 'transform_%s' % self.current, None) |
|
23
|
|
|
return method(data, mark, next) if method else data |
|
24
|
|
|
|
|
25
|
|
|
def look(self, value, current, start): |
|
26
|
|
|
offset = len(start) |
|
27
|
|
|
try: |
|
28
|
|
|
index = len(value) |
|
29
|
|
|
if self.streaming: |
|
30
|
|
|
index -= max(map(len, self.jumps)) |
|
31
|
|
|
size = 0 |
|
32
|
|
|
mark = '' |
|
33
|
|
|
next = None |
|
34
|
|
|
for amark, anext in self.jumps[current].items(): |
|
35
|
|
|
asize = len(amark) |
|
36
|
|
|
aindex = value.find(amark, offset, index + asize) |
|
37
|
|
|
if ( |
|
38
|
|
|
aindex == -1 or |
|
39
|
|
|
aindex > index or |
|
40
|
|
|
aindex == index and asize < size): |
|
41
|
|
|
continue |
|
42
|
|
|
index = aindex |
|
43
|
|
|
size = asize |
|
44
|
|
|
mark = amark |
|
45
|
|
|
next = anext |
|
46
|
|
|
except KeyError: |
|
47
|
|
|
raise KeyError( |
|
48
|
|
|
'Current state %r not defined in %s.jumps.' |
|
49
|
|
|
% (current, self.__class__) |
|
50
|
|
|
) |
|
51
|
|
|
return index, mark, next |
|
52
|
|
|
|
|
53
|
|
|
def __iter__(self): |
|
54
|
|
|
''' |
|
55
|
|
|
Iterate over tramsformation result chunks. |
|
56
|
|
|
|
|
57
|
|
|
On non-streaming mode, flush and yield it on completion. |
|
58
|
|
|
|
|
59
|
|
|
:yields: transformation result chunka |
|
60
|
|
|
:ytype: str |
|
61
|
|
|
''' |
|
62
|
|
|
while True: |
|
63
|
|
|
index, mark, next = self.look( |
|
64
|
|
|
self.pending, self.current, self.start) |
|
65
|
|
|
if next is None: |
|
66
|
|
|
break |
|
67
|
|
|
data = self.transform(self.pending[:index], mark, next) |
|
68
|
|
|
self.start = mark |
|
69
|
|
|
self.current = next |
|
70
|
|
|
self.pending = self.pending[index:] |
|
71
|
|
|
if data: |
|
72
|
|
|
yield data |
|
73
|
|
|
if not self.streaming: |
|
74
|
|
|
data = ( |
|
75
|
|
|
self.transform(self.pending, '', None) |
|
76
|
|
|
if self.pending else |
|
77
|
|
|
'' |
|
78
|
|
|
) |
|
79
|
|
|
self.pending = '' |
|
80
|
|
|
self.start = '' |
|
81
|
|
|
yield data |
|
82
|
|
|
|
|
83
|
|
|
def feed(self, data=''): |
|
84
|
|
|
self.streaming = True |
|
85
|
|
|
self.pending += data |
|
86
|
|
|
for i in self: |
|
87
|
|
|
yield i |
|
88
|
|
|
|
|
89
|
|
|
def finish(self, data=''): |
|
90
|
|
|
self.pending += data |
|
91
|
|
|
self.streaming = False |
|
92
|
|
|
for i in self: |
|
93
|
|
|
yield i |
|
94
|
|
|
|
ergoithz /
browsepy
