Show More
@@ -1,207 +1,206 b'' | |||
|
1 | 1 | # Copyright 2011, Google Inc. |
|
2 | 2 | # All rights reserved. |
|
3 | 3 | # |
|
4 | 4 | # Redistribution and use in source and binary forms, with or without |
|
5 | 5 | # modification, are permitted provided that the following conditions are |
|
6 | 6 | # met: |
|
7 | 7 | # |
|
8 | 8 | # * Redistributions of source code must retain the above copyright |
|
9 | 9 | # notice, this list of conditions and the following disclaimer. |
|
10 | 10 | # * Redistributions in binary form must reproduce the above |
|
11 | 11 | # copyright notice, this list of conditions and the following disclaimer |
|
12 | 12 | # in the documentation and/or other materials provided with the |
|
13 | 13 | # distribution. |
|
14 | 14 | # * Neither the name of Google Inc. nor the names of its |
|
15 | 15 | # contributors may be used to endorse or promote products derived from |
|
16 | 16 | # this software without specific prior written permission. |
|
17 | 17 | |
|
18 | 18 | # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
|
19 | 19 | # "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
|
20 | 20 | # LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR |
|
21 | 21 | # A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT |
|
22 | 22 | # OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, |
|
23 | 23 | # SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT |
|
24 | 24 | # LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
|
25 | 25 | # DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
|
26 | 26 | # THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
|
27 | 27 | # (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
|
28 | 28 | # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
|
29 | 29 | """Reader objects to abstract out different body response types. |
|
30 | 30 | |
|
31 | 31 | This module is package-private. It is not expected that these will |
|
32 | 32 | have any clients outside of httpplus. |
|
33 | 33 | """ |
|
34 | 34 | |
|
35 | 35 | import httplib |
|
36 | 36 | import itertools |
|
37 | 37 | import logging |
|
38 | 38 | |
|
39 | 39 | logger = logging.getLogger(__name__) |
|
40 | 40 | |
|
41 | 41 | |
|
42 | 42 | class ReadNotReady(Exception): |
|
43 | 43 | """Raised when read() is attempted but not enough data is loaded.""" |
|
44 | 44 | |
|
45 | 45 | |
|
46 | 46 | class HTTPRemoteClosedError(httplib.HTTPException): |
|
47 | 47 | """The server closed the remote socket in the middle of a response.""" |
|
48 | 48 | |
|
49 | 49 | |
|
50 | 50 | class AbstractReader(object): |
|
51 | 51 | """Abstract base class for response readers. |
|
52 | 52 | |
|
53 | 53 | Subclasses must implement _load, and should implement _close if |
|
54 | 54 | it's not an error for the server to close their socket without |
|
55 | 55 | some termination condition being detected during _load. |
|
56 | 56 | """ |
|
57 | 57 | def __init__(self): |
|
58 | 58 | self._finished = False |
|
59 | 59 | self._done_chunks = [] |
|
60 | 60 | self.available_data = 0 |
|
61 | 61 | |
|
62 | 62 | def addchunk(self, data): |
|
63 | 63 | self._done_chunks.append(data) |
|
64 | 64 | self.available_data += len(data) |
|
65 | 65 | |
|
66 | 66 | def pushchunk(self, data): |
|
67 | 67 | self._done_chunks.insert(0, data) |
|
68 | 68 | self.available_data += len(data) |
|
69 | 69 | |
|
70 | 70 | def popchunk(self): |
|
71 | 71 | b = self._done_chunks.pop(0) |
|
72 | 72 | self.available_data -= len(b) |
|
73 | 73 | |
|
74 | 74 | return b |
|
75 | 75 | |
|
76 | 76 | def done(self): |
|
77 | 77 | return self._finished |
|
78 | 78 | |
|
79 | 79 | def read(self, amt): |
|
80 | 80 | if self.available_data < amt and not self._finished: |
|
81 | 81 | raise ReadNotReady() |
|
82 |
|
|
|
83 | def pred(s): | |
|
84 | needed = need[0] > 0 | |
|
85 |
|
|
|
86 |
|
|
|
87 | blocks = list(itertools.takewhile(pred, self._done_chunks)) | |
|
88 | self._done_chunks = self._done_chunks[len(blocks):] | |
|
89 | over_read = sum(map(len, blocks)) - amt | |
|
90 | if over_read > 0 and blocks: | |
|
91 | logger.debug('need to reinsert %d data into done chunks', over_read) | |
|
92 | last = blocks[-1] | |
|
93 | blocks[-1], reinsert = last[:-over_read], last[-over_read:] | |
|
94 | self._done_chunks.insert(0, reinsert) | |
|
82 | blocks = [] | |
|
83 | need = amt | |
|
84 | while self._done_chunks: | |
|
85 | b = self.popchunk() | |
|
86 | if len(b) > need: | |
|
87 | nb = b[:need] | |
|
88 | self.pushchunk(b[need:]) | |
|
89 | b = nb | |
|
90 | blocks.append(b) | |
|
91 | need -= len(b) | |
|
92 | if need == 0: | |
|
93 | break | |
|
95 | 94 | result = ''.join(blocks) |
|
96 | 95 | assert len(result) == amt or (self._finished and len(result) < amt) |
|
97 | self.available_data -= amt | |
|
96 | ||
|
98 | 97 | return result |
|
99 | 98 | |
|
100 | 99 | def _load(self, data): # pragma: no cover |
|
101 | 100 | """Subclasses must implement this. |
|
102 | 101 | |
|
103 | 102 | As data is available to be read out of this object, it should |
|
104 | 103 | be placed into the _done_chunks list. Subclasses should not |
|
105 | 104 | rely on data remaining in _done_chunks forever, as it may be |
|
106 | 105 | reaped if the client is parsing data as it comes in. |
|
107 | 106 | """ |
|
108 | 107 | raise NotImplementedError |
|
109 | 108 | |
|
110 | 109 | def _close(self): |
|
111 | 110 | """Default implementation of close. |
|
112 | 111 | |
|
113 | 112 | The default implementation assumes that the reader will mark |
|
114 | 113 | the response as finished on the _finished attribute once the |
|
115 | 114 | entire response body has been read. In the event that this is |
|
116 | 115 | not true, the subclass should override the implementation of |
|
117 | 116 | close (for example, close-is-end responses have to set |
|
118 | 117 | self._finished in the close handler.) |
|
119 | 118 | """ |
|
120 | 119 | if not self._finished: |
|
121 | 120 | raise HTTPRemoteClosedError( |
|
122 | 121 | 'server appears to have closed the socket mid-response') |
|
123 | 122 | |
|
124 | 123 | |
|
125 | 124 | class AbstractSimpleReader(AbstractReader): |
|
126 | 125 | """Abstract base class for simple readers that require no response decoding. |
|
127 | 126 | |
|
128 | 127 | Examples of such responses are Connection: Close (close-is-end) |
|
129 | 128 | and responses that specify a content length. |
|
130 | 129 | """ |
|
131 | 130 | def _load(self, data): |
|
132 | 131 | if data: |
|
133 | 132 | assert not self._finished, ( |
|
134 | 133 | 'tried to add data (%r) to a closed reader!' % data) |
|
135 | 134 | logger.debug('%s read an additional %d data', self.name, len(data)) |
|
136 | 135 | self.addchunk(data) |
|
137 | 136 | |
|
138 | 137 | |
|
139 | 138 | class CloseIsEndReader(AbstractSimpleReader): |
|
140 | 139 | """Reader for responses that specify Connection: Close for length.""" |
|
141 | 140 | name = 'close-is-end' |
|
142 | 141 | |
|
143 | 142 | def _close(self): |
|
144 | 143 | logger.info('Marking close-is-end reader as closed.') |
|
145 | 144 | self._finished = True |
|
146 | 145 | |
|
147 | 146 | |
|
148 | 147 | class ContentLengthReader(AbstractSimpleReader): |
|
149 | 148 | """Reader for responses that specify an exact content length.""" |
|
150 | 149 | name = 'content-length' |
|
151 | 150 | |
|
152 | 151 | def __init__(self, amount): |
|
153 | 152 | AbstractReader.__init__(self) |
|
154 | 153 | self._amount = amount |
|
155 | 154 | if amount == 0: |
|
156 | 155 | self._finished = True |
|
157 | 156 | self._amount_seen = 0 |
|
158 | 157 | |
|
159 | 158 | def _load(self, data): |
|
160 | 159 | AbstractSimpleReader._load(self, data) |
|
161 | 160 | self._amount_seen += len(data) |
|
162 | 161 | if self._amount_seen >= self._amount: |
|
163 | 162 | self._finished = True |
|
164 | 163 | logger.debug('content-length read complete') |
|
165 | 164 | |
|
166 | 165 | |
|
167 | 166 | class ChunkedReader(AbstractReader): |
|
168 | 167 | """Reader for chunked transfer encoding responses.""" |
|
169 | 168 | def __init__(self, eol): |
|
170 | 169 | AbstractReader.__init__(self) |
|
171 | 170 | self._eol = eol |
|
172 | 171 | self._leftover_skip_amt = 0 |
|
173 | 172 | self._leftover_data = '' |
|
174 | 173 | |
|
175 | 174 | def _load(self, data): |
|
176 | 175 | assert not self._finished, 'tried to add data to a closed reader!' |
|
177 | 176 | logger.debug('chunked read an additional %d data', len(data)) |
|
178 | 177 | position = 0 |
|
179 | 178 | if self._leftover_data: |
|
180 | 179 | logger.debug('chunked reader trying to finish block from leftover data') |
|
181 | 180 | # TODO: avoid this string concatenation if possible |
|
182 | 181 | data = self._leftover_data + data |
|
183 | 182 | position = self._leftover_skip_amt |
|
184 | 183 | self._leftover_data = '' |
|
185 | 184 | self._leftover_skip_amt = 0 |
|
186 | 185 | datalen = len(data) |
|
187 | 186 | while position < datalen: |
|
188 | 187 | split = data.find(self._eol, position) |
|
189 | 188 | if split == -1: |
|
190 | 189 | self._leftover_data = data |
|
191 | 190 | self._leftover_skip_amt = position |
|
192 | 191 | return |
|
193 | 192 | amt = int(data[position:split], base=16) |
|
194 | 193 | block_start = split + len(self._eol) |
|
195 | 194 | # If the whole data chunk plus the eol trailer hasn't |
|
196 | 195 | # loaded, we'll wait for the next load. |
|
197 | 196 | if block_start + amt + len(self._eol) > len(data): |
|
198 | 197 | self._leftover_data = data |
|
199 | 198 | self._leftover_skip_amt = position |
|
200 | 199 | return |
|
201 | 200 | if amt == 0: |
|
202 | 201 | self._finished = True |
|
203 | 202 | logger.debug('closing chunked reader due to chunk of length 0') |
|
204 | 203 | return |
|
205 | 204 | self.addchunk(data[block_start:block_start + amt]) |
|
206 | 205 | position = block_start + amt + len(self._eol) |
|
207 | 206 | # no-check-code |
General Comments 0
You need to be logged in to leave comments.
Login now