intermediate changes

ref:cde9a383711a11544ce7e107a78147fb96cc4029

1 files changed, 92 insertions, 0 deletions

diff --git a/library/cpp/monlib/encode/buffered/string_pool.h b/library/cpp/monlib/encode/buffered/string_pool.h
new file mode 100644
index 0000000000..00e5644608
--- /dev/null
+++ b/library/cpp/monlib/encode/buffered/string_pool.h
@@ -0,0 +1,92 @@
+#pragma once
+
+#include <util/generic/hash.h>
+#include <util/generic/vector.h>
+
+namespace NMonitoring {
+    ////////////////////////////////////////////////////////////////////////////////
+    // TStringPoolBuilder
+    ////////////////////////////////////////////////////////////////////////////////
+    class TStringPoolBuilder {
+    public:
+        struct TValue: TNonCopyable {
+            TValue(ui32 idx, ui32 freq)
+                : Index{idx}
+                , Frequency{freq}
+            {
+            }
+
+            ui32 Index;
+            ui32 Frequency;
+        };
+
+    public:
+        const TValue* PutIfAbsent(TStringBuf str);
+        const TValue* GetByIndex(ui32 index) const;
+
+        /// Determines whether pool must be sorted by value frequencies
+        TStringPoolBuilder& SetSorted(bool sorted) {
+            RequiresSorting_ = sorted;
+            return *this;
+        }
+
+        TStringPoolBuilder& Build();
+
+        TStringBuf Get(ui32 index) const {
+            Y_ENSURE(IsBuilt_, "Pool must be sorted first");
+            return StrVector_.at(index).first;
+        }
+
+        TStringBuf Get(const TValue* value) const {
+            return StrVector_.at(value->Index).first;
+        }
+
+        template <typename TConsumer>
+        void ForEach(TConsumer&& c) {
+            Y_ENSURE(IsBuilt_, "Pool must be sorted first");
+            for (const auto& value : StrVector_) {
+                c(value.first, value.second->Index, value.second->Frequency);
+            }
+        }
+
+        size_t BytesSize() const noexcept {
+            return BytesSize_;
+        }
+
+        size_t Count() const noexcept {
+            return StrMap_.size();
+        }
+
+    private:
+        THashMap<TString, TValue> StrMap_;
+        TVector<std::pair<TStringBuf, TValue*>> StrVector_;
+        bool RequiresSorting_ = false;
+        bool IsBuilt_ = false;
+        size_t BytesSize_ = 0;
+    };
+
+    ////////////////////////////////////////////////////////////////////////////////
+    // TStringPool
+    ////////////////////////////////////////////////////////////////////////////////
+    class TStringPool {
+    public:
+        TStringPool(const char* data, ui32 size) {
+            InitIndex(data, size);
+        }
+
+        TStringBuf Get(ui32 i) const {
+            return Index_.at(i);
+        }
+
+        size_t Size() const {
+            return Index_.size();
+        }
+
+    private:
+        void InitIndex(const char* data, ui32 size);
+
+    private:
+        TVector<TStringBuf> Index_;
+    };
+
+}