io.quarkiverse.langchain4j.ollama.OllamaStreamingChatLanguageModel Maven / Gradle / Ivy
package io.quarkiverse.langchain4j.ollama;
import static dev.langchain4j.internal.ValidationUtils.ensureNotEmpty;
import static io.quarkiverse.langchain4j.ollama.MessageMapper.toOllamaMessages;
import java.time.Duration;
import java.util.ArrayList;
import java.util.List;
import java.util.function.Consumer;
import dev.langchain4j.data.message.AiMessage;
import dev.langchain4j.data.message.ChatMessage;
import dev.langchain4j.model.StreamingResponseHandler;
import dev.langchain4j.model.chat.StreamingChatLanguageModel;
import dev.langchain4j.model.output.Response;
import io.smallrye.mutiny.Context;
/**
* Use to have streaming feature on models used trough Ollama.
*/
public class OllamaStreamingChatLanguageModel implements StreamingChatLanguageModel {
private final OllamaClient client;
private final String model;
private final String format;
private final Options options;
private OllamaStreamingChatLanguageModel(OllamaStreamingChatLanguageModel.Builder builder) {
client = new OllamaClient(builder.baseUrl, builder.timeout, builder.logRequests, builder.logResponses);
model = builder.model;
format = builder.format;
options = builder.options;
}
public static OllamaStreamingChatLanguageModel.Builder builder() {
return new OllamaStreamingChatLanguageModel.Builder();
}
@Override
public void generate(List messages, StreamingResponseHandler handler) {
ensureNotEmpty(messages, "messages");
ChatRequest request = ChatRequest.builder()
.model(model)
.messages(toOllamaMessages(messages))
.options(options)
.format(format)
.stream(true)
.build();
Context context = Context.of("response", new ArrayList());
client.streamingChat(request)
.subscribe()
.with(context,
new Consumer() {
@Override
@SuppressWarnings("unchecked")
public void accept(ChatResponse response) {
try {
if ((response == null) || (response.message() == null)
|| (response.message().content() == null)
|| response.message().content().isBlank()) {
return;
}
((List) context.get("response")).add(response);
handler.onNext(response.message().content());
} catch (Exception e) {
handler.onError(e);
}
}
},
new Consumer() {
@Override
public void accept(Throwable error) {
handler.onError(error);
}
},
new Runnable() {
@Override
@SuppressWarnings("unchecked")
public void run() {
var list = ((List) context.get("response"));
StringBuilder builder = new StringBuilder();
for (ChatResponse response : list) {
builder.append(response.message().content());
}
AiMessage message = new AiMessage(builder.toString());
handler.onComplete(Response.from(message));
}
});
}
/**
* Builder for Ollama configuration.
*/
public static final class Builder {
private Builder() {
super();
}
private String baseUrl = "http://localhost:11434";
private Duration timeout = Duration.ofSeconds(10);
private String model;
private String format;
private Options options;
private boolean logRequests = false;
private boolean logResponses = false;
public OllamaStreamingChatLanguageModel.Builder baseUrl(String val) {
baseUrl = val;
return this;
}
public OllamaStreamingChatLanguageModel.Builder timeout(Duration val) {
this.timeout = val;
return this;
}
public OllamaStreamingChatLanguageModel.Builder model(String val) {
model = val;
return this;
}
public OllamaStreamingChatLanguageModel.Builder format(String val) {
format = val;
return this;
}
public OllamaStreamingChatLanguageModel.Builder options(Options val) {
options = val;
return this;
}
public OllamaStreamingChatLanguageModel.Builder logRequests(boolean logRequests) {
this.logRequests = logRequests;
return this;
}
public OllamaStreamingChatLanguageModel.Builder logResponses(boolean logResponses) {
this.logResponses = logResponses;
return this;
}
public OllamaStreamingChatLanguageModel build() {
return new OllamaStreamingChatLanguageModel(this);
}
}
}
© 2015 - 2024 Weber Informatics LLC | Privacy Policy