From 8f52ebbbf333b4bafb1ce55560d2789c746d8ebc Mon Sep 17 00:00:00 2001
From: Tom <thomas.hodson@ecmwf.int>
Date: Tue, 6 Aug 2024 09:48:34 +0100
Subject: [PATCH] Add new LongConstantString Codec behind feature flag

The new codec is enabled for reading but odc will not write data with the new codec unless "ODC_ENABLE_WRITING_LONG_STRING_CODEC" is set in the environment. This is to give some time for installations in the wild to update before files containing the new code are produced by default.
---
 src/odc/codec/CodecOptimizer.h |  2 ++
 src/odc/codec/Constant.cc      |  1 +
 src/odc/codec/Constant.h       | 53 ++++++++++++++++++++++++++++++++++
 3 files changed, 56 insertions(+)

diff --git a/src/odc/codec/CodecOptimizer.h b/src/odc/codec/CodecOptimizer.h
index 48ea6c9f..42fb73eb 100644
--- a/src/odc/codec/CodecOptimizer.h
+++ b/src/odc/codec/CodecOptimizer.h
@@ -96,6 +96,8 @@ int CodecOptimizer::setOptimalCodecs(core::MetaData& columns)
                     n = col.coder().numStrings();
                                         if (n == 1 && col.coder().dataSizeDoubles() == 1)
 						codec = "constant_string";
+                    else if (n == 1 && std::getenv("ODC_ENABLE_WRITING_LONG_STRING_CODEC") != NULL)
+                        codec = "long_constant_string";
 					else if(n < 256)
 						codec = "int8_string";
 					else if(n < 65536)
diff --git a/src/odc/codec/Constant.cc b/src/odc/codec/Constant.cc
index cbcb1864..8596d0f0 100644
--- a/src/odc/codec/Constant.cc
+++ b/src/odc/codec/Constant.cc
@@ -23,6 +23,7 @@ namespace codec {
 namespace {
     core::IntegerCodecBuilder<CodecConstant> constantBuilder;
     core::CodecBuilder<CodecConstantString> constantStringBuilder;
+    core::CodecBuilder<CodecLongConstantString> LongConstantStringBuilder;
 }
 
 //----------------------------------------------------------------------------------------------------------------------
diff --git a/src/odc/codec/Constant.h b/src/odc/codec/Constant.h
index ffc18348..b655121c 100644
--- a/src/odc/codec/Constant.h
+++ b/src/odc/codec/Constant.h
@@ -16,6 +16,7 @@
 #define odc_core_codec_Constant_H
 
 #include "odc/core/Codec.h"
+#include "odc/codec/String.h"
 
 namespace odc {
 namespace codec {
@@ -74,6 +75,25 @@ class CodecConstantString : public CodecConstant<ByteOrder, double> {
     void save(core::DataStream<ByteOrder>& ds) override;
 };
 
+template<typename ByteOrder>
+class CodecLongConstantString : public CodecChars<ByteOrder> {
+
+public: // methods
+    constexpr static const char* codec_name() { return "long_constant_string"; }
+    CodecLongConstantString(api::ColumnType type) : CodecChars<ByteOrder>(type, codec_name()) {};
+    ~CodecLongConstantString() override {}
+
+private: // methods
+    unsigned char* encode(unsigned char* p, const double& d) override;
+    void decode(double* out) override;
+    void skip() override;
+
+    size_t numStrings() const override { return 1; }
+
+    void load(core::DataStream<ByteOrder>& ds) override;
+    void save(core::DataStream<ByteOrder>& ds) override;
+};
+
 //----------------------------------------------------------------------------------------------------------------------
 
 // Implementation of Constant
@@ -151,6 +171,39 @@ void CodecConstantString<ByteOrder>::print(std::ostream& s) const {
 
 //----------------------------------------------------------------------------------------------------------------------
 
+// Implementation of LongConstantString
+
+template <typename ByteOrder>
+unsigned char* CodecLongConstantString<ByteOrder>::encode(unsigned char* p, const double&) {
+    return p;
+}
+
+template <typename ByteOrder>
+void CodecLongConstantString<ByteOrder>::decode(double* out) {
+    ::memset(out, 0, this->decodedSizeDoubles_*sizeof(double));
+    ::memcpy(reinterpret_cast<char*>(out), &this->strings_[0][0], std::min(this->strings_[0].length(), this->decodedSizeDoubles_*sizeof(double)));
+}
+
+template <typename ByteOrder>
+void CodecLongConstantString<ByteOrder>::skip() {}
+
+template <typename ByteOrder>
+void CodecLongConstantString<ByteOrder>::load(core::DataStream<ByteOrder>& ds) {
+    core::DataStreamCodec<ByteOrder>::load(ds);
+    std::string s;
+    ds.read(s);
+    this->decodedSizeDoubles_ = ((s.length()-1) / sizeof(double)) +1;
+    this->strings_.push_back(s);
+}
+
+template <typename ByteOrder>
+void CodecLongConstantString<ByteOrder>::save(core::DataStream<ByteOrder>& ds) {
+    core::DataStreamCodec<ByteOrder>::save(ds);
+    ds.write(this->strings_[0]);
+}
+
+//----------------------------------------------------------------------------------------------------------------------
+
 } // namespace codec
 } // namespace odc