Update app.py

app.py

@@ -5,10 +5,11 @@
 VMware On-Prem → Azure Local Migration Assistant (Gradio)
 - No external API calls. No scikit-learn.
 - Upload design/migration docs (PDF/DOCX/TXT/MD).
-- Ask questions; get RELIABLE, DETAILED answers:
-  • Concept KB (for definitions like “What is Azure Arc-enabled SDN?”)
-  • RAG on uploaded docs (excerpts + gaps/fixes)
-  • Seeded FAQs (migration flows)
+- Ask questions; get RELIABLE, DETAILED, and RELEVANT answers:
+  • RAG on uploaded docs (excerpts + topic-tailored structure)
+  • Seeded FAQs (for migration flows)
+  • Topic-aware fallbacks (no more SDN leakage into unrelated topics)
+
 Run locally:
   pip install gradio PyPDF2 python-docx
   python app.py
@@ -18,7 +19,7 @@ import os
 import io
 import re
 import math
-from typing import List, Tuple, Dict, Any, Optional
+from typing import List, Tuple, Dict, Any
 from collections import Counter, defaultdict
 
 import gradio as gr
@@ -42,14 +43,24 @@ except Exception:
 # =========================
 
 TRUSTED_SOURCES: List[Tuple[str, str]] = [
+    # Core Azure landing/ops
+    ("Cloud Adoption Framework (CAF)", "https://learn.microsoft.com/azure/cloud-adoption-framework/"),
+    ("Azure Well-Architected Framework (WAF)", "https://learn.microsoft.com/azure/architecture/framework/"),
+    # Networking / SDN (used ONLY when topic == 'sdn')
+    ("Azure SDN concepts (HCI)", "https://learn.microsoft.com/azure-stack/hci/concepts/software-defined-networking"),
     ("Azure Arc (overview)", "https://learn.microsoft.com/azure/azure-arc/"),
     ("Azure Stack HCI (Azure Local)", "https://learn.microsoft.com/azure-stack/hci/"),
-    ("Azure SDN concepts (HCI)", "https://learn.microsoft.com/azure-stack/hci/concepts/software-defined-networking"),
+    ("Azure Virtual Network", "https://learn.microsoft.com/azure/virtual-network/"),
+    # Migration
     ("Azure VMware Solution (AVS)", "https://learn.microsoft.com/azure/azure-vmware/"),
     ("Azure Migrate", "https://learn.microsoft.com/azure/migrate/"),
-    ("Cloud Adoption Framework (CAF)", "https://learn.microsoft.com/azure/cloud-adoption-framework/"),
-    ("Azure Well-Architected Framework (WAF)", "https://learn.microsoft.com/azure/architecture/framework/"),
     ("VMware HCX Docs", "https://docs.vmware.com/en/VMware-HCX/index.html"),
+    # DR
+    ("Azure Site Recovery (ASR)", "https://learn.microsoft.com/azure/site-recovery/"),
+    # Security
+    ("Microsoft Defender for Cloud", "https://learn.microsoft.com/azure/defender-for-cloud/"),
+    # Cost
+    ("Azure Cost Management", "https://learn.microsoft.com/azure/cost-management-billing/"),
 ]
 
 FAQ_SEEDS: List[Dict[str, Any]] = [
@@ -77,7 +88,7 @@ FAQ_SEEDS: List[Dict[str, Any]] = [
             "Define RTO/RPO per app. Use immutable backups and soft-delete. "
             "Leverage ASR for DR where appropriate, run failover drills, and document rollback."
         ),
-        "refs": ["Azure Well-Architected Framework (WAF)"],
+        "refs": ["Azure Site Recovery (ASR)"],
     },
 ]
 
@@ -100,6 +111,41 @@ def list_refs(ref_names: List[str]) -> str:
     return " | ".join(links) if links else ""
 
 
+# =========================
+# Topic detection (keeps answers relevant)
+# =========================
+
+def detect_topic(q: str) -> str:
+    """
+    Returns one of: 'sdn', 'migration', 'dr', 'security', 'cost', 'general'
+    """
+    toks = set(tokenize(q))
+    if "sdn" in toks or "software-defined" in toks or "softwaredefined" in toks:
+        return "sdn"
+    if {"migrate", "migration", "hcx", "avs", "vmotion", "cutover"} & toks:
+        return "migration"
+    if {"dr", "disaster", "asr", "rto", "rpo", "failover"} & toks:
+        return "dr"
+    if {"defender", "sentinel", "pim", "mfa", "vault", "identity", "entra"} & toks:
+        return "security"
+    if {"cost", "reservation", "savings", "rightsizing", "tagging"} & toks:
+        return "cost"
+    return "general"
+
+def topic_refs(topic: str) -> List[str]:
+    if topic == "sdn":
+        return ["Azure SDN concepts (HCI)", "Azure Arc (overview)", "Azure Stack HCI (Azure Local)", "Azure Virtual Network"]
+    if topic == "migration":
+        return ["Azure Migrate", "Azure VMware Solution (AVS)", "VMware HCX Docs", "Cloud Adoption Framework (CAF)"]
+    if topic == "dr":
+        return ["Azure Site Recovery (ASR)", "Azure Well-Architected Framework (WAF)"]
+    if topic == "security":
+        return ["Microsoft Defender for Cloud", "Azure Well-Architected Framework (WAF)"]
+    if topic == "cost":
+        return ["Azure Cost Management", "Azure Well-Architected Framework (WAF)"]
+    return ["Cloud Adoption Framework (CAF)", "Azure Well-Architected Framework (WAF)"]
+
+
 # =========================
 # Tiny TF-IDF Index (no sklearn)
 # =========================
@@ -160,7 +206,7 @@ class TinyTfidfIndex:
 
 
 # =========================
-# Rubric for RAG-tailoring
+# Rubric for tailoring RAG output
 # =========================
 
 CHECKS = [
@@ -196,118 +242,6 @@ def score_text_against_checks(text: str) -> Tuple[Dict[str, float], List[Dict[st
     return scores, gaps
 
 
-# =========================
-# Built-in Concept KB (for definitional questions)
-# =========================
-
-class Concept:
-    def __init__(self, name: str, aliases: List[str], builder):
-        self.name = name
-        self.aliases = [tokenize(a) for a in aliases]
-        self.builder = builder  # function(query:str)->str
-
-def _kb_ans_azure_sdn(_: str) -> str:
-    refs = list_refs(["Azure SDN concepts (HCI)", "Azure Arc (overview)", "Azure Stack HCI (Azure Local)"])
-    return (
-        "### Azure SDN — What it is and why it matters\n"
-        "**Definition:** Azure SDN is Microsoft's software-defined networking stack that centralizes network control in software, "
-        "decoupling policy and management from physical hardware. It lets you programmatically create and secure virtual networks, "
-        "subnets, microsegmentation (ACL/NSG-like policies), load balancers and gateways across Azure and Azure Local (Azure Stack HCI) environments.\n\n"
-        "**Key capabilities**\n"
-        "- Central, policy-driven control plane for virtual networking resources.\n"
-        "- Automation & GitOps-friendly configuration for repeatable environments.\n"
-        "- Microsegmentation and traffic filtering for east–west security.\n"
-        "- Software load balancing and gateway services for app connectivity.\n"
-        "- Consistent constructs across cloud and on-prem (with Azure Local).\n\n"
-        "**How it works (high level)**\n"
-        "- A software control plane programs host virtual switches and network functions.\n"
-        "- Network intent (VNets, subnets, policies) is applied consistently across hosts.\n"
-        "- Integrates with Azure identity/management for RBAC and governance.\n\n"
-        "**Common use cases**\n"
-        "- Rapidly provisioning isolated app environments.\n"
-        "- Enforcing zero-trust style segmentation between tiers.\n"
-        "- Hybrid apps spanning Azure and Azure Local.\n\n"
-        f"**Trusted sources:** {refs}"
-    )
-
-def _kb_ans_arc_enabled_sdn(_: str) -> str:
-    refs = list_refs(["Azure SDN concepts (HCI)", "Azure Arc (overview)", "Azure Stack HCI (Azure Local)"])
-    return (
-        "### Azure Arc-enabled SDN — Definition & details\n"
-        "**Definition:** Azure Arc-enabled SDN brings Azure's software-defined networking to on-premises Azure Local (Azure Stack HCI) clusters, "
-        "managed through Azure Arc. It decouples network control from hardware so you can centrally define, automate, and secure "
-        "virtual networks, subnets, and policies in your datacenter using Azure-consistent tools.\n\n"
-        "**Why it matters**\n"
-        "- Gives you Azure-like VNet constructs and policy management on-prem.\n"
-        "- Enables consistent security and segmentation across hybrid estates.\n"
-        "- Supports rapid, software-driven changes without touching physical fabric.\n\n"
-        "**Key capabilities**\n"
-        "- Create/modify on-prem VNets, subnets, and routing policies from Azure.\n"
-        "- Apply microsegmentation rules (policy/ACL-style) for east–west security.\n"
-        "- Software load balancing and gateway services for north–south/east–west flows.\n"
-        "- Integration with Azure RBAC, tagging, and governance for change control.\n\n"
-        "**Core components (conceptual)**\n"
-        "- **Arc resource bridge & agents** — connect your HCI cluster to Azure control.\n"
-        "- **SDN controller & host agents** — program the Hyper-V vSwitch and network functions.\n"
-        "- **Azure portal/CLI/GitOps** — define intent (VNets, subnets, policies) and deploy.\n\n"
-        "**Prerequisites (typical)**\n"
-        "- Azure Local (Azure Stack HCI) cluster connected to Azure Arc.\n"
-        "- Arc resource bridge onboarded; network requirements met.\n"
-        "- Appropriate RBAC roles to manage networking resources.\n\n"
-        "**Use cases**\n"
-        "- Host Azure-consistent app networks on-prem for data locality/regulatory needs.\n"
-        "- Hybrid deployments with identical network constructs across Azure and HCI.\n"
-        "- Rapid rollout of segmented networks for dev/test/prod without hardware changes.\n\n"
-        "**Notes & limitations (high level)**\n"
-        "- Physical underlay still matters (IP design, routing, bandwidth, HA).\n"
-        "- Feature parity with public Azure services may vary; validate per release.\n\n"
-        f"**Trusted sources:** {refs}"
-    )
-
-KB_CONCEPTS: List[Concept] = [
-    Concept(
-        name="azure sdn",
-        aliases=[
-            "azure sdn",
-            "software defined networking azure",
-            "sdn in azure",
-            "azure local sdn",
-            "azure stack hci sdn",
-        ],
-        builder=_kb_ans_azure_sdn,
-    ),
-    Concept(
-        name="azure arc enabled sdn",
-        aliases=[
-            "azure arc enabled sdn",
-            "azure arc-enabled sdn",
-            "arc enabled sdn",
-            "arc-enabled sdn",
-            "arc sdn",
-            "azure local arc sdn",
-            "azure stack hci arc sdn",
-        ],
-        builder=_kb_ans_arc_enabled_sdn,
-    ),
-]
-
-def lookup_concept(query: str) -> Optional[Concept]:
-    q_tokens = set(tokenize(query))
-    best: Optional[Concept] = None
-    best_score = 0.0
-    for c in KB_CONCEPTS:
-        for alias_tokens in c.aliases:
-            if not alias_tokens:
-                continue
-            overlap = len(q_tokens & set(alias_tokens))
-            score = overlap / float(len(set(alias_tokens)))
-            if score > best_score:
-                best_score = score
-                best = c
-    # threshold: intentional but tolerant
-    return best if best_score >= 0.5 else None
-
-
 # =========================
 # File Parsing
 # =========================
@@ -360,31 +294,146 @@ def parse_file(file_obj: Dict[str, Any]) -> Dict[str, str]:
 
 
 # =========================
-# Detailed Answer Composer (for RAG path)
+# Helpers for composing detailed answers
 # =========================
 
-def _compose_detailed_from_snippets(query: str, snippets: List[Dict[str, str]]) -> str:
+def _extract_key_points(text: str, max_points: int = 6) -> List[str]:
+    # naive sentence splitter
+    parts = re.split(r"(?<=[.!?])\s+", text.strip())
+    points = []
+    for p in parts:
+        p = p.strip()
+        if 30 <= len(p) <= 300 and p not in points:
+            points.append(p)
+        if len(points) >= max_points:
+            break
+    return points
+
+def _topic_steps(topic: str) -> List[str]:
+    if topic == "sdn":
+        return [
+            "Define VNets/subnets and segmentation policy.",
+            "Automate configuration (ARM/Bicep/Terraform/GitOps).",
+            "Harden east–west flows with policy-based filtering.",
+            "Plan ingress/egress with load balancers and gateways.",
+            "Integrate with RBAC, logging, and change control.",
+        ]
+    if topic == "migration":
+        return [
+            "Establish governed landing zone (Policy, RBAC, logging).",
+            "Connect networks (ExpressRoute/VPN), validate DNS/MTU.",
+            "Discover/assess with Azure Migrate; classify apps.",
+            "Pilot 2–3 VMs; choose HCX or Azure Migrate cutover.",
+            "Migrate in waves; document rollback and success criteria.",
+        ]
+    if topic == "dr":
+        return [
+            "Define business RTO/RPO per workload.",
+            "Enable ASR where applicable; set up replication.",
+            "Run planned/unplanned failover drills; validate runbooks.",
+            "Harden backups (immutability, soft-delete).",
+            "Document recovery steps and responsibilities.",
+        ]
+    if topic == "security":
+        return [
+            "Centralize secrets in Key Vault; enable RBAC/PIM/MFA.",
+            "Enable Defender for Cloud and configure policies.",
+            "Collect/monitor logs; set alerts and playbooks.",
+            "Segment networks; restrict lateral movement.",
+            "Review identity hygiene and conditional access.",
+        ]
+    if topic == "cost":
+        return [
+            "Right-size compute/storage based on metrics.",
+            "Use reservations or Savings Plans where stable.",
+            "Automate tagging for showback/chargeback.",
+            "Schedule shutdowns for non-prod.",
+            "Monitor cost anomalies and budgets.",
+        ]
+    return [
+        "Clarify objective, constraints, and success criteria.",
+        "Assess current state and dependencies.",
+        "Choose the minimal viable approach first; pilot.",
+        "Define rollout plan, rollback, and verification.",
+        "Measure results and iterate.",
+    ]
+
+def _compose_detailed_from_snippets(query: str, snippets: List[Dict[str, str]], topic: str) -> str:
     combined = "\n\n".join([s.get("excerpt", "") for s in snippets])
     scores, gaps = score_text_against_checks(combined)
-    def _mk_gaps(glist):
-        return "\n".join([f"- ({g['severity']}) {g['id']}: {g['fix']}" for g in glist]) or "- No major issues detected."
-    refs = list_refs([s[0] for s in TRUSTED_SOURCES])
-    details = (
-        f"### Answer (detailed)\n"
-        f"**Your question:** {query}\n\n"
-        f"**Summary:** Migration planning must cover landing zone, connectivity, tooling, security, DR, and cost.\n\n"
-        f"#### Scores\nOverall: {scores.get('overall', 0)}/5.0\n\n"
-        f"#### Gaps & Fixes\n{_mk_gaps(gaps)}\n\n"
-        f"#### Supporting Excerpts\n"
-    )
+    points = _extract_key_points(combined, max_points=6)
+    refs = list_refs(topic_refs(topic))
+
+    md = [
+        "### Answer (detailed)",
+        f"**Your question:** {query}",
+        "",
+        "**Executive summary:**",
+    ]
+    if points:
+        for p in points:
+            md.append(f"- {p}")
+    else:
+        md.append("- Based on your documents, here is a structured plan and key considerations.")
+
+    md += [
+        "",
+        "#### Recommended steps",
+    ]
+    for step in _topic_steps(topic):
+        md.append(f"- {step}")
+
+    md += [
+        "",
+        "#### Supporting excerpts",
+    ]
     for s in snippets:
-        details += f"- {s['file']} (rel {s['relevance']:.2f}): {s['excerpt']}\n"
-    details += f"\n**Trusted sources:** {refs}"
-    return details
+        md.append(f"- **{s['file']}** (relevance {s['relevance']:.2f}): {s['excerpt']}")
+
+    md += [
+        "",
+        f"**Trusted sources:** {refs}"
+    ]
+
+    return "\n".join(md)
+
+def _compose_topic_fallback(query: str, topic: str) -> str:
+    refs = list_refs(topic_refs(topic))
+    headline = {
+        "sdn": "Azure SDN — Overview",
+        "migration": "Azure VMware / Azure Migrate — Overview",
+        "dr": "Azure Site Recovery (DR) — Overview",
+        "security": "Security & Governance in Azure — Overview",
+        "cost": "Cost Optimization in Azure — Overview",
+        "general": "Overview",
+    }[topic]
+
+    md = [
+        f"### {headline}",
+        f"**Your question:** {query}",
+        "",
+        "**Definition/Context:**",
+        "- What it is, the problem it solves, and where it runs (Azure / Azure Local).",
+        "",
+        "**Key capabilities / success factors:**",
+    ]
+    for step in _topic_steps(topic):
+        md.append(f"- {step}")
+
+    md += [
+        "",
+        "**Notes & caveats:**",
+        "- Validate limits and prerequisites for your environment.",
+        "- Align with governance and security baselines.",
+        "- Pilot before broad rollout.",
+        "",
+        f"**Trusted sources:** {refs}",
+    ]
+    return "\n".join(md)
 
 
 # =========================
-# Main Answer Function
+# Main Answer Function (no SDN bias)
 # =========================
 
 def answer_faq_or_approach_detailed(
@@ -398,15 +447,15 @@ def answer_faq_or_approach_detailed(
     if not q:
         return "Please enter a question."
 
-    # 0) Concept KB for definitional questions (e.g., "What is Azure Arc-enabled SDN?")
-    concept = lookup_concept(q)
-    if concept is not None:
-        return concept.builder(q)
+    topic = detect_topic(q)
 
-    # 1) Seeded FAQs → detailed plan when relevant (>=50% overlap with seed)
+    # 1) Seeded FAQs → detailed plan (only for migration-like questions)
     q_tokens = set(tokenize(q))
     for item in FAQ_SEEDS:
         seed_tokens = set(tokenize(item["q"]))
+        # require at least one migration-specific token to avoid hijacking definitional questions
+        if not ({"migrate", "migration", "hcx", "avs"} & q_tokens):
+            continue
         if seed_tokens and (len(seed_tokens & q_tokens) / float(len(seed_tokens))) >= 0.5:
             return (
                 "### Answer (detailed)\n"
@@ -429,20 +478,10 @@ def answer_faq_or_approach_detailed(
                 "excerpt": excerpt
             })
         if snippets:
-            return _compose_detailed_from_snippets(q, snippets)
-
-    # 3) Fallback (no docs) → generic, but structured overview (not migration-only)
-    refs = list_refs(["Azure Arc (overview)", "Azure Stack HCI (Azure Local)", "Azure SDN concepts (HCI)"])
-    return (
-        "### Answer (detailed)\n"
-        "I couldn't match a specific concept or supporting excerpts, so here's a structured overview you can refine:\n\n"
-        "**Definition:** Describe what the service/feature is, what problems it solves, and where it runs (Azure / Azure Local).\n\n"
-        "**Key capabilities:** automation, policy-driven control, security segmentation, connectivity services.\n\n"
-        "**How it works:** control plane programs host/network functions; policies applied consistently; integrates with RBAC/governance.\n\n"
-        "**Prerequisites:** identity/RBAC, connectivity to Azure (for Arc), supported host/cluster versions.\n\n"
-        "**Use cases:** hybrid deployments, zero-trust segmentation, rapid environment provisioning.\n\n"
-        f"**Trusted sources:** {refs}"
-    )
+            return _compose_detailed_from_snippets(q, snippets, topic)
+
+    # 3) Topic-aware fallback (no SDN unless you asked about SDN)
+    return _compose_topic_fallback(q, topic)
 
 
 # =========================
@@ -472,7 +511,9 @@ def build_index(files: List[Dict[str, Any]]):
 with gr.Blocks(title="VMware → Azure Migration Assistant", fill_height=True) as demo:
     gr.Markdown(
         "## VMware On-Prem → Azure Local Migration Assistant\n"
-        "Upload documents and ask questions. Detailed answers will be provided."
+        "- Upload documents (PDF/DOCX/TXT/MD)\n"
+        "- Click **Build Index**\n"
+        "- Ask a question. Answers are **detailed** and **topic-relevant**\n"
     )
 
     with gr.Row():
@@ -485,7 +526,7 @@ with gr.Blocks(title="VMware → Azure Migration Assistant", fill_height=True) a
             build_btn = gr.Button("Build Index", variant="primary")
 
         with gr.Column(scale=3):
-            question = gr.Textbox(label="Ask a question", placeholder="e.g., What is Azure Arc-enabled SDN, and why would I use it?")
+            question = gr.Textbox(label="Ask a question", placeholder="e.g., What is Azure Arc-enabled SDN?  or  What's the best way to minimize downtime for our AVS migration?")
             use_docs = gr.Checkbox(label="Use uploaded docs (RAG)", value=True)
             ask_btn = gr.Button("Ask", variant="primary")
             answer_box = gr.Markdown("")