app.py

import os
import re
import time
import requests
import pandas as pd
import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt
import gradio as gr

from bs4 import BeautifulSoup
from langchain_core.documents import Document
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain_huggingface import HuggingFaceEmbeddings
from langchain_community.vectorstores import Chroma
from rank_bm25 import BM25Okapi
from sentence_transformers import CrossEncoder


PERSIST_DIR = "k8s_chroma_db"

URLS = {
    "pods": "https://kubernetes.io/docs/concepts/workloads/pods/",
    "deployments": "https://kubernetes.io/docs/concepts/workloads/controllers/deployment/",
    "services": "https://kubernetes.io/docs/concepts/services-networking/service/",
    "namespaces": "https://kubernetes.io/docs/concepts/overview/working-with-objects/namespaces/",
    "nodes": "https://kubernetes.io/docs/concepts/architecture/nodes/",
    "statefulsets": "https://kubernetes.io/docs/concepts/workloads/controllers/statefulset/",
    "rbac": "https://kubernetes.io/docs/reference/access-authn-authz/rbac/",
    "persistent-volumes": "https://kubernetes.io/docs/concepts/storage/persistent-volumes/",
    "ingress": "https://kubernetes.io/docs/concepts/services-networking/ingress/",
    "autoscaling": "https://kubernetes.io/docs/tasks/run-application/horizontal-pod-autoscale/",
}


# ================= Knowledge Base ================= #

def scrape_page(name, url):
    try:
        response = requests.get(url, timeout=20)
        response.raise_for_status()
        soup = BeautifulSoup(response.text, "html.parser")
        content = soup.find("div", class_="td-content")

        if not content:
            return None

        text = content.get_text(separator="\n").strip()
        return Document(page_content=text, metadata={"doc_id": name, "url": url})
    except Exception as e:
        print(f"[ERROR] scraping {url}: {e}")
        return None


def build_or_load_kb():
    print("[INFO] Loading embedding model...")
    embedding_model = HuggingFaceEmbeddings(
        model_name="sentence-transformers/all-MiniLM-L6-v2"
    )

    if os.path.isdir(PERSIST_DIR):
        print("[INFO] Loading existing vector DB...")
        vectordb = Chroma(
            embedding_function=embedding_model,
            persist_directory=PERSIST_DIR,
        )
        raw = vectordb._collection.get(include=["documents", "metadatas"])
        chunks = [
            Document(page_content=d, metadata=m)
            for d, m in zip(raw["documents"], raw["metadatas"])
        ]
        return vectordb, chunks

    print("[INFO] No DB found — scraping docs...")
    docs = []
    for name, url in URLS.items():
        doc = scrape_page(name, url)
        if doc:
            docs.append(doc)
    print(f"[INFO] Scraped {len(docs)} docs")

    splitter = RecursiveCharacterTextSplitter(chunk_size=900, chunk_overlap=200)
    chunks = splitter.split_documents(docs)

    vectordb = Chroma.from_documents(chunks, embedding_model, persist_directory=PERSIST_DIR)
    vectordb.persist()

    print("[INFO] Vector DB built & saved.")
    return vectordb, chunks


vectordb, chunks = build_or_load_kb()


# ================= Search & Reranker ================= #

bm25 = BM25Okapi([c.page_content.split() for c in chunks])
reranker = CrossEncoder("cross-encoder/ms-marco-MiniLM-L-12-v2")

retriever = vectordb.as_retriever(
    search_type="similarity_score_threshold",
    search_kwargs={"k": 8, "score_threshold": 0.35},
)


def hybrid_search(query, top_k=5):
    vector_results = retriever.invoke(query)

    bm_scores = bm25.get_scores(query.lower().split())
    bm_ranked = sorted(zip(bm_scores, chunks), reverse=True)
    bm_results = [doc for _, doc in bm_ranked[:top_k]]

    unique_docs = []
    seen = set()
    for doc in vector_results + bm_results:
        key = (doc.metadata.get("doc_id"), doc.page_content[:50])
        if key not in seen:
            seen.add(key)
            unique_docs.append(doc)

    if not unique_docs:
        return []

    rerank_pairs = [(query, doc.page_content) for doc in unique_docs]
    scores = reranker.predict(rerank_pairs)
    ranked = sorted(zip(scores, unique_docs), reverse=True)[:top_k]

    for s, doc in ranked:
        doc.metadata["rerank_score"] = float(s)

    return [doc for _, doc in ranked]


# ================= LLM ================= #

def call_llm(prompt):
    api_key = os.getenv("OPENROUTER_API_KEY")
    if not api_key:
        return "⚠️ Missing API key.\nGroundedness: 0%"

    try:
        res = requests.post(
            "https://openrouter.ai/api/v1/chat/completions",
            headers={
                "Authorization": f"Bearer {api_key}",
                "HTTP-Referer": "https://huggingface.co/",
                "X-Title": "Kubernetes RAG Assistant",
            },
            json={
                "model": "meta-llama/llama-3.1-8b-instruct",
                "messages": [{"role": "user", "content": prompt}],
                "temperature": 0.0,
                "max_tokens": 400,
            },
        )
        res.raise_for_status()
        data = res.json()
        return data["choices"][0]["message"]["content"]
    except Exception as e:
        return f"⚠️ LLM Error: {e}\nGroundedness: 0%"


# ================= Analytics ================= #

def classify_query(q):
    q = q.lower()
    if "how" in q:
        return "how-to"
    if "error" in q or "fail" in q:
        return "debug"
    return "general"


METRICS = {"q": [], "lat": [], "tok": [], "g": [], "r": [], "c": [], "t": []}


# ================= Chat Handler ================= #

def answer_question(query, history):
    start = time.time()
    docs = hybrid_search(query)

    if not docs:
        reply = "Not found in docs.\nGroundedness: 0%"
        return history + [
            {"role": "user", "content": query},
            {"role": "assistant", "content": reply}
        ], ""

    ctx = ""
    sources = []
    scores = []
    for i, d in enumerate(docs, 1):
        label = f"[{i}]"
        ctx += f"{label} {d.page_content[:900]}\nSource: {d.metadata['url']}\n\n"
        sources.append(f"{label} → {d.metadata['url']}")
        scores.append(d.metadata["rerank_score"])

    prompt = f"""
Answer the question ONLY using the context below.
Each sentence MUST end with a citation like [1].

Question: {query}

Context:
{ctx}

End with: Groundedness: XX%
"""

    answer = call_llm(prompt)
    latency = time.time() - start

    grounded = 0
    m = re.search(r"Groundedness:\s*(\d+)%", answer)
    if m:
        grounded = int(m.group(1"))

    cites = len(set(re.findall(r"\[(\d+)\]", answer)))
    avg_score = sum(scores) / len(scores) if scores else 0

    final = answer + "\n\n---\nSources:\n" + "\n".join(sources)

    METRICS["q"].append(query)
    METRICS["lat"].append(latency)
    METRICS["tok"].append(len(answer.split()))
    METRICS["g"].append(grounded)
    METRICS["r"].append(avg_score)
    METRICS["c"].append(cites)
    METRICS["t"].append(classify_query(query))

    history.append({"role": "user", "content": query})
    history.append({"role": "assistant", "content": final})
    return history, ""


def update_dashboard():
    rows = list(zip(
        range(1, len(METRICS["q"])+1),
        METRICS["q"],
        METRICS["lat"],
        METRICS["tok"],
        METRICS["g"],
        METRICS["r"],
        METRICS["c"],
        METRICS["t"],
    ))

    avgG = round(sum(METRICS["g"]) / len(METRICS["g"]), 2)
    avgL = round(sum(METRICS["lat"]) / len(METRICS["lat"]), 2)
    avgT = round(sum(METRICS["tok"]) / len(METRICS["tok"]), 2)

    return rows, avgG, avgL, avgT


# ================= UI ================= #

with gr.Blocks(title="Kubernetes RAG Assistant") as app:
    gr.Markdown("# ☸ Kubernetes RAG Assistant")

    with gr.Tab("Chat"):
        chat = gr.Chatbot(height=450)
        user_in = gr.Textbox(label="Ask anything about Kubernetes")
        reset = gr.Button("Reset")

        user_in.submit(answer_question, [user_in, chat], [chat, user_in])
        reset.click(lambda: ([], ""), None, [chat, user_in])

    with gr.Tab("Analytics"):
        gr.Markdown("### 📊 Analytics This Session")
        table = gr.DataFrame(headers=[
            "ID","Query","Latency","Tokens","Grounded","Rerank","Citations","Type"
        ], interactive=False)
        avgG = gr.Number(label="Avg Groundedness")
        avgL = gr.Number(label="Avg Latency")
        avgT = gr.Number(label="Avg Tokens")
        refresh = gr.Button("Refresh")
        refresh.click(update_dashboard, None, [table, avgG, avgL, avgT])

app.launch()