snapshots: move encoding modules to common (#2374)

silkworm/db/snapshots/rec_split/encoding/elias_fano.hpp → silkworm/db/snapshots/common/encoding/elias_fano.hpp

@@ -58,16 +58,17 @@
 #include <silkworm/core/common/base.hpp>
 #include <silkworm/core/common/bytes.hpp>
 #include <silkworm/core/common/endian.hpp>
-#include <silkworm/db/snapshots/rec_split/common/common.hpp>
-#include <silkworm/db/snapshots/rec_split/encoding/sequence.hpp>
 #include <silkworm/infra/common/ensure.hpp>
 #include <silkworm/infra/common/log.hpp>
 
+#include "sequence.hpp"
+#include "util.hpp"
+
 // EliasFano algo overview https://www.antoniomallia.it/sorted-integers-compression-with-elias-fano-encoding.html
 // P. Elias. Efficient storage and retrieval by content and address of static files. J. ACM, 21(2):246–260, 1974.
 // Partitioned Elias-Fano Indexes http://groups.di.unipi.it/~ottavian/files/elias_fano_sigir14.pdf
 
-namespace silkworm::snapshots::rec_split::encoding {
+namespace silkworm::snapshots::encoding {
 
 //! Log2Q = Log2(Quantum)
 static constexpr uint64_t kLog2q = 8;
@@ -569,4 +570,4 @@ class DoubleEliasFanoList16 {
     }
 };
 
-}  // namespace silkworm::snapshots::rec_split::encoding
+}  // namespace silkworm::snapshots::encoding

silkworm/db/snapshots/rec_split/encoding/elias_fano_test.cpp → silkworm/db/snapshots/common/encoding/elias_fano_test.cpp

@@ -26,7 +26,7 @@
 #include <silkworm/core/common/util.hpp>
 #include <silkworm/infra/test_util/log.hpp>
 
-namespace silkworm::snapshots::rec_split::encoding {
+namespace silkworm::snapshots::encoding {
 
 struct EliasFanoList32Test {
     std::vector<uint64_t> offsets;
@@ -168,4 +168,4 @@ TEST_CASE("DoubleEliasFanoList16", "[silkworm][recsplit][elias_fano]") {
                     "0000000000000000010000000000000000000000000000000000000000000000"));
 }
 
-}  // namespace silkworm::snapshots::rec_split::encoding
+}  // namespace silkworm::snapshots::encoding

silkworm/db/snapshots/rec_split/encoding/golomb_rice.hpp → silkworm/db/snapshots/common/encoding/golomb_rice.hpp

@@ -50,11 +50,12 @@
 #include <iostream>
 
 #include <silkworm/core/common/assert.hpp>
-#include <silkworm/db/snapshots/rec_split/common/common.hpp>
-#include <silkworm/db/snapshots/rec_split/encoding/sequence.hpp>
 #include <silkworm/infra/common/log.hpp>
 
-namespace silkworm::snapshots::rec_split::encoding {
+#include "sequence.hpp"
+#include "util.hpp"
+
+namespace silkworm::snapshots::encoding {
 
 //! Storage for Golomb-Rice codes of a RecSplit bucket.
 class GolombRiceVector {
@@ -253,4 +254,4 @@ class GolombRiceVector {
     }
 };
 
-}  // namespace silkworm::snapshots::rec_split::encoding
+}  // namespace silkworm::snapshots::encoding

silkworm/db/snapshots/rec_split/encoding/golomb_rice_test.cpp → silkworm/db/snapshots/common/encoding/golomb_rice_test.cpp

@@ -21,10 +21,9 @@
 #include <catch2/catch_test_macros.hpp>
 
 #include <silkworm/core/common/random_number.hpp>
-#include <silkworm/db/snapshots/rec_split/encoding/sequence.hpp>
 #include <silkworm/infra/test_util/log.hpp>
 
-namespace silkworm::snapshots::rec_split::encoding {
+namespace silkworm::snapshots::encoding {
 
 static const std::size_t kGolombRiceTestNumKeys{128};
 static const std::size_t kGolombRiceTestNumTrees{1'000};
@@ -86,4 +85,4 @@ TEST_CASE("GolombRiceVector", "[silkworm][recsplit][golomb_rice]") {
     }
 }
 
-}  // namespace silkworm::snapshots::rec_split::encoding
+}  // namespace silkworm::snapshots::encoding

silkworm/db/snapshots/rec_split/encoding/sequence.hpp → silkworm/db/snapshots/common/encoding/sequence.hpp

@@ -25,7 +25,7 @@
 #include <silkworm/core/common/endian.hpp>
 #include <silkworm/infra/common/ensure.hpp>
 
-namespace silkworm::snapshots::rec_split::encoding {
+namespace silkworm::snapshots::encoding {
 
 template <UnsignedIntegral T>
 using UnsignedIntegralSequence = std::vector<T>;
@@ -64,4 +64,4 @@ std::istream& operator>>(std::istream& is, UnsignedIntegralSequence<T>& s) {
     return is;
 }
 
-}  // namespace silkworm::snapshots::rec_split::encoding
+}  // namespace silkworm::snapshots::encoding

silkworm/db/snapshots/rec_split/encoding/sequence_test.cpp → silkworm/db/snapshots/common/encoding/sequence_test.cpp

@@ -25,7 +25,7 @@
 #include <silkworm/core/common/endian.hpp>
 #include <silkworm/infra/test_util/log.hpp>
 
-namespace silkworm::snapshots::rec_split::encoding {
+namespace silkworm::snapshots::encoding {
 
 TEST_CASE("Uint64Sequence", "[silkworm][snapshots][recsplit][sequence]") {
     test_util::SetLogVerbosityGuard guard{log::Level::kNone};
@@ -52,4 +52,4 @@ TEST_CASE("Uint64Sequence: size too big", "[silkworm][snapshots][recsplit][seque
     CHECK_THROWS_AS((ss >> input_sequence), std::logic_error);
 }
 
-}  // namespace silkworm::snapshots::rec_split::encoding
+}  // namespace silkworm::snapshots::encoding

silkworm/db/snapshots/rec_split/common/common.hpp → silkworm/db/snapshots/common/encoding/util.hpp

@@ -58,7 +58,7 @@
 
 #include <silkworm/core/common/assert.hpp>
 
-namespace silkworm::snapshots::rec_split {
+namespace silkworm::snapshots::encoding {
 
 using std::memcpy;
 
@@ -206,4 +206,4 @@ inline uint64_t select64(uint64_t x, uint64_t k) {
 #endif
 }
 
-}  // namespace silkworm::snapshots::rec_split
+}  // namespace silkworm::snapshots::encoding

silkworm/db/snapshots/index/btree_index.hpp

@@ -23,13 +23,13 @@
 
 #include <silkworm/infra/common/memory_mapped_file.hpp>
 
-#include "../rec_split/encoding/elias_fano.hpp"  // TODO(canepat) move to snapshots/common
+#include "../common/encoding/elias_fano.hpp"
 #include "../seg/decompressor.hpp"
 #include "btree.hpp"
 
 namespace silkworm::snapshots::index {
 
-using rec_split::encoding::EliasFanoList32;  // TODO(canepat) remove after moving
+using encoding::EliasFanoList32;
 
 class BTreeIndex {
   public:

silkworm/db/snapshots/rec_split/rec_split.hpp

@@ -72,9 +72,9 @@
 #include <silkworm/core/common/math.hpp>
 #include <silkworm/core/common/util.hpp>
 #include <silkworm/db/snapshots/common/bitmask_operators.hpp>
-#include <silkworm/db/snapshots/rec_split/common/murmur_hash3.hpp>
-#include <silkworm/db/snapshots/rec_split/encoding/elias_fano.hpp>
-#include <silkworm/db/snapshots/rec_split/encoding/golomb_rice.hpp>
+#include <silkworm/db/snapshots/common/encoding/elias_fano.hpp>
+#include <silkworm/db/snapshots/common/encoding/golomb_rice.hpp>
+#include <silkworm/db/snapshots/rec_split/murmur_hash3.hpp>
 #include <silkworm/infra/common/directories.hpp>
 #include <silkworm/infra/common/ensure.hpp>
 #include <silkworm/infra/common/log.hpp>
@@ -87,6 +87,7 @@
 namespace silkworm::snapshots::rec_split {
 
 using namespace std::chrono;
+using encoding::remap16, encoding::remap128;
 
 //! Assumed *maximum* size of a bucket. Works with high probability up to average bucket size ~2000
 static const int kMaxBucketSize = 3000;