Patch RapidJSON to make sure it builds on all platforms (#8818)

* patch RapidJSON to make sure it builds on all platforms

* remove unnecessary sed command from Docker recipe

---------

Co-authored-by: ptosco <paolo.tosco@novartis.com>

.gitignore

@@ -76,7 +76,7 @@ __pycache__/
 /External/RingFamilies/RingDecomposerLib/
 /External/YAeHMOP/yaehmop/
 /External/catch/catch/
-/External/rapidjson*/
+/External/rapidjson/rapidjson-*/
 
 /rdkit/RDPaths.py
 /rdkit/Chem/inchi.py

Code/GraphMol/MolInterchange/CMakeLists.txt

@@ -1,24 +1,23 @@
 
-if(NOT EXISTS "${RDKit_ExternalDir}/rapidjson-1.1.0")
+if(NOT EXISTS "${RDKit_ExternalDir}/rapidjson/rapidjson-1.1.0")
     downloadAndCheckMD5("https://github.com/Tencent/rapidjson/archive/v1.1.0.tar.gz"
-          "${RDKit_ExternalDir}/rapidjson-1.1.0.tar.gz"
+          "${RDKit_ExternalDir}/rapidjson/rapidjson-1.1.0.tar.gz"
           "badd12c511e081fec6c89c43a7027bce")
     execute_process(COMMAND ${CMAKE_COMMAND} -E tar zxf
-      ${RDKit_ExternalDir}/rapidjson-1.1.0.tar.gz
-      WORKING_DIRECTORY ${RDKit_ExternalDir})
-    # comment out a line which causes a compilation error on some platforms
+      ${RDKit_ExternalDir}/rapidjson/rapidjson-1.1.0.tar.gz
+      WORKING_DIRECTORY ${RDKit_ExternalDir}/rapidjson)
+    # backport some master branch improvements to avoid a compilation error on some platforms
     # (based on the change which has already been applied to the RapidJSON master branch, see
-    # https://github.com/Tencent/rapidjson/blob/ab1842a2dae061284c0a62dca1cc6d5e7e37e346/include/rapidjson/document.h#L414)
-    file(READ ${RDKit_ExternalDir}/rapidjson-1.1.0/include/rapidjson/document.h RAPIDJSON_DOCUMENT_H)
-    string(REGEX REPLACE
-           "( *)(GenericStringRef& operator=\\(const GenericStringRef& rhs\\) { s = rhs\\.s. length = rhs\\.length. })" "\\1//\\2"
-           RAPIDJSON_DOCUMENT_H "${RAPIDJSON_DOCUMENT_H}")
-    file(WRITE ${RDKit_ExternalDir}/rapidjson-1.1.0/include/rapidjson/document.h "${RAPIDJSON_DOCUMENT_H}")
+    # https://github.com/Tencent/rapidjson/blob/24b5e7a8b27f42fa16b96fc70aade9106cf7102f/include/rapidjson/document.h#L414
+    # and https://patch-diff.githubusercontent.com/raw/Tencent/rapidjson/pull/1698.diff
+    execute_process(COMMAND git --git-dir= apply
+      ${RDKit_ExternalDir}/rapidjson/patch-rapidjson-1.1.0.diff
+      WORKING_DIRECTORY ${RDKit_ExternalDir}/rapidjson)
 else()
-    message("-- Found RapidJSON source in ${RDKit_ExternalDir}")
+    message("-- Found RapidJSON source in ${RDKit_ExternalDir}/rapidjson")
 endif()
 
-include_directories(${RDKit_ExternalDir}/rapidjson-1.1.0/include)
+include_directories(${RDKit_ExternalDir}/rapidjson/rapidjson-1.1.0/include)
 
 
 rdkit_library(MolInterchange

Code/MinimalLib/CMakeLists.txt

@@ -1,5 +1,5 @@
 include_directories(${RDKit_ExternalDir})
-include_directories(${RDKit_ExternalDir}/rapidjson-1.1.0/include)
+include_directories(${RDKit_ExternalDir}/rapidjson/rapidjson-1.1.0/include)
 
 if(RDK_BUILD_MINIMAL_LIB)
     set(MINIMAL_LIB_LIBRARIES "MolInterchange;Abbreviations;"

Code/MinimalLib/docker/Dockerfile_3_rdkit_build

@@ -83,12 +83,6 @@ RUN emcmake cmake -DRDK_BUILD_FREETYPE_SUPPORT=ON -DRDK_BUILD_MINIMAL_LIB=ON \
 RUN cp /src/rdkit/External/INCHI-API/src/INCHI_BASE/src/util.c /src/rdkit/External/INCHI-API/src/INCHI_BASE/src/util.c.bak && \
   sed 's/&& defined(__APPLE__)//' /src/rdkit/External/INCHI-API/src/INCHI_BASE/src/util.c.bak > /src/rdkit/External/INCHI-API/src/INCHI_BASE/src/util.c
 
-# comment out a line which causes a compilation error on some platforms
-# (based on the change which has already been applied to the RapidJSON master branch, see
-# https://github.com/Tencent/rapidjson/blob/ab1842a2dae061284c0a62dca1cc6d5e7e37e346/include/rapidjson/document.h#L414)
-RUN sed -i 's|^\( *\)\(GenericStringRef\& operator=(const GenericStringRef\& rhs) { s = rhs.s; length = rhs.length; } *\)$|\1//\2|' \
-  /src/rdkit/External/rapidjson-1.1.0/include/rapidjson/document.h
-
 # build and "install"
 RUN make -j2 RDKit_minimal && \
   cp Code/MinimalLib/RDKit_minimal.* ../Code/MinimalLib/demo/

External/rapidjson/patch-rapidjson-1.1.0.diff

@@ -0,0 +1,100 @@
+--- a/rapidjson-1.1.0/include/rapidjson/document.h	2016-08-25 09:59:05.000000000 +0200
++++ b/rapidjson-1.1.0/include/rapidjson/document.h	2025-09-25 11:37:06.694007282 +0200
+@@ -22,6 +22,8 @@
+ #include "internal/strfunc.h"
+ #include "memorystream.h"
+ #include "encodedstream.h"
++
++#include <algorithm>
+ #include <new>      // placement new
+ #include <limits>
+ 
+@@ -68,9 +70,48 @@
+     https://code.google.com/p/rapidjson/issues/detail?id=64
+ */
+ template <typename Encoding, typename Allocator> 
+-struct GenericMember { 
++class GenericMember {
++public:
++    // Allow default construction as it is needed during copying.
++    GenericMember() {}
++
+     GenericValue<Encoding, Allocator> name;     //!< name of member (must be a string)
+     GenericValue<Encoding, Allocator> value;    //!< value of member.
++
++#if RAPIDJSON_HAS_CXX11_RVALUE_REFS
++    //! Move constructor in C++11
++    GenericMember(GenericMember&& rhs) RAPIDJSON_NOEXCEPT
++        : name(std::move(rhs.name)),
++          value(std::move(rhs.value))
++    {
++    }
++
++    //! Move assignment in C++11
++    GenericMember& operator=(GenericMember&& rhs) RAPIDJSON_NOEXCEPT {
++        return *this = static_cast<GenericMember&>(rhs);
++    }
++#endif
++
++    //! Assignment with move semantics.
++    /*! \param rhs Source of the assignment. Its name and value will become a null value after assignment.
++    */
++    GenericMember& operator=(GenericMember& rhs) RAPIDJSON_NOEXCEPT {
++        if (RAPIDJSON_LIKELY(this != &rhs)) {
++            name = rhs.name;
++            value = rhs.value;
++        }
++        return *this;
++    }
++
++    // swap() for std::sort() and other potential use in STL.
++    friend inline void swap(GenericMember& a, GenericMember& b) RAPIDJSON_NOEXCEPT {
++        a.name.Swap(b.name);
++        a.value.Swap(b.value);
++    }
++
++private:
++    //! Copy constructor is not permitted.
++    GenericMember(const GenericMember& rhs);
+ };
+ 
+ ///////////////////////////////////////////////////////////////////////////////
+@@ -316,8 +357,6 @@
+ 
+     GenericStringRef(const GenericStringRef& rhs) : s(rhs.s), length(rhs.length) {}
+ 
+-    GenericStringRef& operator=(const GenericStringRef& rhs) { s = rhs.s; length = rhs.length; }
+-
+     //! implicit conversion to plain CharType pointer
+     operator const Ch *() const { return s; }
+ 
+@@ -1934,9 +1973,11 @@
+     void SetArrayRaw(GenericValue* values, SizeType count, Allocator& allocator) {
+         data_.f.flags = kArrayFlag;
+         if (count) {
+-            GenericValue* e = static_cast<GenericValue*>(allocator.Malloc(count * sizeof(GenericValue)));
+-            SetElementsPointer(e);
+-            std::memcpy(e, values, count * sizeof(GenericValue));
++            auto arr = static_cast<GenericValue*>(allocator.Malloc(count * sizeof(GenericValue)));
++            for (SizeType idx = 0; idx < count; ++idx)
++                new (arr + idx) GenericValue;
++            SetElementsPointer(arr);
++            std::copy_n(values, count, arr);
+         }
+         else
+             SetElementsPointer(0);
+@@ -1947,9 +1988,11 @@
+     void SetObjectRaw(Member* members, SizeType count, Allocator& allocator) {
+         data_.f.flags = kObjectFlag;
+         if (count) {
+-            Member* m = static_cast<Member*>(allocator.Malloc(count * sizeof(Member)));
+-            SetMembersPointer(m);
+-            std::memcpy(m, members, count * sizeof(Member));
++            auto arr = static_cast<Member*>(allocator.Malloc(count * sizeof(Member)));
++            for (SizeType idx = 0; idx < count; ++idx)
++                new (arr + idx) Member;
++            SetMembersPointer(arr);
++            std::copy_n(members, count, arr);
+         }
+         else
+             SetMembersPointer(0);