Update how schemas are generated and stored and loaded into C++

See #20

.gitignore

@@ -14,4 +14,7 @@
 /**/.DS_Store
 doc/source/_static/doxygen
 doc/source/api/*.rst
-doc/source/examples/*.pdf
+doc/source/examples/*.pdf
+interface/CPP/model_schemas.cpp
+
+interface/CPP/model_schemas.json

CMakeLists.txt

@@ -23,6 +23,15 @@ else()
     message(FATAL_ERROR "tar.xz of boost sources needed for teqp cannot be found")
 endif()
 
+set(ZIPFN "${CMAKE_CURRENT_SOURCE_DIR}/dev/model_schemas.tar.xz")
+if (EXISTS ${ZIPFN})
+  execute_process(COMMAND ${CMAKE_COMMAND} -E tar -xJf ${ZIPFN} 
+                  WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}/interface/CPP")
+  message(STATUS "Untarred model schemas")
+else()
+  message(FATAL_ERROR "tar.xz of thermodynamic model schemas needed for teqp cannot be found")
+endif()
+
 add_library(teqpinterface INTERFACE)
 
 target_include_directories(teqpinterface INTERFACE "${CMAKE_CURRENT_SOURCE_DIR}/include")
@@ -101,14 +110,8 @@ if (NOT TEQP_NO_TEQPCPP)
   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 ${MODEL_JSON_SCHEMA_TEMPLATE}
-       @ONLY)
-  target_sources(teqpcpp PRIVATE model_schemas.cpp)
+  # Add the schema file
+  target_sources(teqpcpp PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/interface/CPP/model_schemas.cpp")
 
   if (TEQP_TESTTEQPCPP)
     add_executable(test_teqpcpp "${CMAKE_CURRENT_SOURCE_DIR}/interface/CPP/test/test_teqpcpp.cpp")

notebooks/Schema Builder.ipynb

@@ -289,18 +289,7 @@
    "execution_count": 9,
    "id": "e83a283c",
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "SAFTVRMieWithCoeffs(kmat=None, polar_model='GrayGubbins+GubbinsTwu', SAFTVRMie_flags=None, polar_flags=None, coeffs=[SAFTVRMieCoeffsigmaAngstrom(name='R1234YF', m=1.3656, epsilon_over_k=299.424, BibTeXKey='Paricaud', lambda_r=21.7779, lambda_a=6.0, mu_Cm=None, mu_D=2.2814, Q_Cm2=None, Q_DA=1.4151, nmu=1.0, nQ=1.0, sigma_Angstrom=4.5307)])"
-      ]
-     },
-     "execution_count": 9,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "outputs": [],
    "source": [
     "class SAFTVRMieCoeffBase(NoExtraBaseModel):\n",
     "    name: str\n",
@@ -338,7 +327,7 @@
     "        \n",
     "SAFTVRMie = TypeAdapter(Union[SAFTVRMieWithNames, SAFTVRMieWithCoeffs])\n",
     "# SAFTVRMieWithCoeffs.validate({\"coeffs\":[{\"BibTeXKey\":\"me\",\"epsilon_over_k\":100,\"lambda_a\":6,\"lambda_r\": 12,\"m\":1.0,\"name\":\"Stockmayer126\",\"sigma_m\":3e-10}]})\n",
-    "SAFTVRMie.validate_python({\"coeffs\":[{\"BibTeXKey\":\"Paricaud\",\"Q_DA\":1.4151,\"epsilon_over_k\":299.424, \"lambda_a\":6.0,\"lambda_r\":21.7779,\"m\":1.3656,\"mu_D\":2.2814,\"nQ\":1.0,\"name\":\"R1234YF\",\"nmu\":1.0,\"sigma_Angstrom\":4.5307}],\"polar_model\":\"GrayGubbins+GubbinsTwu\"})"
+    "# SAFTVRMie.validate_python({\"coeffs\":[{\"BibTeXKey\":\"Paricaud\",\"Q_DA\":1.4151,\"epsilon_over_k\":299.424, \"lambda_a\":6.0,\"lambda_r\":21.7779,\"m\":1.3656,\"mu_D\":2.2814,\"nQ\":1.0,\"name\":\"R1234YF\",\"nmu\":1.0,\"sigma_Angstrom\":4.5307}],\"polar_model\":\"GrayGubbins+GubbinsTwu\"})"
    ]
   },
   {
@@ -387,9 +376,66 @@
     "        else:\n",
     "            schemas[klass.__name__] = klass.model_json_schema()\n",
     "    \n",
-    "with open(\"schemas.json\", 'w') as fp:\n",
+    "with open(\"model_schemas.json\", 'w') as fp:\n",
     "    fp.write(json.dumps(schemas, indent=2))"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "id": "463514bd",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import json\n",
+    "def string_to_hexchunk(jsonstring):\n",
+    "    \n",
+    "    def to_chunks(l, n):\n",
+    "        if n < 1:\n",
+    "            n = 1\n",
+    "        return [l[i:i + n] for i in range(0, len(l), n)]\n",
+    "    \n",
+    "    try:\n",
+    "        h = [\"0x{:02x}\".format(ord(b)) for b in jsonstring] + ['0x00']\n",
+    "    except TypeError:\n",
+    "        h = [\"0x{:02x}\".format(int(b)) for b in jsonstring] + ['0x00']\n",
+    "\n",
+    "    # Break up the file into lines of 16 hex characters\n",
+    "    # because some compilers don't like VERY long lines\n",
+    "    chunks = to_chunks(h, 16)\n",
+    "\n",
+    "    # Put the lines back together again\n",
+    "    # The chunks are joined together with commas, and then EOL are used to join the rest\n",
+    "    hex_string = '{' + ',\\n'.join([', '.join(chunk) for chunk in chunks]) + '}'\n",
+    "    return hex_string\n",
+    "            \n",
+    "chunk = string_to_hexchunk(json.dumps(schemas, indent=0))\n",
+    "INFO = \"\"\"// Due to limitations in MSVC, very long string literals are \n",
+    "// not allowed. Thus the string must be re-encoded as binary. The \n",
+    "// contents of the string are in the JSON file next to this file\n",
+    "\n",
+    "#include \"nlohmann/json.hpp\"\n",
+    "#include <string>\n",
+    "\"\"\"\n",
+    "with open('model_schemas.cpp','w') as fp:\n",
+    "    fp.write(INFO)\n",
+    "    fp.write('extern const auto model_schema_library = nlohmann::json::parse(std::string(\\n' + chunk + '\\n)); ')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "id": "d141efd0",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import tarfile\n",
+    "\n",
+    "path = '../dev/model_schemas.tar.xz'\n",
+    "with tarfile.open(path, mode='w:xz') as tar:\n",
+    "    tar.add('model_schemas.json')\n",
+    "    tar.add('model_schemas.cpp')"
+   ]
   }
  ],
  "metadata": {