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

feat(api): Add image multimodal support for LLMNode (#17372)

Browse Source 
Enhance `LLMNode` with multimodal capability, introducing support for
image outputs.

This implementation extracts base64-encoded images from LLM responses,
saves them to the storage service, and records the file metadata in the
`ToolFile` table. In conversations, these images are rendered as
markdown-based inline images.
Additionally, the images are included in the LLMNode's output as
file variables, enabling subsequent nodes in the workflow to utilize them.

To integrate file outputs into workflows, adjustments to the frontend code
are necessary.

For multimodal output functionality, updates to related model configurations
are required. Currently, this capability has been applied exclusively to
Google's Gemini models.

Close #15814.

Signed-off-by: -LAN- <laipz8200@outlook.com>
Co-authored-by: -LAN- <laipz8200@outlook.com>
This commit is contained in:
QuantumGhost
2025-04-30 17:28:02 +08:00
committed by GitHub
parent 6c9a9d344a
commit 349c3cf7b8
24 changed files with 971 additions and 191 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
api/core/file/models.py
View File

@@ -4,11 +4,11 @@ from typing import Any, Optional
from pydantic import BaseModel, Field, model_validator
from core.model_runtime.entities.message_entities import ImagePromptMessageContent
from core.tools.signature import sign_tool_file
from . import helpers
from .constants import FILE_MODEL_IDENTITY
from .enums import FileTransferMethod, FileType
from .tool_file_parser import ToolFileParser
class ImageConfig(BaseModel):
@@ -34,13 +34,21 @@ class FileUploadConfig(BaseModel):
class File(BaseModel):
# NOTE: dify_model_identity is a special identifier used to distinguish between
# new and old data formats during serialization and deserialization.
dify_model_identity: str = FILE_MODEL_IDENTITY
id: Optional[str] = None # message file id
tenant_id: str
type: FileType
transfer_method: FileTransferMethod
# If `transfer_method` is `FileTransferMethod.remote_url`, the
# `remote_url` attribute must not be `None`.
remote_url: Optional[str] = None # remote url
# If `transfer_method` is `FileTransferMethod.local_file` or
# `FileTransferMethod.tool_file`, the `related_id` attribute must not be `None`.
#
# It should be set to `ToolFile.id` when `transfer_method` is `tool_file`.
related_id: Optional[str] = None
filename: Optional[str] = None
extension: Optional[str] = Field(default=None, description="File extension, should contains dot")
@@ -110,9 +118,7 @@ class File(BaseModel):
elif self.transfer_method == FileTransferMethod.TOOL_FILE:
assert self.related_id is not None
assert self.extension is not None
return ToolFileParser.get_tool_file_manager().sign_file(
tool_file_id=self.related_id, extension=self.extension
)
return sign_tool_file(tool_file_id=self.related_id, extension=self.extension)
def to_plugin_parameter(self) -> dict[str, Any]:
return {