Update app.py

app.py

@@ -1,21 +1,17 @@
-"""
-Governance‑GPT Quiz · Instruction‑tuned FLAN‑T5 summary with full logging
-"""
-
 import datetime, tempfile, re, traceback
 import gradio as gr
 import pandas as pd
 from fpdf import FPDF
 from transformers import pipeline
 
-print("[INIT] loading FLAN‑T5‑base … (≈20‑30 s on first start)")
+print("[INIT] loading FLAN-T5-base ...")
 summariser = pipeline(
     "text2text-generation",
     model="google/flan-t5-base",
     tokenizer="google/flan-t5-base",
     max_new_tokens=180,
 )
-print("[INIT] FLAN ready")
+print("[INIT] FLAN ready.")
 
 QUESTIONS = [
     "Governance framework is documented and communicated across the organisation.",
@@ -51,50 +47,46 @@ TIERS = {
     "Optimized":  (4.51,5.00),
 }
 
-def score_to_tier(x):  # helper
+def score_to_tier(x):
     for t,(lo,hi) in TIERS.items():
-        if lo<=x<=hi:
-            return t
+        if lo<=x<=hi: return t
     return "Unclassified"
 
-def latin1(t):  # PDF font safety
+def latin1(t):
     return (t.replace("–","-").replace("—","-").replace("•","-")
             .encode("latin-1","replace").decode("latin-1"))
 
-def llm_remediation(product,b_avgs,overall_tier):
-    bucket_txt="\n".join(f"- {b}: {v:.2f}" for b,v in b_avgs.items())
+def llm_remediation(product, b_avgs, overall_tier):
+    bucket_lines = "\n".join(f"{b}: {v:.2f}" for b,v in b_avgs.items())
     prompt = (
-        "You are an AI governance consultant.\n"
-        f"Product: {product}\n"
-        f"Overall tier: {overall_tier}\n"
-        "Bucket scores:\n"
-        f"{bucket_txt}\n\n"
-        "Provide:\n"
-        "1) One-sentence overall assessment.\n"
-        "2) 3‑5 bullet remediation actions referencing bucket names.\n"
-        "### Response:\n"
+        f"Summarise the governance maturity for the product '{product}' at tier '{overall_tier}' based on these bucket scores:\n"
+        f"{bucket_lines}\n\n"
+        "First, write a one-sentence overall assessment. Then, provide 3-5 markdown bullets suggesting next actions for improvement, mentioning bucket names."
     )
     try:
         print("[LLM] prompt >>>\n", prompt)
-        raw = summariser(prompt, temperature=0)[0]["generated_text"]
+        raw = summariser(prompt, temperature=0.0, do_sample=False)[0]["generated_text"]
         print("[LLM] raw >>>\n", raw)
     except Exception as e:
         print("[LLM] ERROR", e)
         traceback.print_exc()
         return "LLM summary unavailable."
 
-    # keep only text after the marker
-    cleaned = raw.split("### Response:")[-1].strip().replace("•","- ")
-    # very short => fallback
-    if len(cleaned) < 20:
-        cleaned = (f"{product} is at a '{overall_tier}' maturity level. "
-                   "Focus on formalising governance charters, embedding privacy impact assessments, "
-                   "and creating incident playbooks across all buckets.")
-    print("[LLM] cleaned >>>\n", cleaned)
-    return cleaned
+    # extract bullets if present, otherwise take everything
+    match = re.search(r"(?s)(.+?)(?:\n|$)(- .+)", raw)
+    if match:
+        summary = match.group(1).strip() + "\n" + match.group(2).strip()
+    else:
+        summary = raw.strip()
+    # ensure not too short
+    if len(summary) < 20:
+        summary = (f"{product} is at '{overall_tier}' level. "
+                   "Consider improving governance, privacy, risk and security processes in relevant buckets.")
+    print("[LLM] cleaned >>>\n", summary)
+    return summary
 
 def build_pdf(product,df,avg,tier,path,summary):
-    print("[PDF] building …")
+    print("[PDF] building ...")
     pdf=FPDF(); pdf.set_auto_page_break(auto=True,margin=15); pdf.add_page()
     pdf.set_font("Helvetica","B",16)
     pdf.cell(0,10,latin1(f"AI Governance Maturity Report - {product}"),ln=1,align="C")
@@ -104,7 +96,6 @@ def build_pdf(product,df,avg,tier,path,summary):
     pdf.cell(0,8,latin1(f"Overall Score: {avg:.2f} | Tier: {tier}"),ln=1)
     pdf.set_font("Helvetica","",11)
     pdf.multi_cell(0,6,latin1(summary)); pdf.ln(4)
-
     pdf.set_font("Helvetica","B",11)
     pdf.cell(80,8,"Bucket",1); pdf.cell(35,8,"Avg",1); pdf.cell(35,8,"Tier",1,ln=1)
     pdf.set_font("Helvetica","",10)
@@ -116,25 +107,26 @@ def build_pdf(product,df,avg,tier,path,summary):
     pdf.output(path)
     print("[PDF] saved at", path)
 
-def generate_report(name,*scores):
-    product=name.strip() or "your product"
-    scores=list(scores)
-    b_avgs={b:sum(scores[i] for i in idx)/len(idx) for b,idx in BUCKETS.items()}
-    avg=sum(scores)/len(scores); tier=score_to_tier(avg)
-    df=pd.DataFrame({"Bucket":b_avgs.keys(),
-                     "Avg":b_avgs.values(),
-                     "Tier":[score_to_tier(v) for v in b_avgs.values()]})
-    summary=llm_remediation(product,b_avgs,tier)
-    tmp=tempfile.NamedTemporaryFile(delete=False,suffix=".pdf")
-    build_pdf(product,df,avg,tier,tmp.name,summary)
-    return summary,tmp.name
+def generate_report(name, *scores):
+    product = name.strip() or "your product"
+    scores = list(scores)
+    b_avgs = {b: sum(scores[i] for i in idx)/len(idx) for b,idx in BUCKETS.items()}
+    avg = sum(scores)/len(scores)
+    tier = score_to_tier(avg)
+    df = pd.DataFrame({"Bucket":b_avgs.keys(),
+                       "Avg":b_avgs.values(),
+                       "Tier":[score_to_tier(v) for v in b_avgs.values()]})
+    summary = llm_remediation(product, b_avgs, tier)
+    tmp = tempfile.NamedTemporaryFile(delete=False,suffix=".pdf")
+    build_pdf(product, df, avg, tier, tmp.name, summary)
+    return summary, tmp.name
 
-with gr.Blocks(title="Governance‑GPT Quiz") as demo:
-    gr.Markdown("# Governance‑GPT Quiz")
-    pname=gr.Textbox(label="Product / System Name",placeholder="e.g. AcmeAI Recommender")
-    sliders=[gr.Slider(1,5,3,1,label=q) for q in QUESTIONS]
-    btn=gr.Button("Generate PDF Report")
-    md_out=gr.Markdown(); file_out=gr.File(label="⬇️ Download PDF")
+with gr.Blocks(title="Governance-GPT Quiz") as demo:
+    gr.Markdown("# Governance-GPT Quiz")
+    pname = gr.Textbox(label="Product / System Name",placeholder="e.g. AcmeAI Recommender")
+    sliders = [gr.Slider(1,5,3,1,label=q) for q in QUESTIONS]
+    btn = gr.Button("Generate PDF Report")
+    md_out = gr.Markdown(); file_out = gr.File(label="⬇️ Download PDF")
     btn.click(generate_report,[pname]+sliders,[md_out,file_out])
 
-demo.launch(share=True)   # share=True gives you a public link
+demo.launch(share=True)