From 25a4a3b76f6b53114aab2150cf9a60094608f6aa Mon Sep 17 00:00:00 2001
From: Viktor Govako <viktor.govako@gmail.com>
Date: Sat, 5 Jul 2025 11:19:30 -0300
Subject: [PATCH] [coding] Added varint short arrays ser/des.

Signed-off-by: Viktor Govako <viktor.govako@gmail.com>

# Conflicts:
#	libs/coding/coding_tests/varint_test.cpp
---
 libs/coding/coding_tests/varint_test.cpp | 29 ++++++++++++-
 libs/coding/varint.hpp                   | 54 +++++++++++++++++++++---
 2 files changed, 77 insertions(+), 6 deletions(-)

diff --git a/libs/coding/coding_tests/varint_test.cpp b/libs/coding/coding_tests/varint_test.cpp
index f18278347..4f28e18ac 100644
--- a/libs/coding/coding_tests/varint_test.cpp
+++ b/libs/coding/coding_tests/varint_test.cpp
@@ -1,7 +1,8 @@
-#include "coding/varint.hpp"
 #include "testing/testing.hpp"
 
 #include "coding/byte_stream.hpp"
+#include "coding/reader.hpp"
+#include "coding/varint.hpp"
 
 #include "base/macros.hpp"
 #include "base/stl_helpers.hpp"
@@ -185,3 +186,29 @@ UNIT_TEST(ReadVarInt64Array)
     }
   }
 }
+
+UNIT_TEST(VarInt_ShortSortedArray)
+{
+  uint32_t constexpr maxVal = (uint32_t(1) << 30) - 1;
+  std::vector<uint32_t> samples[] = {
+      {0},
+      {10, 10000},
+      {maxVal - 2, maxVal - 1, maxVal},
+  };
+
+  for (auto const & s : samples)
+  {
+    std::vector<uint8_t> buffer;
+    PushBackByteSink sink(buffer);
+
+    WriteVarUInt32SortedShortArray(s, sink);
+
+    MemReader reader(buffer.data(), buffer.size());
+    ReaderSource src(reader);
+
+    std::vector<uint32_t> actual;
+    ReadVarUInt32SortedShortArray(src, actual);
+
+    TEST_EQUAL(s, actual, ());
+  }
+}
diff --git a/libs/coding/varint.hpp b/libs/coding/varint.hpp
index b34430118..e401f7494 100644
--- a/libs/coding/varint.hpp
+++ b/libs/coding/varint.hpp
@@ -3,13 +3,10 @@
 #include "coding/write_to_sink.hpp"
 
 #include "base/assert.hpp"
-#include "base/base.hpp"
 #include "base/bits.hpp"
 #include "base/exception.hpp"
 #include "base/stl_helpers.hpp"
 
-#include <cstddef>
-#include <cstdint>
 #include <type_traits>
 
 // Writes any unsigned integer type using optimal bytes count, platform-independent.
@@ -296,6 +293,53 @@ void const * ReadVarUint64Array(void const * pBeg, size_t count, F f)
 template <class Cont, class Sink>
 void WriteVarUintArray(Cont const & v, Sink & sink)
 {
-  for (size_t i = 0; i != v.size(); ++i)
-    WriteVarUint(sink, v[i]);
+  for (auto e : v)
+    WriteVarUint(sink, e);
 }
+
+/// @name Special functions to read/write varint array when each value stores a control bit "has next".
+/// Useful for short (>90% have 1 value) arrays.
+/// @{
+template <class Sink>
+void WriteVarUInt32SortedShortArray(std::vector<uint32_t> const & cont, Sink & sink)
+{
+  CHECK(!cont.empty(), ());
+  CHECK(base::IsSortedAndUnique(cont), ());
+
+  uint32_t prev = 0;
+  size_t const sz = cont.size();
+  for (size_t i = 0; i < sz; ++i)
+  {
+    uint32_t v = cont[i] - prev;
+    prev = cont[i];
+
+    CHECK((v & (uint32_t(3) << 30)) == 0, ());  // leading 2 bits are zeros
+    v <<= 1;
+    // set 1-bit if have a next value
+    if (i + 1 < sz)
+      v |= 0x1;
+
+    WriteVarUint(sink, v);
+  }
+}
+
+/// @todo Refactor on raw pointers (same as ReadXXX functions above).
+/// ArrayByteSource isn't as fast as pointers :)
+template <class Source, class ContT>
+void ReadVarUInt32SortedShortArray(Source & src, ContT & cont)
+{
+  cont.clear();
+
+  bool hasNext = false;
+  uint32_t prev = 0;
+  do
+  {
+    uint32_t const v = ReadVarUint<uint32_t>(src);
+    hasNext = (v & 0x1) != 0;
+
+    prev += (v >> 1);
+    cont.push_back(prev);
+  }
+  while (hasNext);
+}
+/// @}