q792602257/DataMate
0
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

feat(annotation): 替换模板配置表单为树形编辑器组件

Browse Source 
- 移除 TemplateConfigurationForm 组件并引入 TemplateConfigurationTreeEditor
- 使用 useTagConfig Hook 获取标签配置
- 将自定义XML状态 customXml 替换为 labelConfig
- 删除模板编辑标签页和选择模板状态管理
- 更新XML解析逻辑支持更多对象和标注控件类型
- 添加配置验证功能确保至少包含数据对象和标注控件
- 在模板详情页面使用树形编辑器显示配置详情
- 更新任务创建页面集成新的树形配置编辑器
- 调整预览数据生成功能适配新的XML解析方式
This commit is contained in:
Jerry Yan
2026-01-23 16:11:59 +08:00
parent 76d06b9809
commit 3f566a0b08
14 changed files with 1383 additions and 900 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
runtime/datamate-python/app/db/models/annotation_management.py
View File

@@ -16,8 +16,8 @@ class AnnotationTemplate(Base):
description = Column(String(500), nullable=True, comment="模板描述")
data_type = Column(String(50), nullable=False, comment="数据类型: image/text/audio/video/timeseries/pdf/chat/html/table")
labeling_type = Column(String(50), nullable=False, comment="标注类型: asr/ner/object-detection/等")
configuration = Column(JSON, nullable=False, comment="标注配置（包含labels定义等）")
label_config = Column(Text, nullable=True, comment="Label Studio XML配置（内置模板预定义，自定义模板自动生成）")
configuration = Column(JSON, nullable=True, comment="标注配置（兼容字段，主配置为 label_config）")
label_config = Column(Text, nullable=True, comment="Label Studio XML配置（模板主配置）")
style = Column(String(32), nullable=False, comment="样式配置: horizontal/vertical")
category = Column(String(50), default='custom', comment="模板分类: audio-speech/chat/computer-vision/nlp/等")
built_in = Column(Boolean, default=False, comment="是否系统内置模板")

64
runtime/datamate-python/app/module/annotation/config/label_studio_tags.yaml
View File

@@ -23,11 +23,6 @@ objects:
required_attrs: [name, value]
optional_attrs: []
category: document
ParagraphLabels:
description: "Display paragraphs with label support"
required_attrs: [name, value]
optional_attrs: []
category: text
Timeseries:
description: "Display timeseries data"
required_attrs: [name, value]
@@ -118,7 +113,7 @@ controls:
default: 3
description: "Maximum depth of taxonomy tree"
requires_children: true
child_tag: Path
child_tag: Choice
child_required_attrs: [value]
category: labeling
@@ -135,7 +130,7 @@ controls:
requires_children: true
child_tag: Choice
child_required_attrs: [value]
category: layout
category: labeling
List:
description: "List selection control"
@@ -150,11 +145,11 @@ controls:
requires_children: true
child_tag: Item
child_required_attrs: [value]
category: layout
category: labeling
Filter:
description: "Filter control for annotation"
required_attrs: [name, toName]
required_attrs: []
optional_attrs:
required:
type: boolean
@@ -163,7 +158,7 @@ controls:
Collapse:
description: "Collapsible UI section"
required_attrs: [name]
required_attrs: []
optional_attrs:
collapsed:
type: boolean
@@ -173,18 +168,18 @@ controls:
Header:
description: "Section header for UI grouping"
required_attrs: [name]
required_attrs: [value]
optional_attrs:
level:
size:
type: number
default: 1
description: "Header level (1-6)"
default: 3
description: "Header size"
requires_children: false
category: layout
Shortcut:
description: "Keyboard shortcut definition"
required_attrs: [name, toName]
required_attrs: []
optional_attrs:
key:
type: string
@@ -194,11 +189,8 @@ controls:
Style:
description: "Custom style for annotation UI"
required_attrs: [name]
optional_attrs:
value:
type: string
description: "CSS style value"
required_attrs: []
optional_attrs: {}
requires_children: false
category: layout
@@ -247,23 +239,14 @@ controls:
child_required_attrs: [value]
category: labeling
Relation:
description: "Draw relation between objects"
required_attrs: [name, toName]
optional_attrs:
required:
type: boolean
requires_children: false
category: layout
Relations:
description: "Draw multiple relations between objects"
required_attrs: [name, toName]
optional_attrs:
required:
type: boolean
requires_children: false
category: layout
required_attrs: []
optional_attrs: {}
requires_children: true
child_tag: Relation
child_required_attrs: [value]
category: labeling
Pairwise:
description: "Pairwise comparison control"
@@ -272,7 +255,7 @@ controls:
required:
type: boolean
requires_children: false
category: layout
category: labeling
DateTime:
description: "Date and time input"
@@ -350,6 +333,15 @@ controls:
child_required_attrs: [value]
category: labeling
HyperTextLabels:
description: "Labels for hypertext entities"
required_attrs: [name, toName]
optional_attrs: [required]
requires_children: true
child_tag: Label
child_required_attrs: [value]
category: labeling
KeyPointLabels:
description: "Keypoint annotations with labels"
required_attrs: [name, toName]

2
runtime/datamate-python/app/module/annotation/interface/template.py
View File

@@ -35,7 +35,7 @@ async def create_template(
- **description**: 模板描述（可选，最多500字符）
- **dataType**: 数据类型（必填）
- **labelingType**: 标注类型（必填）
- **configuration**: 标注配置（必填，包含labels和objects）
- **labelConfig**: Label Studio XML 配置（必填）
- **style**: 样式配置（默认horizontal）
- **category**: 模板分类（默认custom）
"""

8
runtime/datamate-python/app/module/annotation/schema/template.py
View File

@@ -45,7 +45,7 @@ class CreateAnnotationTemplateRequest(BaseModel):
description: Optional[str] = Field(None, max_length=500, description="模板描述")
data_type: str = Field(alias="dataType", description="数据类型")
labeling_type: str = Field(alias="labelingType", description="标注类型")
configuration: TemplateConfiguration = Field(..., description="标注配置")
label_config: str = Field(alias="labelConfig", description="Label Studio XML 配置")
style: str = Field(default="horizontal", description="样式配置")
category: str = Field(default="custom", description="模板分类")
@@ -58,7 +58,7 @@ class UpdateAnnotationTemplateRequest(BaseModel):
description: Optional[str] = Field(None, max_length=500, description="模板描述")
data_type: Optional[str] = Field(None, alias="dataType", description="数据类型")
labeling_type: Optional[str] = Field(None, alias="labelingType", description="标注类型")
configuration: Optional[TemplateConfiguration] = Field(None, description="标注配置")
label_config: Optional[str] = Field(None, alias="labelConfig", description="Label Studio XML 配置")
style: Optional[str] = Field(None, description="样式配置")
category: Optional[str] = Field(None, description="模板分类")
@@ -72,8 +72,8 @@ class AnnotationTemplateResponse(BaseModel):
description: Optional[str] = Field(None, description="模板描述")
data_type: str = Field(alias="dataType", description="数据类型")
labeling_type: str = Field(alias="labelingType", description="标注类型")
configuration: TemplateConfiguration = Field(..., description="标注配置")
label_config: Optional[str] = Field(None, alias="labelConfig", description="生成的Label Studio XML配置")
configuration: Optional[TemplateConfiguration] = Field(None, description="标注配置")
label_config: Optional[str] = Field(None, alias="labelConfig", description="Label Studio XML配置")
style: str = Field(..., description="样式配置")
category: str = Field(..., description="模板分类")
built_in: bool = Field(alias="builtIn", description="是否内置模板")

62
runtime/datamate-python/app/module/annotation/service/template.py
View File

@@ -114,28 +114,20 @@ class AnnotationTemplateService:
Returns:
创建的模板响应
"""
# 验证配置JSON
config_dict = request.configuration.model_dump(mode='json', by_alias=False)
valid, error = LabelStudioConfigValidator.validate_configuration_json(config_dict)
if not valid:
raise HTTPException(status_code=400, detail=f"Invalid configuration: {error}")
# 生成Label Studio XML配置（用于验证，但不存储）
label_config = self.generate_label_studio_config(request.configuration)
# 验证生成的XML
label_config = request.label_config
valid, error = LabelStudioConfigValidator.validate_xml(label_config)
if not valid:
raise HTTPException(status_code=400, detail=f"Generated XML is invalid: {error}")
# 创建模板对象（不包含label_config字段）
raise HTTPException(status_code=400, detail=f"Invalid labelConfig: {error}")
# 创建模板对象
template = AnnotationTemplate(
id=str(uuid4()),
name=request.name,
description=request.description,
data_type=request.data_type,
labeling_type=request.labeling_type,
configuration=config_dict,
configuration=None,
label_config=label_config,
style=request.style,
category=request.category,
built_in=False,
@@ -280,24 +272,11 @@ class AnnotationTemplateService:
update_data = request.model_dump(exclude_unset=True, by_alias=False)
for field, value in update_data.items():
if field == 'configuration' and value is not None:
# 验证配置JSON
config = value if isinstance(value, TemplateConfiguration) else TemplateConfiguration.model_validate(value)
config_dict = config.model_dump(mode='json', by_alias=False)
valid, error = LabelStudioConfigValidator.validate_configuration_json(config_dict)
if field == "label_config" and value is not None:
valid, error = LabelStudioConfigValidator.validate_xml(value)
if not valid:
raise HTTPException(status_code=400, detail=f"Invalid configuration: {error}")
# 重新生成Label Studio XML配置（用于验证）
label_config = self.generate_label_studio_config(config)
# 验证生成的XML
valid, error = LabelStudioConfigValidator.validate_xml(label_config)
if not valid:
raise HTTPException(status_code=400, detail=f"Generated XML is invalid: {error}")
# 只更新configuration字段，不存储label_config
setattr(template, field, config_dict)
raise HTTPException(status_code=400, detail=f"Invalid labelConfig: {error}")
setattr(template, field, value)
else:
setattr(template, field, value)
@@ -350,20 +329,17 @@ class AnnotationTemplateService:
Returns:
模板响应对象
"""
# 将配置JSON转换为TemplateConfiguration对象
from typing import cast, Dict, Any
config_dict = cast(Dict[str, Any], template.configuration)
config = TemplateConfiguration(**config_dict)
config = None
if template.configuration:
try:
from typing import cast, Dict, Any
config_dict = cast(Dict[str, Any], template.configuration)
config = TemplateConfiguration(**config_dict)
except Exception:
config = None
# 优先使用预定义的 label_config，否则动态生成
if template.label_config:
label_config = template.label_config
else:
label_config = self.generate_label_studio_config(config)
# 使用model_validate从ORM对象创建响应对象
response = AnnotationTemplateResponse.model_validate(template)
response.configuration = config
response.label_config = label_config # type: ignore
response.label_config = template.label_config # type: ignore
return response

111
runtime/datamate-python/app/module/annotation/utils/config_validator.py
View File

@@ -1,7 +1,7 @@
"""
Label Studio Configuration Validation Utilities
"""
from typing import Dict, List, Tuple, Optional
from typing import Dict, List, Tuple, Optional, Set
import xml.etree.ElementTree as ET
from app.module.annotation.config import LabelStudioTagConfig
@@ -13,6 +13,23 @@ class LabelStudioConfigValidator:
def _get_config() -> LabelStudioTagConfig:
"""获取标签配置实例"""
return LabelStudioTagConfig()
@staticmethod
def _get_required_attrs(tag_type: str, is_control: bool) -> List[str]:
config = LabelStudioConfigValidator._get_config()
tag_config = (
config.get_control_config(tag_type)
if is_control
else config.get_object_config(tag_type)
)
required_attrs = tag_config.get("required_attrs", []) if tag_config else []
return required_attrs if isinstance(required_attrs, list) else []
@staticmethod
def _get_control_category(tag_type: str) -> Optional[str]:
config = LabelStudioConfigValidator._get_config()
tag_config = config.get_control_config(tag_type) or {}
return tag_config.get("category")
@staticmethod
def validate_xml(xml_string: str) -> Tuple[bool, Optional[str]]:
@@ -33,24 +50,49 @@ class LabelStudioConfigValidator:
if root.tag != 'View':
return False, "Root element must be <View>"
# 检查是否有对象定义
object_types = config.get_object_types()
objects = [elem for elem in root.iter() if elem is not root and elem.tag in object_types]
control_types = config.get_control_types()
objects = [elem for elem in root.iter() if elem.tag in object_types]
controls = [elem for elem in root.iter() if elem.tag in control_types]
labeling_controls = [
control
for control in controls
if LabelStudioConfigValidator._get_control_category(control.tag) == "labeling"
]
if not objects:
return False, "No data objects (Image, Text, etc.) found"
# 检查是否有控件定义
control_types = config.get_control_types()
controls = [elem for elem in root.iter() if elem is not root and elem.tag in control_types]
if not controls:
return False, "No annotation controls found"
# 验证每个控件
if not labeling_controls:
return False, "No labeling controls found"
object_names = {
obj.get("name") for obj in objects if obj.get("name")
}
# 校验对象必填属性
for obj in objects:
required_attrs = LabelStudioConfigValidator._get_required_attrs(
obj.tag, is_control=False
)
for attr in required_attrs:
if not obj.attrib.get(attr):
return False, f"Object {obj.tag} missing '{attr}' attribute"
if obj.attrib.get("value") and not obj.attrib.get("value", "").startswith("$"):
return False, "Object value must start with '$' (e.g., '$image')"
# 校验控件（布局类仅提示，不作为失败条件）
for control in controls:
valid, error = LabelStudioConfigValidator._validate_control(control)
if not valid:
category = LabelStudioConfigValidator._get_control_category(control.tag)
strict = category == "labeling"
valid, error = LabelStudioConfigValidator._validate_control(
control, object_names, strict
)
if not valid and strict:
return False, f"Control {control.tag}: {error}"
return True, None
except ET.ParseError as e:
@@ -59,7 +101,11 @@ class LabelStudioConfigValidator:
return False, f"Validation error: {str(e)}"
@staticmethod
def _validate_control(control: ET.Element) -> Tuple[bool, Optional[str]]:
def _validate_control(
control: ET.Element,
object_names: Set[str],
strict: bool
) -> Tuple[bool, Optional[str]]:
"""
验证单个控件元素
@@ -72,26 +118,37 @@ class LabelStudioConfigValidator:
config = LabelStudioConfigValidator._get_config()
# 检查必需属性
if 'name' not in control.attrib:
return False, "Missing 'name' attribute"
if 'toName' not in control.attrib:
return False, "Missing 'toName' attribute"
required_attrs = LabelStudioConfigValidator._get_required_attrs(
control.tag, is_control=True
)
for attr in required_attrs:
if not control.attrib.get(attr):
return (False, f"Missing '{attr}' attribute") if strict else (True, None)
# 校验 toName 指向对象
if strict and control.attrib.get("toName"):
to_names = [
name.strip()
for name in control.attrib.get("toName", "").split(",")
if name.strip()
]
invalid = [name for name in to_names if name not in object_names]
if invalid:
return False, f"toName references unknown object(s): {', '.join(invalid)}"
# 检查控件是否需要子元素
if config.requires_children(control.tag):
child_tag = config.get_child_tag(control.tag)
if not child_tag:
return False, f"Configuration error: no child_tag defined for {control.tag}"
return (False, f"Configuration error: no child_tag defined for {control.tag}") if strict else (True, None)
children = control.findall(child_tag)
if not children:
return False, f"{control.tag} must have at least one <{child_tag}> child"
# 检查每个子元素是否有value
return (False, f"{control.tag} must have at least one <{child_tag}> child") if strict else (True, None)
for child in children:
if 'value' not in child.attrib:
return False, f"{child_tag} missing 'value' attribute"
if "value" not in child.attrib or not child.attrib.get("value"):
return (False, f"{child_tag} missing 'value' attribute") if strict else (True, None)
return True, None