Jinja2PromptTemplate Class

Creates and renders Jinja2 prompt templates to text.

Jinja2 templates support advanced features such as variable substitution, control structures, and inheritance, making it possible to dynamically generate text based on input arguments and predefined functions. This class leverages Jinja2's flexibility to render prompts that can include conditional logic, loops, and functions, based on the provided template configuration and arguments.

Note that the fully qualified function name (in the form of "plugin-function") is not allowed in Jinja2 because of the hyphen. Therefore, the function name is replaced with an underscore, which are allowed in Python function names.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Inheritance

PromptTemplateBase

Jinja2PromptTemplate

Constructor

Jinja2PromptTemplate(*, prompt_template_config: PromptTemplateConfig, allow_dangerously_set_content: bool = False)

Parameters

Name	Description
prompt_template_config Required	<xref:semantic_kernel.prompt_template.jinja2_prompt_template.PromptTemplateConfig> The configuration object for the prompt template. This should specify the template format as 'jinja2' and include any necessary configuration details required for rendering the template.
allow_dangerously_set_content Required	<xref:<xref:semantic_kernel.prompt_template.jinja2_prompt_template.bool = False>> Allow content without encoding throughout, this overrides the same settings in the prompt template config and input variables. This reverts the behavior to unencoded input.

Keyword-Only Parameters

Name	Description
prompt_template_config Required
allow_dangerously_set_content Required

Methods

model_post_init	We need to both initialize private attributes and call the user-defined model_post_init method.
render	Render the prompt template. Using the prompt template, replace the variables with their values and execute the functions replacing their reference with the function result.
validate_template_format	Validate the template format.

model_post_init

We need to both initialize private attributes and call the user-defined model_post_init method.

model_post_init(context: Any, /) -> None

Positional-Only Parameters

Name	Description
context Required

render

Render the prompt template.

Using the prompt template, replace the variables with their values and execute the functions replacing their reference with the function result.

async render(kernel: Kernel, arguments: KernelArguments | None = None) -> str

Positional-Only Parameters

Name	Description
context Required

Parameters

Name	Description
kernel Required	The kernel instance
arguments	The kernel arguments Default value: None

Returns

Type	Description
	The prompt template ready to be used for an AI request

validate_template_format

Validate the template format.

validate_template_format(v: PromptTemplateConfig) -> PromptTemplateConfig

Positional-Only Parameters

Name	Description
context Required

Parameters

Name	Description
v Required

Attributes

model_computed_fields

A dictionary of computed field names and their corresponding ComputedFieldInfo objects.

model_computed_fields: ClassVar[Dict[str, ComputedFieldInfo]] = {}

model_config

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

model_config: ClassVar[ConfigDict] = {'arbitrary_types_allowed': True, 'populate_by_name': True, 'validate_assignment': True}

model_fields

Metadata about the fields defined on the model, mapping of field names to [FieldInfo][pydantic.fields.FieldInfo] objects.

This replaces Model.fields from Pydantic V1.

model_fields: ClassVar[Dict[str, FieldInfo]] = {'allow_dangerously_set_content': FieldInfo(annotation=bool, required=False, default=False), 'prompt_template_config': FieldInfo(annotation=PromptTemplateConfig, required=True)}

allow_dangerously_set_content

allow_dangerously_set_content: bool

prompt_template_config

prompt_template_config: PromptTemplateConfig