File size: 6,939 Bytes
cd55ee8 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 |
import os
import torch
from qdrant_client import QdrantClient, models
from sentence_transformers import SentenceTransformer, CrossEncoder
from pymongo import MongoClient
from bson import ObjectId
from typing import List, Dict
import google.generativeai as genai
from groq import Groq
def build_content(doc: dict, entity_type: str) -> str:
"""Convert MongoDB document into natural text for embeddings."""
parts = [f"{entity_type} ID: {doc.get('id', str(doc.get('_id', '')))}"]
for k, v in doc.items():
if k in ["_id"]: # skip ObjectId
continue
if isinstance(v, list):
parts.append(f"{k}: {', '.join(map(str, v))}")
elif isinstance(v, dict):
nested = "; ".join([f"{nk}: {nv}" for nk, nv in v.items() if nv])
parts.append(f"{k}: {nested}")
else:
if v:
parts.append(f"{k}: {v}")
return "\n".join(parts)
class ErrorBot:
"""Chatbot using RAG (Qdrant + Gemini API)."""
def __init__(self, embedding_model_name: str, llm_model_name: str, google_api_key: str):
print("π Initializing ErrorBot...")
# --- Embedding model
self.device = "cuda" if torch.cuda.is_available() else "cpu"
print(f"Using device: {self.device}")
self.embedding_model = SentenceTransformer(embedding_model_name, device=self.device)
self.embedding_dim = self.embedding_model.get_sentence_embedding_dimension()
# --- Qdrant client
print("Connecting to Qdrant...")
self.qdrant = QdrantClient(
url=os.getenv("QDRANT_URL"),
api_key=os.getenv("QDRANT_API_KEY"),
)
self.collection_name = "technical_errors"
self._setup_collection()
# --- Gemini LLM
genai.configure(api_key=google_api_key)
self.llm_model_name = llm_model_name
self.llm = genai.GenerativeModel(llm_model_name)
# --- Cross encoder reranker
print("Loading cross-encoder reranker...")
self.reranker = CrossEncoder("cross-encoder/ms-marco-MiniLM-L-6-v2")
print("β
ErrorBot ready.")
def _setup_collection(self):
if not self.qdrant.collection_exists(self.collection_name):
self.qdrant.create_collection(
collection_name=self.collection_name,
vectors_config=models.VectorParams(
size=self.embedding_dim,
distance=models.Distance.COSINE,
),
)
def ingest_from_mongodb(self, mongo_uri: str, db_name: str, batch_size: int = 32):
client = MongoClient(mongo_uri)
db = client[db_name]
collections = {
"ProblemReport": db["problemReports"],
"FaultAnalysis": db["faultanalysis"],
"Correction": db["corrections"],
}
docs = []
for entity_type, coll in collections.items():
for doc in coll.find():
if "_id" in doc and isinstance(doc["_id"], ObjectId):
doc["_id"] = str(doc["_id"])
docs.append({"entity_type": entity_type, "data": doc})
contents = [build_content(d["data"], d["entity_type"]) for d in docs]
all_embeddings = []
for i in range(0, len(contents), batch_size):
batch_contents = contents[i:i + batch_size]
embeddings = self.embedding_model.encode(batch_contents, show_progress_bar=True).tolist()
all_embeddings.extend(embeddings)
self.qdrant.upsert(
collection_name=self.collection_name,
points=[
models.PointStruct(
id=i,
vector=emb,
payload={
"id": d["data"].get("id", str(d["data"].get("_id", i))),
"entity_type": d["entity_type"],
"raw": d["data"],
"content": c,
},
)
for i, (d, emb, c) in enumerate(zip(docs, all_embeddings, contents))
],
wait=True,
)
print(f"β
Ingested {len(docs)} documents into '{self.collection_name}'")
def retrieve(self, query: str, top_k: int = 5, score_threshold: float = 0.3, rerank: bool = True):
query_embedding = self.embedding_model.encode(query).tolist()
hits = self.qdrant.query_points(
collection_name=self.collection_name,
query=query_embedding,
limit=top_k * 3 if rerank else top_k,
with_payload=True,
score_threshold=score_threshold,
).points
candidates = [
{
"id": hit.payload.get("id"),
"entity_type": hit.payload.get("entity_type", ""),
"content": hit.payload.get("content", ""),
"score": hit.score,
}
for hit in hits
]
if rerank and candidates:
pairs = [(query, c["content"]) for c in candidates]
scores = self.reranker.predict(pairs)
for i, score in enumerate(scores):
candidates[i]["rerank_score"] = float(score)
candidates = sorted(candidates, key=lambda x: x["rerank_score"], reverse=True)
return candidates[:top_k]
def generate_answer(self, query: str, context: List[Dict], history: list = None):
context_str = "\n---\n".join(
[f"{c['entity_type']} (Score: {c['score']:.2f}):\n{c['content']}" for c in context]
)
convo_str = ""
if history:
for msg in history:
role = "User" if msg["role"] == "user" else "Assistant"
convo_str += f"{role}: {msg['content']}\n"
convo_str += f"User: {query}\nAssistant:"
prompt = f"""
You are a technical assistant. You have access to Problem Reports (PR), Fault Analyses (FA), and Corrections (CR).
Use the provided context and conversation history to answer the question clearly and concisely.
If context is not relevant, say you do not have enough information.
### Context
{context_str}
### Conversation
{convo_str}
"""
response = self.llm.generate_content(prompt)
return response.text.strip()
def ask(self, query: str, history: list = None):
print(f"\nβ Query: {query}")
retrieved_context = self.retrieve(query)
if not retrieved_context:
print("π¬ No relevant context found.")
return "I could not find any relevant information."
print(f"β
Retrieved {len(retrieved_context)} documents.")
for i, doc in enumerate(retrieved_context):
print(f" - Context {i+1} ({doc['entity_type']}, ID: {doc['id']}, Score: {doc['score']:.2f})")
answer = self.generate_answer(query, retrieved_context, history)
print(f"\nπ€ Answer: {answer}")
return answer
|