blob: d990e580499ad93746811234a48dc6ac8fc4f657 [file] [log] [blame]
// Copyright 2012 The Chromium Authors
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
#ifndef BASE_PICKLE_H_
#define BASE_PICKLE_H_
#include <stddef.h>
#include <stdint.h>
#include <optional>
#include <string>
#include <string_view>
#include "base/base_export.h"
#include "base/check_op.h"
#include "base/containers/span.h"
#include "base/gtest_prod_util.h"
#include "base/memory/raw_ptr_exclusion.h"
#include "base/memory/ref_counted.h"
#include "base/strings/string_piece.h"
namespace base {
class Pickle;
// PickleIterator reads data from a Pickle. The Pickle object must remain valid
// while the PickleIterator object is in use.
class BASE_EXPORT PickleIterator {
PickleIterator() : payload_(nullptr), read_index_(0), end_index_(0) {}
explicit PickleIterator(const Pickle& pickle);
// Methods for reading the payload of the Pickle. To read from the start of
// the Pickle, create a PickleIterator from a Pickle. If successful, these
// methods return true. Otherwise, false is returned to indicate that the
// result could not be extracted. It is not possible to read from the iterator
// after that.
[[nodiscard]] bool ReadBool(bool* result);
[[nodiscard]] bool ReadInt(int* result);
[[nodiscard]] bool ReadLong(long* result);
[[nodiscard]] bool ReadUInt16(uint16_t* result);
[[nodiscard]] bool ReadUInt32(uint32_t* result);
[[nodiscard]] bool ReadInt64(int64_t* result);
[[nodiscard]] bool ReadUInt64(uint64_t* result);
[[nodiscard]] bool ReadFloat(float* result);
[[nodiscard]] bool ReadDouble(double* result);
[[nodiscard]] bool ReadString(std::string* result);
// The StringPiece data will only be valid for the lifetime of the message.
[[nodiscard]] bool ReadStringPiece(StringPiece* result);
[[nodiscard]] bool ReadString16(std::u16string* result);
// The StringPiece16 data will only be valid for the lifetime of the message.
[[nodiscard]] bool ReadStringPiece16(StringPiece16* result);
// A pointer to the data will be placed in |*data|, and the length will be
// placed in |*length|. The pointer placed into |*data| points into the
// message's buffer so it will be scoped to the lifetime of the message (or
// until the message data is mutated). Do not keep the pointer around!
[[nodiscard]] bool ReadData(const char** data, size_t* length);
// Similar, but using base::span for convenience.
[[nodiscard]] std::optional<base::span<const uint8_t>> ReadData();
// A pointer to the data will be placed in |*data|. The caller specifies the
// number of bytes to read, and ReadBytes will validate this length. The
// pointer placed into |*data| points into the message's buffer so it will be
// scoped to the lifetime of the message (or until the message data is
// mutated). Do not keep the pointer around!
[[nodiscard]] bool ReadBytes(const char** data, size_t length);
// A version of ReadInt() that checks for the result not being negative. Use
// it for reading the object sizes.
[[nodiscard]] bool ReadLength(size_t* result) {
int result_int;
if (!ReadInt(&result_int) || result_int < 0)
return false;
*result = static_cast<size_t>(result_int);
return true;
// Skips bytes in the read buffer and returns true if there are at least
// num_bytes available. Otherwise, does nothing and returns false.
[[nodiscard]] bool SkipBytes(size_t num_bytes) {
return !!GetReadPointerAndAdvance(num_bytes);
bool ReachedEnd() const { return read_index_ == end_index_; }
// Read Type from Pickle.
template <typename Type>
bool ReadBuiltinType(Type* result);
// Advance read_index_ but do not allow it to exceed end_index_.
// Keeps read_index_ aligned.
void Advance(size_t size);
// Get read pointer for Type and advance read pointer.
template<typename Type>
const char* GetReadPointerAndAdvance();
// Get read pointer for |num_bytes| and advance read pointer. This method
// checks num_bytes for wrapping.
const char* GetReadPointerAndAdvance(size_t num_bytes);
// Get read pointer for (num_elements * size_element) bytes and advance read
// pointer. This method checks for overflow and wrapping.
const char* GetReadPointerAndAdvance(size_t num_elements,
size_t size_element);
const char* payload_; // Start of our pickle's payload.
size_t read_index_; // Offset of the next readable byte in payload.
size_t end_index_; // Payload size.
FRIEND_TEST_ALL_PREFIXES(PickleTest, GetReadPointerAndAdvance);
// This class provides facilities for basic binary value packing and unpacking.
// The Pickle class supports appending primitive values (ints, strings, etc.)
// to a pickle instance. The Pickle instance grows its internal memory buffer
// dynamically to hold the sequence of primitive values. The internal memory
// buffer is exposed as the "data" of the Pickle. This "data" can be passed
// to a Pickle object to initialize it for reading.
// When reading from a Pickle object, it is important for the consumer to know
// what value types to read and in what order to read them as the Pickle does
// not keep track of the type of data written to it.
// The Pickle's data has a header which contains the size of the Pickle's
// payload. It can optionally support additional space in the header. That
// space is controlled by the header_size parameter passed to the Pickle
// constructor.
class BASE_EXPORT Pickle {
// Auxiliary data attached to a Pickle. Pickle must be subclassed along with
// this interface in order to provide a concrete implementation of support
// for attachments. The base Pickle implementation does not accept
// attachments.
class BASE_EXPORT Attachment : public RefCountedThreadSafe<Attachment> {
Attachment(const Attachment&) = delete;
Attachment& operator=(const Attachment&) = delete;
friend class RefCountedThreadSafe<Attachment>;
virtual ~Attachment();
// Initialize a Pickle object using the default header size.
// Initialize a Pickle object with the specified header size in bytes, which
// must be greater-than-or-equal-to sizeof(Pickle::Header). The header size
// will be rounded up to ensure that the header size is 32bit-aligned.
explicit Pickle(size_t header_size);
// Initializes a Pickle from a const block of data. The data is not copied;
// instead the data is merely referenced by this Pickle. Only const methods
// should be used on the Pickle when initialized this way. The header
// padding size is deduced from the data length.
explicit Pickle(span<const uint8_t> data);
// TODO( Migrate callers of this overload to the span
// version.
Pickle(const char* data, size_t data_len);
// Initializes a Pickle as a deep copy of another Pickle.
Pickle(const Pickle& other);
// Note: Other classes are derived from this class, and they may well
// delete through this parent class, e.g. std::uniuqe_ptr<Pickle> exists
// in several places the code.
virtual ~Pickle();
// Performs a deep copy.
Pickle& operator=(const Pickle& other);
// Returns the number of bytes written in the Pickle, including the header.
size_t size() const {
return header_ ? header_size_ + header_->payload_size : 0;
// Returns the data for this Pickle.
const uint8_t* data() const {
return reinterpret_cast<const uint8_t*>(header_);
// Handy method to simplify calling data() with a reinterpret_cast.
const char* data_as_char() const {
return reinterpret_cast<const char*>(data());
// Returns the effective memory capacity of this Pickle, that is, the total
// number of bytes currently dynamically allocated or 0 in the case of a
// read-only Pickle. This should be used only for diagnostic / profiling
// purposes.
size_t GetTotalAllocatedSize() const;
// Methods for adding to the payload of the Pickle. These values are
// appended to the end of the Pickle's payload. When reading values from a
// Pickle, it is important to read them in the order in which they were added
// to the Pickle.
void WriteBool(bool value) { WriteInt(value ? 1 : 0); }
void WriteInt(int value) { WritePOD(value); }
void WriteLong(long value) {
// Always write long as a 64-bit value to ensure compatibility between
// 32-bit and 64-bit processes.
void WriteUInt16(uint16_t value) { WritePOD(value); }
void WriteUInt32(uint32_t value) { WritePOD(value); }
void WriteInt64(int64_t value) { WritePOD(value); }
void WriteUInt64(uint64_t value) { WritePOD(value); }
void WriteFloat(float value) { WritePOD(value); }
void WriteDouble(double value) { WritePOD(value); }
void WriteString(const StringPiece& value);
void WriteString16(const StringPiece16& value);
// "Data" is a blob with a length. When you read it out you will be given the
// length. See also WriteBytes.
// TODO( Migrate callers to the span versions.
void WriteData(const char* data, size_t length);
void WriteData(span<const uint8_t> data);
void WriteData(std::string_view data);
// "Bytes" is a blob with no length. The caller must specify the length both
// when reading and writing. It is normally used to serialize PoD types of a
// known size. See also WriteData.
// TODO( Migrate callers to the span version.
void WriteBytes(const void* data, size_t length);
void WriteBytes(span<const uint8_t> data);
// WriteAttachment appends |attachment| to the pickle. It returns
// false iff the set is full or if the Pickle implementation does not support
// attachments.
virtual bool WriteAttachment(scoped_refptr<Attachment> attachment);
// ReadAttachment parses an attachment given the parsing state |iter| and
// writes it to |*attachment|. It returns true on success.
virtual bool ReadAttachment(base::PickleIterator* iter,
scoped_refptr<Attachment>* attachment) const;
// Indicates whether the pickle has any attachments.
virtual bool HasAttachments() const;
// Reserves space for upcoming writes when multiple writes will be made and
// their sizes are computed in advance. It can be significantly faster to call
// Reserve() before calling WriteFoo() multiple times.
void Reserve(size_t additional_capacity);
// Payload follows after allocation of Header (header size is customizable).
struct Header {
uint32_t payload_size; // Specifies the size of the payload.
// Returns the header, cast to a user-specified type T. The type T must be a
// subclass of Header and its size must correspond to the header_size passed
// to the Pickle constructor.
template <class T>
T* headerT() {
DCHECK_EQ(header_size_, sizeof(T));
return static_cast<T*>(header_);
template <class T>
const T* headerT() const {
DCHECK_EQ(header_size_, sizeof(T));
return static_cast<const T*>(header_);
// The payload is the pickle data immediately following the header.
size_t payload_size() const {
return header_ ? header_->payload_size : 0;
base::span<const uint8_t> payload_bytes() const {
return base::as_bytes(base::make_span(payload(), payload_size()));
// Returns size of the header, which can have default value, set by user or
// calculated by passed raw data.
size_t header_size() const { return header_size_; }
const char* payload() const {
return reinterpret_cast<const char*>(header_) + header_size_;
// Returns the address of the byte immediately following the currently valid
// header + payload.
const char* end_of_payload() const {
// This object may be invalid.
return header_ ? payload() + payload_size() : NULL;
char* mutable_payload() {
return reinterpret_cast<char*>(header_) + header_size_;
size_t capacity_after_header() const {
return capacity_after_header_;
// Resize the capacity, note that the input value should not include the size
// of the header.
void Resize(size_t new_capacity);
// Claims |num_bytes| bytes of payload. This is similar to Reserve() in that
// it may grow the capacity, but it also advances the write offset of the
// pickle by |num_bytes|. Claimed memory, including padding, is zeroed.
// Returns the address of the first byte claimed.
void* ClaimBytes(size_t num_bytes);
// Find the end of the pickled data that starts at range_start. Returns NULL
// if the entire Pickle is not found in the given data range.
static const char* FindNext(size_t header_size,
const char* range_start,
const char* range_end);
// Parse pickle header and return total size of the pickle. Data range
// doesn't need to contain entire pickle.
// Returns true if pickle header was found and parsed. Callers must check
// returned |pickle_size| for sanity (against maximum message size, etc).
// NOTE: when function successfully parses a header, but encounters an
// overflow during pickle size calculation, it sets |pickle_size| to the
// maximum size_t value and returns true.
static bool PeekNext(size_t header_size,
const char* range_start,
const char* range_end,
size_t* pickle_size);
// The allocation granularity of the payload.
static const size_t kPayloadUnit;
friend class PickleIterator;
// `header_` is not a raw_ptr<...> for performance reasons (based on analysis
// of sampling profiler data).
RAW_PTR_EXCLUSION Header* header_;
size_t header_size_; // Supports extra data between header and payload.
// Allocation size of payload (or -1 if allocation is const). Note: this
// doesn't count the header.
size_t capacity_after_header_;
// The offset at which we will write the next field. Note: this doesn't count
// the header.
size_t write_offset_;
// Just like WriteBytes, but with a compile-time size, for performance.
template<size_t length> void BASE_EXPORT WriteBytesStatic(const void* data);
// Writes a POD by copying its bytes.
template <typename T> bool WritePOD(const T& data) {
return true;
inline void* ClaimUninitializedBytesInternal(size_t num_bytes);
inline void WriteBytesCommon(span<const uint8_t> data);
FRIEND_TEST_ALL_PREFIXES(PickleTest, DeepCopyResize);
FRIEND_TEST_ALL_PREFIXES(PickleTest, PeekNextOverflow);
FRIEND_TEST_ALL_PREFIXES(PickleTest, FindNextWithIncompleteHeader);
FRIEND_TEST_ALL_PREFIXES(PickleTest, FindNextOverflow);
} // namespace base
#endif // BASE_PICKLE_H_