admin/dify
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

feat: knowledge pipeline (#25360)

Browse Source 
Signed-off-by: -LAN- <laipz8200@outlook.com>
Co-authored-by: twwu <twwu@dify.ai>
Co-authored-by: crazywoola <100913391+crazywoola@users.noreply.github.com>
Co-authored-by: jyong <718720800@qq.com>
Co-authored-by: Wu Tianwei <30284043+WTW0313@users.noreply.github.com>
Co-authored-by: QuantumGhost <obelisk.reg+git@gmail.com>
Co-authored-by: lyzno1 <yuanyouhuilyz@gmail.com>
Co-authored-by: quicksand <quicksandzn@gmail.com>
Co-authored-by: Jyong <76649700+JohnJyong@users.noreply.github.com>
Co-authored-by: lyzno1 <92089059+lyzno1@users.noreply.github.com>
Co-authored-by: zxhlyh <jasonapring2015@outlook.com>
Co-authored-by: Yongtao Huang <yongtaoh2022@gmail.com>
Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>
Co-authored-by: Joel <iamjoel007@gmail.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: nite-knite <nkCoding@gmail.com>
Co-authored-by: Hanqing Zhao <sherry9277@gmail.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Harry <xh001x@hotmail.com>
This commit is contained in:
-LAN-
2025-09-18 12:49:10 +08:00
committed by GitHub
parent 7dadb33003
commit 85cda47c70
1772 changed files with 102407 additions and 31710 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
api/core/workflow/nodes/llm/entities.py
View File

@@ -5,8 +5,8 @@ from pydantic import BaseModel, Field, field_validator
from core.model_runtime.entities import ImagePromptMessageContent, LLMMode
from core.prompt.entities.advanced_prompt_entities import ChatModelMessage, CompletionModelPromptTemplate, MemoryConfig
from core.workflow.entities.variable_entities import VariableSelector
from core.workflow.nodes.base import BaseNodeData
from core.workflow.nodes.base.entities import VariableSelector
class ModelConfig(BaseModel):

2
api/core/workflow/nodes/llm/file_saver.py
View File

@@ -8,7 +8,7 @@ from core.file import File, FileTransferMethod, FileType
from core.helper import ssrf_proxy
from core.tools.signature import sign_tool_file
from core.tools.tool_file_manager import ToolFileManager
from models import db as global_db
from extensions.ext_database import db as global_db
class LLMFileSaver(tp.Protocol):

6
api/core/workflow/nodes/llm/llm_utils.py
View File

@@ -13,16 +13,16 @@ from core.model_manager import ModelInstance, ModelManager
from core.model_runtime.entities.llm_entities import LLMUsage
from core.model_runtime.entities.model_entities import ModelType
from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
from core.plugin.entities.plugin import ModelProviderID
from core.prompt.entities.advanced_prompt_entities import MemoryConfig
from core.variables.segments import ArrayAnySegment, ArrayFileSegment, FileSegment, NoneSegment, StringSegment
from core.workflow.entities.variable_pool import VariablePool
from core.workflow.entities import VariablePool
from core.workflow.enums import SystemVariableKey
from core.workflow.nodes.llm.entities import ModelConfig
from extensions.ext_database import db
from libs.datetime_utils import naive_utc_now
from models import db
from models.model import Conversation
from models.provider import Provider, ProviderType
from models.provider_ids import ModelProviderID
from .exc import InvalidVariableTypeError, LLMModeRequiredError, ModelNotExistError

117
api/core/workflow/nodes/llm/node.py
View File

@@ -4,7 +4,7 @@ import json
import logging
import re
from collections.abc import Generator, Mapping, Sequence
from typing import TYPE_CHECKING, Any, Literal, Union
from typing import TYPE_CHECKING, Any, Literal
from core.app.entities.app_invoke_entities import ModelConfigWithCredentialsEntity
from core.file import FileType, file_manager
@@ -51,22 +51,25 @@ from core.variables import (
StringSegment,
)
from core.workflow.constants import SYSTEM_VARIABLE_NODE_ID
from core.workflow.entities.node_entities import NodeRunResult
from core.workflow.entities.variable_entities import VariableSelector
from core.workflow.entities.variable_pool import VariablePool
from core.workflow.entities.workflow_node_execution import WorkflowNodeExecutionMetadataKey, WorkflowNodeExecutionStatus
from core.workflow.enums import SystemVariableKey
from core.workflow.nodes.base import BaseNode
from core.workflow.nodes.base.entities import BaseNodeData, RetryConfig
from core.workflow.nodes.enums import ErrorStrategy, NodeType
from core.workflow.nodes.event import (
ModelInvokeCompletedEvent,
NodeEvent,
RunCompletedEvent,
RunRetrieverResourceEvent,
RunStreamChunkEvent,
from core.workflow.entities import GraphInitParams, VariablePool
from core.workflow.enums import (
ErrorStrategy,
NodeType,
SystemVariableKey,
WorkflowNodeExecutionMetadataKey,
WorkflowNodeExecutionStatus,
)
from core.workflow.utils.variable_template_parser import VariableTemplateParser
from core.workflow.node_events import (
ModelInvokeCompletedEvent,
NodeEventBase,
NodeRunResult,
RunRetrieverResourceEvent,
StreamChunkEvent,
StreamCompletedEvent,
)
from core.workflow.nodes.base.entities import BaseNodeData, RetryConfig, VariableSelector
from core.workflow.nodes.base.node import Node
from core.workflow.nodes.base.variable_template_parser import VariableTemplateParser
from . import llm_utils
from .entities import (
@@ -89,14 +92,13 @@ from .file_saver import FileSaverImpl, LLMFileSaver
if TYPE_CHECKING:
from core.file.models import File
from core.workflow.graph_engine import Graph, GraphInitParams, GraphRuntimeState
from core.workflow.graph_engine.entities.event import InNodeEvent
from core.workflow.entities import GraphRuntimeState
logger = logging.getLogger(__name__)
class LLMNode(BaseNode):
_node_type = NodeType.LLM
class LLMNode(Node):
node_type = NodeType.LLM
_node_data: LLMNodeData
@@ -114,10 +116,7 @@ class LLMNode(BaseNode):
id: str,
config: Mapping[str, Any],
graph_init_params: "GraphInitParams",
graph: "Graph",
graph_runtime_state: "GraphRuntimeState",
previous_node_id: str | None = None,
thread_pool_id: str | None = None,
*,
llm_file_saver: LLMFileSaver | None = None,
):
@@ -125,10 +124,7 @@ class LLMNode(BaseNode):
id=id,
config=config,
graph_init_params=graph_init_params,
graph=graph,
graph_runtime_state=graph_runtime_state,
previous_node_id=previous_node_id,
thread_pool_id=thread_pool_id,
)
# LLM file outputs, used for MultiModal outputs.
self._file_outputs: list[File] = []
@@ -165,9 +161,9 @@ class LLMNode(BaseNode):
def version(cls) -> str:
return "1"
def _run(self) -> Generator[Union[NodeEvent, "InNodeEvent"], None, None]:
node_inputs: dict[str, Any] | None = None
process_data = None
def _run(self) -> Generator:
node_inputs: dict[str, Any] = {}
process_data: dict[str, Any] = {}
result_text = ""
usage = LLMUsage.empty_usage()
finish_reason = None
@@ -187,8 +183,6 @@ class LLMNode(BaseNode):
# merge inputs
inputs.update(jinja_inputs)
node_inputs = {}
# fetch files
files = (
llm_utils.fetch_files(
@@ -206,9 +200,8 @@ class LLMNode(BaseNode):
generator = self._fetch_context(node_data=self._node_data)
context = None
for event in generator:
if isinstance(event, RunRetrieverResourceEvent):
context = event.context
yield event
context = event.context
yield event
if context:
node_inputs["#context#"] = context
@@ -260,14 +253,15 @@ class LLMNode(BaseNode):
structured_output=self._node_data.structured_output,
file_saver=self._llm_file_saver,
file_outputs=self._file_outputs,
node_id=self.node_id,
node_id=self._node_id,
node_type=self.node_type,
reasoning_format=self._node_data.reasoning_format,
)
structured_output: LLMStructuredOutput | None = None
for event in generator:
if isinstance(event, RunStreamChunkEvent):
if isinstance(event, StreamChunkEvent):
yield event
elif isinstance(event, ModelInvokeCompletedEvent):
# Raw text
@@ -309,11 +303,18 @@ class LLMNode(BaseNode):
}
if structured_output:
outputs["structured_output"] = structured_output.structured_output
if self._file_outputs is not None:
if self._file_outputs:
outputs["files"] = ArrayFileSegment(value=self._file_outputs)
yield RunCompletedEvent(
run_result=NodeRunResult(
# Send final chunk event to indicate streaming is complete
yield StreamChunkEvent(
selector=[self._node_id, "text"],
chunk="",
is_final=True,
)
yield StreamCompletedEvent(
node_run_result=NodeRunResult(
status=WorkflowNodeExecutionStatus.SUCCEEDED,
inputs=node_inputs,
process_data=process_data,
@@ -327,8 +328,8 @@ class LLMNode(BaseNode):
)
)
except ValueError as e:
yield RunCompletedEvent(
run_result=NodeRunResult(
yield StreamCompletedEvent(
node_run_result=NodeRunResult(
status=WorkflowNodeExecutionStatus.FAILED,
error=str(e),
inputs=node_inputs,
@@ -338,8 +339,8 @@ class LLMNode(BaseNode):
)
except Exception as e:
logger.exception("error while executing llm node")
yield RunCompletedEvent(
run_result=NodeRunResult(
yield StreamCompletedEvent(
node_run_result=NodeRunResult(
status=WorkflowNodeExecutionStatus.FAILED,
error=str(e),
inputs=node_inputs,
@@ -360,8 +361,9 @@ class LLMNode(BaseNode):
file_saver: LLMFileSaver,
file_outputs: list["File"],
node_id: str,
node_type: NodeType,
reasoning_format: Literal["separated", "tagged"] = "tagged",
) -> Generator[NodeEvent | LLMStructuredOutput, None, None]:
) -> Generator[NodeEventBase | LLMStructuredOutput, None, None]:
model_schema = model_instance.model_type_instance.get_model_schema(
node_data_model.name, model_instance.credentials
)
@@ -397,6 +399,7 @@ class LLMNode(BaseNode):
file_saver=file_saver,
file_outputs=file_outputs,
node_id=node_id,
node_type=node_type,
reasoning_format=reasoning_format,
)
@@ -407,8 +410,9 @@ class LLMNode(BaseNode):
file_saver: LLMFileSaver,
file_outputs: list["File"],
node_id: str,
node_type: NodeType,
reasoning_format: Literal["separated", "tagged"] = "tagged",
) -> Generator[NodeEvent | LLMStructuredOutput, None, None]:
) -> Generator[NodeEventBase | LLMStructuredOutput, None, None]:
# For blocking mode
if isinstance(invoke_result, LLMResult):
event = LLMNode.handle_blocking_result(
@@ -440,7 +444,11 @@ class LLMNode(BaseNode):
file_outputs=file_outputs,
):
full_text_buffer.write(text_part)
yield RunStreamChunkEvent(chunk_content=text_part, from_variable_selector=[node_id, "text"])
yield StreamChunkEvent(
selector=[node_id, "text"],
chunk=text_part,
is_final=False,
)
# Update the whole metadata
if not model and result.model:
@@ -890,14 +898,14 @@ class LLMNode(BaseNode):
node_id: str,
node_data: Mapping[str, Any],
) -> Mapping[str, Sequence[str]]:
# graph_config is not used in this node type
_ = graph_config # Explicitly mark as unused
# Create typed NodeData from dict
typed_node_data = LLMNodeData.model_validate(node_data)
prompt_template = typed_node_data.prompt_template
variable_selectors = []
if isinstance(prompt_template, list) and all(
isinstance(prompt, LLMNodeChatModelMessage) for prompt in prompt_template
):
if isinstance(prompt_template, list):
for prompt in prompt_template:
if prompt.edition_type != "jinja2":
variable_template_parser = VariableTemplateParser(template=prompt.text)
@@ -951,7 +959,7 @@ class LLMNode(BaseNode):
return variable_mapping
@classmethod
def get_default_config(cls, filters: dict | None = None):
def get_default_config(cls, filters: Mapping[str, object] | None = None) -> Mapping[str, object]:
return {
"type": "llm",
"config": {
@@ -1146,7 +1154,7 @@ class LLMNode(BaseNode):
return
if isinstance(contents, str):
yield contents
elif isinstance(contents, list):
else:
for item in contents:
if isinstance(item, TextPromptMessageContent):
yield item.data
@@ -1160,13 +1168,6 @@ class LLMNode(BaseNode):
else:
logger.warning("unknown item type encountered, type=%s", type(item))
yield str(item)
else:
logger.warning("unknown contents type encountered, type=%s", type(contents))
yield str(contents)
@property
def continue_on_error(self) -> bool:
return self._node_data.error_strategy is not None
@property
def retry(self) -> bool: