src/google/protobuf/compiler/cpp/helpers.h

// 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.

#ifndef GOOGLE_PROTOBUF_COMPILER_CPP_HELPERS_H__
#define GOOGLE_PROTOBUF_COMPILER_CPP_HELPERS_H__

#include <algorithm>
#include <cstdint>
#include <iterator>
#include <map>
#include <string>

#include <google/protobuf/compiler/scc.h>
#include <google/protobuf/compiler/code_generator.h>
#include <google/protobuf/compiler/cpp/names.h>
#include <google/protobuf/compiler/cpp/options.h>
#include <google/protobuf/descriptor.pb.h>
#include <google/protobuf/io/printer.h>
#include <google/protobuf/descriptor.h>
#include <google/protobuf/port.h>
#include <google/protobuf/stubs/strutil.h>

// Must be included last.
#include <google/protobuf/port_def.inc>

namespace google {
namespace protobuf {
namespace compiler {
namespace cpp {

enum class ArenaDtorNeeds { kNone = 0, kOnDemand = 1, kRequired = 2 };

inline std::string ProtobufNamespace(const Options& /* options */) {
  return "PROTOBUF_NAMESPACE_ID";
}

inline std::string MacroPrefix(const Options& /* options */) {
  return "GOOGLE_PROTOBUF";
}

inline std::string DeprecatedAttribute(const Options& /* options */,
                                       const FieldDescriptor* d) {
  return d->options().deprecated() ? "PROTOBUF_DEPRECATED " : "";
}

inline std::string DeprecatedAttribute(const Options& /* options */,
                                       const EnumValueDescriptor* d) {
  return d->options().deprecated() ? "PROTOBUF_DEPRECATED_ENUM " : "";
}

// Commonly-used separator comments.  Thick is a line of '=', thin is a line
// of '-'.
extern const char kThickSeparator[];
extern const char kThinSeparator[];

void SetCommonVars(const Options& options,
                   std::map<std::string, std::string>* variables);

// Variables to access message data from the message scope.
void SetCommonMessageDataVariables(
    const Descriptor* descriptor,
    std::map<std::string, std::string>* variables);

void SetUnknownFieldsVariable(const Descriptor* descriptor,
                              const Options& options,
                              std::map<std::string, std::string>* variables);

bool GetBootstrapBasename(const Options& options, const std::string& basename,
                          std::string* bootstrap_basename);
bool MaybeBootstrap(const Options& options, GeneratorContext* generator_context,
                    bool bootstrap_flag, std::string* basename);
bool IsBootstrapProto(const Options& options, const FileDescriptor* file);

// Name space of the proto file. This namespace is such that the string
// "<namespace>::some_name" is the correct fully qualified namespace.
// This means if the package is empty the namespace is "", and otherwise
// the namespace is "::foo::bar::...::baz" without trailing semi-colons.
std::string Namespace(const FileDescriptor* d, const Options& options);
std::string Namespace(const Descriptor* d, const Options& options);
std::string Namespace(const FieldDescriptor* d, const Options& options);
std::string Namespace(const EnumDescriptor* d, const Options& options);

// Returns true if it's safe to reset "field" to zero.
bool CanInitializeByZeroing(const FieldDescriptor* field);

std::string ClassName(const Descriptor* descriptor);
std::string ClassName(const EnumDescriptor* enum_descriptor);

std::string QualifiedClassName(const Descriptor* d, const Options& options);
std::string QualifiedClassName(const EnumDescriptor* d, const Options& options);

std::string QualifiedClassName(const Descriptor* d);
std::string QualifiedClassName(const EnumDescriptor* d);

// DEPRECATED just use ClassName or QualifiedClassName, a boolean is very
// unreadable at the callsite.
// Returns the non-nested type name for the given type.  If "qualified" is
// true, prefix the type with the full namespace.  For example, if you had:
//   package foo.bar;
//   message Baz { message Moo {} }
// Then the qualified ClassName for Moo would be:
//   ::foo::bar::Baz_Moo
// While the non-qualified version would be:
//   Baz_Moo
inline std::string ClassName(const Descriptor* descriptor, bool qualified) {
  return qualified ? QualifiedClassName(descriptor, Options())
                   : ClassName(descriptor);
}

inline std::string ClassName(const EnumDescriptor* descriptor, bool qualified) {
  return qualified ? QualifiedClassName(descriptor, Options())
                   : ClassName(descriptor);
}

// Returns the extension name prefixed with the class name if nested but without
// the package name.
std::string ExtensionName(const FieldDescriptor* d);

std::string QualifiedExtensionName(const FieldDescriptor* d,
                                   const Options& options);
std::string QualifiedExtensionName(const FieldDescriptor* d);

// Type name of default instance.
std::string DefaultInstanceType(const Descriptor* descriptor,
                                const Options& options, bool split = false);

// Non-qualified name of the default_instance of this message.
std::string DefaultInstanceName(const Descriptor* descriptor,
                                const Options& options, bool split = false);

// Non-qualified name of the default instance pointer. This is used only for
// implicit weak fields, where we need an extra indirection.
std::string DefaultInstancePtr(const Descriptor* descriptor,
                               const Options& options, bool split = false);

// Fully qualified name of the default_instance of this message.
std::string QualifiedDefaultInstanceName(const Descriptor* descriptor,
                                         const Options& options,
                                         bool split = false);

// Fully qualified name of the default instance pointer.
std::string QualifiedDefaultInstancePtr(const Descriptor* descriptor,
                                        const Options& options,
                                        bool split = false);

// DescriptorTable variable name.
std::string DescriptorTableName(const FileDescriptor* file,
                                const Options& options);

// When declaring symbol externs from another file, this macro will supply the
// dllexport needed for the target file, if any.
std::string FileDllExport(const FileDescriptor* file, const Options& options);

// Name of the base class: google::protobuf::Message or google::protobuf::MessageLite.
std::string SuperClassName(const Descriptor* descriptor,
                           const Options& options);

// Adds an underscore if necessary to prevent conflicting with a keyword.
std::string ResolveKeyword(const std::string& name);

// Get the (unqualified) name that should be used for this field in C++ code.
// The name is coerced to lower-case to emulate proto1 behavior.  People
// should be using lowercase-with-underscores style for proto field names
// anyway, so normally this just returns field->name().
std::string FieldName(const FieldDescriptor* field);

// Returns the (unqualified) private member name for this field in C++ code.
std::string FieldMemberName(const FieldDescriptor* field, bool split);

// Returns an estimate of the compiler's alignment for the field.  This
// can't guarantee to be correct because the generated code could be compiled on
// different systems with different alignment rules.  The estimates below assume
// 64-bit pointers.
int EstimateAlignmentSize(const FieldDescriptor* field);

// Get the unqualified name that should be used for a field's field
// number constant.
std::string FieldConstantName(const FieldDescriptor* field);

// Returns the scope where the field was defined (for extensions, this is
// different from the message type to which the field applies).
inline const Descriptor* FieldScope(const FieldDescriptor* field) {
  return field->is_extension() ? field->extension_scope()
                               : field->containing_type();
}

// Returns the fully-qualified type name field->message_type().  Usually this
// is just ClassName(field->message_type(), true);
std::string FieldMessageTypeName(const FieldDescriptor* field,
                                 const Options& options);

// Get the C++ type name for a primitive type (e.g. "double", "::google::protobuf::int32", etc.).
const char* PrimitiveTypeName(FieldDescriptor::CppType type);
std::string PrimitiveTypeName(const Options& options,
                              FieldDescriptor::CppType type);

// Get the declared type name in CamelCase format, as is used e.g. for the
// methods of WireFormat.  For example, TYPE_INT32 becomes "Int32".
const char* DeclaredTypeMethodName(FieldDescriptor::Type type);

// Return the code that evaluates to the number when compiled.
std::string Int32ToString(int number);

// Get code that evaluates to the field's default value.
std::string DefaultValue(const Options& options, const FieldDescriptor* field);

// Compatibility function for callers outside proto2.
std::string DefaultValue(const FieldDescriptor* field);

// Convert a file name into a valid identifier.
std::string FilenameIdentifier(const std::string& filename);

// For each .proto file generates a unique name. To prevent collisions of
// symbols in the global namespace
std::string UniqueName(const std::string& name, const std::string& filename,
                       const Options& options);
inline std::string UniqueName(const std::string& name, const FileDescriptor* d,
                              const Options& options) {
  return UniqueName(name, d->name(), options);
}
inline std::string UniqueName(const std::string& name, const Descriptor* d,
                              const Options& options) {
  return UniqueName(name, d->file(), options);
}
inline std::string UniqueName(const std::string& name, const EnumDescriptor* d,
                              const Options& options) {
  return UniqueName(name, d->file(), options);
}
inline std::string UniqueName(const std::string& name,
                              const ServiceDescriptor* d,
                              const Options& options) {
  return UniqueName(name, d->file(), options);
}

// Versions for call sites that only support the internal runtime (like proto1
// support).
inline Options InternalRuntimeOptions() {
  Options options;
  options.opensource_runtime = false;
  return options;
}
inline std::string UniqueName(const std::string& name,
                              const std::string& filename) {
  return UniqueName(name, filename, InternalRuntimeOptions());
}
inline std::string UniqueName(const std::string& name,
                              const FileDescriptor* d) {
  return UniqueName(name, d->name(), InternalRuntimeOptions());
}
inline std::string UniqueName(const std::string& name, const Descriptor* d) {
  return UniqueName(name, d->file(), InternalRuntimeOptions());
}
inline std::string UniqueName(const std::string& name,
                              const EnumDescriptor* d) {
  return UniqueName(name, d->file(), InternalRuntimeOptions());
}
inline std::string UniqueName(const std::string& name,
                              const ServiceDescriptor* d) {
  return UniqueName(name, d->file(), InternalRuntimeOptions());
}

// Return the qualified C++ name for a file level symbol.
std::string QualifiedFileLevelSymbol(const FileDescriptor* file,
                                     const std::string& name,
                                     const Options& options);

// Escape C++ trigraphs by escaping question marks to \?
std::string EscapeTrigraphs(const std::string& to_escape);

// Escaped function name to eliminate naming conflict.
std::string SafeFunctionName(const Descriptor* descriptor,
                             const FieldDescriptor* field,
                             const std::string& prefix);

// Returns true if generated messages have public unknown fields accessors
inline bool PublicUnknownFieldsAccessors(const Descriptor* message) {
  return message->file()->syntax() != FileDescriptor::SYNTAX_PROTO3;
}

// Returns the optimize mode for <file>, respecting <options.enforce_lite>.
FileOptions_OptimizeMode GetOptimizeFor(const FileDescriptor* file,
                                        const Options& options);

// Determines whether unknown fields will be stored in an UnknownFieldSet or
// a string.
inline bool UseUnknownFieldSet(const FileDescriptor* file,
                               const Options& options) {
  return GetOptimizeFor(file, options) != FileOptions::LITE_RUNTIME;
}

inline bool IsWeak(const FieldDescriptor* field, const Options& options) {
  if (field->options().weak()) {
    GOOGLE_CHECK(!options.opensource_runtime);
    return true;
  }
  return false;
}

bool IsStringInlined(const FieldDescriptor* descriptor, const Options& options);

// For a string field, returns the effective ctype.  If the actual ctype is
// not supported, returns the default of STRING.
FieldOptions::CType EffectiveStringCType(const FieldDescriptor* field,
                                         const Options& options);

inline bool IsCord(const FieldDescriptor* field, const Options& options) {
  return field->cpp_type() == FieldDescriptor::CPPTYPE_STRING &&
         EffectiveStringCType(field, options) == FieldOptions::CORD;
}

inline bool IsString(const FieldDescriptor* field, const Options& options) {
  return field->cpp_type() == FieldDescriptor::CPPTYPE_STRING &&
         EffectiveStringCType(field, options) == FieldOptions::STRING;
}

inline bool IsStringPiece(const FieldDescriptor* field,
                          const Options& options) {
  return field->cpp_type() == FieldDescriptor::CPPTYPE_STRING &&
         EffectiveStringCType(field, options) == FieldOptions::STRING_PIECE;
}

class MessageSCCAnalyzer;

// Does the given FileDescriptor use lazy fields?
bool HasLazyFields(const FileDescriptor* file, const Options& options,
                   MessageSCCAnalyzer* scc_analyzer);

// Is the given field a supported lazy field?
bool IsLazy(const FieldDescriptor* field, const Options& options,
            MessageSCCAnalyzer* scc_analyzer);

// Is this an explicit (non-profile driven) lazy field, as denoted by
// lazy/unverified_lazy in the descriptor?
inline bool IsExplicitLazy(const FieldDescriptor* field) {
  return field->options().lazy() || field->options().unverified_lazy();
}

bool IsEagerlyVerifiedLazy(const FieldDescriptor* field, const Options& options,
                           MessageSCCAnalyzer* scc_analyzer);

bool IsLazilyVerifiedLazy(const FieldDescriptor* field, const Options& options);

// Is the given message being split (go/pdsplit)?
bool ShouldSplit(const Descriptor* desc, const Options& options);

// Is the given field being split out?
bool ShouldSplit(const FieldDescriptor* field, const Options& options);

inline bool IsFieldUsed(const FieldDescriptor* /* field */,
                        const Options& /* options */) {
  return true;
}

// Returns true if "field" is stripped.
inline bool IsFieldStripped(const FieldDescriptor* /*field*/,
                            const Options& /*options*/) {
  return false;
}

// Does the file contain any definitions that need extension_set.h?
bool HasExtensionsOrExtendableMessage(const FileDescriptor* file);

// Does the file have any repeated fields, necessitating the file to include
// repeated_field.h? This does not include repeated extensions, since those are
// all stored internally in an ExtensionSet, not a separate RepeatedField*.
bool HasRepeatedFields(const FileDescriptor* file);

// Does the file have any string/bytes fields with ctype=STRING_PIECE? This
// does not include extensions, since ctype is ignored for extensions.
bool HasStringPieceFields(const FileDescriptor* file, const Options& options);

// Does the file have any string/bytes fields with ctype=CORD? This does not
// include extensions, since ctype is ignored for extensions.
bool HasCordFields(const FileDescriptor* file, const Options& options);

// Does the file have any map fields, necessitating the file to include
// map_field_inl.h and map.h.
bool HasMapFields(const FileDescriptor* file);

// Does this file have any enum type definitions?
bool HasEnumDefinitions(const FileDescriptor* file);

// Does this file have generated parsing, serialization, and other
// standard methods for which reflection-based fallback implementations exist?
inline bool HasGeneratedMethods(const FileDescriptor* file,
                                const Options& options) {
  return GetOptimizeFor(file, options) != FileOptions::CODE_SIZE;
}

// Do message classes in this file have descriptor and reflection methods?
inline bool HasDescriptorMethods(const FileDescriptor* file,
                                 const Options& options) {
  return GetOptimizeFor(file, options) != FileOptions::LITE_RUNTIME;
}

// Should we generate generic services for this file?
inline bool HasGenericServices(const FileDescriptor* file,
                               const Options& options) {
  return file->service_count() > 0 &&
         GetOptimizeFor(file, options) != FileOptions::LITE_RUNTIME &&
         file->options().cc_generic_services();
}

inline bool IsProto2MessageSet(const Descriptor* descriptor,
                               const Options& options) {
  return !options.opensource_runtime &&
         options.enforce_mode != EnforceOptimizeMode::kLiteRuntime &&
         !options.lite_implicit_weak_fields &&
         descriptor->options().message_set_wire_format() &&
         descriptor->full_name() == "google.protobuf.bridge.MessageSet";
}

inline bool IsMapEntryMessage(const Descriptor* descriptor) {
  return descriptor->options().map_entry();
}

// Returns true if the field's CPPTYPE is string or message.
bool IsStringOrMessage(const FieldDescriptor* field);

std::string UnderscoresToCamelCase(const std::string& input,
                                   bool cap_next_letter);

inline bool IsProto3(const FileDescriptor* file) {
  return file->syntax() == FileDescriptor::SYNTAX_PROTO3;
}

inline bool HasHasbit(const FieldDescriptor* field) {
  // This predicate includes proto3 message fields only if they have "optional".
  //   Foo submsg1 = 1;           // HasHasbit() == false
  //   optional Foo submsg2 = 2;  // HasHasbit() == true
  // This is slightly odd, as adding "optional" to a singular proto3 field does
  // not change the semantics or API. However whenever any field in a message
  // has a hasbit, it forces reflection to include hasbit offsets for *all*
  // fields, even if almost all of them are set to -1 (no hasbit). So to avoid
  // causing a sudden size regression for ~all proto3 messages, we give proto3
  // message fields a hasbit only if "optional" is present. If the user is
  // explicitly writing "optional", it is likely they are writing it on
  // primitive fields also.
  return (field->has_optional_keyword() || field->is_required()) &&
         !field->options().weak();
}

// Returns true if 'enum' semantics are such that unknown values are preserved
// in the enum field itself, rather than going to the UnknownFieldSet.
inline bool HasPreservingUnknownEnumSemantics(const FieldDescriptor* field) {
  return field->file()->syntax() == FileDescriptor::SYNTAX_PROTO3;
}

inline bool IsCrossFileMessage(const FieldDescriptor* field) {
  return field->type() == FieldDescriptor::TYPE_MESSAGE &&
         field->message_type()->file() != field->file();
}

inline std::string MakeDefaultName(const FieldDescriptor* field) {
  return StrCat("_i_give_permission_to_break_this_code_default_",
                      FieldName(field), "_");
}

// Semantically distinct from MakeDefaultName in that it gives the C++ code
// referencing a default field from the message scope, rather than just the
// variable name.
// For example, declarations of default variables should always use just
// MakeDefaultName to produce code like:
//   Type _i_give_permission_to_break_this_code_default_field_;
//
// Code that references these should use MakeDefaultFieldName, in case the field
// exists at some nested level like:
//   internal_container_._i_give_permission_to_break_this_code_default_field_;
inline std::string MakeDefaultFieldName(const FieldDescriptor* field) {
  return StrCat("Impl_::", MakeDefaultName(field));
}

inline std::string MakeVarintCachedSizeName(const FieldDescriptor* field) {
  return StrCat("_", FieldName(field), "_cached_byte_size_");
}

// Semantically distinct from MakeVarintCachedSizeName in that it gives the C++
// code referencing the object from the message scope, rather than just the
// variable name.
// For example, declarations of default variables should always use just
// MakeVarintCachedSizeName to produce code like:
//   Type _field_cached_byte_size_;
//
// Code that references these variables should use
// MakeVarintCachedSizeFieldName, in case the field exists at some nested level
// like:
//   internal_container_._field_cached_byte_size_;
inline std::string MakeVarintCachedSizeFieldName(const FieldDescriptor* field,
                                                 bool split) {
  return StrCat("_impl_.", split ? "_split_->" : "", "_",
                      FieldName(field), "_cached_byte_size_");
}

// Note: A lot of libraries detect Any protos based on Descriptor::full_name()
// while the two functions below use FileDescriptor::name(). In a sane world the
// two approaches should be equivalent. But if you are dealing with descriptors
// from untrusted sources, you might need to match semantics across libraries.
bool IsAnyMessage(const FileDescriptor* descriptor, const Options& options);
bool IsAnyMessage(const Descriptor* descriptor, const Options& options);

bool IsWellKnownMessage(const FileDescriptor* descriptor);

inline std::string IncludeGuard(const FileDescriptor* file, bool pb_h,
                                const Options& options) {
  // If we are generating a .pb.h file and the proto_h option is enabled, then
  // the .pb.h gets an extra suffix.
  std::string filename_identifier = FilenameIdentifier(
      file->name() + (pb_h && options.proto_h ? ".pb.h" : ""));

  if (IsWellKnownMessage(file)) {
    // For well-known messages we need third_party/protobuf and net/proto2 to
    // have distinct include guards, because some source files include both and
    // both need to be defined (the third_party copies will be in the
    // google::protobuf_opensource namespace).
    return MacroPrefix(options) + "_INCLUDED_" + filename_identifier;
  } else {
    // Ideally this case would use distinct include guards for opensource and
    // google3 protos also.  (The behavior of "first #included wins" is not
    // ideal).  But unfortunately some legacy code includes both and depends on
    // the identical include guards to avoid compile errors.
    //
    // We should clean this up so that this case can be removed.
    return "GOOGLE_PROTOBUF_INCLUDED_" + filename_identifier;
  }
}

// Returns the OptimizeMode for this file, furthermore it updates a status
// bool if has_opt_codesize_extension is non-null. If this status bool is true
// it means this file contains an extension that itself is defined as
// optimized_for = CODE_SIZE.
FileOptions_OptimizeMode GetOptimizeFor(const FileDescriptor* file,
                                        const Options& options,
                                        bool* has_opt_codesize_extension);
inline FileOptions_OptimizeMode GetOptimizeFor(const FileDescriptor* file,
                                               const Options& options) {
  return GetOptimizeFor(file, options, nullptr);
}
inline bool NeedsEagerDescriptorAssignment(const FileDescriptor* file,
                                           const Options& options) {
  bool has_opt_codesize_extension;
  if (GetOptimizeFor(file, options, &has_opt_codesize_extension) ==
          FileOptions::CODE_SIZE &&
      has_opt_codesize_extension) {
    // If this filedescriptor contains an extension from another file which
    // is optimized_for = CODE_SIZE. We need to be careful in the ordering so
    // we eagerly build the descriptors in the dependencies before building
    // the descriptors of this file.
    return true;
  } else {
    // If we have a generated code based parser we never need eager
    // initialization of descriptors of our deps.
    return false;
  }
}

// This orders the messages in a .pb.cc as it's outputted by file.cc
void FlattenMessagesInFile(const FileDescriptor* file,
                           std::vector<const Descriptor*>* result);
inline std::vector<const Descriptor*> FlattenMessagesInFile(
    const FileDescriptor* file) {
  std::vector<const Descriptor*> result;
  FlattenMessagesInFile(file, &result);
  return result;
}

template <typename F>
void ForEachMessage(const Descriptor* descriptor, F&& func) {
  for (int i = 0; i < descriptor->nested_type_count(); i++)
    ForEachMessage(descriptor->nested_type(i), std::forward<F&&>(func));
  func(descriptor);
}

template <typename F>
void ForEachMessage(const FileDescriptor* descriptor, F&& func) {
  for (int i = 0; i < descriptor->message_type_count(); i++)
    ForEachMessage(descriptor->message_type(i), std::forward<F&&>(func));
}

bool HasWeakFields(const Descriptor* desc, const Options& options);
bool HasWeakFields(const FileDescriptor* desc, const Options& options);

// Returns true if the "required" restriction check should be ignored for the
// given field.
inline static bool ShouldIgnoreRequiredFieldCheck(const FieldDescriptor* field,
                                                  const Options& options) {
  // Do not check "required" for lazily verified lazy fields.
  return IsLazilyVerifiedLazy(field, options);
}

struct MessageAnalysis {
  bool is_recursive = false;
  bool contains_cord = false;
  bool contains_extension = false;
  bool contains_required = false;
  bool contains_weak = false;  // Implicit weak as well.
};

// This class is used in FileGenerator, to ensure linear instead of
// quadratic performance, if we do this per message we would get O(V*(V+E)).
// Logically this is just only used in message.cc, but in the header for
// FileGenerator to help share it.
class PROTOC_EXPORT MessageSCCAnalyzer {
 public:
  explicit MessageSCCAnalyzer(const Options& options) : options_(options) {}

  MessageAnalysis GetSCCAnalysis(const SCC* scc);

  bool HasRequiredFields(const Descriptor* descriptor) {
    MessageAnalysis result = GetSCCAnalysis(GetSCC(descriptor));
    return result.contains_required || result.contains_extension;
  }
  bool HasWeakField(const Descriptor* descriptor) {
    MessageAnalysis result = GetSCCAnalysis(GetSCC(descriptor));
    return result.contains_weak;
  }
  const SCC* GetSCC(const Descriptor* descriptor) {
    return analyzer_.GetSCC(descriptor);
  }

 private:
  struct DepsGenerator {
    std::vector<const Descriptor*> operator()(const Descriptor* desc) const {
      std::vector<const Descriptor*> deps;
      for (int i = 0; i < desc->field_count(); i++) {
        if (desc->field(i)->message_type()) {
          deps.push_back(desc->field(i)->message_type());
        }
      }
      return deps;
    }
  };
  SCCAnalyzer<DepsGenerator> analyzer_;
  Options options_;
  std::map<const SCC*, MessageAnalysis> analysis_cache_;
};

void ListAllFields(const Descriptor* d,
                   std::vector<const FieldDescriptor*>* fields);
void ListAllFields(const FileDescriptor* d,
                   std::vector<const FieldDescriptor*>* fields);

template <class T>
void ForEachField(const Descriptor* d, T&& func) {
  for (int i = 0; i < d->nested_type_count(); i++) {
    ForEachField(d->nested_type(i), std::forward<T&&>(func));
  }
  for (int i = 0; i < d->extension_count(); i++) {
    func(d->extension(i));
  }
  for (int i = 0; i < d->field_count(); i++) {
    func(d->field(i));
  }
}

template <class T>
void ForEachField(const FileDescriptor* d, T&& func) {
  for (int i = 0; i < d->message_type_count(); i++) {
    ForEachField(d->message_type(i), std::forward<T&&>(func));
  }
  for (int i = 0; i < d->extension_count(); i++) {
    func(d->extension(i));
  }
}

void ListAllTypesForServices(const FileDescriptor* fd,
                             std::vector<const Descriptor*>* types);

// Indicates whether we should use implicit weak fields for this file.
bool UsingImplicitWeakFields(const FileDescriptor* file,
                             const Options& options);

// Indicates whether to treat this field as implicitly weak.
bool IsImplicitWeakField(const FieldDescriptor* field, const Options& options,
                         MessageSCCAnalyzer* scc_analyzer);

inline bool HasSimpleBaseClass(const Descriptor* desc, const Options& options) {
  if (!HasDescriptorMethods(desc->file(), options)) return false;
  if (desc->extension_range_count() != 0) return false;
  if (desc->field_count() == 0) return true;
  // TODO(jorg): Support additional common message types with only one
  // or two fields
  return false;
}

inline bool HasSimpleBaseClasses(const FileDescriptor* file,
                                 const Options& options) {
  bool v = false;
  ForEachMessage(file, [&v, &options](const Descriptor* desc) {
    v |= HasSimpleBaseClass(desc, options);
  });
  return v;
}

inline std::string SimpleBaseClass(const Descriptor* desc,
                                   const Options& options) {
  if (!HasDescriptorMethods(desc->file(), options)) return "";
  if (desc->extension_range_count() != 0) return "";
  if (desc->field_count() == 0) {
    return "ZeroFieldsBase";
  }
  // TODO(jorg): Support additional common message types with only one
  // or two fields
  return "";
}

// Returns true if this message has a _tracker_ field.
inline bool HasTracker(const Descriptor* desc, const Options& options) {
  return options.field_listener_options.inject_field_listener_events &&
         desc->file()->options().optimize_for() !=
             google::protobuf::FileOptions::LITE_RUNTIME;
}

// Returns true if this message needs an Impl_ struct for it's data.
inline bool HasImplData(const Descriptor* desc, const Options& options) {
  return !HasSimpleBaseClass(desc, options);
}

// Formatter is a functor class which acts as a closure around printer and
// the variable map. It's much like printer->Print except it supports both named
// variables that are substituted using a key value map and direct arguments. In
// the format string $1$, $2$, etc... are substituted for the first, second, ...
// direct argument respectively in the format call, it accepts both strings and
// integers. The implementation verifies all arguments are used and are "first"
// used in order of appearance in the argument list. For example,
//
// Format("return array[$1$];", 3) -> "return array[3];"
// Format("array[$2$] = $1$;", "Bla", 3) -> FATAL error (wrong order)
// Format("array[$1$] = $2$;", 3, "Bla") -> "array[3] = Bla;"
//
// The arguments can be used more than once like
//
// Format("array[$1$] = $2$;  // Index = $1$", 3, "Bla") ->
//        "array[3] = Bla;  // Index = 3"
//
// If you use more arguments use the following style to help the reader,
//
// Format("int $1$() {\n"
//        "  array[$2$] = $3$;\n"
//        "  return $4$;"
//        "}\n",
//        funname, // 1
//        idx,  // 2
//        varname,  // 3
//        retval);  // 4
//
// but consider using named variables. Named variables like $foo$, with some
// identifier foo, are looked up in the map. One additional feature is that
// spaces are accepted between the '$' delimiters, $ foo$ will
// substitute to " bar" if foo stands for "bar", but in case it's empty
// will substitute to "". Hence, for example,
//
// Format(vars, "$dllexport $void fun();") -> "void fun();"
//                                            "__declspec(export) void fun();"
//
// which is convenient to prevent double, leading or trailing spaces.
class PROTOC_EXPORT Formatter {
 public:
  explicit Formatter(io::Printer* printer) : printer_(printer) {}
  Formatter(io::Printer* printer,
            const std::map<std::string, std::string>& vars)
      : printer_(printer), vars_(vars) {}

  template <typename T>
  void Set(const std::string& key, const T& value) {
    vars_[key] = ToString(value);
  }

  void AddMap(const std::map<std::string, std::string>& vars) {
    for (const auto& keyval : vars) vars_[keyval.first] = keyval.second;
  }

  template <typename... Args>
  void operator()(const char* format, const Args&... args) const {
    printer_->FormatInternal({ToString(args)...}, vars_, format);
  }

  void Indent() const { printer_->Indent(); }
  void Outdent() const { printer_->Outdent(); }
  io::Printer* printer() const { return printer_; }

  class PROTOC_EXPORT ScopedIndenter {
   public:
    explicit ScopedIndenter(Formatter* format) : format_(format) {
      format_->Indent();
    }
    ~ScopedIndenter() { format_->Outdent(); }

   private:
    Formatter* format_;
  };

  PROTOBUF_NODISCARD ScopedIndenter ScopedIndent() {
    return ScopedIndenter(this);
  }
  template <typename... Args>
  PROTOBUF_NODISCARD ScopedIndenter ScopedIndent(const char* format,
                                                 const Args&&... args) {
    (*this)(format, static_cast<Args&&>(args)...);
    return ScopedIndenter(this);
  }

  class PROTOC_EXPORT SaveState {
   public:
    explicit SaveState(Formatter* format)
        : format_(format), vars_(format->vars_) {}
    ~SaveState() { format_->vars_.swap(vars_); }

   private:
    Formatter* format_;
    std::map<std::string, std::string> vars_;
  };

 private:
  io::Printer* printer_;
  std::map<std::string, std::string> vars_;

  // Convenience overloads to accept different types as arguments.
  static std::string ToString(const std::string& s) { return s; }
  template <typename I, typename = typename std::enable_if<
                            std::is_integral<I>::value>::type>
  static std::string ToString(I x) {
    return StrCat(x);
  }
  static std::string ToString(strings::Hex x) { return StrCat(x); }
  static std::string ToString(const FieldDescriptor* d) { return Payload(d); }
  static std::string ToString(const Descriptor* d) { return Payload(d); }
  static std::string ToString(const EnumDescriptor* d) { return Payload(d); }
  static std::string ToString(const EnumValueDescriptor* d) {
    return Payload(d);
  }
  static std::string ToString(const OneofDescriptor* d) { return Payload(d); }

  template <typename Descriptor>
  static std::string Payload(const Descriptor* descriptor) {
    std::vector<int> path;
    descriptor->GetLocationPath(&path);
    GeneratedCodeInfo::Annotation annotation;
    for (int index : path) {
      annotation.add_path(index);
    }
    annotation.set_source_file(descriptor->file()->name());
    return annotation.SerializeAsString();
  }
};

template <class T>
void PrintFieldComment(const Formatter& format, const T* field) {
  // Print the field's (or oneof's) proto-syntax definition as a comment.
  // We don't want to print group bodies so we cut off after the first
  // line.
  DebugStringOptions options;
  options.elide_group_body = true;
  options.elide_oneof_body = true;
  std::string def = field->DebugStringWithOptions(options);
  format("// $1$\n", def.substr(0, def.find_first_of('\n')));
}

class PROTOC_EXPORT NamespaceOpener {
 public:
  explicit NamespaceOpener(const Formatter& format)
      : printer_(format.printer()) {}
  NamespaceOpener(const std::string& name, const Formatter& format)
      : NamespaceOpener(format) {
    ChangeTo(name);
  }
  ~NamespaceOpener() { ChangeTo(""); }

  void ChangeTo(const std::string& name) {
    std::vector<std::string> new_stack_ =
        Split(name, "::", true);
    size_t len = std::min(name_stack_.size(), new_stack_.size());
    size_t common_idx = 0;
    while (common_idx < len) {
      if (name_stack_[common_idx] != new_stack_[common_idx]) break;
      common_idx++;
    }
    for (auto it = name_stack_.crbegin();
         it != name_stack_.crend() - common_idx; ++it) {
      if (*it == "PROTOBUF_NAMESPACE_ID") {
        printer_->Print("PROTOBUF_NAMESPACE_CLOSE\n");
      } else {
        printer_->Print("}  // namespace $ns$\n", "ns", *it);
      }
    }
    name_stack_.swap(new_stack_);
    for (size_t i = common_idx; i < name_stack_.size(); ++i) {
      if (name_stack_[i] == "PROTOBUF_NAMESPACE_ID") {
        printer_->Print("PROTOBUF_NAMESPACE_OPEN\n");
      } else {
        printer_->Print("namespace $ns$ {\n", "ns", name_stack_[i]);
      }
    }
  }

 private:
  io::Printer* printer_;
  std::vector<std::string> name_stack_;
};

enum class Utf8CheckMode {
  kStrict = 0,  // Parsing will fail if non UTF-8 data is in string fields.
  kVerify = 1,  // Only log an error but parsing will succeed.
  kNone = 2,    // No UTF-8 check.
};

Utf8CheckMode GetUtf8CheckMode(const FieldDescriptor* field,
                               const Options& options);

void GenerateUtf8CheckCodeForString(const FieldDescriptor* field,
                                    const Options& options, bool for_parse,
                                    const char* parameters,
                                    const Formatter& format);

void GenerateUtf8CheckCodeForCord(const FieldDescriptor* field,
                                  const Options& options, bool for_parse,
                                  const char* parameters,
                                  const Formatter& format);

template <typename T>
struct FieldRangeImpl {
  struct Iterator {
    using iterator_category = std::forward_iterator_tag;
    using value_type = const FieldDescriptor*;
    using difference_type = int;

    value_type operator*() { return descriptor->field(idx); }

    friend bool operator==(const Iterator& a, const Iterator& b) {
      GOOGLE_DCHECK(a.descriptor == b.descriptor);
      return a.idx == b.idx;
    }
    friend bool operator!=(const Iterator& a, const Iterator& b) {
      return !(a == b);
    }

    Iterator& operator++() {
      idx++;
      return *this;
    }

    int idx;
    const T* descriptor;
  };

  Iterator begin() const { return {0, descriptor}; }
  Iterator end() const { return {descriptor->field_count(), descriptor}; }

  const T* descriptor;
};

template <typename T>
FieldRangeImpl<T> FieldRange(const T* desc) {
  return {desc};
}

struct OneOfRangeImpl {
  struct Iterator {
    using iterator_category = std::forward_iterator_tag;
    using value_type = const OneofDescriptor*;
    using difference_type = int;

    value_type operator*() { return descriptor->oneof_decl(idx); }

    friend bool operator==(const Iterator& a, const Iterator& b) {
      GOOGLE_DCHECK(a.descriptor == b.descriptor);
      return a.idx == b.idx;
    }
    friend bool operator!=(const Iterator& a, const Iterator& b) {
      return !(a == b);
    }

    Iterator& operator++() {
      idx++;
      return *this;
    }

    int idx;
    const Descriptor* descriptor;
  };

  Iterator begin() const { return {0, descriptor}; }
  Iterator end() const {
    return {descriptor->real_oneof_decl_count(), descriptor};
  }

  const Descriptor* descriptor;
};

inline OneOfRangeImpl OneOfRange(const Descriptor* desc) { return {desc}; }

PROTOC_EXPORT std::string StripProto(const std::string& filename);

bool EnableMessageOwnedArena(const Descriptor* desc, const Options& options);

bool EnableMessageOwnedArenaTrial(const Descriptor* desc,
                                  const Options& options);

bool ShouldVerify(const Descriptor* descriptor, const Options& options,
                  MessageSCCAnalyzer* scc_analyzer);
bool ShouldVerify(const FileDescriptor* file, const Options& options,
                  MessageSCCAnalyzer* scc_analyzer);

// Indicates whether to use predefined verify methods for a given message. If a
// message is "simple" and needs no special verification per field (e.g. message
// field, repeated packed, UTF8 string, etc.), we can use either VerifySimple or
// VerifySimpleAlwaysCheckInt32 methods as all verification can be done based on
// the wire type.
//
// Otherwise, we need "custom" verify methods tailored to a message to pass
// which field needs a special verification; i.e. InternalVerify.
enum class VerifySimpleType {
  kSimpleInt32Never,   // Use VerifySimple
  kSimpleInt32Always,  // Use VerifySimpleAlwaysCheckInt32
  kCustom,             // Use InternalVerify and check only for int32
  kCustomInt32Never,   // Use InternalVerify but never check for int32
  kCustomInt32Always,  // Use InternalVerify and always check for int32
};

// Returns VerifySimpleType if messages can be verified by predefined methods.
VerifySimpleType ShouldVerifySimple(const Descriptor* descriptor);

bool IsUtf8String(const FieldDescriptor* field);

bool HasMessageFieldOrExtension(const Descriptor* desc);

}  // namespace cpp
}  // namespace compiler
}  // namespace protobuf
}  // namespace google

#include <google/protobuf/port_undef.inc>

#endif  // GOOGLE_PROTOBUF_COMPILER_CPP_HELPERS_H__