Lattice

A Python package that provides support for a schema-based building data model framework, currently under development as ASHRAE Standard 232P, where schema are described in compliant YAML source files. Lattice performs:

• Data model validation: Ensures the YAML schema source files are well-formed.
• Schema generation: Translates the YAML schema source files into equivalent JSON Schema.
• Data file validation: Validates data files against the generated JSON Schema and additional validation requirements not supported by JSON Schema (e.g., reference checking).
• Data model documentation: Generates web documentation of the data model from the YAML schema source files and templated markdown files. This web documentation can be automatically deployed to GitHub pages.

Future additions under development include:

• Generation of PDF documentation of the data model.
• Generation of C/C++ source code for processing compliant data files.

Installing

To install Lattice, simply:

pip install lattice

Example Usage

lattice is Python package defining the Lattice class. Lattice relies on a predetermined structure of subdirectories:

• schema (optional): Contains YAML source schema files describing the data model. Alternatively, if YAML source schema files are not provided in a "schema" directory, they must be in the root directory.
• docs (optional): Contains markdown templates that are used to render model documentation. An optional subdirectory of "docs" called "web" contains additional content required for generating the web documentation including configuration settings, graphics, and supplementary content.
• examples (optional): Example data files compliant with the data model.

The Lattice class is instantiated with the following parameters:

• root_directory: This is the directory containing the source subdirectories.The default is the current working directory.

• build_directory: This is the path to the directory where the content related to lattice is stored. The content itself will be located in a subdirectory determined by build_output_directory_name (below). It includes intermediate meta-schema(s), JSON Schema(s), generated markdown files, and the generated web documentation. The default is root_directory.

• build_output_directory_name: The name of the lattice output content directory. The default is ".lattice/".

• build_validation: A boolean indicator to automatically generate meta-schema, validate the data model, generate the schemas and validate the example data files upon instantiation. If false, these tasks must be executed after instantiation using the generate_meta_schemas, validate_schemas, generate_json_schemas, and validate_example_files methods. The default is True.

The repository's examples directory contains sample data models exemplifying different model options, such as Data Group Templates or scoped references.

More complete examples of projects using the ASHRAE Standard 232P framework include:

• IBPSA-USA Climate Information (uses lattice)
• ASHRAE Standard 205 (transitioning to lattice)
• ASHRAE Standard 229 (does not use lattice...yet)

C++ Library Code Generation

Lattice's C++ code generation is achieved by calling the function generate_cpp_project(). Its only parameter is a list of submodule URLs.

Translations

Schema can be converted into C++ classes with the following mappings:

Object Type	C++ type
Data Group	struct
Data Element	public data member
Enumerator	enum
Data Group Template	base class*

|

*with optional methods, including initialize()

Data Type	C++ type
Integer	int
Numeric	float
String	std::string
{}	struct
<>	enum
[]	std::vector
list "(A,B)"	std::unique_ptr<BaseOfAB>

|

Inheritance

The code generator will assume that Data Group schema elements with a Data Group Template parameter use that template as the element's superclass. An #include statement for the expected superclass file is listed at the top of the schema's implemenation (cpp) file (note: see Big Ladder file naming conventions). If the superclass file is not found, the C++ generator will create a stub file for the class. If it is found, any virtual functions in the superclass will appear as overridden function stubs in the subclassed Data Group's struct. Any additional code (members, methods) for the superclass itself must be provided by the lattice user.

In the event that the source schema contains a Data Element with a "selector constraint" (i.e. a list of possible Data Types combined with an associated list of possible enumerator values for the selector Data Element), the C++ generated code will assume that the Data Types in the list all derive from a common base class, named by the Data Group Template of the first Data Type in the list. (The Data Group Template may be defined in a different schema than the Data Type.) An #include statement for the base class will be generated, as above. The code that populates the Data Group (the struct's from_json function) will use a conditional statement to create a new object of the correct subclass and assign it to a member unique_ptr, calling directly the derived class' auto-generated from_json functions.

Note: If the first Data Type in a selector constraint list does not have a Data Group Template tag in its schema, the unique_ptr's base class will default to "MissingType."

Build information

In addition to .h and .cpp files containing translated schema data, the code generator adds Git repository and CMake build support to the schema code, creating most of the structure necessary to test-build the schema code as a library. Necessary submodules are also downloaded: fmt, json, and courier. To build a generated project, navigate to your lattice project's build directory, cpp subdirectory (e.g. /.lattice/cpp), and use a standard cmake build sequence:

cmake -B build

cmake --build build --config release