386 lines
12 KiB
Python
386 lines
12 KiB
Python
"""AI Chat Completions API
|
||
|
||
OpenAI-compatible Chat Completions endpoint with xAI/LangChain backend.
|
||
|
||
Features:
|
||
- File Search (RAG) via xAI Collections
|
||
- Web Search via xAI web_search tool
|
||
- Aktenzeichen-based automatic collection lookup
|
||
- Multiple tools simultaneously
|
||
- Clean, reusable architecture for future LLM endpoints
|
||
|
||
Note: Streaming is not supported (Motia limitation - returns clear error).
|
||
|
||
Reusability:
|
||
- extract_request_params(): Parse requests for any LLM endpoint
|
||
- resolve_collection_id(): Auto-detect Aktenzeichen, lookup collection
|
||
- initialize_model_with_tools(): Bind tools to any LangChain model
|
||
- invoke_and_format_response(): Standard OpenAI response formatting
|
||
"""
|
||
import time
|
||
from typing import Any, Dict, List, Optional
|
||
from motia import FlowContext, http, ApiRequest, ApiResponse
|
||
|
||
config = {
|
||
"name": "AI Chat Completions API",
|
||
"description": "OpenAI-compatible Chat Completions API with xAI backend",
|
||
"flows": ["ai-general"],
|
||
"triggers": [
|
||
http("POST", "/ai/v1/chat/completions"),
|
||
http("POST", "/v1/chat/completions")
|
||
],
|
||
}
|
||
|
||
|
||
# ============================================================================
|
||
# MAIN HANDLER
|
||
# ============================================================================
|
||
|
||
async def handler(request: ApiRequest, ctx: FlowContext[Any]) -> ApiResponse:
|
||
"""
|
||
OpenAI-compatible Chat Completions endpoint.
|
||
|
||
Returns:
|
||
ApiResponse with chat completion or error
|
||
"""
|
||
ctx.logger.info("=" * 80)
|
||
ctx.logger.info("🤖 AI Chat Completions API")
|
||
ctx.logger.info("=" * 80)
|
||
|
||
try:
|
||
# 1. Parse and validate request
|
||
params = extract_request_params(request, ctx)
|
||
|
||
# 2. Check streaming (not supported)
|
||
if params['stream']:
|
||
return ApiResponse(
|
||
status=501,
|
||
body={
|
||
'error': {
|
||
'message': 'Streaming is not supported. Please set stream=false.',
|
||
'type': 'not_implemented',
|
||
'param': 'stream'
|
||
}
|
||
}
|
||
)
|
||
|
||
# 3. Resolve collection (explicit ID or Aktenzeichen lookup)
|
||
collection_id = await resolve_collection_id(
|
||
params['collection_id'],
|
||
params['messages'],
|
||
params['enable_web_search'],
|
||
ctx
|
||
)
|
||
|
||
# 4. Validate: collection or web_search required
|
||
if not collection_id and not params['enable_web_search']:
|
||
return ApiResponse(
|
||
status=400,
|
||
body={
|
||
'error': {
|
||
'message': 'Either collection_id or enable_web_search must be provided',
|
||
'type': 'invalid_request_error'
|
||
}
|
||
}
|
||
)
|
||
|
||
# 5. Initialize LLM with tools
|
||
model_with_tools = await initialize_model_with_tools(
|
||
model_name=params['model'],
|
||
temperature=params['temperature'],
|
||
max_tokens=params['max_tokens'],
|
||
collection_id=collection_id,
|
||
enable_web_search=params['enable_web_search'],
|
||
web_search_config=params['web_search_config'],
|
||
ctx=ctx
|
||
)
|
||
|
||
# 6. Invoke LLM
|
||
completion_id = f"chatcmpl-{int(time.time())}"
|
||
response = await invoke_and_format_response(
|
||
model=model_with_tools,
|
||
messages=params['messages'],
|
||
completion_id=completion_id,
|
||
model_name=params['model'],
|
||
ctx=ctx
|
||
)
|
||
|
||
ctx.logger.info(f"✅ Completion successful – {len(response.body['choices'][0]['message']['content'])} chars")
|
||
return response
|
||
|
||
except ValueError as e:
|
||
ctx.logger.error(f"❌ Validation error: {e}")
|
||
return ApiResponse(
|
||
status=400,
|
||
body={'error': {'message': str(e), 'type': 'invalid_request_error'}}
|
||
)
|
||
except Exception as e:
|
||
ctx.logger.error(f"❌ Error: {e}")
|
||
return ApiResponse(
|
||
status=500,
|
||
body={'error': {'message': 'Internal server error', 'type': 'server_error'}}
|
||
)
|
||
|
||
|
||
# ============================================================================
|
||
# REUSABLE HELPER FUNCTIONS
|
||
# ============================================================================
|
||
|
||
def extract_request_params(request: ApiRequest, ctx: FlowContext) -> Dict[str, Any]:
|
||
"""
|
||
Extract and validate request parameters.
|
||
|
||
Returns:
|
||
Dict with validated parameters
|
||
|
||
Raises:
|
||
ValueError: If validation fails
|
||
"""
|
||
body = request.body or {}
|
||
|
||
if not isinstance(body, dict):
|
||
raise ValueError("Request body must be JSON object")
|
||
|
||
messages = body.get('messages', [])
|
||
if not messages or not isinstance(messages, list):
|
||
raise ValueError("messages must be non-empty array")
|
||
|
||
# Extract parameters with defaults
|
||
params = {
|
||
'model': body.get('model', 'grok-4-1-fast-reasoning'),
|
||
'messages': messages,
|
||
'temperature': body.get('temperature', 0.7),
|
||
'max_tokens': body.get('max_tokens'),
|
||
'stream': body.get('stream', False),
|
||
'extra_body': body.get('extra_body', {}),
|
||
}
|
||
|
||
# Handle enable_web_search (body or extra_body)
|
||
params['enable_web_search'] = body.get(
|
||
'enable_web_search',
|
||
params['extra_body'].get('enable_web_search', False)
|
||
)
|
||
|
||
# Handle web_search_config
|
||
params['web_search_config'] = body.get(
|
||
'web_search_config',
|
||
params['extra_body'].get('web_search_config', {})
|
||
)
|
||
|
||
# Handle collection_id (multiple sources)
|
||
params['collection_id'] = (
|
||
body.get('collection_id') or
|
||
body.get('custom_collection_id') or
|
||
params['extra_body'].get('collection_id')
|
||
)
|
||
|
||
# Log concisely
|
||
ctx.logger.info(f"📋 Model: {params['model']} | Stream: {params['stream']}")
|
||
ctx.logger.info(f"📋 Web Search: {params['enable_web_search']} | Collection: {params['collection_id'] or 'auto'}")
|
||
ctx.logger.info(f"📨 Messages: {len(messages)}")
|
||
|
||
return params
|
||
|
||
|
||
async def resolve_collection_id(
|
||
explicit_collection_id: Optional[str],
|
||
messages: List[Dict[str, Any]],
|
||
enable_web_search: bool,
|
||
ctx: FlowContext
|
||
) -> Optional[str]:
|
||
"""
|
||
Resolve collection ID from explicit ID or Aktenzeichen auto-detection.
|
||
|
||
Args:
|
||
explicit_collection_id: Explicitly provided collection ID
|
||
messages: Chat messages (for Aktenzeichen extraction)
|
||
enable_web_search: Whether web search is enabled
|
||
ctx: Motia context
|
||
|
||
Returns:
|
||
Collection ID or None
|
||
"""
|
||
# Explicit collection ID takes precedence
|
||
if explicit_collection_id:
|
||
ctx.logger.info(f"🔍 Using explicit collection: {explicit_collection_id}")
|
||
return explicit_collection_id
|
||
|
||
# Try Aktenzeichen auto-detection from first user message
|
||
from services.aktenzeichen_utils import (
|
||
extract_aktenzeichen,
|
||
normalize_aktenzeichen,
|
||
remove_aktenzeichen
|
||
)
|
||
|
||
for msg in messages:
|
||
if msg.get('role') == 'user':
|
||
content = msg.get('content', '')
|
||
aktenzeichen_raw = extract_aktenzeichen(content)
|
||
|
||
if aktenzeichen_raw:
|
||
aktenzeichen = normalize_aktenzeichen(aktenzeichen_raw)
|
||
ctx.logger.info(f"🔍 Aktenzeichen detected: {aktenzeichen}")
|
||
|
||
collection_id = await lookup_collection_by_aktenzeichen(aktenzeichen, ctx)
|
||
|
||
if collection_id:
|
||
# Clean Aktenzeichen from message
|
||
msg['content'] = remove_aktenzeichen(content)
|
||
ctx.logger.info(f"✅ Collection found: {collection_id}")
|
||
return collection_id
|
||
else:
|
||
ctx.logger.warning(f"⚠️ No collection for Aktenzeichen: {aktenzeichen}")
|
||
break # Only check first user message
|
||
|
||
return None
|
||
|
||
|
||
async def initialize_model_with_tools(
|
||
model_name: str,
|
||
temperature: float,
|
||
max_tokens: Optional[int],
|
||
collection_id: Optional[str],
|
||
enable_web_search: bool,
|
||
web_search_config: Dict[str, Any],
|
||
ctx: FlowContext
|
||
) -> Any:
|
||
"""
|
||
Initialize LangChain model with tool bindings (file_search, web_search).
|
||
|
||
Returns:
|
||
Model instance with tools bound
|
||
"""
|
||
from services.langchain_xai_service import LangChainXAIService
|
||
|
||
service = LangChainXAIService(ctx)
|
||
|
||
# Create base model
|
||
model = service.get_chat_model(
|
||
model=model_name,
|
||
temperature=temperature,
|
||
max_tokens=max_tokens
|
||
)
|
||
|
||
# Bind tools
|
||
model_with_tools = service.bind_tools(
|
||
model=model,
|
||
collection_id=collection_id,
|
||
enable_web_search=enable_web_search,
|
||
web_search_config=web_search_config,
|
||
max_num_results=10
|
||
)
|
||
|
||
return model_with_tools
|
||
|
||
|
||
async def invoke_and_format_response(
|
||
model: Any,
|
||
messages: List[Dict[str, Any]],
|
||
completion_id: str,
|
||
model_name: str,
|
||
ctx: FlowContext
|
||
) -> ApiResponse:
|
||
"""
|
||
Invoke LLM and format response in OpenAI-compatible format.
|
||
|
||
Returns:
|
||
ApiResponse with chat completion
|
||
"""
|
||
from services.langchain_xai_service import LangChainXAIService
|
||
|
||
service = LangChainXAIService(ctx)
|
||
result = await service.invoke_chat(model, messages)
|
||
|
||
# Extract content (handle structured responses)
|
||
if hasattr(result, 'content'):
|
||
raw = result.content
|
||
if isinstance(raw, list):
|
||
# Extract text parts from structured response
|
||
text_parts = [
|
||
item.get('text', '')
|
||
for item in raw
|
||
if isinstance(item, dict) and item.get('type') == 'text'
|
||
]
|
||
content = ''.join(text_parts) or str(raw)
|
||
else:
|
||
content = raw
|
||
else:
|
||
content = str(result)
|
||
|
||
# Extract usage metadata (if available)
|
||
usage = {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0}
|
||
if hasattr(result, 'usage_metadata'):
|
||
u = result.usage_metadata
|
||
usage = {
|
||
"prompt_tokens": getattr(u, 'input_tokens', 0),
|
||
"completion_tokens": getattr(u, 'output_tokens', 0),
|
||
"total_tokens": getattr(u, 'input_tokens', 0) + getattr(u, 'output_tokens', 0)
|
||
}
|
||
|
||
# Log complete LLM response
|
||
ctx.logger.info("=" * 80)
|
||
ctx.logger.info("📤 LLM RESPONSE")
|
||
ctx.logger.info("-" * 80)
|
||
ctx.logger.info(f"Model: {model_name}")
|
||
ctx.logger.info(f"Completion ID: {completion_id}")
|
||
ctx.logger.info(f"Usage: {usage['prompt_tokens']} prompt + {usage['completion_tokens']} completion = {usage['total_tokens']} total tokens")
|
||
ctx.logger.info("-" * 80)
|
||
ctx.logger.info("Content:")
|
||
ctx.logger.info(content)
|
||
ctx.logger.info("=" * 80)
|
||
|
||
# Format OpenAI-compatible response
|
||
response_body = {
|
||
'id': completion_id,
|
||
'object': 'chat.completion',
|
||
'created': int(time.time()),
|
||
'model': model_name,
|
||
'choices': [{
|
||
'index': 0,
|
||
'message': {'role': 'assistant', 'content': content},
|
||
'finish_reason': 'stop'
|
||
}],
|
||
'usage': usage
|
||
}
|
||
|
||
return ApiResponse(status=200, body=response_body)
|
||
|
||
|
||
async def lookup_collection_by_aktenzeichen(
|
||
aktenzeichen: str,
|
||
ctx: FlowContext
|
||
) -> Optional[str]:
|
||
"""
|
||
Lookup xAI Collection ID by Aktenzeichen via EspoCRM.
|
||
|
||
Args:
|
||
aktenzeichen: Normalized Aktenzeichen (e.g., "1234/56")
|
||
ctx: Motia context
|
||
|
||
Returns:
|
||
Collection ID or None if not found
|
||
"""
|
||
try:
|
||
from services.espocrm import EspoCRMAPI
|
||
|
||
espocrm = EspoCRMAPI(ctx)
|
||
|
||
search_result = await espocrm.search_entities(
|
||
entity_type='Raeumungsklage',
|
||
where=[{
|
||
'type': 'equals',
|
||
'attribute': 'advowareAkteBezeichner',
|
||
'value': aktenzeichen
|
||
}],
|
||
select=['id', 'xaiCollectionId'],
|
||
maxSize=1
|
||
)
|
||
|
||
if search_result and len(search_result) > 0:
|
||
return search_result[0].get('xaiCollectionId')
|
||
|
||
return None
|
||
|
||
except Exception as e:
|
||
ctx.logger.error(f"❌ Collection lookup failed: {e}")
|
||
return None |