mirror of
https://github.com/langgenius/dify.git
synced 2026-05-06 02:18:08 +08:00
feat(skill-compiler): skill compiler
This commit is contained in:
@ -1,12 +1,7 @@
|
||||
from .entities import (
|
||||
AssetItem,
|
||||
FileAsset,
|
||||
FileReference,
|
||||
SkillAsset,
|
||||
SkillMetadata,
|
||||
ToolConfiguration,
|
||||
ToolFieldConfig,
|
||||
ToolReference,
|
||||
)
|
||||
from .packager import AssetPackager, ZipPackager
|
||||
from .parser import AssetItemParser, AssetParser, FileAssetParser, SkillAssetParser
|
||||
@ -20,12 +15,7 @@ __all__ = [
|
||||
"AssetPaths",
|
||||
"FileAsset",
|
||||
"FileAssetParser",
|
||||
"FileReference",
|
||||
"SkillAsset",
|
||||
"SkillAssetParser",
|
||||
"SkillMetadata",
|
||||
"ToolConfiguration",
|
||||
"ToolFieldConfig",
|
||||
"ToolReference",
|
||||
"ZipPackager",
|
||||
]
|
||||
|
||||
12
api/core/app_assets/builder/__init__.py
Normal file
12
api/core/app_assets/builder/__init__.py
Normal file
@ -0,0 +1,12 @@
|
||||
from .base import AssetBuilder, BuildContext
|
||||
from .file_builder import FileBuilder
|
||||
from .pipeline import AssetBuildPipeline
|
||||
from .skill_builder import SkillBuilder
|
||||
|
||||
__all__ = [
|
||||
"AssetBuildPipeline",
|
||||
"AssetBuilder",
|
||||
"BuildContext",
|
||||
"FileBuilder",
|
||||
"SkillBuilder",
|
||||
]
|
||||
20
api/core/app_assets/builder/base.py
Normal file
20
api/core/app_assets/builder/base.py
Normal file
@ -0,0 +1,20 @@
|
||||
from dataclasses import dataclass
|
||||
from typing import Protocol
|
||||
|
||||
from core.app.entities.app_asset_entities import AppAssetFileTree, AppAssetNode
|
||||
from core.app_assets.entities import AssetItem
|
||||
|
||||
|
||||
@dataclass
|
||||
class BuildContext:
|
||||
tenant_id: str
|
||||
app_id: str
|
||||
build_id: str
|
||||
|
||||
|
||||
class AssetBuilder(Protocol):
|
||||
def accept(self, node: AppAssetNode) -> bool: ...
|
||||
|
||||
def collect(self, node: AppAssetNode, path: str, ctx: BuildContext) -> None: ...
|
||||
|
||||
def build(self, tree: AppAssetFileTree, ctx: BuildContext) -> list[AssetItem]: ...
|
||||
30
api/core/app_assets/builder/file_builder.py
Normal file
30
api/core/app_assets/builder/file_builder.py
Normal file
@ -0,0 +1,30 @@
|
||||
from core.app.entities.app_asset_entities import AppAssetFileTree, AppAssetNode
|
||||
from core.app_assets.entities import AssetItem, FileAsset
|
||||
from core.app_assets.paths import AssetPaths
|
||||
|
||||
from .base import BuildContext
|
||||
|
||||
|
||||
class FileBuilder:
|
||||
_nodes: list[tuple[AppAssetNode, str]]
|
||||
|
||||
def __init__(self) -> None:
|
||||
self._nodes = []
|
||||
|
||||
def accept(self, node: AppAssetNode) -> bool:
|
||||
return True
|
||||
|
||||
def collect(self, node: AppAssetNode, path: str, ctx: BuildContext) -> None:
|
||||
self._nodes.append((node, path))
|
||||
|
||||
def build(self, tree: AppAssetFileTree, ctx: BuildContext) -> list[AssetItem]:
|
||||
return [
|
||||
FileAsset(
|
||||
asset_id=node.id,
|
||||
path=path,
|
||||
file_name=node.name,
|
||||
extension=node.extension or "",
|
||||
storage_key=AssetPaths.draft_file(ctx.tenant_id, ctx.app_id, node.id),
|
||||
)
|
||||
for node, path in self._nodes
|
||||
]
|
||||
29
api/core/app_assets/builder/pipeline.py
Normal file
29
api/core/app_assets/builder/pipeline.py
Normal file
@ -0,0 +1,29 @@
|
||||
from core.app.entities.app_asset_entities import AppAssetFileTree
|
||||
from core.app_assets.builder.file_builder import FileBuilder
|
||||
from core.app_assets.builder.skill_builder import SkillBuilder
|
||||
from core.app_assets.entities import AssetItem
|
||||
|
||||
from .base import AssetBuilder, BuildContext
|
||||
|
||||
|
||||
class AssetBuildPipeline:
|
||||
_builders: list[AssetBuilder]
|
||||
|
||||
def __init__(self, builders: list[AssetBuilder] | None = None) -> None:
|
||||
self._builders = builders or [SkillBuilder(), FileBuilder()]
|
||||
|
||||
def build_all(self, tree: AppAssetFileTree, ctx: BuildContext) -> list[AssetItem]:
|
||||
# 1. Distribute: each node goes to first accepting builder
|
||||
for node in tree.walk_files():
|
||||
path = tree.get_path(node.id)
|
||||
for builder in self._builders:
|
||||
if builder.accept(node):
|
||||
builder.collect(node, path, ctx)
|
||||
break
|
||||
|
||||
# 2. Each builder builds its collected nodes
|
||||
results: list[AssetItem] = []
|
||||
for builder in self._builders:
|
||||
results.extend(builder.build(tree, ctx))
|
||||
|
||||
return results
|
||||
85
api/core/app_assets/builder/skill_builder.py
Normal file
85
api/core/app_assets/builder/skill_builder.py
Normal file
@ -0,0 +1,85 @@
|
||||
import json
|
||||
|
||||
from core.app.entities.app_asset_entities import AppAssetFileTree, AppAssetNode
|
||||
from core.app_assets.entities import AssetItem, FileAsset
|
||||
from core.app_assets.paths import AssetPaths
|
||||
from core.skill.entities.skill_document import SkillDocument
|
||||
from core.skill.skill_compiler import SkillCompiler
|
||||
from core.skill.skill_manager import SkillManager
|
||||
from extensions.ext_storage import storage
|
||||
|
||||
from .base import BuildContext
|
||||
|
||||
|
||||
class SkillBuilder:
|
||||
_nodes: list[tuple[AppAssetNode, str]]
|
||||
|
||||
def __init__(self) -> None:
|
||||
self._nodes = []
|
||||
|
||||
def accept(self, node: AppAssetNode) -> bool:
|
||||
return node.extension == "md"
|
||||
|
||||
def collect(self, node: AppAssetNode, path: str, ctx: BuildContext) -> None:
|
||||
self._nodes.append((node, path))
|
||||
|
||||
def build(self, tree: AppAssetFileTree, ctx: BuildContext) -> list[AssetItem]:
|
||||
if not self._nodes:
|
||||
return []
|
||||
|
||||
# 1. Load and create documents
|
||||
documents: list[SkillDocument] = []
|
||||
for node, _ in self._nodes:
|
||||
draft_key = AssetPaths.draft_file(ctx.tenant_id, ctx.app_id, node.id)
|
||||
try:
|
||||
data = json.loads(storage.load_once(draft_key))
|
||||
content = data.get("content", "") if isinstance(data, dict) else ""
|
||||
metadata = data.get("metadata", {}) if isinstance(data, dict) else {}
|
||||
except Exception:
|
||||
content = ""
|
||||
metadata = {}
|
||||
|
||||
documents.append(
|
||||
SkillDocument(
|
||||
skill_id=node.id,
|
||||
content=content,
|
||||
metadata=metadata,
|
||||
)
|
||||
)
|
||||
|
||||
# 2. Compile all skills
|
||||
compiler = SkillCompiler()
|
||||
artifact_set = compiler.compile_all(documents, tree, ctx.build_id)
|
||||
|
||||
# 3. Save tool artifact
|
||||
SkillManager.save_tool_artifact(
|
||||
ctx.tenant_id,
|
||||
ctx.app_id,
|
||||
ctx.build_id,
|
||||
artifact_set.get_tool_artifact(),
|
||||
)
|
||||
|
||||
# 4. Save compiled content to storage and return FileAssets
|
||||
results: list[AssetItem] = []
|
||||
for node, path in self._nodes:
|
||||
artifact = artifact_set.get(node.id)
|
||||
if artifact is None:
|
||||
continue
|
||||
|
||||
# Write compiled content to storage
|
||||
resolved_key = AssetPaths.build_resolved_file(
|
||||
ctx.tenant_id, ctx.app_id, ctx.build_id, node.id
|
||||
)
|
||||
storage.save(resolved_key, artifact.content.encode("utf-8"))
|
||||
|
||||
results.append(
|
||||
FileAsset(
|
||||
asset_id=node.id,
|
||||
path=path,
|
||||
file_name=node.name,
|
||||
extension=node.extension or "",
|
||||
storage_key=resolved_key,
|
||||
)
|
||||
)
|
||||
|
||||
return results
|
||||
@ -1,20 +1,8 @@
|
||||
from .assets import AssetItem, FileAsset
|
||||
from .skill import (
|
||||
FileReference,
|
||||
SkillAsset,
|
||||
SkillMetadata,
|
||||
ToolConfiguration,
|
||||
ToolFieldConfig,
|
||||
ToolReference,
|
||||
)
|
||||
from .skill import SkillAsset
|
||||
|
||||
__all__ = [
|
||||
"AssetItem",
|
||||
"FileAsset",
|
||||
"FileReference",
|
||||
"SkillAsset",
|
||||
"SkillMetadata",
|
||||
"ToolConfiguration",
|
||||
"ToolFieldConfig",
|
||||
"ToolReference",
|
||||
]
|
||||
|
||||
@ -4,7 +4,7 @@ from dataclasses import dataclass
|
||||
|
||||
@dataclass
|
||||
class AssetItem(ABC):
|
||||
node_id: str
|
||||
asset_id: str
|
||||
path: str
|
||||
file_name: str
|
||||
extension: str
|
||||
|
||||
@ -1,59 +1,14 @@
|
||||
from dataclasses import dataclass
|
||||
from collections.abc import Mapping
|
||||
from dataclasses import dataclass, field
|
||||
from typing import Any
|
||||
|
||||
from pydantic import BaseModel, ConfigDict, Field
|
||||
|
||||
from core.tools.entities.tool_entities import ToolProviderType
|
||||
|
||||
from .assets import AssetItem
|
||||
|
||||
|
||||
class ToolFieldConfig(BaseModel):
|
||||
model_config = ConfigDict(extra="forbid")
|
||||
|
||||
id: str
|
||||
value: Any
|
||||
auto: bool = False
|
||||
|
||||
|
||||
class ToolConfiguration(BaseModel):
|
||||
model_config = ConfigDict(extra="forbid")
|
||||
|
||||
fields: list[ToolFieldConfig] = Field(default_factory=list)
|
||||
|
||||
def default_values(self) -> dict[str, Any]:
|
||||
return {field.id: field.value for field in self.fields if field.value is not None}
|
||||
|
||||
|
||||
class ToolReference(BaseModel):
|
||||
model_config = ConfigDict(extra="forbid")
|
||||
|
||||
uuid: str = Field(description="Unique identifier for this tool reference")
|
||||
type: ToolProviderType = Field(description="Tool provider type")
|
||||
provider: str = Field(description="Tool provider")
|
||||
tool_name: str = Field(description="Tool name")
|
||||
credential_id: str | None = Field(default=None, description="Credential ID")
|
||||
configuration: ToolConfiguration | None = Field(default=None, description="Tool configuration")
|
||||
|
||||
|
||||
class FileReference(BaseModel):
|
||||
model_config = ConfigDict(extra="forbid")
|
||||
|
||||
source: str = Field(description="Source location or identifier of the file")
|
||||
uuid: str = Field(description="Unique identifier for this file reference")
|
||||
|
||||
|
||||
class SkillMetadata(BaseModel):
|
||||
model_config = ConfigDict(extra="allow")
|
||||
|
||||
tools: dict[str, ToolReference] = Field(default_factory=dict, description="Map of tool references by UUID")
|
||||
files: list[FileReference] = Field(default_factory=list, description="List of file references")
|
||||
|
||||
|
||||
@dataclass
|
||||
class SkillAsset(AssetItem):
|
||||
storage_key: str
|
||||
metadata: SkillMetadata
|
||||
metadata: Mapping[str, Any] = field(default_factory=dict)
|
||||
|
||||
def get_storage_key(self) -> str:
|
||||
return self.storage_key
|
||||
|
||||
@ -7,7 +7,7 @@ class AssetItemParser(ABC):
|
||||
@abstractmethod
|
||||
def parse(
|
||||
self,
|
||||
node_id: str,
|
||||
asset_id: str,
|
||||
path: str,
|
||||
file_name: str,
|
||||
extension: str,
|
||||
@ -19,14 +19,14 @@ class AssetItemParser(ABC):
|
||||
class FileAssetParser(AssetItemParser):
|
||||
def parse(
|
||||
self,
|
||||
node_id: str,
|
||||
asset_id: str,
|
||||
path: str,
|
||||
file_name: str,
|
||||
extension: str,
|
||||
storage_key: str,
|
||||
) -> FileAsset:
|
||||
return FileAsset(
|
||||
node_id=node_id,
|
||||
asset_id=asset_id,
|
||||
path=path,
|
||||
file_name=file_name,
|
||||
extension=extension,
|
||||
|
||||
@ -1,161 +1,57 @@
|
||||
import json
|
||||
import logging
|
||||
import re
|
||||
from typing import Any
|
||||
|
||||
from core.app.entities.app_asset_entities import AppAssetFileTree, AppAssetNode
|
||||
from core.app_assets.entities import (
|
||||
SkillAsset,
|
||||
SkillMetadata,
|
||||
)
|
||||
from core.app_assets.entities.skill import FileReference, ToolConfiguration, ToolReference
|
||||
from core.app_assets.paths import AssetPaths
|
||||
from core.tools.entities.tool_entities import ToolProviderType
|
||||
from core.app_assets.entities import SkillAsset
|
||||
from core.app_assets.entities.assets import AssetItem, FileAsset
|
||||
from extensions.ext_storage import storage
|
||||
|
||||
from .base import AssetItemParser
|
||||
|
||||
TOOL_REFERENCE_PATTERN = re.compile(r"§\[tool\]\.\[([^\]]+)\]\.\[([^\]]+)\]\.\[([^\]]+)\]§")
|
||||
FILE_REFERENCE_PATTERN = re.compile(r"§\[file\]\.\[([^\]]+)\]\.\[([^\]]+)\]§")
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
class SkillAssetParser(AssetItemParser):
|
||||
def __init__(
|
||||
self,
|
||||
tenant_id: str,
|
||||
app_id: str,
|
||||
assets_id: str,
|
||||
tree: AppAssetFileTree,
|
||||
) -> None:
|
||||
self._tenant_id = tenant_id
|
||||
self._app_id = app_id
|
||||
self._assets_id = assets_id
|
||||
self._tree = tree
|
||||
"""
|
||||
Parser for skill assets.
|
||||
|
||||
Responsibilities:
|
||||
- Read file from storage
|
||||
- Parse JSON structure
|
||||
- Return SkillAsset with raw metadata (no parsing/resolution)
|
||||
|
||||
Metadata parsing and content resolution are handled by SkillCompiler.
|
||||
"""
|
||||
|
||||
def parse(
|
||||
self,
|
||||
node_id: str,
|
||||
asset_id: str,
|
||||
path: str,
|
||||
file_name: str,
|
||||
extension: str,
|
||||
storage_key: str,
|
||||
) -> SkillAsset:
|
||||
try:
|
||||
return self._parse_skill_asset(node_id, path, file_name, extension, storage_key)
|
||||
except Exception:
|
||||
logger.exception("Failed to parse skill asset %s", node_id)
|
||||
# handle as plain text
|
||||
return SkillAsset(
|
||||
node_id=node_id,
|
||||
path=path,
|
||||
file_name=file_name,
|
||||
extension=extension,
|
||||
storage_key=storage_key,
|
||||
metadata=SkillMetadata(),
|
||||
)
|
||||
|
||||
def _parse_skill_asset(
|
||||
self, node_id: str, path: str, file_name: str, extension: str, storage_key: str
|
||||
) -> SkillAsset:
|
||||
) -> AssetItem:
|
||||
try:
|
||||
data = json.loads(storage.load_once(storage_key))
|
||||
except (json.JSONDecodeError, UnicodeDecodeError):
|
||||
# handle as plain text
|
||||
if not isinstance(data, dict):
|
||||
raise ValueError(f"Skill document {asset_id} must be a JSON object")
|
||||
|
||||
metadata_raw: dict[str, Any] = data.get("metadata", {})
|
||||
|
||||
return SkillAsset(
|
||||
node_id=node_id,
|
||||
asset_id=asset_id,
|
||||
path=path,
|
||||
file_name=file_name,
|
||||
extension=extension,
|
||||
storage_key=storage_key,
|
||||
metadata=SkillMetadata(),
|
||||
metadata=metadata_raw,
|
||||
)
|
||||
|
||||
if not isinstance(data, dict):
|
||||
raise ValueError(f"Skill document {node_id} must be a JSON object")
|
||||
|
||||
data_dict: dict[str, Any] = data
|
||||
metadata_raw = data_dict.get("metadata", {})
|
||||
content = data_dict.get("content", "")
|
||||
|
||||
if not isinstance(content, str):
|
||||
raise ValueError(f"Skill document {node_id} 'content' must be a string")
|
||||
|
||||
resolved_key = AssetPaths.build_resolved_file(self._tenant_id, self._app_id, self._assets_id, node_id)
|
||||
current_file = self._tree.get(node_id)
|
||||
if current_file is None:
|
||||
raise ValueError(f"File not found for id={node_id}")
|
||||
|
||||
metadata = self._resolve_metadata(content, metadata_raw)
|
||||
storage.save(resolved_key, self._resolve_content(current_file, content, metadata).encode("utf-8"))
|
||||
|
||||
return SkillAsset(
|
||||
node_id=node_id,
|
||||
path=path,
|
||||
file_name=file_name,
|
||||
extension=extension,
|
||||
storage_key=resolved_key,
|
||||
metadata=metadata,
|
||||
)
|
||||
|
||||
def _resolve_content(self, current_file: AppAssetNode, content: str, metadata: SkillMetadata) -> str:
|
||||
for match in FILE_REFERENCE_PATTERN.finditer(content):
|
||||
# replace with file relative path
|
||||
file_id = match.group(2)
|
||||
file = self._tree.get(file_id)
|
||||
if file is None:
|
||||
logger.warning("File not found for id=%s, skipping", file_id)
|
||||
# replace with file not found placeholder
|
||||
content = content.replace(match.group(0), "[File not found]")
|
||||
continue
|
||||
content = content.replace(match.group(0), self._tree.relative_path(current_file, file))
|
||||
|
||||
for match in TOOL_REFERENCE_PATTERN.finditer(content):
|
||||
tool_id = match.group(3)
|
||||
tool = metadata.tools.get(tool_id)
|
||||
if tool is None:
|
||||
logger.warning("Tool not found for id=%s, skipping", tool_id)
|
||||
# replace with tool not found placeholder
|
||||
content = content.replace(match.group(0), f"[Tool not found: {tool_id}]")
|
||||
continue
|
||||
content = content.replace(match.group(0), f"[Bash Command: {tool.tool_name}_{tool_id}]")
|
||||
return content
|
||||
|
||||
def _resolve_file_references(self, content: str) -> list[FileReference]:
|
||||
file_references: list[FileReference] = []
|
||||
for match in FILE_REFERENCE_PATTERN.finditer(content):
|
||||
file_references.append(
|
||||
FileReference(
|
||||
source=match.group(1),
|
||||
uuid=match.group(2),
|
||||
)
|
||||
except Exception:
|
||||
logger.exception("Failed to parse skill asset %s", asset_id)
|
||||
return FileAsset(
|
||||
asset_id=asset_id,
|
||||
path=path,
|
||||
file_name=file_name,
|
||||
extension=extension,
|
||||
storage_key=storage_key,
|
||||
)
|
||||
return file_references
|
||||
|
||||
def _resolve_tool_references(self, content: str, tools: dict[str, Any]) -> dict[str, ToolReference]:
|
||||
tool_references: dict[str, ToolReference] = {}
|
||||
for match in TOOL_REFERENCE_PATTERN.finditer(content):
|
||||
tool_id = match.group(3)
|
||||
tool_name = match.group(2)
|
||||
tool_provider = match.group(1)
|
||||
metadata = tools.get(tool_id)
|
||||
if metadata is None:
|
||||
raise ValueError(f"Tool metadata for {tool_id} not found")
|
||||
|
||||
configuration = ToolConfiguration.model_validate(metadata.get("configuration", {}))
|
||||
tool_references[tool_id] = ToolReference(
|
||||
uuid=tool_id,
|
||||
type=ToolProviderType.value_of(metadata.get("type", None)),
|
||||
provider=tool_provider,
|
||||
tool_name=tool_name,
|
||||
credential_id=metadata.get("credential_id", None),
|
||||
configuration=configuration,
|
||||
)
|
||||
return tool_references
|
||||
|
||||
def _resolve_metadata(self, content: str, metadata: dict[str, Any]) -> SkillMetadata:
|
||||
return SkillMetadata(
|
||||
files=self._resolve_file_references(content=content),
|
||||
tools=self._resolve_tool_references(content=content, tools=metadata.get("tools", {})),
|
||||
)
|
||||
|
||||
@ -16,3 +16,7 @@ class AssetPaths:
|
||||
@staticmethod
|
||||
def build_tool_artifact(tenant_id: str, app_id: str, assets_id: str) -> str:
|
||||
return f"{AssetPaths._BASE}/{tenant_id}/{app_id}/artifacts/{assets_id}/tool_artifact.json"
|
||||
|
||||
@staticmethod
|
||||
def build_skill_artifact_set(tenant_id: str, app_id: str, assets_id: str) -> str:
|
||||
return f"{AssetPaths._BASE}/{tenant_id}/{app_id}/artifacts/{assets_id}/skill_artifact_set.json"
|
||||
|
||||
Reference in New Issue
Block a user