feat: add AuditPage for analyzing audit logs with visualization and m… #18

ChingEnLin · 2026-01-12T10:13:38Z

This pull request introduces a new audit log query and visualization feature to both the backend and frontend, enabling users to ask natural language questions about audit logs and receive summarized results with recommended chart visualizations. The backend now supports an /audit/query endpoint that translates user questions into SQL, executes them safely, and summarizes results. The frontend adds navigation for audit logs, unified authentication handling, and a new chart component for displaying visualizations.

Backend: Audit Log Query & Summarization

Added a new audit route (backend/routes/audit.py) with a /audit/query endpoint that accepts a natural language question, validates auth, and returns SQL, results, summary, and visualization config.
Implemented the audit service (backend/services/audit_service.py) to orchestrate: NL-to-SQL conversion, safe SQL execution (SELECT-only), and result summarization with visualization config using Gemini.
Extended Gemini service (backend/services/gemini_service.py) with prompt templates and logic for generating audit SQL and summarizing results, including a structured response for chart visualization. [1] [2]
Registered the audit router in the FastAPI app (backend/main.py). [1] [2]
Added unit tests for audit service logic (backend/tests/test_audit_service.py).

Frontend: Audit Log UI & Visualization

Added a new ChartDisplay component (frontend/components/ChartDisplay.tsx) to render bar, line, and pie charts based on backend-provided visualization config.
Updated navigation bar (frontend/components/NavBar.tsx) to include an "Audit Log" section and theme/user controls.
Implemented a unified authentication hook (frontend/hooks/useUnifiedAuth.ts) to support both MSAL and context-based auth flows.
Updated dependencies to include recharts and react-markdown for charting and markdown rendering. (frontend/package.json)

…arkdown support - Added AuditPage component with form for user queries - Integrated ReactMarkdown for displaying AI insights - Implemented data fetching from the API with error handling - Created AuditPage.module.css for styling the new page - Updated router to include protected route for AuditPage - Added new dependencies: react-markdown and recharts

Copilot

Pull request overview

This pull request adds a comprehensive audit log analysis feature that enables users to query audit logs using natural language, with AI-powered summarization and automatic visualization recommendations. The implementation spans both backend and frontend, introducing new endpoints, services, UI components, and authentication utilities.

Changes:

Added backend /audit/query endpoint with NL-to-SQL translation, safe query execution, and AI summarization with visualization configs
Implemented frontend AuditPage with chart visualization support using Recharts library
Created unified authentication hook to support both MSAL and context-based auth flows

Reviewed changes

Copilot reviewed 12 out of 13 changed files in this pull request and generated 19 comments.

Show a summary per file

File	Description
backend/routes/audit.py	New audit endpoint for processing natural language queries
backend/services/audit_service.py	Orchestrates SQL generation, execution, and result summarization
backend/services/gemini_service.py	Extended with audit SQL generation and result summarization using Gemini
backend/main.py	Registered audit router
backend/tests/test_audit_service.py	Unit tests for audit service
frontend/pages/AuditPage.tsx	New page for audit log querying with visualization
frontend/components/ChartDisplay.tsx	Recharts-based component for rendering bar, line, and pie charts
frontend/components/NavBar.tsx	Navigation bar with audit log link and theme controls
frontend/hooks/useUnifiedAuth.ts	Unified auth hook supporting MSAL and bypass modes
frontend/router.tsx	Added /audit route with protection
frontend/package.json	Added react-markdown and recharts dependencies

Files not reviewed (1)

frontend/package-lock.json: Language not supported

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-01-12T10:21:06Z

frontend/components/ChartDisplay.tsx

+    const chartData = data.map(item => ({
+        ...item,
+        // Create a display label for X axis if x_key exists
+        ...(config.x_key && { [config.x_key]: item[config.x_key] })
+    }));


The visualization data processing in ChartDisplay doesn't validate that the required keys (x_key, y_key) actually exist in the data. If the backend provides a visualization config with keys that don't match the actual data structure, this will cause rendering errors. Add validation to check that all referenced keys exist in the data before rendering.

Copilot · 2026-01-12T10:21:06Z

frontend/components/ChartDisplay.tsx

+                            cx="50%"
+                            cy="50%"
+                            labelLine={false}
+                            label={({ name, percent }) => `${name} ${(percent * 100).toFixed(0)}%`}


The pie chart label renders percentages but doesn't handle edge cases where the percentage calculation might fail or produce NaN. When data values are zero or invalid, the label expression (percent * 100).toFixed(0) could produce unexpected results. Add defensive checks for valid numeric values before rendering.

Suggested change

label={({ name, percent }) => `${name} ${(percent * 100).toFixed(0)}%`}

label={({ name, percent }) => {

const isValidPercent = typeof percent === 'number' && Number.isFinite(percent);

const percentValue = isValidPercent ? (percent * 100).toFixed(0) : '0';

return `${name} ${percentValue}%`;

}}

Copilot · 2026-01-12T10:21:06Z

backend/services/audit_service.py

+    # If the generator returned an error query or invalid SQL, return it
+    if "Error:" in sql_query:
+        return {
+            "sql_query": sql_query,
+            "results": [],
+            "summary": "Could not generate a valid query for your request."
+        }


The process_audit_question function checks for errors using string matching ("Error:" in sql_query) which is fragile and could produce false positives if legitimate queries contain the word "Error:". Use a more robust error handling mechanism such as custom exception types or a Result wrapper type.

Copilot · 2026-01-12T10:21:07Z

frontend/pages/AuditPage.tsx

+                                    <table className={styles.table}>
+                                        <thead>
+                                            <tr>
+                                                {Object.keys(data.results[0]).map((key) => (
+                                                    <th key={key}>{key}</th>
+                                                ))}
+                                            </tr>
+                                        </thead>
+                                        <tbody>
+                                            {data.results.map((row, i) => (
+                                                <tr key={i}>
+                                                    {Object.values(row).map((val: any, j) => (
+                                                        <td key={j}>
+                                                            {typeof val === 'object' ? JSON.stringify(val) : String(val)}
+                                                        </td>
+                                                    ))}
+                                                </tr>
+                                            ))}
+                                        </tbody>
+                                    </table>


The table rendering assumes all rows have the same keys as the first row (line 102). If the SQL results have inconsistent structure across rows (e.g., some rows with additional fields), this will cause incorrect rendering. Consider collecting all unique keys across all rows or handling missing keys gracefully.

Suggested change

<table className={styles.table}>

<thead>

<tr>

{Object.keys(data.results[0]).map((key) => (

<th key={key}>{key}</th>

))}

</tr>

</thead>

<tbody>

{data.results.map((row, i) => (

<tr key={i}>

{Object.values(row).map((val: any, j) => (

<td key={j}>

{typeof val === 'object' ? JSON.stringify(val) : String(val)}

</td>

))}

</tr>

))}

</tbody>

</table>

{(() => {

const allKeys = Array.from(

new Set(

data.results.flatMap((row) =>

row && typeof row === 'object'

? Object.keys(row)

: []

)

)

);

return (

<table className={styles.table}>

<thead>

<tr>

{allKeys.map((key) => (

<th key={key}>{key}</th>

))}

</tr>

</thead>

<tbody>

{data.results.map((row, i) => (

<tr key={i}>

{allKeys.map((key, j) => {

const val =

row && typeof row === 'object'

? (row as any)[key]

: undefined;

const display =

val === undefined || val === null

? ''

: typeof val === 'object'

? JSON.stringify(val)

: String(val);

return <td key={j}>{display}</td>;

})}

</tr>

))}

</tbody>

</table>

);

})()}

Copilot · 2026-01-12T10:21:07Z

backend/services/audit_service.py

+        conn = get_connection()
+        cur = conn.cursor()
+        cur.execute(sql_query)
+
+        # Get column names
+        columns = [desc[0] for desc in cur.description]
+        results = [dict(zip(columns, row)) for row in cur.fetchall()]
+
+        # Serialize datetime and json objects
+        for row in results:
+            for key, value in row.items():
+                if hasattr(value, 'isoformat'):
+                    row[key] = value.isoformat()
+                elif isinstance(value, dict):
+                    # Ensure dicts (like diff_data) are kept as dicts for the frontend
+                    pass
+
+        cur.close()
+        conn.close()
+        return results


The audit service closes database connections after each query execution. This is correct for preventing connection leaks, but the code should use a context manager pattern or try-finally block to ensure connections are properly closed even if an exception occurs before reaching the close() calls.

Copilot · 2026-01-12T10:21:10Z

backend/routes/audit.py

+
+    # We might want to validate the token here even if we don't use it for the pg connection directly yet
+    # user_token = authorization.replace("Bearer ", "")
+    # access_token = exchange_token_obo(user_token) 
+


The audit route extracts the user token but doesn't validate it or use it for authorization. The commented code suggests token validation was intended but not implemented. This leaves the endpoint vulnerable to unauthorized access. Either implement proper token validation using exchange_token_obo or add a clear TODO comment explaining why validation is deferred.

Suggested change

# We might want to validate the token here even if we don't use it for the pg connection directly yet

# user_token = authorization.replace("Bearer ", "")

# access_token = exchange_token_obo(user_token)

# Validate the user token using OBO flow; this ensures only authorized callers can access this route

user_token = authorization.replace("Bearer ", "").strip()

try:

# We currently don't use the exchanged access token further in this handler,

# but calling this function validates the incoming user token.

exchange_token_obo(user_token)

except Exception as exc:

raise HTTPException(status_code=401, detail="Invalid or expired token") from exc

Copilot · 2026-01-12T10:21:10Z

backend/routes/audit.py

+from pydantic import BaseModel
+from typing import List, Dict, Any, Optional
+from services.audit_service import process_audit_question
+from services.azure_auth import exchange_token_obo


Import of 'exchange_token_obo' is not used.

Suggested change

from services.azure_auth import exchange_token_obo

Copilot · 2026-01-12T10:21:11Z

backend/routes/audit.py

+from typing import List, Dict, Any, Optional
+from services.audit_service import process_audit_question
+from services.azure_auth import exchange_token_obo
+from services.gemini_service import VisualizationConfig, AuditSummaryResponse


Import of 'AuditSummaryResponse' is not used.

Suggested change

from services.gemini_service import VisualizationConfig, AuditSummaryResponse

from services.gemini_service import VisualizationConfig

Copilot · 2026-01-12T10:21:11Z

backend/services/audit_service.py

+from typing import Dict, Any
+from services.pg_connection import get_connection
+from services.gemini_service import generate_audit_sql, summarize_audit_results
+import json


Import of 'json' is not used.

Suggested change

import json

Copilot · 2026-01-12T10:21:11Z

backend/services/gemini_service.py

 from google.genai import types
 from models.schemas import GeneratedCode, CollectionContext, DebugSuggestionResponse
+from pydantic import BaseModel, Field
+from typing import Optional, List, Any, Dict


Import of 'Any' is not used.
Import of 'Dict' is not used.

Suggested change

from typing import Optional, List, Any, Dict

from typing import Optional, List

github-actions · 2026-01-12T11:01:42Z

🎉 This PR is included in version 2.5.0 🎉

The release is available on GitHub release

Your semantic-release bot 📦🚀

Copilot AI review requested due to automatic review settings January 12, 2026 10:13

Copilot started reviewing on behalf of ChingEnLin January 12, 2026 10:14 View session

Copilot AI reviewed Jan 12, 2026

View reviewed changes

ChingEnLin added 2 commits January 12, 2026 11:26

fix: static code analysis

d57c968

fix: mock return of test

495a85f

ChingEnLin merged commit 254e838 into dev Jan 12, 2026
3 checks passed

ChingEnLin deleted the feat/data_analysis branch January 12, 2026 10:59

github-actions bot added the released label Jan 12, 2026

-                            label={({ name, percent }) => `${name} ${(percent * 100).toFixed(0)}%`}
+                            label={({ name, percent }) => {
+                                const isValidPercent = typeof percent === 'number' && Number.isFinite(percent);
+                                const percentValue = isValidPercent ? (percent * 100).toFixed(0) : '0';
+                                return `${name} ${percentValue}%`;
+                            }}

-    # We might want to validate the token here even if we don't use it for the pg connection directly yet
-    # user_token = authorization.replace("Bearer ", "")
-    # access_token = exchange_token_obo(user_token)
+    # Validate the user token using OBO flow; this ensures only authorized callers can access this route
+    user_token = authorization.replace("Bearer ", "").strip()
+    try:
+        # We currently don't use the exchanged access token further in this handler,
+        # but calling this function validates the incoming user token.
+        exchange_token_obo(user_token)
+    except Exception as exc:
+        raise HTTPException(status_code=401, detail="Invalid or expired token") from exc

	from services.gemini_service import VisualizationConfig, AuditSummaryResponse
	from services.gemini_service import VisualizationConfig

	from typing import Optional, List, Any, Dict
	from typing import Optional, List

feat: add AuditPage for analyzing audit logs with visualization and m… #18

feat: add AuditPage for analyzing audit logs with visualization and m… #18

Uh oh!

Conversation

ChingEnLin commented Jan 12, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Jan 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 12, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

github-actions bot commented Jan 12, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants