20#ifndef COMPACT_THETA_SKETCH_PARSER_IMPL_HPP_
21#define COMPACT_THETA_SKETCH_PARSER_IMPL_HPP_
30T whole_bytes_to_hold_bits(T bits) {
31 static_assert(std::is_integral<T>::value,
"integral type expected");
32 return (bits >> 3) + ((bits & 7) > 0);
36auto compact_theta_sketch_parser<dummy>::parse(
const void* ptr,
size_t size, uint64_t seed,
bool dump_on_error) -> compact_theta_sketch_data {
37 check_memory_size(ptr, size, 8, dump_on_error);
38 checker<true>::check_sketch_type(
reinterpret_cast<const uint8_t*
>(ptr)[COMPACT_SKETCH_TYPE_BYTE], COMPACT_SKETCH_TYPE);
39 uint8_t serial_version =
reinterpret_cast<const uint8_t*
>(ptr)[COMPACT_SKETCH_SERIAL_VERSION_BYTE];
40 switch(serial_version) {
43 const uint16_t seed_hash =
reinterpret_cast<const uint16_t*
>(ptr)[COMPACT_SKETCH_SEED_HASH_U16];
44 checker<true>::check_seed_hash(seed_hash, compute_seed_hash(seed));
45 const bool has_theta =
reinterpret_cast<const uint8_t*
>(ptr)[COMPACT_SKETCH_PRE_LONGS_BYTE] > 1;
48 check_memory_size(ptr, size, 16, dump_on_error);
49 theta =
reinterpret_cast<const uint64_t*
>(ptr)[COMPACT_SKETCH_V4_THETA_U64];
51 const uint8_t num_entries_bytes =
reinterpret_cast<const uint8_t*
>(ptr)[COMPACT_SKETCH_V4_NUM_ENTRIES_BYTES_BYTE];
52 check_v4_num_entries_bytes(num_entries_bytes);
53 size_t data_offset_bytes = has_theta ? COMPACT_SKETCH_V4_PACKED_DATA_ESTIMATION_BYTE : COMPACT_SKETCH_V4_PACKED_DATA_EXACT_BYTE;
54 check_memory_size(ptr, size, data_offset_bytes + num_entries_bytes, dump_on_error);
55 uint32_t num_entries = 0;
56 const uint8_t* num_entries_ptr =
reinterpret_cast<const uint8_t*
>(ptr) + data_offset_bytes;
57 for (
unsigned i = 0; i < num_entries_bytes; ++i) {
58 num_entries |= (*num_entries_ptr++) << (i << 3);
60 data_offset_bytes += num_entries_bytes;
61 const uint8_t entry_bits =
reinterpret_cast<const uint8_t*
>(ptr)[COMPACT_SKETCH_V4_ENTRY_BITS_BYTE];
62 check_v4_entry_bits(entry_bits);
63 const uint64_t expected_bits =
static_cast<uint64_t
>(entry_bits) * num_entries;
64 const size_t expected_size_bytes = data_offset_bytes + whole_bytes_to_hold_bits(expected_bits);
65 check_memory_size(ptr, size, expected_size_bytes, dump_on_error);
66 return {
false,
true, seed_hash, num_entries, theta,
67 reinterpret_cast<const uint8_t*
>(ptr) + data_offset_bytes, entry_bits};
71 const uint16_t seed_hash =
reinterpret_cast<const uint16_t*
>(ptr)[COMPACT_SKETCH_SEED_HASH_U16];
72 if (
reinterpret_cast<const uint8_t*
>(ptr)[COMPACT_SKETCH_FLAGS_BYTE] & (1 << COMPACT_SKETCH_IS_EMPTY_FLAG)) {
73 return {
true,
true, seed_hash, 0, theta,
nullptr, 64};
75 checker<true>::check_seed_hash(seed_hash, compute_seed_hash(seed));
76 const bool has_theta =
reinterpret_cast<const uint8_t*
>(ptr)[COMPACT_SKETCH_PRE_LONGS_BYTE] > 2;
78 check_memory_size(ptr, size, (COMPACT_SKETCH_THETA_U64 + 1) *
sizeof(uint64_t), dump_on_error);
79 theta =
reinterpret_cast<const uint64_t*
>(ptr)[COMPACT_SKETCH_THETA_U64];
81 if (
reinterpret_cast<const uint8_t*
>(ptr)[COMPACT_SKETCH_PRE_LONGS_BYTE] == 1) {
82 check_memory_size(ptr, size, 16, dump_on_error);
83 return {
false,
true, seed_hash, 1, theta,
reinterpret_cast<const uint64_t*
>(ptr) + COMPACT_SKETCH_SINGLE_ENTRY_U64, 64};
85 const size_t entries_start_u64 = has_theta ? COMPACT_SKETCH_ENTRIES_ESTIMATION_U64 : COMPACT_SKETCH_ENTRIES_EXACT_U64;
86 check_memory_size(ptr, size, entries_start_u64 *
sizeof(uint64_t), dump_on_error);
87 const uint32_t num_entries =
reinterpret_cast<const uint32_t*
>(ptr)[COMPACT_SKETCH_NUM_ENTRIES_U32];
88 const uint64_t* entries =
reinterpret_cast<const uint64_t*
>(ptr) + entries_start_u64;
89 const size_t expected_size_bytes = (entries_start_u64 + num_entries) *
sizeof(uint64_t);
90 check_memory_size(ptr, size, expected_size_bytes, dump_on_error);
91 const bool is_ordered =
reinterpret_cast<const uint8_t*
>(ptr)[COMPACT_SKETCH_FLAGS_BYTE] & (1 << COMPACT_SKETCH_IS_ORDERED_FLAG);
92 return {
false, is_ordered, seed_hash, num_entries, theta, entries, 64};
95 uint16_t seed_hash = compute_seed_hash(seed);
96 check_memory_size(ptr, size, COMPACT_SKETCH_ENTRIES_ESTIMATION_U64 *
sizeof(uint64_t), dump_on_error);
97 const uint32_t num_entries =
reinterpret_cast<const uint32_t*
>(ptr)[COMPACT_SKETCH_NUM_ENTRIES_U32];
98 uint64_t theta =
reinterpret_cast<const uint64_t*
>(ptr)[COMPACT_SKETCH_THETA_U64];
100 if (is_empty)
return {
true,
true, seed_hash, 0, theta,
nullptr, 64};
101 const uint64_t* entries =
reinterpret_cast<const uint64_t*
>(ptr) + COMPACT_SKETCH_ENTRIES_ESTIMATION_U64;
102 const size_t expected_size_bytes = (COMPACT_SKETCH_ENTRIES_ESTIMATION_U64 + num_entries) *
sizeof(uint64_t);
103 check_memory_size(ptr, size, expected_size_bytes, dump_on_error);
104 return {
false,
true, seed_hash, num_entries, theta, entries, 64};
107 uint8_t preamble_size =
reinterpret_cast<const uint8_t*
>(ptr)[COMPACT_SKETCH_PRE_LONGS_BYTE];
108 const uint16_t seed_hash =
reinterpret_cast<const uint16_t*
>(ptr)[COMPACT_SKETCH_SEED_HASH_U16];
109 checker<true>::check_seed_hash(seed_hash, compute_seed_hash(seed));
110 if (preamble_size == 1) {
112 }
else if (preamble_size == 2) {
113 check_memory_size(ptr, size, COMPACT_SKETCH_ENTRIES_EXACT_U64 *
sizeof(uint64_t), dump_on_error);
114 const uint32_t num_entries =
reinterpret_cast<const uint32_t*
>(ptr)[COMPACT_SKETCH_NUM_ENTRIES_U32];
115 if (num_entries == 0) {
118 const size_t expected_size_bytes = (preamble_size +
static_cast<size_t>(num_entries)) << 3;
119 check_memory_size(ptr, size, expected_size_bytes, dump_on_error);
120 const uint64_t* entries =
reinterpret_cast<const uint64_t*
>(ptr) + COMPACT_SKETCH_ENTRIES_EXACT_U64;
123 }
else if (preamble_size == 3) {
124 check_memory_size(ptr, size, COMPACT_SKETCH_ENTRIES_ESTIMATION_U64 *
sizeof(uint64_t), dump_on_error);
125 const uint32_t num_entries =
reinterpret_cast<const uint32_t*
>(ptr)[COMPACT_SKETCH_NUM_ENTRIES_U32];
126 uint64_t theta =
reinterpret_cast<const uint64_t*
>(ptr)[COMPACT_SKETCH_THETA_U64];
128 if (is_empty)
return {
true,
true, seed_hash, 0, theta,
nullptr, 64};
129 const uint64_t* entries =
reinterpret_cast<const uint64_t*
>(ptr) + COMPACT_SKETCH_ENTRIES_ESTIMATION_U64;
130 const size_t expected_size_bytes = (COMPACT_SKETCH_ENTRIES_ESTIMATION_U64 + num_entries) *
sizeof(uint64_t);
131 check_memory_size(ptr, size, expected_size_bytes, dump_on_error);
132 return {
false,
true, seed_hash, num_entries, theta, entries, 64};
134 throw std::invalid_argument(std::to_string(preamble_size) +
" longs of premable, but expected 1, 2, or 3");
138 throw std::invalid_argument(
"unsupported serial version " + std::to_string(serial_version));
143void compact_theta_sketch_parser<dummy>::check_memory_size(
const void* ptr,
size_t actual_bytes,
size_t expected_bytes,
bool dump_on_error) {
144 if (actual_bytes < expected_bytes)
throw std::out_of_range(
"at least " + std::to_string(expected_bytes)
145 +
" bytes expected, actual " + std::to_string(actual_bytes)
146 + (dump_on_error ? (
", sketch dump: " + hex_dump(
reinterpret_cast<const uint8_t*
>(ptr), actual_bytes)) :
""));
150void compact_theta_sketch_parser<dummy>::check_v4_entry_bits(uint8_t entry_bits) {
152 if (entry_bits == 0 || entry_bits > 63) {
153 throw std::invalid_argument(
"entry bits must be in [1, 63], actual " + std::to_string(entry_bits));
158void compact_theta_sketch_parser<dummy>::check_v4_num_entries_bytes(uint8_t num_entries_bytes) {
159 if (num_entries_bytes == 0 || num_entries_bytes >
sizeof(uint32_t)) {
160 throw std::invalid_argument(
"num entries bytes must be in [1, 4], actual " + std::to_string(num_entries_bytes));
165std::string compact_theta_sketch_parser<dummy>::hex_dump(
const uint8_t* ptr,
size_t size) {
167 s << std::hex << std::setfill(
'0') << std::uppercase;
168 for (
size_t i = 0; i < size; ++i) s << std::setw(2) << (ptr[i] & 0xff);
const uint64_t MAX_THETA
max theta - signed max for compatibility with Java
Definition theta_constants.hpp:36
DataSketches namespace.
Definition binomial_bounds.hpp:38