Model input schema validation

.gitmodules

@@ -13,3 +13,6 @@
 [submodule "externals/Catch2"]
 	path = externals/Catch2
 	url = https://github.com/catchorg/Catch2.git
+[submodule "externals/json-schema-validator"]
+	path = externals/json-schema-validator
+	url = https://github.com/pboettch/json-schema-validator

CMakeLists.txt

@@ -32,6 +32,7 @@ target_include_directories(teqpinterface INTERFACE "${CMAKE_CURRENT_SOURCE_DIR}/
 target_include_directories(teqpinterface INTERFACE "${CMAKE_CURRENT_SOURCE_DIR}/boost_teqp")
 
 add_subdirectory("${CMAKE_CURRENT_SOURCE_DIR}/externals/Catch2")
+add_subdirectory("${CMAKE_CURRENT_SOURCE_DIR}/externals/json-schema-validator")
 
 set(EIGEN3_INCLUDE_DIR "${CMAKE_CURRENT_SOURCE_DIR}/externals/Eigen" CACHE INTERNAL "Path to Eigen, for autodiff")
 set(EIGEN3_VERSION_OK TRUE CACHE BOOL "Yes eigen is fine")
@@ -93,11 +94,20 @@ if (NOT TEQP_NO_TEQPCPP)
   # doesn't require a full compile for a single LOC change
   file(GLOB sources "${CMAKE_CURRENT_SOURCE_DIR}/interface/CPP/*.cpp")
   add_library(teqpcpp STATIC ${sources})
-  target_link_libraries(teqpcpp PUBLIC teqpinterface PUBLIC autodiff)
+  target_link_libraries(teqpcpp PUBLIC nlohmann_json_schema_validator PUBLIC teqpinterface PUBLIC autodiff)
   target_include_directories(teqpcpp PUBLIC "${CMAKE_CURRENT_SOURCE_DIR}/interface/CPP")
   set_property(TARGET teqpcpp PROPERTY POSITION_INDEPENDENT_CODE ON)
   target_compile_definitions(teqpcpp PRIVATE -DMULTICOMPLEX_NO_MULTIPRECISION)
   target_compile_definitions(teqpcpp PUBLIC -DUSE_AUTODIFF)
+
+  # Populate the model schema cpp file with the contents
+  file(READ "${CMAKE_CURRENT_SOURCE_DIR}/notebooks/schemas.json" MODEL_SCHEMA_CONTENTS)
+  file(READ "${CMAKE_CURRENT_SOURCE_DIR}/dev/templates/model_schema.cpp.in" MODEL_JSON_SCHEMA_TEMPLATE)
+  file(CONFIGURE
+       OUTPUT model_schemas.cpp
+       CONTENT ${MODEL_JSON_SCHEMA_TEMPLATE}
+       @ONLY)
+  target_sources(teqpcpp PRIVATE model_schemas.cpp)
 
   if (TEQP_TESTTEQPCPP)
     add_executable(test_teqpcpp "${CMAKE_CURRENT_SOURCE_DIR}/interface/CPP/test/test_teqpcpp.cpp")
@@ -134,7 +144,7 @@ if (NOT TEQP_NO_PYTHON)
     file(GLOB pybind11_files "${CMAKE_CURRENT_SOURCE_DIR}/interface/*.cpp")
     pybind11_add_module(teqp "${pybind11_files}")
     target_include_directories(teqp PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/externals/pybind11_json/include")
-    target_link_libraries(teqp PRIVATE autodiff PRIVATE teqpinterface PRIVATE teqpcpp)
+    target_link_libraries(teqp PRIVATE teqpcpp PRIVATE autodiff PRIVATE teqpinterface )
     target_compile_definitions(teqp PUBLIC -DUSE_AUTODIFF)
     if (MSVC)
       target_compile_options(teqp PRIVATE "/Zm1000")
@@ -151,7 +161,8 @@ if (NOT TEQP_NO_TESTS)
   endif()
   target_compile_definitions(catch_tests PRIVATE -DTEQPC_CATCH)
   target_compile_definitions(catch_tests PRIVATE -DTEQP_MULTICOMPLEX_ENABLED)
-  target_link_libraries(catch_tests PRIVATE autodiff PRIVATE teqpinterface PRIVATE Catch2WithMain PUBLIC teqpcpp)
+  target_compile_definitions(catch_tests PRIVATE -DTEQP_MULTIPRECISION_ENABLED)
+  target_link_libraries(catch_tests PUBLIC teqpcpp PRIVATE autodiff PRIVATE teqpinterface PRIVATE Catch2WithMain)
   add_test(normal_tests catch_tests)
 endif()
 
@@ -202,7 +213,7 @@ if (TEQP_SNIPPETS)
       target_sources(${snippet_exe} PUBLIC "${CMAKE_CURRENT_SOURCE_DIR}/externals/Eigen/debug/msvc/eigen.natvis")
     endif()
 
-    target_link_libraries(${snippet_exe} PRIVATE autodiff PRIVATE teqpinterface PRIVATE Catch2WithMain PRIVATE teqpcpp)
+    target_link_libraries(${snippet_exe} PRIVATE teqpcpp PRIVATE autodiff PRIVATE teqpinterface PRIVATE Catch2WithMain )
     if(UNIX)
       target_link_libraries (${snippet_exe} PRIVATE ${CMAKE_DL_LIBS})
     endif()

dev/templates/model_schema.cpp.in

@@ -0,0 +1,9 @@
+#include "nlohmann/json.hpp"
+
+// The contents of this file are populated by CMake
+// Note the protectparens( ... )protectparens is to ensure that ( can be used inside the raw string literal
+extern const nlohmann::json model_schema_library = R"protectparens(
+@MODEL_SCHEMA_CONTENTS@
+)protectparens"_json\;
+
+

doc/source/models/model_potentials.ipynb

@@ -178,7 +178,7 @@
     "  'model': {\n",
     "      \"author\": \"2CLJF_Lisal\",\n",
     "      'L^*': 0.5,\n",
-    "      '(mu^*)^2': 0.1\n",
+    "      '(Q^*)^2': 0.1\n",
     "  }\n",
     "})\n",
     "print(model.solve_pure_critical(1.3, 0.3))"

include/teqp/cpp/teqpcpp.hpp

@@ -175,5 +175,8 @@ namespace teqp {
         );
 
         std::unique_ptr<AbstractModel> build_model_ptr(const nlohmann::json& json);
+
+        /// Return the schema for the given model kind
+        nlohmann::json get_model_schema(const std::string& kind);
     }
 }

include/teqp/exceptions.hpp

@@ -43,4 +43,20 @@ namespace teqp {
         NotImplementedError(const std::string& msg) : teqpException(200, msg) {};
     };
 
+    /// Validation of a JSON schema failed
+    class JSONValidationError : public teqpException {
+    private:
+        auto errors_to_string(const std::vector<std::string> &errors, const std::string delim = "|/|\\|"){
+            std::string o = "";
+            if (errors.empty()){ return o; }
+            o = errors[0];
+            for (auto j = 1; j < errors.size(); ++j){
+                o += delim + errors[j];
+            }
+            return o;
+        }
+    public:
+        JSONValidationError(const std::vector<std::string>& errors) : teqpException(300, errors_to_string(errors)) {};
+    };
+
 }; // namespace teqp

include/teqp/json_tools.hpp

@@ -1,11 +1,17 @@
 #pragma once
 #include "nlohmann/json.hpp"
+#include <nlohmann/json-schema.hpp>
 
 #include <set>
 #include <filesystem>
 #include <fstream>
 #include "teqp/exceptions.hpp"
 
+#include <Eigen/Dense>
+
+using nlohmann::json;
+using nlohmann::json_schema::json_validator;
+
 namespace teqp{
 
     /// Load a JSON file from a specified file
@@ -105,4 +111,43 @@ namespace teqp{
             throw teqp::InvalidArgument("Unable to load the argument to multilevel_JSON_load");
         }
     }
+
+    /**
+    This class is not thread-safe for construction because the validator is not
+     */
+    class JSONValidator{
+    public:
+        const nlohmann::json schema;
+
+        json_validator validator; // create validator
+
+        // Instantiate the validator object, will throw if the schema is invalid
+        JSONValidator(const nlohmann::json& schema) : schema(schema) {
+            validator.set_root_schema(schema); // insert root-schema
+        }
+
+        // Return the validation errors when trying to validate the JSON
+        std::vector<std::string> get_validation_errors(const nlohmann::json& j) const{
+
+            /* Custom error handler */
+            class custom_error_handler : public nlohmann::json_schema::basic_error_handler
+            {
+            public:
+                std::vector<std::string> errors;
+                void error(const nlohmann::json::json_pointer &ptr, const json &instance, const std::string &message) override
+                {
+                    nlohmann::json_schema::basic_error_handler::error(ptr, instance, message);
+                    std::stringstream ss;
+                    ss << ptr << ":" << instance << "': " << message << "\n";
+                    errors.push_back(ss.str());
+                }
+            } handler;
+            validator.validate(j, handler); // validate the document
+            return handler.errors;
+        }
+
+        // A quick checker for validation of the JSON
+        bool is_valid(const nlohmann::json&j ) const { return get_validation_errors(j).empty(); }
+    };
+
 }

include/teqp/models/model_potentials/2center_ljf.hpp

@@ -424,7 +424,7 @@ namespace teqp {
         }
 
         // build the 2-center Lennard-Jones model with quadrupole
-        inline auto build_two_center_model_quadrupole(const std::string& model_version, const double& L = 0.0, const double& mu_sq = 0.0) {
+        inline auto build_two_center_model_quadrupole(const std::string& model_version, const double& L = 0.0, const double& Q_sq = 0.0) {
 
             // Get reducing for temperature and density
             auto DC_funcs = get_density_reducing(model_version);
@@ -438,7 +438,7 @@ namespace teqp {
             auto EOS_quadrupolar = get_Quadrupolar_contribution(model_version);
 
             // Build the 2-center Lennard-Jones model
-            auto model = Twocenterljf(std::move(DC_funcs), std::move(TC_func), std::move(EOS_hard), std::move(EOS_att), std::move(EOS_quadrupolar), L, mu_sq);
+            auto model = Twocenterljf(std::move(DC_funcs), std::move(TC_func), std::move(EOS_hard), std::move(EOS_att), std::move(EOS_quadrupolar), L, Q_sq);
 
             return model;
         }

interface/CPP/teqp_impl_factory.cpp

@@ -2,13 +2,19 @@
 #include "teqp/json_builder.hpp"
 #include "teqp/cpp/deriv_adapter.hpp"
 
+// This large block of schema definitions is populated by cmake
+// at cmake configuration time
+extern const nlohmann::json model_schema_library;
+
 namespace teqp {
     namespace cppinterface {
 
         std::unique_ptr<teqp::cppinterface::AbstractModel> make_SAFTVRMie(const nlohmann::json &j);
 
         using makefunc = std::function<std::unique_ptr<teqp::cppinterface::AbstractModel>(const nlohmann::json &j)>;
         using namespace teqp::cppinterface::adapter;
+
+        nlohmann::json get_model_schema(const std::string& kind) { model_schema_library.at(kind); }
 
         static std::unordered_map<std::string, makefunc> pointer_factory = {
             {"vdW1", [](const nlohmann::json& spec){ return make_owned(vdWEOS1(spec.at("a"), spec.at("b"))); }},
@@ -30,7 +36,7 @@ namespace teqp {
             {"LJ126_Johnson1993", [](const nlohmann::json& spec){ return make_owned(LJ126Johnson1993());}},
             {"Mie_Pohl2023", [](const nlohmann::json& spec){ return make_owned(Mie::Mie6Pohl2023(spec.at("lambda_a")));}},
             {"2CLJF-Dipole", [](const nlohmann::json& spec){ return make_owned(twocenterljf::build_two_center_model_dipole(spec.at("author"), spec.at("L^*"), spec.at("(mu^*)^2")));}},
-            {"2CLJF-Quadrupole", [](const nlohmann::json& spec){ return make_owned(twocenterljf::build_two_center_model_quadrupole(spec.at("author"), spec.at("L^*"), spec.at("(mu^*)^2")));}},
+            {"2CLJF-Quadrupole", [](const nlohmann::json& spec){ return make_owned(twocenterljf::build_two_center_model_quadrupole(spec.at("author"), spec.at("L^*"), spec.at("(Q^*)^2")));}},
             {"IdealHelmholtz", [](const nlohmann::json& spec){ return make_owned(IdealHelmholtz(spec));}},
 
             // Implemented in its own compilation unit to help with compilation time
@@ -45,6 +51,13 @@ namespace teqp {
 
             auto itr = pointer_factory.find(kind);
             if (itr != pointer_factory.end()){
+                if (model_schema_library.contains(kind)){
+                    // This block is not thread-safe, needs a mutex or something
+                    JSONValidator validator(model_schema_library.at(kind));
+                    if (!validator.is_valid(spec)){
+                        throw teqp::JSONValidationError(validator.get_validation_errors(spec));
+                    }
+                }
                 return (itr->second)(spec);
             }
             else{