serialize.h

// Copyright (c) 2013-2014 Sandstorm Development Group, Inc. and contributors
// Licensed under the MIT License:
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
// THE SOFTWARE.

// This file implements a simple serialization format for Cap'n Proto messages.  The format
// is as follows:
//
// * 32-bit little-endian segment count (4 bytes).
// * 32-bit little-endian size of each segment (4*(segment count) bytes).
// * Padding so that subsequent data is 64-bit-aligned (0 or 4 bytes).  (I.e., if there are an even
//     number of segments, there are 4 bytes of zeros here, otherwise there is no padding.)
// * Data from each segment, in order (8*sum(segment sizes) bytes)
//
// This format has some important properties:
// - It is self-delimiting, so multiple messages may be written to a stream without any external
//   delimiter.
// - The total size and position of each segment can be determined by reading only the first part
//   of the message, allowing lazy and random-access reading of the segment data.
// - A message is always at least 8 bytes.
// - A single-segment message can be read entirely in two system calls with no buffering.
// - A multi-segment message can be read entirely in three system calls with no buffering.
// - The format is appropriate for mmap()ing since all data is aligned.

#pragma once

#if defined(__GNUC__) && !defined(CAPNP_HEADER_WARNINGS)
#pragma GCC system_header
#endif

#include "message.h"
#include <kj/io.h>

namespace capnp {

class UnalignedFlatArrayMessageReader: public MessageReader {
  // Like FlatArrayMessageReader, but skips checking that the array is properly-aligned.
  //
  // WARNING: This only works on architectures that support unaligned reads, like x86/x64 and
  //   modern ARM. Unaligned access may incur a performance penalty on these platforms. On many
  //   other platforms, the program will simply crash on unaligned reads. Also note that unaligned
  //   data access may be considered undefined behavior by compilers; use at your own risk. If at
  //   all possible, try to ensure your data ends up in aligned buffers rather than rely on this
  //   class.

public:
  UnalignedFlatArrayMessageReader(
      kj::ArrayPtr<const word> array, ReaderOptions options = ReaderOptions());
  kj::ArrayPtr<const word> getSegment(uint id) override;
  const word* getEnd() const { return end; }

private:
  // Optimize for single-segment case.
  kj::ArrayPtr<const word> segment0;
  kj::Array<kj::ArrayPtr<const word>> moreSegments;
  const word* end;
};

class FlatArrayMessageReader: public UnalignedFlatArrayMessageReader {
  // Parses a message from a flat array.  Note that it makes sense to use this together with mmap()
  // for extremely fast parsing.

public:
  FlatArrayMessageReader(kj::ArrayPtr<const word> array, ReaderOptions options = ReaderOptions());
  // The array must remain valid until the MessageReader is destroyed.

  const word* getEnd() const { return UnalignedFlatArrayMessageReader::getEnd(); }
  // Get a pointer just past the end of the message as determined by reading the message header.
  // This could actually be before the end of the input array.  This pointer is useful e.g. if
  // you know that the input array has extra stuff appended after the message and you want to
  // get at it.

private:
  static kj::ArrayPtr<const word> checkAlignment(kj::ArrayPtr<const word> array);
};

kj::ArrayPtr<const word> initMessageBuilderFromFlatArrayCopy(
    kj::ArrayPtr<const word> array, MessageBuilder& target,
    ReaderOptions options = ReaderOptions());
// Convenience function which reads a message using `FlatArrayMessageReader` then copies the
// content into the target `MessageBuilder`, verifying that the message structure is valid
// (although not necessarily that it matches the desired schema).
//
// Returns an ArrayPtr containing any words left over in the array after consuming the whole
// message. This is useful when reading multiple messages that have been concatenated. See also
// FlatArrayMessageReader::getEnd().
//
// (Note that it's also possible to initialize a `MessageBuilder` directly without a copy using one
// of `MessageBuilder`'s constructors. However, this approach skips the validation step and is not
// safe to use on untrusted input. Therefore, we do not provide a convenience method for it.)

kj::Array<word> messageToFlatArray(MessageBuilder& builder);
// Constructs a flat array containing the entire content of the given message.
//
// To output the message as bytes, use `.asBytes()` on the returned word array. Keep in mind that
// `asBytes()` returns an ArrayPtr, so you have to save the Array as well to prevent it from being
// deleted. For example:
//
//     kj::Array<capnp::word> words = messageToFlatArray(myMessage);
//     kj::ArrayPtr<kj::byte> bytes = words.asBytes();
//     write(fd, bytes.begin(), bytes.size());

kj::Array<word> messageToFlatArray(kj::ArrayPtr<const kj::ArrayPtr<const word>> segments);
// Version of messageToFlatArray that takes a raw segment array.

size_t computeSerializedSizeInWords(MessageBuilder& builder);
// Returns the size, in words, that will be needed to serialize the message, including the header.

size_t computeSerializedSizeInWords(kj::ArrayPtr<const kj::ArrayPtr<const word>> segments);
// Version of computeSerializedSizeInWords that takes a raw segment array.

size_t expectedSizeInWordsFromPrefix(kj::ArrayPtr<const word> messagePrefix);
// Given a prefix of a serialized message, try to determine the expected total size of the message,
// in words. The returned size is based on the information known so far; it may be an underestimate
// if the prefix doesn't contain the full segment table.
//
// If the returned value is greater than `messagePrefix.size()`, then the message is not yet
// complete and the app cannot parse it yet. If the returned value is less than or equal to
// `messagePrefix.size()`, then the returned value is the exact total size of the message; any
// remaining bytes are part of the next message.
//
// This function is useful when reading messages from a stream in an asynchronous way, but when
// using the full KJ async infrastructure would be too difficult. Each time bytes are received,
// use this function to determine if an entire message is ready to be parsed.

// =======================================================================================

class InputStreamMessageReader: public MessageReader {
  // A MessageReader that reads from an abstract kj::InputStream. See also StreamFdMessageReader
  // for a subclass specific to file descriptors.

public:
  InputStreamMessageReader(kj::InputStream& inputStream,
                           ReaderOptions options = ReaderOptions(),
                           kj::ArrayPtr<word> scratchSpace = nullptr);
  ~InputStreamMessageReader() noexcept(false);

  // implements MessageReader ----------------------------------------
  kj::ArrayPtr<const word> getSegment(uint id) override;

private:
  kj::InputStream& inputStream;
  byte* readPos;

  // Optimize for single-segment case.
  kj::ArrayPtr<const word> segment0;
  kj::Array<kj::ArrayPtr<const word>> moreSegments;

  kj::Array<word> ownedSpace;
  // Only if scratchSpace wasn't big enough.

  kj::UnwindDetector unwindDetector;
};

void readMessageCopy(kj::InputStream& input, MessageBuilder& target,
                     ReaderOptions options = ReaderOptions(),
                     kj::ArrayPtr<word> scratchSpace = nullptr);
// Convenience function which reads a message using `InputStreamMessageReader` then copies the
// content into the target `MessageBuilder`, verifying that the message structure is valid
// (although not necessarily that it matches the desired schema).
//
// (Note that it's also possible to initialize a `MessageBuilder` directly without a copy using one
// of `MessageBuilder`'s constructors. However, this approach skips the validation step and is not
// safe to use on untrusted input. Therefore, we do not provide a convenience method for it.)

void writeMessage(kj::OutputStream& output, MessageBuilder& builder);
// Write the message to the given output stream.

void writeMessage(kj::OutputStream& output, kj::ArrayPtr<const kj::ArrayPtr<const word>> segments);
// Write the segment array to the given output stream.

// =======================================================================================
// Specializations for reading from / writing to file descriptors.

class StreamFdMessageReader: private kj::FdInputStream, public InputStreamMessageReader {
  // A MessageReader that reads from a stream-based file descriptor.

public:
  StreamFdMessageReader(int fd, ReaderOptions options = ReaderOptions(),
                        kj::ArrayPtr<word> scratchSpace = nullptr)
      : FdInputStream(fd), InputStreamMessageReader(*this, options, scratchSpace) {}
  // Read message from a file descriptor, without taking ownership of the descriptor.

  StreamFdMessageReader(kj::AutoCloseFd fd, ReaderOptions options = ReaderOptions(),
                        kj::ArrayPtr<word> scratchSpace = nullptr)
      : FdInputStream(kj::mv(fd)), InputStreamMessageReader(*this, options, scratchSpace) {}
  // Read a message from a file descriptor, taking ownership of the descriptor.

  ~StreamFdMessageReader() noexcept(false);
};

void readMessageCopyFromFd(int fd, MessageBuilder& target,
                           ReaderOptions options = ReaderOptions(),
                           kj::ArrayPtr<word> scratchSpace = nullptr);
// Convenience function which reads a message using `StreamFdMessageReader` then copies the
// content into the target `MessageBuilder`, verifying that the message structure is valid
// (although not necessarily that it matches the desired schema).
//
// (Note that it's also possible to initialize a `MessageBuilder` directly without a copy using one
// of `MessageBuilder`'s constructors. However, this approach skips the validation step and is not
// safe to use on untrusted input. Therefore, we do not provide a convenience method for it.)

void writeMessageToFd(int fd, MessageBuilder& builder);
// Write the message to the given file descriptor.
//
// This function throws an exception on any I/O error.  If your code is not exception-safe, be sure
// you catch this exception at the call site.  If throwing an exception is not acceptable, you
// can implement your own OutputStream with arbitrary error handling and then use writeMessage().

void writeMessageToFd(int fd, kj::ArrayPtr<const kj::ArrayPtr<const word>> segments);
// Write the segment array to the given file descriptor.
//
// This function throws an exception on any I/O error.  If your code is not exception-safe, be sure
// you catch this exception at the call site.  If throwing an exception is not acceptable, you
// can implement your own OutputStream with arbitrary error handling and then use writeMessage().

// =======================================================================================
// inline stuff

inline FlatArrayMessageReader::FlatArrayMessageReader(
    kj::ArrayPtr<const word> array, ReaderOptions options)
#ifdef KJ_DEBUG
    : UnalignedFlatArrayMessageReader(checkAlignment(array), options) {}
#else
    : UnalignedFlatArrayMessageReader(array, options) {}
#endif

inline kj::Array<word> messageToFlatArray(MessageBuilder& builder) {
  return messageToFlatArray(builder.getSegmentsForOutput());
}

inline size_t computeSerializedSizeInWords(MessageBuilder& builder) {
  return computeSerializedSizeInWords(builder.getSegmentsForOutput());
}

inline void writeMessage(kj::OutputStream& output, MessageBuilder& builder) {
  writeMessage(output, builder.getSegmentsForOutput());
}

inline void writeMessageToFd(int fd, MessageBuilder& builder) {
  writeMessageToFd(fd, builder.getSegmentsForOutput());
}

}  // namespace capnp