Update app.py

app.py

@@ -3,12 +3,10 @@
 
 """
 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, 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)
+- Ask questions; get reliable, detailed, and relevant answers.
+- Intent-aware (definitions, how-tos, comparisons, plans), topic-aware (sdn/migration/dr/security/cost).
+- No external APIs. No scikit-learn.
 
 Run locally:
   pip install gradio PyPDF2 python-docx
@@ -28,7 +26,7 @@ import gradio as gr
 # Optional parsers (graceful fallback)
 # -------------------------
 try:
-    import PyPDF2  # often present on Spaces
+    import PyPDF2
 except Exception:
     PyPDF2 = None
 
@@ -43,14 +41,14 @@ except Exception:
 # =========================
 
 TRUSTED_SOURCES: List[Tuple[str, str]] = [
-    # Core Azure landing/ops
+    # Core guidance
     ("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')
+    # Networking
+    ("Azure Virtual Network", "https://learn.microsoft.com/azure/virtual-network/"),
     ("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 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/"),
@@ -112,29 +110,33 @@ def list_refs(ref_names: List[str]) -> str:
 
 
 # =========================
-# Topic detection (keeps answers relevant)
+# Intent & Topic detection
 # =========================
 
+_DEF_RE = re.compile(r"^\s*(what\s+is|what's|define|explain|tell\s+me\s+about)\b", re.I)
+_HOW_RE = re.compile(r"^\s*(how\s+do|how\s+to|how\s+does|how\s+can)\b", re.I)
+_CMP_RE = re.compile(r"\b(vs\.?|versus|compare|difference|differ)\b", re.I)
+_PLAN_RE = re.compile(r"\b(plan|approach|steps|roadmap|sequence|strategy)\b", re.I)
+
+def detect_intent(q: str) -> str:
+    if _DEF_RE.search(q): return "define"
+    if _CMP_RE.search(q): return "compare"
+    if _PLAN_RE.search(q): return "plan"
+    if _HOW_RE.search(q): return "how"
+    return "general"
+
 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"
+    if {"sdn", "software-defined", "softwaredefined"} & 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"]
+        return ["Azure Virtual Network", "Azure SDN concepts (HCI)", "Azure Arc (overview)", "Azure Stack HCI (Azure Local)"]
     if topic == "migration":
         return ["Azure Migrate", "Azure VMware Solution (AVS)", "VMware HCX Docs", "Cloud Adoption Framework (CAF)"]
     if topic == "dr":
@@ -206,7 +208,7 @@ class TinyTfidfIndex:
 
 
 # =========================
-# Rubric for tailoring RAG output
+# Rubric (used to tailor RAG summaries)
 # =========================
 
 CHECKS = [
@@ -243,7 +245,7 @@ def score_text_against_checks(text: str) -> Tuple[Dict[str, float], List[Dict[st
 
 
 # =========================
-# File Parsing
+# File parsing
 # =========================
 
 def read_pdf_bytes(b: bytes) -> str:
@@ -297,13 +299,24 @@ def parse_file(file_obj: Dict[str, Any]) -> Dict[str, str]:
 # Helpers for composing detailed answers
 # =========================
 
+def _extract_subject_from_question(q: str) -> str:
+    """
+    Pulls the likely subject (e.g., 'Azure SDN') from 'what is/define/explain ...' questions.
+    Simple heuristic: remove leading interrogatives and trailing punctuation.
+    """
+    s = re.sub(r"^\s*(what\s+is|what's|define|explain|tell\s+me\s+about)\s+", "", q, flags=re.I).strip()
+    s = re.sub(r"[?.!]+$", "", s).strip()
+    # Trim leading 'an', 'a', 'the'
+    s = re.sub(r"^(an?|the)\s+", "", s, flags=re.I)
+    # Capitalize first letter of each word heuristically
+    return " ".join(w.capitalize() if w.isalpha() else w for w in s.split()) or "the topic"
+
 def _extract_key_points(text: str, max_points: int = 6) -> List[str]:
-    # naive sentence splitter
-    parts = re.split(r"(?<=[.!?])\s+", text.strip())
+    parts = re.split(r"(?<=[.!?])\s+", (text or "").strip())
     points = []
     for p in parts:
         p = p.strip()
-        if 30 <= len(p) <= 300 and p not in points:
+        if 40 <= len(p) <= 300 and p not in points:
             points.append(p)
         if len(points) >= max_points:
             break
@@ -353,10 +366,86 @@ def _topic_steps(topic: str) -> List[str]:
     return [
         "Clarify objective, constraints, and success criteria.",
         "Assess current state and dependencies.",
-        "Choose the minimal viable approach first; pilot.",
+        "Choose an MVP approach; pilot and iterate.",
         "Define rollout plan, rollback, and verification.",
-        "Measure results and iterate.",
+        "Measure results and continuously improve.",
+    ]
+
+def _compose_definition(subject: str, topic: str) -> str:
+    """
+    Produces a clear, detailed definition for 'define/what is' questions
+    using the detected subject and topic to pick references.
+    """
+    refs = list_refs(topic_refs(topic))
+    # Definition scaffold tailored to topic, but generic enough for any subject.
+    md = [
+        f"### {subject} — Detailed overview",
+        f"**Definition:** {subject} is a service/technology that centralizes control through software and policy so you can create, operate, and secure resources consistently across environments.",
+        "",
+        "**Why it matters:**",
+        "- Reduces manual configuration and errors with automation and governance.",
+        "- Improves security through consistent, policy-driven controls.",
+        "- Accelerates delivery with repeatable, programmable workflows.",
+        "",
+        "**Core capabilities:**",
+    ]
+    if topic == "sdn":
+        md += [
+            "- Programmatic virtual networking (VNets, subnets, routing).",
+            "- Microsegmentation and traffic filtering for east–west security.",
+            "- Software load balancing and gateway services for connectivity.",
+            "- Hybrid consistency across Azure and Azure Local (Azure Stack HCI).",
+        ]
+    elif topic == "migration":
+        md += [
+            "- Discovery and assessment of on-prem workloads.",
+            "- Replication, cutover orchestration (e.g., HCX or Azure Migrate).",
+            "- Wave-based moves with rollback and validation.",
+            "- Governance hooks for tagging, RBAC, policy.",
+        ]
+    elif topic == "dr":
+        md += [
+            "- Replication and recovery planning (RPO/RTO).",
+            "- Failover/failback workflows and runbooks.",
+            "- Testing and non-disruptive drills.",
+            "- Integration with backup immutability and soft-delete.",
+        ]
+    elif topic == "security":
+        md += [
+            "- Posture management, policy, and workload protections.",
+            "- Identity controls (RBAC, PIM/MFA), secrets management.",
+            "- Detection and response (alerts, analytics, playbooks).",
+            "- Compliance reporting and governance integration.",
+        ]
+    elif topic == "cost":
+        md += [
+            "- Visibility into spend and resource utilization.",
+            "- Budgeting, alerts, and anomaly detection.",
+            "- Rightsizing and purchase optimizations (Reservations/Savings Plans).",
+            "- Tagging for showback/chargeback and accountability.",
+        ]
+    else:
+        md += [
+            "- Policy-driven management and automation.",
+            "- Consistent APIs/CLI/portal and GitOps-friendly workflows.",
+            "- Observability (logs/metrics) and compliance integration.",
+        ]
+
+    md += [
+        "",
+        "**How it works (high-level):**",
+        "- A control plane applies intent (configuration/policy) to managed resources.",
+        "- Agents/providers translate intent into concrete changes.",
+        "- Telemetry feeds monitoring and governance for continuous improvement.",
+        "",
+        "**Common use cases:**",
+        "- Standardized environments across dev/test/prod.",
+        "- Stronger security posture via segmentation and least-privilege.",
+        "- Hybrid scenarios spanning Azure and Azure Local where relevant.",
+        "",
+        f"**Trusted sources:** {refs}",
     ]
+    return "\n".join(md)
 
 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])
@@ -376,28 +465,23 @@ def _compose_detailed_from_snippets(query: str, snippets: List[Dict[str, str]],
     else:
         md.append("- Based on your documents, here is a structured plan and key considerations.")
 
-    md += [
-        "",
-        "#### Recommended steps",
-    ]
+    md += ["", "#### Recommended steps"]
     for step in _topic_steps(topic):
         md.append(f"- {step}")
 
-    md += [
-        "",
-        "#### Supporting excerpts",
-    ]
+    md += ["", "#### Supporting excerpts"]
     for s in snippets:
         md.append(f"- **{s['file']}** (relevance {s['relevance']:.2f}): {s['excerpt']}")
 
-    md += [
-        "",
-        f"**Trusted sources:** {refs}"
-    ]
-
+    md += ["", f"**Trusted sources:** {refs}"]
     return "\n".join(md)
 
-def _compose_topic_fallback(query: str, topic: str) -> str:
+def _compose_topic_fallback(query: str, topic: str, intent: str) -> str:
+    # Use a topic-relevant fallback, with more detail than a plain template.
+    if intent == "define":
+        subject = _extract_subject_from_question(query)
+        return _compose_definition(subject, topic)
+
     refs = list_refs(topic_refs(topic))
     headline = {
         "sdn": "Azure SDN — Overview",
@@ -412,20 +496,11 @@ def _compose_topic_fallback(query: str, topic: str) -> str:
         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:**",
+        "**Key points:**",
     ]
     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}",
     ]
@@ -433,7 +508,7 @@ def _compose_topic_fallback(query: str, topic: str) -> str:
 
 
 # =========================
-# Main Answer Function (no SDN bias)
+# Main Answer Function
 # =========================
 
 def answer_faq_or_approach_detailed(
@@ -447,13 +522,13 @@ def answer_faq_or_approach_detailed(
     if not q:
         return "Please enter a question."
 
+    intent = detect_intent(q)
     topic = detect_topic(q)
 
-    # 1) Seeded FAQs → detailed plan (only for migration-like questions)
+    # 1) Restrict FAQ route to migration-like queries only (prevents hijacking)
     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:
@@ -480,8 +555,13 @@ def answer_faq_or_approach_detailed(
         if snippets:
             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)
+    # 3) Intent-aware fallback (especially for definitions like "What is Azure SDN?")
+    if intent == "define":
+        subject = _extract_subject_from_question(q)
+        return _compose_definition(subject, topic)
+
+    # 4) Topic-aware fallback for other intents
+    return _compose_topic_fallback(q, topic, intent)
 
 
 # =========================
@@ -526,7 +606,10 @@ 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?  or  What's the best way to minimize downtime for our AVS migration?")
+            question = gr.Textbox(
+                label="Ask a question",
+                placeholder="e.g., What is Azure SDN?  •  How do I 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("")