Create app.py

app.py

@@ -0,0 +1,168 @@
+# --- Imports ---
+import os
+import re
+import gradio as gr
+import openai
+from datetime import datetime
+from bs4 import BeautifulSoup
+
+# --- API Keys ---
+openai_api_key = os.environ.get("OPENAI_API_KEY")
+if not openai_api_key:
+    raise ValueError("OPENAI_API_KEY environment variable is not set.")
+
+client = openai.OpenAI(api_key=openai_api_key)
+
+# --- Exadata Specs ---
+exadata_specs = {
+    "X7": {"Quarter Rack": {"max_iops": 350000, "max_throughput": 25},
+           "Half Rack": {"max_iops": 700000, "max_throughput": 50},
+           "Full Rack": {"max_iops": 1400000, "max_throughput": 100}},
+    "X8": {"Quarter Rack": {"max_iops": 380000, "max_throughput": 28},
+           "Half Rack": {"max_iops": 760000, "max_throughput": 56},
+           "Full Rack": {"max_iops": 1520000, "max_throughput": 112}},
+    "X9": {"Quarter Rack": {"max_iops": 450000, "max_throughput": 30},
+           "Half Rack": {"max_iops": 900000, "max_throughput": 60},
+           "Full Rack": {"max_iops": 1800000, "max_throughput": 120}},
+    "X10": {"Quarter Rack": {"max_iops": 500000, "max_throughput": 35},
+            "Half Rack": {"max_iops": 1000000, "max_throughput": 70},
+            "Full Rack": {"max_iops": 2000000, "max_throughput": 140}},
+    "X11M": {"Quarter Rack": {"max_iops": 600000, "max_throughput": 40},
+             "Half Rack": {"max_iops": 1200000, "max_throughput": 80},
+             "Full Rack": {"max_iops": 2400000, "max_throughput": 160}},
+}
+
+# --- Preprocessor ---
+def clean_awr_content(content):
+    if "<html" in content.lower():
+        soup = BeautifulSoup(content, "html.parser")
+        text = soup.get_text()
+    else:
+        text = content
+    cleaned = "\n".join([line.strip() for line in text.splitlines() if line.strip()])
+    return cleaned
+
+
+# --- Agent Classes ---
+
+SYSTEM_PROMPT = """
+You are an expert Oracle Exadata and RAC performance consultant.
+Prioritize CRITICAL SYSTEM HEALTH issues first. Provide DBA-level observations and recommendations.
+"""
+
+class HealthRiskAgent:
+    def __init__(self, model="gpt-4o"):
+        self.model = model
+
+    def analyze(self, data):
+        prompt = f"""
+======== BEGIN DATA ========
+{data}
+======== END DATA ========
+
+Identify CRITICAL SYSTEM HEALTH issues (Flash Cache degraded, Confined Disks, Redo Stress, RAC GC waits, IO Errors). 
+If issues found, output "⚠️ CRITICAL ALERTS DETECTED" + Explanation + DBA Actions.
+If clean, output "✅ None Detected".
+"""
+        response = client.chat.completions.create(
+            model=self.model,
+            messages=[
+                {"role": "system", "content": SYSTEM_PROMPT},
+                {"role": "user", "content": prompt}
+            ]
+        )
+        return response.choices[0].message.content.strip()
+
+
+class PerformanceAnalyzerAgent:
+    def __init__(self, model="gpt-4o"):
+        self.model = model
+
+    def analyze(self, data, exadata_model=None, rack_size=None):
+        prompt = f"""
+======== BEGIN DATA ========
+{data}
+======== END DATA ========
+
+Please provide:
+- Performance Summary
+- Detailed Bottleneck Analysis
+- Forecast / Predictions
+- Monitoring Suggestions
+- Exadata Stats Summary
+- Recommended Next Steps
+
+If this is a performance test:
+- Compare observed vs theoretical for Exadata
+- Recommend gap-closing actions.
+"""
+        if exadata_model and rack_size:
+            specs = exadata_specs.get(exadata_model, {}).get(rack_size, {})
+            if specs:
+                prompt += f"""
+Theoretical Max for Oracle Exadata {exadata_model} {rack_size}:
+- Max IOPS: {specs['max_iops']}
+- Max Throughput: {specs['max_throughput']} GB/s
+"""
+        response = client.chat.completions.create(
+            model=self.model,
+            messages=[
+                {"role": "system", "content": SYSTEM_PROMPT},
+                {"role": "user", "content": prompt}
+            ]
+        )
+        return response.choices[0].message.content.strip()
+
+
+class AWRAgentCoordinator:
+    def __init__(self):
+        self.health_agent = HealthRiskAgent()
+        self.performance_agent = PerformanceAnalyzerAgent()
+
+    def analyze(self, awr_data, exadata_model=None, rack_size=None):
+        # Run both agents
+        health_result = self.health_agent.analyze(awr_data)
+        perf_result = self.performance_agent.analyze(awr_data, exadata_model, rack_size)
+
+        return health_result, perf_result
+
+
+# --- Gradio UI ---
+agent = AWRAgentCoordinator()
+
+with gr.Blocks() as demo:
+    gr.Markdown("# 📊 Exadata + RAC AWR Analyzer (Multi-Agent View)")
+
+    awr_text = gr.Textbox(label="Paste AWR Report (HTML or TXT)", lines=30, placeholder="Paste AWR report here...")
+    performance_test_mode = gr.Checkbox(label="Performance Test Mode")
+    exadata_model = gr.Dropdown(choices=["X7", "X8", "X9", "X10", "X11M"], label="Exadata Model", visible=False)
+    rack_size = gr.Dropdown(choices=["Quarter Rack", "Half Rack", "Full Rack"], label="Rack Size", visible=False)
+
+    def toggle_visibility(mode):
+        return gr.update(visible=mode), gr.update(visible=mode)
+
+    performance_test_mode.change(toggle_visibility, inputs=performance_test_mode, outputs=[exadata_model, rack_size])
+
+    analyze_btn = gr.Button("Analyze AWR Report")
+
+    with gr.Row():
+        health_output = gr.Textbox(label="Health Risk Agent (Critical Alerts + Actions)", lines=20)
+        performance_output = gr.Textbox(label="Performance Analyzer Agent (Full Analysis)", lines=20)
+
+    def run_analysis(awr_text, performance_test_mode, exadata_model, rack_size):
+        if not awr_text.strip():
+            return "❗ Please paste the AWR report first.", ""
+
+        cleaned = clean_awr_content(awr_text)
+
+        if performance_test_mode:
+            health, perf = agent.analyze(cleaned, exadata_model, rack_size)
+        else:
+            health, perf = agent.analyze(cleaned)
+
+        return health, perf
+
+    analyze_btn.click(run_analysis, inputs=[awr_text, performance_test_mode, exadata_model, rack_size],
+                      outputs=[health_output, performance_output])
+
+demo.launch(debug=True)