add pdf report route

.github/workflows/docker.yml

@@ -2,7 +2,7 @@ name: Docker
 
 on:
     push:
-        branches: [main]
+        branches: [convert-docx-to-pdf]
 
 jobs:
     build-backend:
@@ -27,16 +27,16 @@ jobs:
               with:
                   context: ./backend
                   push: true
-                  tags: ghcr.io/socfortress/copilot-backend:latest
+                  tags: ghcr.io/socfortress/copilot-backend:lab
                   build-args: |
                       COPILOT_API_KEY=${{ secrets.COPILOT_API_KEY }}
 
-            - name: Notify Discord
-              uses: appleboy/[email protected]
-              with:
-                  webhook_id: ${{ secrets.DISCORD_WEBHOOK_ID }}
-                  webhook_token: ${{ secrets.DISCORD_WEBHOOK_TOKEN }}
-                  message: "Docker image for backend has been updated."
+            # - name: Notify Discord
+            #   uses: appleboy/[email protected]
+            #   with:
+            #       webhook_id: ${{ secrets.DISCORD_WEBHOOK_ID }}
+            #       webhook_token: ${{ secrets.DISCORD_WEBHOOK_TOKEN }}
+            #       message: "Docker image for backend has been updated."
 
     build-frontend:
         runs-on: ubuntu-latest
@@ -60,11 +60,11 @@ jobs:
               with:
                   context: ./frontend
                   push: true
-                  tags: ghcr.io/socfortress/copilot-frontend:latest
+                  tags: ghcr.io/socfortress/copilot-frontend:lab
 
-            - name: Notify Discord
-              uses: appleboy/[email protected]
-              with:
-                  webhook_id: ${{ secrets.DISCORD_WEBHOOK_ID }}
-                  webhook_token: ${{ secrets.DISCORD_WEBHOOK_TOKEN }}
-                  message: "Docker image for frontend has been updated."
+            # - name: Notify Discord
+            #   uses: appleboy/[email protected]
+            #   with:
+            #       webhook_id: ${{ secrets.DISCORD_WEBHOOK_ID }}
+            #       webhook_token: ${{ secrets.DISCORD_WEBHOOK_TOKEN }}
+            #       message: "Docker image for frontend has been updated."

backend/app/incidents/routes/db_operations.py

@@ -933,13 +933,18 @@ async def upload_case_report_template_endpoint(
     file: UploadFile = File(...),
     db: AsyncSession = Depends(get_db),
 ):
-    # Check if the file type is a .docx
+    # Check if the file type is a .docx or .html
     mime_type, _ = mimetypes.guess_type(file.filename)
-    if mime_type != "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
-        raise HTTPException(status_code=400, detail="Invalid file type. Only .docx files are allowed.")
+    allowed_mime_types = [
+        "application/vnd.openxmlformats-officedocument.wordprocessingml.document",  # .docx
+        "text/html"  # .html
+    ]
+    if mime_type not in allowed_mime_types:
+        raise HTTPException(status_code=400, detail="Invalid file type. Only .docx and .html files are allowed.")
 
     if await report_template_exists(file.filename, db):
         raise HTTPException(status_code=400, detail="File name already exists for this template")
+
     return CaseReportTemplateDataStoreResponse(
         case_report_template_data_store=await upload_report_template(file, db),
         success=True,

backend/app/incidents/routes/incident_report.py

@@ -29,6 +29,12 @@
 from app.incidents.services.reports import download_template
 from app.incidents.services.reports import render_document_with_context
 from app.incidents.services.reports import save_template_to_tempfile
+from app.incidents.services.reports_pdf import convert_html_to_pdf
+from app.incidents.services.reports_pdf import create_file_response_pdf
+from app.incidents.services.reports_pdf import download_template_pdf
+from app.incidents.services.reports_pdf import render_html_template
+from app.incidents.services.reports_pdf import create_case_context_pdf
+
 
 incidents_report_router = APIRouter()
 
@@ -202,5 +208,36 @@ async def get_cases_export_docx_route(
 
     return response
 
+@incidents_report_router.post(
+    "/generate-report-pdf",
+    description="Generate a PDF report for a case.",
+)
+async def get_cases_export_pdf_route(
+    request: CaseDownloadDocxRequest,
+    session: AsyncSession = Depends(get_db),
+) -> FileResponse:
+    case = await fetch_case_by_id(session, request.case_id)
+    if not case:
+        raise HTTPException(status_code=404, detail="No cases found")
+
+    context = create_case_context_pdf(case)
+
+    # Download and save the template
+    tmp_template_name = await download_template_pdf(request.template_name)
+
+    # Render the HTML template with the context
+    rendered_html_file_name = render_html_template(tmp_template_name, context)
+
+    # Convert HTML to PDF using WeasyPrint
+    rendered_pdf_file_name = convert_html_to_pdf(rendered_html_file_name)
+
+    # Create the FileResponse for PDF
+    response = create_file_response_pdf(file_path=rendered_pdf_file_name, file_name=request.file_name.replace(".docx", ".pdf"))
+
+    # Clean up temporary files
+    cleanup_temp_files([tmp_template_name, rendered_html_file_name])
+
+    return response
+
 
 # ! TODO: ROUTE FOR MARKDOWN TEMPLATE ! #

backend/app/incidents/services/reports_pdf.py

@@ -0,0 +1,145 @@
+import os
+import pdfkit
+from tempfile import NamedTemporaryFile
+from typing import Dict, Optional
+from jinja2 import Environment, FileSystemLoader
+from fastapi.responses import FileResponse
+import platform
+from app.data_store.data_store_operations import download_data_store
+
+async def download_template_pdf(template_name: str) -> str:
+    """Retrieve the template file content from the data store and save it to a temporary file."""
+    template_content = await download_data_store(bucket_name="copilot-case-report-templates", object_name=template_name)
+    with NamedTemporaryFile(delete=False, suffix=".html") as tmp_template:
+        tmp_template.write(template_content)
+        return tmp_template.name
+
+def create_case_context_pdf(case) -> Dict[str, Dict[str, str]]:
+    """Prepare the context for the Jinja template."""
+    return {
+        "case": {
+            "name": case.case_name,
+            "description": case.case_description,
+            "assigned_to": case.assigned_to,
+            "case_creation_time": case.case_creation_time,
+            "id": case.id,
+            "alerts": [
+                {
+                    "alert_name": alert.alert.alert_name,
+                    "alert_description": alert.alert.alert_description,
+                    "status": alert.alert.status,
+                    "tags": [tag.tag.tag for tag in alert.alert.tags],
+                    "assets": [
+                        {
+                            "asset_name": asset.asset_name,
+                            "agent_id": asset.agent_id,
+                        }
+                        for asset in alert.alert.assets
+                    ],
+                    "comments": [
+                        {
+                            "comment": comment.comment,
+                            "user_name": comment.user_name,
+                            "created_at": comment.created_at,
+                        }
+                        for comment in alert.alert.comments
+                    ],
+                    "context": {
+                        "source": alert.alert.assets[0].alert_context.source
+                        if alert.alert.assets and alert.alert.assets[0].alert_context
+                        else None,
+                        "context": alert.alert.assets[0].alert_context.context
+                        if alert.alert.assets and alert.alert.assets[0].alert_context
+                        else None,
+                    }
+                    if alert.alert.assets
+                    else None,
+                    "iocs": [
+                        {
+                            "ioc_value": ioc.ioc.value,
+                            "ioc_type": ioc.ioc.type,
+                            "ioc_description": ioc.ioc.description,
+                        }
+                        for ioc in alert.alert.iocs
+                    ],
+                }
+                for alert in case.alerts
+            ],
+        },
+    }
+
+def render_html_template(template_path: str, context: Dict[str, Dict[str, str]]) -> str:
+    """Render the Jinja HTML template with the provided context."""
+    template_dir = os.path.dirname(template_path)
+    template_name = os.path.basename(template_path)
+
+    env = Environment(loader=FileSystemLoader(template_dir))
+    template = env.get_template(template_name)
+    rendered_html = template.render(context)
+
+    # Save rendered HTML to a temporary file
+    with NamedTemporaryFile(delete=False, suffix=".html") as tmp:
+        tmp.write(rendered_html.encode("utf-8"))
+        return tmp.name
+
+def convert_html_to_pdf(html_path: str) -> str:
+    """Convert the HTML file to a PDF using wkhtmltopdf via pdfkit, with dynamic path detection for different platforms."""
+    pdf_path = html_path.replace(".html", ".pdf")
+    wkhtmltopdf_paths = []
+
+    # Determine paths to wkhtmltopdf based on the current platform
+    try:
+        if platform.system() == "Windows":
+            # Common installation paths for wkhtmltopdf on Windows
+            wkhtmltopdf_paths = [
+                r"C:\Program Files\wkhtmltopdf\bin\wkhtmltopdf.exe",
+                r"C:\Program Files (x86)\wkhtmltopdf\bin\wkhtmltopdf.exe"
+            ]
+        elif platform.system() == "Darwin":  # macOS
+            # Common installation paths for wkhtmltopdf on macOS
+            wkhtmltopdf_paths = [
+                "/usr/local/bin/wkhtmltopdf",
+                "/opt/homebrew/bin/wkhtmltopdf"  # For macOS ARM (M1/M2) using Homebrew
+            ]
+        elif platform.system() == "Linux":
+            # Common installation paths for wkhtmltopdf on Linux (Debian-based)
+            wkhtmltopdf_paths = [
+                "/usr/bin/wkhtmltopdf",
+                "/usr/local/bin/wkhtmltopdf"
+            ]
+
+        # Try each path until a valid executable is found
+        path_to_wkhtmltopdf = None
+        for path in wkhtmltopdf_paths:
+            try:
+                # Check if the executable can be accessed
+                config = pdfkit.configuration(wkhtmltopdf=path)
+                path_to_wkhtmltopdf = path
+                break
+            except OSError:
+                continue
+
+        # Raise an exception if no valid wkhtmltopdf path is found
+        if path_to_wkhtmltopdf is None:
+            raise FileNotFoundError("No valid wkhtmltopdf executable found. Ensure wkhtmltopdf is installed and accessible.")
+
+        # Generate the PDF from HTML using the valid wkhtmltopdf path
+        pdfkit.from_file(html_path, pdf_path, configuration=config)
+    except Exception as e:
+        raise RuntimeError(f"Failed to convert HTML to PDF: {str(e)}")
+
+    return pdf_path
+
+def create_file_response_pdf(file_path: str, file_name: Optional[str] = "case_report.pdf") -> FileResponse:
+    """Create a FileResponse object for the rendered document."""
+    return FileResponse(
+        file_path,
+        filename=file_name,
+        media_type="application/pdf",
+    )
+
+def cleanup_temp_files(file_paths: list):
+    """Clean up the temporary files."""
+    for file_path in file_paths:
+        if os.path.exists(file_path):
+            os.remove(file_path)

backend/app/incidents/templates/case_report_jinja_template.html

@@ -0,0 +1,50 @@
+<!DOCTYPE html>
+<html>
+<head>
+    <title>Case Report</title>
+    <style>
+        body { font-family: Arial, sans-serif; }
+        .case-info { margin-bottom: 20px; }
+        .alert { margin-bottom: 15px; }
+    </style>
+</head>
+<body>
+    <h1>Case Report</h1>
+    <div class="case-info">
+        <p><strong>Name of Case:</strong> {{ case.name }}</p>
+        <p><strong>Description:</strong> {{ case.description }}</p>
+        <p><strong>Assigned To:</strong> {{ case.assigned_to }}</p>
+        <p><strong>Case Creation Time:</strong> {{ case.case_creation_time }}</p>
+        <p><strong>Case ID:</strong> {{ case.id }}</p>
+    </div>
+
+    <h2>Alerts:</h2>
+    {% for alert in case.alerts %}
+    <div class="alert">
+        <p><strong>Alert Name:</strong> {{ alert.alert_name }}</p>
+        <p><strong>Description:</strong> {{ alert.alert_description }}</p>
+        <p><strong>Status:</strong> {{ alert.status }}</p>
+        <p><strong>Tags:</strong> {{ alert.tags | join(', ') }}</p>
+
+        <h3>Assets:</h3>
+        {% for asset in alert.assets %}
+        <p>- <strong>Asset Name:</strong> {{ asset.asset_name }} | <strong>Agent ID:</strong> {{ asset.agent_id }}</p>
+        {% endfor %}
+
+        <h3>Comments:</h3>
+        {% for comment in alert.comments %}
+        <p>- "{{ comment.comment }}" by {{ comment.user_name }} at {{ comment.created_at }}</p>
+        {% endfor %}
+
+        <h3>Context:</h3>
+        <p><strong>Source:</strong> {{ alert.context.source }}</p>
+        <p><strong>Context Details:</strong> {{ alert.context.context }}</p>
+
+        <h3>IoCs:</h3>
+        {% for ioc in alert.iocs %}
+        <p>- <strong>IoC Value:</strong> {{ ioc.ioc_value }} | <strong>Type:</strong> {{ ioc.ioc_type }} | <strong>Description:</strong> {{ ioc.ioc_description }}</p>
+        {% endfor %}
+    </div>
+    {% endfor %}
+</body>
+</html>