Mercurial > public > mercurial-scm > hg-stable
annotate mercurial/utils/cborutil.py @ 39440:babad5ebaf0a
cborutil: add a buffering decoder
The sansiodecoder leaves it up to the callers to feed in data that
wasn't fully consumed last time.
This commit implements a decoder that performs buffering of
leftover chunks from the previous invocation. It otherwise
behaves identically to sansiodecoder.
Differential Revision: https://phab.mercurial-scm.org/D4434
author | Gregory Szorc <gregory.szorc@gmail.com> |
---|---|
date | Wed, 29 Aug 2018 14:29:01 -0700 |
parents | a40d3da89b7d |
children | 8d858fbf2759 |
rev | line source |
---|---|
37711
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
1 # cborutil.py - CBOR extensions |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
2 # |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
3 # Copyright 2018 Gregory Szorc <gregory.szorc@gmail.com> |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
4 # |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
5 # This software may be used and distributed according to the terms of the |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
6 # GNU General Public License version 2 or any later version. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
7 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
8 from __future__ import absolute_import |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
9 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
10 import struct |
39438
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
11 import sys |
37711
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
12 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
13 # Very short very of RFC 7049... |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
14 # |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
15 # Each item begins with a byte. The 3 high bits of that byte denote the |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
16 # "major type." The lower 5 bits denote the "subtype." Each major type |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
17 # has its own encoding mechanism. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
18 # |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
19 # Most types have lengths. However, bytestring, string, array, and map |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
20 # can be indefinite length. These are denotes by a subtype with value 31. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
21 # Sub-components of those types then come afterwards and are terminated |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
22 # by a "break" byte. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
23 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
24 MAJOR_TYPE_UINT = 0 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
25 MAJOR_TYPE_NEGINT = 1 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
26 MAJOR_TYPE_BYTESTRING = 2 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
27 MAJOR_TYPE_STRING = 3 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
28 MAJOR_TYPE_ARRAY = 4 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
29 MAJOR_TYPE_MAP = 5 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
30 MAJOR_TYPE_SEMANTIC = 6 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
31 MAJOR_TYPE_SPECIAL = 7 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
32 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
33 SUBTYPE_MASK = 0b00011111 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
34 |
39438
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
35 SUBTYPE_FALSE = 20 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
36 SUBTYPE_TRUE = 21 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
37 SUBTYPE_NULL = 22 |
37711
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
38 SUBTYPE_HALF_FLOAT = 25 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
39 SUBTYPE_SINGLE_FLOAT = 26 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
40 SUBTYPE_DOUBLE_FLOAT = 27 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
41 SUBTYPE_INDEFINITE = 31 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
42 |
39438
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
43 SEMANTIC_TAG_FINITE_SET = 258 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
44 |
37711
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
45 # Indefinite types begin with their major type ORd with information value 31. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
46 BEGIN_INDEFINITE_BYTESTRING = struct.pack( |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
47 r'>B', MAJOR_TYPE_BYTESTRING << 5 | SUBTYPE_INDEFINITE) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
48 BEGIN_INDEFINITE_ARRAY = struct.pack( |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
49 r'>B', MAJOR_TYPE_ARRAY << 5 | SUBTYPE_INDEFINITE) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
50 BEGIN_INDEFINITE_MAP = struct.pack( |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
51 r'>B', MAJOR_TYPE_MAP << 5 | SUBTYPE_INDEFINITE) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
52 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
53 ENCODED_LENGTH_1 = struct.Struct(r'>B') |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
54 ENCODED_LENGTH_2 = struct.Struct(r'>BB') |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
55 ENCODED_LENGTH_3 = struct.Struct(r'>BH') |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
56 ENCODED_LENGTH_4 = struct.Struct(r'>BL') |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
57 ENCODED_LENGTH_5 = struct.Struct(r'>BQ') |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
58 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
59 # The break ends an indefinite length item. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
60 BREAK = b'\xff' |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
61 BREAK_INT = 255 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
62 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
63 def encodelength(majortype, length): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
64 """Obtain a value encoding the major type and its length.""" |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
65 if length < 24: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
66 return ENCODED_LENGTH_1.pack(majortype << 5 | length) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
67 elif length < 256: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
68 return ENCODED_LENGTH_2.pack(majortype << 5 | 24, length) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
69 elif length < 65536: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
70 return ENCODED_LENGTH_3.pack(majortype << 5 | 25, length) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
71 elif length < 4294967296: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
72 return ENCODED_LENGTH_4.pack(majortype << 5 | 26, length) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
73 else: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
74 return ENCODED_LENGTH_5.pack(majortype << 5 | 27, length) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
75 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
76 def streamencodebytestring(v): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
77 yield encodelength(MAJOR_TYPE_BYTESTRING, len(v)) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
78 yield v |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
79 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
80 def streamencodebytestringfromiter(it): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
81 """Convert an iterator of chunks to an indefinite bytestring. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
82 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
83 Given an input that is iterable and each element in the iterator is |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
84 representable as bytes, emit an indefinite length bytestring. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
85 """ |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
86 yield BEGIN_INDEFINITE_BYTESTRING |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
87 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
88 for chunk in it: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
89 yield encodelength(MAJOR_TYPE_BYTESTRING, len(chunk)) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
90 yield chunk |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
91 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
92 yield BREAK |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
93 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
94 def streamencodeindefinitebytestring(source, chunksize=65536): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
95 """Given a large source buffer, emit as an indefinite length bytestring. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
96 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
97 This is a generator of chunks constituting the encoded CBOR data. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
98 """ |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
99 yield BEGIN_INDEFINITE_BYTESTRING |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
100 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
101 i = 0 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
102 l = len(source) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
103 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
104 while True: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
105 chunk = source[i:i + chunksize] |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
106 i += len(chunk) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
107 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
108 yield encodelength(MAJOR_TYPE_BYTESTRING, len(chunk)) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
109 yield chunk |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
110 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
111 if i >= l: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
112 break |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
113 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
114 yield BREAK |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
115 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
116 def streamencodeint(v): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
117 if v >= 18446744073709551616 or v < -18446744073709551616: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
118 raise ValueError('big integers not supported') |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
119 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
120 if v >= 0: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
121 yield encodelength(MAJOR_TYPE_UINT, v) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
122 else: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
123 yield encodelength(MAJOR_TYPE_NEGINT, abs(v) - 1) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
124 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
125 def streamencodearray(l): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
126 """Encode a known size iterable to an array.""" |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
127 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
128 yield encodelength(MAJOR_TYPE_ARRAY, len(l)) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
129 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
130 for i in l: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
131 for chunk in streamencode(i): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
132 yield chunk |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
133 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
134 def streamencodearrayfromiter(it): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
135 """Encode an iterator of items to an indefinite length array.""" |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
136 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
137 yield BEGIN_INDEFINITE_ARRAY |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
138 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
139 for i in it: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
140 for chunk in streamencode(i): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
141 yield chunk |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
142 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
143 yield BREAK |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
144 |
37942
2ae6a3134362
cborutil: port to Python 3
Augie Fackler <augie@google.com>
parents:
37711
diff
changeset
|
145 def _mixedtypesortkey(v): |
2ae6a3134362
cborutil: port to Python 3
Augie Fackler <augie@google.com>
parents:
37711
diff
changeset
|
146 return type(v).__name__, v |
2ae6a3134362
cborutil: port to Python 3
Augie Fackler <augie@google.com>
parents:
37711
diff
changeset
|
147 |
37711
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
148 def streamencodeset(s): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
149 # https://www.iana.org/assignments/cbor-tags/cbor-tags.xhtml defines |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
150 # semantic tag 258 for finite sets. |
39438
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
151 yield encodelength(MAJOR_TYPE_SEMANTIC, SEMANTIC_TAG_FINITE_SET) |
37711
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
152 |
37942
2ae6a3134362
cborutil: port to Python 3
Augie Fackler <augie@google.com>
parents:
37711
diff
changeset
|
153 for chunk in streamencodearray(sorted(s, key=_mixedtypesortkey)): |
37711
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
154 yield chunk |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
155 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
156 def streamencodemap(d): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
157 """Encode dictionary to a generator. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
158 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
159 Does not supporting indefinite length dictionaries. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
160 """ |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
161 yield encodelength(MAJOR_TYPE_MAP, len(d)) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
162 |
37942
2ae6a3134362
cborutil: port to Python 3
Augie Fackler <augie@google.com>
parents:
37711
diff
changeset
|
163 for key, value in sorted(d.iteritems(), |
2ae6a3134362
cborutil: port to Python 3
Augie Fackler <augie@google.com>
parents:
37711
diff
changeset
|
164 key=lambda x: _mixedtypesortkey(x[0])): |
37711
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
165 for chunk in streamencode(key): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
166 yield chunk |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
167 for chunk in streamencode(value): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
168 yield chunk |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
169 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
170 def streamencodemapfromiter(it): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
171 """Given an iterable of (key, value), encode to an indefinite length map.""" |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
172 yield BEGIN_INDEFINITE_MAP |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
173 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
174 for key, value in it: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
175 for chunk in streamencode(key): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
176 yield chunk |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
177 for chunk in streamencode(value): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
178 yield chunk |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
179 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
180 yield BREAK |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
181 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
182 def streamencodebool(b): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
183 # major type 7, simple value 20 and 21. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
184 yield b'\xf5' if b else b'\xf4' |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
185 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
186 def streamencodenone(v): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
187 # major type 7, simple value 22. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
188 yield b'\xf6' |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
189 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
190 STREAM_ENCODERS = { |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
191 bytes: streamencodebytestring, |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
192 int: streamencodeint, |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
193 list: streamencodearray, |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
194 tuple: streamencodearray, |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
195 dict: streamencodemap, |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
196 set: streamencodeset, |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
197 bool: streamencodebool, |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
198 type(None): streamencodenone, |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
199 } |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
200 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
201 def streamencode(v): |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
202 """Encode a value in a streaming manner. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
203 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
204 Given an input object, encode it to CBOR recursively. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
205 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
206 Returns a generator of CBOR encoded bytes. There is no guarantee |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
207 that each emitted chunk fully decodes to a value or sub-value. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
208 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
209 Encoding is deterministic - unordered collections are sorted. |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
210 """ |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
211 fn = STREAM_ENCODERS.get(v.__class__) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
212 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
213 if not fn: |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
214 raise ValueError('do not know how to encode %s' % type(v)) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
215 |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
216 return fn(v) |
65a23cc8e75b
cborutil: implement support for streaming encoding, bytestring decoding
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
217 |
39438
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
218 class CBORDecodeError(Exception): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
219 """Represents an error decoding CBOR.""" |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
220 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
221 if sys.version_info.major >= 3: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
222 def _elementtointeger(b, i): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
223 return b[i] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
224 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
225 def _elementtointeger(b, i): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
226 return ord(b[i]) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
227 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
228 STRUCT_BIG_UBYTE = struct.Struct(r'>B') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
229 STRUCT_BIG_USHORT = struct.Struct('>H') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
230 STRUCT_BIG_ULONG = struct.Struct('>L') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
231 STRUCT_BIG_ULONGLONG = struct.Struct('>Q') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
232 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
233 SPECIAL_NONE = 0 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
234 SPECIAL_START_INDEFINITE_BYTESTRING = 1 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
235 SPECIAL_START_ARRAY = 2 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
236 SPECIAL_START_MAP = 3 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
237 SPECIAL_START_SET = 4 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
238 SPECIAL_INDEFINITE_BREAK = 5 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
239 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
240 def decodeitem(b, offset=0): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
241 """Decode a new CBOR value from a buffer at offset. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
242 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
243 This function attempts to decode up to one complete CBOR value |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
244 from ``b`` starting at offset ``offset``. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
245 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
246 The beginning of a collection (such as an array, map, set, or |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
247 indefinite length bytestring) counts as a single value. For these |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
248 special cases, a state flag will indicate that a special value was seen. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
249 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
250 When called, the function either returns a decoded value or gives |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
251 a hint as to how many more bytes are needed to do so. By calling |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
252 the function repeatedly given a stream of bytes, the caller can |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
253 build up the original values. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
254 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
255 Returns a tuple with the following elements: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
256 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
257 * Bool indicating whether a complete value was decoded. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
258 * A decoded value if first value is True otherwise None |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
259 * Integer number of bytes. If positive, the number of bytes |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
260 read. If negative, the number of bytes we need to read to |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
261 decode this value or the next chunk in this value. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
262 * One of the ``SPECIAL_*`` constants indicating special treatment |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
263 for this value. ``SPECIAL_NONE`` means this is a fully decoded |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
264 simple value (such as an integer or bool). |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
265 """ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
266 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
267 initial = _elementtointeger(b, offset) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
268 offset += 1 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
269 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
270 majortype = initial >> 5 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
271 subtype = initial & SUBTYPE_MASK |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
272 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
273 if majortype == MAJOR_TYPE_UINT: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
274 complete, value, readcount = decodeuint(subtype, b, offset) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
275 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
276 if complete: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
277 return True, value, readcount + 1, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
278 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
279 return False, None, readcount, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
280 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
281 elif majortype == MAJOR_TYPE_NEGINT: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
282 # Negative integers are the same as UINT except inverted minus 1. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
283 complete, value, readcount = decodeuint(subtype, b, offset) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
284 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
285 if complete: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
286 return True, -value - 1, readcount + 1, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
287 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
288 return False, None, readcount, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
289 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
290 elif majortype == MAJOR_TYPE_BYTESTRING: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
291 # Beginning of bytestrings are treated as uints in order to |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
292 # decode their length, which may be indefinite. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
293 complete, size, readcount = decodeuint(subtype, b, offset, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
294 allowindefinite=True) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
295 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
296 # We don't know the size of the bytestring. It must be a definitive |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
297 # length since the indefinite subtype would be encoded in the initial |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
298 # byte. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
299 if not complete: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
300 return False, None, readcount, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
301 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
302 # We know the length of the bytestring. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
303 if size is not None: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
304 # And the data is available in the buffer. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
305 if offset + readcount + size <= len(b): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
306 value = b[offset + readcount:offset + readcount + size] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
307 return True, value, readcount + size + 1, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
308 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
309 # And we need more data in order to return the bytestring. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
310 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
311 wanted = len(b) - offset - readcount - size |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
312 return False, None, wanted, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
313 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
314 # It is an indefinite length bytestring. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
315 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
316 return True, None, 1, SPECIAL_START_INDEFINITE_BYTESTRING |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
317 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
318 elif majortype == MAJOR_TYPE_STRING: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
319 raise CBORDecodeError('string major type not supported') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
320 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
321 elif majortype == MAJOR_TYPE_ARRAY: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
322 # Beginning of arrays are treated as uints in order to decode their |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
323 # length. We don't allow indefinite length arrays. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
324 complete, size, readcount = decodeuint(subtype, b, offset) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
325 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
326 if complete: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
327 return True, size, readcount + 1, SPECIAL_START_ARRAY |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
328 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
329 return False, None, readcount, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
330 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
331 elif majortype == MAJOR_TYPE_MAP: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
332 # Beginning of maps are treated as uints in order to decode their |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
333 # number of elements. We don't allow indefinite length arrays. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
334 complete, size, readcount = decodeuint(subtype, b, offset) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
335 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
336 if complete: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
337 return True, size, readcount + 1, SPECIAL_START_MAP |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
338 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
339 return False, None, readcount, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
340 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
341 elif majortype == MAJOR_TYPE_SEMANTIC: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
342 # Semantic tag value is read the same as a uint. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
343 complete, tagvalue, readcount = decodeuint(subtype, b, offset) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
344 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
345 if not complete: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
346 return False, None, readcount, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
347 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
348 # This behavior here is a little wonky. The main type being "decorated" |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
349 # by this semantic tag follows. A more robust parser would probably emit |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
350 # a special flag indicating this as a semantic tag and let the caller |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
351 # deal with the types that follow. But since we don't support many |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
352 # semantic tags, it is easier to deal with the special cases here and |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
353 # hide complexity from the caller. If we add support for more semantic |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
354 # tags, we should probably move semantic tag handling into the caller. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
355 if tagvalue == SEMANTIC_TAG_FINITE_SET: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
356 if offset + readcount >= len(b): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
357 return False, None, -1, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
358 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
359 complete, size, readcount2, special = decodeitem(b, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
360 offset + readcount) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
361 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
362 if not complete: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
363 return False, None, readcount2, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
364 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
365 if special != SPECIAL_START_ARRAY: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
366 raise CBORDecodeError('expected array after finite set ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
367 'semantic tag') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
368 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
369 return True, size, readcount + readcount2 + 1, SPECIAL_START_SET |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
370 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
371 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
372 raise CBORDecodeError('semantic tag %d not allowed' % tagvalue) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
373 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
374 elif majortype == MAJOR_TYPE_SPECIAL: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
375 # Only specific values for the information field are allowed. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
376 if subtype == SUBTYPE_FALSE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
377 return True, False, 1, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
378 elif subtype == SUBTYPE_TRUE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
379 return True, True, 1, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
380 elif subtype == SUBTYPE_NULL: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
381 return True, None, 1, SPECIAL_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
382 elif subtype == SUBTYPE_INDEFINITE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
383 return True, None, 1, SPECIAL_INDEFINITE_BREAK |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
384 # If value is 24, subtype is in next byte. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
385 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
386 raise CBORDecodeError('special type %d not allowed' % subtype) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
387 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
388 assert False |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
389 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
390 def decodeuint(subtype, b, offset=0, allowindefinite=False): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
391 """Decode an unsigned integer. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
392 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
393 ``subtype`` is the lower 5 bits from the initial byte CBOR item |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
394 "header." ``b`` is a buffer containing bytes. ``offset`` points to |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
395 the index of the first byte after the byte that ``subtype`` was |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
396 derived from. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
397 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
398 ``allowindefinite`` allows the special indefinite length value |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
399 indicator. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
400 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
401 Returns a 3-tuple of (successful, value, count). |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
402 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
403 The first element is a bool indicating if decoding completed. The 2nd |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
404 is the decoded integer value or None if not fully decoded or the subtype |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
405 is 31 and ``allowindefinite`` is True. The 3rd value is the count of bytes. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
406 If positive, it is the number of additional bytes decoded. If negative, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
407 it is the number of additional bytes needed to decode this value. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
408 """ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
409 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
410 # Small values are inline. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
411 if subtype < 24: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
412 return True, subtype, 0 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
413 # Indefinite length specifier. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
414 elif subtype == 31: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
415 if allowindefinite: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
416 return True, None, 0 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
417 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
418 raise CBORDecodeError('indefinite length uint not allowed here') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
419 elif subtype >= 28: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
420 raise CBORDecodeError('unsupported subtype on integer type: %d' % |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
421 subtype) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
422 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
423 if subtype == 24: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
424 s = STRUCT_BIG_UBYTE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
425 elif subtype == 25: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
426 s = STRUCT_BIG_USHORT |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
427 elif subtype == 26: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
428 s = STRUCT_BIG_ULONG |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
429 elif subtype == 27: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
430 s = STRUCT_BIG_ULONGLONG |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
431 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
432 raise CBORDecodeError('bounds condition checking violation') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
433 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
434 if len(b) - offset >= s.size: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
435 return True, s.unpack_from(b, offset)[0], s.size |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
436 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
437 return False, None, len(b) - offset - s.size |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
438 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
439 class bytestringchunk(bytes): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
440 """Represents a chunk/segment in an indefinite length bytestring. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
441 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
442 This behaves like a ``bytes`` but in addition has the ``isfirst`` |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
443 and ``islast`` attributes indicating whether this chunk is the first |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
444 or last in an indefinite length bytestring. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
445 """ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
446 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
447 def __new__(cls, v, first=False, last=False): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
448 self = bytes.__new__(cls, v) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
449 self.isfirst = first |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
450 self.islast = last |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
451 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
452 return self |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
453 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
454 class sansiodecoder(object): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
455 """A CBOR decoder that doesn't perform its own I/O. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
456 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
457 To use, construct an instance and feed it segments containing |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
458 CBOR-encoded bytes via ``decode()``. The return value from ``decode()`` |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
459 indicates whether a fully-decoded value is available, how many bytes |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
460 were consumed, and offers a hint as to how many bytes should be fed |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
461 in next time to decode the next value. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
462 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
463 The decoder assumes it will decode N discrete CBOR values, not just |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
464 a single value. i.e. if the bytestream contains uints packed one after |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
465 the other, the decoder will decode them all, rather than just the initial |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
466 one. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
467 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
468 When ``decode()`` indicates a value is available, call ``getavailable()`` |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
469 to return all fully decoded values. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
470 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
471 ``decode()`` can partially decode input. It is up to the caller to keep |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
472 track of what data was consumed and to pass unconsumed data in on the |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
473 next invocation. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
474 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
475 The decoder decodes atomically at the *item* level. See ``decodeitem()``. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
476 If an *item* cannot be fully decoded, the decoder won't record it as |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
477 partially consumed. Instead, the caller will be instructed to pass in |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
478 the initial bytes of this item on the next invocation. This does result |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
479 in some redundant parsing. But the overhead should be minimal. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
480 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
481 This decoder only supports a subset of CBOR as required by Mercurial. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
482 It lacks support for: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
483 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
484 * Indefinite length arrays |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
485 * Indefinite length maps |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
486 * Use of indefinite length bytestrings as keys or values within |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
487 arrays, maps, or sets. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
488 * Nested arrays, maps, or sets within sets |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
489 * Any semantic tag that isn't a mathematical finite set |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
490 * Floating point numbers |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
491 * Undefined special value |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
492 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
493 CBOR types are decoded to Python types as follows: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
494 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
495 uint -> int |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
496 negint -> int |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
497 bytestring -> bytes |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
498 map -> dict |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
499 array -> list |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
500 True -> bool |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
501 False -> bool |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
502 null -> None |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
503 indefinite length bytestring chunk -> [bytestringchunk] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
504 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
505 The only non-obvious mapping here is an indefinite length bytestring |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
506 to the ``bytestringchunk`` type. This is to facilitate streaming |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
507 indefinite length bytestrings out of the decoder and to differentiate |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
508 a regular bytestring from an indefinite length bytestring. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
509 """ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
510 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
511 _STATE_NONE = 0 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
512 _STATE_WANT_MAP_KEY = 1 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
513 _STATE_WANT_MAP_VALUE = 2 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
514 _STATE_WANT_ARRAY_VALUE = 3 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
515 _STATE_WANT_SET_VALUE = 4 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
516 _STATE_WANT_BYTESTRING_CHUNK_FIRST = 5 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
517 _STATE_WANT_BYTESTRING_CHUNK_SUBSEQUENT = 6 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
518 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
519 def __init__(self): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
520 # TODO add support for limiting size of bytestrings |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
521 # TODO add support for limiting number of keys / values in collections |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
522 # TODO add support for limiting size of buffered partial values |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
523 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
524 self.decodedbytecount = 0 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
525 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
526 self._state = self._STATE_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
527 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
528 # Stack of active nested collections. Each entry is a dict describing |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
529 # the collection. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
530 self._collectionstack = [] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
531 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
532 # Fully decoded key to use for the current map. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
533 self._currentmapkey = None |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
534 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
535 # Fully decoded values available for retrieval. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
536 self._decodedvalues = [] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
537 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
538 @property |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
539 def inprogress(self): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
540 """Whether the decoder has partially decoded a value.""" |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
541 return self._state != self._STATE_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
542 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
543 def decode(self, b, offset=0): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
544 """Attempt to decode bytes from an input buffer. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
545 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
546 ``b`` is a collection of bytes and ``offset`` is the byte |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
547 offset within that buffer from which to begin reading data. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
548 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
549 ``b`` must support ``len()`` and accessing bytes slices via |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
550 ``__slice__``. Typically ``bytes`` instances are used. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
551 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
552 Returns a tuple with the following fields: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
553 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
554 * Bool indicating whether values are available for retrieval. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
555 * Integer indicating the number of bytes that were fully consumed, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
556 starting from ``offset``. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
557 * Integer indicating the number of bytes that are desired for the |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
558 next call in order to decode an item. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
559 """ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
560 if not b: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
561 return bool(self._decodedvalues), 0, 0 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
562 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
563 initialoffset = offset |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
564 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
565 # We could easily split the body of this loop into a function. But |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
566 # Python performance is sensitive to function calls and collections |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
567 # are composed of many items. So leaving as a while loop could help |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
568 # with performance. One thing that may not help is the use of |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
569 # if..elif versus a lookup/dispatch table. There may be value |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
570 # in switching that. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
571 while offset < len(b): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
572 # Attempt to decode an item. This could be a whole value or a |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
573 # special value indicating an event, such as start or end of a |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
574 # collection or indefinite length type. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
575 complete, value, readcount, special = decodeitem(b, offset) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
576 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
577 if readcount > 0: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
578 self.decodedbytecount += readcount |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
579 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
580 if not complete: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
581 assert readcount < 0 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
582 return ( |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
583 bool(self._decodedvalues), |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
584 offset - initialoffset, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
585 -readcount, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
586 ) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
587 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
588 offset += readcount |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
589 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
590 # No nested state. We either have a full value or beginning of a |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
591 # complex value to deal with. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
592 if self._state == self._STATE_NONE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
593 # A normal value. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
594 if special == SPECIAL_NONE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
595 self._decodedvalues.append(value) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
596 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
597 elif special == SPECIAL_START_ARRAY: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
598 self._collectionstack.append({ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
599 'remaining': value, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
600 'v': [], |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
601 }) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
602 self._state = self._STATE_WANT_ARRAY_VALUE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
603 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
604 elif special == SPECIAL_START_MAP: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
605 self._collectionstack.append({ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
606 'remaining': value, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
607 'v': {}, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
608 }) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
609 self._state = self._STATE_WANT_MAP_KEY |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
610 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
611 elif special == SPECIAL_START_SET: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
612 self._collectionstack.append({ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
613 'remaining': value, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
614 'v': set(), |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
615 }) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
616 self._state = self._STATE_WANT_SET_VALUE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
617 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
618 elif special == SPECIAL_START_INDEFINITE_BYTESTRING: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
619 self._state = self._STATE_WANT_BYTESTRING_CHUNK_FIRST |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
620 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
621 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
622 raise CBORDecodeError('unhandled special state: %d' % |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
623 special) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
624 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
625 # This value becomes an element of the current array. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
626 elif self._state == self._STATE_WANT_ARRAY_VALUE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
627 # Simple values get appended. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
628 if special == SPECIAL_NONE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
629 c = self._collectionstack[-1] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
630 c['v'].append(value) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
631 c['remaining'] -= 1 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
632 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
633 # self._state doesn't need changed. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
634 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
635 # An array nested within an array. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
636 elif special == SPECIAL_START_ARRAY: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
637 lastc = self._collectionstack[-1] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
638 newvalue = [] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
639 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
640 lastc['v'].append(newvalue) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
641 lastc['remaining'] -= 1 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
642 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
643 self._collectionstack.append({ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
644 'remaining': value, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
645 'v': newvalue, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
646 }) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
647 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
648 # self._state doesn't need changed. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
649 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
650 # A map nested within an array. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
651 elif special == SPECIAL_START_MAP: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
652 lastc = self._collectionstack[-1] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
653 newvalue = {} |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
654 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
655 lastc['v'].append(newvalue) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
656 lastc['remaining'] -= 1 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
657 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
658 self._collectionstack.append({ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
659 'remaining': value, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
660 'v': newvalue |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
661 }) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
662 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
663 self._state = self._STATE_WANT_MAP_KEY |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
664 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
665 elif special == SPECIAL_START_SET: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
666 lastc = self._collectionstack[-1] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
667 newvalue = set() |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
668 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
669 lastc['v'].append(newvalue) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
670 lastc['remaining'] -= 1 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
671 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
672 self._collectionstack.append({ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
673 'remaining': value, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
674 'v': newvalue, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
675 }) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
676 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
677 self._state = self._STATE_WANT_SET_VALUE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
678 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
679 elif special == SPECIAL_START_INDEFINITE_BYTESTRING: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
680 raise CBORDecodeError('indefinite length bytestrings ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
681 'not allowed as array values') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
682 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
683 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
684 raise CBORDecodeError('unhandled special item when ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
685 'expecting array value: %d' % special) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
686 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
687 # This value becomes the key of the current map instance. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
688 elif self._state == self._STATE_WANT_MAP_KEY: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
689 if special == SPECIAL_NONE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
690 self._currentmapkey = value |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
691 self._state = self._STATE_WANT_MAP_VALUE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
692 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
693 elif special == SPECIAL_START_INDEFINITE_BYTESTRING: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
694 raise CBORDecodeError('indefinite length bytestrings ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
695 'not allowed as map keys') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
696 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
697 elif special in (SPECIAL_START_ARRAY, SPECIAL_START_MAP, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
698 SPECIAL_START_SET): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
699 raise CBORDecodeError('collections not supported as map ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
700 'keys') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
701 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
702 # We do not allow special values to be used as map keys. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
703 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
704 raise CBORDecodeError('unhandled special item when ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
705 'expecting map key: %d' % special) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
706 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
707 # This value becomes the value of the current map key. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
708 elif self._state == self._STATE_WANT_MAP_VALUE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
709 # Simple values simply get inserted into the map. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
710 if special == SPECIAL_NONE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
711 lastc = self._collectionstack[-1] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
712 lastc['v'][self._currentmapkey] = value |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
713 lastc['remaining'] -= 1 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
714 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
715 self._state = self._STATE_WANT_MAP_KEY |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
716 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
717 # A new array is used as the map value. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
718 elif special == SPECIAL_START_ARRAY: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
719 lastc = self._collectionstack[-1] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
720 newvalue = [] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
721 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
722 lastc['v'][self._currentmapkey] = newvalue |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
723 lastc['remaining'] -= 1 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
724 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
725 self._collectionstack.append({ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
726 'remaining': value, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
727 'v': newvalue, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
728 }) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
729 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
730 self._state = self._STATE_WANT_ARRAY_VALUE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
731 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
732 # A new map is used as the map value. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
733 elif special == SPECIAL_START_MAP: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
734 lastc = self._collectionstack[-1] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
735 newvalue = {} |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
736 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
737 lastc['v'][self._currentmapkey] = newvalue |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
738 lastc['remaining'] -= 1 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
739 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
740 self._collectionstack.append({ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
741 'remaining': value, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
742 'v': newvalue, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
743 }) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
744 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
745 self._state = self._STATE_WANT_MAP_KEY |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
746 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
747 # A new set is used as the map value. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
748 elif special == SPECIAL_START_SET: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
749 lastc = self._collectionstack[-1] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
750 newvalue = set() |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
751 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
752 lastc['v'][self._currentmapkey] = newvalue |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
753 lastc['remaining'] -= 1 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
754 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
755 self._collectionstack.append({ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
756 'remaining': value, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
757 'v': newvalue, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
758 }) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
759 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
760 self._state = self._STATE_WANT_SET_VALUE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
761 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
762 elif special == SPECIAL_START_INDEFINITE_BYTESTRING: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
763 raise CBORDecodeError('indefinite length bytestrings not ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
764 'allowed as map values') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
765 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
766 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
767 raise CBORDecodeError('unhandled special item when ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
768 'expecting map value: %d' % special) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
769 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
770 self._currentmapkey = None |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
771 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
772 # This value is added to the current set. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
773 elif self._state == self._STATE_WANT_SET_VALUE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
774 if special == SPECIAL_NONE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
775 lastc = self._collectionstack[-1] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
776 lastc['v'].add(value) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
777 lastc['remaining'] -= 1 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
778 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
779 elif special == SPECIAL_START_INDEFINITE_BYTESTRING: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
780 raise CBORDecodeError('indefinite length bytestrings not ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
781 'allowed as set values') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
782 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
783 elif special in (SPECIAL_START_ARRAY, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
784 SPECIAL_START_MAP, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
785 SPECIAL_START_SET): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
786 raise CBORDecodeError('collections not allowed as set ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
787 'values') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
788 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
789 # We don't allow non-trivial types to exist as set values. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
790 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
791 raise CBORDecodeError('unhandled special item when ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
792 'expecting set value: %d' % special) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
793 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
794 # This value represents the first chunk in an indefinite length |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
795 # bytestring. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
796 elif self._state == self._STATE_WANT_BYTESTRING_CHUNK_FIRST: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
797 # We received a full chunk. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
798 if special == SPECIAL_NONE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
799 self._decodedvalues.append(bytestringchunk(value, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
800 first=True)) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
801 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
802 self._state = self._STATE_WANT_BYTESTRING_CHUNK_SUBSEQUENT |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
803 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
804 # The end of stream marker. This means it is an empty |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
805 # indefinite length bytestring. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
806 elif special == SPECIAL_INDEFINITE_BREAK: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
807 # We /could/ convert this to a b''. But we want to preserve |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
808 # the nature of the underlying data so consumers expecting |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
809 # an indefinite length bytestring get one. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
810 self._decodedvalues.append(bytestringchunk(b'', |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
811 first=True, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
812 last=True)) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
813 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
814 # Since indefinite length bytestrings can't be used in |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
815 # collections, we must be at the root level. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
816 assert not self._collectionstack |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
817 self._state = self._STATE_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
818 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
819 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
820 raise CBORDecodeError('unexpected special value when ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
821 'expecting bytestring chunk: %d' % |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
822 special) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
823 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
824 # This value represents the non-initial chunk in an indefinite |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
825 # length bytestring. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
826 elif self._state == self._STATE_WANT_BYTESTRING_CHUNK_SUBSEQUENT: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
827 # We received a full chunk. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
828 if special == SPECIAL_NONE: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
829 self._decodedvalues.append(bytestringchunk(value)) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
830 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
831 # The end of stream marker. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
832 elif special == SPECIAL_INDEFINITE_BREAK: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
833 self._decodedvalues.append(bytestringchunk(b'', last=True)) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
834 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
835 # Since indefinite length bytestrings can't be used in |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
836 # collections, we must be at the root level. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
837 assert not self._collectionstack |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
838 self._state = self._STATE_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
839 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
840 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
841 raise CBORDecodeError('unexpected special value when ' |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
842 'expecting bytestring chunk: %d' % |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
843 special) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
844 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
845 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
846 raise CBORDecodeError('unhandled decoder state: %d' % |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
847 self._state) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
848 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
849 # We could have just added the final value in a collection. End |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
850 # all complete collections at the top of the stack. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
851 while True: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
852 # Bail if we're not waiting on a new collection item. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
853 if self._state not in (self._STATE_WANT_ARRAY_VALUE, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
854 self._STATE_WANT_MAP_KEY, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
855 self._STATE_WANT_SET_VALUE): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
856 break |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
857 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
858 # Or we are expecting more items for this collection. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
859 lastc = self._collectionstack[-1] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
860 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
861 if lastc['remaining']: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
862 break |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
863 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
864 # The collection at the top of the stack is complete. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
865 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
866 # Discard it, as it isn't needed for future items. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
867 self._collectionstack.pop() |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
868 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
869 # If this is a nested collection, we don't emit it, since it |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
870 # will be emitted by its parent collection. But we do need to |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
871 # update state to reflect what the new top-most collection |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
872 # on the stack is. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
873 if self._collectionstack: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
874 self._state = { |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
875 list: self._STATE_WANT_ARRAY_VALUE, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
876 dict: self._STATE_WANT_MAP_KEY, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
877 set: self._STATE_WANT_SET_VALUE, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
878 }[type(self._collectionstack[-1]['v'])] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
879 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
880 # If this is the root collection, emit it. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
881 else: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
882 self._decodedvalues.append(lastc['v']) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
883 self._state = self._STATE_NONE |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
884 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
885 return ( |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
886 bool(self._decodedvalues), |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
887 offset - initialoffset, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
888 0, |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
889 ) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
890 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
891 def getavailable(self): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
892 """Returns an iterator over fully decoded values. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
893 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
894 Once values are retrieved, they won't be available on the next call. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
895 """ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
896 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
897 l = list(self._decodedvalues) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
898 self._decodedvalues = [] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
899 return l |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
900 |
39440
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
901 class bufferingdecoder(object): |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
902 """A CBOR decoder that buffers undecoded input. |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
903 |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
904 This is a glorified wrapper around ``sansiodecoder`` that adds a buffering |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
905 layer. All input that isn't consumed by ``sansiodecoder`` will be buffered |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
906 and concatenated with any new input that arrives later. |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
907 |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
908 TODO consider adding limits as to the maximum amount of data that can |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
909 be buffered. |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
910 """ |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
911 def __init__(self): |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
912 self._decoder = sansiodecoder() |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
913 self._leftover = None |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
914 |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
915 def decode(self, b): |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
916 """Attempt to decode bytes to CBOR values. |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
917 |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
918 Returns a tuple with the following fields: |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
919 |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
920 * Bool indicating whether new values are available for retrieval. |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
921 * Integer number of bytes decoded from the new input. |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
922 * Integer number of bytes wanted to decode the next value. |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
923 """ |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
924 |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
925 if self._leftover: |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
926 oldlen = len(self._leftover) |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
927 b = self._leftover + b |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
928 self._leftover = None |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
929 else: |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
930 b = b |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
931 oldlen = 0 |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
932 |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
933 available, readcount, wanted = self._decoder.decode(b) |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
934 |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
935 if readcount < len(b): |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
936 self._leftover = b[readcount:] |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
937 |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
938 return available, readcount - oldlen, wanted |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
939 |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
940 def getavailable(self): |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
941 return self._decoder.getavailable() |
babad5ebaf0a
cborutil: add a buffering decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
39439
diff
changeset
|
942 |
39438
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
943 def decodeall(b): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
944 """Decode all CBOR items present in an iterable of bytes. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
945 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
946 In addition to regular decode errors, raises CBORDecodeError if the |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
947 entirety of the passed buffer does not fully decode to complete CBOR |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
948 values. This includes failure to decode any value, incomplete collection |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
949 types, incomplete indefinite length items, and extra data at the end of |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
950 the buffer. |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
951 """ |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
952 if not b: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
953 return [] |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
954 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
955 decoder = sansiodecoder() |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
956 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
957 havevalues, readcount, wantbytes = decoder.decode(b) |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
958 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
959 if readcount != len(b): |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
960 raise CBORDecodeError('input data not fully consumed') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
961 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
962 if decoder.inprogress: |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
963 raise CBORDecodeError('input data not complete') |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
964 |
aeb551a3bb8a
cborutil: implement sans I/O decoder
Gregory Szorc <gregory.szorc@gmail.com>
parents:
37942
diff
changeset
|
965 return decoder.getavailable() |