[ML] Release allocator cache

docs/CHANGELOG.asciidoc

@@ -28,11 +28,19 @@
 
 //=== Regressions
 
+== {es} version 8.14.1
+
+=== Enhancements
+
+* Improve memory allocation management for JSON processing to reduce memory usage.
+  (See {ml-pull}2676[#2676].)
+
 == {es} version 8.14.0
 
 === Bug Fixes
 
 * Remove ineffective optimizations for duplicate strings. (See {ml-pull}2652[#2652], issue: {ml-issue}2130[#2130].)
+* Use custom Boost.JSON resource allocator. (See {ml-pull}2674[#2674].)
 
 == {es} version 8.13.0
 

include/api/CJsonOutputWriter.h

@@ -273,6 +273,9 @@ class API_EXPORT CJsonOutputWriter {
     //! \p allocatorName A unique identifier for the allocator
     void pushAllocator(const std::string& allocatorName);
 
+    //! release the allocator
+    void releaseAllocator(const std::string& allocatorName);
+
     //! revert to using the previous allocator for JSON output processing
     void popAllocator();
 

include/core/CBoostJsonPoolAllocator.h

@@ -20,14 +20,34 @@ namespace json = boost::json;
 
 namespace ml {
 namespace core {
+
+namespace {
+
+class custom_resource : public boost::container::pmr::memory_resource {
+private:
+    void* do_allocate(std::size_t bytes, std::size_t /*align*/) override {
+        return ::operator new(bytes);
+    }
+
+    void do_deallocate(void* ptr, std::size_t /*bytes*/, std::size_t /*align*/) override {
+        return ::operator delete(ptr);
+    }
+
+    bool do_is_equal(memory_resource const& other) const noexcept override {
+        // since the global allocation and de-allocation functions are used,
+        // any instance of a custom_resource can deallocate memory allocated
+        // by another instance of a logging_resource
+        return dynamic_cast<custom_resource const*>(&other) != nullptr;
+    }
+};
+}
 //! \brief
-//! A boost::json memory allocator using a fixed size buffer
+//! A custom boost::json memory allocator
 //!
 //! DESCRIPTION:\n
-//! Encapsulates a boost::json monotonic_resource optimized with a fixed size buffer, see https://www.boost.org/doc/libs/1_83_0/libs/json/doc/html/json/allocators/storage_ptr.html
+//! Encapsulates a custom boost::json memory_resource, see https://www.boost.org/doc/libs/1_83_0/libs/json/doc/html/json/allocators/storage_ptr.html
 //!
 //! IMPLEMENTATION DECISIONS:\n
-//! Use a fixed size buffer for the allocator for performance reasons
 //!
 //! Retain documents created to ensure that the associated memory allocator exists for the documents
 //! lifetime
@@ -58,16 +78,12 @@ class CBoostJsonPoolAllocator {
     json::storage_ptr& get() { return m_JsonStoragePointer; }
 
 private:
-    //! Size of the fixed buffer to allocate for parsing JSON
-    static const size_t FIXED_BUFFER_SIZE = 4096;
-
-private:
-    //! fixed size memory buffer used to optimize allocator performance
-    unsigned char m_FixedBuffer[FIXED_BUFFER_SIZE];
-
     //! storage pointer to use for allocating boost::json objects
-    json::storage_ptr m_JsonStoragePointer{
-        json::make_shared_resource<json::monotonic_resource>(m_FixedBuffer)};
+    //! We use a custom resource allocator for more predictable
+    //! and timely allocation/de-allocations, see
+    //! https://www.boost.org/doc/libs/1_83_0/libs/json/doc/html/json/allocators/storage_ptr.html#json.allocators.storage_ptr.user_defined_resource
+    //! for more details.
+    json::storage_ptr m_JsonStoragePointer{json::make_shared_resource<custom_resource>()};
 
     //! Container used to persist boost::json documents
     TDocumentPtrVec m_JsonDocumentStore;

include/core/CBoostJsonWriterBase.h

@@ -24,10 +24,7 @@
 #include <boost/unordered_set.hpp>
 
 #include <cmath>
-#include <iomanip>
-#include <iostream>
 #include <memory>
-#include <regex>
 #include <stack>
 
 namespace json = boost::json;
@@ -134,6 +131,14 @@ class CBoostJsonWriterBase {
         return this->getAllocator()->get();
     }
 
+    void releaseAllocator(const std::string& allocatorName) {
+        if (m_AllocatorCache.find(allocatorName) != m_AllocatorCache.end()) {
+            TPoolAllocatorPtr allocator = m_AllocatorCache[allocatorName];
+            allocator.reset();
+            m_AllocatorCache.erase(allocatorName);
+        }
+    }
+
     bool isComplete() const {
         bool ret = m_Levels.empty() || m_Levels.top() == 0;
         return ret;

include/core/CScopedBoostJsonPoolAllocator.h

@@ -12,6 +12,7 @@
 #define INCLUDED_ml_core_CScopedBoostJsonPoolAllocator_h
 
 #include <boost/json.hpp>
+#include <string>
 
 namespace ml {
 namespace core {
@@ -31,14 +32,18 @@ class CScopedBoostJsonPoolAllocator {
     //! \p allocatorName Unique identifier for the allocator
     //! \p jsonOutputWriter JSON output writer that will make use of the allocator
     CScopedBoostJsonPoolAllocator(const std::string& allocatorName, T& writer)
-        : m_Writer(writer) {
+        : m_Writer(writer), m_AllocatorName(allocatorName) {
         m_Writer.pushAllocator(allocatorName);
     }
 
-    ~CScopedBoostJsonPoolAllocator() { m_Writer.popAllocator(); }
+    ~CScopedBoostJsonPoolAllocator() {
+        m_Writer.popAllocator();
+        m_Writer.releaseAllocator(m_AllocatorName);
+    }
 
 private:
     T& m_Writer;
+    std::string m_AllocatorName;
 };
 }
 }

lib/api/CJsonOutputWriter.cc

@@ -850,6 +850,10 @@ void CJsonOutputWriter::pushAllocator(const std::string& allocatorName) {
     m_Writer.pushAllocator(allocatorName);
 }
 
+void CJsonOutputWriter::releaseAllocator(const std::string& allocatorName) {
+    m_Writer.releaseAllocator(allocatorName);
+}
+
 void CJsonOutputWriter::popAllocator() {
     m_Writer.popAllocator();
 }