Package org.vrspace.server.connect

Class OllamaConnector

java.lang.Object

org.vrspace.server.connect.OllamaConnector

@Component
@ConditionalOnBean(OllamaConfig.class)
public class OllamaConnector
extends Object

Component that connects to Ollama, and provides services and tools to agents, bots, and other components. It can generate image description from the image of a 3D model, and store it to local vector database for later search. Image processing is done asynchronously, one by one, and optionally paused during search, to keep down resource consumption. Models used for images and tool calls (search) are configured in properties file, and can be different, or the same one.

Author:

joe

Constructor Summary

Constructors

Constructor	Description
OllamaConnector()

Method Summary

Modifier and Type	Method	Description
String	describeImage(String url)	Describe the image: issue API call to Ollama vision model, and post-process the image.
String	localSearch(String query, Integer maxSize, Boolean animated, Boolean rigged, Integer maxResults)
void	shutdown()
String	sketchfabSearch(String keywords, Integer maxSize, Boolean animated, Boolean rigged, Integer maxResults)
void	startImageProcessing()	(Re)start stopped image processing.
void	stopImageProcessing()	Stop image processing to speed up general chat.
org.springframework.ai.ollama.OllamaChatModel	toolsModel()	Returns configured tools model.
void	updateDescriptionFromThumbnail(GltfModel model)	Passed as a post-processing function to sketchfab connector search.
org.springframework.ai.ollama.OllamaChatModel	visionModel()	Returns configured vision model.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

OllamaConnector

public OllamaConnector()

Method Details

describeImage

public String describeImage(String url)

Describe the image: issue API call to Ollama vision model, and post-process the image.

Parameters:

url - image URL

Returns:

generated and post-processed description

updateDescriptionFromThumbnail

public void updateDescriptionFromThumbnail(GltfModel model)

Passed as a post-processing function to sketchfab connector search. Adds a task to image processing executor that processes the image with vision model, and asynchronously saves generated description of the model to the object and vector databases.

visionModel

public org.springframework.ai.ollama.OllamaChatModel visionModel()

Returns configured vision model.

toolsModel

public org.springframework.ai.ollama.OllamaChatModel toolsModel()

Returns configured tools model.

stopImageProcessing

public void stopImageProcessing()

Stop image processing to speed up general chat.

startImageProcessing

public void startImageProcessing()

(Re)start stopped image processing.

shutdown

@PreDestroy
public void shutdown()

sketchfabSearch

@Tool(description="Sketchfab 3D model search web API")
public String sketchfabSearch(@ToolParam(description="Search keywords")
 String keywords,
 @ToolParam(description="Maximum model size, in megabytes")
 Integer maxSize,
 @ToolParam(description="Request only animated models")
 Boolean animated,
 @ToolParam(description="Request only rigged models")
 Boolean rigged,
 @ToolParam(description="Maximum number of results")
 Integer maxResults)

localSearch

@Tool(description="Search local vector database for known 3D models")
public String localSearch(@ToolParam(description="Search query")
 String query,
 @ToolParam(description="Maximum model size, in megabytes")
 Integer maxSize,
 @ToolParam(description="Request only animated models")
 Boolean animated,
 @ToolParam(description="Request only rigged models")
 Boolean rigged,
 @ToolParam(description="Maximum number of results")
 Integer maxResults)