From c8b007631d3d6c26197f9c7724a65016bb498550 Mon Sep 17 00:00:00 2001 From: end-4 <97237370+end-4@users.noreply.github.com> Date: Fri, 25 Jul 2025 20:14:37 +0700 Subject: [PATCH] ai: refractor api formats --- .config/quickshell/ii/services/Ai.qml | 362 ++++-------------- .../ii/services/{ => ai}/AiMessageData.qml | 1 - .config/quickshell/ii/services/ai/AiModel.qml | 33 ++ .../quickshell/ii/services/ai/ApiStrategy.qml | 10 + .../ii/services/ai/GeminiApiStrategy.qml | 156 ++++++++ .../ii/services/ai/OpenAiApiStrategy.qml | 97 +++++ 6 files changed, 379 insertions(+), 280 deletions(-) rename .config/quickshell/ii/services/{ => ai}/AiMessageData.qml (95%) create mode 100644 .config/quickshell/ii/services/ai/AiModel.qml create mode 100644 .config/quickshell/ii/services/ai/ApiStrategy.qml create mode 100644 .config/quickshell/ii/services/ai/GeminiApiStrategy.qml create mode 100644 .config/quickshell/ii/services/ai/OpenAiApiStrategy.qml diff --git a/.config/quickshell/ii/services/Ai.qml b/.config/quickshell/ii/services/Ai.qml index 82386fad..f7365255 100644 --- a/.config/quickshell/ii/services/Ai.qml +++ b/.config/quickshell/ii/services/Ai.qml @@ -7,6 +7,7 @@ import qs import Quickshell import Quickshell.Io import QtQuick +import "./ai/" /** * Basic service to handle LLM chats. Supports Google's and OpenAI's API formats. @@ -14,9 +15,13 @@ import QtQuick Singleton { id: root + property Component aiMessageComponent: AiMessageData {} + property Component aiModelComponent: AiModel {} + property Component geminiApiStrategy: GeminiApiStrategy {} + property Component openaiApiStrategy: OpenAiApiStrategy {} readonly property string interfaceRole: "interface" readonly property string apiKeyEnvVarName: "API_KEY" - property Component aiMessageComponent: AiMessageData {} + property string systemPrompt: Config.options?.ai?.systemPrompt ?? "" // property var messages: [] property var messageIDs: [] @@ -126,7 +131,7 @@ Singleton { // - tools: List of tools that the model can use. Each tool is an object with the tool name as the key and an empty object as the value. // - extraParams: Extra parameters to be passed to the model. This is a JSON object. property var models: { - "gemini-2.0-flash-search": { + "gemini-2.0-flash-search": aiModelComponent.createObject(this, { "name": "Gemini 2.0 Flash (Search)", "icon": "google-gemini-symbolic", "description": Translation.tr("Online | Google's model\nGives up-to-date information with search."), @@ -141,8 +146,8 @@ Singleton { "tools": [{ "google_search": {} }] - }, - "gemini-2.0-flash-tools": { + }), + "gemini-2.0-flash-tools": aiModelComponent.createObject(this, { "name": "Gemini 2.0 Flash (Tools)", "icon": "google-gemini-symbolic", "description": Translation.tr("Experimental | Online | Google's model\nCan do a little more but takes an extra turn to perform search"), @@ -155,8 +160,8 @@ Singleton { "key_get_description": Translation.tr("**Pricing**: free. Data used for training.\n\n**Instructions**: Log into Google account, allow AI Studio to create Google Cloud project or whatever it asks, go back and click Get API key"), "api_format": "gemini", "tools": root.tools["gemini"], - }, - "gemini-2.5-flash-search": { + }), + "gemini-2.5-flash-search": aiModelComponent.createObject(this, { "name": "Gemini 2.5 Flash (Search)", "icon": "google-gemini-symbolic", "description": Translation.tr("Online | Google's model\nGives up-to-date information with search."), @@ -171,8 +176,8 @@ Singleton { "tools": [{ "google_search": {} }] - }, - "gemini-2.5-flash-tools": { + }), + "gemini-2.5-flash-tools": aiModelComponent.createObject(this, { "name": "Gemini 2.5 Flash (Tools)", "icon": "google-gemini-symbolic", "description": Translation.tr("Experimental | Online | Google's model\nCan do a little more but takes an extra turn to perform search"), @@ -185,21 +190,8 @@ Singleton { "key_get_description": Translation.tr("**Pricing**: free. Data used for training.\n\n**Instructions**: Log into Google account, allow AI Studio to create Google Cloud project or whatever it asks, go back and click Get API key"), "api_format": "gemini", "tools": root.tools["gemini"], - }, - "gemini-2.5-flash-lite": { - "name": "Gemini 2.5 Flash-Lite", - "icon": "google-gemini-symbolic", - "description": Translation.tr("Experimental | Online | Google's model\nA Gemini 2.5 Flash model optimized for cost-efficiency and high throughput."), - "homepage": "https://aistudio.google.com", - "endpoint": "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash-lite:streamGenerateContent", - "model": "gemini-2.5-flash-lite", - "requires_key": true, - "key_id": "gemini", - "key_get_link": "https://aistudio.google.com/app/apikey", - "key_get_description": Translation.tr("**Pricing**: free. Data used for training.\n\n**Instructions**: Log into Google account, allow AI Studio to create Google Cloud project or whatever it asks, go back and click Get API key"), - "api_format": "gemini", - }, - "gemini-2.5-flash-lite-search": { + }), + "gemini-2.5-flash-lite-search": aiModelComponent.createObject(this, { "name": "Gemini 2.5 Flash-Lite (Search)", "icon": "google-gemini-symbolic", "description": Translation.tr("Experimental | Online | Google's model\nA Gemini 2.5 Flash model optimized for cost-efficiency and high throughput."), @@ -214,8 +206,22 @@ Singleton { "tools": [{ "google_search": {} }] - }, - "openrouter-llama4-maverick": { + }), + "gemini-2.5-flash-lite": aiModelComponent.createObject(this, { + "name": "Gemini 2.5 Flash-Lite", + "icon": "google-gemini-symbolic", + "description": Translation.tr("Experimental | Online | Google's model\nA Gemini 2.5 Flash model optimized for cost-efficiency and high throughput."), + "homepage": "https://aistudio.google.com", + "endpoint": "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash-lite:streamGenerateContent", + "model": "gemini-2.5-flash-lite", + "requires_key": true, + "key_id": "gemini", + "key_get_link": "https://aistudio.google.com/app/apikey", + "key_get_description": Translation.tr("**Pricing**: free. Data used for training.\n\n**Instructions**: Log into Google account, allow AI Studio to create Google Cloud project or whatever it asks, go back and click Get API key"), + "api_format": "gemini", + "tools": root.tools["gemini"], + }), + "openrouter-llama4-maverick": aiModelComponent.createObject(this, { "name": "Llama 4 Maverick", "icon": "ollama-symbolic", "description": Translation.tr("Online via %1 | %2's model").arg("OpenRouter").arg("Meta"), @@ -226,8 +232,8 @@ Singleton { "key_id": "openrouter", "key_get_link": "https://openrouter.ai/settings/keys", "key_get_description": Translation.tr("**Pricing**: free. Data use policy varies depending on your OpenRouter account settings.\n\n**Instructions**: Log into OpenRouter account, go to Keys on the topright menu, click Create API Key"), - }, - "openrouter-deepseek-r1": { + }), + "openrouter-deepseek-r1": aiModelComponent.createObject(this, { "name": "DeepSeek R1", "icon": "deepseek-symbolic", "description": Translation.tr("Online via %1 | %2's model").arg("OpenRouter").arg("DeepSeek"), @@ -238,11 +244,17 @@ Singleton { "key_id": "openrouter", "key_get_link": "https://openrouter.ai/settings/keys", "key_get_description": Translation.tr("**Pricing**: free. Data use policy varies depending on your OpenRouter account settings.\n\n**Instructions**: Log into OpenRouter account, go to Keys on the topright menu, click Create API Key"), - }, + }), } property var modelList: Object.keys(root.models) property var currentModelId: Persistent.states?.ai?.model || modelList[0] + property var apiStrategies: { + "openai": openaiApiStrategy.createObject(this), + "gemini": geminiApiStrategy.createObject(this), + } + property ApiStrategy currentApiStrategy: apiStrategies[models[currentModelId]?.api_format || "openai"] + Component.onCompleted: { setModel(currentModelId, false, false); // Do necessary setup for model } @@ -280,14 +292,15 @@ Singleton { root.modelList = [...root.modelList, ...dataJson]; dataJson.forEach(model => { const safeModelName = root.safeModelName(model); - root.models[safeModelName] = { + root.models[safeModelName] = aiModelComponent.createObject(this, { "name": guessModelName(model), "icon": guessModelLogo(model), "description": Translation.tr("Local Ollama model | %1").arg(model), "homepage": `https://ollama.com/library/${model}`, "endpoint": "http://localhost:11434/v1/chat/completions", "model": model, - } + "requires_key": false, + }) }); root.modelList = Object.keys(root.models); @@ -473,24 +486,16 @@ Singleton { function clearMessages() { root.messageIDs = []; root.messageByID = ({}); + root.tokenCount.input = -1; + root.tokenCount.output = -1; + root.tokenCount.total = -1; } Process { id: requester property var baseCommand: ["bash", "-c"] - property var message - property bool isReasoning - property string apiFormat: "openai" - property string geminiBuffer: "" - - function buildGeminiEndpoint(model) { - // console.log("ENDPOINT: " + model.endpoint + `?key=\$\{${root.apiKeyEnvVarName}\}`) - return model.endpoint + `?key=\$\{${root.apiKeyEnvVarName}\}`; - } - - function buildOpenAIEndpoint(model) { - return model.endpoint; - } + property AiMessageData message + property ApiStrategy currentStrategy function markDone() { requester.message.done = true; @@ -501,84 +506,20 @@ Singleton { root.saveChat("lastSession") } - function buildGeminiRequestData(model, messages) { - const tools = [ - ...(model.tools ?? root.tools[model.api_format]), - ] - // console.log("Tools", JSON.stringify(tools, null, 2)); - let baseData = { - "contents": messages.filter(message => (message.role != Ai.interfaceRole)).map(message => { - const geminiApiRoleName = (message.role === "assistant") ? "model" : message.role; - const usingSearch = tools[0].google_search != undefined - if (!usingSearch && message.functionCall != undefined && message.functionCall.length > 0) { - return { - "role": geminiApiRoleName, - "parts": [{ - functionCall: { - "name": message.functionName, - } - }] - } - } - if (!usingSearch && message.functionResponse != undefined && message.functionResponse.length > 0) { - return { - "role": geminiApiRoleName, - "parts": [{ - functionResponse: { - "name": message.functionName, - "response": { "content": message.functionResponse } - } - }] - } - } - return { - "role": geminiApiRoleName, - "parts": [{ - text: message.rawContent, - }] - } - }), - "tools": tools, - "system_instruction": { - "parts": [{ text: root.systemPrompt }] - }, - "generationConfig": { - "temperature": root.temperature, - }, - }; - return model.extraParams ? Object.assign({}, baseData, model.extraParams) : baseData; - } - - function buildOpenAIRequestData(model, messages) { - let baseData = { - "model": model.model, - "messages": [ - {role: "system", content: root.systemPrompt}, - ...messages.filter(message => (message.role != Ai.interfaceRole)).map(message => { - return { - "role": message.role, - "content": message.rawContent, - } - }), - ], - "stream": true, - "temperature": root.temperature, - }; - return model.extraParams ? Object.assign({}, baseData, model.extraParams) : baseData; - } - function makeRequest() { const model = models[currentModelId]; - requester.apiFormat = model.api_format ?? "openai"; + requester.currentStrategy = root.currentApiStrategy; + requester.currentStrategy.reset(); // Reset strategy state /* Put API key in environment variable */ if (model.requires_key) requester.environment[`${root.apiKeyEnvVarName}`] = root.apiKeys ? (root.apiKeys[model.key_id] ?? "") : "" /* Build endpoint, request data */ - const endpoint = (apiFormat === "gemini") ? buildGeminiEndpoint(model) : buildOpenAIEndpoint(model); + const endpoint = root.currentApiStrategy.buildEndpoint(model); const messageArray = root.messageIDs.map(id => root.messageByID[id]); - const data = (apiFormat === "gemini") ? buildGeminiRequestData(model, messageArray) : buildOpenAIRequestData(model, messageArray); - // console.log("REQUEST DATA: ", JSON.stringify(data, null, 2)); + const filteredMessageArray = messageArray.filter(message => message.role !== Ai.interfaceRole); + const data = root.currentApiStrategy.buildRequestData(model, filteredMessageArray, root.systemPrompt, root.temperature); + // console.log("[Ai] Request data: ", JSON.stringify(data, null, 2)); let requestHeaders = { "Content-Type": "application/json", @@ -606,166 +547,45 @@ Singleton { // console.log("Request headers: ", JSON.stringify(requestHeaders)); // console.log("Header string: ", headerString); + /* Get authorization header from strategy */ + const authHeader = requester.currentStrategy.buildAuthorizationHeader(root.apiKeyEnvVarName); + /* Create command string */ const requestCommandString = `curl --no-buffer "${endpoint}"` + ` ${headerString}` - + ((apiFormat == "gemini") ? "" : ` -H "Authorization: Bearer \$\{${root.apiKeyEnvVarName}\}"`) + + (authHeader ? ` ${authHeader}` : "") + ` -d '${CF.StringUtils.shellSingleQuoteEscape(JSON.stringify(data))}'` - // console.log("Request command: ", requestCommandString); + + /* Send the request */ requester.command = baseCommand.concat([requestCommandString]); - - /* Reset vars and make the request */ - requester.isReasoning = false requester.running = true } - function parseGeminiBuffer() { - // console.log("BUFFER DATA: ", requester.geminiBuffer); - try { - if (requester.geminiBuffer.length === 0) return; - const dataJson = JSON.parse(requester.geminiBuffer); - if (!dataJson.candidates) return; - - if (dataJson.candidates[0]?.finishReason) { - requester.markDone(); - } - // Function call handling - if (dataJson.candidates[0]?.content?.parts[0]?.functionCall) { - const functionCall = dataJson.candidates[0]?.content?.parts[0]?.functionCall; - requester.message.functionName = functionCall.name; - requester.message.functionCall = functionCall.name; - const newContent = `\n\n[[ Function: ${functionCall.name}(${JSON.stringify(functionCall.args, null, 2)}) ]]\n` - requester.message.rawContent += newContent; - requester.message.content += newContent; - root.handleGeminiFunctionCall(functionCall.name, functionCall.args); - return - } - - // Normal text response - const responseContent = dataJson.candidates[0]?.content?.parts[0]?.text - requester.message.rawContent += responseContent; - requester.message.content += responseContent; - const annotationSources = dataJson.candidates[0]?.groundingMetadata?.groundingChunks?.map(chunk => { - return { - "type": "url_citation", - "text": chunk?.web?.title, - "url": chunk?.web?.uri, - } - }) ?? []; - - // Handle annotations and search queries - const annotations = dataJson.candidates[0]?.groundingMetadata?.groundingSupports?.map(citation => { - return { - "type": "url_citation", - "start_index": citation.segment?.startIndex, - "end_index": citation.segment?.endIndex, - "text": citation?.segment.text, - "url": annotationSources[citation.groundingChunkIndices[0]]?.url, - "sources": citation.groundingChunkIndices - } - }); - requester.message.annotationSources = annotationSources; - requester.message.annotations = annotations; - requester.message.searchQueries = dataJson.candidates[0]?.groundingMetadata?.webSearchQueries ?? []; - // console.log("[AI] Gemini: Search queries: ", JSON.stringify(requester.message.searchQueries, null, 2)); - - // Usage - root.tokenCount.input = dataJson.usageMetadata?.promptTokenCount ?? -1; - root.tokenCount.output = dataJson.usageMetadata?.candidatesTokenCount ?? -1; - root.tokenCount.total = dataJson.usageMetadata?.totalTokenCount ?? -1; - // console.log("[AI] Gemini: Token count: ", root.tokenCount); - - // Last logging - // console.log(JSON.stringify(requester.message, null, 2)); - } catch (e) { - console.log("[AI] Gemini: Could not parse buffer: ", e); - requester.message.rawContent += requester.geminiBuffer; - requester.message.content += requester.geminiBuffer - } finally { - requester.geminiBuffer = ""; - } - } - - function handleGeminiResponseLine(line) { - if (line.startsWith("[")) { - requester.geminiBuffer += line.slice(1).trim(); - } else if (line == "]") { - requester.geminiBuffer += line.slice(0, -1).trim(); - parseGeminiBuffer(); - } else if (line.startsWith(",")) { // end of one entry - parseGeminiBuffer(); - } else { - requester.geminiBuffer += line.trim(); - } - } - - function handleOpenAIResponseLine(line) { - // Remove 'data: ' prefix if present and trim whitespace - let cleanData = line.trim(); - if (cleanData.startsWith("data:")) { - cleanData = cleanData.slice(5).trim(); - } - // console.log("Clean data: ", cleanData); - if (!cleanData || cleanData.startsWith(":")) return; - - if (cleanData === "[DONE]") { - requester.markDone(); - return; - } - const dataJson = JSON.parse(cleanData); - - let newContent = ""; - const responseContent = dataJson.choices[0]?.delta?.content || dataJson.message?.content; - const responseReasoning = dataJson.choices[0]?.delta?.reasoning || dataJson.choices[0]?.delta?.reasoning_content; - - if (responseContent && responseContent.length > 0) { - if (requester.isReasoning) { - requester.isReasoning = false; - const endBlock = "\n\n\n\n"; - requester.message.content += endBlock; - requester.message.rawContent += endBlock; - } - newContent = dataJson.choices[0]?.delta?.content || dataJson.message.content; - } else if (responseReasoning && responseReasoning.length > 0) { - // console.log("Reasoning content: ", dataJson.choices[0].delta.reasoning); - if (!requester.isReasoning) { - requester.isReasoning = true; - const startBlock = "\n\n\n\n"; - requester.message.rawContent += startBlock; - requester.message.content += startBlock; - } - newContent = dataJson.choices[0].delta.reasoning || dataJson.choices[0].delta.reasoning_content; - } - - requester.message.content += newContent; - requester.message.rawContent += newContent; - - if (dataJson.done) { - requester.markDone(); - } - } - stdout: SplitParser { onRead: data => { - // console.log("RAW DATA: ", data); + // console.log("[Ai] Raw response line: ", data); if (data.length === 0) return; + if (requester.message.thinking) requester.message.thinking = false; // Handle response line - if (requester.message.thinking) requester.message.thinking = false; try { - if (requester.apiFormat === "gemini") { - requester.handleGeminiResponseLine(data); + const result = requester.currentStrategy.parseResponseLine(data, requester.message); + // console.log("[Ai] Parsed response result: ", JSON.stringify(result, null, 2)); + + if (result.functionCall) { + root.handleFunctionCall(result.functionCall.name, result.functionCall.args); } - else if (requester.apiFormat === "openai") { - requester.handleOpenAIResponseLine(data); + if (result.tokenUsage) { + root.tokenCount.input = result.tokenUsage.input; + root.tokenCount.output = result.tokenUsage.output; + root.tokenCount.total = result.tokenUsage.total; } - else { - console.log("Unknown API format: ", requester.apiFormat); - requester.message.rawContent += data; - requester.message.content += data; + if (result.finished) { + requester.markDone(); } + } catch (e) { - console.log("[AI] Could not parse response from stream: ", e); + console.log("[AI] Could not parse response: ", e); requester.message.rawContent += data; requester.message.content += data; } @@ -773,18 +593,15 @@ Singleton { } onExited: (exitCode, exitStatus) => { - if (requester.apiFormat == "gemini") requester.parseGeminiBuffer(); - else requester.markDone(); - - try { // to parse full response into json for error handling - // console.log("Full response: ", requester.message.content + "]"); - const parsedResponse = JSON.parse(requester.message.rawContent + "]"); - requester.message.rawContent = `\`\`\`json\n${JSON.stringify(parsedResponse, null, 2)}\n\`\`\``; - requester.message.content = requester.message.rawContent; - } catch (e) { - // console.log("[AI] Could not parse response on exit: ", e); + const result = requester.currentStrategy.onRequestFinished(requester.message); + + if (result.finished) { + requester.markDone(); + } else if (!requester.message.done) { + requester.markDone(); } + // Handle error responses if (requester.message.content.includes("API key not valid")) { root.addApiKeyAdvice(models[requester.message.model]); } @@ -814,20 +631,7 @@ Singleton { root.messageByID[id] = aiMessage; } - function buildGeminiFunctionOutput(name, output) { - const functionResponsePart = { - "name": name, - "response": { "content": output } - } - return { - "role": "user", - "parts": [{ - functionResponse: functionResponsePart, - }] - } - } - - function handleGeminiFunctionCall(name, args) { + function handleFunctionCall(name, args) { if (name === "switch_to_search_mode") { const modelId = root.currentModelId; if (modelId.endsWith("-tools")) { diff --git a/.config/quickshell/ii/services/AiMessageData.qml b/.config/quickshell/ii/services/ai/AiMessageData.qml similarity index 95% rename from .config/quickshell/ii/services/AiMessageData.qml rename to .config/quickshell/ii/services/ai/AiMessageData.qml index 4695afe5..8813a70b 100644 --- a/.config/quickshell/ii/services/AiMessageData.qml +++ b/.config/quickshell/ii/services/ai/AiMessageData.qml @@ -1,4 +1,3 @@ -import qs.modules.common import QtQuick; /** diff --git a/.config/quickshell/ii/services/ai/AiModel.qml b/.config/quickshell/ii/services/ai/AiModel.qml new file mode 100644 index 00000000..5544cf16 --- /dev/null +++ b/.config/quickshell/ii/services/ai/AiModel.qml @@ -0,0 +1,33 @@ +import QtQuick; + +/** + * An AI model representation. + * - name: Friendly name of the model + * - icon: Icon name of the model + * - description: Description of the model + * - endpoint: Endpoint of the model + * - model: Model code (like gpt-4.1 or gemini-2.5-flash) + * - requires_key: Whether the model requires an API key + * - key_id: The identifier of the API key. Use the same identifier for models that can be accessed with the same key. + * - key_get_link: Link to get an API key + * - key_get_description: Description of pricing and how to get an API key + * - api_format: The API format of the model. Can be "openai" or "gemini". Default is "openai". + * - tools: List of tools that the model can use. + * - extraParams: Extra parameters to be passed to the model. This is a JSON object. + */ + +QtObject { + property string name + property string icon + property string description + property string homepage + property string endpoint + property string model + property bool requires_key: true + property string key_id + property string key_get_link + property string key_get_description + property string api_format: "openai" + property var tools + property var extraParams: ({}) +} diff --git a/.config/quickshell/ii/services/ai/ApiStrategy.qml b/.config/quickshell/ii/services/ai/ApiStrategy.qml new file mode 100644 index 00000000..d63f932d --- /dev/null +++ b/.config/quickshell/ii/services/ai/ApiStrategy.qml @@ -0,0 +1,10 @@ +import QtQuick + +QtObject { + function buildEndpoint(model: AiModel): string { throw new Error("Not implemented") } + function buildRequestData(model: AiModel, messages, systemPrompt: string, temperature: real) { throw new Error("Not implemented") } + function buildAuthorizationHeader(apiKeyEnvVarName: string): string { throw new Error("Not implemented") } + function parseResponseLine(line: string, message: AiMessageData) { throw new Error("Not implemented") } + function onRequestFinished(message: AiMessageData): var { return {} } // Default: no special handling + function reset() { } // Reset any internal state if needed +} diff --git a/.config/quickshell/ii/services/ai/GeminiApiStrategy.qml b/.config/quickshell/ii/services/ai/GeminiApiStrategy.qml new file mode 100644 index 00000000..733dda51 --- /dev/null +++ b/.config/quickshell/ii/services/ai/GeminiApiStrategy.qml @@ -0,0 +1,156 @@ +import QtQuick + +ApiStrategy { + property string buffer: "" + + function buildEndpoint(model: AiModel): string { + const result = model.endpoint + `?key=\$\{${root.apiKeyEnvVarName}\}` + console.log("[AI] Endpoint: " + result); + return result; + } + + function buildRequestData(model: AiModel, messages, systemPrompt: string, temperature: real) { + const tools = model.tools ?? []; + let baseData = { + "contents": messages.map(message => { + const geminiApiRoleName = (message.role === "assistant") ? "model" : message.role; + const usingSearch = tools[0].google_search != undefined + if (!usingSearch && message.functionCall != undefined && message.functionCall.length > 0) { + return { + "role": geminiApiRoleName, + "parts": [{ + functionCall: { + "name": message.functionName, + } + }] + } + } + if (!usingSearch && message.functionResponse != undefined && message.functionResponse.length > 0) { + return { + "role": geminiApiRoleName, + "parts": [{ + functionResponse: { + "name": message.functionName, + "response": { "content": message.functionResponse } + } + }] + } + } + return { + "role": geminiApiRoleName, + "parts": [{ + text: message.rawContent, + }] + } + }), + "tools": tools, + "system_instruction": { + "parts": [{ text: systemPrompt }] + }, + "generationConfig": { + "temperature": temperature, + }, + }; + return model.extraParams ? Object.assign({}, baseData, model.extraParams) : baseData; + } + + function buildAuthorizationHeader(apiKeyEnvVarName: string): string { + // Gemini doesn't use Authorization header, key is in URL + return ""; + } + + function parseResponseLine(line, message) { + if (line.startsWith("[")) { + buffer += line.slice(1).trim(); + } else if (line === "]") { + buffer += line.slice(0, -1).trim(); + return parseBuffer(message); + } else if (line.startsWith(",")) { + return parseBuffer(message); + } else { + buffer += line.trim(); + } + return {}; + } + + function parseBuffer(message) { + // console.log("[Ai] Gemini buffer: ", buffer); + let finished = false; + try { + if (buffer.length === 0) return {}; + const dataJson = JSON.parse(buffer); + if (!dataJson.candidates) return {}; + + if (dataJson.candidates[0]?.finishReason) { + finished = true; + } + + // Function call handling + if (dataJson.candidates[0]?.content?.parts[0]?.functionCall) { + const functionCall = dataJson.candidates[0]?.content?.parts[0]?.functionCall; + message.functionName = functionCall.name; + message.functionCall = functionCall.name; + const newContent = `\n\n[[ Function: ${functionCall.name}(${JSON.stringify(functionCall.args, null, 2)}) ]]\n` + message.rawContent += newContent; + message.content += newContent; + return { functionCall: { name: functionCall.name, args: functionCall.args }, finished: finished }; + } + + // Normal text response + const responseContent = dataJson.candidates[0]?.content?.parts[0]?.text + message.rawContent += responseContent; + message.content += responseContent; + + // Handle annotations and metadata + const annotationSources = dataJson.candidates[0]?.groundingMetadata?.groundingChunks?.map(chunk => { + return { + "type": "url_citation", + "text": chunk?.web?.title, + "url": chunk?.web?.uri, + } + }) ?? []; + + const annotations = dataJson.candidates[0]?.groundingMetadata?.groundingSupports?.map(citation => { + return { + "type": "url_citation", + "start_index": citation.segment?.startIndex, + "end_index": citation.segment?.endIndex, + "text": citation?.segment.text, + "url": annotationSources[citation.groundingChunkIndices[0]]?.url, + "sources": citation.groundingChunkIndices + } + }); + message.annotationSources = annotationSources; + message.annotations = annotations; + message.searchQueries = dataJson.candidates[0]?.groundingMetadata?.webSearchQueries ?? []; + + // Usage metadata + if (dataJson.usageMetadata) { + return { + tokenUsage: { + input: dataJson.usageMetadata.promptTokenCount ?? -1, + output: dataJson.usageMetadata.candidatesTokenCount ?? -1, + total: dataJson.usageMetadata.totalTokenCount ?? -1 + }, + finished: finished + }; + } + + } catch (e) { + console.log("[AI] Gemini: Could not parse buffer: ", e); + message.rawContent += buffer; + message.content += buffer; + } finally { + buffer = ""; + } + return { finished: finished }; + } + + function onRequestFinished(message) { + return parseBuffer(message); + } + + function reset() { + buffer = ""; + } +} diff --git a/.config/quickshell/ii/services/ai/OpenAiApiStrategy.qml b/.config/quickshell/ii/services/ai/OpenAiApiStrategy.qml new file mode 100644 index 00000000..ebc00970 --- /dev/null +++ b/.config/quickshell/ii/services/ai/OpenAiApiStrategy.qml @@ -0,0 +1,97 @@ +import QtQuick + +ApiStrategy { + property bool isReasoning: false + + function buildEndpoint(model: AiModel): string { + console.log("[AI] Endpoint: " + model.endpoint); + return model.endpoint; + } + + function buildRequestData(model: AiModel, messages, systemPrompt: string, temperature: real) { + let baseData = { + "model": model.model, + "messages": [ + {role: "system", content: systemPrompt}, + ...messages.map(message => { + return { + "role": message.role, + "content": message.rawContent, + } + }), + ], + "stream": true, + "temperature": temperature, + }; + return model.extraParams ? Object.assign({}, baseData, model.extraParams) : baseData; + } + + function buildAuthorizationHeader(apiKeyEnvVarName: string): string { + return `-H "Authorization: Bearer \$\{${apiKeyEnvVarName}\}"`; + } + + function parseResponseLine(line, message) { + // Remove 'data: ' prefix if present and trim whitespace + let cleanData = line.trim(); + if (cleanData.startsWith("data:")) { + cleanData = cleanData.slice(5).trim(); + } + + // Handle special cases + if (!cleanData || cleanData.startsWith(":")) return {}; + if (cleanData === "[DONE]") { + return { finished: true }; + } + + // Real stuff + try { + const dataJson = JSON.parse(cleanData); + let newContent = ""; + + const responseContent = dataJson.choices[0]?.delta?.content || dataJson.message?.content; + const responseReasoning = dataJson.choices[0]?.delta?.reasoning || dataJson.choices[0]?.delta?.reasoning_content; + + if (responseContent && responseContent.length > 0) { + if (isReasoning) { + isReasoning = false; + const endBlock = "\n\n\n\n"; + message.content += endBlock; + message.rawContent += endBlock; + } + newContent = responseContent; + } else if (responseReasoning && responseReasoning.length > 0) { + if (!isReasoning) { + isReasoning = true; + const startBlock = "\n\n\n\n"; + message.rawContent += startBlock; + message.content += startBlock; + } + newContent = responseReasoning; + } + + message.content += newContent; + message.rawContent += newContent; + + if (dataJson.done) { + return { finished: true }; + } + + } catch (e) { + console.log("[AI] OpenAI: Could not parse line: ", e); + message.rawContent += line; + message.content += line; + } + + return {}; + } + + function onRequestFinished(message) { + // OpenAI format doesn't need special finish handling + return {}; + } + + function reset() { + isReasoning = false; + } + +}