Spaces:
Sleeping
Sleeping
Update main.py
Browse files
main.py
CHANGED
|
@@ -1,33 +1,63 @@
|
|
| 1 |
from fastapi import FastAPI, UploadFile, File, Form, HTTPException
|
| 2 |
from fastapi.responses import JSONResponse
|
|
|
|
| 3 |
from pydantic import BaseModel
|
| 4 |
-
from typing import Optional, List
|
| 5 |
from PIL import Image
|
| 6 |
import io
|
| 7 |
import numpy as np
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
|
| 9 |
from embedding_service import JinaClipEmbeddingService
|
| 10 |
from qdrant_service import QdrantVectorService
|
| 11 |
|
| 12 |
# Initialize FastAPI app
|
| 13 |
app = FastAPI(
|
| 14 |
-
title="Event Social Media Embeddings API",
|
| 15 |
-
description="API để embeddings
|
| 16 |
-
version="
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
)
|
| 18 |
|
| 19 |
# Initialize services
|
| 20 |
print("Initializing services...")
|
| 21 |
embedding_service = JinaClipEmbeddingService(model_path="jinaai/jina-clip-v2")
|
|
|
|
|
|
|
| 22 |
qdrant_service = QdrantVectorService(
|
| 23 |
-
|
| 24 |
-
collection_name="event_social_media",
|
| 25 |
vector_size=embedding_service.get_embedding_dimension()
|
| 26 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
print("✓ Services initialized successfully")
|
| 28 |
|
| 29 |
|
| 30 |
-
# Pydantic models
|
| 31 |
class SearchRequest(BaseModel):
|
| 32 |
text: Optional[str] = None
|
| 33 |
limit: int = 10
|
|
@@ -48,15 +78,62 @@ class IndexResponse(BaseModel):
|
|
| 48 |
message: str
|
| 49 |
|
| 50 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
@app.get("/")
|
| 52 |
async def root():
|
| 53 |
"""Health check endpoint"""
|
| 54 |
return {
|
| 55 |
"status": "running",
|
| 56 |
-
"service": "Event Social Media Embeddings API",
|
| 57 |
"embedding_model": "Jina CLIP v2",
|
| 58 |
"vector_db": "Qdrant",
|
| 59 |
-
"language_support": "Vietnamese + 88 other languages"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
}
|
| 61 |
|
| 62 |
|
|
@@ -342,6 +419,279 @@ async def get_stats():
|
|
| 342 |
raise HTTPException(status_code=500, detail=f"Lỗi khi lấy stats: {str(e)}")
|
| 343 |
|
| 344 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 345 |
if __name__ == "__main__":
|
| 346 |
import uvicorn
|
| 347 |
uvicorn.run(
|
|
|
|
| 1 |
from fastapi import FastAPI, UploadFile, File, Form, HTTPException
|
| 2 |
from fastapi.responses import JSONResponse
|
| 3 |
+
from fastapi.middleware.cors import CORSMiddleware
|
| 4 |
from pydantic import BaseModel
|
| 5 |
+
from typing import Optional, List, Dict
|
| 6 |
from PIL import Image
|
| 7 |
import io
|
| 8 |
import numpy as np
|
| 9 |
+
import os
|
| 10 |
+
from datetime import datetime
|
| 11 |
+
from pymongo import MongoClient
|
| 12 |
+
from huggingface_hub import InferenceClient
|
| 13 |
|
| 14 |
from embedding_service import JinaClipEmbeddingService
|
| 15 |
from qdrant_service import QdrantVectorService
|
| 16 |
|
| 17 |
# Initialize FastAPI app
|
| 18 |
app = FastAPI(
|
| 19 |
+
title="Event Social Media Embeddings & ChatbotRAG API",
|
| 20 |
+
description="API để embeddings, search và ChatbotRAG với Jina CLIP v2 + Qdrant + MongoDB + LLM",
|
| 21 |
+
version="2.0.0"
|
| 22 |
+
)
|
| 23 |
+
|
| 24 |
+
# CORS middleware
|
| 25 |
+
app.add_middleware(
|
| 26 |
+
CORSMiddleware,
|
| 27 |
+
allow_origins=["*"],
|
| 28 |
+
allow_credentials=True,
|
| 29 |
+
allow_methods=["*"],
|
| 30 |
+
allow_headers=["*"],
|
| 31 |
)
|
| 32 |
|
| 33 |
# Initialize services
|
| 34 |
print("Initializing services...")
|
| 35 |
embedding_service = JinaClipEmbeddingService(model_path="jinaai/jina-clip-v2")
|
| 36 |
+
|
| 37 |
+
collection_name = os.getenv("COLLECTION_NAME", "event_social_media")
|
| 38 |
qdrant_service = QdrantVectorService(
|
| 39 |
+
collection_name=collection_name,
|
|
|
|
| 40 |
vector_size=embedding_service.get_embedding_dimension()
|
| 41 |
)
|
| 42 |
+
print(f"✓ Qdrant collection: {collection_name}")
|
| 43 |
+
|
| 44 |
+
# MongoDB connection
|
| 45 |
+
mongodb_uri = os.getenv("MONGODB_URI", "mongodb+srv://truongtn7122003:7KaI9OT5KTUxWjVI@truongtn7122003.xogin4q.mongodb.net/")
|
| 46 |
+
mongo_client = MongoClient(mongodb_uri)
|
| 47 |
+
db = mongo_client[os.getenv("MONGODB_DB_NAME", "chatbot_rag")]
|
| 48 |
+
documents_collection = db["documents"]
|
| 49 |
+
chat_history_collection = db["chat_history"]
|
| 50 |
+
print("✓ MongoDB connected")
|
| 51 |
+
|
| 52 |
+
# Hugging Face token
|
| 53 |
+
hf_token = os.getenv("HUGGINGFACE_TOKEN")
|
| 54 |
+
if hf_token:
|
| 55 |
+
print("✓ Hugging Face token configured")
|
| 56 |
+
|
| 57 |
print("✓ Services initialized successfully")
|
| 58 |
|
| 59 |
|
| 60 |
+
# Pydantic models for embeddings
|
| 61 |
class SearchRequest(BaseModel):
|
| 62 |
text: Optional[str] = None
|
| 63 |
limit: int = 10
|
|
|
|
| 78 |
message: str
|
| 79 |
|
| 80 |
|
| 81 |
+
# Pydantic models for ChatbotRAG
|
| 82 |
+
class ChatRequest(BaseModel):
|
| 83 |
+
message: str
|
| 84 |
+
use_rag: bool = True
|
| 85 |
+
top_k: int = 3
|
| 86 |
+
system_message: Optional[str] = "You are a helpful AI assistant."
|
| 87 |
+
max_tokens: int = 512
|
| 88 |
+
temperature: float = 0.7
|
| 89 |
+
top_p: float = 0.95
|
| 90 |
+
hf_token: Optional[str] = None
|
| 91 |
+
|
| 92 |
+
|
| 93 |
+
class ChatResponse(BaseModel):
|
| 94 |
+
response: str
|
| 95 |
+
context_used: List[Dict]
|
| 96 |
+
timestamp: str
|
| 97 |
+
|
| 98 |
+
|
| 99 |
+
class AddDocumentRequest(BaseModel):
|
| 100 |
+
text: str
|
| 101 |
+
metadata: Optional[Dict] = None
|
| 102 |
+
|
| 103 |
+
|
| 104 |
+
class AddDocumentResponse(BaseModel):
|
| 105 |
+
success: bool
|
| 106 |
+
doc_id: str
|
| 107 |
+
message: str
|
| 108 |
+
|
| 109 |
+
|
| 110 |
@app.get("/")
|
| 111 |
async def root():
|
| 112 |
"""Health check endpoint"""
|
| 113 |
return {
|
| 114 |
"status": "running",
|
| 115 |
+
"service": "Event Social Media Embeddings & ChatbotRAG API",
|
| 116 |
"embedding_model": "Jina CLIP v2",
|
| 117 |
"vector_db": "Qdrant",
|
| 118 |
+
"language_support": "Vietnamese + 88 other languages",
|
| 119 |
+
"endpoints": {
|
| 120 |
+
"embeddings": {
|
| 121 |
+
"POST /index": "Index data với text/image",
|
| 122 |
+
"POST /search": "Hybrid search",
|
| 123 |
+
"POST /search/text": "Text search",
|
| 124 |
+
"POST /search/image": "Image search",
|
| 125 |
+
"DELETE /delete/{doc_id}": "Delete document",
|
| 126 |
+
"GET /document/{doc_id}": "Get document",
|
| 127 |
+
"GET /stats": "Collection statistics"
|
| 128 |
+
},
|
| 129 |
+
"chatbot_rag": {
|
| 130 |
+
"POST /chat": "Chat với RAG",
|
| 131 |
+
"POST /documents": "Add document to knowledge base",
|
| 132 |
+
"POST /rag/search": "Search in knowledge base",
|
| 133 |
+
"GET /history": "Get chat history",
|
| 134 |
+
"DELETE /documents/{doc_id}": "Delete document from knowledge base"
|
| 135 |
+
}
|
| 136 |
+
}
|
| 137 |
}
|
| 138 |
|
| 139 |
|
|
|
|
| 419 |
raise HTTPException(status_code=500, detail=f"Lỗi khi lấy stats: {str(e)}")
|
| 420 |
|
| 421 |
|
| 422 |
+
# ============================================
|
| 423 |
+
# ChatbotRAG Endpoints
|
| 424 |
+
# ============================================
|
| 425 |
+
|
| 426 |
+
@app.post("/chat", response_model=ChatResponse)
|
| 427 |
+
async def chat(request: ChatRequest):
|
| 428 |
+
"""
|
| 429 |
+
Chat endpoint với RAG
|
| 430 |
+
|
| 431 |
+
Body:
|
| 432 |
+
- message: User message
|
| 433 |
+
- use_rag: Enable RAG retrieval (default: true)
|
| 434 |
+
- top_k: Number of documents to retrieve (default: 3)
|
| 435 |
+
- system_message: System prompt (optional)
|
| 436 |
+
- max_tokens: Max tokens for response (default: 512)
|
| 437 |
+
- temperature: Temperature for generation (default: 0.7)
|
| 438 |
+
- hf_token: Hugging Face token (optional, sẽ dùng env nếu không truyền)
|
| 439 |
+
|
| 440 |
+
Returns:
|
| 441 |
+
- response: Generated response
|
| 442 |
+
- context_used: Retrieved context documents
|
| 443 |
+
- timestamp: Response timestamp
|
| 444 |
+
"""
|
| 445 |
+
try:
|
| 446 |
+
# Retrieve context if RAG enabled
|
| 447 |
+
context_used = []
|
| 448 |
+
if request.use_rag:
|
| 449 |
+
# Generate query embedding
|
| 450 |
+
query_embedding = embedding_service.encode_text(request.message)
|
| 451 |
+
|
| 452 |
+
# Search in Qdrant
|
| 453 |
+
results = qdrant_service.search(
|
| 454 |
+
query_embedding=query_embedding,
|
| 455 |
+
limit=request.top_k,
|
| 456 |
+
score_threshold=0.5
|
| 457 |
+
)
|
| 458 |
+
context_used = results
|
| 459 |
+
|
| 460 |
+
# Build context text
|
| 461 |
+
context_text = ""
|
| 462 |
+
if context_used:
|
| 463 |
+
context_text = "\n\nRelevant Context:\n"
|
| 464 |
+
for i, doc in enumerate(context_used, 1):
|
| 465 |
+
doc_text = doc["metadata"].get("text", "")
|
| 466 |
+
confidence = doc["confidence"]
|
| 467 |
+
context_text += f"\n[{i}] (Confidence: {confidence:.2f})\n{doc_text}\n"
|
| 468 |
+
|
| 469 |
+
# Add context to system message
|
| 470 |
+
system_message = f"{request.system_message}\n{context_text}\n\nPlease use the above context to answer the user's question when relevant."
|
| 471 |
+
else:
|
| 472 |
+
system_message = request.system_message
|
| 473 |
+
|
| 474 |
+
# Use token from request or fallback to env
|
| 475 |
+
token = request.hf_token or hf_token
|
| 476 |
+
|
| 477 |
+
# Generate response
|
| 478 |
+
if not token:
|
| 479 |
+
response = f"""[LLM Response Placeholder]
|
| 480 |
+
|
| 481 |
+
Context retrieved: {len(context_used)} documents
|
| 482 |
+
User question: {request.message}
|
| 483 |
+
|
| 484 |
+
To enable actual LLM generation:
|
| 485 |
+
1. Set HUGGINGFACE_TOKEN environment variable, OR
|
| 486 |
+
2. Pass hf_token in request body
|
| 487 |
+
|
| 488 |
+
Example:
|
| 489 |
+
{{
|
| 490 |
+
"message": "Your question",
|
| 491 |
+
"hf_token": "hf_xxxxxxxxxxxxx"
|
| 492 |
+
}}
|
| 493 |
+
"""
|
| 494 |
+
else:
|
| 495 |
+
try:
|
| 496 |
+
client = InferenceClient(
|
| 497 |
+
token=token,
|
| 498 |
+
model="openai/gpt-oss-20b"
|
| 499 |
+
)
|
| 500 |
+
|
| 501 |
+
# Build messages
|
| 502 |
+
messages = [
|
| 503 |
+
{"role": "system", "content": system_message},
|
| 504 |
+
{"role": "user", "content": request.message}
|
| 505 |
+
]
|
| 506 |
+
|
| 507 |
+
# Generate response
|
| 508 |
+
response = ""
|
| 509 |
+
for msg in client.chat_completion(
|
| 510 |
+
messages,
|
| 511 |
+
max_tokens=request.max_tokens,
|
| 512 |
+
stream=True,
|
| 513 |
+
temperature=request.temperature,
|
| 514 |
+
top_p=request.top_p,
|
| 515 |
+
):
|
| 516 |
+
choices = msg.choices
|
| 517 |
+
if len(choices) and choices[0].delta.content:
|
| 518 |
+
response += choices[0].delta.content
|
| 519 |
+
|
| 520 |
+
except Exception as e:
|
| 521 |
+
response = f"Error generating response with LLM: {str(e)}\n\nContext was retrieved successfully, but LLM generation failed."
|
| 522 |
+
|
| 523 |
+
# Save to history
|
| 524 |
+
chat_data = {
|
| 525 |
+
"user_message": request.message,
|
| 526 |
+
"assistant_response": response,
|
| 527 |
+
"context_used": context_used,
|
| 528 |
+
"timestamp": datetime.utcnow()
|
| 529 |
+
}
|
| 530 |
+
chat_history_collection.insert_one(chat_data)
|
| 531 |
+
|
| 532 |
+
return ChatResponse(
|
| 533 |
+
response=response,
|
| 534 |
+
context_used=context_used,
|
| 535 |
+
timestamp=datetime.utcnow().isoformat()
|
| 536 |
+
)
|
| 537 |
+
|
| 538 |
+
except Exception as e:
|
| 539 |
+
raise HTTPException(status_code=500, detail=f"Error: {str(e)}")
|
| 540 |
+
|
| 541 |
+
|
| 542 |
+
@app.post("/documents", response_model=AddDocumentResponse)
|
| 543 |
+
async def add_document(request: AddDocumentRequest):
|
| 544 |
+
"""
|
| 545 |
+
Add document to knowledge base
|
| 546 |
+
|
| 547 |
+
Body:
|
| 548 |
+
- text: Document text
|
| 549 |
+
- metadata: Additional metadata (optional)
|
| 550 |
+
|
| 551 |
+
Returns:
|
| 552 |
+
- success: True/False
|
| 553 |
+
- doc_id: MongoDB document ID
|
| 554 |
+
- message: Status message
|
| 555 |
+
"""
|
| 556 |
+
try:
|
| 557 |
+
# Save to MongoDB
|
| 558 |
+
doc_data = {
|
| 559 |
+
"text": request.text,
|
| 560 |
+
"metadata": request.metadata or {},
|
| 561 |
+
"created_at": datetime.utcnow()
|
| 562 |
+
}
|
| 563 |
+
result = documents_collection.insert_one(doc_data)
|
| 564 |
+
doc_id = str(result.inserted_id)
|
| 565 |
+
|
| 566 |
+
# Generate embedding
|
| 567 |
+
embedding = embedding_service.encode_text(request.text)
|
| 568 |
+
|
| 569 |
+
# Index to Qdrant
|
| 570 |
+
qdrant_service.index_data(
|
| 571 |
+
doc_id=doc_id,
|
| 572 |
+
embedding=embedding,
|
| 573 |
+
metadata={
|
| 574 |
+
"text": request.text,
|
| 575 |
+
"source": "api",
|
| 576 |
+
**(request.metadata or {})
|
| 577 |
+
}
|
| 578 |
+
)
|
| 579 |
+
|
| 580 |
+
return AddDocumentResponse(
|
| 581 |
+
success=True,
|
| 582 |
+
doc_id=doc_id,
|
| 583 |
+
message=f"Document added successfully with ID: {doc_id}"
|
| 584 |
+
)
|
| 585 |
+
|
| 586 |
+
except Exception as e:
|
| 587 |
+
raise HTTPException(status_code=500, detail=f"Error: {str(e)}")
|
| 588 |
+
|
| 589 |
+
|
| 590 |
+
@app.post("/rag/search", response_model=List[SearchResponse])
|
| 591 |
+
async def rag_search(
|
| 592 |
+
query: str = Form(...),
|
| 593 |
+
top_k: int = Form(5),
|
| 594 |
+
score_threshold: Optional[float] = Form(0.5)
|
| 595 |
+
):
|
| 596 |
+
"""
|
| 597 |
+
Search in knowledge base
|
| 598 |
+
|
| 599 |
+
Body:
|
| 600 |
+
- query: Search query
|
| 601 |
+
- top_k: Number of results (default: 5)
|
| 602 |
+
- score_threshold: Minimum score (default: 0.5)
|
| 603 |
+
|
| 604 |
+
Returns:
|
| 605 |
+
- results: List of matching documents
|
| 606 |
+
"""
|
| 607 |
+
try:
|
| 608 |
+
# Generate query embedding
|
| 609 |
+
query_embedding = embedding_service.encode_text(query)
|
| 610 |
+
|
| 611 |
+
# Search in Qdrant
|
| 612 |
+
results = qdrant_service.search(
|
| 613 |
+
query_embedding=query_embedding,
|
| 614 |
+
limit=top_k,
|
| 615 |
+
score_threshold=score_threshold
|
| 616 |
+
)
|
| 617 |
+
|
| 618 |
+
return [
|
| 619 |
+
SearchResponse(
|
| 620 |
+
id=result["id"],
|
| 621 |
+
confidence=result["confidence"],
|
| 622 |
+
metadata=result["metadata"]
|
| 623 |
+
)
|
| 624 |
+
for result in results
|
| 625 |
+
]
|
| 626 |
+
|
| 627 |
+
except Exception as e:
|
| 628 |
+
raise HTTPException(status_code=500, detail=f"Error: {str(e)}")
|
| 629 |
+
|
| 630 |
+
|
| 631 |
+
@app.get("/history")
|
| 632 |
+
async def get_history(limit: int = 10, skip: int = 0):
|
| 633 |
+
"""
|
| 634 |
+
Get chat history
|
| 635 |
+
|
| 636 |
+
Query params:
|
| 637 |
+
- limit: Number of messages to return (default: 10)
|
| 638 |
+
- skip: Number of messages to skip (default: 0)
|
| 639 |
+
|
| 640 |
+
Returns:
|
| 641 |
+
- history: List of chat messages
|
| 642 |
+
"""
|
| 643 |
+
try:
|
| 644 |
+
history = list(
|
| 645 |
+
chat_history_collection
|
| 646 |
+
.find({}, {"_id": 0})
|
| 647 |
+
.sort("timestamp", -1)
|
| 648 |
+
.skip(skip)
|
| 649 |
+
.limit(limit)
|
| 650 |
+
)
|
| 651 |
+
|
| 652 |
+
# Convert datetime to string
|
| 653 |
+
for msg in history:
|
| 654 |
+
if "timestamp" in msg:
|
| 655 |
+
msg["timestamp"] = msg["timestamp"].isoformat()
|
| 656 |
+
|
| 657 |
+
return {
|
| 658 |
+
"history": history,
|
| 659 |
+
"total": chat_history_collection.count_documents({})
|
| 660 |
+
}
|
| 661 |
+
|
| 662 |
+
except Exception as e:
|
| 663 |
+
raise HTTPException(status_code=500, detail=f"Error: {str(e)}")
|
| 664 |
+
|
| 665 |
+
|
| 666 |
+
@app.delete("/documents/{doc_id}")
|
| 667 |
+
async def delete_document_from_kb(doc_id: str):
|
| 668 |
+
"""
|
| 669 |
+
Delete document from knowledge base
|
| 670 |
+
|
| 671 |
+
Args:
|
| 672 |
+
- doc_id: Document ID (MongoDB ObjectId)
|
| 673 |
+
|
| 674 |
+
Returns:
|
| 675 |
+
- success: True/False
|
| 676 |
+
- message: Status message
|
| 677 |
+
"""
|
| 678 |
+
try:
|
| 679 |
+
# Delete from MongoDB
|
| 680 |
+
result = documents_collection.delete_one({"_id": doc_id})
|
| 681 |
+
|
| 682 |
+
# Delete from Qdrant
|
| 683 |
+
if result.deleted_count > 0:
|
| 684 |
+
qdrant_service.delete_by_id(doc_id)
|
| 685 |
+
return {"success": True, "message": f"Document {doc_id} deleted from knowledge base"}
|
| 686 |
+
else:
|
| 687 |
+
raise HTTPException(status_code=404, detail=f"Document {doc_id} not found")
|
| 688 |
+
|
| 689 |
+
except HTTPException:
|
| 690 |
+
raise
|
| 691 |
+
except Exception as e:
|
| 692 |
+
raise HTTPException(status_code=500, detail=f"Error: {str(e)}")
|
| 693 |
+
|
| 694 |
+
|
| 695 |
if __name__ == "__main__":
|
| 696 |
import uvicorn
|
| 697 |
uvicorn.run(
|