descriptor

Protocol Buffers - Google’s data interchange format Copyright 2008 Google Inc. All rights reserved. https://developers.google.com/protocol-buffers/

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

• Redistributions of source code must retain the above copyright

notice, this list of conditions and the following disclaimer.

• Redistributions in binary form must reproduce the above

copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.

• Neither the name of Google Inc. nor the names of its

contributors may be used to endorse or promote products derived from this software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. Author: kenton@google.com (Kenton Varda)

Based on original Protocol Buffers design by Sanjay Ghemawat, Jeff Dean, and others.

The messages in this file describe the definitions found in .proto files. A valid .proto file can be translated directly to a FileDescriptorProto without any other information (e.g. without reading its imports).

Options Each of the definitions above may have “options” attached. These are just annotations which may cause code to be generated slightly differently or may contain hints for code that manipulates protocol messages.

Clients may define custom options as extensions of the *Options messages. These extensions may not yet be known at parsing time, so the parser cannot store the values in them. Instead it stores them in a field in the *Options message called uninterpreted_option. This field must have the same name across all *Options messages. We then use this field to populate the extensions when we build a descriptor, at which point all protos have been parsed and so all extensions are known.

Extension numbers for custom options may be chosen as follows: * For options which will only be used within a single application or

organization, or for experimental options, use field numbers 50000 through 99999. It is up to you to ensure that you do not use the same number for multiple options.

• For options which will be published and used publicly by multiple independent entities, e-mail protobuf-global-extension-registry@google.com to reserve extension numbers. Simply provide your project name (e.g. Objective-C plugin) and your project website (if available) – there’s no need to explain how you intend to use them. Usually you only need one extension number. You can declare multiple options with only one extension number by putting them in a sub-message. See the Custom Options section of the docs for examples: https://developers.google.com/protocol-buffers/docs/proto#options If this turns out to be popular, a web service will be set up to automatically assign option numbers.

Note: Field numbers 1 through 32 are reserved for Google’s internal RPC

framework. We apologize for hoarding these numbers to ourselves, but we were already using them long before we decided to release Protocol Buffers.

Note: Field numbers 1 through 32 are reserved for Google’s internal RPC

framework. We apologize for hoarding these numbers to ourselves, but we were already using them long before we decided to release Protocol Buffers.

Optional source code info

message FileDescriptorSet

Fields:

• file (list of FileDescriptorProto) –

The protocol compiler can output a FileDescriptorSet containing the .proto files it parses.

message FileDescriptorProto

Fields:

• name (string) – file name, relative to root of source tree
• package (string) – e.g. “foo”, “foo.bar”, etc.
• dependency (string) – Names of files imported by this file.
• public_dependency (integer) – Indexes of the public imported files in the dependency list above.
• weak_dependency (integer) – Indexes of the weak imported files in the dependency list. For Google-internal migration only. Do not use.
• message_type (list of DescriptorProto) – All top-level definitions in this file.
• enum_type (list of EnumDescriptorProto) –
• service (list of ServiceDescriptorProto) –
• extension (list of FieldDescriptorProto) –
• options (FileOptions) –
• source_code_info (SourceCodeInfo) – This field contains optional information about the original source code. You may safely remove this entire field without harming runtime functionality of the descriptors – the information is needed only by development tools.
• syntax (string) – The syntax of the proto file. The supported values are “proto2” and “proto3”.

Describes a complete .proto file.

message DescriptorProto

Fields:

• name (string) –
• field (list of FieldDescriptorProto) –
• extension (list of FieldDescriptorProto) –
• nested_type (list of DescriptorProto) –
• enum_type (list of EnumDescriptorProto) –
• extension_range (list of ExtensionRange) –
• oneof_decl (list of OneofDescriptorProto) –
• options (MessageOptions) –
• reserved_range (list of ReservedRange) –
• reserved_name (string) – Reserved field names, which may not be used by fields in the same message. A given name may only be reserved once.

Describes a message type.

message ExtensionRange

Fields:

• start (integer) –
• end (integer) –

message ReservedRange

Fields:

• start (integer) –
• end (integer) –

message FieldDescriptorProto

Fields:

• name (string) –
• number (integer) –
• label (Label) –
• type (Type) – If type_name is set, this need not be set. If both this and type_name are set, this must be one of TYPE_ENUM, TYPE_MESSAGE or TYPE_GROUP.
• type_name (string) – For message and enum types, this is the name of the type. If the name starts with a ‘.’, it is fully-qualified. Otherwise, C++-like scoping rules are used to find the type (i.e. first the nested types within this message are searched, then within the parent, on up to the root namespace).
• extendee (string) – For extensions, this is the name of the type being extended. It is resolved in the same manner as type_name.
• default_value (string) – For numeric types, contains the original text representation of the value. For booleans, “true” or “false”. For strings, contains the default text contents (not escaped in any way). For bytes, contains the C escaped value. All bytes >= 128 are escaped. TODO(kenton): Base-64 encode?
• oneof_index (integer) – If set, gives the index of a oneof in the containing type’s oneof_decl list. This field is a member of that oneof.
• json_name (string) – JSON name of this field. The value is set by protocol compiler. If the user has set a “json_name” option on this field, that option’s value will be used. Otherwise, it’s deduced from the field’s name by converting it to camelCase.
• options (FieldOptions) –

Describes a field within a message.

enum Type

Symbols:TYPE_DOUBLE|TYPE_FLOAT|TYPE_INT64|TYPE_UINT64|TYPE_INT32|TYPE_FIXED64|TYPE_FIXED32|TYPE_BOOL|TYPE_STRING|TYPE_GROUP|TYPE_MESSAGE|TYPE_BYTES|TYPE_UINT32|TYPE_ENUM|TYPE_SFIXED32|TYPE_SFIXED64|TYPE_SINT32|TYPE_SINT64

• TYPE_DOUBLE:
• TYPE_FLOAT:
• TYPE_INT64:
• TYPE_UINT64:
• TYPE_INT32:
• TYPE_FIXED64:
• TYPE_FIXED32:
• TYPE_BOOL:
• TYPE_STRING:
• TYPE_GROUP:
• TYPE_MESSAGE:
• TYPE_BYTES:
• TYPE_UINT32:
• TYPE_ENUM:
• TYPE_SFIXED32:
• TYPE_SFIXED64:
• TYPE_SINT32:
• TYPE_SINT64:

enum Label

Symbols:LABEL_OPTIONAL|LABEL_REQUIRED|LABEL_REPEATED

• LABEL_OPTIONAL:
• LABEL_REQUIRED:
• LABEL_REPEATED:

message OneofDescriptorProto

Fields:

• name (string) –
• options (OneofOptions) –

Describes a oneof.

message EnumDescriptorProto

Fields:

• name (string) –
• value (list of EnumValueDescriptorProto) –
• options (EnumOptions) –

Describes an enum type.

message EnumValueDescriptorProto

Fields:

• name (string) –
• number (integer) –
• options (EnumValueOptions) –

Describes a value within an enum.

message ServiceDescriptorProto

Fields:

• name (string) –
• method (list of MethodDescriptorProto) –
• options (ServiceOptions) –

Describes a service.

message MethodDescriptorProto

Fields:

• name (string) –
• input_type (string) – Input and output type names. These are resolved in the same way as FieldDescriptorProto.type_name, but must refer to a message type.
• output_type (string) –
• options (MethodOptions) –
• client_streaming (boolean) – Identifies if client streams multiple client messages
• server_streaming (boolean) – Identifies if server streams multiple server messages

Describes a method of a service.

message FileOptions

Fields:

• java_package (string) – Sets the Java package where classes generated from this .proto will be placed. By default, the proto package is used, but this is often inappropriate because proto packages do not normally start with backwards domain names.
• java_outer_classname (string) – If set, all the classes from the .proto file are wrapped in a single outer class with the given name. This applies to both Proto1 (equivalent to the old “–one_java_file” option) and Proto2 (where a .proto always translates to a single class, but you may want to explicitly choose the class name).
• java_multiple_files (boolean) – If set true, then the Java code generator will generate a separate .java file for each top-level message, enum, and service defined in the .proto file. Thus, these types will not be nested inside the outer class named by java_outer_classname. However, the outer class will still be generated to contain the file’s getDescriptor() method as well as any top-level extensions defined in the file.
• java_generate_equals_and_hash (boolean) – If set true, then the Java code generator will generate equals() and hashCode() methods for all messages defined in the .proto file. This increases generated code size, potentially substantially for large protos, which may harm a memory-constrained application. - In the full runtime this is a speed optimization, as the AbstractMessage base class includes reflection-based implementations of these methods. - In the lite runtime, setting this option changes the semantics of equals() and hashCode() to more closely match those of the full runtime; the generated methods compute their results based on field values rather than object identity. (Implementations should not assume that hashcodes will be consistent across runtimes or versions of the protocol compiler.)
• java_string_check_utf8 (boolean) – If set true, then the Java2 code generator will generate code that throws an exception whenever an attempt is made to assign a non-UTF-8 byte sequence to a string field. Message reflection will do the same. However, an extension field still accepts non-UTF-8 byte sequences. This option has no effect on when used with the lite runtime.
• optimize_for (OptimizeMode) –
• go_package (string) – Sets the Go package where structs generated from this .proto will be placed. If omitted, the Go package will be derived from the following: - The basename of the package import path, if provided. - Otherwise, the package statement in the .proto file, if present. - Otherwise, the basename of the .proto file, without extension.
• cc_generic_services (boolean) –

  Should generic services be generated in each language? “Generic” services are not specific to any particular RPC system. They are generated by the main code generators in each language (without additional plugins). Generic services were the only kind of service generation supported by early versions of google.protobuf.

  Generic services are now considered deprecated in favor of using plugins that generate code specific to your particular RPC system. Therefore, these default to false. Old code which depends on generic services should explicitly set them to true.

• java_generic_services (boolean) –
• py_generic_services (boolean) –
• deprecated (boolean) – Is this file deprecated? Depending on the target platform, this can emit Deprecated annotations for everything in the file, or it will be completely ignored; in the very least, this is a formalization for deprecating files.
• cc_enable_arenas (boolean) – Enables the use of arenas for the proto messages in this file. This applies only to generated classes for C++.
• objc_class_prefix (string) – Sets the objective c class prefix which is prepended to all objective c generated classes from this .proto. There is no default.
• csharp_namespace (string) – Namespace for generated classes; defaults to the package.
• uninterpreted_option (list of UninterpretedOption) – The parser stores options it doesn’t recognize here. See above.

enum OptimizeMode

Symbols:SPEED|CODE_SIZE|LITE_RUNTIME

• SPEED:
• CODE_SIZE:
• LITE_RUNTIME:

message MessageOptions

Fields:

• message_set_wire_format (boolean) –

  Set true to use the old proto1 MessageSet wire format for extensions. This is provided for backwards-compatibility with the MessageSet wire format. You should not use this for any other reason: It’s less efficient, has fewer features, and is more complicated.

  The message must be defined exactly as follows: message Foo { option message_set_wire_format = true; extensions 4 to max; } Note that the message cannot have any defined fields; MessageSets only have extensions.

  All extensions of your type must be singular messages; e.g. they cannot be int32s, enums, or repeated messages.

  Because this is an option, the above two restrictions are not enforced by the protocol compiler.

• no_standard_descriptor_accessor (boolean) – Disables the generation of the standard “descriptor()” accessor, which can conflict with a field of the same name. This is meant to make migration from proto1 easier; new code should avoid fields named “descriptor”.
• deprecated (boolean) – Is this message deprecated? Depending on the target platform, this can emit Deprecated annotations for the message, or it will be completely ignored; in the very least, this is a formalization for deprecating messages.
• map_entry (boolean) –

  Whether the message is an automatically generated map entry type for the maps field.

  For maps fields: map<KeyType, ValueType> map_field = 1; The parsed descriptor looks like: message MapFieldEntry { option map_entry = true; optional KeyType key = 1; optional ValueType value = 2; } repeated MapFieldEntry map_field = 1;

  Implementations may choose not to generate the map_entry=true message, but use a native map in the target language to hold the keys and values. The reflection APIs in such implementions still need to work as if the field is a repeated message field.

  NOTE: Do not set the option in .proto files. Always use the maps syntax instead. The option should only be implicitly set by the proto compiler parser.

• uninterpreted_option (list of UninterpretedOption) – The parser stores options it doesn’t recognize here. See above.

message FieldOptions

Fields:

• ctype (CType) – The ctype option instructs the C++ code generator to use a different representation of the field than it normally would. See the specific options below. This option is not yet implemented in the open source release – sorry, we’ll try to include it in a future version!
• packed (boolean) – The packed option can be enabled for repeated primitive fields to enable a more efficient representation on the wire. Rather than repeatedly writing the tag and type for each element, the entire array is encoded as a single length-delimited blob. In proto3, only explicit setting it to false will avoid using packed encoding.
• jstype (JSType) – The jstype option determines the JavaScript type used for values of the field. The option is permitted only for 64 bit integral and fixed types (int64, uint64, sint64, fixed64, sfixed64). By default these types are represented as JavaScript strings. This avoids loss of precision that can happen when a large value is converted to a floating point JavaScript numbers. Specifying JS_NUMBER for the jstype causes the generated JavaScript code to use the JavaScript “number” type instead of strings. This option is an enum to permit additional types to be added, e.g. goog.math.Integer.
• lazy (boolean) –

  Should this field be parsed lazily? Lazy applies only to message-type fields. It means that when the outer message is initially parsed, the inner message’s contents will not be parsed but instead stored in encoded form. The inner message will actually be parsed when it is first accessed.

  This is only a hint. Implementations are free to choose whether to use eager or lazy parsing regardless of the value of this option. However, setting this option true suggests that the protocol author believes that using lazy parsing on this field is worth the additional bookkeeping overhead typically needed to implement it.

  This option does not affect the public interface of any generated code; all method signatures remain the same. Furthermore, thread-safety of the interface is not affected by this option; const methods remain safe to call from multiple threads concurrently, while non-const methods continue to require exclusive access.

  Note that implementations may choose not to check required fields within a lazy sub-message. That is, calling IsInitialized() on the outher message may return true even if the inner message has missing required fields. This is necessary because otherwise the inner message would have to be parsed in order to perform the check, defeating the purpose of lazy parsing. An implementation which chooses not to check required fields must be consistent about it. That is, for any particular sub-message, the implementation must either always check its required fields, or never check its required fields, regardless of whether or not the message has been parsed.

• deprecated (boolean) – Is this field deprecated? Depending on the target platform, this can emit Deprecated annotations for accessors, or it will be completely ignored; in the very least, this is a formalization for deprecating fields.
• weak (boolean) – For Google-internal migration only. Do not use.
• uninterpreted_option (list of UninterpretedOption) – The parser stores options it doesn’t recognize here. See above.

enum CType

Symbols:STRING|CORD|STRING_PIECE

• STRING:
• CORD:
• STRING_PIECE:

enum JSType

Symbols:JS_NORMAL|JS_STRING|JS_NUMBER

• JS_NORMAL:
• JS_STRING:
• JS_NUMBER:

message OneofOptions

Fields:

• uninterpreted_option (list of UninterpretedOption) – The parser stores options it doesn’t recognize here. See above.

message EnumOptions

Fields:

• allow_alias (boolean) – Set this option to true to allow mapping different tag names to the same value.
• deprecated (boolean) – Is this enum deprecated? Depending on the target platform, this can emit Deprecated annotations for the enum, or it will be completely ignored; in the very least, this is a formalization for deprecating enums.
• uninterpreted_option (list of UninterpretedOption) – The parser stores options it doesn’t recognize here. See above.

message EnumValueOptions

Fields:

• deprecated (boolean) – Is this enum value deprecated? Depending on the target platform, this can emit Deprecated annotations for the enum value, or it will be completely ignored; in the very least, this is a formalization for deprecating enum values.
• uninterpreted_option (list of UninterpretedOption) – The parser stores options it doesn’t recognize here. See above.

message ServiceOptions

Fields:

• deprecated (boolean) – Is this service deprecated? Depending on the target platform, this can emit Deprecated annotations for the service, or it will be completely ignored; in the very least, this is a formalization for deprecating services.
• uninterpreted_option (list of UninterpretedOption) – The parser stores options it doesn’t recognize here. See above.

message MethodOptions

Fields:

• deprecated (boolean) – Is this method deprecated? Depending on the target platform, this can emit Deprecated annotations for the method, or it will be completely ignored; in the very least, this is a formalization for deprecating methods.
• uninterpreted_option (list of UninterpretedOption) – The parser stores options it doesn’t recognize here. See above.

message UninterpretedOption

Fields:

• name (list of NamePart) –
• identifier_value (string) – The value of the uninterpreted option, in whatever type the tokenizer identified it as during parsing. Exactly one of these should be set.
• positive_int_value (uint64) –
• negative_int_value (long) –
• double_value (double) –
• string_value (bytes) –
• aggregate_value (string) –

A message representing a option the parser does not recognize. This only appears in options protos created by the compiler::Parser class. DescriptorPool resolves these when building Descriptor objects. Therefore, options protos in descriptor objects (e.g. returned by Descriptor::options(), or produced by Descriptor::CopyTo()) will never have UninterpretedOptions in them.

message NamePart

Fields:

• name_part (string) –
• is_extension (boolean) –

message SourceCodeInfo

Fields:

• location (list of Location) –

  A Location identifies a piece of source code in a .proto file which corresponds to a particular definition. This information is intended to be useful to IDEs, code indexers, documentation generators, and similar tools.

  For example, say we have a file like: message Foo { optional string foo = 1; } Let’s look at just the field definition: optional string foo = 1; ^ ^^ ^^ ^ ^^^ a bc de f ghi We have the following locations: span path represents [a,i) [ 4, 0, 2, 0 ] The whole field definition. [a,b) [ 4, 0, 2, 0, 4 ] The label (optional). [c,d) [ 4, 0, 2, 0, 5 ] The type (string). [e,f) [ 4, 0, 2, 0, 1 ] The name (foo). [g,h) [ 4, 0, 2, 0, 3 ] The number (1).

  Notes: - A location may refer to a repeated field itself (i.e. not to any particular index within it). This is used whenever a set of elements are logically enclosed in a single code segment. For example, an entire extend block (possibly containing multiple extension definitions) will have an outer location whose path refers to the “extensions” repeated field without an index. - Multiple locations may have the same path. This happens when a single logical declaration is spread out across multiple places. The most obvious example is the “extend” block again – there may be multiple extend blocks in the same scope, each of which will have the same path. - A location’s span is not always a subset of its parent’s span. For example, the “extendee” of an extension declaration appears at the beginning of the “extend” block and is shared by all extensions within the block. - Just because a location’s span is a subset of some other location’s span does not mean that it is a descendent. For example, a “group” defines both a type and a field in a single declaration. Thus, the locations corresponding to the type and field and their components will overlap. - Code which tries to interpret locations should probably be designed to ignore those that it doesn’t understand, as more types of locations could be recorded in the future.

Encapsulates information about the original source file from which a FileDescriptorProto was generated.

message Location

Fields:

• path (integer) –
• span (integer) –
• leading_comments (string) –
• trailing_comments (string) –
• leading_detached_comments (string) –

message GeneratedCodeInfo

Fields:

• annotation (list of Annotation) – An Annotation connects some span of text in generated code to an element of its generating .proto file.

Describes the relationship between generated code and its original source file. A GeneratedCodeInfo message is associated with only one generated source file, but may contain references to different source .proto files.

message Annotation

Fields:

• path (integer) –
• source_file (string) –
• begin (integer) –
• end (integer) –