feat: bump version to 0.3.18 (#1000)

api/config.py

@@ -100,7 +100,7 @@ class Config:
         self.CONSOLE_URL = get_env('CONSOLE_URL')
         self.API_URL = get_env('API_URL')
         self.APP_URL = get_env('APP_URL')
-        self.CURRENT_VERSION = "0.3.17"
+        self.CURRENT_VERSION = "0.3.18"
         self.COMMIT_SHA = get_env('COMMIT_SHA')
         self.EDITION = "SELF_HOSTED"
         self.DEPLOY_ENV = get_env('DEPLOY_ENV')

api/core/model_providers/providers/xinference_provider.py

@@ -2,7 +2,6 @@ import json
 from typing import Type
 
 import requests
-from xinference.client import RESTfulGenerateModelHandle, RESTfulChatModelHandle, RESTfulChatglmCppChatModelHandle
 
 from core.helper import encrypter
 from core.model_providers.models.embedding.xinference_embedding import XinferenceEmbedding
@@ -73,7 +72,7 @@ class XinferenceProvider(BaseModelProvider):
                 top_p=KwargRule[float](min=0, max=1, default=0.7),
                 presence_penalty=KwargRule[float](enabled=False),
                 frequency_penalty=KwargRule[float](enabled=False),
-                max_tokens=KwargRule[int](alias='max_new_tokens', min=10, max=4000, default=256),
+                max_tokens=KwargRule[int](min=10, max=4000, default=256),
             )
 
 

api/core/prompt/generate_prompts/baichuan_chat.json

@@ -1,7 +1,7 @@
 {
   "human_prefix": "用户",
   "assistant_prefix": "助手",
-  "context_prompt": "用户在与一个客观的助手对话。助手会尊重找到的材料，给出全面专业的解释，但不会过度演绎。同时回答中不会暴露引用的材料：\n\n```\n引用材料\n{{context}}\n```\n\n",
+  "context_prompt": "用户在与一个客观的助手对话。助手会尊重找到的材料，给出全面专业的解释，但不会过度演绎。同时回答中不会暴露引用的材料：\n\n```\n{{context}}\n```\n\n",
   "histories_prompt": "用户和助手的历史对话内容如下：\n```\n{{histories}}\n```\n\n",
   "system_prompt_orders": [
     "context_prompt",

api/core/prompt/generate_prompts/baichuan_completion.json

@@ -1,5 +1,5 @@
 {
-  "context_prompt": "用户在与一个客观的助手对话。助手会尊重找到的材料，给出全面专业的解释，但不会过度演绎。同时回答中不会暴露引用的材料：\n\n```\n引用材料\n{{context}}\n```\n",
+  "context_prompt": "用户在与一个客观的助手对话。助手会尊重找到的材料，给出全面专业的解释，但不会过度演绎。同时回答中不会暴露引用的材料：\n\n```\n{{context}}\n```\n",
   "system_prompt_orders": [
     "context_prompt",
     "pre_prompt"

api/core/third_party/langchain/llms/xinference_llm.py

@@ -9,11 +9,11 @@ from xinference.client import RESTfulChatglmCppChatModelHandle, \
 
 class XinferenceLLM(Xinference):
     def _call(
-        self,
-        prompt: str,
-        stop: Optional[List[str]] = None,
-        run_manager: Optional[CallbackManagerForLLMRun] = None,
-        **kwargs: Any,
+            self,
+            prompt: str,
+            stop: Optional[List[str]] = None,
+            run_manager: Optional[CallbackManagerForLLMRun] = None,
+            **kwargs: Any,
     ) -> str:
         """Call the xinference model and return the output.
 
@@ -56,10 +56,10 @@ class XinferenceLLM(Xinference):
             if generate_config and generate_config.get("stream"):
                 combined_text_output = ""
                 for token in self._stream_generate(
-                    model=model,
-                    prompt=prompt,
-                    run_manager=run_manager,
-                    generate_config=generate_config,
+                        model=model,
+                        prompt=prompt,
+                        run_manager=run_manager,
+                        generate_config=generate_config,
                 ):
                     combined_text_output += token
                 return combined_text_output
@@ -73,10 +73,10 @@ class XinferenceLLM(Xinference):
             if generate_config and generate_config.get("stream"):
                 combined_text_output = ""
                 for token in self._stream_generate(
-                    model=model,
-                    prompt=prompt,
-                    run_manager=run_manager,
-                    generate_config=generate_config,
+                        model=model,
+                        prompt=prompt,
+                        run_manager=run_manager,
+                        generate_config=generate_config,
                 ):
                     combined_text_output += token
                 completion = combined_text_output
@@ -89,13 +89,13 @@ class XinferenceLLM(Xinference):
 
             return completion
 
-
     def _stream_generate(
-        self,
-        model: Union["RESTfulGenerateModelHandle", "RESTfulChatModelHandle", "RESTfulChatglmCppChatModelHandle"],
-        prompt: str,
-        run_manager: Optional[CallbackManagerForLLMRun] = None,
-        generate_config: Optional[Union["LlamaCppGenerateConfig", "PytorchGenerateConfig", "ChatglmCppGenerateConfig"]] = None,
+            self,
+            model: Union["RESTfulGenerateModelHandle", "RESTfulChatModelHandle", "RESTfulChatglmCppChatModelHandle"],
+            prompt: str,
+            run_manager: Optional[CallbackManagerForLLMRun] = None,
+            generate_config: Optional[
+                Union["LlamaCppGenerateConfig", "PytorchGenerateConfig", "ChatglmCppGenerateConfig"]] = None,
     ) -> Generator[str, None, None]:
         """
         Args:
@@ -123,6 +123,10 @@ class XinferenceLLM(Xinference):
                 if choices:
                     choice = choices[0]
                     if isinstance(choice, dict):
+                        if 'finish_reason' in choice and choice['finish_reason'] \
+                                and choice['finish_reason'] in ['stop', 'length']:
+                            break
+
                         if 'text' in choice:
                             token = choice.get("text", "")
                         elif 'delta' in choice and 'content' in choice['delta']:

api/requirements.txt

@@ -49,4 +49,5 @@ huggingface_hub~=0.16.4
 transformers~=4.31.0
 stripe~=5.5.0
 pandas==1.5.3
-xinference==0.2.1
+xinference==0.2.1
+safetensors==0.3.2

docker/docker-compose.yaml

@@ -2,7 +2,7 @@ version: '3.1'
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.3.17
+    image: langgenius/dify-api:0.3.18
     restart: always
     environment:
       # Startup mode, 'api' starts the API server.
@@ -124,7 +124,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.3.17
+    image: langgenius/dify-api:0.3.18
     restart: always
     environment:
       # Startup mode, 'worker' starts the Celery worker for processing the queue.
@@ -176,7 +176,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.3.17
+    image: langgenius/dify-web:0.3.18
     restart: always
     environment:
       EDITION: SELF_HOSTED

web/app/(commonLayout)/apps/Apps.tsx

@@ -1,15 +1,14 @@
 'use client'
 
-import { useEffect, useRef, useState } from 'react'
+import { useCallback, useEffect, useRef, useState } from 'react'
 import { useRouter, useSearchParams } from 'next/navigation'
 import useSWRInfinite from 'swr/infinite'
-import { debounce } from 'lodash-es'
 import { useTranslation } from 'react-i18next'
 import AppCard from './AppCard'
 import NewAppCard from './NewAppCard'
 import type { AppListResponse } from '@/models/app'
 import { fetchAppList } from '@/service/apps'
-import { useAppContext, useSelector } from '@/context/app-context'
+import { useAppContext } from '@/context/app-context'
 import { NEED_REFRESH_APP_LIST_KEY } from '@/config'
 import { ProviderEnum } from '@/app/components/header/account-setting/model-page/declarations'
 import Confirm from '@/app/components/base/confirm/common'
@@ -24,15 +23,18 @@ const Apps = () => {
   const { t } = useTranslation()
   const { isCurrentWorkspaceManager } = useAppContext()
   const { data, isLoading, setSize, mutate } = useSWRInfinite(getKey, fetchAppList, { revalidateFirstPage: false })
-  const loadingStateRef = useRef(false)
-  const pageContainerRef = useSelector(state => state.pageContainerRef)
-  const anchorRef = useRef<HTMLAnchorElement>(null)
+  const anchorRef = useRef<HTMLDivElement>(null)
   const searchParams = useSearchParams()
   const router = useRouter()
   const payProviderName = searchParams.get('provider_name')
   const payStatus = searchParams.get('payment_result')
   const [showPayStatusModal, setShowPayStatusModal] = useState(false)
 
+  const handleCancelShowPayStatusModal = useCallback(() => {
+    setShowPayStatusModal(false)
+    router.replace('/', { forceOptimisticNavigation: false })
+  }, [router])
+
   useEffect(() => {
     document.title = `${t('app.title')} -  Dify`
     if (localStorage.getItem(NEED_REFRESH_APP_LIST_KEY) === '1') {
@@ -41,35 +43,24 @@ const Apps = () => {
     }
     if (payProviderName === ProviderEnum.anthropic && (payStatus === 'succeeded' || payStatus === 'cancelled'))
       setShowPayStatusModal(true)
-  }, [])
+  }, [mutate, payProviderName, payStatus, t])
 
   useEffect(() => {
-    loadingStateRef.current = isLoading
-  }, [isLoading])
-
-  useEffect(() => {
-    const onScroll = debounce(() => {
-      if (!loadingStateRef.current) {
-        const { scrollTop, clientHeight } = pageContainerRef.current!
-        const anchorOffset = anchorRef.current!.offsetTop
-        if (anchorOffset - scrollTop - clientHeight < 100)
+    let observer: IntersectionObserver | undefined
+    if (anchorRef.current) {
+      observer = new IntersectionObserver((entries) => {
+        if (entries[0].isIntersecting)
           setSize(size => size + 1)
-      }
-    }, 50)
-
-    pageContainerRef.current?.addEventListener('scroll', onScroll)
-    return () => pageContainerRef.current?.removeEventListener('scroll', onScroll)
-  }, [])
-
-  const handleCancelShowPayStatusModal = () => {
-    setShowPayStatusModal(false)
-    router.replace('/', { forceOptimisticNavigation: false })
-  }
+      }, { rootMargin: '100px' })
+      observer.observe(anchorRef.current)
+    }
+    return () => observer?.disconnect()
+  }, [isLoading, setSize, anchorRef, mutate])
 
   return (
-    <nav className='grid content-start grid-cols-1 gap-4 px-12 pt-8 sm:grid-cols-2 md:grid-cols-3 lg:grid-cols-4 grow shrink-0'>
+    <><nav className='grid content-start grid-cols-1 gap-4 px-12 pt-8 sm:grid-cols-2 md:grid-cols-3 lg:grid-cols-4 grow shrink-0'>
       { isCurrentWorkspaceManager
-      && <NewAppCard ref={anchorRef} onSuccess={mutate} />}
+      && <NewAppCard onSuccess={mutate} />}
       {data?.map(({ data: apps }) => apps.map(app => (
         <AppCard key={app.id} app={app} onRefresh={mutate} />
       )))}
@@ -95,6 +86,8 @@ const Apps = () => {
         )
       }
     </nav>
+    <div ref={anchorRef} className='h-0'> </div>
+    </>
   )
 }
 

web/app/(shareLayout)/layout.tsx

@@ -1,4 +1,4 @@
-import React from "react";
+import React from 'react'
 import type { FC } from 'react'
 import GA, { GaType } from '@/app/components/base/ga'
 
@@ -6,13 +6,11 @@ const Layout: FC<{
   children: React.ReactNode
 }> = ({ children }) => {
   return (
-    <div className="overflow-x-auto">
-      <div className="w-screen h-screen min-w-[300px]">
-        <GA gaType={GaType.webapp} />
-        {children}
-      </div>
+    <div className="min-w-[300px]">
+      <GA gaType={GaType.webapp} />
+      {children}
     </div>
   )
 }
 
-export default Layout
+export default Layout

web/app/components/share/chat/index.tsx

@@ -574,7 +574,7 @@ const Main: FC<IMainProps> = ({
     return <Loading type='app' />
 
   return (
-    <div className='bg-gray-100'>
+    <div className='bg-gray-100 flex w-full h-full'>
       {!isInstalledApp && (
         <Header
           title={siteInfo.title}
@@ -588,7 +588,7 @@ const Main: FC<IMainProps> = ({
 
       <div
         className={cn(
-          'flex rounded-t-2xl bg-white overflow-hidden',
+          'flex rounded-t-2xl bg-white overflow-hidden h-full w-full',
           isInstalledApp && 'rounded-b-2xl',
         )}
         style={isInstalledApp
@@ -611,7 +611,7 @@ const Main: FC<IMainProps> = ({
         )}
         {/* main */}
         <div className={cn(
-          isInstalledApp ? s.installedApp : 'h-[calc(100vh_-_3rem)]',
+          isInstalledApp ? s.installedApp : '',
           'flex-grow flex flex-col overflow-y-auto',
         )
         }>

web/app/components/share/chat/sidebar/index.tsx

@@ -85,7 +85,7 @@ const Sidebar: FC<ISidebarProps> = ({
     <div
       className={
         cn(
-          (isInstalledApp || isUniversalChat) ? 'tablet:h-[calc(100vh_-_74px)]' : 'tablet:h-[calc(100vh_-_3rem)]',
+          (isInstalledApp || isUniversalChat) ? 'tablet:h-[calc(100vh_-_74px)]' : '',
           'shrink-0 flex flex-col bg-white pc:w-[244px] tablet:w-[192px] mobile:w-[240px]  border-r border-gray-200 mobile:h-screen',
         )
       }

web/app/components/share/chatbot/index.tsx

@@ -510,7 +510,7 @@ const Main: FC<IMainProps> = ({
 
       <div className={'flex bg-white overflow-hidden'}>
         <div className={cn(
-          isInstalledApp ? s.installedApp : 'h-[calc(100vh_-_3rem)]',
+          isInstalledApp ? s.installedApp : '',
           'flex-grow flex flex-col overflow-y-auto',
         )
         }>

web/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dify-web",
-  "version": "0.3.17",
+  "version": "0.3.18",
   "private": true,
   "scripts": {
     "dev": "next dev",