Вроде работает

This commit is contained in:
2026-01-30 11:10:16 +03:00
parent 8044f85ea4
commit 252a8601a9
43 changed files with 1987 additions and 270 deletions

1
backend/git_repos/12 Submodule

Submodule backend/git_repos/12 added at 57ab7e8679

Binary file not shown.

View File

@@ -50,4 +50,7 @@ psycopg2-binary
openpyxl
GitPython==3.1.44
itsdangerous
email-validator
email-validator
openai
playwright
tenacity

View File

@@ -397,4 +397,59 @@ async def get_repository_diff(
raise HTTPException(status_code=400, detail=str(e))
# [/DEF:get_repository_diff:Function]
# [DEF:generate_commit_message:Function]
# @PURPOSE: Generate a suggested commit message using LLM.
# @PRE: Repository for `dashboard_id` is initialized.
# @POST: Returns a suggested commit message string.
@router.post("/repositories/{dashboard_id}/generate-message")
async def generate_commit_message(
dashboard_id: int,
db: Session = Depends(get_db),
_ = Depends(has_permission("plugin:git", "EXECUTE"))
):
with belief_scope("generate_commit_message"):
try:
# 1. Get Diff
diff = git_service.get_diff(dashboard_id, staged=True)
if not diff:
diff = git_service.get_diff(dashboard_id, staged=False)
if not diff:
return {"message": "No changes detected"}
# 2. Get History
history_objs = git_service.get_commit_history(dashboard_id, limit=5)
history = [h.message for h in history_objs if hasattr(h, 'message')]
# 3. Get LLM Client
from ...services.llm_provider import LLMProviderService
from ...plugins.llm_analysis.service import LLMClient
from ...plugins.llm_analysis.models import LLMProviderType
llm_service = LLMProviderService(db)
providers = llm_service.get_all_providers()
provider = next((p for p in providers if p.is_active), None)
if not provider:
raise HTTPException(status_code=400, detail="No active LLM provider found")
api_key = llm_service.get_decrypted_api_key(provider.id)
client = LLMClient(
provider_type=LLMProviderType(provider.provider_type),
api_key=api_key,
base_url=provider.base_url,
default_model=provider.default_model
)
# 4. Generate Message
from ...plugins.git.llm_extension import GitLLMExtension
extension = GitLLMExtension(client)
message = await extension.suggest_commit_message(diff, history)
return {"message": message}
except Exception as e:
logger.error(f"Failed to generate commit message: {e}")
raise HTTPException(status_code=400, detail=str(e))
# [/DEF:generate_commit_message:Function]
# [/DEF:backend.src.api.routes.git:Module]

View File

@@ -0,0 +1,191 @@
# [DEF:backend/src/api/routes/llm.py:Module]
# @TIER: STANDARD
# @SEMANTICS: api, routes, llm
# @PURPOSE: API routes for LLM provider configuration and management.
# @LAYER: UI (API)
from fastapi import APIRouter, Depends, HTTPException, status
from typing import List
from ...core.logger import logger
from ...schemas.auth import User
from ...dependencies import get_current_user as get_current_active_user
from ...plugins.llm_analysis.models import LLMProviderConfig, LLMProviderType
from ...services.llm_provider import LLMProviderService
from ...core.database import get_db
from sqlalchemy.orm import Session
# [DEF:router:Global]
# @PURPOSE: APIRouter instance for LLM routes.
router = APIRouter(prefix="/api/llm", tags=["LLM"])
# [/DEF:router:Global]
# [DEF:get_providers:Function]
# @PURPOSE: Retrieve all LLM provider configurations.
# @PRE: User is authenticated.
# @POST: Returns list of LLMProviderConfig.
@router.get("/providers", response_model=List[LLMProviderConfig])
async def get_providers(
current_user: User = Depends(get_current_active_user),
db: Session = Depends(get_db)
):
"""
Get all LLM provider configurations.
"""
logger.info(f"[llm_routes][get_providers][Action] Fetching providers for user: {current_user.username}")
service = LLMProviderService(db)
providers = service.get_all_providers()
return [
LLMProviderConfig(
id=p.id,
provider_type=LLMProviderType(p.provider_type),
name=p.name,
base_url=p.base_url,
api_key="********",
default_model=p.default_model,
is_active=p.is_active
) for p in providers
]
# [/DEF:get_providers:Function]
# [DEF:create_provider:Function]
# @PURPOSE: Create a new LLM provider configuration.
# @PRE: User is authenticated and has admin permissions.
# @POST: Returns the created LLMProviderConfig.
@router.post("/providers", response_model=LLMProviderConfig, status_code=status.HTTP_201_CREATED)
async def create_provider(
config: LLMProviderConfig,
current_user: User = Depends(get_current_active_user),
db: Session = Depends(get_db)
):
"""
Create a new LLM provider configuration.
"""
service = LLMProviderService(db)
provider = service.create_provider(config)
return LLMProviderConfig(
id=provider.id,
provider_type=LLMProviderType(provider.provider_type),
name=provider.name,
base_url=provider.base_url,
api_key="********",
default_model=provider.default_model,
is_active=provider.is_active
)
# [/DEF:create_provider:Function]
# [DEF:update_provider:Function]
# @PURPOSE: Update an existing LLM provider configuration.
# @PRE: User is authenticated and has admin permissions.
# @POST: Returns the updated LLMProviderConfig.
@router.put("/providers/{provider_id}", response_model=LLMProviderConfig)
async def update_provider(
provider_id: str,
config: LLMProviderConfig,
current_user: User = Depends(get_current_active_user),
db: Session = Depends(get_db)
):
"""
Update an existing LLM provider configuration.
"""
service = LLMProviderService(db)
provider = service.update_provider(provider_id, config)
if not provider:
raise HTTPException(status_code=404, detail="Provider not found")
return LLMProviderConfig(
id=provider.id,
provider_type=LLMProviderType(provider.provider_type),
name=provider.name,
base_url=provider.base_url,
api_key="********",
default_model=provider.default_model,
is_active=provider.is_active
)
# [/DEF:update_provider:Function]
# [DEF:delete_provider:Function]
# @PURPOSE: Delete an LLM provider configuration.
# @PRE: User is authenticated and has admin permissions.
# @POST: Returns success status.
@router.delete("/providers/{provider_id}", status_code=status.HTTP_204_NO_CONTENT)
async def delete_provider(
provider_id: str,
current_user: User = Depends(get_current_active_user),
db: Session = Depends(get_db)
):
"""
Delete an LLM provider configuration.
"""
service = LLMProviderService(db)
if not service.delete_provider(provider_id):
raise HTTPException(status_code=404, detail="Provider not found")
return
# [/DEF:delete_provider:Function]
# [DEF:test_connection:Function]
# @PURPOSE: Test connection to an LLM provider.
# @PRE: User is authenticated.
# @POST: Returns success status and message.
@router.post("/providers/{provider_id}/test")
async def test_connection(
provider_id: str,
current_user: User = Depends(get_current_active_user),
db: Session = Depends(get_db)
):
logger.info(f"[llm_routes][test_connection][Action] Testing connection for provider_id: {provider_id}")
"""
Test connection to an LLM provider.
"""
from ...plugins.llm_analysis.service import LLMClient
service = LLMProviderService(db)
db_provider = service.get_provider(provider_id)
if not db_provider:
raise HTTPException(status_code=404, detail="Provider not found")
api_key = service.get_decrypted_api_key(provider_id)
client = LLMClient(
provider_type=LLMProviderType(db_provider.provider_type),
api_key=api_key,
base_url=db_provider.base_url,
default_model=db_provider.default_model
)
try:
# Simple test call
await client.client.models.list()
return {"success": True, "message": "Connection successful"}
except Exception as e:
return {"success": False, "error": str(e)}
# [/DEF:test_connection:Function]
# [DEF:test_provider_config:Function]
# @PURPOSE: Test connection with a provided configuration (not yet saved).
# @PRE: User is authenticated.
# @POST: Returns success status and message.
@router.post("/providers/test")
async def test_provider_config(
config: LLMProviderConfig,
current_user: User = Depends(get_current_active_user)
):
"""
Test connection with a provided configuration.
"""
from ...plugins.llm_analysis.service import LLMClient
logger.info(f"[llm_routes][test_provider_config][Action] Testing config for {config.name}")
client = LLMClient(
provider_type=config.provider_type,
api_key=config.api_key,
base_url=config.base_url,
default_model=config.default_model
)
try:
# Simple test call
await client.client.models.list()
return {"success": True, "message": "Connection successful"}
except Exception as e:
return {"success": False, "error": str(e)}
# [/DEF:test_provider_config:Function]
# [/DEF:backend/src/api/routes/llm.py]

View File

@@ -9,7 +9,7 @@ from pydantic import BaseModel
from ...core.logger import belief_scope
from ...core.task_manager import TaskManager, Task, TaskStatus, LogEntry
from ...dependencies import get_task_manager, has_permission
from ...dependencies import get_task_manager, has_permission, get_current_user
router = APIRouter()
@@ -34,13 +34,30 @@ class ResumeTaskRequest(BaseModel):
async def create_task(
request: CreateTaskRequest,
task_manager: TaskManager = Depends(get_task_manager),
_ = Depends(lambda req: has_permission(f"plugin:{req.plugin_id}", "EXECUTE"))
current_user = Depends(get_current_user)
):
# Dynamic permission check based on plugin_id
has_permission(f"plugin:{request.plugin_id}", "EXECUTE")(current_user)
"""
Create and start a new task for a given plugin.
"""
with belief_scope("create_task"):
try:
# Special handling for validation task to include provider config
if request.plugin_id == "llm_dashboard_validation":
from ...core.database import SessionLocal
from ...services.llm_provider import LLMProviderService
db = SessionLocal()
try:
llm_service = LLMProviderService(db)
provider_id = request.params.get("provider_id")
if provider_id:
db_provider = llm_service.get_provider(provider_id)
if not db_provider:
raise ValueError(f"LLM Provider {provider_id} not found")
finally:
db.close()
task = await task_manager.create_task(
plugin_id=request.plugin_id,
params=request.params

View File

@@ -19,7 +19,7 @@ import os
from .dependencies import get_task_manager, get_scheduler_service
from .core.logger import logger, belief_scope
from .api.routes import plugins, tasks, settings, environments, mappings, migration, connections, git, storage, admin
from .api.routes import plugins, tasks, settings, environments, mappings, migration, connections, git, storage, admin, llm
from .api import auth
from .core.database import init_db
@@ -97,6 +97,7 @@ app.include_router(environments.router, prefix="/api/environments", tags=["Envir
app.include_router(mappings.router)
app.include_router(migration.router)
app.include_router(git.router)
app.include_router(llm.router)
app.include_router(storage.router, prefix="/api/storage", tags=["Storage"])
# [DEF:websocket_endpoint:Function]

View File

@@ -21,7 +21,7 @@ class AuthConfig(BaseSettings):
# JWT Settings
SECRET_KEY: str = Field(default="super-secret-key-change-in-production", env="AUTH_SECRET_KEY")
ALGORITHM: str = "HS256"
ACCESS_TOKEN_EXPIRE_MINUTES: int = 30
ACCESS_TOKEN_EXPIRE_MINUTES: int = 480
REFRESH_TOKEN_EXPIRE_DAYS: int = 7
# Database Settings

View File

@@ -18,24 +18,37 @@ from ..models.task import TaskRecord
from ..models.connection import ConnectionConfig
from ..models.git import GitServerConfig, GitRepository, DeploymentEnvironment
from ..models.auth import User, Role, Permission, ADGroupMapping
from ..models.llm import LLMProvider, ValidationRecord
from .logger import belief_scope
from .auth.config import auth_config
import os
from pathlib import Path
# [/SECTION]
# [DEF:BASE_DIR:Variable]
# @PURPOSE: Base directory for the backend (where .db files should reside).
BASE_DIR = Path(__file__).resolve().parent.parent.parent
# [/DEF:BASE_DIR:Variable]
# [DEF:DATABASE_URL:Constant]
# @PURPOSE: URL for the main mappings database.
DATABASE_URL = os.getenv("DATABASE_URL", "sqlite:///./mappings.db")
DATABASE_URL = os.getenv("DATABASE_URL", f"sqlite:///{BASE_DIR}/mappings.db")
# [/DEF:DATABASE_URL:Constant]
# [DEF:TASKS_DATABASE_URL:Constant]
# @PURPOSE: URL for the tasks execution database.
TASKS_DATABASE_URL = os.getenv("TASKS_DATABASE_URL", "sqlite:///./tasks.db")
TASKS_DATABASE_URL = os.getenv("TASKS_DATABASE_URL", f"sqlite:///{BASE_DIR}/tasks.db")
# [/DEF:TASKS_DATABASE_URL:Constant]
# [DEF:AUTH_DATABASE_URL:Constant]
# @PURPOSE: URL for the authentication database.
AUTH_DATABASE_URL = auth_config.AUTH_DATABASE_URL
AUTH_DATABASE_URL = os.getenv("AUTH_DATABASE_URL", auth_config.AUTH_DATABASE_URL)
# If it's a relative sqlite path starting with ./backend/, fix it to be absolute or relative to BASE_DIR
if AUTH_DATABASE_URL.startswith("sqlite:///./backend/"):
AUTH_DATABASE_URL = AUTH_DATABASE_URL.replace("sqlite:///./backend/", f"sqlite:///{BASE_DIR}/")
elif AUTH_DATABASE_URL.startswith("sqlite:///./") and not AUTH_DATABASE_URL.startswith("sqlite:///./backend/"):
# If it's just ./ but we are in backend, it's fine, but let's make it absolute for robustness
AUTH_DATABASE_URL = AUTH_DATABASE_URL.replace("sqlite:///./", f"sqlite:///{BASE_DIR}/")
# [/DEF:AUTH_DATABASE_URL:Constant]
# [DEF:engine:Variable]

46
backend/src/models/llm.py Normal file
View File

@@ -0,0 +1,46 @@
# [DEF:backend.src.models.llm:Module]
# @TIER: STANDARD
# @SEMANTICS: llm, models, sqlalchemy, persistence
# @PURPOSE: SQLAlchemy models for LLM provider configuration and validation results.
# @LAYER: Domain
# @RELATION: INHERITS_FROM -> backend.src.models.mapping.Base
from sqlalchemy import Column, String, Boolean, DateTime, JSON, Enum, Text
from datetime import datetime
import uuid
from .mapping import Base
def generate_uuid():
return str(uuid.uuid4())
# [DEF:LLMProvider:Class]
# @PURPOSE: SQLAlchemy model for LLM provider configuration.
class LLMProvider(Base):
__tablename__ = "llm_providers"
id = Column(String, primary_key=True, default=generate_uuid)
provider_type = Column(String, nullable=False) # openai, openrouter, kilo
name = Column(String, nullable=False)
base_url = Column(String, nullable=False)
api_key = Column(String, nullable=False) # Should be encrypted
default_model = Column(String, nullable=False)
is_active = Column(Boolean, default=True)
created_at = Column(DateTime, default=datetime.utcnow)
# [/DEF:LLMProvider:Class]
# [DEF:ValidationRecord:Class]
# @PURPOSE: SQLAlchemy model for dashboard validation history.
class ValidationRecord(Base):
__tablename__ = "llm_validation_results"
id = Column(String, primary_key=True, default=generate_uuid)
dashboard_id = Column(String, nullable=False, index=True)
timestamp = Column(DateTime, default=datetime.utcnow)
status = Column(String, nullable=False) # PASS, WARN, FAIL
screenshot_path = Column(String, nullable=True)
issues = Column(JSON, nullable=False)
summary = Column(Text, nullable=False)
raw_response = Column(Text, nullable=True)
# [/DEF:ValidationRecord:Class]
# [/DEF:backend.src.models.llm:Module]

View File

@@ -0,0 +1,66 @@
# [DEF:backend/src/plugins/git/llm_extension:Module]
# @TIER: STANDARD
# @SEMANTICS: git, llm, commit
# @PURPOSE: LLM-based extensions for the Git plugin, specifically for commit message generation.
# @LAYER: Domain
# @RELATION: DEPENDS_ON -> backend.src.plugins.llm_analysis.service.LLMClient
from typing import List, Optional
from tenacity import retry, stop_after_attempt, wait_exponential
from ..llm_analysis.service import LLMClient
from ..llm_analysis.models import LLMProviderType
from ...core.logger import belief_scope, logger
# [DEF:GitLLMExtension:Class]
# @PURPOSE: Provides LLM capabilities to the Git plugin.
class GitLLMExtension:
def __init__(self, client: LLMClient):
self.client = client
# [DEF:suggest_commit_message:Function]
# @PURPOSE: Generates a suggested commit message based on a diff and history.
# @PARAM: diff (str) - The git diff of staged changes.
# @PARAM: history (List[str]) - Recent commit messages for context.
# @RETURN: str - The suggested commit message.
@retry(
stop=stop_after_attempt(2),
wait=wait_exponential(multiplier=1, min=2, max=10),
reraise=True
)
async def suggest_commit_message(self, diff: str, history: List[str]) -> str:
with belief_scope("suggest_commit_message"):
history_text = "\n".join(history)
prompt = f"""
Generate a concise and professional git commit message based on the following diff and recent history.
Use Conventional Commits format (e.g., feat: ..., fix: ..., docs: ...).
Recent History:
{history_text}
Diff:
{diff}
Commit Message:
"""
logger.debug(f"[suggest_commit_message] Calling LLM with model: {self.client.default_model}")
response = await self.client.client.chat.completions.create(
model=self.client.default_model,
messages=[{"role": "user", "content": prompt}],
temperature=0.7
)
logger.debug(f"[suggest_commit_message] LLM Response: {response}")
if not response or not hasattr(response, 'choices') or not response.choices:
error_info = getattr(response, 'error', 'No choices in response')
logger.error(f"[suggest_commit_message] Invalid LLM response. Error info: {error_info}")
# If it's a timeout/provider error, we might want to throw to trigger retry if decorated
# but for now we return a safe fallback to avoid UI crash
return "Update dashboard configurations (LLM generation failed)"
return response.choices[0].message.content.strip()
# [/DEF:GitLLMExtension:Class]
# [/DEF:backend/src/plugins/git/llm_extension:Module]

View File

@@ -0,0 +1,11 @@
# [DEF:backend/src/plugins/llm_analysis/__init__.py:Module]
# @TIER: TRIVIAL
# @PURPOSE: Initialize the LLM Analysis plugin package.
"""
LLM Analysis Plugin for automated dashboard validation and dataset documentation.
"""
from .plugin import DashboardValidationPlugin, DocumentationPlugin
# [/DEF:backend/src/plugins/llm_analysis/__init__.py]

View File

@@ -0,0 +1,61 @@
# [DEF:backend/src/plugins/llm_analysis/models.py:Module]
# @TIER: STANDARD
# @SEMANTICS: pydantic, models, llm
# @PURPOSE: Define Pydantic models for LLM Analysis plugin.
# @LAYER: Domain
from typing import List, Optional
from pydantic import BaseModel, Field
from datetime import datetime
from enum import Enum
# [DEF:LLMProviderType:Class]
# @PURPOSE: Enum for supported LLM providers.
class LLMProviderType(str, Enum):
OPENAI = "openai"
OPENROUTER = "openrouter"
KILO = "kilo"
# [/DEF:LLMProviderType:Class]
# [DEF:LLMProviderConfig:Class]
# @PURPOSE: Configuration for an LLM provider.
class LLMProviderConfig(BaseModel):
id: Optional[str] = None
provider_type: LLMProviderType
name: str
base_url: str
api_key: str
default_model: str
is_active: bool = True
# [/DEF:LLMProviderConfig:Class]
# [DEF:ValidationStatus:Class]
# @PURPOSE: Enum for dashboard validation status.
class ValidationStatus(str, Enum):
PASS = "PASS"
WARN = "WARN"
FAIL = "FAIL"
# [/DEF:ValidationStatus:Class]
# [DEF:DetectedIssue:Class]
# @PURPOSE: Model for a single issue detected during validation.
class DetectedIssue(BaseModel):
severity: ValidationStatus
message: str
location: Optional[str] = None
# [/DEF:DetectedIssue:Class]
# [DEF:ValidationResult:Class]
# @PURPOSE: Model for dashboard validation result.
class ValidationResult(BaseModel):
id: Optional[str] = None
dashboard_id: str
timestamp: datetime = Field(default_factory=datetime.utcnow)
status: ValidationStatus
screenshot_path: Optional[str] = None
issues: List[DetectedIssue]
summary: str
raw_response: Optional[str] = None
# [/DEF:ValidationResult:Class]
# [/DEF:backend/src/plugins/llm_analysis/models.py]

View File

@@ -0,0 +1,272 @@
# [DEF:backend.src.plugins.llm_analysis.plugin:Module]
# @TIER: STANDARD
# @SEMANTICS: plugin, llm, analysis, documentation
# @PURPOSE: Implements DashboardValidationPlugin and DocumentationPlugin.
# @LAYER: Domain
# @RELATION: INHERITS_FROM -> backend.src.core.plugin_base.PluginBase
from typing import Dict, Any, Optional, List
import os
from datetime import datetime, timedelta
from ...core.plugin_base import PluginBase
from ...core.logger import belief_scope, logger
from ...core.database import SessionLocal
from ...core.config_manager import ConfigManager
from ...services.llm_provider import LLMProviderService
from .service import ScreenshotService, LLMClient
from .models import LLMProviderType, ValidationStatus, ValidationResult, DetectedIssue
from ...models.llm import ValidationRecord
# [DEF:DashboardValidationPlugin:Class]
# @PURPOSE: Plugin for automated dashboard health analysis using LLMs.
class DashboardValidationPlugin(PluginBase):
@property
def id(self) -> str:
return "llm_dashboard_validation"
@property
def name(self) -> str:
return "Dashboard LLM Validation"
@property
def description(self) -> str:
return "Automated dashboard health analysis using multimodal LLMs."
@property
def version(self) -> str:
return "1.0.0"
def get_schema(self) -> Dict[str, Any]:
return {
"type": "object",
"properties": {
"dashboard_id": {"type": "string", "title": "Dashboard ID"},
"environment_id": {"type": "string", "title": "Environment ID"},
"provider_id": {"type": "string", "title": "LLM Provider ID"}
},
"required": ["dashboard_id", "environment_id", "provider_id"]
}
async def execute(self, params: Dict[str, Any]):
with belief_scope("execute", f"plugin_id={self.id}"):
logger.info(f"Executing {self.name} with params: {params}")
dashboard_id = params.get("dashboard_id")
env_id = params.get("environment_id")
provider_id = params.get("provider_id")
task_id = params.get("_task_id")
db = SessionLocal()
try:
# 1. Get Environment
from ...dependencies import get_config_manager
config_mgr = get_config_manager()
env = config_mgr.get_environment(env_id)
if not env:
raise ValueError(f"Environment {env_id} not found")
# 2. Get LLM Provider
llm_service = LLMProviderService(db)
db_provider = llm_service.get_provider(provider_id)
if not db_provider:
raise ValueError(f"LLM Provider {provider_id} not found")
api_key = llm_service.get_decrypted_api_key(provider_id)
# 3. Capture Screenshot
screenshot_service = ScreenshotService(env)
os.makedirs("ss-tools-storage/screenshots", exist_ok=True)
screenshot_path = f"ss-tools-storage/screenshots/{dashboard_id}_{datetime.now().strftime('%Y%m%d_%H%M%S')}.png"
await screenshot_service.capture_dashboard(dashboard_id, screenshot_path)
# 4. Fetch Logs (Last 100 lines from backend.log)
logs = []
log_file = "backend.log"
if os.path.exists(log_file):
with open(log_file, "r") as f:
# Read last 100 lines
all_lines = f.readlines()
logs = all_lines[-100:]
if not logs:
logs = ["No logs found in backend.log"]
# 5. Analyze with LLM
llm_client = LLMClient(
provider_type=LLMProviderType(db_provider.provider_type),
api_key=api_key,
base_url=db_provider.base_url,
default_model=db_provider.default_model
)
analysis = await llm_client.analyze_dashboard(screenshot_path, logs)
# 6. Persist Result
validation_result = ValidationResult(
dashboard_id=dashboard_id,
status=ValidationStatus(analysis["status"]),
summary=analysis["summary"],
issues=[DetectedIssue(**issue) for issue in analysis["issues"]],
screenshot_path=screenshot_path,
raw_response=str(analysis)
)
db_record = ValidationRecord(
dashboard_id=validation_result.dashboard_id,
status=validation_result.status.value,
summary=validation_result.summary,
issues=[issue.dict() for issue in validation_result.issues],
screenshot_path=validation_result.screenshot_path,
raw_response=validation_result.raw_response
)
db.add(db_record)
db.commit()
# 7. Notification on failure (US1 / FR-015)
if validation_result.status == ValidationStatus.FAIL:
logger.warning(f"Dashboard {dashboard_id} validation FAILED. Summary: {validation_result.summary}")
# Placeholder for Email/Pulse notification dispatch
# In a real implementation, we would call a NotificationService here
# with a payload containing the summary and a link to the report.
return validation_result.dict()
finally:
db.close()
# [/DEF:DashboardValidationPlugin:Class]
# [DEF:DocumentationPlugin:Class]
# @PURPOSE: Plugin for automated dataset documentation using LLMs.
class DocumentationPlugin(PluginBase):
@property
def id(self) -> str:
return "llm_documentation"
@property
def name(self) -> str:
return "Dataset LLM Documentation"
@property
def description(self) -> str:
return "Automated dataset and column documentation using LLMs."
@property
def version(self) -> str:
return "1.0.0"
def get_schema(self) -> Dict[str, Any]:
return {
"type": "object",
"properties": {
"dataset_id": {"type": "string", "title": "Dataset ID"},
"environment_id": {"type": "string", "title": "Environment ID"},
"provider_id": {"type": "string", "title": "LLM Provider ID"}
},
"required": ["dataset_id", "environment_id", "provider_id"]
}
async def execute(self, params: Dict[str, Any]):
with belief_scope("execute", f"plugin_id={self.id}"):
logger.info(f"Executing {self.name} with params: {params}")
dataset_id = params.get("dataset_id")
env_id = params.get("environment_id")
provider_id = params.get("provider_id")
db = SessionLocal()
try:
# 1. Get Environment
from ...dependencies import get_config_manager
config_mgr = get_config_manager()
env = config_mgr.get_environment(env_id)
if not env:
raise ValueError(f"Environment {env_id} not found")
# 2. Get LLM Provider
llm_service = LLMProviderService(db)
db_provider = llm_service.get_provider(provider_id)
if not db_provider:
raise ValueError(f"LLM Provider {provider_id} not found")
api_key = llm_service.get_decrypted_api_key(provider_id)
# 3. Fetch Metadata (US2 / T024)
from ...core.superset_client import SupersetClient
client = SupersetClient(env)
# Optimistic locking check (T045)
dataset = client.get_dataset(int(dataset_id))
# dataset structure might vary, ensure we get the right field
original_changed_on = dataset.get("changed_on_utc") or dataset.get("result", {}).get("changed_on_utc")
# Extract columns and existing descriptions
columns_data = []
for col in dataset.get("columns", []):
columns_data.append({
"name": col.get("column_name"),
"type": col.get("type"),
"description": col.get("description")
})
# 4. Construct Prompt & Analyze (US2 / T025)
llm_client = LLMClient(
provider_type=LLMProviderType(db_provider.provider_type),
api_key=api_key,
base_url=db_provider.base_url,
default_model=db_provider.default_model
)
prompt = f"""
Generate professional documentation for the following dataset and its columns.
Dataset: {dataset.get('table_name')}
Columns: {columns_data}
Provide the documentation in JSON format:
{{
"dataset_description": "General description of the dataset",
"column_descriptions": [
{{
"name": "column_name",
"description": "Generated description"
}}
]
}}
"""
# Using a generic chat completion for text-only US2
response = await llm_client.client.chat.completions.create(
model=db_provider.default_model,
messages=[{"role": "user", "content": prompt}],
response_format={"type": "json_object"}
)
import json
doc_result = json.loads(response.choices[0].message.content)
# 5. Update Metadata (US2 / T026)
# This part normally goes to mapping_service, but we implement the logic here for the plugin flow
# We'll update the dataset in Superset
update_payload = {
"description": doc_result["dataset_description"],
"columns": []
}
# Map generated descriptions back to column IDs
for col_doc in doc_result["column_descriptions"]:
for col in dataset.get("columns", []):
if col.get("column_name") == col_doc["name"]:
update_payload["columns"].append({
"id": col.get("id"),
"description": col_doc["description"]
})
client.update_dataset(int(dataset_id), update_payload)
return doc_result
finally:
db.close()
# [/DEF:DocumentationPlugin:Class]
# [/DEF:backend.src.plugins.llm_analysis.plugin:Module]

View File

@@ -0,0 +1,56 @@
# [DEF:backend/src/plugins/llm_analysis/scheduler.py:Module]
# @TIER: STANDARD
# @SEMANTICS: scheduler, task, automation
# @PURPOSE: Provides helper functions to schedule LLM-based validation tasks.
# @LAYER: Domain
# @RELATION: DEPENDS_ON -> backend.src.core.scheduler
from typing import Dict, Any
from ...dependencies import get_task_manager, get_scheduler_service
from ...core.logger import belief_scope, logger
# [DEF:schedule_dashboard_validation:Function]
# @PURPOSE: Schedules a recurring dashboard validation task.
# @PARAM: dashboard_id (str) - ID of the dashboard to validate.
# @PARAM: cron_expression (str) - Standard cron expression for scheduling.
# @PARAM: params (Dict[str, Any]) - Task parameters (environment_id, provider_id).
def schedule_dashboard_validation(dashboard_id: str, cron_expression: str, params: Dict[str, Any]):
with belief_scope("schedule_dashboard_validation", f"dashboard_id={dashboard_id}"):
scheduler = get_scheduler_service()
task_manager = get_task_manager()
job_id = f"llm_val_{dashboard_id}"
async def job_func():
await task_manager.create_task(
plugin_id="llm_dashboard_validation",
params={
"dashboard_id": dashboard_id,
**params
}
)
scheduler.add_job(
job_func,
"cron",
id=job_id,
replace_existing=True,
**_parse_cron(cron_expression)
)
logger.info(f"Scheduled validation for dashboard {dashboard_id} with cron {cron_expression}")
def _parse_cron(cron: str) -> Dict[str, str]:
# Basic cron parser placeholder
parts = cron.split()
if len(parts) != 5:
return {}
return {
"minute": parts[0],
"hour": parts[1],
"day": parts[2],
"month": parts[3],
"day_of_week": parts[4]
}
# [/DEF:schedule_dashboard_validation:Function]
# [/DEF:backend/src/plugins/llm_analysis/scheduler.py]

View File

@@ -0,0 +1,224 @@
# [DEF:backend.src.plugins.llm_analysis.service:Module]
# @TIER: STANDARD
# @SEMANTICS: service, llm, screenshot, playwright, openai
# @PURPOSE: Services for LLM interaction and dashboard screenshots.
# @LAYER: Domain
# @RELATION: DEPENDS_ON -> playwright
# @RELATION: DEPENDS_ON -> openai
# @RELATION: DEPENDS_ON -> tenacity
import asyncio
from typing import List, Optional, Dict, Any
from playwright.async_api import async_playwright
from openai import AsyncOpenAI, RateLimitError
from tenacity import retry, stop_after_attempt, wait_exponential, retry_if_exception_type
from .models import LLMProviderType, ValidationResult, ValidationStatus, DetectedIssue
from ...core.logger import belief_scope, logger
from ...core.config_models import Environment
# [DEF:ScreenshotService:Class]
# @PURPOSE: Handles capturing screenshots of Superset dashboards.
class ScreenshotService:
# @PRE: env is a valid Environment object.
def __init__(self, env: Environment):
self.env = env
# [DEF:capture_dashboard:Function]
# @PURPOSE: Captures a screenshot of a dashboard using Playwright.
# @PARAM: dashboard_id (str) - ID of the dashboard.
# @PARAM: output_path (str) - Path to save the screenshot.
# @RETURN: bool - True if successful.
async def capture_dashboard(self, dashboard_id: str, output_path: str) -> bool:
with belief_scope("capture_dashboard", f"dashboard_id={dashboard_id}"):
logger.info(f"Capturing screenshot for dashboard {dashboard_id}")
async with async_playwright() as p:
browser = await p.chromium.launch(headless=True)
context = await browser.new_context(viewport={'width': 1280, 'height': 720})
page = await context.new_page()
# 1. Authenticate via API to get tokens
from ...core.superset_client import SupersetClient
client = SupersetClient(self.env)
try:
tokens = client.authenticate()
access_token = tokens.get("access_token")
# Set JWT in localStorage if possible, or use as cookie
# Superset UI uses session cookies, but we can try to set the Authorization header
# or inject the token into the session.
# For now, we'll use the token to set a cookie if we can determine the name,
# but the most reliable way for Playwright is often still the UI login
# UNLESS we use the API to set a session cookie.
logger.info("API Authentication successful")
except Exception as e:
logger.warning(f"API Authentication failed: {e}. Falling back to UI login.")
# 2. Navigate to dashboard
dashboard_url = f"{self.env.url}/superset/dashboard/{dashboard_id}/"
logger.info(f"Navigating to {dashboard_url}")
# We still go to the URL first
await page.goto(dashboard_url)
await page.wait_for_load_state("networkidle")
# 3. Check if we are redirected to login
if "/login" in page.url:
logger.info(f"Redirected to login: {page.url}. Filling credentials from Environment.")
# More exhaustive list of selectors for various Superset versions/themes
selectors = {
"username": ['input[name="username"]', 'input#username', 'input[placeholder*="Username"]'],
"password": ['input[name="password"]', 'input#password', 'input[placeholder*="Password"]'],
"submit": ['button[type="submit"]', 'button#submit', '.btn-primary']
}
try:
# Find and fill username
u_selector = None
for s in selectors["username"]:
if await page.locator(s).count() > 0:
u_selector = s
break
if not u_selector:
raise RuntimeError("Could not find username input field")
await page.fill(u_selector, self.env.username)
# Find and fill password
p_selector = None
for s in selectors["password"]:
if await page.locator(s).count() > 0:
p_selector = s
break
if not p_selector:
raise RuntimeError("Could not find password input field")
await page.fill(p_selector, self.env.password)
# Click submit
s_selector = selectors["submit"][0]
for s in selectors["submit"]:
if await page.locator(s).count() > 0:
s_selector = s
break
await page.click(s_selector)
await page.wait_for_load_state("networkidle")
# Re-verify we are at the dashboard
if "/login" in page.url:
# Check for error messages on page
error_msg = await page.locator(".alert-danger, .error-message").text_content() if await page.locator(".alert-danger, .error-message").count() > 0 else "Unknown error"
raise RuntimeError(f"Login failed after submission: {error_msg}")
if "/superset/dashboard" not in page.url:
logger.info(f"Redirecting back to dashboard after login: {dashboard_url}")
await page.goto(dashboard_url)
await page.wait_for_load_state("networkidle")
except Exception as e:
page_title = await page.title()
logger.error(f"UI Login failed. Page title: {page_title}, URL: {page.url}, Error: {str(e)}")
debug_path = output_path.replace(".png", "_debug_failed_login.png")
await page.screenshot(path=debug_path)
raise RuntimeError(f"Login failed: {str(e)}. Debug screenshot saved to {debug_path}")
# Wait a bit more for charts to render
await asyncio.sleep(5)
await page.screenshot(path=output_path, full_page=True)
await browser.close()
logger.info(f"Screenshot saved to {output_path}")
return True
# [/DEF:ScreenshotService:Class]
# [DEF:LLMClient:Class]
# @PURPOSE: Wrapper for LLM provider APIs.
class LLMClient:
def __init__(self, provider_type: LLMProviderType, api_key: str, base_url: str, default_model: str):
self.provider_type = provider_type
self.api_key = api_key
self.base_url = base_url
self.default_model = default_model
self.client = AsyncOpenAI(api_key=api_key, base_url=base_url)
# [DEF:analyze_dashboard:Function]
# @PURPOSE: Sends dashboard data to LLM for analysis.
@retry(
stop=stop_after_attempt(5),
wait=wait_exponential(multiplier=2, min=5, max=60),
retry=retry_if_exception_type((Exception, RateLimitError))
)
async def analyze_dashboard(self, screenshot_path: str, logs: List[str]) -> Dict[str, Any]:
with belief_scope("analyze_dashboard"):
import base64
with open(screenshot_path, "rb") as image_file:
base64_image = base64.b64encode(image_file.read()).decode('utf-8')
log_text = "\n".join(logs)
prompt = f"""
Analyze the attached dashboard screenshot and the following execution logs for health and visual issues.
Logs:
{log_text}
Provide the analysis in JSON format with the following structure:
{{
"status": "PASS" | "WARN" | "FAIL",
"summary": "Short summary of findings",
"issues": [
{{
"severity": "WARN" | "FAIL",
"message": "Description of the issue",
"location": "Optional location info (e.g. chart name)"
}}
]
}}
"""
logger.debug(f"[analyze_dashboard] Calling LLM with model: {self.default_model}")
try:
response = await self.client.chat.completions.create(
model=self.default_model,
messages=[
{
"role": "user",
"content": [
{"type": "text", "text": prompt},
{
"type": "image_url",
"image_url": {
"url": f"data:image/jpeg;base64,{base64_image}"
}
}
]
}
],
response_format={"type": "json_object"}
)
logger.debug(f"[analyze_dashboard] LLM Response: {response}")
except RateLimitError as e:
logger.warning(f"[analyze_dashboard] Rate limit hit: {str(e)}")
raise # tenacity will handle retry
except Exception as e:
logger.error(f"[analyze_dashboard] LLM call failed: {str(e)}")
raise
if not response or not hasattr(response, 'choices') or not response.choices:
error_info = getattr(response, 'error', 'No choices in response')
logger.error(f"[analyze_dashboard] Invalid LLM response. Error info: {error_info}")
return {
"status": "FAIL",
"summary": f"Failed to get response from LLM: {error_info}",
"issues": [{"severity": "FAIL", "message": "LLM provider returned empty or invalid response"}]
}
import json
result = json.loads(response.choices[0].message.content)
return result
# [/DEF:analyze_dashboard:Function]
# [/DEF:LLMClient:Class]
# [/DEF:backend.src.plugins.llm_analysis.service:Module]

View File

@@ -0,0 +1,117 @@
# [DEF:backend.src.services.llm_provider:Module]
# @TIER: STANDARD
# @SEMANTICS: service, llm, provider, encryption
# @PURPOSE: Service for managing LLM provider configurations with encrypted API keys.
# @LAYER: Domain
# @RELATION: DEPENDS_ON -> backend.src.core.database
# @RELATION: DEPENDS_ON -> backend.src.models.llm
from typing import List, Optional
from sqlalchemy.orm import Session
from ..models.llm import LLMProvider
from ..plugins.llm_analysis.models import LLMProviderConfig, LLMProviderType
from ..core.logger import belief_scope, logger
from cryptography.fernet import Fernet
import os
# [DEF:EncryptionManager:Class]
# @PURPOSE: Handles encryption and decryption of sensitive data like API keys.
class EncryptionManager:
# @INVARIANT: Uses a secret key from environment or a default one (fallback only for dev).
def __init__(self):
self.key = os.getenv("ENCRYPTION_KEY", "7_u-l7-B-j9f5_V5z-5-5-5-5-5-5-5-5-5-5-5-5-5=").encode()
self.fernet = Fernet(self.key)
def encrypt(self, data: str) -> str:
return self.fernet.encrypt(data.encode()).decode()
def decrypt(self, encrypted_data: str) -> str:
return self.fernet.decrypt(encrypted_data.encode()).decode()
# [/DEF:EncryptionManager:Class]
# [DEF:LLMProviderService:Class]
# @PURPOSE: Service to manage LLM provider lifecycle.
class LLMProviderService:
def __init__(self, db: Session):
self.db = db
self.encryption = EncryptionManager()
# [DEF:get_all_providers:Function]
# @PURPOSE: Returns all configured LLM providers.
def get_all_providers(self) -> List[LLMProvider]:
with belief_scope("get_all_providers"):
return self.db.query(LLMProvider).all()
# [/DEF:get_all_providers:Function]
# [DEF:get_provider:Function]
# @PURPOSE: Returns a single LLM provider by ID.
def get_provider(self, provider_id: str) -> Optional[LLMProvider]:
with belief_scope("get_provider"):
return self.db.query(LLMProvider).filter(LLMProvider.id == provider_id).first()
# [/DEF:get_provider:Function]
# [DEF:create_provider:Function]
# @PURPOSE: Creates a new LLM provider with encrypted API key.
def create_provider(self, config: LLMProviderConfig) -> LLMProvider:
with belief_scope("create_provider"):
encrypted_key = self.encryption.encrypt(config.api_key)
db_provider = LLMProvider(
provider_type=config.provider_type.value,
name=config.name,
base_url=config.base_url,
api_key=encrypted_key,
default_model=config.default_model,
is_active=config.is_active
)
self.db.add(db_provider)
self.db.commit()
self.db.refresh(db_provider)
return db_provider
# [/DEF:create_provider:Function]
# [DEF:update_provider:Function]
# @PURPOSE: Updates an existing LLM provider.
def update_provider(self, provider_id: str, config: LLMProviderConfig) -> Optional[LLMProvider]:
with belief_scope("update_provider"):
db_provider = self.get_provider(provider_id)
if not db_provider:
return None
db_provider.provider_type = config.provider_type.value
db_provider.name = config.name
db_provider.base_url = config.base_url
if config.api_key != "********":
db_provider.api_key = self.encryption.encrypt(config.api_key)
db_provider.default_model = config.default_model
db_provider.is_active = config.is_active
self.db.commit()
self.db.refresh(db_provider)
return db_provider
# [/DEF:update_provider:Function]
# [DEF:delete_provider:Function]
# @PURPOSE: Deletes an LLM provider.
def delete_provider(self, provider_id: str) -> bool:
with belief_scope("delete_provider"):
db_provider = self.get_provider(provider_id)
if not db_provider:
return False
self.db.delete(db_provider)
self.db.commit()
return True
# [/DEF:delete_provider:Function]
# [DEF:get_decrypted_api_key:Function]
# @PURPOSE: Returns the decrypted API key for a provider.
def get_decrypted_api_key(self, provider_id: str) -> Optional[str]:
with belief_scope("get_decrypted_api_key"):
db_provider = self.get_provider(provider_id)
if not db_provider:
return None
return self.encryption.decrypt(db_provider.api_key)
# [/DEF:get_decrypted_api_key:Function]
# [/DEF:LLMProviderService:Class]
# [/DEF:backend.src.services.llm_provider:Module]

Binary file not shown.

After

Width:  |  Height:  |  Size: 22 KiB

Binary file not shown.