merlyn/server/utils/AiProviders/gemini/index.js
2026-02-02 20:11:18 -08:00

458 lines
14 KiB
JavaScript

const fs = require("fs");
const path = require("path");
const { NativeEmbedder } = require("../../EmbeddingEngines/native");
const {
LLMPerformanceMonitor,
} = require("../../helpers/chat/LLMPerformanceMonitor");
const {
formatChatHistory,
handleDefaultStreamResponseV2,
} = require("../../helpers/chat/responses");
const { MODEL_MAP } = require("../modelMap");
const { defaultGeminiModels, v1BetaModels } = require("./defaultModels");
const { safeJsonParse } = require("../../http");
const cacheFolder = path.resolve(
process.env.STORAGE_DIR
? path.resolve(process.env.STORAGE_DIR, "models", "gemini")
: path.resolve(__dirname, `../../../storage/models/gemini`)
);
const NO_SYSTEM_PROMPT_MODELS = [
"gemma-3-1b-it",
"gemma-3-4b-it",
"gemma-3-12b-it",
"gemma-3-27b-it",
];
class GeminiLLM {
constructor(embedder = null, modelPreference = null) {
if (!process.env.GEMINI_API_KEY)
throw new Error("No Gemini API key was set.");
this.className = "GeminiLLM";
const { OpenAI: OpenAIApi } = require("openai");
this.model =
modelPreference ||
process.env.GEMINI_LLM_MODEL_PREF ||
"gemini-2.0-flash-lite";
const isExperimental = this.isExperimentalModel(this.model);
this.openai = new OpenAIApi({
apiKey: process.env.GEMINI_API_KEY,
// Even models that are v1 in gemini API can be used with v1beta/openai/ endpoint and nobody knows why.
baseURL: "https://generativelanguage.googleapis.com/v1beta/openai/",
});
this.limits = {
history: this.promptWindowLimit() * 0.15,
system: this.promptWindowLimit() * 0.15,
user: this.promptWindowLimit() * 0.7,
};
this.embedder = embedder ?? new NativeEmbedder();
this.defaultTemp = 0.7;
if (!fs.existsSync(cacheFolder))
fs.mkdirSync(cacheFolder, { recursive: true });
this.cacheModelPath = path.resolve(cacheFolder, "models.json");
this.cacheAtPath = path.resolve(cacheFolder, ".cached_at");
this.#log(
`Initialized with model: ${this.model} ${isExperimental ? "[Experimental v1beta]" : "[Stable v1]"} - ctx: ${this.promptWindowLimit()}`
);
}
/**
* Checks if the model supports system prompts
* This is a static list of models that are known to not support system prompts
* since this information is not available in the API model response.
* @returns {boolean}
*/
get supportsSystemPrompt() {
return !NO_SYSTEM_PROMPT_MODELS.includes(this.model);
}
#log(text, ...args) {
console.log(`\x1b[32m[${this.className}]\x1b[0m ${text}`, ...args);
}
// This checks if the .cached_at file has a timestamp that is more than 1Week (in millis)
// from the current date. If it is, then we will refetch the API so that all the models are up
// to date.
static cacheIsStale() {
const MAX_STALE = 8.64e7; // 1 day in MS
if (!fs.existsSync(path.resolve(cacheFolder, ".cached_at"))) return true;
const now = Number(new Date());
const timestampMs = Number(
fs.readFileSync(path.resolve(cacheFolder, ".cached_at"))
);
return now - timestampMs > MAX_STALE;
}
#appendContext(contextTexts = []) {
if (!contextTexts || !contextTexts.length) return "";
return (
"\nContext:\n" +
contextTexts
.map((text, i) => {
return `[CONTEXT ${i}]:\n${text}\n[END CONTEXT ${i}]\n\n`;
})
.join("")
);
}
streamingEnabled() {
return "streamGetChatCompletion" in this;
}
static promptWindowLimit(modelName) {
try {
const cacheModelPath = path.resolve(cacheFolder, "models.json");
if (!fs.existsSync(cacheModelPath))
return MODEL_MAP.get("gemini", modelName) ?? 30_720;
const models = safeJsonParse(fs.readFileSync(cacheModelPath));
const model = models.find((model) => model.id === modelName);
if (!model)
throw new Error(
"Model not found in cache - falling back to default model."
);
return model.contextWindow;
} catch (e) {
console.error(`GeminiLLM:promptWindowLimit`, e.message);
return MODEL_MAP.get("gemini", modelName) ?? 30_720;
}
}
promptWindowLimit() {
try {
if (!fs.existsSync(this.cacheModelPath))
return MODEL_MAP.get("gemini", this.model) ?? 30_720;
const models = safeJsonParse(fs.readFileSync(this.cacheModelPath));
const model = models.find((model) => model.id === this.model);
if (!model)
throw new Error(
"Model not found in cache - falling back to default model."
);
return model.contextWindow;
} catch (e) {
console.error(`GeminiLLM:promptWindowLimit`, e.message);
return MODEL_MAP.get("gemini", this.model) ?? 30_720;
}
}
/**
* Checks if a model is experimental by reading from the cache if available, otherwise it will perform
* a blind check against the v1BetaModels list - which is manually maintained and updated.
* @param {string} modelName - The name of the model to check
* @returns {boolean} A boolean indicating if the model is experimental
*/
isExperimentalModel(modelName) {
if (
fs.existsSync(cacheFolder) &&
fs.existsSync(path.resolve(cacheFolder, "models.json"))
) {
const models = safeJsonParse(
fs.readFileSync(path.resolve(cacheFolder, "models.json"))
);
const model = models.find((model) => model.id === modelName);
if (!model) return false;
return model.experimental;
}
return modelName.includes("exp") || v1BetaModels.includes(modelName);
}
/**
* Fetches Gemini models from the Google Generative AI API
* @param {string} apiKey - The API key to use for the request
* @param {number} limit - The maximum number of models to fetch
* @param {string} pageToken - The page token to use for pagination
* @returns {Promise<[{id: string, name: string, contextWindow: number, experimental: boolean}]>} A promise that resolves to an array of Gemini models
*/
static async fetchModels(apiKey, limit = 1_000, pageToken = null) {
if (!apiKey) return [];
if (fs.existsSync(cacheFolder) && !this.cacheIsStale()) {
console.log(
`\x1b[32m[GeminiLLM]\x1b[0m Using cached models API response.`
);
return safeJsonParse(
fs.readFileSync(path.resolve(cacheFolder, "models.json"))
);
}
const stableModels = [];
const allModels = [];
// Fetch from v1
try {
const url = new URL(
"https://generativelanguage.googleapis.com/v1/models"
);
url.searchParams.set("pageSize", limit);
url.searchParams.set("key", apiKey);
if (pageToken) url.searchParams.set("pageToken", pageToken);
await fetch(url.toString(), {
method: "GET",
headers: { "Content-Type": "application/json" },
})
.then((res) => res.json())
.then((data) => {
if (data.error) throw new Error(data.error.message);
return data.models ?? [];
})
.then((models) => {
return models
.filter(
(model) => !model.displayName?.toLowerCase()?.includes("tuning")
) // remove tuning models
.filter(
(model) =>
!model.description?.toLowerCase()?.includes("deprecated")
) // remove deprecated models (in comment)
.filter((model) =>
// Only generateContent is supported
model.supportedGenerationMethods.includes("generateContent")
)
.map((model) => {
stableModels.push(model.name);
allModels.push({
id: model.name.split("/").pop(),
name: model.displayName,
contextWindow: model.inputTokenLimit,
experimental: false,
});
});
})
.catch((e) => {
console.error(`Gemini:getGeminiModelsV1`, e.message);
return;
});
} catch (e) {
console.error(`Gemini:getGeminiModelsV1`, e.message);
}
// Fetch from v1beta
try {
const url = new URL(
"https://generativelanguage.googleapis.com/v1beta/models"
);
url.searchParams.set("pageSize", limit);
url.searchParams.set("key", apiKey);
if (pageToken) url.searchParams.set("pageToken", pageToken);
await fetch(url.toString(), {
method: "GET",
headers: { "Content-Type": "application/json" },
})
.then((res) => res.json())
.then((data) => {
if (data.error) throw new Error(data.error.message);
return data.models ?? [];
})
.then((models) => {
return models
.filter((model) => !stableModels.includes(model.name)) // remove stable models that are already in the v1 list
.filter(
(model) => !model.displayName?.toLowerCase()?.includes("tuning")
) // remove tuning models
.filter(
(model) =>
!model.description?.toLowerCase()?.includes("deprecated")
) // remove deprecated models (in comment)
.filter((model) =>
// Only generateContent is supported
model.supportedGenerationMethods.includes("generateContent")
)
.map((model) => {
allModels.push({
id: model.name.split("/").pop(),
name: model.displayName,
contextWindow: model.inputTokenLimit,
experimental: true,
});
});
})
.catch((e) => {
console.error(`Gemini:getGeminiModelsV1beta`, e.message);
return;
});
} catch (e) {
console.error(`Gemini:getGeminiModelsV1beta`, e.message);
}
if (allModels.length === 0) {
console.error(`Gemini:getGeminiModels - No models found`);
return defaultGeminiModels();
}
console.log(
`\x1b[32m[GeminiLLM]\x1b[0m Writing cached models API response to disk.`
);
if (!fs.existsSync(cacheFolder))
fs.mkdirSync(cacheFolder, { recursive: true });
fs.writeFileSync(
path.resolve(cacheFolder, "models.json"),
JSON.stringify(allModels)
);
fs.writeFileSync(
path.resolve(cacheFolder, ".cached_at"),
new Date().getTime().toString()
);
return allModels;
}
/**
* Checks if a model is valid for chat completion (unused)
* @deprecated
* @param {string} modelName - The name of the model to check
* @returns {Promise<boolean>} A promise that resolves to a boolean indicating if the model is valid
*/
async isValidChatCompletionModel(modelName = "") {
const models = await this.fetchModels(process.env.GEMINI_API_KEY);
return models.some((model) => model.id === modelName);
}
/**
* Generates appropriate content array for a message + attachments.
* @param {{userPrompt:string, attachments: import("../../helpers").Attachment[]}}
* @returns {string|object[]}
*/
#generateContent({ userPrompt, attachments = [] }) {
if (!attachments.length) return userPrompt;
const content = [{ type: "text", text: userPrompt }];
for (let attachment of attachments) {
content.push({
type: "image_url",
image_url: {
url: attachment.contentString,
detail: "high",
},
});
}
return content.flat();
}
/**
* Construct the user prompt for this model.
* @param {{attachments: import("../../helpers").Attachment[]}} param0
* @returns
*/
constructPrompt({
systemPrompt = "",
contextTexts = [],
chatHistory = [],
userPrompt = "",
attachments = [], // This is the specific attachment for only this prompt
}) {
let prompt = [];
if (this.supportsSystemPrompt) {
prompt.push({
role: "system",
content: `${systemPrompt}${this.#appendContext(contextTexts)}`,
});
} else {
this.#log(
`${this.model} - does not support system prompts - emulating...`
);
prompt.push(
{
role: "user",
content: `${systemPrompt}${this.#appendContext(contextTexts)}`,
},
{
role: "assistant",
content: "Okay.",
}
);
}
return [
...prompt,
...formatChatHistory(chatHistory, this.#generateContent),
{
role: "user",
content: this.#generateContent({ userPrompt, attachments }),
},
];
}
async getChatCompletion(messages = null, { temperature = 0.7 }) {
const result = await LLMPerformanceMonitor.measureAsyncFunction(
this.openai.chat.completions
.create({
model: this.model,
messages,
temperature: temperature,
})
.catch((e) => {
console.error(e);
throw new Error(e.message);
})
);
if (
!result.output.hasOwnProperty("choices") ||
result.output.choices.length === 0
)
return null;
return {
textResponse: result.output.choices[0].message.content,
metrics: {
prompt_tokens: result.output.usage.prompt_tokens || 0,
completion_tokens: result.output.usage.completion_tokens || 0,
total_tokens: result.output.usage.total_tokens || 0,
outputTps: result.output.usage.completion_tokens / result.duration,
duration: result.duration,
model: this.model,
provider: this.className,
timestamp: new Date(),
},
};
}
async streamGetChatCompletion(messages = null, { temperature = 0.7 }) {
const measuredStreamRequest = await LLMPerformanceMonitor.measureStream({
func: this.openai.chat.completions.create({
model: this.model,
stream: true,
messages,
temperature: temperature,
stream_options: {
include_usage: true,
},
}),
messages,
runPromptTokenCalculation: false,
modelTag: this.model,
provider: this.className,
});
return measuredStreamRequest;
}
handleStream(response, stream, responseProps) {
return handleDefaultStreamResponseV2(response, stream, responseProps);
}
async compressMessages(promptArgs = {}, rawHistory = []) {
const { messageArrayCompressor } = require("../../helpers/chat");
const messageArray = this.constructPrompt(promptArgs);
return await messageArrayCompressor(this, messageArray, rawHistory);
}
// Simple wrapper for dynamic embedder & normalize interface for all LLM implementations
async embedTextInput(textInput) {
return await this.embedder.embedTextInput(textInput);
}
async embedChunks(textChunks = []) {
return await this.embedder.embedChunks(textChunks);
}
}
module.exports = {
GeminiLLM,
NO_SYSTEM_PROMPT_MODELS,
};