generate_samples

# PDF Style Engine v2.0
# (c) 2025 Claudia Ortega
# Licensed under CC BY-NC-ND 4.0
# https://creativecommons.org/licenses/by-nc-nd/4.0/

#!/usr/bin/env python3
"""
generate_samples.py
===================
CLI tool to generate sample PDFs for both available styles.

Usage
-----
    python generate_samples.py                      # generates both styles
    python generate_samples.py --style professional # single style
    python generate_samples.py --style cyber
    python generate_samples.py --out /tmp/my_docs  # custom output directory
"""

import argparse
import os
import sys

# Allow running from any working directory
sys.path.insert(0, os.path.dirname(__file__))
from pdf_generator import generate_pdf


# ---------------------------------------------------------------------------
# Sample data definitions
# ---------------------------------------------------------------------------

PROFESSIONAL_DATA = {
    "title":     "Annual Technology Review 2025",
    "subtitle":  "Strategic Infrastructure Assessment",
    "date":      "2025-05-06",
    "reference": "TEC-2025-0042",
    "company":   "Nexus Consulting Group",
    "contact":   "info@nexusconsulting.io  |  +1 (800) 555-0199",
    # "logo_path": "/path/to/logo.png",   # uncomment and set a real path
    "sections": [
        {
            "heading": "Executive Summary",
            "body": (
                "This report provides a comprehensive assessment of the organisation's "
                "technology infrastructure for the fiscal year 2025. Key findings indicate "
                "that cloud adoption has accelerated by 34% year-over-year, while on-premise "
                "legacy costs have been reduced by 18%. The strategic roadmap outlined herein "
                "prioritises security hardening, AI-assisted observability, and edge computing "
                "initiatives for the coming 18 months."
            ),
        },
        {
            "heading": "Infrastructure Overview",
            "body": (
                "The current stack consists of a hybrid multi-cloud deployment spanning AWS "
                "and Azure tenants, augmented by an on-premise Kubernetes cluster for latency-"
                "sensitive workloads. Network segmentation follows a zero-trust architecture "
                "enforced via service mesh policies. Total managed endpoints have grown from "
                "1,240 to 1,890 units, with a corresponding increase in EDR coverage to 99.4%."
            ),
        },
        {
            "heading": "Security Posture",
            "body": (
                "Penetration testing conducted in Q1 identified three medium-severity findings, "
                "all of which were remediated within the agreed 30-day SLA. The SIEM platform "
                "processed an average of 4.2 million events per day with a mean detection time "
                "of 6 minutes for critical alerts. Phishing simulation success rates improved "
                "from 12% to 4% following the updated security awareness programme."
            ),
        },
        {
            "heading": "Recommendations",
            "body": (
                "1. Migrate remaining on-premise workloads to containerised environments by Q3. "
                "2. Implement automated patch management for all Tier-1 systems. "
                "3. Expand SIEM retention to 18 months to meet emerging compliance requirements. "
                "4. Evaluate AI-driven anomaly detection platforms for SOC augmentation."
            ),
        },
    ],
    "table": {
        "headers": ["System", "Status", "Coverage", "Next Review"],
        "rows": [
            ["AWS Cloud",        "Operational", "100%",   "2025-08-01"],
            ["Azure Tenant",     "Operational", "100%",   "2025-08-01"],
            ["On-Prem K8s",      "Degraded",    "94.1%",  "2025-06-15"],
            ["EDR Platform",     "Operational", "99.4%",  "2025-09-01"],
            ["SIEM / SOAR",      "Operational", "100%",   "2025-07-30"],
            ["Backup Systems",   "Operational", "97.8%",  "2025-08-20"],
        ],
    },
}


CYBER_DATA = {
    "title":     "THREAT_INTEL_BRIEF",
    "subtitle":  "Red Team Operation — Shadow Nexus // Phase II",
    "date":      "2025-05-06T03:14:15Z",
    "reference": "OPS-SN-0xBEEF",
    "hash_id":   "A3F9C1D7E2B08541",
    "ascii_art": (
        "  ██████╗██╗   ██╗██████╗ ███████╗██████╗ \n"
        " ██╔════╝╚██╗ ██╔╝██╔══██╗██╔════╝██╔══██╗\n"
        " ██║      ╚████╔╝ ██████╔╝█████╗  ██████╔╝\n"
        " ██║       ╚██╔╝  ██╔══██╗██╔══╝  ██╔══██╗\n"
        " ╚██████╗   ██║   ██████╔╝███████╗██║  ██║\n"
        "  ╚═════╝   ╚═╝   ╚═════╝ ╚══════╝╚═╝  ╚═╝"
    ),
    "sections": [
        {
            "heading": "Mission Brief",
            "body": (
                "OPERATION SHADOW NEXUS — Phase II authorisation granted at 03:00 UTC. "
                "Objective: enumerate lateral movement vectors across the 10.14.0.0/16 segment "
                "and validate persistence mechanisms planted during Phase I. Rules of engagement "
                "restrict destructive payloads. All traffic to be tunnelled through VPN_GATEWAY_7."
            ),
        },
        {
            "heading": "Reconnaissance Summary",
            "body": (
                "Host discovery completed via masscan at 100k PPS. 847 live hosts identified. "
                "Service fingerprinting revealed 23 exposed RDP instances, 14 unpatched SMB "
                "services (CVE-2020-0796), and 3 internet-facing Jenkins nodes running outdated "
                "versions. Credential spray returned 6 valid accounts in the CORP domain. "
                "Kerberoastable SPNs: 11 identified, 4 cracked offline within 40 minutes."
            ),
        },
        {
            "heading": "Exploitation Log",
            "body": (
                "[+] 10.14.3.22  — SMBGhost RCE — SYSTEM shell obtained 04:17 UTC\n"
                "[+] 10.14.7.55  — Jenkins Script Console — reverse shell spawned 04:31 UTC\n"
                "[+] 10.14.1.9   — Pass-the-Hash — lateral move to DC01 05:02 UTC\n"
                "[!] 10.14.2.100 — EDR blocked Mimikatz execution — alternate method used\n"
                "[+] DC01        — DCSync executed — NTDS.dit extracted 05:44 UTC"
            ),
        },
        {
            "heading": "Indicators of Compromise",
            "body": (
                "SHA256: e3b0c44298fc1c149afb4c8996fb92427ae41e4649b934ca495991b7852b855 "
                "C2 Domain: nx-relay.shadowed.io  |  C2 IP: 198.51.100.77:4444 "
                "Persistence: HKCU\\Software\\Microsoft\\Windows\\CurrentVersion\\Run — key 'svcupd' "
                "Scheduled Task: \\Microsoft\\Windows\\svchost_upd  — triggers every 15 min"
            ),
        },
        {
            "heading": "Recommendations",
            "body": (
                "PRIORITY 1 — Patch CVE-2020-0796 on all affected hosts immediately. "
                "PRIORITY 2 — Rotate all domain credentials; enforce MFA on RDP endpoints. "
                "PRIORITY 3 — Harden Kerberos delegation; remove unnecessary SPNs. "
                "PRIORITY 4 — Segment Jenkins CI/CD pipeline from corporate network. "
                "PRIORITY 5 — Deploy canary tokens on NTDS.dit and sensitive share paths."
            ),
        },
    ],
    "table": {
        "headers": ["HOST", "CVE", "SEVERITY", "STATUS"],
        "rows": [
            ["10.14.3.22",  "CVE-2020-0796", "CRITICAL", "EXPLOITED"],
            ["10.14.7.55",  "N/A (Misconfig)", "HIGH",   "EXPLOITED"],
            ["10.14.1.9",   "PtH Attack",   "HIGH",      "PIVOTED"],
            ["10.14.2.100", "CVE-2021-34527", "CRITICAL", "BLOCKED"],
            ["10.14.5.11",  "CVE-2019-0708", "CRITICAL", "PENDING"],
            ["DC01",        "DCSync / AD",  "CRITICAL",  "PWNED"],
        ],
    },
}


# ---------------------------------------------------------------------------
# CLI
# ---------------------------------------------------------------------------

def main() -> None:
    parser = argparse.ArgumentParser(
        description="Generate sample PDFs using pdf_generator.py"
    )
    parser.add_argument(
        "--style",
        choices=["professional", "cyber", "both"],
        default="both",
        help="Which style to generate (default: both)",
    )
    parser.add_argument(
        "--out",
        default=os.path.join(os.path.dirname(__file__), "output"),
        help="Output directory (default: ./output)",
    )
    args = parser.parse_args()

    os.makedirs(args.out, exist_ok=True)

    styles_to_run = (
        ["professional", "cyber"] if args.style == "both" else [args.style]
    )

    for style in styles_to_run:
        data = PROFESSIONAL_DATA if style == "professional" else CYBER_DATA
        out  = os.path.join(args.out, f"sample_{style}.pdf")
        path = generate_pdf(style, data, out)
        print(f"[✓] {style.upper():14s}  →  {path}")


if __name__ == "__main__":
    main()