v4.6.9-alpha (#918)

Co-authored-by: Mufei <327958099@qq.com> Co-authored-by: heheer <71265218+newfish-cmyk@users.noreply.github.com>
2024-03-04 00:05:25 +08:00
parent f9f0b4bffd
commit 42a8184ea0
153 changed files with 4906 additions and 4307 deletions
--- a/projects/app/src/service/moduleDispatch/agent/classifyQuestion.ts
+++ b/projects/app/src/service/moduleDispatch/agent/classifyQuestion.ts
@@ -1,6 +1,7 @@
 import { adaptChat2GptMessages } from '@fastgpt/global/core/chat/adapt';
-import { ChatContextFilter, countMessagesChars } from '@fastgpt/service/core/chat/utils';
-import type { moduleDispatchResType, ChatItemType } from '@fastgpt/global/core/chat/type.d';
+import { ChatContextFilter } from '@fastgpt/service/core/chat/utils';
+import { countMessagesTokens } from '@fastgpt/global/common/string/tiktoken';
+import type { ChatItemType } from '@fastgpt/global/core/chat/type.d';
 import { ChatRoleEnum } from '@fastgpt/global/core/chat/constants';
 import { getAIApi } from '@fastgpt/service/core/ai/config';
 import type {
@@ -14,7 +15,7 @@ import { Prompt_CQJson } from '@/global/core/prompt/agent';
 import { LLMModelItemType } from '@fastgpt/global/core/ai/model.d';
 import { ModelTypeEnum, getLLMModel } from '@fastgpt/service/core/ai/model';
 import { getHistories } from '../utils';
-import { formatModelChars2Points } from '@/service/support/wallet/usage/utils';
+import { formatModelChars2Points } from '@fastgpt/service/support/wallet/usage/utils';

 type Props = ModuleDispatchProps<{
  [ModuleInputKeyEnum.aiModel]: string;
@@ -46,7 +47,7 @@ export const dispatchClassifyQuestion = async (props: Props): Promise<CQResponse

  const chatHistories = getHistories(history, histories);

-  const { arg, charsLength } = await (async () => {
+  const { arg, tokens } = await (async () => {
    if (cqModel.toolChoice) {
      return toolChoice({
        ...props,
@@ -65,7 +66,7 @@ export const dispatchClassifyQuestion = async (props: Props): Promise<CQResponse

  const { totalPoints, modelName } = formatModelChars2Points({
    model: cqModel.model,
-    charsLength,
+    tokens,
    modelType: ModelTypeEnum.llm
  });

@@ -75,7 +76,7 @@ export const dispatchClassifyQuestion = async (props: Props): Promise<CQResponse
      totalPoints: user.openaiAccount?.key ? 0 : totalPoints,
      model: modelName,
      query: userChatInput,
-      charsLength,
+      tokens,
      cqList: agents,
      cqResult: result.value,
      contextTotalLen: chatHistories.length + 2
@@ -85,7 +86,7 @@ export const dispatchClassifyQuestion = async (props: Props): Promise<CQResponse
        moduleName: name,
        totalPoints: user.openaiAccount?.key ? 0 : totalPoints,
        model: modelName,
-        charsLength
+        tokens
      }
    ]
  };
@@ -136,6 +137,13 @@ ${systemPrompt}
      required: ['type']
    }
  };
+  const tools: any = [
+    {
+      type: 'function',
+      function: agentFunction
+    }
+  ];
+
  const ai = getAIApi({
    userKey: user.openaiAccount,
    timeout: 480000
@@ -144,13 +152,8 @@ ${systemPrompt}
  const response = await ai.chat.completions.create({
    model: cqModel.model,
    temperature: 0,
-    messages: [...adaptMessages],
-    tools: [
-      {
-        type: 'function',
-        function: agentFunction
-      }
-    ],
+    messages: adaptMessages,
+    tools,
    tool_choice: { type: 'function', function: { name: agentFunName } }
  });

@@ -158,13 +161,10 @@ ${systemPrompt}
    const arg = JSON.parse(
      response?.choices?.[0]?.message?.tool_calls?.[0]?.function?.arguments || ''
    );
-    const functionChars =
-      agentFunction.description.length +
-      agentFunction.parameters.properties.type.description.length;

    return {
      arg,
-      charsLength: countMessagesChars(messages) + functionChars
+      tokens: countMessagesTokens(messages, tools)
    };
  } catch (error) {
    console.log(agentFunction.parameters);
@@ -174,7 +174,7 @@ ${systemPrompt}

    return {
      arg: {},
-      charsLength: 0
+      tokens: 0
    };
  }
 }
@@ -216,7 +216,7 @@ async function completions({
    agents.find((item) => answer.includes(item.key) || answer.includes(item.value))?.key || '';

  return {
-    charsLength: countMessagesChars(messages),
+    tokens: countMessagesTokens(messages),
    arg: { type: id }
  };
 }
--- a/projects/app/src/service/moduleDispatch/agent/extract.ts
+++ b/projects/app/src/service/moduleDispatch/agent/extract.ts
@@ -1,6 +1,7 @@
 import { adaptChat2GptMessages } from '@fastgpt/global/core/chat/adapt';
-import { ChatContextFilter, countMessagesChars } from '@fastgpt/service/core/chat/utils';
-import type { moduleDispatchResType, ChatItemType } from '@fastgpt/global/core/chat/type.d';
+import { ChatContextFilter } from '@fastgpt/service/core/chat/utils';
+import type { ChatItemType } from '@fastgpt/global/core/chat/type.d';
+import { countMessagesTokens } from '@fastgpt/global/common/string/tiktoken';
 import { ChatRoleEnum } from '@fastgpt/global/core/chat/constants';
 import { getAIApi } from '@fastgpt/service/core/ai/config';
 import type {
@@ -14,7 +15,7 @@ import { replaceVariable } from '@fastgpt/global/common/string/tools';
 import { LLMModelItemType } from '@fastgpt/global/core/ai/model.d';
 import { getHistories } from '../utils';
 import { ModelTypeEnum, getLLMModel } from '@fastgpt/service/core/ai/model';
-import { formatModelChars2Points } from '@/service/support/wallet/usage/utils';
+import { formatModelChars2Points } from '@fastgpt/service/support/wallet/usage/utils';

 type Props = ModuleDispatchProps<{
  [ModuleInputKeyEnum.history]?: ChatItemType[];
@@ -46,7 +47,7 @@ export async function dispatchContentExtract(props: Props): Promise<Response> {
  const extractModel = getLLMModel(model);
  const chatHistories = getHistories(history, histories);

-  const { arg, charsLength } = await (async () => {
+  const { arg, tokens } = await (async () => {
    if (extractModel.toolChoice) {
      return toolChoice({
        ...props,
@@ -85,7 +86,7 @@ export async function dispatchContentExtract(props: Props): Promise<Response> {

  const { totalPoints, modelName } = formatModelChars2Points({
    model: extractModel.model,
-    charsLength,
+    tokens,
    modelType: ModelTypeEnum.llm
  });

@@ -98,7 +99,7 @@ export async function dispatchContentExtract(props: Props): Promise<Response> {
      totalPoints: user.openaiAccount?.key ? 0 : totalPoints,
      model: modelName,
      query: content,
-      charsLength,
+      tokens,
      extractDescription: description,
      extractResult: arg,
      contextTotalLen: chatHistories.length + 2
@@ -108,7 +109,7 @@ export async function dispatchContentExtract(props: Props): Promise<Response> {
        moduleName: name,
        totalPoints: user.openaiAccount?.key ? 0 : totalPoints,
        model: modelName,
-        charsLength
+        tokens
      }
    ]
  };
@@ -170,6 +171,12 @@ ${description || '根据用户要求获取适当的 JSON 字符串。'}
      required: extractKeys.filter((item) => item.required).map((item) => item.key)
    }
  };
+  const tools: any = [
+    {
+      type: 'function',
+      function: agentFunction
+    }
+  ];

  const ai = getAIApi({
    userKey: user.openaiAccount,
@@ -180,12 +187,7 @@ ${description || '根据用户要求获取适当的 JSON 字符串。'}
    model: extractModel.model,
    temperature: 0,
    messages: [...adaptMessages],
-    tools: [
-      {
-        type: 'function',
-        function: agentFunction
-      }
-    ],
+    tools,
    tool_choice: { type: 'function', function: { name: agentFunName } }
  });

@@ -202,12 +204,9 @@ ${description || '根据用户要求获取适当的 JSON 字符串。'}
    }
  })();

-  const functionChars =
-    description.length + extractKeys.reduce((sum, item) => sum + item.desc.length, 0);
-
  return {
    rawResponse: response?.choices?.[0]?.message?.tool_calls?.[0]?.function?.arguments || '',
-    charsLength: countMessagesChars(messages) + functionChars,
+    tokens: countMessagesTokens(messages, tools),
    arg
  };
 }
@@ -257,7 +256,7 @@ Human: ${content}`
  if (start === -1 || end === -1)
    return {
      rawResponse: answer,
-      charsLength: countMessagesChars(messages),
+      tokens: countMessagesTokens(messages),
      arg: {}
    };

@@ -269,14 +268,14 @@ Human: ${content}`
  try {
    return {
      rawResponse: answer,
-      charsLength: countMessagesChars(messages),
+      tokens: countMessagesTokens(messages),

      arg: JSON.parse(jsonStr) as Record<string, any>
    };
  } catch (error) {
    return {
      rawResponse: answer,
-      charsLength: countMessagesChars(messages),
+      tokens: countMessagesTokens(messages),
      arg: {}
    };
  }
--- a/projects/app/src/service/moduleDispatch/chat/oneapi.ts
+++ b/projects/app/src/service/moduleDispatch/chat/oneapi.ts
@@ -1,17 +1,17 @@
 import type { NextApiResponse } from 'next';
-import { ChatContextFilter, countMessagesChars } from '@fastgpt/service/core/chat/utils';
-import type { moduleDispatchResType, ChatItemType } from '@fastgpt/global/core/chat/type.d';
+import { ChatContextFilter } from '@fastgpt/service/core/chat/utils';
+import type { ChatItemType } from '@fastgpt/global/core/chat/type.d';
 import { ChatRoleEnum } from '@fastgpt/global/core/chat/constants';
 import { sseResponseEventEnum } from '@fastgpt/service/common/response/constant';
 import { textAdaptGptResponse } from '@/utils/adapt';
 import { getAIApi } from '@fastgpt/service/core/ai/config';
 import type { ChatCompletion, StreamChatType } from '@fastgpt/global/core/ai/type.d';
-import { formatModelChars2Points } from '@/service/support/wallet/usage/utils';
+import { formatModelChars2Points } from '@fastgpt/service/support/wallet/usage/utils';
 import type { LLMModelItemType } from '@fastgpt/global/core/ai/model.d';
 import { postTextCensor } from '@/service/common/censor';
 import { ChatCompletionRequestMessageRoleEnum } from '@fastgpt/global/core/ai/constant';
 import type { ModuleDispatchResponse, ModuleItemType } from '@fastgpt/global/core/module/type.d';
-import { countMessagesTokens, sliceMessagesTB } from '@fastgpt/global/common/string/tiktoken';
+import { countMessagesTokens } from '@fastgpt/global/common/string/tiktoken';
 import { adaptChat2GptMessages } from '@fastgpt/global/core/chat/adapt';
 import { Prompt_QuotePromptList, Prompt_QuoteTemplateList } from '@/global/core/prompt/AIChat';
 import type { AIChatModuleProps } from '@fastgpt/global/core/module/node/type.d';
@@ -98,7 +98,7 @@ export const dispatchChatCompletion = async (props: ChatProps): Promise<ChatResp
    userChatInput,
    systemPrompt
  });
-  const { max_tokens } = getMaxTokens({
+  const { max_tokens } = await getMaxTokens({
    model: modelConstantsData,
    maxToken,
    filterMessages
@@ -137,8 +137,6 @@ export const dispatchChatCompletion = async (props: ChatProps): Promise<ChatResp

  const response = await ai.chat.completions.create(
    {
-      presence_penalty: 0,
-      frequency_penalty: 0,
      ...modelConstantsData?.defaultConfig,
      model: modelConstantsData.model,
      temperature,
@@ -189,10 +187,10 @@ export const dispatchChatCompletion = async (props: ChatProps): Promise<ChatResp
    }
  })();

-  const charsLength = countMessagesChars(completeMessages);
+  const tokens = countMessagesTokens(completeMessages);
  const { totalPoints, modelName } = formatModelChars2Points({
    model,
-    charsLength,
+    tokens,
    modelType: ModelTypeEnum.llm
  });

@@ -201,7 +199,7 @@ export const dispatchChatCompletion = async (props: ChatProps): Promise<ChatResp
    [ModuleOutputKeyEnum.responseData]: {
      totalPoints: user.openaiAccount?.key ? 0 : totalPoints,
      model: modelName,
-      charsLength,
+      tokens,
      query: `${userChatInput}`,
      maxToken: max_tokens,
      quoteList: filterQuoteQA,
@@ -213,7 +211,7 @@ export const dispatchChatCompletion = async (props: ChatProps): Promise<ChatResp
        moduleName: name,
        totalPoints: user.openaiAccount?.key ? 0 : totalPoints,
        model: modelName,
-        charsLength
+        tokens
      }
    ],
    history: completeMessages
@@ -292,7 +290,7 @@ function getChatMessages({

  const filterMessages = ChatContextFilter({
    messages,
-    maxTokens: Math.ceil(model.maxContext - 300) // filter token. not response maxToken
+    maxTokens: model.maxContext - 300 // filter token. not response maxToken
  });

  const adaptMessages = adaptChat2GptMessages({ messages: filterMessages, reserveId: false });
@@ -315,11 +313,12 @@ function getMaxTokens({
  const tokensLimit = model.maxContext;

  /* count response max token */
-  const promptsToken = countMessagesTokens({
-    messages: filterMessages
-  });
+  const promptsToken = countMessagesTokens(filterMessages);
  maxToken = promptsToken + maxToken > tokensLimit ? tokensLimit - promptsToken : maxToken;

+  if (maxToken <= 0) {
+    return Promise.reject('Over max token');
+  }
  return {
    max_tokens: maxToken
  };
--- a/projects/app/src/service/moduleDispatch/dataset/search.ts
+++ b/projects/app/src/service/moduleDispatch/dataset/search.ts
@@ -1,5 +1,5 @@
 import type { moduleDispatchResType } from '@fastgpt/global/core/chat/type.d';
-import { formatModelChars2Points } from '@/service/support/wallet/usage/utils';
+import { formatModelChars2Points } from '@fastgpt/service/support/wallet/usage/utils';
 import type { SelectedDatasetType } from '@fastgpt/global/core/module/api.d';
 import type { SearchDataResponseItemType } from '@fastgpt/global/core/dataset/type';
 import type {
@@ -12,7 +12,7 @@ import { ModuleInputKeyEnum, ModuleOutputKeyEnum } from '@fastgpt/global/core/mo
 import { DatasetSearchModeEnum } from '@fastgpt/global/core/dataset/constants';
 import { getHistories } from '../utils';
 import { datasetSearchQueryExtension } from '@fastgpt/service/core/dataset/search/utils';
-import { ChatModuleBillType } from '@fastgpt/global/support/wallet/bill/type';
+import { ChatModuleUsageType } from '@fastgpt/global/support/wallet/bill/type';
 import { checkTeamReRankPermission } from '@fastgpt/service/support/permission/teamLimit';

 type DatasetSearchProps = ModuleDispatchProps<{
@@ -85,7 +85,7 @@ export async function dispatchDatasetSearch(
  // start search
  const {
    searchRes,
-    charsLength,
+    tokens,
    usingSimilarityFilter,
    usingReRank: searchUsingReRank
  } = await searchDatasetData({
@@ -104,37 +104,37 @@ export async function dispatchDatasetSearch(
  // vector
  const { totalPoints, modelName } = formatModelChars2Points({
    model: vectorModel.model,
-    charsLength,
+    tokens,
    modelType: ModelTypeEnum.vector
  });
  const responseData: moduleDispatchResType & { totalPoints: number } = {
    totalPoints,
    query: concatQueries.join('\n'),
    model: modelName,
-    charsLength,
+    tokens,
    similarity: usingSimilarityFilter ? similarity : undefined,
    limit,
    searchMode,
    searchUsingReRank: searchUsingReRank
  };
-  const moduleDispatchBills: ChatModuleBillType[] = [
+  const moduleDispatchBills: ChatModuleUsageType[] = [
    {
      totalPoints,
      moduleName: module.name,
      model: modelName,
-      charsLength
+      tokens
    }
  ];

  if (aiExtensionResult) {
    const { totalPoints, modelName } = formatModelChars2Points({
      model: aiExtensionResult.model,
-      charsLength: aiExtensionResult.charsLength,
+      tokens: aiExtensionResult.tokens,
      modelType: ModelTypeEnum.llm
    });

    responseData.totalPoints += totalPoints;
-    responseData.charsLength = aiExtensionResult.charsLength;
+    responseData.tokens = aiExtensionResult.tokens;
    responseData.extensionModel = modelName;
    responseData.extensionResult =
      aiExtensionResult.extensionQueries?.join('\n') ||
@@ -144,7 +144,7 @@ export async function dispatchDatasetSearch(
      totalPoints,
      moduleName: 'core.module.template.Query extension',
      model: modelName,
-      charsLength: aiExtensionResult.charsLength
+      tokens: aiExtensionResult.tokens
    });
  }

--- a/projects/app/src/service/moduleDispatch/index.ts
+++ b/projects/app/src/service/moduleDispatch/index.ts
@@ -28,7 +28,7 @@ import { dispatchRunPlugin } from './plugin/run';
 import { dispatchPluginInput } from './plugin/runInput';
 import { dispatchPluginOutput } from './plugin/runOutput';
 import { valueTypeFormat } from './utils';
-import { ChatModuleBillType } from '@fastgpt/global/support/wallet/bill/type';
+import { ChatModuleUsageType } from '@fastgpt/global/support/wallet/bill/type';

 const callbackMap: Record<`${FlowNodeTypeEnum}`, Function> = {
  [FlowNodeTypeEnum.historyNode]: dispatchHistory,
@@ -83,7 +83,7 @@ export async function dispatchModules({
  // let storeData: Record<string, any> = {}; // after module used
  let chatResponse: ChatHistoryItemResType[] = []; // response request and save to database
  let chatAnswerText = ''; // AI answer
-  let chatModuleBills: ChatModuleBillType[] = [];
+  let chatModuleBills: ChatModuleUsageType[] = [];
  let runningTime = Date.now();

  function pushStore(
@@ -95,7 +95,7 @@ export async function dispatchModules({
    }: {
      answerText?: string;
      responseData?: ChatHistoryItemResType | ChatHistoryItemResType[];
-      moduleDispatchBills?: ChatModuleBillType[];
+      moduleDispatchBills?: ChatModuleUsageType[];
    }
  ) {
    const time = Date.now();
@@ -165,7 +165,6 @@ export async function dispatchModules({
    const filterModules = nextRunModules.filter((module) => {
      if (set.has(module.moduleId)) return false;
      set.add(module.moduleId);
-      ``;
      return true;
    });

--- a/projects/app/src/service/moduleDispatch/plugin/run.ts
+++ b/projects/app/src/service/moduleDispatch/plugin/run.ts
@@ -95,7 +95,7 @@ export const dispatchRunPlugin = async (props: RunPluginProps): Promise<RunPlugi
        moduleName: plugin.name,
        totalPoints: moduleDispatchBills.reduce((sum, item) => sum + (item.totalPoints || 0), 0),
        model: plugin.name,
-        charsLength: 0
+        tokens: 0
      }
    ],
    ...(output ? output.pluginOutput : {})
--- a/projects/app/src/service/moduleDispatch/tools/http468.ts
+++ b/projects/app/src/service/moduleDispatch/tools/http468.ts
@@ -10,6 +10,7 @@ import {
 import axios from 'axios';
 import { valueTypeFormat } from '../utils';
 import { SERVICE_LOCAL_HOST } from '@fastgpt/service/common/system/tools';
+import { addLog } from '@fastgpt/service/common/system/log';

 type PropsArrType = {
  key: string;
@@ -130,7 +131,7 @@ export const dispatchHttp468Request = async (props: HttpRequestProps): Promise<H
      ...results
    };
  } catch (error) {
-    const err = httpRequestErrorResponseData(error)
+    addLog.error('Http request error', error);
    return {
      [ModuleOutputKeyEnum.failed]: true,
      [ModuleOutputKeyEnum.responseData]: {
@@ -138,7 +139,7 @@ export const dispatchHttp468Request = async (props: HttpRequestProps): Promise<H
        params: Object.keys(params).length > 0 ? params : undefined,
        body: Object.keys(requestBody).length > 0 ? requestBody : undefined,
        headers: Object.keys(headers).length > 0 ? headers : undefined,
-        httpResult: { error: err }
+        httpResult: { error: formatHttpError(error) }
      }
    };
  }
@@ -280,21 +281,14 @@ function removeUndefinedSign(obj: Record<string, any>) {
  }
  return obj;
 }
-function httpRequestErrorResponseData(error: any) {
-    try {
-        return {
-            message: error?.message || undefined,
-            name: error?.name || undefined,
-            method: error?.config?.method || undefined,
-            baseURL: error?.config?.baseURL || undefined,
-            url: error?.config?.url || undefined,
-            code: error?.code || undefined,
-            status: error?.status || undefined
-        }
-    } catch (error) {
-        return {
-            message: 'Request Failed',
-            name: "AxiosError",
-        };
-    }
+function formatHttpError(error: any) {
+  return {
+    message: error?.message,
+    name: error?.name,
+    method: error?.config?.method,
+    baseURL: error?.config?.baseURL,
+    url: error?.config?.url,
+    code: error?.code,
+    status: error?.status
+  };
 }
--- a/projects/app/src/service/moduleDispatch/tools/queryExternsion.ts
+++ b/projects/app/src/service/moduleDispatch/tools/queryExternsion.ts
@@ -5,7 +5,7 @@ import type {
 } from '@fastgpt/global/core/module/type.d';
 import { ModuleInputKeyEnum, ModuleOutputKeyEnum } from '@fastgpt/global/core/module/constants';
 import { ModelTypeEnum, getLLMModel } from '@fastgpt/service/core/ai/model';
-import { formatModelChars2Points } from '@/service/support/wallet/usage/utils';
+import { formatModelChars2Points } from '@fastgpt/service/support/wallet/usage/utils';
 import { queryExtension } from '@fastgpt/service/core/ai/functions/queryExtension';
 import { getHistories } from '../utils';
 import { hashStr } from '@fastgpt/global/common/string/tools';
@@ -32,7 +32,7 @@ export const dispatchQueryExtension = async ({
  const queryExtensionModel = getLLMModel(model);
  const chatHistories = getHistories(history, histories);

-  const { extensionQueries, charsLength } = await queryExtension({
+  const { extensionQueries, tokens } = await queryExtension({
    chatBg: systemPrompt,
    query: userChatInput,
    histories: chatHistories,
@@ -43,7 +43,7 @@ export const dispatchQueryExtension = async ({

  const { totalPoints, modelName } = formatModelChars2Points({
    model: queryExtensionModel.model,
-    charsLength,
+    tokens,
    modelType: ModelTypeEnum.llm
  });

@@ -60,7 +60,7 @@ export const dispatchQueryExtension = async ({
    [ModuleOutputKeyEnum.responseData]: {
      totalPoints,
      model: modelName,
-      charsLength,
+      tokens,
      query: userChatInput,
      textOutput: JSON.stringify(filterSameQueries)
    },
@@ -69,7 +69,7 @@ export const dispatchQueryExtension = async ({
        moduleName: module.name,
        totalPoints,
        model: modelName,
-        charsLength
+        tokens
      }
    ],
    [ModuleOutputKeyEnum.text]: JSON.stringify(filterSameQueries)