feat(build-push): Avoid updating the latest version tag.

Signed-off-by: -LAN- <laipz8200@outlook.com>

.github/workflows/build-push.yml

@@ -5,8 +5,8 @@ on:
     branches:
       - "main"
       - "deploy/dev"
-  release:
-    types: [published]
+    tags:
+      - "*"
 
 concurrency:
   group: build-push-${{ github.head_ref || github.run_id }}
@@ -125,7 +125,6 @@ jobs:
         with:
           images: ${{ env[matrix.image_name_env] }}
           tags: |
-            type=raw,value=latest,enable=${{ startsWith(github.ref, 'refs/tags/') && !contains(github.ref, '-') }}
             type=ref,event=branch
             type=sha,enable=true,priority=100,prefix=,suffix=,format=long
             type=raw,value=${{ github.ref_name }},enable=${{ startsWith(github.ref, 'refs/tags/') }}

.markdownlint.json

@@ -0,0 +1,4 @@
+{
+    "MD024": false,
+    "MD013": false
+}

CHANGELOG.md

@@ -0,0 +1,45 @@
+# Changelog
+
+All notable changes to Dify will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [0.15.8] - 2025-05-30
+
+### Added
+
+- Added gunicorn keepalive setting (#19537)
+
+### Fixed
+
+- Fixed database configuration to allow DB_EXTRAS to set search_path via options (#16a4f77)
+- Fixed frontend third-party package security issues (#19655)
+- Updated dependencies: huggingface-hub (~0.16.4 to ~0.31.0), transformers (~4.35.0 to ~4.39.0), and resend (~0.7.0 to ~2.9.0) (#19563)
+- Downgrade boto3 from 1.36 to 1.35 (#19736)
+
+## [0.15.7] - 2025-04-27
+
+### Added
+
+- Added support for GPT-4.1 in model providers (#18912)
+- Added support for Amazon Bedrock DeepSeek-R1 model (#18908)
+- Added support for Amazon Bedrock Claude Sonnet 3.7 model (#18788)
+- Refined version compatibility logic in app DSL service
+
+### Fixed
+
+- Fixed issue with creating apps from template categories (#18807, #18868)
+- Fixed DSL version check when creating apps from explore templates (#18872, #18878)
+
+## [0.15.6] - 2025-04-22
+
+### Security
+
+- Fixed clickjacking vulnerability (#18552)
+- Fixed reset password security issue (#18366)
+- Updated reset password token when email code verification succeeds (#18362)
+
+### Fixed
+
+- Fixed Vertex AI Gemini 2.0 Flash 001 schema (#18405)

api/.env.example

@@ -430,4 +430,7 @@ CREATE_TIDB_SERVICE_JOB_ENABLED=false
 # Maximum number of submitted thread count in a ThreadPool for parallel node execution
 MAX_SUBMIT_COUNT=100
 # Lockout duration in seconds
-LOGIN_LOCKOUT_DURATION=86400
+LOGIN_LOCKOUT_DURATION=86400
+
+# Prevent Clickjacking
+ALLOW_EMBED=false

api/configs/middleware/__init__.py

@@ -1,5 +1,5 @@
 from typing import Any, Literal, Optional
-from urllib.parse import quote_plus
+from urllib.parse import parse_qsl, quote_plus
 
 from pydantic import Field, NonNegativeInt, PositiveFloat, PositiveInt, computed_field
 from pydantic_settings import BaseSettings
@@ -166,14 +166,28 @@ class DatabaseConfig(BaseSettings):
         default=False,
     )
 
-    @computed_field
+    @computed_field  # type: ignore[misc]
+    @property
     def SQLALCHEMY_ENGINE_OPTIONS(self) -> dict[str, Any]:
+        # Parse DB_EXTRAS for 'options'
+        db_extras_dict = dict(parse_qsl(self.DB_EXTRAS))
+        options = db_extras_dict.get("options", "")
+        # Always include timezone
+        timezone_opt = "-c timezone=UTC"
+        if options:
+            # Merge user options and timezone
+            merged_options = f"{options} {timezone_opt}"
+        else:
+            merged_options = timezone_opt
+
+        connect_args = {"options": merged_options}
+
         return {
             "pool_size": self.SQLALCHEMY_POOL_SIZE,
             "max_overflow": self.SQLALCHEMY_MAX_OVERFLOW,
             "pool_recycle": self.SQLALCHEMY_POOL_RECYCLE,
             "pool_pre_ping": self.SQLALCHEMY_POOL_PRE_PING,
-            "connect_args": {"options": "-c timezone=UTC"},
+            "connect_args": connect_args,
         }
 
 

api/configs/packaging/__init__.py

@@ -9,7 +9,7 @@ class PackagingInfo(BaseSettings):
 
     CURRENT_VERSION: str = Field(
         description="Dify version",
-        default="0.15.3",
+        default="0.15.8",
     )
 
     COMMIT_SHA: str = Field(

api/controllers/console/auth/forgot_password.py

@@ -8,7 +8,7 @@ from constants.languages import languages
 from controllers.console import api
 from controllers.console.auth.error import EmailCodeError, InvalidEmailError, InvalidTokenError, PasswordMismatchError
 from controllers.console.error import AccountInFreezeError, AccountNotFound, EmailSendIpLimitError
-from controllers.console.wraps import setup_required
+from controllers.console.wraps import email_password_login_enabled, setup_required
 from events.tenant_event import tenant_was_created
 from extensions.ext_database import db
 from libs.helper import email, extract_remote_ip
@@ -22,6 +22,7 @@ from services.feature_service import FeatureService
 
 class ForgotPasswordSendEmailApi(Resource):
     @setup_required
+    @email_password_login_enabled
     def post(self):
         parser = reqparse.RequestParser()
         parser.add_argument("email", type=email, required=True, location="json")
@@ -53,6 +54,7 @@ class ForgotPasswordSendEmailApi(Resource):
 
 class ForgotPasswordCheckApi(Resource):
     @setup_required
+    @email_password_login_enabled
     def post(self):
         parser = reqparse.RequestParser()
         parser.add_argument("email", type=str, required=True, location="json")
@@ -72,11 +74,20 @@ class ForgotPasswordCheckApi(Resource):
         if args["code"] != token_data.get("code"):
             raise EmailCodeError()
 
-        return {"is_valid": True, "email": token_data.get("email")}
+        # Verified, revoke the first token
+        AccountService.revoke_reset_password_token(args["token"])
+
+        # Refresh token data by generating a new token
+        _, new_token = AccountService.generate_reset_password_token(
+            user_email, code=args["code"], additional_data={"phase": "reset"}
+        )
+
+        return {"is_valid": True, "email": token_data.get("email"), "token": new_token}
 
 
 class ForgotPasswordResetApi(Resource):
     @setup_required
+    @email_password_login_enabled
     def post(self):
         parser = reqparse.RequestParser()
         parser.add_argument("token", type=str, required=True, nullable=False, location="json")
@@ -95,6 +106,9 @@ class ForgotPasswordResetApi(Resource):
 
         if reset_data is None:
             raise InvalidTokenError()
+        # Must use token in reset phase
+        if reset_data.get("phase", "") != "reset":
+            raise InvalidTokenError()
 
         AccountService.revoke_reset_password_token(token)
 

api/controllers/console/auth/login.py

@@ -22,7 +22,7 @@ from controllers.console.error import (
     EmailSendIpLimitError,
     NotAllowedCreateWorkspace,
 )
-from controllers.console.wraps import setup_required
+from controllers.console.wraps import email_password_login_enabled, setup_required
 from events.tenant_event import tenant_was_created
 from libs.helper import email, extract_remote_ip
 from libs.password import valid_password
@@ -38,6 +38,7 @@ class LoginApi(Resource):
     """Resource for user login."""
 
     @setup_required
+    @email_password_login_enabled
     def post(self):
         """Authenticate user and login."""
         parser = reqparse.RequestParser()
@@ -110,6 +111,7 @@ class LogoutApi(Resource):
 
 class ResetPasswordSendEmailApi(Resource):
     @setup_required
+    @email_password_login_enabled
     def post(self):
         parser = reqparse.RequestParser()
         parser.add_argument("email", type=email, required=True, location="json")

api/controllers/console/wraps.py

@@ -154,3 +154,16 @@ def enterprise_license_required(view):
         return view(*args, **kwargs)
 
     return decorated
+
+
+def email_password_login_enabled(view):
+    @wraps(view)
+    def decorated(*args, **kwargs):
+        features = FeatureService.get_system_features()
+        if features.enable_email_password_login:
+            return view(*args, **kwargs)
+
+        # otherwise, return 403
+        abort(403)
+
+    return decorated

api/core/agent/cot_agent_runner.py

@@ -104,7 +104,6 @@ class CotAgentRunner(BaseAgentRunner, ABC):
 
             # recalc llm max tokens
             prompt_messages = self._organize_prompt_messages()
-            self.recalc_llm_max_tokens(self.model_config, prompt_messages)
             # invoke model
             chunks = model_instance.invoke_llm(
                 prompt_messages=prompt_messages,

api/core/agent/fc_agent_runner.py

@@ -84,7 +84,6 @@ class FunctionCallAgentRunner(BaseAgentRunner):
 
             # recalc llm max tokens
             prompt_messages = self._organize_prompt_messages()
-            self.recalc_llm_max_tokens(self.model_config, prompt_messages)
             # invoke model
             chunks: Union[Generator[LLMResultChunk, None, None], LLMResult] = model_instance.invoke_llm(
                 prompt_messages=prompt_messages,

api/core/app/apps/agent_chat/app_runner.py

@@ -55,20 +55,6 @@ class AgentChatAppRunner(AppRunner):
         query = application_generate_entity.query
         files = application_generate_entity.files
 
-        # Pre-calculate the number of tokens of the prompt messages,
-        # and return the rest number of tokens by model context token size limit and max token size limit.
-        # If the rest number of tokens is not enough, raise exception.
-        # Include: prompt template, inputs, query(optional), files(optional)
-        # Not Include: memory, external data, dataset context
-        self.get_pre_calculate_rest_tokens(
-            app_record=app_record,
-            model_config=application_generate_entity.model_conf,
-            prompt_template_entity=app_config.prompt_template,
-            inputs=inputs,
-            files=files,
-            query=query,
-        )
-
         memory = None
         if application_generate_entity.conversation_id:
             # get memory of conversation (read-only)

api/core/app/apps/base_app_runner.py

@@ -15,10 +15,8 @@ from core.app.features.annotation_reply.annotation_reply import AnnotationReplyF
 from core.app.features.hosting_moderation.hosting_moderation import HostingModerationFeature
 from core.external_data_tool.external_data_fetch import ExternalDataFetch
 from core.memory.token_buffer_memory import TokenBufferMemory
-from core.model_manager import ModelInstance
 from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk, LLMResultChunkDelta, LLMUsage
 from core.model_runtime.entities.message_entities import AssistantPromptMessage, PromptMessage
-from core.model_runtime.entities.model_entities import ModelPropertyKey
 from core.model_runtime.errors.invoke import InvokeBadRequestError
 from core.moderation.input_moderation import InputModeration
 from core.prompt.advanced_prompt_transform import AdvancedPromptTransform
@@ -31,106 +29,6 @@ if TYPE_CHECKING:
 
 
 class AppRunner:
-    def get_pre_calculate_rest_tokens(
-        self,
-        app_record: App,
-        model_config: ModelConfigWithCredentialsEntity,
-        prompt_template_entity: PromptTemplateEntity,
-        inputs: Mapping[str, str],
-        files: Sequence["File"],
-        query: Optional[str] = None,
-    ) -> int:
-        """
-        Get pre calculate rest tokens
-        :param app_record: app record
-        :param model_config: model config entity
-        :param prompt_template_entity: prompt template entity
-        :param inputs: inputs
-        :param files: files
-        :param query: query
-        :return:
-        """
-        # Invoke model
-        model_instance = ModelInstance(
-            provider_model_bundle=model_config.provider_model_bundle, model=model_config.model
-        )
-
-        model_context_tokens = model_config.model_schema.model_properties.get(ModelPropertyKey.CONTEXT_SIZE)
-
-        max_tokens = 0
-        for parameter_rule in model_config.model_schema.parameter_rules:
-            if parameter_rule.name == "max_tokens" or (
-                parameter_rule.use_template and parameter_rule.use_template == "max_tokens"
-            ):
-                max_tokens = (
-                    model_config.parameters.get(parameter_rule.name)
-                    or model_config.parameters.get(parameter_rule.use_template or "")
-                ) or 0
-
-        if model_context_tokens is None:
-            return -1
-
-        if max_tokens is None:
-            max_tokens = 0
-
-        # get prompt messages without memory and context
-        prompt_messages, stop = self.organize_prompt_messages(
-            app_record=app_record,
-            model_config=model_config,
-            prompt_template_entity=prompt_template_entity,
-            inputs=inputs,
-            files=files,
-            query=query,
-        )
-
-        prompt_tokens = model_instance.get_llm_num_tokens(prompt_messages)
-
-        rest_tokens: int = model_context_tokens - max_tokens - prompt_tokens
-        if rest_tokens < 0:
-            raise InvokeBadRequestError(
-                "Query or prefix prompt is too long, you can reduce the prefix prompt, "
-                "or shrink the max token, or switch to a llm with a larger token limit size."
-            )
-
-        return rest_tokens
-
-    def recalc_llm_max_tokens(
-        self, model_config: ModelConfigWithCredentialsEntity, prompt_messages: list[PromptMessage]
-    ):
-        # recalc max_tokens if sum(prompt_token +  max_tokens) over model token limit
-        model_instance = ModelInstance(
-            provider_model_bundle=model_config.provider_model_bundle, model=model_config.model
-        )
-
-        model_context_tokens = model_config.model_schema.model_properties.get(ModelPropertyKey.CONTEXT_SIZE)
-
-        max_tokens = 0
-        for parameter_rule in model_config.model_schema.parameter_rules:
-            if parameter_rule.name == "max_tokens" or (
-                parameter_rule.use_template and parameter_rule.use_template == "max_tokens"
-            ):
-                max_tokens = (
-                    model_config.parameters.get(parameter_rule.name)
-                    or model_config.parameters.get(parameter_rule.use_template or "")
-                ) or 0
-
-        if model_context_tokens is None:
-            return -1
-
-        if max_tokens is None:
-            max_tokens = 0
-
-        prompt_tokens = model_instance.get_llm_num_tokens(prompt_messages)
-
-        if prompt_tokens + max_tokens > model_context_tokens:
-            max_tokens = max(model_context_tokens - prompt_tokens, 16)
-
-            for parameter_rule in model_config.model_schema.parameter_rules:
-                if parameter_rule.name == "max_tokens" or (
-                    parameter_rule.use_template and parameter_rule.use_template == "max_tokens"
-                ):
-                    model_config.parameters[parameter_rule.name] = max_tokens
-
     def organize_prompt_messages(
         self,
         app_record: App,

api/core/app/apps/chat/app_runner.py

@@ -50,20 +50,6 @@ class ChatAppRunner(AppRunner):
         query = application_generate_entity.query
         files = application_generate_entity.files
 
-        # Pre-calculate the number of tokens of the prompt messages,
-        # and return the rest number of tokens by model context token size limit and max token size limit.
-        # If the rest number of tokens is not enough, raise exception.
-        # Include: prompt template, inputs, query(optional), files(optional)
-        # Not Include: memory, external data, dataset context
-        self.get_pre_calculate_rest_tokens(
-            app_record=app_record,
-            model_config=application_generate_entity.model_conf,
-            prompt_template_entity=app_config.prompt_template,
-            inputs=inputs,
-            files=files,
-            query=query,
-        )
-
         memory = None
         if application_generate_entity.conversation_id:
             # get memory of conversation (read-only)
@@ -194,9 +180,6 @@ class ChatAppRunner(AppRunner):
         if hosting_moderation_result:
             return
 
-        # Re-calculate the max tokens if sum(prompt_token +  max_tokens) over model token limit
-        self.recalc_llm_max_tokens(model_config=application_generate_entity.model_conf, prompt_messages=prompt_messages)
-
         # Invoke model
         model_instance = ModelInstance(
             provider_model_bundle=application_generate_entity.model_conf.provider_model_bundle,

api/core/app/apps/completion/app_runner.py

@@ -43,20 +43,6 @@ class CompletionAppRunner(AppRunner):
         query = application_generate_entity.query
         files = application_generate_entity.files
 
-        # Pre-calculate the number of tokens of the prompt messages,
-        # and return the rest number of tokens by model context token size limit and max token size limit.
-        # If the rest number of tokens is not enough, raise exception.
-        # Include: prompt template, inputs, query(optional), files(optional)
-        # Not Include: memory, external data, dataset context
-        self.get_pre_calculate_rest_tokens(
-            app_record=app_record,
-            model_config=application_generate_entity.model_conf,
-            prompt_template_entity=app_config.prompt_template,
-            inputs=inputs,
-            files=files,
-            query=query,
-        )
-
         # organize all inputs and template to prompt messages
         # Include: prompt template, inputs, query(optional), files(optional)
         prompt_messages, stop = self.organize_prompt_messages(
@@ -152,9 +138,6 @@ class CompletionAppRunner(AppRunner):
         if hosting_moderation_result:
             return
 
-        # Re-calculate the max tokens if sum(prompt_token +  max_tokens) over model token limit
-        self.recalc_llm_max_tokens(model_config=application_generate_entity.model_conf, prompt_messages=prompt_messages)
-
         # Invoke model
         model_instance = ModelInstance(
             provider_model_bundle=application_generate_entity.model_conf.provider_model_bundle,

api/core/memory/token_buffer_memory.py

@@ -26,7 +26,7 @@ class TokenBufferMemory:
         self.model_instance = model_instance
 
     def get_history_prompt_messages(
-        self, max_token_limit: int = 2000, message_limit: Optional[int] = None
+        self, max_token_limit: int = 100000, message_limit: Optional[int] = None
     ) -> Sequence[PromptMessage]:
         """
         Get history prompt messages.

api/core/model_runtime/model_providers/bedrock/llm/eu.anthropic.claude-3.7-sonnet-v1.yaml

@@ -0,0 +1,115 @@
+model: us.anthropic.claude-3-7-sonnet-20250219-v1:0
+label:
+  en_US: Claude 3.7 Sonnet(US.Cross Region Inference)
+icon: icon_s_en.svg
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 200000
+# docs: https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters-anthropic-claude-messages.html
+parameter_rules:
+  - name: enable_cache
+    label:
+      zh_Hans: 启用提示缓存
+      en_US: Enable Prompt Cache
+    type: boolean
+    required: false
+    default: true
+    help:
+      zh_Hans: 启用提示缓存可以提高性能并降低成本。Claude 3.7 Sonnet支持在system、messages和tools字段中使用缓存检查点。
+      en_US: Enable prompt caching to improve performance and reduce costs. Claude 3.7 Sonnet supports cache checkpoints in system, messages, and tools fields.
+  - name: reasoning_type
+    label:
+      zh_Hans: 推理配置
+      en_US: Reasoning Type
+    type: boolean
+    required: false
+    default: false
+    placeholder:
+      zh_Hans: 设置推理配置
+      en_US: Set reasoning configuration
+    help:
+      zh_Hans: 控制模型的推理能力。启用时，temperature将固定为1且top_p将被禁用。
+      en_US: Controls the model's reasoning capability. When enabled, temperature will be fixed to 1 and top_p will be disabled.
+  - name: reasoning_budget
+    show_on:
+      - variable: reasoning_type
+        value: true
+    label:
+      zh_Hans: 推理预算
+      en_US: Reasoning Budget
+    type: int
+    default: 1024
+    min: 0
+    max: 128000
+    help:
+      zh_Hans: 推理的预算限制（最小1024），必须小于max_tokens。仅在推理类型为enabled时可用。
+      en_US: Budget limit for reasoning (minimum 1024), must be less than max_tokens. Only available when reasoning type is enabled.
+
+  - name: max_tokens
+    use_template: max_tokens
+    required: true
+    label:
+      zh_Hans: 最大token数
+      en_US: Max Tokens
+    type: int
+    default: 8192
+    min: 1
+    max: 128000
+    help:
+      zh_Hans: 停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
+      en_US: The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
+  - name: temperature
+    use_template: temperature
+    required: false
+    label:
+      zh_Hans: 模型温度
+      en_US: Model Temperature
+    type: float
+    default: 1
+    min: 0.0
+    max: 1.0
+    help:
+      zh_Hans: 生成内容的随机性。当推理功能启用时，该值将被固定为1。
+      en_US: The amount of randomness injected into the response. When reasoning is enabled, this value will be fixed to 1.
+  - name: top_p
+    show_on:
+      - variable: reasoning_type
+        value: disabled
+    use_template: top_p
+    label:
+      zh_Hans: Top P
+      en_US: Top P
+    required: false
+    type: float
+    default: 0.999
+    min: 0.000
+    max: 1.000
+    help:
+      zh_Hans: 在核采样中的概率阈值。当推理功能启用时，该参数将被禁用。
+      en_US: The probability threshold in nucleus sampling. When reasoning is enabled, this parameter will be disabled.
+  - name: top_k
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    required: false
+    type: int
+    default: 0
+    min: 0
+    # tip docs from aws has error, max value is 500
+    max: 500
+    help:
+      zh_Hans: 对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
+      en_US: Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.003'
+  output: '0.015'
+  unit: '0.001'
+  currency: USD

api/core/model_runtime/model_providers/bedrock/llm/llm.py

@@ -58,6 +58,7 @@ class BedrockLargeLanguageModel(LargeLanguageModel):
     # TODO There is invoke issue: context limit on Cohere Model, will add them after fixed.
     CONVERSE_API_ENABLED_MODEL_INFO = [
         {"prefix": "anthropic.claude-v2", "support_system_prompts": True, "support_tool_use": False},
+        {"prefix": "us.deepseek", "support_system_prompts": True, "support_tool_use": False},
         {"prefix": "anthropic.claude-v1", "support_system_prompts": True, "support_tool_use": False},
         {"prefix": "us.anthropic.claude-3", "support_system_prompts": True, "support_tool_use": True},
         {"prefix": "eu.anthropic.claude-3", "support_system_prompts": True, "support_tool_use": True},

api/core/model_runtime/model_providers/bedrock/llm/us.deepseek-r1.yaml

@@ -0,0 +1,63 @@
+model: us.deepseek.r1-v1:0
+label:
+  en_US: DeepSeek-R1(US.Cross Region Inference)
+icon: icon_s_en.svg
+model_type: llm
+features:
+  - agent-thought
+  - vision
+  - tool-call
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: max_tokens
+    use_template: max_tokens
+    required: true
+    label:
+      zh_Hans: 最大token数
+      en_US: Max Tokens
+    type: int
+    default: 8192
+    min: 1
+    max: 128000
+    help:
+      zh_Hans: 停止前生成的最大令牌数。
+      en_US: The maximum number of tokens to generate before stopping.
+  - name: temperature
+    use_template: temperature
+    required: false
+    label:
+      zh_Hans: 模型温度
+      en_US: Model Temperature
+    type: float
+    default: 1
+    min: 0.0
+    max: 1.0
+    help:
+      zh_Hans: 生成内容的随机性。当推理功能启用时，该值将被固定为1。
+      en_US: The amount of randomness injected into the response. When reasoning is enabled, this value will be fixed to 1.
+  - name: top_p
+    show_on:
+      - variable: reasoning_type
+        value: disabled
+    use_template: top_p
+    label:
+      zh_Hans: Top P
+      en_US: Top P
+    required: false
+    type: float
+    default: 0.999
+    min: 0.000
+    max: 1.000
+    help:
+      zh_Hans: 在核采样中的概率阈值。当推理功能启用时，该参数将被禁用。
+      en_US: The probability threshold in nucleus sampling. When reasoning is enabled, this parameter will be disabled.
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.001'
+  output: '0.005'
+  unit: '0.001'
+  currency: USD

api/core/model_runtime/model_providers/google/google.py

@@ -19,8 +19,8 @@ class GoogleProvider(ModelProvider):
         try:
             model_instance = self.get_model_instance(ModelType.LLM)
 
-            # Use `gemini-pro` model for validate,
-            model_instance.validate_credentials(model="gemini-pro", credentials=credentials)
+            # Use `gemini-2.0-flash` model for validate,
+            model_instance.validate_credentials(model="gemini-2.0-flash", credentials=credentials)
         except CredentialsValidateFailedError as ex:
             raise ex
         except Exception as ex:

api/core/model_runtime/model_providers/google/llm/_position.yaml

@@ -19,5 +19,3 @@
 - gemini-exp-1206
 - gemini-exp-1121
 - gemini-exp-1114
-- gemini-pro
-- gemini-pro-vision

api/core/model_runtime/model_providers/google/llm/gemini-pro-vision.yaml

@@ -1,35 +0,0 @@
-model: gemini-pro-vision
-label:
-  en_US: Gemini Pro Vision
-model_type: llm
-features:
-  - vision
-model_properties:
-  mode: chat
-  context_size: 12288
-parameter_rules:
-  - name: temperature
-    use_template: temperature
-  - name: top_p
-    use_template: top_p
-  - name: top_k
-    label:
-      zh_Hans: 取样数量
-      en_US: Top k
-    type: int
-    help:
-      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
-      en_US: Only sample from the top K options for each subsequent token.
-    required: false
-  - name: max_tokens_to_sample
-    use_template: max_tokens
-    required: true
-    default: 4096
-    min: 1
-    max: 4096
-pricing:
-  input: '0.00'
-  output: '0.00'
-  unit: '0.000001'
-  currency: USD
-deprecated: true

api/core/model_runtime/model_providers/google/llm/gemini-pro.yaml

@@ -1,39 +0,0 @@
-model: gemini-pro
-label:
-  en_US: Gemini Pro
-model_type: llm
-features:
-  - agent-thought
-  - tool-call
-  - stream-tool-call
-model_properties:
-  mode: chat
-  context_size: 30720
-parameter_rules:
-  - name: temperature
-    use_template: temperature
-  - name: top_p
-    use_template: top_p
-  - name: top_k
-    label:
-      zh_Hans: 取样数量
-      en_US: Top k
-    type: int
-    help:
-      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
-      en_US: Only sample from the top K options for each subsequent token.
-    required: false
-  - name: max_tokens_to_sample
-    use_template: max_tokens
-    required: true
-    default: 2048
-    min: 1
-    max: 2048
-  - name: response_format
-    use_template: response_format
-pricing:
-  input: '0.00'
-  output: '0.00'
-  unit: '0.000001'
-  currency: USD
-deprecated: true

api/core/model_runtime/model_providers/openai/llm/_position.yaml

@@ -1,3 +1,4 @@
+- gpt-4.1
 - o1
 - o1-2024-12-17
 - o1-mini

api/core/model_runtime/model_providers/openai/llm/gpt-4-1.yaml

@@ -0,0 +1,60 @@
+model: gpt-4.1
+label:
+  zh_Hans: gpt-4.1
+  en_US: gpt-4.1
+model_type: llm
+features:
+  - multi-tool-call
+  - agent-thought
+  - stream-tool-call
+  - vision
+model_properties:
+  mode: chat
+  context_size: 1047576
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
+  - name: max_tokens
+    use_template: max_tokens
+    default: 512
+    min: 1
+    max: 32768
+  - name: reasoning_effort
+    label:
+      zh_Hans: 推理工作
+      en_US: Reasoning Effort
+    type: string
+    help:
+      zh_Hans: 限制推理模型的推理工作
+      en_US: Constrains effort on reasoning for reasoning models
+    required: false
+    options:
+      - low
+      - medium
+      - high
+  - name: response_format
+    label:
+      zh_Hans: 回复格式
+      en_US: Response Format
+    type: string
+    help:
+      zh_Hans: 指定模型必须输出的格式
+      en_US: specifying the format that the model must output
+    required: false
+    options:
+      - text
+      - json_object
+      - json_schema
+  - name: json_schema
+    use_template: json_schema
+pricing:
+  input: '2.00'
+  output: '8.00'
+  unit: '0.000001'
+  currency: USD

api/core/model_runtime/model_providers/openai/llm/llm.py

@@ -1049,6 +1049,9 @@ class OpenAILargeLanguageModel(_CommonOpenAI, LargeLanguageModel):
         """Calculate num tokens for gpt-3.5-turbo and gpt-4 with tiktoken package.
 
         Official documentation: https://github.com/openai/openai-cookbook/blob/main/examples/How_to_format_inputs_to_ChatGPT_models.ipynb"""
+        if not messages and not tools:
+            return 0
+
         if model.startswith("ft:"):
             model = model.split(":")[1]
 
@@ -1057,18 +1060,18 @@ class OpenAILargeLanguageModel(_CommonOpenAI, LargeLanguageModel):
             model = "gpt-4o"
 
         try:
-            encoding = tiktoken.encoding_for_model(model)
-        except KeyError:
-            logger.warning("Warning: model not found. Using cl100k_base encoding.")
-            model = "cl100k_base"
             encoding = tiktoken.get_encoding(model)
+        except (KeyError, ValueError) as e:
+            logger.warning("Warning: model not found. Using cl100k_base encoding.")
+            encoding_name = "cl100k_base"
+            encoding = tiktoken.get_encoding(encoding_name)
 
         if model.startswith("gpt-3.5-turbo-0301"):
             # every message follows <im_start>{role/name}\n{content}<im_end>\n
             tokens_per_message = 4
             # if there's a name, the role is omitted
             tokens_per_name = -1
-        elif model.startswith("gpt-3.5-turbo") or model.startswith("gpt-4") or model.startswith(("o1", "o3")):
+        elif model.startswith("gpt-3.5-turbo") or model.startswith("gpt-4") or model.startswith(("o1", "o3", "o4")):
             tokens_per_message = 3
             tokens_per_name = 1
         else:

api/core/model_runtime/model_providers/vertex_ai/llm/gemini-2.0-flash-001.yaml

@@ -5,11 +5,6 @@ model_type: llm
 features:
   - agent-thought
   - vision
-  - tool-call
-  - stream-tool-call
-  - document
-  - video
-  - audio
 model_properties:
   mode: chat
   context_size: 1048576
@@ -20,20 +15,21 @@ parameter_rules:
     use_template: top_p
   - name: top_k
     label:
-      zh_Hans: 取样数量
       en_US: Top k
     type: int
     help:
-      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
       en_US: Only sample from the top K options for each subsequent token.
     required: false
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
   - name: max_output_tokens
     use_template: max_tokens
+    required: true
     default: 8192
     min: 1
     max: 8192
-  - name: json_schema
-    use_template: json_schema
 pricing:
   input: '0.00'
   output: '0.00'

api/core/tools/provider/_position.yaml

@@ -77,5 +77,4 @@
 - onebot
 - regex
 - trello
-- vanna
 - fal

api/core/tools/provider/builtin/vanna/tools/vanna.py

@@ -1,134 +0,0 @@
-from typing import Any, Union
-
-from vanna.remote import VannaDefault  # type: ignore
-
-from core.tools.entities.tool_entities import ToolInvokeMessage
-from core.tools.errors import ToolProviderCredentialValidationError
-from core.tools.tool.builtin_tool import BuiltinTool
-
-
-class VannaTool(BuiltinTool):
-    def _invoke(
-        self, user_id: str, tool_parameters: dict[str, Any]
-    ) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
-        """
-        invoke tools
-        """
-        # Ensure runtime and credentials
-        if not self.runtime or not self.runtime.credentials:
-            raise ToolProviderCredentialValidationError("Tool runtime or credentials are missing")
-        api_key = self.runtime.credentials.get("api_key", None)
-        if not api_key:
-            raise ToolProviderCredentialValidationError("Please input api key")
-
-        model = tool_parameters.get("model", "")
-        if not model:
-            return self.create_text_message("Please input RAG model")
-
-        prompt = tool_parameters.get("prompt", "")
-        if not prompt:
-            return self.create_text_message("Please input prompt")
-
-        url = tool_parameters.get("url", "")
-        if not url:
-            return self.create_text_message("Please input URL/Host/DSN")
-
-        db_name = tool_parameters.get("db_name", "")
-        username = tool_parameters.get("username", "")
-        password = tool_parameters.get("password", "")
-        port = tool_parameters.get("port", 0)
-
-        base_url = self.runtime.credentials.get("base_url", None)
-        vn = VannaDefault(model=model, api_key=api_key, config={"endpoint": base_url})
-
-        db_type = tool_parameters.get("db_type", "")
-        if db_type in {"Postgres", "MySQL", "Hive", "ClickHouse"}:
-            if not db_name:
-                return self.create_text_message("Please input database name")
-            if not username:
-                return self.create_text_message("Please input username")
-            if port < 1:
-                return self.create_text_message("Please input port")
-
-        schema_sql = "SELECT * FROM INFORMATION_SCHEMA.COLUMNS"
-        match db_type:
-            case "SQLite":
-                schema_sql = "SELECT type, sql FROM sqlite_master WHERE sql is not null"
-                vn.connect_to_sqlite(url)
-            case "Postgres":
-                vn.connect_to_postgres(host=url, dbname=db_name, user=username, password=password, port=port)
-            case "DuckDB":
-                vn.connect_to_duckdb(url=url)
-            case "SQLServer":
-                vn.connect_to_mssql(url)
-            case "MySQL":
-                vn.connect_to_mysql(host=url, dbname=db_name, user=username, password=password, port=port)
-            case "Oracle":
-                vn.connect_to_oracle(user=username, password=password, dsn=url)
-            case "Hive":
-                vn.connect_to_hive(host=url, dbname=db_name, user=username, password=password, port=port)
-            case "ClickHouse":
-                vn.connect_to_clickhouse(host=url, dbname=db_name, user=username, password=password, port=port)
-
-        enable_training = tool_parameters.get("enable_training", False)
-        reset_training_data = tool_parameters.get("reset_training_data", False)
-        if enable_training:
-            if reset_training_data:
-                existing_training_data = vn.get_training_data()
-                if len(existing_training_data) > 0:
-                    for _, training_data in existing_training_data.iterrows():
-                        vn.remove_training_data(training_data["id"])
-
-            ddl = tool_parameters.get("ddl", "")
-            question = tool_parameters.get("question", "")
-            sql = tool_parameters.get("sql", "")
-            memos = tool_parameters.get("memos", "")
-            training_metadata = tool_parameters.get("training_metadata", False)
-
-            if training_metadata:
-                if db_type == "SQLite":
-                    df_ddl = vn.run_sql(schema_sql)
-                    for ddl in df_ddl["sql"].to_list():
-                        vn.train(ddl=ddl)
-                else:
-                    df_information_schema = vn.run_sql(schema_sql)
-                    plan = vn.get_training_plan_generic(df_information_schema)
-                    vn.train(plan=plan)
-
-            if ddl:
-                vn.train(ddl=ddl)
-
-            if sql:
-                if question:
-                    vn.train(question=question, sql=sql)
-                else:
-                    vn.train(sql=sql)
-            if memos:
-                vn.train(documentation=memos)
-
-        #########################################################################################
-        # Due to CVE-2024-5565, we have to disable the chart generation feature
-        # The Vanna library uses a prompt function to present the user with visualized results,
-        # it is possible to alter the prompt using prompt injection and run arbitrary Python code
-        # instead of the intended visualization code.
-        # Specifically - allowing external input to the library’s “ask” method
-        # with "visualize" set to True (default behavior) leads to remote code execution.
-        # Affected versions: <= 0.5.5
-        #########################################################################################
-        allow_llm_to_see_data = tool_parameters.get("allow_llm_to_see_data", False)
-        res = vn.ask(
-            prompt, print_results=False, auto_train=True, visualize=False, allow_llm_to_see_data=allow_llm_to_see_data
-        )
-
-        result = []
-
-        if res is not None:
-            result.append(self.create_text_message(res[0]))
-            if len(res) > 1 and res[1] is not None:
-                result.append(self.create_text_message(res[1].to_markdown()))
-            if len(res) > 2 and res[2] is not None:
-                result.append(
-                    self.create_blob_message(blob=res[2].to_image(format="svg"), meta={"mime_type": "image/svg+xml"})
-                )
-
-        return result

api/core/tools/provider/builtin/vanna/tools/vanna.yaml

@@ -1,213 +0,0 @@
-identity:
-  name: vanna
-  author: QCTC
-  label:
-    en_US: Vanna.AI
-    zh_Hans: Vanna.AI
-description:
-  human:
-    en_US: The fastest way to get actionable insights from your database just by asking questions.
-    zh_Hans: 一个基于大模型和RAG的Text2SQL工具。
-  llm: A tool for converting text to SQL.
-parameters:
-  - name: prompt
-    type: string
-    required: true
-    label:
-      en_US: Prompt
-      zh_Hans: 提示词
-      pt_BR: Prompt
-    human_description:
-      en_US: used for generating SQL
-      zh_Hans: 用于生成SQL
-    llm_description: key words for generating SQL
-    form: llm
-  - name: model
-    type: string
-    required: true
-    label:
-      en_US: RAG Model
-      zh_Hans: RAG模型
-    human_description:
-      en_US: RAG Model for your database DDL
-      zh_Hans: 存储数据库训练数据的RAG模型
-    llm_description: RAG Model for generating SQL
-    form: llm
-  - name: db_type
-    type: select
-    required: true
-    options:
-      - value: SQLite
-        label:
-          en_US: SQLite
-          zh_Hans: SQLite
-      - value: Postgres
-        label:
-          en_US: Postgres
-          zh_Hans: Postgres
-      - value: DuckDB
-        label:
-          en_US: DuckDB
-          zh_Hans: DuckDB
-      - value: SQLServer
-        label:
-          en_US: Microsoft SQL Server
-          zh_Hans: 微软 SQL Server
-      - value: MySQL
-        label:
-          en_US: MySQL
-          zh_Hans: MySQL
-      - value: Oracle
-        label:
-          en_US: Oracle
-          zh_Hans: Oracle
-      - value: Hive
-        label:
-          en_US: Hive
-          zh_Hans: Hive
-      - value: ClickHouse
-        label:
-          en_US: ClickHouse
-          zh_Hans: ClickHouse
-    default: SQLite
-    label:
-      en_US: DB Type
-      zh_Hans: 数据库类型
-    human_description:
-      en_US: Database type.
-      zh_Hans: 选择要链接的数据库类型。
-    form: form
-  - name: url
-    type: string
-    required: true
-    label:
-      en_US: URL/Host/DSN
-      zh_Hans: URL/Host/DSN
-    human_description:
-      en_US: Please input depending on DB type, visit https://vanna.ai/docs/ for more specification
-      zh_Hans: 请根据数据库类型，填入对应值，详情参考https://vanna.ai/docs/
-    form: form
-  - name: db_name
-    type: string
-    required: false
-    label:
-      en_US: DB name
-      zh_Hans: 数据库名
-    human_description:
-      en_US: Database name
-      zh_Hans: 数据库名
-    form: form
-  - name: username
-    type: string
-    required: false
-    label:
-      en_US: Username
-      zh_Hans: 用户名
-    human_description:
-      en_US: Username
-      zh_Hans: 用户名
-    form: form
-  - name: password
-    type: secret-input
-    required: false
-    label:
-      en_US: Password
-      zh_Hans: 密码
-    human_description:
-      en_US: Password
-      zh_Hans: 密码
-    form: form
-  - name: port
-    type: number
-    required: false
-    label:
-      en_US: Port
-      zh_Hans: 端口
-    human_description:
-      en_US: Port
-      zh_Hans: 端口
-    form: form
-  - name: ddl
-    type: string
-    required: false
-    label:
-      en_US: Training DDL
-      zh_Hans: 训练DDL
-    human_description:
-      en_US: DDL statements for training data
-      zh_Hans: 用于训练RAG Model的建表语句
-    form: llm
-  - name: question
-    type: string
-    required: false
-    label:
-      en_US: Training Question
-      zh_Hans: 训练问题
-    human_description:
-      en_US: Question-SQL Pairs
-      zh_Hans: Question-SQL中的问题
-    form: llm
-  - name: sql
-    type: string
-    required: false
-    label:
-      en_US: Training SQL
-      zh_Hans: 训练SQL
-    human_description:
-      en_US: SQL queries to your training data
-      zh_Hans: 用于训练RAG Model的SQL语句
-    form: llm
-  - name: memos
-    type: string
-    required: false
-    label:
-      en_US: Training Memos
-      zh_Hans: 训练说明
-    human_description:
-      en_US: Sometimes you may want to add documentation about your business terminology or definitions
-      zh_Hans: 添加更多关于数据库的业务说明
-    form: llm
-  - name: enable_training
-    type: boolean
-    required: false
-    default: false
-    label:
-      en_US: Training Data
-      zh_Hans: 训练数据
-    human_description:
-      en_US: You only need to train once. Do not train again unless you want to add more training data
-      zh_Hans: 训练数据无更新时，训练一次即可
-    form: form
-  - name: reset_training_data
-    type: boolean
-    required: false
-    default: false
-    label:
-      en_US: Reset Training Data
-      zh_Hans: 重置训练数据
-    human_description:
-      en_US: Remove all training data in the current RAG Model
-      zh_Hans: 删除当前RAG Model中的所有训练数据
-    form: form
-  - name: training_metadata
-    type: boolean
-    required: false
-    default: false
-    label:
-      en_US: Training Metadata
-      zh_Hans: 训练元数据
-    human_description:
-      en_US: If enabled, it will attempt to train on the metadata of that database
-      zh_Hans: 是否自动从数据库获取元数据来训练
-    form: form
-  - name: allow_llm_to_see_data
-    type: boolean
-    required: false
-    default: false
-    label:
-      en_US: Whether to allow the LLM to see the data
-      zh_Hans: 是否允许LLM查看数据
-    human_description:
-      en_US: Whether to allow the LLM to see the data
-      zh_Hans: 是否允许LLM查看数据
-    form: form

api/core/tools/provider/builtin/vanna/vanna.py

@@ -1,46 +0,0 @@
-import re
-from typing import Any
-from urllib.parse import urlparse
-
-from core.tools.errors import ToolProviderCredentialValidationError
-from core.tools.provider.builtin.vanna.tools.vanna import VannaTool
-from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
-
-
-class VannaProvider(BuiltinToolProviderController):
-    def _get_protocol_and_main_domain(self, url):
-        parsed_url = urlparse(url)
-        protocol = parsed_url.scheme
-        hostname = parsed_url.hostname
-        port = f":{parsed_url.port}" if parsed_url.port else ""
-
-        # Check if the hostname is an IP address
-        is_ip = re.match(r"^\d{1,3}(\.\d{1,3}){3}$", hostname) is not None
-
-        # Return the full hostname (with port if present) for IP addresses, otherwise return the main domain
-        main_domain = f"{hostname}{port}" if is_ip else ".".join(hostname.split(".")[-2:]) + port
-        return f"{protocol}://{main_domain}"
-
-    def _validate_credentials(self, credentials: dict[str, Any]) -> None:
-        base_url = credentials.get("base_url")
-        if not base_url:
-            base_url = "https://ask.vanna.ai/rpc"
-        else:
-            base_url = base_url.removesuffix("/")
-        credentials["base_url"] = base_url
-        try:
-            VannaTool().fork_tool_runtime(
-                runtime={
-                    "credentials": credentials,
-                }
-            ).invoke(
-                user_id="",
-                tool_parameters={
-                    "model": "chinook",
-                    "db_type": "SQLite",
-                    "url": f"{self._get_protocol_and_main_domain(credentials['base_url'])}/Chinook.sqlite",
-                    "query": "What are the top 10 customers by sales?",
-                },
-            )
-        except Exception as e:
-            raise ToolProviderCredentialValidationError(str(e))

api/core/tools/provider/builtin/vanna/vanna.yaml

@@ -1,35 +0,0 @@
-identity:
-  author: QCTC
-  name: vanna
-  label:
-    en_US: Vanna.AI
-    zh_Hans: Vanna.AI
-  description:
-    en_US: The fastest way to get actionable insights from your database just by asking questions.
-    zh_Hans: 一个基于大模型和RAG的Text2SQL工具。
-  icon: icon.png
-  tags:
-    - utilities
-    - productivity
-credentials_for_provider:
-  api_key:
-    type: secret-input
-    required: true
-    label:
-      en_US: API key
-      zh_Hans: API key
-    placeholder:
-      en_US: Please input your API key
-      zh_Hans: 请输入你的 API key
-      pt_BR: Please input your API key
-    help:
-      en_US: Get your API key from Vanna.AI
-      zh_Hans: 从 Vanna.AI 获取你的 API key
-    url: https://vanna.ai/account/profile
-  base_url:
-    type: text-input
-    required: false
-    label:
-      en_US: Vanna.AI Endpoint Base URL
-    placeholder:
-      en_US: https://ask.vanna.ai/rpc

api/core/tools/utils/web_reader_tool.py

@@ -1,21 +1,13 @@
-import hashlib
-import json
 import mimetypes
-import os
 import re
-import site
-import subprocess
-import tempfile
-import unicodedata
-from contextlib import contextmanager
-from pathlib import Path
-from typing import Any, Literal, Optional, cast
+from collections.abc import Sequence
+from dataclasses import dataclass
+from typing import Any, Optional, cast
 from urllib.parse import unquote
 
 import chardet
 import cloudscraper  # type: ignore
-from bs4 import BeautifulSoup, CData, Comment, NavigableString  # type: ignore
-from regex import regex  # type: ignore
+from readabilipy import simple_json_from_html_string  # type: ignore
 
 from core.helper import ssrf_proxy
 from core.rag.extractor import extract_processor
@@ -23,9 +15,7 @@ from core.rag.extractor.extract_processor import ExtractProcessor
 
 FULL_TEMPLATE = """
 TITLE: {title}
-AUTHORS: {authors}
-PUBLISH DATE: {publish_date}
-TOP_IMAGE_URL: {top_image}
+AUTHOR: {author}
 TEXT:
 
 {text}
@@ -73,8 +63,8 @@ def get_url(url: str, user_agent: Optional[str] = None) -> str:
         response = ssrf_proxy.get(url, headers=headers, follow_redirects=True, timeout=(120, 300))
     elif response.status_code == 403:
         scraper = cloudscraper.create_scraper()
-        scraper.perform_request = ssrf_proxy.make_request
-        response = scraper.get(url, headers=headers, follow_redirects=True, timeout=(120, 300))
+        scraper.perform_request = ssrf_proxy.make_request  # type: ignore
+        response = scraper.get(url, headers=headers, follow_redirects=True, timeout=(120, 300))  # type: ignore
 
     if response.status_code != 200:
         return "URL returned status code {}.".format(response.status_code)
@@ -90,273 +80,36 @@ def get_url(url: str, user_agent: Optional[str] = None) -> str:
     else:
         content = response.text
 
-    a = extract_using_readabilipy(content)
+    article = extract_using_readabilipy(content)
 
-    if not a["plain_text"] or not a["plain_text"].strip():
+    if not article.text:
         return ""
 
     res = FULL_TEMPLATE.format(
-        title=a["title"],
-        authors=a["byline"],
-        publish_date=a["date"],
-        top_image="",
-        text=a["plain_text"] or "",
+        title=article.title,
+        author=article.auther,
+        text=article.text,
     )
 
     return res
 
 
-def extract_using_readabilipy(html):
-    with tempfile.NamedTemporaryFile(delete=False, mode="w+") as f_html:
-        f_html.write(html)
-        f_html.close()
-    html_path = f_html.name
-
-    # Call Mozilla's Readability.js Readability.parse() function via node, writing output to a temporary file
-    article_json_path = html_path + ".json"
-    jsdir = os.path.join(find_module_path("readabilipy"), "javascript")
-    with chdir(jsdir):
-        subprocess.check_call(["node", "ExtractArticle.js", "-i", html_path, "-o", article_json_path])
-
-    # Read output of call to Readability.parse() from JSON file and return as Python dictionary
-    input_json = json.loads(Path(article_json_path).read_text(encoding="utf-8"))
-
-    # Deleting files after processing
-    os.unlink(article_json_path)
-    os.unlink(html_path)
-
-    article_json: dict[str, Any] = {
-        "title": None,
-        "byline": None,
-        "date": None,
-        "content": None,
-        "plain_content": None,
-        "plain_text": None,
-    }
-    # Populate article fields from readability fields where present
-    if input_json:
-        if input_json.get("title"):
-            article_json["title"] = input_json["title"]
-        if input_json.get("byline"):
-            article_json["byline"] = input_json["byline"]
-        if input_json.get("date"):
-            article_json["date"] = input_json["date"]
-        if input_json.get("content"):
-            article_json["content"] = input_json["content"]
-            article_json["plain_content"] = plain_content(article_json["content"], False, False)
-            article_json["plain_text"] = extract_text_blocks_as_plain_text(article_json["plain_content"])
-        if input_json.get("textContent"):
-            article_json["plain_text"] = input_json["textContent"]
-            article_json["plain_text"] = re.sub(r"\n\s*\n", "\n", article_json["plain_text"])
-
-    return article_json
+@dataclass
+class Article:
+    title: str
+    auther: str
+    text: Sequence[dict]
 
 
-def find_module_path(module_name):
-    for package_path in site.getsitepackages():
-        potential_path = os.path.join(package_path, module_name)
-        if os.path.exists(potential_path):
-            return potential_path
-
-    return None
-
-
-@contextmanager
-def chdir(path):
-    """Change directory in context and return to original on exit"""
-    # From https://stackoverflow.com/a/37996581, couldn't find a built-in
-    original_path = os.getcwd()
-    os.chdir(path)
-    try:
-        yield
-    finally:
-        os.chdir(original_path)
-
-
-def extract_text_blocks_as_plain_text(paragraph_html):
-    # Load article as DOM
-    soup = BeautifulSoup(paragraph_html, "html.parser")
-    # Select all lists
-    list_elements = soup.find_all(["ul", "ol"])
-    # Prefix text in all list items with "* " and make lists paragraphs
-    for list_element in list_elements:
-        plain_items = "".join(
-            list(filter(None, [plain_text_leaf_node(li)["text"] for li in list_element.find_all("li")]))
-        )
-        list_element.string = plain_items
-        list_element.name = "p"
-    # Select all text blocks
-    text_blocks = [s.parent for s in soup.find_all(string=True)]
-    text_blocks = [plain_text_leaf_node(block) for block in text_blocks]
-    # Drop empty paragraphs
-    text_blocks = list(filter(lambda p: p["text"] is not None, text_blocks))
-    return text_blocks
-
-
-def plain_text_leaf_node(element):
-    # Extract all text, stripped of any child HTML elements and normalize it
-    plain_text = normalize_text(element.get_text())
-    if plain_text != "" and element.name == "li":
-        plain_text = "* {}, ".format(plain_text)
-    if plain_text == "":
-        plain_text = None
-    if "data-node-index" in element.attrs:
-        plain = {"node_index": element["data-node-index"], "text": plain_text}
-    else:
-        plain = {"text": plain_text}
-    return plain
-
-
-def plain_content(readability_content, content_digests, node_indexes):
-    # Load article as DOM
-    soup = BeautifulSoup(readability_content, "html.parser")
-    # Make all elements plain
-    elements = plain_elements(soup.contents, content_digests, node_indexes)
-    if node_indexes:
-        # Add node index attributes to nodes
-        elements = [add_node_indexes(element) for element in elements]
-    # Replace article contents with plain elements
-    soup.contents = elements
-    return str(soup)
-
-
-def plain_elements(elements, content_digests, node_indexes):
-    # Get plain content versions of all elements
-    elements = [plain_element(element, content_digests, node_indexes) for element in elements]
-    if content_digests:
-        # Add content digest attribute to nodes
-        elements = [add_content_digest(element) for element in elements]
-    return elements
-
-
-def plain_element(element, content_digests, node_indexes):
-    # For lists, we make each item plain text
-    if is_leaf(element):
-        # For leaf node elements, extract the text content, discarding any HTML tags
-        # 1. Get element contents as text
-        plain_text = element.get_text()
-        # 2. Normalize the extracted text string to a canonical representation
-        plain_text = normalize_text(plain_text)
-        # 3. Update element content to be plain text
-        element.string = plain_text
-    elif is_text(element):
-        if is_non_printing(element):
-            # The simplified HTML may have come from Readability.js so might
-            # have non-printing text (e.g. Comment or CData). In this case, we
-            # keep the structure, but ensure that the string is empty.
-            element = type(element)("")
-        else:
-            plain_text = element.string
-            plain_text = normalize_text(plain_text)
-            element = type(element)(plain_text)
-    else:
-        # If not a leaf node or leaf type call recursively on child nodes, replacing
-        element.contents = plain_elements(element.contents, content_digests, node_indexes)
-    return element
-
-
-def add_node_indexes(element, node_index="0"):
-    # Can't add attributes to string types
-    if is_text(element):
-        return element
-    # Add index to current element
-    element["data-node-index"] = node_index
-    # Add index to child elements
-    for local_idx, child in enumerate([c for c in element.contents if not is_text(c)], start=1):
-        # Can't add attributes to leaf string types
-        child_index = "{stem}.{local}".format(stem=node_index, local=local_idx)
-        add_node_indexes(child, node_index=child_index)
-    return element
-
-
-def normalize_text(text):
-    """Normalize unicode and whitespace."""
-    # Normalize unicode first to try and standardize whitespace characters as much as possible before normalizing them
-    text = strip_control_characters(text)
-    text = normalize_unicode(text)
-    text = normalize_whitespace(text)
-    return text
-
-
-def strip_control_characters(text):
-    """Strip out unicode control characters which might break the parsing."""
-    # Unicode control characters
-    #   [Cc]: Other, Control [includes new lines]
-    #   [Cf]: Other, Format
-    #   [Cn]: Other, Not Assigned
-    #   [Co]: Other, Private Use
-    #   [Cs]: Other, Surrogate
-    control_chars = {"Cc", "Cf", "Cn", "Co", "Cs"}
-    retained_chars = ["\t", "\n", "\r", "\f"]
-
-    # Remove non-printing control characters
-    return "".join(
-        [
-            "" if (unicodedata.category(char) in control_chars) and (char not in retained_chars) else char
-            for char in text
-        ]
+def extract_using_readabilipy(html: str):
+    json_article: dict[str, Any] = simple_json_from_html_string(html, use_readability=True)
+    article = Article(
+        title=json_article.get("title") or "",
+        auther=json_article.get("byline") or "",
+        text=json_article.get("plain_text") or [],
     )
 
-
-def normalize_unicode(text):
-    """Normalize unicode such that things that are visually equivalent map to the same unicode string where possible."""
-    normal_form: Literal["NFC", "NFD", "NFKC", "NFKD"] = "NFKC"
-    text = unicodedata.normalize(normal_form, text)
-    return text
-
-
-def normalize_whitespace(text):
-    """Replace runs of whitespace characters with a single space as this is what happens when HTML text is displayed."""
-    text = regex.sub(r"\s+", " ", text)
-    # Remove leading and trailing whitespace
-    text = text.strip()
-    return text
-
-
-def is_leaf(element):
-    return element.name in {"p", "li"}
-
-
-def is_text(element):
-    return isinstance(element, NavigableString)
-
-
-def is_non_printing(element):
-    return any(isinstance(element, _e) for _e in [Comment, CData])
-
-
-def add_content_digest(element):
-    if not is_text(element):
-        element["data-content-digest"] = content_digest(element)
-    return element
-
-
-def content_digest(element):
-    digest: Any
-    if is_text(element):
-        # Hash
-        trimmed_string = element.string.strip()
-        if trimmed_string == "":
-            digest = ""
-        else:
-            digest = hashlib.sha256(trimmed_string.encode("utf-8")).hexdigest()
-    else:
-        contents = element.contents
-        num_contents = len(contents)
-        if num_contents == 0:
-            # No hash when no child elements exist
-            digest = ""
-        elif num_contents == 1:
-            # If single child, use digest of child
-            digest = content_digest(contents[0])
-        else:
-            # Build content digest from the "non-empty" digests of child nodes
-            digest = hashlib.sha256()
-            child_digests = list(filter(lambda x: x != "", [content_digest(content) for content in contents]))
-            for child in child_digests:
-                digest.update(child.encode("utf-8"))
-            digest = digest.hexdigest()
-    return digest
+    return article
 
 
 def get_image_upload_file_ids(content):

api/core/workflow/nodes/code/code_node.py

@@ -195,7 +195,7 @@ class CodeNode(BaseNode[CodeNodeData]):
             if output_config.type == "object":
                 # check if output is object
                 if not isinstance(result.get(output_name), dict):
-                    if isinstance(result.get(output_name), type(None)):
+                    if result.get(output_name) is None:
                         transformed_result[output_name] = None
                     else:
                         raise OutputValidationError(
@@ -223,7 +223,7 @@ class CodeNode(BaseNode[CodeNodeData]):
             elif output_config.type == "array[number]":
                 # check if array of number available
                 if not isinstance(result[output_name], list):
-                    if isinstance(result[output_name], type(None)):
+                    if result[output_name] is None:
                         transformed_result[output_name] = None
                     else:
                         raise OutputValidationError(
@@ -244,7 +244,7 @@ class CodeNode(BaseNode[CodeNodeData]):
             elif output_config.type == "array[string]":
                 # check if array of string available
                 if not isinstance(result[output_name], list):
-                    if isinstance(result[output_name], type(None)):
+                    if result[output_name] is None:
                         transformed_result[output_name] = None
                     else:
                         raise OutputValidationError(
@@ -265,7 +265,7 @@ class CodeNode(BaseNode[CodeNodeData]):
             elif output_config.type == "array[object]":
                 # check if array of object available
                 if not isinstance(result[output_name], list):
-                    if isinstance(result[output_name], type(None)):
+                    if result[output_name] is None:
                         transformed_result[output_name] = None
                     else:
                         raise OutputValidationError(

api/core/workflow/nodes/llm/node.py

@@ -968,14 +968,12 @@ def _handle_memory_chat_mode(
     *,
     memory: TokenBufferMemory | None,
     memory_config: MemoryConfig | None,
-    model_config: ModelConfigWithCredentialsEntity,
+    model_config: ModelConfigWithCredentialsEntity,  # TODO(-LAN-): Needs to remove
 ) -> Sequence[PromptMessage]:
     memory_messages: Sequence[PromptMessage] = []
     # Get messages from memory for chat model
     if memory and memory_config:
-        rest_tokens = _calculate_rest_token(prompt_messages=[], model_config=model_config)
         memory_messages = memory.get_history_prompt_messages(
-            max_token_limit=rest_tokens,
             message_limit=memory_config.window.size if memory_config.window.enabled else None,
         )
     return memory_messages

api/docker/entrypoint.sh

@@ -35,6 +35,7 @@ else
       --worker-class ${SERVER_WORKER_CLASS:-gevent} \
       --worker-connections ${SERVER_WORKER_CONNECTIONS:-10} \
       --timeout ${GUNICORN_TIMEOUT:-200} \
+      --keep-alive ${GUNICORN_KEEP_ALIVE:-2} \
       app:app
   fi
 fi

api/pyproject.toml

@@ -21,7 +21,7 @@ azure-ai-inference = "~1.0.0b8"
 azure-ai-ml = "~1.20.0"
 azure-identity = "1.16.1"
 beautifulsoup4 = "4.12.2"
-boto3 = "1.36.12"
+boto3 = "~1.35.0"
 bs4 = "~0.0.1"
 cachetools = "~5.3.0"
 celery = "~5.4.0"
@@ -48,7 +48,7 @@ google-generativeai = "0.8.1"
 googleapis-common-protos = "1.63.0"
 gunicorn = "~23.0.0"
 httpx = { version = "~0.27.0", extras = ["socks"] }
-huggingface-hub = "~0.16.4"
+huggingface-hub = "~0.31.0"
 jieba = "0.42.1"
 langfuse = "~2.51.3"
 langsmith = "~0.1.77"
@@ -78,18 +78,18 @@ pyyaml = "~6.0.1"
 readabilipy = "0.2.0"
 redis = { version = "~5.0.3", extras = ["hiredis"] }
 replicate = "~0.22.0"
-resend = "~0.7.0"
+resend = "~2.9.0"
 sagemaker = "~2.231.0"
 scikit-learn = "~1.5.1"
 sentry-sdk = { version = "~1.44.1", extras = ["flask"] }
 sqlalchemy = "~2.0.29"
 starlette = "0.41.0"
 tencentcloud-sdk-python-hunyuan = "~3.0.1294"
-tiktoken = "~0.8.0"
+tiktoken = "^0.9.0"
 tokenizers = "~0.15.0"
-transformers = "~4.35.0"
+transformers = "~4.39.0"
 unstructured = { version = "~0.16.1", extras = ["docx", "epub", "md", "msg", "ppt", "pptx"] }
-validators = "0.21.0"
+validators = "0.22.0"
 volcengine-python-sdk = {extras = ["ark"], version = "~1.0.98"}
 websocket-client = "~1.7.0"
 xinference-client = "0.15.2"
@@ -112,7 +112,7 @@ safetensors = "~0.4.3"
 # [ Tools ] dependency group
 ############################################################
 [tool.poetry.group.tools.dependencies]
-arxiv = "2.1.0"
+arxiv = "2.2.0"
 cloudscraper = "1.2.71"
 duckduckgo-search = "~6.3.0"
 jsonpath-ng = "1.6.1"
@@ -166,7 +166,7 @@ tcvectordb = "1.3.2"
 tidb-vector = "0.0.9"
 upstash-vector = "0.6.0"
 volcengine-compat = "~1.0.156"
-weaviate-client = "~3.21.0"
+weaviate-client = "~3.26.0"
 
 ############################################################
 # [ Dev ] dependency group

api/services/account_service.py

@@ -406,10 +406,8 @@ class AccountService:
 
             raise PasswordResetRateLimitExceededError()
 
-        code = "".join([str(random.randint(0, 9)) for _ in range(6)])
-        token = TokenManager.generate_token(
-            account=account, email=email, token_type="reset_password", additional_data={"code": code}
-        )
+        code, token = cls.generate_reset_password_token(account_email, account)
+
         send_reset_password_mail_task.delay(
             language=language,
             to=account_email,
@@ -418,6 +416,22 @@ class AccountService:
         cls.reset_password_rate_limiter.increment_rate_limit(account_email)
         return token
 
+    @classmethod
+    def generate_reset_password_token(
+        cls,
+        email: str,
+        account: Optional[Account] = None,
+        code: Optional[str] = None,
+        additional_data: dict[str, Any] = {},
+    ):
+        if not code:
+            code = "".join([str(random.randint(0, 9)) for _ in range(6)])
+        additional_data["code"] = code
+        token = TokenManager.generate_token(
+            account=account, email=email, token_type="reset_password", additional_data=additional_data
+        )
+        return code, token
+
     @classmethod
     def revoke_reset_password_token(cls, token: str):
         TokenManager.revoke_token(token, "reset_password")

api/services/app_dsl_service.py

@@ -55,13 +55,19 @@ def _check_version_compatibility(imported_version: str) -> ImportStatus:
     except version.InvalidVersion:
         return ImportStatus.FAILED
 
-    # Compare major version and minor version
-    if current_ver.major != imported_ver.major or current_ver.minor != imported_ver.minor:
+    # If imported version is newer than current, always return PENDING
+    if imported_ver > current_ver:
         return ImportStatus.PENDING
 
-    if current_ver.micro != imported_ver.micro:
+    # If imported version is older than current's major, return PENDING
+    if imported_ver.major < current_ver.major:
+        return ImportStatus.PENDING
+
+    # If imported version is older than current's minor, return COMPLETED_WITH_WARNINGS
+    if imported_ver.minor < current_ver.minor:
         return ImportStatus.COMPLETED_WITH_WARNINGS
 
+    # If imported version equals or is older than current's micro, return COMPLETED
     return ImportStatus.COMPLETED
 
 

docker/.env.example

@@ -142,6 +142,9 @@ CELERY_WORKER_CLASS=
 # it is recommended to set it to 360 to support a longer sse connection time.
 GUNICORN_TIMEOUT=360
 
+# The number of seconds to wait for requests on a Keep-Alive connection, default to 2
+GUNICORN_KEEP_ALIVE=2
+
 # The number of Celery workers. The default is 1, and can be set as needed.
 CELERY_WORKER_AMOUNT=
 
@@ -932,3 +935,6 @@ MAX_SUBMIT_COUNT=100
 
 # The maximum number of top-k value for RAG.
 TOP_K_MAX_VALUE=10
+
+# Prevent Clickjacking
+ALLOW_EMBED=false

docker/docker-compose-template.yaml

@@ -1,8 +1,8 @@
-x-shared-env: &shared-api-worker-env
+x-shared-env: &shared-api-worker-env 
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.15.3
+    image: langgenius/dify-api:0.15.8
     restart: always
     environment:
       # Use the shared environment variables.
@@ -25,7 +25,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.15.3
+    image: langgenius/dify-api:0.15.8
     restart: always
     environment:
       # Use the shared environment variables.
@@ -47,7 +47,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.15.3
+    image: langgenius/dify-web:0.15.8
     restart: always
     environment:
       CONSOLE_API_URL: ${CONSOLE_API_URL:-}
@@ -56,6 +56,7 @@ services:
       NEXT_TELEMETRY_DISABLED: ${NEXT_TELEMETRY_DISABLED:-0}
       TEXT_GENERATION_TIMEOUT_MS: ${TEXT_GENERATION_TIMEOUT_MS:-60000}
       CSP_WHITELIST: ${CSP_WHITELIST:-}
+      ALLOW_EMBED: ${ALLOW_EMBED:-false}
       TOP_K_MAX_VALUE: ${TOP_K_MAX_VALUE:-}
       INDEXING_MAX_SEGMENTATION_TOKENS_LENGTH: ${INDEXING_MAX_SEGMENTATION_TOKENS_LENGTH:-}
 
@@ -98,7 +99,7 @@ services:
 
   # The DifySandbox
   sandbox:
-    image: langgenius/dify-sandbox:0.2.10
+    image: langgenius/dify-sandbox:0.2.11
     restart: always
     environment:
       # The DifySandbox configurations

docker/docker-compose.middleware.yaml

@@ -43,7 +43,7 @@ services:
 
   # The DifySandbox
   sandbox:
-    image: langgenius/dify-sandbox:0.2.10
+    image: langgenius/dify-sandbox:0.2.11
     restart: always
     environment:
       # The DifySandbox configurations

docker/docker-compose.yaml

@@ -37,6 +37,7 @@ x-shared-env: &shared-api-worker-env
   SERVER_WORKER_CONNECTIONS: ${SERVER_WORKER_CONNECTIONS:-10}
   CELERY_WORKER_CLASS: ${CELERY_WORKER_CLASS:-}
   GUNICORN_TIMEOUT: ${GUNICORN_TIMEOUT:-360}
+  GUNICORN_KEEP_ALIVE: ${GUNICORN_KEEP_ALIVE:-2}
   CELERY_WORKER_AMOUNT: ${CELERY_WORKER_AMOUNT:-}
   CELERY_AUTO_SCALE: ${CELERY_AUTO_SCALE:-false}
   CELERY_MAX_WORKERS: ${CELERY_MAX_WORKERS:-}
@@ -389,11 +390,12 @@ x-shared-env: &shared-api-worker-env
   CREATE_TIDB_SERVICE_JOB_ENABLED: ${CREATE_TIDB_SERVICE_JOB_ENABLED:-false}
   MAX_SUBMIT_COUNT: ${MAX_SUBMIT_COUNT:-100}
   TOP_K_MAX_VALUE: ${TOP_K_MAX_VALUE:-10}
+  ALLOW_EMBED: ${ALLOW_EMBED:-false}
 
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.15.3
+    image: langgenius/dify-api:0.15.8
     restart: always
     environment:
       # Use the shared environment variables.
@@ -416,7 +418,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.15.3
+    image: langgenius/dify-api:0.15.8
     restart: always
     environment:
       # Use the shared environment variables.
@@ -438,7 +440,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.15.3
+    image: langgenius/dify-web:0.15.8
     restart: always
     environment:
       CONSOLE_API_URL: ${CONSOLE_API_URL:-}
@@ -447,6 +449,7 @@ services:
       NEXT_TELEMETRY_DISABLED: ${NEXT_TELEMETRY_DISABLED:-0}
       TEXT_GENERATION_TIMEOUT_MS: ${TEXT_GENERATION_TIMEOUT_MS:-60000}
       CSP_WHITELIST: ${CSP_WHITELIST:-}
+      ALLOW_EMBED: ${ALLOW_EMBED:-false}
       TOP_K_MAX_VALUE: ${TOP_K_MAX_VALUE:-}
       INDEXING_MAX_SEGMENTATION_TOKENS_LENGTH: ${INDEXING_MAX_SEGMENTATION_TOKENS_LENGTH:-}
 
@@ -489,7 +492,7 @@ services:
 
   # The DifySandbox
   sandbox:
-    image: langgenius/dify-sandbox:0.2.10
+    image: langgenius/dify-sandbox:0.2.11
     restart: always
     environment:
       # The DifySandbox configurations

web/.env.example

@@ -31,3 +31,6 @@ NEXT_PUBLIC_TOP_K_MAX_VALUE=10
 
 # The maximum number of tokens for segmentation
 NEXT_PUBLIC_INDEXING_MAX_SEGMENTATION_TOKENS_LENGTH=4000
+
+# Default is not allow to embed into iframe to prevent Clickjacking: https://owasp.org/www-community/attacks/Clickjacking
+NEXT_PUBLIC_ALLOW_EMBED=

web/app/components/app/create-app-dialog/app-list/index.tsx

@@ -186,15 +186,17 @@ const Apps = ({
         <div className='w-[180px] h-8'></div>
       </div>
       <div className='relative flex flex-1 overflow-y-auto'>
-        {!searchKeywords && <div className='w-[200px] h-full p-4'>
-          <Sidebar current={currCategory as AppCategories} onClick={(category) => { setCurrCategory(category) }} onCreateFromBlank={onCreateFromBlank} />
+        {!searchKeywords && <div className='h-full w-[200px] p-4'>
+          <Sidebar current={currCategory as AppCategories} categories={categories} onClick={(category) => { setCurrCategory(category) }} onCreateFromBlank={onCreateFromBlank} />
         </div>}
         <div className='flex-1 h-full overflow-auto shrink-0 grow p-6 pt-2 border-l border-divider-burn'>
           {searchFilteredList && searchFilteredList.length > 0 && <>
             <div className='pt-4 pb-1'>
               {searchKeywords
                 ? <p className='title-md-semi-bold text-text-tertiary'>{searchFilteredList.length > 1 ? t('app.newApp.foundResults', { count: searchFilteredList.length }) : t('app.newApp.foundResult', { count: searchFilteredList.length })}</p>
-                : <AppCategoryLabel category={currCategory as AppCategories} className='title-md-semi-bold text-text-primary' />}
+                : <div className='flex h-[22px] items-center'>
+                  <AppCategoryLabel category={currCategory as AppCategories} className='title-md-semi-bold text-text-primary' />
+                </div>}
             </div>
             <div
               className={cn(

web/app/components/app/create-app-dialog/app-list/sidebar.tsx

@@ -1,39 +1,29 @@
 'use client'
-import { RiAppsFill, RiChatSmileAiFill, RiExchange2Fill, RiPassPendingFill, RiQuillPenAiFill, RiSpeakAiFill, RiStickyNoteAddLine, RiTerminalBoxFill, RiThumbUpFill } from '@remixicon/react'
+import { RiStickyNoteAddLine, RiThumbUpLine } from '@remixicon/react'
 import { useTranslation } from 'react-i18next'
 import classNames from '@/utils/classnames'
 import Divider from '@/app/components/base/divider'
 
 export enum AppCategories {
   RECOMMENDED = 'Recommended',
-  ASSISTANT = 'Assistant',
-  AGENT = 'Agent',
-  HR = 'HR',
-  PROGRAMMING = 'Programming',
-  WORKFLOW = 'Workflow',
-  WRITING = 'Writing',
 }
 
 type SidebarProps = {
-  current: AppCategories
-  onClick?: (category: AppCategories) => void
+  current: AppCategories | string
+  categories: string[]
+  onClick?: (category: AppCategories | string) => void
   onCreateFromBlank?: () => void
 }
 
-export default function Sidebar({ current, onClick, onCreateFromBlank }: SidebarProps) {
+export default function Sidebar({ current, categories, onClick, onCreateFromBlank }: SidebarProps) {
   const { t } = useTranslation()
-  return <div className="w-full h-full flex flex-col">
-    <ul>
+  return <div className="flex h-full w-full flex-col">
+    <ul className='pt-0.5'>
       <CategoryItem category={AppCategories.RECOMMENDED} active={current === AppCategories.RECOMMENDED} onClick={onClick} />
     </ul>
-    <div className='px-3 pt-2 pb-1 system-xs-medium-uppercase text-text-tertiary'>{t('app.newAppFromTemplate.byCategories')}</div>
-    <ul className='flex-grow flex flex-col gap-0.5'>
-      <CategoryItem category={AppCategories.ASSISTANT} active={current === AppCategories.ASSISTANT} onClick={onClick} />
-      <CategoryItem category={AppCategories.AGENT} active={current === AppCategories.AGENT} onClick={onClick} />
-      <CategoryItem category={AppCategories.HR} active={current === AppCategories.HR} onClick={onClick} />
-      <CategoryItem category={AppCategories.PROGRAMMING} active={current === AppCategories.PROGRAMMING} onClick={onClick} />
-      <CategoryItem category={AppCategories.WORKFLOW} active={current === AppCategories.WORKFLOW} onClick={onClick} />
-      <CategoryItem category={AppCategories.WRITING} active={current === AppCategories.WRITING} onClick={onClick} />
+    <div className='system-xs-medium-uppercase mb-0.5 mt-3 px-3 pb-1 pt-2 text-text-tertiary'>{t('app.newAppFromTemplate.byCategories')}</div>
+    <ul className='flex grow flex-col gap-0.5'>
+      {categories.map(category => (<CategoryItem key={category} category={category} active={current === category} onClick={onClick} />))}
     </ul>
     <Divider bgStyle='gradient' />
     <div className='px-3 py-1 flex items-center gap-1 text-text-tertiary cursor-pointer' onClick={onCreateFromBlank}>
@@ -45,47 +35,26 @@ export default function Sidebar({ current, onClick, onCreateFromBlank }: Sidebar
 
 type CategoryItemProps = {
   active: boolean
-  category: AppCategories
-  onClick?: (category: AppCategories) => void
+  category: AppCategories | string
+  onClick?: (category: AppCategories | string) => void
 }
 function CategoryItem({ category, active, onClick }: CategoryItemProps) {
   return <li
-    className={classNames('p-1 pl-3 rounded-lg flex items-center gap-2 group cursor-pointer hover:bg-state-base-hover [&.active]:bg-state-base-active', active && 'active')}
+    className={classNames('p-1 pl-3 h-8 rounded-lg flex items-center gap-2 group cursor-pointer hover:bg-state-base-hover [&.active]:bg-state-base-active', active && 'active')}
     onClick={() => { onClick?.(category) }}>
-    <div className='w-5 h-5 inline-flex items-center justify-center rounded-md border border-divider-regular bg-components-icon-bg-midnight-solid group-[.active]:bg-components-icon-bg-blue-solid'>
-      <AppCategoryIcon category={category} />
-    </div>
+    {category === AppCategories.RECOMMENDED && <div className='inline-flex h-5 w-5 items-center justify-center rounded-md'>
+      <RiThumbUpLine className='h-4 w-4 text-components-menu-item-text group-[.active]:text-components-menu-item-text-active' />
+    </div>}
     <AppCategoryLabel category={category}
       className={classNames('system-sm-medium text-components-menu-item-text group-[.active]:text-components-menu-item-text-active group-hover:text-components-menu-item-text-hover', active && 'system-sm-semibold')} />
   </li >
 }
 
 type AppCategoryLabelProps = {
-  category: AppCategories
+  category: AppCategories | string
   className?: string
 }
 export function AppCategoryLabel({ category, className }: AppCategoryLabelProps) {
   const { t } = useTranslation()
-  return <span className={className}>{t(`app.newAppFromTemplate.sidebar.${category}`)}</span>
-}
-
-type AppCategoryIconProps = {
-  category: AppCategories
-}
-function AppCategoryIcon({ category }: AppCategoryIconProps) {
-  if (category === AppCategories.AGENT)
-    return <RiSpeakAiFill className='w-3.5 h-3.5 text-components-avatar-shape-fill-stop-100' />
-  if (category === AppCategories.ASSISTANT)
-    return <RiChatSmileAiFill className='w-3.5 h-3.5 text-components-avatar-shape-fill-stop-100' />
-  if (category === AppCategories.HR)
-    return <RiPassPendingFill className='w-3.5 h-3.5 text-components-avatar-shape-fill-stop-100' />
-  if (category === AppCategories.PROGRAMMING)
-    return <RiTerminalBoxFill className='w-3.5 h-3.5 text-components-avatar-shape-fill-stop-100' />
-  if (category === AppCategories.RECOMMENDED)
-    return <RiThumbUpFill className='w-3.5 h-3.5 text-components-avatar-shape-fill-stop-100' />
-  if (category === AppCategories.WRITING)
-    return <RiQuillPenAiFill className='w-3.5 h-3.5 text-components-avatar-shape-fill-stop-100' />
-  if (category === AppCategories.WORKFLOW)
-    return <RiExchange2Fill className='w-3.5 h-3.5 text-components-avatar-shape-fill-stop-100' />
-  return <RiAppsFill className='w-3.5 h-3.5 text-components-avatar-shape-fill-stop-100' />
+  return <span className={className}>{category === AppCategories.RECOMMENDED ? t('app.newAppFromTemplate.sidebar.Recommended') : category}</span>
 }

web/app/components/app/create-app-modal/index.tsx

@@ -312,7 +312,7 @@ function AppPreview({ mode }: { mode: AppMode }) {
     'chat': {
       title: t('app.types.chatbot'),
       description: t('app.newApp.chatbotUserDescription'),
-      link: 'https://docs.dify.ai/guides/application-orchestrate/conversation-application?fallback=true',
+      link: 'https://docs.dify.ai/guides/application-orchestrate#application_type',
     },
     'advanced-chat': {
       title: t('app.types.advanced'),

web/app/components/app/create-from-dsl-modal/dsl-confirm-modal.tsx

@@ -0,0 +1,46 @@
+import { useTranslation } from 'react-i18next'
+import Modal from '@/app/components/base/modal'
+import Button from '@/app/components/base/button'
+
+type DSLConfirmModalProps = {
+  versions?: {
+    importedVersion: string
+    systemVersion: string
+  }
+  onCancel: () => void
+  onConfirm: () => void
+  confirmDisabled?: boolean
+}
+const DSLConfirmModal = ({
+  versions = { importedVersion: '', systemVersion: '' },
+  onCancel,
+  onConfirm,
+  confirmDisabled = false,
+}: DSLConfirmModalProps) => {
+  const { t } = useTranslation()
+
+  return (
+    <Modal
+      isShow
+      onClose={() => onCancel()}
+      className='w-[480px]'
+    >
+      <div className='flex flex-col items-start gap-2 self-stretch pb-4'>
+        <div className='title-2xl-semi-bold text-text-primary'>{t('app.newApp.appCreateDSLErrorTitle')}</div>
+        <div className='system-md-regular flex grow flex-col text-text-secondary'>
+          <div>{t('app.newApp.appCreateDSLErrorPart1')}</div>
+          <div>{t('app.newApp.appCreateDSLErrorPart2')}</div>
+          <br />
+          <div>{t('app.newApp.appCreateDSLErrorPart3')}<span className='system-md-medium'>{versions.importedVersion}</span></div>
+          <div>{t('app.newApp.appCreateDSLErrorPart4')}<span className='system-md-medium'>{versions.systemVersion}</span></div>
+        </div>
+      </div>
+      <div className='flex items-start justify-end gap-2 self-stretch pt-6'>
+        <Button variant='secondary' onClick={() => onCancel()}>{t('app.newApp.Cancel')}</Button>
+        <Button variant='primary' destructive onClick={onConfirm} disabled={confirmDisabled}>{t('app.newApp.Confirm')}</Button>
+      </div>
+    </Modal>
+  )
+}
+
+export default DSLConfirmModal

web/app/components/app/overview/embedded/index.tsx

@@ -24,7 +24,7 @@ const OPTION_MAP = {
   iframe: {
     getContent: (url: string, token: string) =>
       `<iframe
- src="${url}/chatbot/${token}"
+ src="${url}/chat/${token}"
  style="width: 100%; height: 100%; min-height: 700px"
  frameborder="0"
  allow="microphone">
@@ -35,12 +35,12 @@ const OPTION_MAP = {
       `<script>
  window.difyChatbotConfig = {
   token: '${token}'${isTestEnv
-  ? `,
+        ? `,
   isDev: true`
-  : ''}${IS_CE_EDITION
-  ? `,
+        : ''}${IS_CE_EDITION
+          ? `,
   baseUrl: '${url}'`
-  : ''}
+          : ''}
  }
 </script>
 <script

web/app/components/base/chat/embedded-chatbot/hooks.tsx

@@ -11,10 +11,12 @@ import { useLocalStorageState } from 'ahooks'
 import produce from 'immer'
 import type {
   ChatConfig,
+  ChatItem,
   Feedback,
 } from '../types'
 import { CONVERSATION_ID_INFO } from '../constants'
-import { getPrevChatList, getProcessedInputsFromUrlParams } from '../utils'
+import { buildChatItemTree, getProcessedInputsFromUrlParams } from '../utils'
+import { getProcessedFilesFromResponse } from '../../file-uploader/utils'
 import {
   fetchAppInfo,
   fetchAppMeta,
@@ -32,6 +34,33 @@ import { useToastContext } from '@/app/components/base/toast'
 import { changeLanguage } from '@/i18n/i18next-config'
 import { InputVarType } from '@/app/components/workflow/types'
 import { TransferMethod } from '@/types/app'
+import { addFileInfos, sortAgentSorts } from '@/app/components/tools/utils'
+
+function getFormattedChatList(messages: any[]) {
+  const newChatList: ChatItem[] = []
+  messages.forEach((item) => {
+    const questionFiles = item.message_files?.filter((file: any) => file.belongs_to === 'user') || []
+    newChatList.push({
+      id: `question-${item.id}`,
+      content: item.query,
+      isAnswer: false,
+      message_files: getProcessedFilesFromResponse(questionFiles.map((item: any) => ({ ...item, related_id: item.id }))),
+      parentMessageId: item.parent_message_id || undefined,
+    })
+    const answerFiles = item.message_files?.filter((file: any) => file.belongs_to === 'assistant') || []
+    newChatList.push({
+      id: item.id,
+      content: item.answer,
+      agent_thoughts: addFileInfos(item.agent_thoughts ? sortAgentSorts(item.agent_thoughts) : item.agent_thoughts, item.message_files),
+      feedback: item.feedback,
+      isAnswer: true,
+      citation: item.retriever_resources,
+      message_files: getProcessedFilesFromResponse(answerFiles.map((item: any) => ({ ...item, related_id: item.id }))),
+      parentMessageId: `question-${item.id}`,
+    })
+  })
+  return newChatList
+}
 
 export const useEmbeddedChatbot = () => {
   const isInstalledApp = false
@@ -77,7 +106,7 @@ export const useEmbeddedChatbot = () => {
 
   const appPrevChatList = useMemo(
     () => (currentConversationId && appChatListData?.data.length)
-      ? getPrevChatList(appChatListData.data)
+      ? buildChatItemTree(getFormattedChatList(appChatListData.data))
       : [],
     [appChatListData, currentConversationId],
   )

web/app/components/base/chat/utils.ts

@@ -1,6 +1,8 @@
 import { UUID_NIL } from './constants'
 import type { IChatItem } from './chat/type'
 import type { ChatItem, ChatItemInTree } from './types'
+import { addFileInfos, sortAgentSorts } from '../../tools/utils'
+import { getProcessedFilesFromResponse } from '../file-uploader/utils'
 
 async function decodeBase64AndDecompress(base64String: string) {
   const binaryString = atob(base64String)
@@ -19,6 +21,60 @@ function getProcessedInputsFromUrlParams(): Record<string, any> {
   return inputs
 }
 
+function appendQAToChatList(chatList: ChatItem[], item: any) {
+  // we append answer first and then question since will reverse the whole chatList later
+  const answerFiles = item.message_files?.filter((file: any) => file.belongs_to === 'assistant') || []
+  chatList.push({
+    id: item.id,
+    content: item.answer,
+    agent_thoughts: addFileInfos(item.agent_thoughts ? sortAgentSorts(item.agent_thoughts) : item.agent_thoughts, item.message_files),
+    feedback: item.feedback,
+    isAnswer: true,
+    citation: item.retriever_resources,
+    message_files: getProcessedFilesFromResponse(answerFiles.map((item: any) => ({ ...item, related_id: item.id }))),
+  })
+  const questionFiles = item.message_files?.filter((file: any) => file.belongs_to === 'user') || []
+  chatList.push({
+    id: `question-${item.id}`,
+    content: item.query,
+    isAnswer: false,
+    message_files: getProcessedFilesFromResponse(questionFiles.map((item: any) => ({ ...item, related_id: item.id }))),
+  })
+}
+
+/**
+ * Computes the latest thread messages from all messages of the conversation.
+ * Same logic as backend codebase `api/core/prompt/utils/extract_thread_messages.py`
+ *
+ * @param fetchedMessages - The history chat list data from the backend, sorted by created_at in descending order. This includes all flattened history messages of the conversation.
+ * @returns An array of ChatItems representing the latest thread.
+ */
+
+
+function getPrevChatList(fetchedMessages: any[]) {
+  const ret: ChatItem[] = []
+  let nextMessageId = null
+
+  for (const item of fetchedMessages) {
+    if (!item.parent_message_id) {
+      appendQAToChatList(ret, item)
+      break
+    }
+
+    if (!nextMessageId) {
+      appendQAToChatList(ret, item)
+      nextMessageId = item.parent_message_id
+    }
+    else {
+      if (item.id === nextMessageId || nextMessageId === UUID_NIL) {
+        appendQAToChatList(ret, item)
+        nextMessageId = item.parent_message_id
+      }
+    }
+  }
+  return ret.reverse()
+}
+
 function isValidGeneratedAnswer(item?: ChatItem | ChatItemInTree): boolean {
   return !!item && item.isAnswer && !item.id.startsWith('answer-placeholder-') && !item.isOpeningStatement
 }
@@ -164,6 +220,7 @@ function getThreadMessages(tree: ChatItemInTree[], targetMessageId?: string): Ch
 export {
   getProcessedInputsFromUrlParams,
   isValidGeneratedAnswer,
+  getPrevChatList,
   getLastAnswer,
   buildChatItemTree,
   getThreadMessages,

web/app/components/base/markdown.tsx

@@ -10,6 +10,7 @@ import SyntaxHighlighter from 'react-syntax-highlighter'
 import { atelierHeathLight } from 'react-syntax-highlighter/dist/esm/styles/hljs'
 import { Component, memo, useMemo, useRef, useState } from 'react'
 import type { CodeComponent } from 'react-markdown/lib/ast-to-react'
+import SVGRenderer from './svg-gallery'
 import cn from '@/utils/classnames'
 import CopyBtn from '@/app/components/base/copy-btn'
 import SVGBtn from '@/app/components/base/svg'
@@ -18,7 +19,7 @@ import ImageGallery from '@/app/components/base/image-gallery'
 import { useChatContext } from '@/app/components/base/chat/chat/context'
 import VideoGallery from '@/app/components/base/video-gallery'
 import AudioGallery from '@/app/components/base/audio-gallery'
-import SVGRenderer from '@/app/components/base/svg-gallery'
+// import SVGRenderer from '@/app/components/base/svg-gallery'
 import MarkdownButton from '@/app/components/base/markdown-blocks/button'
 import MarkdownForm from '@/app/components/base/markdown-blocks/form'
 

web/app/components/base/svg-gallery/index.tsx

@@ -1,5 +1,6 @@
 import { useEffect, useRef, useState } from 'react'
 import { SVG } from '@svgdotjs/svg.js'
+import DOMPurify from 'dompurify'
 import ImagePreview from '@/app/components/base/image-uploader/image-preview'
 
 export const SVGRenderer = ({ content }: { content: string }) => {
@@ -44,7 +45,7 @@ export const SVGRenderer = ({ content }: { content: string }) => {
 
         svgRef.current.style.width = `${Math.min(originalWidth, 298)}px`
 
-        const rootElement = draw.svg(content)
+        const rootElement = draw.svg(DOMPurify.sanitize(content))
 
         rootElement.click(() => {
           setImagePreview(svgToDataURL(svgElement as Element))

web/app/components/explore/app-list/index.tsx

@@ -1,12 +1,10 @@
 'use client'
 
-import React, { useMemo, useState } from 'react'
-import { useRouter } from 'next/navigation'
+import React, { useCallback, useMemo, useState } from 'react'
 import { useTranslation } from 'react-i18next'
 import { useContext } from 'use-context-selector'
 import useSWR from 'swr'
 import { useDebounceFn } from 'ahooks'
-import Toast from '../../base/toast'
 import s from './style.module.css'
 import cn from '@/utils/classnames'
 import ExploreContext from '@/context/explore-context'
@@ -14,17 +12,17 @@ import type { App } from '@/models/explore'
 import Category from '@/app/components/explore/category'
 import AppCard from '@/app/components/explore/app-card'
 import { fetchAppDetail, fetchAppList } from '@/service/explore'
-import { importDSL } from '@/service/apps'
 import { useTabSearchParams } from '@/hooks/use-tab-searchparams'
 import CreateAppModal from '@/app/components/explore/create-app-modal'
 import AppTypeSelector from '@/app/components/app/type-selector'
 import type { CreateAppModalProps } from '@/app/components/explore/create-app-modal'
 import Loading from '@/app/components/base/loading'
-import { NEED_REFRESH_APP_LIST_KEY } from '@/config'
-import { useAppContext } from '@/context/app-context'
-import { getRedirection } from '@/utils/app-redirection'
 import Input from '@/app/components/base/input'
-import { DSLImportMode } from '@/models/app'
+import {
+  DSLImportMode,
+} from '@/models/app'
+import { useImportDSL } from '@/hooks/use-import-dsl'
+import DSLConfirmModal from '@/app/components/app/create-from-dsl-modal/dsl-confirm-modal'
 
 type AppsProps = {
   pageType?: PageType
@@ -41,8 +39,6 @@ const Apps = ({
   onSuccess,
 }: AppsProps) => {
   const { t } = useTranslation()
-  const { isCurrentWorkspaceEditor } = useAppContext()
-  const { push } = useRouter()
   const { hasEditPermission } = useContext(ExploreContext)
   const allCategoriesEn = t('explore.apps.allCategories', { lng: 'en' })
 
@@ -117,6 +113,14 @@ const Apps = ({
 
   const [currApp, setCurrApp] = React.useState<App | null>(null)
   const [isShowCreateModal, setIsShowCreateModal] = React.useState(false)
+
+  const {
+    handleImportDSL,
+    handleImportDSLConfirm,
+    versions,
+    isFetching,
+  } = useImportDSL()
+  const [showDSLConfirmModal, setShowDSLConfirmModal] = useState(false)
   const onCreate: CreateAppModalProps['onConfirm'] = async ({
     name,
     icon_type,
@@ -127,31 +131,31 @@ const Apps = ({
     const { export_data } = await fetchAppDetail(
       currApp?.app.id as string,
     )
-    try {
-      const app = await importDSL({
-        mode: DSLImportMode.YAML_CONTENT,
-        yaml_content: export_data,
-        name,
-        icon_type,
-        icon,
-        icon_background,
-        description,
-      })
-      setIsShowCreateModal(false)
-      Toast.notify({
-        type: 'success',
-        message: t('app.newApp.appCreated'),
-      })
-      if (onSuccess)
-        onSuccess()
-      localStorage.setItem(NEED_REFRESH_APP_LIST_KEY, '1')
-      getRedirection(isCurrentWorkspaceEditor, { id: app.app_id }, push)
-    }
-    catch (e) {
-      Toast.notify({ type: 'error', message: t('app.newApp.appCreateFailed') })
+    const payload = {
+      mode: DSLImportMode.YAML_CONTENT,
+      yaml_content: export_data,
+      name,
+      icon_type,
+      icon,
+      icon_background,
+      description,
     }
+    await handleImportDSL(payload, {
+      onSuccess: () => {
+        setIsShowCreateModal(false)
+      },
+      onPending: () => {
+        setShowDSLConfirmModal(true)
+      },
+    })
   }
 
+  const onConfirmDSL = useCallback(async () => {
+    await handleImportDSLConfirm({
+      onSuccess,
+    })
+  }, [handleImportDSLConfirm, onSuccess])
+
   if (!categories || categories.length === 0) {
     return (
       <div className="flex h-full items-center">
@@ -234,9 +238,20 @@ const Apps = ({
           appDescription={currApp?.app.description || ''}
           show={isShowCreateModal}
           onConfirm={onCreate}
+          confirmDisabled={isFetching}
           onHide={() => setIsShowCreateModal(false)}
         />
       )}
+      {
+        showDSLConfirmModal && (
+          <DSLConfirmModal
+            versions={versions}
+            onCancel={() => setShowDSLConfirmModal(false)}
+            onConfirm={onConfirmDSL}
+            confirmDisabled={isFetching}
+          />
+        )
+      }
     </div>
   )
 }

web/app/components/explore/create-app-modal/index.tsx

@@ -33,6 +33,7 @@ export type CreateAppModalProps = {
     description: string
     use_icon_as_answer_icon?: boolean
   }) => Promise<void>
+  confirmDisabled?: boolean
   onHide: () => void
 }
 
@@ -48,6 +49,7 @@ const CreateAppModal = ({
   appMode,
   appUseIconAsAnswerIcon,
   onConfirm,
+  confirmDisabled,
   onHide,
 }: CreateAppModalProps) => {
   const { t } = useTranslation()
@@ -145,7 +147,7 @@ const CreateAppModal = ({
           {!isEditModal && isAppsFull && <AppsFull loc='app-explore-create' />}
         </div>
         <div className='flex flex-row-reverse'>
-          <Button disabled={!isEditModal && isAppsFull} className='w-24 ml-2' variant='primary' onClick={submit}>{!isEditModal ? t('common.operation.create') : t('common.operation.save')}</Button>
+          <Button disabled={(!isEditModal && isAppsFull) || !name.trim() || confirmDisabled} className='w-24 ml-2' variant='primary' onClick={submit}>{!isEditModal ? t('common.operation.create') : t('common.operation.save')}</Button>
           <Button className='w-24' onClick={onHide}>{t('common.operation.cancel')}</Button>
         </div>
       </Modal>

web/app/reset-password/check-code/page.tsx

@@ -39,7 +39,11 @@ export default function CheckCode() {
       }
       setIsLoading(true)
       const ret = await verifyResetPasswordCode({ email, code, token })
-      ret.is_valid && router.push(`/reset-password/set-password?${searchParams.toString()}`)
+      if (ret.is_valid) {
+        const params = new URLSearchParams(searchParams)
+        params.set('token', encodeURIComponent(ret.token))
+        router.push(`/reset-password/set-password?${params.toString()}`)
+      }
     }
     catch (error) { console.error(error) }
     finally {

web/docker/entrypoint.sh

@@ -23,6 +23,7 @@ export NEXT_TELEMETRY_DISABLED=${NEXT_TELEMETRY_DISABLED}
 
 export NEXT_PUBLIC_TEXT_GENERATION_TIMEOUT_MS=${TEXT_GENERATION_TIMEOUT_MS}
 export NEXT_PUBLIC_CSP_WHITELIST=${CSP_WHITELIST}
+export NEXT_PUBLIC_ALLOW_EMBED=${ALLOW_EMBED}
 export NEXT_PUBLIC_TOP_K_MAX_VALUE=${TOP_K_MAX_VALUE}
 export NEXT_PUBLIC_INDEXING_MAX_SEGMENTATION_TOKENS_LENGTH=${INDEXING_MAX_SEGMENTATION_TOKENS_LENGTH}
 

web/hooks/use-import-dsl.ts

@@ -0,0 +1,158 @@
+import {
+  useCallback,
+  useRef,
+  useState,
+} from 'react'
+import { useTranslation } from 'react-i18next'
+import { useRouter } from 'next/navigation'
+import type {
+  DSLImportMode,
+  DSLImportResponse,
+} from '@/models/app'
+import { DSLImportStatus } from '@/models/app'
+import {
+  importDSL,
+  importDSLConfirm,
+} from '@/service/apps'
+import type { AppIconType } from '@/types/app'
+import { useToastContext } from '@/app/components/base/toast'
+import { getRedirection } from '@/utils/app-redirection'
+import { useSelector } from '@/context/app-context'
+import { NEED_REFRESH_APP_LIST_KEY } from '@/config'
+
+type DSLPayload = {
+  mode: DSLImportMode
+  yaml_content?: string
+  yaml_url?: string
+  name?: string
+  icon_type?: AppIconType
+  icon?: string
+  icon_background?: string
+  description?: string
+}
+type ResponseCallback = {
+  onSuccess?: () => void
+  onPending?: (payload: DSLImportResponse) => void
+  onFailed?: () => void
+}
+export const useImportDSL = () => {
+  const { t } = useTranslation()
+  const { notify } = useToastContext()
+  const [isFetching, setIsFetching] = useState(false)
+  const isCurrentWorkspaceEditor = useSelector(s => s.isCurrentWorkspaceEditor)
+  const { push } = useRouter()
+  const [versions, setVersions] = useState<{ importedVersion: string; systemVersion: string }>()
+  const importIdRef = useRef<string>('')
+
+  const handleImportDSL = useCallback(async (
+    payload: DSLPayload,
+    {
+      onSuccess,
+      onPending,
+      onFailed,
+    }: ResponseCallback,
+  ) => {
+    if (isFetching)
+      return
+    setIsFetching(true)
+
+    try {
+      const response = await importDSL(payload)
+
+      if (!response)
+        return
+
+      const {
+        id,
+        status,
+        app_id,
+        imported_dsl_version,
+        current_dsl_version,
+      } = response
+
+      if (status === DSLImportStatus.COMPLETED || status === DSLImportStatus.COMPLETED_WITH_WARNINGS) {
+        if (!app_id)
+          return
+
+        notify({
+          type: status === DSLImportStatus.COMPLETED ? 'success' : 'warning',
+          message: t(status === DSLImportStatus.COMPLETED ? 'app.newApp.appCreated' : 'app.newApp.caution'),
+          children: status === DSLImportStatus.COMPLETED_WITH_WARNINGS && t('app.newApp.appCreateDSLWarning'),
+        })
+        onSuccess?.()
+        localStorage.setItem(NEED_REFRESH_APP_LIST_KEY, '1')
+        getRedirection(isCurrentWorkspaceEditor, { id: app_id }, push)
+      }
+      else if (status === DSLImportStatus.PENDING) {
+        setVersions({
+          importedVersion: imported_dsl_version ?? '',
+          systemVersion: current_dsl_version ?? '',
+        })
+        importIdRef.current = id
+        onPending?.(response)
+      }
+      else {
+        notify({ type: 'error', message: t('app.newApp.appCreateFailed') })
+        onFailed?.()
+      }
+    }
+    catch {
+      notify({ type: 'error', message: t('app.newApp.appCreateFailed') })
+      onFailed?.()
+    }
+    finally {
+      setIsFetching(false)
+    }
+  }, [t, notify, isCurrentWorkspaceEditor, push, isFetching])
+
+  const handleImportDSLConfirm = useCallback(async (
+    {
+      onSuccess,
+      onFailed,
+    }: Pick<ResponseCallback, 'onSuccess' | 'onFailed'>,
+  ) => {
+    if (isFetching)
+      return
+    setIsFetching(true)
+    if (!importIdRef.current)
+      return
+
+    try {
+      const response = await importDSLConfirm({
+        import_id: importIdRef.current,
+      })
+
+      const { status, app_id } = response
+      if (!app_id)
+        return
+
+      if (status === DSLImportStatus.COMPLETED) {
+        onSuccess?.()
+        notify({
+          type: 'success',
+          message: t('app.newApp.appCreated'),
+        })
+        localStorage.setItem(NEED_REFRESH_APP_LIST_KEY, '1')
+        getRedirection(isCurrentWorkspaceEditor, { id: app_id! }, push)
+      }
+      else if (status === DSLImportStatus.FAILED) {
+        notify({ type: 'error', message: t('app.newApp.appCreateFailed') })
+        onFailed?.()
+      }
+    }
+    catch {
+      notify({ type: 'error', message: t('app.newApp.appCreateFailed') })
+      onFailed?.()
+    }
+    finally {
+      setIsFetching(false)
+    }
+  }, [t, notify, isCurrentWorkspaceEditor, push, isFetching])
+
+  return {
+    handleImportDSL,
+    handleImportDSLConfirm,
+    versions,
+    isFetching,
+  }
+}

web/middleware.ts

@@ -3,10 +3,26 @@ import { NextResponse } from 'next/server'
 
 const NECESSARY_DOMAIN = '*.sentry.io http://localhost:* http://127.0.0.1:* https://analytics.google.com googletagmanager.com *.googletagmanager.com https://www.google-analytics.com https://api.github.com'
 
+const wrapResponseWithXFrameOptions = (response: NextResponse, pathname: string) => {
+  // prevent clickjacking: https://owasp.org/www-community/attacks/Clickjacking
+  // Chatbot page should be allowed to be embedded in iframe. It's a feature
+  if (process.env.NEXT_PUBLIC_ALLOW_EMBED !== 'true' && !pathname.startsWith('/chat'))
+    response.headers.set('X-Frame-Options', 'DENY')
+
+  return response
+}
 export function middleware(request: NextRequest) {
+  const { pathname } = request.nextUrl
+  const requestHeaders = new Headers(request.headers)
+  const response = NextResponse.next({
+    request: {
+      headers: requestHeaders,
+    },
+  })
+
   const isWhiteListEnabled = !!process.env.NEXT_PUBLIC_CSP_WHITELIST && process.env.NODE_ENV === 'production'
   if (!isWhiteListEnabled)
-    return NextResponse.next()
+    return wrapResponseWithXFrameOptions(response, pathname)
 
   const whiteList = `${process.env.NEXT_PUBLIC_CSP_WHITELIST} ${NECESSARY_DOMAIN}`
   const nonce = Buffer.from(crypto.randomUUID()).toString('base64')
@@ -33,7 +49,6 @@ export function middleware(request: NextRequest) {
     .replace(/\s{2,}/g, ' ')
     .trim()
 
-  const requestHeaders = new Headers(request.headers)
   requestHeaders.set('x-nonce', nonce)
 
   requestHeaders.set(
@@ -41,17 +56,12 @@ export function middleware(request: NextRequest) {
     contentSecurityPolicyHeaderValue,
   )
 
-  const response = NextResponse.next({
-    request: {
-      headers: requestHeaders,
-    },
-  })
   response.headers.set(
     'Content-Security-Policy',
     contentSecurityPolicyHeaderValue,
   )
 
-  return response
+  return wrapResponseWithXFrameOptions(response, pathname)
 }
 
 export const config = {
@@ -73,4 +83,4 @@ export const config = {
       // ],
     },
   ],
-}
+}

web/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dify-web",
-  "version": "0.15.3",
+  "version": "0.15.8",
   "private": true,
   "engines": {
     "node": ">=18.17.0"
@@ -51,6 +51,7 @@
     "crypto-js": "^4.2.0",
     "dayjs": "^1.11.7",
     "decimal.js": "^10.4.3",
+    "dompurify": "^3.2.4",
     "echarts": "^5.5.1",
     "echarts-for-react": "^3.0.2",
     "elkjs": "^0.9.3",
@@ -70,7 +71,7 @@
     "mermaid": "11.4.1",
     "mime": "^4.0.4",
     "negotiator": "^0.6.3",
-    "next": "^14.2.10",
+    "next": "^14.2.25",
     "pinyin-pro": "^3.23.0",
     "qrcode.react": "^3.1.0",
     "qs": "^6.11.1",
@@ -160,7 +161,7 @@
     "jest-environment-jsdom": "^29.7.0",
     "lint-staged": "^13.2.2",
     "magicast": "^0.3.4",
-    "postcss": "^8.4.31",
+    "postcss": "^8.4.47",
     "sass": "^1.61.0",
     "storybook": "^8.3.5",
     "tailwindcss": "^3.4.4",
@@ -171,7 +172,10 @@
   "resolutions": {
     "@types/react": "~18.2.0",
     "@types/react-dom": "~18.2.0",
-    "string-width": "4.2.3"
+    "string-width": "4.2.3",
+    "nanoid": "~3.3.8",
+    "esbuild": "~0.25.0",
+    "serialize-javascript": "~6.0.2"
   },
   "lint-staged": {
     "**/*.js?(x)": [

web/service/common.ts

@@ -40,7 +40,7 @@ import type { SystemFeatures } from '@/types/feature'
 
 type LoginSuccess = {
   result: 'success'
-  data: { access_token: string;refresh_token: string }
+  data: { access_token: string; refresh_token: string }
 }
 type LoginFail = {
   result: 'fail'
@@ -331,20 +331,20 @@ export const uploadRemoteFileInfo = (url: string, isPublic?: boolean) => {
 export const sendEMailLoginCode = (email: string, language = 'en-US') =>
   post<CommonResponse & { data: string }>('/email-code-login', { body: { email, language } })
 
-export const emailLoginWithCode = (data: { email: string;code: string;token: string }) =>
+export const emailLoginWithCode = (data: { email: string; code: string; token: string }) =>
   post<LoginResponse>('/email-code-login/validity', { body: data })
 
 export const sendResetPasswordCode = (email: string, language = 'en-US') =>
-  post<CommonResponse & { data: string;message?: string ;code?: string }>('/forgot-password', { body: { email, language } })
+  post<CommonResponse & { data: string; message?: string; code?: string }>('/forgot-password', { body: { email, language } })
 
-export const verifyResetPasswordCode = (body: { email: string;code: string;token: string }) =>
-  post<CommonResponse & { is_valid: boolean }>('/forgot-password/validity', { body })
+export const verifyResetPasswordCode = (body: { email: string; code: string; token: string }) =>
+  post<CommonResponse & { is_valid: boolean; token: string }>('/forgot-password/validity', { body })
 
 export const sendDeleteAccountCode = () =>
   get<CommonResponse & { data: string }>('/account/delete/verify')
 
-export const verifyDeleteAccountCode = (body: { code: string;token: string }) =>
+export const verifyDeleteAccountCode = (body: { code: string; token: string }) =>
   post<CommonResponse & { is_valid: boolean }>('/account/delete', { body })
 
-export const submitDeleteAccountFeedback = (body: { feedback: string;email: string }) =>
+export const submitDeleteAccountFeedback = (body: { feedback: string; email: string }) =>
   post<CommonResponse>('/account/delete/feedback', { body })