From 15bc01f622d0eaf720c5ae4311cf7595e19c6288 Mon Sep 17 00:00:00 2001
From: Tom <thomas.hodson@ecmwf.int>
Date: Wed, 31 Jul 2024 11:37:42 +0100
Subject: [PATCH] Initial Implementation

---
 src/odc/codec/CodecOptimizer.h |  2 ++
 src/odc/codec/Integer.h        | 41 +++++++++++++++++++++++++++++++++-
 src/odc/codec/String.cc        |  1 +
 src/odc/codec/String.h         |  8 +++++++
 4 files changed, 51 insertions(+), 1 deletion(-)

diff --git a/src/odc/codec/CodecOptimizer.h b/src/odc/codec/CodecOptimizer.h
index 48ea6c9f..3e2d257e 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)
+                        codec = "long_constant_string";
 					else if(n < 256)
 						codec = "int8_string";
 					else if(n < 65536)
diff --git a/src/odc/codec/Integer.h b/src/odc/codec/Integer.h
index 90bc4a99..ca48c82f 100644
--- a/src/odc/codec/Integer.h
+++ b/src/odc/codec/Integer.h
@@ -121,7 +121,6 @@ class CodecIntegerOffset : public BaseCodecInteger<ByteOrder, ValueType> {
     }
 };
 
-
 //----------------------------------------------------------------------------------------------------------------------
 
 template <typename ByteOrder,
@@ -185,6 +184,46 @@ struct CodecInt32 : public CodecIntegerDirect<ByteOrder, ValueType, int32_t, Cod
     using CodecIntegerDirect<ByteOrder, ValueType, int32_t, CodecInt32<ByteOrder, ValueType>>::CodecIntegerDirect;
 };
 
+//----------------------------------------------------------------------------------------------------------------------
+//This is a codec with the same functionality as CodecInt8/16/32
+// Designed so that elsewhere we can do 
+// template<typename ByteOrder>
+// struct CodecLongConstantString : public IntStringCodecBase<ByteOrder, CodecZeroBitInteger<ByteOrder>>
+// CodecZeroBitInteger doesn't encode any data, it just asserts that the provided indices are all zero on encoding
+// and on decoding it just returns zero, which always returns the single string saved in the header by IntStringCodecBase
+
+template <typename ByteOrder, typename ValueType, class DerivedCodec>
+class CodecZeroBitInteger: public BaseCodecInteger<ByteOrder, ValueType> {
+
+public: // methods
+
+    CodecZeroBitInteger(api::ColumnType type) : BaseCodecInteger<ByteOrder, ValueType>(type, DerivedCodec::codec_name()) {}
+    ~CodecZeroBitInteger() override {}
+
+private: // methods
+
+    unsigned char* encode(unsigned char* p, const double& d) override {
+        static_assert(sizeof(ValueType) == sizeof(d), "unsafe casting check");
+        // Static cast to uint64_t ensures that the narrowing cast doesn't trigger the
+        // undefined behaviour sanitizer. It is correct that we can generate an integer
+        // that is too big, and throw it away, if invalid data gets here.
+        const ValueType& val(reinterpret_cast<const ValueType&>(d));
+        ASSERT(val == 0);
+        return p; // no data was encoded so this pointer is unchanged
+    }
+
+    void decode(double* out) override {
+        static_assert(sizeof(ValueType) == sizeof(out), "unsafe casting check");
+
+        ValueType* val_out = reinterpret_cast<ValueType*>(out);
+        (*val_out) = 0;
+    }
+
+    void skip() override {
+        // no data was encoded so we don't need to skip anything
+    }
+};
+
 //----------------------------------------------------------------------------------------------------------------------
 
 } // namespace codec
diff --git a/src/odc/codec/String.cc b/src/odc/codec/String.cc
index 0082ba69..17112b56 100644
--- a/src/odc/codec/String.cc
+++ b/src/odc/codec/String.cc
@@ -21,6 +21,7 @@ namespace codec {
 
 namespace {
     core::CodecBuilder<CodecChars> charsBuilder;
+    core::CodecBuilder<CodecLongConstantString> LongConstantStringBuilder;
     core::CodecBuilder<CodecInt8String> int8StringBuilder;
     core::CodecBuilder<CodecInt16String> int16StringBuilder;
 }
diff --git a/src/odc/codec/String.h b/src/odc/codec/String.h
index c17584f8..6accb7d4 100644
--- a/src/odc/codec/String.h
+++ b/src/odc/codec/String.h
@@ -210,6 +210,14 @@ class IntStringCodecBase : public CodecChars<ByteOrder> {
 
 //----------------------------------------------------------------------------------------------------------------------
 
+// This class uses the same machinery as CodecInt8String/CodecInt16String etc but it only ever stores one string in the header
+// CodecZeroBitInteger just doesn't encode any data and always returns index 0 when decoding.
+template<typename ByteOrder>
+struct CodecLongConstantString : public IntStringCodecBase<ByteOrder, CodecZeroBitInteger<ByteOrder, int64_t, CodecLongConstantString<ByteOrder>>> {
+    constexpr static const char* codec_name() { return "long_constant_string"; }
+    CodecLongConstantString(api::ColumnType type) : IntStringCodecBase<ByteOrder, CodecZeroBitInteger<ByteOrder, int64_t, CodecLongConstantString<ByteOrder>>>(type, codec_name()) {}
+    ~CodecLongConstantString() override {}
+};
 
 template<typename ByteOrder>
 struct CodecInt8String : public IntStringCodecBase<ByteOrder, CodecInt8<ByteOrder, int64_t>> {