| temporal | 40ee551 | 2008-07-10 02:12:20 +0000 | [diff] [blame] | 1 | # Protocol Buffers - Google's data interchange format |
| 2 | # Copyright 2008 Google Inc. |
| 3 | # http://code.google.com/p/protobuf/ |
| 4 | # |
| 5 | # Licensed under the Apache License, Version 2.0 (the "License"); |
| 6 | # you may not use this file except in compliance with the License. |
| 7 | # You may obtain a copy of the License at |
| 8 | # |
| 9 | # http://www.apache.org/licenses/LICENSE-2.0 |
| 10 | # |
| 11 | # Unless required by applicable law or agreed to in writing, software |
| 12 | # distributed under the License is distributed on an "AS IS" BASIS, |
| 13 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 14 | # See the License for the specific language governing permissions and |
| 15 | # limitations under the License. |
| 16 | |
| 17 | # TODO(robinson): We should just make these methods all "pure-virtual" and move |
| 18 | # all implementation out, into reflection.py for now. |
| 19 | |
| 20 | |
| 21 | """Contains an abstract base class for protocol messages.""" |
| 22 | |
| 23 | __author__ = 'robinson@google.com (Will Robinson)' |
| 24 | |
| 25 | from google.protobuf import text_format |
| 26 | |
| 27 | class Error(Exception): pass |
| 28 | class DecodeError(Error): pass |
| 29 | class EncodeError(Error): pass |
| 30 | |
| 31 | |
| 32 | class Message(object): |
| 33 | |
| 34 | """Abstract base class for protocol messages. |
| 35 | |
| 36 | Protocol message classes are almost always generated by the protocol |
| 37 | compiler. These generated types subclass Message and implement the methods |
| 38 | shown below. |
| 39 | |
| 40 | TODO(robinson): Link to an HTML document here. |
| 41 | |
| 42 | TODO(robinson): Document that instances of this class will also |
| 43 | have an Extensions attribute with __getitem__ and __setitem__. |
| 44 | Again, not sure how to best convey this. |
| 45 | |
| 46 | TODO(robinson): Document that the class must also have a static |
| 47 | RegisterExtension(extension_field) method. |
| 48 | Not sure how to best express at this point. |
| 49 | """ |
| 50 | |
| 51 | # TODO(robinson): Document these fields and methods. |
| 52 | |
| 53 | __slots__ = [] |
| 54 | |
| 55 | DESCRIPTOR = None |
| 56 | |
| 57 | def __eq__(self, other_msg): |
| 58 | raise NotImplementedError |
| 59 | |
| 60 | def __ne__(self, other_msg): |
| 61 | # Can't just say self != other_msg, since that would infinitely recurse. :) |
| 62 | return not self == other_msg |
| 63 | |
| 64 | def __str__(self): |
| 65 | return text_format.MessageToString(self) |
| 66 | |
| 67 | def MergeFrom(self, other_msg): |
| 68 | raise NotImplementedError |
| 69 | |
| 70 | def CopyFrom(self, other_msg): |
| 71 | raise NotImplementedError |
| 72 | |
| 73 | def Clear(self): |
| 74 | raise NotImplementedError |
| 75 | |
| 76 | def IsInitialized(self): |
| 77 | raise NotImplementedError |
| 78 | |
| 79 | # TODO(robinson): MergeFromString() should probably return None and be |
| 80 | # implemented in terms of a helper that returns the # of bytes read. Our |
| 81 | # deserialization routines would use the helper when recursively |
| 82 | # deserializing, but the end user would almost always just want the no-return |
| 83 | # MergeFromString(). |
| 84 | |
| 85 | def MergeFromString(self, serialized): |
| 86 | """Merges serialized protocol buffer data into this message. |
| 87 | |
| 88 | When we find a field in |serialized| that is already present |
| 89 | in this message: |
| 90 | - If it's a "repeated" field, we append to the end of our list. |
| 91 | - Else, if it's a scalar, we overwrite our field. |
| 92 | - Else, (it's a nonrepeated composite), we recursively merge |
| 93 | into the existing composite. |
| 94 | |
| 95 | TODO(robinson): Document handling of unknown fields. |
| 96 | |
| 97 | Args: |
| 98 | serialized: Any object that allows us to call buffer(serialized) |
| 99 | to access a string of bytes using the buffer interface. |
| 100 | |
| 101 | TODO(robinson): When we switch to a helper, this will return None. |
| 102 | |
| 103 | Returns: |
| 104 | The number of bytes read from |serialized|. |
| 105 | For non-group messages, this will always be len(serialized), |
| 106 | but for messages which are actually groups, this will |
| 107 | generally be less than len(serialized), since we must |
| 108 | stop when we reach an END_GROUP tag. Note that if |
| 109 | we *do* stop because of an END_GROUP tag, the number |
| 110 | of bytes returned does not include the bytes |
| 111 | for the END_GROUP tag information. |
| 112 | """ |
| 113 | raise NotImplementedError |
| 114 | |
| 115 | def ParseFromString(self, serialized): |
| 116 | """Like MergeFromString(), except we clear the object first.""" |
| 117 | self.Clear() |
| 118 | self.MergeFromString(serialized) |
| 119 | |
| 120 | def SerializeToString(self): |
| 121 | raise NotImplementedError |
| 122 | |
| 123 | # TODO(robinson): Decide whether we like these better |
| 124 | # than auto-generated has_foo() and clear_foo() methods |
| 125 | # on the instances themselves. This way is less consistent |
| 126 | # with C++, but it makes reflection-type access easier and |
| 127 | # reduces the number of magically autogenerated things. |
| 128 | # |
| 129 | # TODO(robinson): Be sure to document (and test) exactly |
| 130 | # which field names are accepted here. Are we case-sensitive? |
| 131 | # What do we do with fields that share names with Python keywords |
| 132 | # like 'lambda' and 'yield'? |
| 133 | # |
| 134 | # nnorwitz says: |
| 135 | # """ |
| 136 | # Typically (in python), an underscore is appended to names that are |
| 137 | # keywords. So they would become lambda_ or yield_. |
| 138 | # """ |
| 139 | def ListFields(self, field_name): |
| 140 | """Returns a list of (FieldDescriptor, value) tuples for all |
| 141 | fields in the message which are not empty. A singular field is non-empty |
| 142 | if HasField() would return true, and a repeated field is non-empty if |
| 143 | it contains at least one element. The fields are ordered by field |
| 144 | number""" |
| 145 | raise NotImplementedError |
| 146 | |
| 147 | def HasField(self, field_name): |
| 148 | raise NotImplementedError |
| 149 | |
| 150 | def ClearField(self, field_name): |
| 151 | raise NotImplementedError |
| 152 | |
| 153 | def HasExtension(self, extension_handle): |
| 154 | raise NotImplementedError |
| 155 | |
| 156 | def ClearExtension(self, extension_handle): |
| 157 | raise NotImplementedError |
| 158 | |
| 159 | def ByteSize(self): |
| 160 | """Returns the serialized size of this message. |
| 161 | Recursively calls ByteSize() on all contained messages. |
| 162 | """ |
| 163 | raise NotImplementedError |
| 164 | |
| 165 | def _SetListener(self, message_listener): |
| 166 | """Internal method used by the protocol message implementation. |
| 167 | Clients should not call this directly. |
| 168 | |
| 169 | Sets a listener that this message will call on certain state transitions. |
| 170 | |
| 171 | The purpose of this method is to register back-edges from children to |
| 172 | parents at runtime, for the purpose of setting "has" bits and |
| 173 | byte-size-dirty bits in the parent and ancestor objects whenever a child or |
| 174 | descendant object is modified. |
| 175 | |
| 176 | If the client wants to disconnect this Message from the object tree, she |
| 177 | explicitly sets callback to None. |
| 178 | |
| 179 | If message_listener is None, unregisters any existing listener. Otherwise, |
| 180 | message_listener must implement the MessageListener interface in |
| 181 | internal/message_listener.py, and we discard any listener registered |
| 182 | via a previous _SetListener() call. |
| 183 | """ |
| 184 | raise NotImplementedError |