Printing data¶

In Draive you often need to surface multimodal data in logs, dashboards, or CLI output. Every model state exposes a few helpers that flatten structured content into printable strings:

Helper	Output
`x.to_mapping(recursive: bool = True)`	`Mapping` representation
`x.to_json(indent: int \\| None = None)`	JSON string
`x.to_str()`	Model-defined string, defaults to `str(x)`
`str(x)`	`ClassName(field: value)` for typed `State` models

Different model instances implement .to_str() in their own way, so the rendered view varies by content type. The sections below walk through the most common multimodal classes and show how their printing helpers behave.

TextContent¶

Plain text output that mirrors the original text value without any extra markup or quoting.

return self.text

Example:

text: TextContent = TextContent.of('Hello world!')
ctx.log_info(text.to_str())

Hello world!

ResourceContent¶

Generates a Markdown media reference. Depending on include_data, the reference either embeds the base64 data URI directly or leaves the link empty.

Note

kind variable will be one of: 'image' | 'audio' | 'video' | ''

When include_data=True, ResourceContent.to_str() returns the full base64 payload:

return f"![{kind}]({self.to_data_uri()})"

Example:

resource: ResourceContent = ResourceContent.of(b'FF', mime_type='application/octet-stream')
ctx.log_info(resource.to_str(include_data=True))

![](data:application/octet-stream;base64,RkY=)

With the default include_data=False, it keeps the media kind visible without leaking the bytes:

return f"![{kind}]()"

Example:

with open('./src/dog.jpg', 'rb') as img:
    img_data: bytes = img.read()
resource: ResourceContent = ResourceContent.of(img_data, mime_type='image/jpeg')
ctx.log_info(resource.to_str())

![image]()

ArtifactContent¶

Renders the wrapped artifact payload as indented JSON when visible, otherwise suppresses content entirely for hidden artifacts.

Hidden artifacts render as an empty string:

return ""

Visible artifacts are serialized to JSON:

return json.dumps(self.artifact, indent=2)

Example:

from draive import ArtifactContent, State, ctx


class User(State, serializable=True):
    first_name: str


artifact: ArtifactContent = ArtifactContent.of(User(first_name="James"), category="profile")
ctx.log_info(artifact.to_str())

{
  "first_name": "James"
}

MultimodalTag¶

Formats XML-like tags, optionally wrapping the rendered child content. Empty content yields a self-closing tag.

When self.content is empty, the printer emits a self-closing tag:

return f"<{self.name}{_tag_attributes(self.meta)}/>"

Example:

<TAG_NAME attr_1="True" attr_2="val_2"/>

Otherwise it wraps the rendered child content:

return f"<{self.name}{_tag_attributes(self.meta)}>{self.content.to_str()}</{self.name}>"

Example:

<TAG_NAME attr_1="True" attr_2="val_2">Hello World!</TAG_NAME>

Important

MultimodalTag is the only multimodal element that exposes metadata inline. Values stored in meta appear as quoted XML-style tag attributes.

MultimodalContent¶

Concatenates the string form of each part, resulting in a single composite response without delimiters. An optional artifact_to_str argument overrides the rendering of ArtifactContent parts.

return "".join(part.to_str() for part in self.parts)

Example:

from draive import MultimodalContent, TextContent, ctx

multimodal: MultimodalContent = MultimodalContent.of(
    TextContent.of('Hello '),
    TextContent.of('World!'),
)
ctx.log_info(multimodal.to_str())

Result:

Hello World!