mirror of
https://github.com/Palm1r/QodeAssist.git
synced 2025-10-06 03:57:07 -04:00
400 lines
13 KiB
C++
400 lines
13 KiB
C++
/*
|
|
* Copyright (C) 2024-2025 Petr Mironychev
|
|
*
|
|
* This file is part of QodeAssist.
|
|
*
|
|
* QodeAssist is free software: you can redistribute it and/or modify
|
|
* it under the terms of the GNU General Public License as published by
|
|
* the Free Software Foundation, either version 3 of the License, or
|
|
* (at your option) any later version.
|
|
*
|
|
* QodeAssist is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
* GNU General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License
|
|
* along with QodeAssist. If not, see <https://www.gnu.org/licenses/>.
|
|
*/
|
|
|
|
#include "LlamaCppProvider.hpp"
|
|
|
|
#include "llmcore/ValidationUtils.hpp"
|
|
#include "logger/Logger.hpp"
|
|
#include "settings/ChatAssistantSettings.hpp"
|
|
#include "settings/CodeCompletionSettings.hpp"
|
|
|
|
#include <QEventLoop>
|
|
#include <QJsonArray>
|
|
#include <QJsonDocument>
|
|
#include <QJsonObject>
|
|
#include <QNetworkReply>
|
|
|
|
namespace QodeAssist::Providers {
|
|
|
|
LlamaCppProvider::LlamaCppProvider(QObject *parent)
|
|
: LLMCore::Provider(parent)
|
|
, m_toolsManager(new Tools::ToolsManager(this))
|
|
{
|
|
connect(
|
|
m_toolsManager,
|
|
&Tools::ToolsManager::toolExecutionComplete,
|
|
this,
|
|
&LlamaCppProvider::onToolExecutionComplete);
|
|
}
|
|
|
|
QString LlamaCppProvider::name() const
|
|
{
|
|
return "llama.cpp";
|
|
}
|
|
|
|
QString LlamaCppProvider::url() const
|
|
{
|
|
return "http://localhost:8080";
|
|
}
|
|
|
|
QString LlamaCppProvider::completionEndpoint() const
|
|
{
|
|
return "/infill";
|
|
}
|
|
|
|
QString LlamaCppProvider::chatEndpoint() const
|
|
{
|
|
return "/v1/chat/completions";
|
|
}
|
|
|
|
bool LlamaCppProvider::supportsModelListing() const
|
|
{
|
|
return false;
|
|
}
|
|
|
|
void LlamaCppProvider::prepareRequest(
|
|
QJsonObject &request,
|
|
LLMCore::PromptTemplate *prompt,
|
|
LLMCore::ContextData context,
|
|
LLMCore::RequestType type)
|
|
{
|
|
if (!prompt->isSupportProvider(providerID())) {
|
|
LOG_MESSAGE(QString("Template %1 doesn't support %2 provider").arg(name(), prompt->name()));
|
|
}
|
|
|
|
prompt->prepareRequest(request, context);
|
|
|
|
auto applyModelParams = [&request](const auto &settings) {
|
|
request["max_tokens"] = settings.maxTokens();
|
|
request["temperature"] = settings.temperature();
|
|
|
|
if (settings.useTopP())
|
|
request["top_p"] = settings.topP();
|
|
if (settings.useTopK())
|
|
request["top_k"] = settings.topK();
|
|
if (settings.useFrequencyPenalty())
|
|
request["frequency_penalty"] = settings.frequencyPenalty();
|
|
if (settings.usePresencePenalty())
|
|
request["presence_penalty"] = settings.presencePenalty();
|
|
};
|
|
|
|
if (type == LLMCore::RequestType::CodeCompletion) {
|
|
applyModelParams(Settings::codeCompletionSettings());
|
|
} else {
|
|
applyModelParams(Settings::chatAssistantSettings());
|
|
}
|
|
|
|
if (supportsTools() && type == LLMCore::RequestType::Chat
|
|
&& Settings::chatAssistantSettings().useTools()) {
|
|
auto toolsDefinitions = m_toolsManager->getToolsDefinitions(Tools::ToolSchemaFormat::OpenAI);
|
|
if (!toolsDefinitions.isEmpty()) {
|
|
request["tools"] = toolsDefinitions;
|
|
LOG_MESSAGE(QString("Added %1 tools to llama.cpp request").arg(toolsDefinitions.size()));
|
|
}
|
|
}
|
|
}
|
|
|
|
QList<QString> LlamaCppProvider::getInstalledModels(const QString &url)
|
|
{
|
|
return {};
|
|
}
|
|
|
|
QList<QString> LlamaCppProvider::validateRequest(
|
|
const QJsonObject &request, LLMCore::TemplateType type)
|
|
{
|
|
if (type == LLMCore::TemplateType::FIM) {
|
|
const auto infillReq = QJsonObject{
|
|
{"model", {}},
|
|
{"input_prefix", {}},
|
|
{"input_suffix", {}},
|
|
{"input_extra", {}},
|
|
{"prompt", {}},
|
|
{"temperature", {}},
|
|
{"top_p", {}},
|
|
{"top_k", {}},
|
|
{"max_tokens", {}},
|
|
{"frequency_penalty", {}},
|
|
{"presence_penalty", {}},
|
|
{"stop", QJsonArray{}},
|
|
{"stream", {}}};
|
|
|
|
return LLMCore::ValidationUtils::validateRequestFields(request, infillReq);
|
|
} else {
|
|
const auto chatReq = QJsonObject{
|
|
{"model", {}},
|
|
{"messages", QJsonArray{{QJsonObject{{"role", {}}, {"content", {}}}}}},
|
|
{"temperature", {}},
|
|
{"max_tokens", {}},
|
|
{"top_p", {}},
|
|
{"top_k", {}},
|
|
{"frequency_penalty", {}},
|
|
{"presence_penalty", {}},
|
|
{"stop", QJsonArray{}},
|
|
{"stream", {}},
|
|
{"tools", {}}};
|
|
|
|
return LLMCore::ValidationUtils::validateRequestFields(request, chatReq);
|
|
}
|
|
}
|
|
|
|
QString LlamaCppProvider::apiKey() const
|
|
{
|
|
return {};
|
|
}
|
|
|
|
void LlamaCppProvider::prepareNetworkRequest(QNetworkRequest &networkRequest) const
|
|
{
|
|
networkRequest.setHeader(QNetworkRequest::ContentTypeHeader, "application/json");
|
|
}
|
|
|
|
LLMCore::ProviderID LlamaCppProvider::providerID() const
|
|
{
|
|
return LLMCore::ProviderID::LlamaCpp;
|
|
}
|
|
|
|
void LlamaCppProvider::sendRequest(
|
|
const LLMCore::RequestID &requestId, const QUrl &url, const QJsonObject &payload)
|
|
{
|
|
if (!m_messages.contains(requestId)) {
|
|
m_dataBuffers[requestId].clear();
|
|
}
|
|
|
|
m_requestUrls[requestId] = url;
|
|
m_originalRequests[requestId] = payload;
|
|
|
|
QNetworkRequest networkRequest(url);
|
|
prepareNetworkRequest(networkRequest);
|
|
|
|
LLMCore::HttpRequest
|
|
request{.networkRequest = networkRequest, .requestId = requestId, .payload = payload};
|
|
|
|
LOG_MESSAGE(
|
|
QString("LlamaCppProvider: Sending request %1 to %2").arg(requestId, url.toString()));
|
|
|
|
emit httpClient()->sendRequest(request);
|
|
}
|
|
|
|
bool LlamaCppProvider::supportsTools() const
|
|
{
|
|
return true;
|
|
}
|
|
|
|
void LlamaCppProvider::cancelRequest(const LLMCore::RequestID &requestId)
|
|
{
|
|
LOG_MESSAGE(QString("LlamaCppProvider: Cancelling request %1").arg(requestId));
|
|
LLMCore::Provider::cancelRequest(requestId);
|
|
cleanupRequest(requestId);
|
|
}
|
|
|
|
void LlamaCppProvider::onDataReceived(
|
|
const QodeAssist::LLMCore::RequestID &requestId, const QByteArray &data)
|
|
{
|
|
LLMCore::DataBuffers &buffers = m_dataBuffers[requestId];
|
|
QStringList lines = buffers.rawStreamBuffer.processData(data);
|
|
|
|
for (const QString &line : lines) {
|
|
if (line.trimmed().isEmpty() || line == "data: [DONE]") {
|
|
continue;
|
|
}
|
|
|
|
QJsonObject chunk = parseEventLine(line);
|
|
if (chunk.isEmpty())
|
|
continue;
|
|
|
|
if (chunk.contains("content")) {
|
|
QString content = chunk["content"].toString();
|
|
if (!content.isEmpty()) {
|
|
buffers.responseContent += content;
|
|
emit partialResponseReceived(requestId, content);
|
|
}
|
|
if (chunk["stop"].toBool()) {
|
|
emit fullResponseReceived(requestId, buffers.responseContent);
|
|
m_dataBuffers.remove(requestId);
|
|
}
|
|
} else if (chunk.contains("choices")) {
|
|
processStreamChunk(requestId, chunk);
|
|
}
|
|
}
|
|
}
|
|
|
|
void LlamaCppProvider::onRequestFinished(
|
|
const QodeAssist::LLMCore::RequestID &requestId, bool success, const QString &error)
|
|
{
|
|
if (!success) {
|
|
LOG_MESSAGE(QString("LlamaCppProvider request %1 failed: %2").arg(requestId, error));
|
|
emit requestFailed(requestId, error);
|
|
cleanupRequest(requestId);
|
|
return;
|
|
}
|
|
|
|
if (m_messages.contains(requestId)) {
|
|
OpenAIMessage *message = m_messages[requestId];
|
|
if (message->state() == LLMCore::MessageState::RequiresToolExecution) {
|
|
LOG_MESSAGE(QString("Waiting for tools to complete for %1").arg(requestId));
|
|
m_dataBuffers.remove(requestId);
|
|
return;
|
|
}
|
|
}
|
|
|
|
if (m_dataBuffers.contains(requestId)) {
|
|
const LLMCore::DataBuffers &buffers = m_dataBuffers[requestId];
|
|
if (!buffers.responseContent.isEmpty()) {
|
|
LOG_MESSAGE(QString("Emitting full response for %1").arg(requestId));
|
|
emit fullResponseReceived(requestId, buffers.responseContent);
|
|
}
|
|
}
|
|
|
|
cleanupRequest(requestId);
|
|
}
|
|
|
|
void LlamaCppProvider::onToolExecutionComplete(
|
|
const QString &requestId, const QHash<QString, QString> &toolResults)
|
|
{
|
|
if (!m_messages.contains(requestId) || !m_requestUrls.contains(requestId)) {
|
|
LOG_MESSAGE(QString("ERROR: Missing data for continuation request %1").arg(requestId));
|
|
cleanupRequest(requestId);
|
|
return;
|
|
}
|
|
|
|
LOG_MESSAGE(QString("Tool execution complete for llama.cpp request %1").arg(requestId));
|
|
|
|
OpenAIMessage *message = m_messages[requestId];
|
|
QJsonObject continuationRequest = m_originalRequests[requestId];
|
|
QJsonArray messages = continuationRequest["messages"].toArray();
|
|
|
|
messages.append(message->toProviderFormat());
|
|
|
|
QJsonArray toolResultMessages = message->createToolResultMessages(toolResults);
|
|
for (const auto &toolMsg : toolResultMessages) {
|
|
messages.append(toolMsg);
|
|
}
|
|
|
|
continuationRequest["messages"] = messages;
|
|
|
|
LOG_MESSAGE(QString("Sending continuation request for %1 with %2 tool results")
|
|
.arg(requestId)
|
|
.arg(toolResults.size()));
|
|
|
|
sendRequest(requestId, m_requestUrls[requestId], continuationRequest);
|
|
}
|
|
|
|
void LlamaCppProvider::processStreamChunk(const QString &requestId, const QJsonObject &chunk)
|
|
{
|
|
QJsonArray choices = chunk["choices"].toArray();
|
|
if (choices.isEmpty()) {
|
|
return;
|
|
}
|
|
|
|
QJsonObject choice = choices[0].toObject();
|
|
QJsonObject delta = choice["delta"].toObject();
|
|
QString finishReason = choice["finish_reason"].toString();
|
|
|
|
OpenAIMessage *message = m_messages.value(requestId);
|
|
if (!message) {
|
|
message = new OpenAIMessage(this);
|
|
m_messages[requestId] = message;
|
|
LOG_MESSAGE(QString("Created NEW OpenAIMessage for llama.cpp request %1").arg(requestId));
|
|
}
|
|
|
|
if (delta.contains("content") && !delta["content"].isNull()) {
|
|
QString content = delta["content"].toString();
|
|
message->handleContentDelta(content);
|
|
|
|
LLMCore::DataBuffers &buffers = m_dataBuffers[requestId];
|
|
buffers.responseContent += content;
|
|
emit partialResponseReceived(requestId, content);
|
|
}
|
|
|
|
if (delta.contains("tool_calls")) {
|
|
QJsonArray toolCalls = delta["tool_calls"].toArray();
|
|
for (const auto &toolCallValue : toolCalls) {
|
|
QJsonObject toolCall = toolCallValue.toObject();
|
|
int index = toolCall["index"].toInt();
|
|
|
|
if (toolCall.contains("id")) {
|
|
QString id = toolCall["id"].toString();
|
|
QJsonObject function = toolCall["function"].toObject();
|
|
QString name = function["name"].toString();
|
|
message->handleToolCallStart(index, id, name);
|
|
}
|
|
|
|
if (toolCall.contains("function")) {
|
|
QJsonObject function = toolCall["function"].toObject();
|
|
if (function.contains("arguments")) {
|
|
QString args = function["arguments"].toString();
|
|
message->handleToolCallDelta(index, args);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
if (!finishReason.isEmpty() && finishReason != "null") {
|
|
for (int i = 0; i < 10; ++i) {
|
|
message->handleToolCallComplete(i);
|
|
}
|
|
|
|
message->handleFinishReason(finishReason);
|
|
handleMessageComplete(requestId);
|
|
}
|
|
}
|
|
|
|
void LlamaCppProvider::handleMessageComplete(const QString &requestId)
|
|
{
|
|
if (!m_messages.contains(requestId))
|
|
return;
|
|
|
|
OpenAIMessage *message = m_messages[requestId];
|
|
|
|
if (message->state() == LLMCore::MessageState::RequiresToolExecution) {
|
|
LOG_MESSAGE(QString("llama.cpp message requires tool execution for %1").arg(requestId));
|
|
|
|
auto toolUseContent = message->getCurrentToolUseContent();
|
|
|
|
if (toolUseContent.isEmpty()) {
|
|
LOG_MESSAGE(QString("No tools to execute for %1").arg(requestId));
|
|
return;
|
|
}
|
|
|
|
for (auto toolContent : toolUseContent) {
|
|
m_toolsManager->executeToolCall(
|
|
requestId, toolContent->id(), toolContent->name(), toolContent->input());
|
|
}
|
|
|
|
} else {
|
|
LOG_MESSAGE(QString("llama.cpp message marked as complete for %1").arg(requestId));
|
|
}
|
|
}
|
|
|
|
void LlamaCppProvider::cleanupRequest(const LLMCore::RequestID &requestId)
|
|
{
|
|
LOG_MESSAGE(QString("Cleaning up llama.cpp request %1").arg(requestId));
|
|
|
|
if (m_messages.contains(requestId)) {
|
|
OpenAIMessage *message = m_messages.take(requestId);
|
|
message->deleteLater();
|
|
}
|
|
|
|
m_dataBuffers.remove(requestId);
|
|
m_requestUrls.remove(requestId);
|
|
m_originalRequests.remove(requestId);
|
|
m_toolsManager->cleanupRequest(requestId);
|
|
}
|
|
|
|
} // namespace QodeAssist::Providers
|