demo_data.py

"""
Demo data generator for the IT Project Health Dashboard.

Produces realistic project portfolio data so the dashboard can run
without a ServiceNow instance.  Deterministic (seeded random) so that
repeated runs yield the same dataset.

Usage:
    from demo_data import build_demo_data
    sections_active, sections_closed, grand_all, grand_key, mail_all = build_demo_data()
"""
from __future__ import annotations

import hashlib
import random
import urllib.parse as ul
from datetime import date, datetime, timedelta
from typing import Any, Dict, List, Tuple

# ────────────────────────────────────────────────────────────────────
# Constants
# ────────────────────────────────────────────────────────────────────

PORTFOLIOS = [
    "Cloud & Infrastructure",
    "Cybersecurity",
    "Data & Analytics",
    "Digital Workplace",
    "ERP & Finance Systems",
    "Manufacturing & OT",
    "Customer Experience",
    "Integration & API Platform",
]

PROJECT_PREFIXES = {
    "Cloud & Infrastructure": [
        "Azure Landing Zone Expansion",
        "AWS Cost Optimization",
        "Hybrid Cloud Migration",
        "Network Modernization",
        "DC Consolidation Phase",
        "Container Platform Rollout",
        "SD-WAN Global Deployment",
        "Cloud DR Site Setup",
        "Infrastructure Monitoring Upgrade",
        "Serverless Compute Pilot",
    ],
    "Cybersecurity": [
        "Zero Trust Architecture",
        "SOC Modernization",
        "PAM Implementation",
        "Endpoint Detection & Response",
        "Cloud Security Posture Mgmt",
        "Identity Governance Refresh",
        "Vulnerability Management Overhaul",
        "Security Awareness Training Platform",
        "DLP Rollout",
        "Threat Intelligence Platform",
    ],
    "Data & Analytics": [
        "Enterprise Data Lake",
        "BI Dashboards Modernization",
        "Data Quality Framework",
        "ML Ops Platform",
        "Customer 360 Analytics",
        "Real-Time Streaming Pipeline",
        "Master Data Management",
        "Self-Service Analytics Portal",
        "Data Catalog Implementation",
        "Predictive Maintenance Analytics",
    ],
    "Digital Workplace": [
        "M365 Copilot Rollout",
        "Intranet Redesign",
        "Digital Signage Platform",
        "Unified Communications Upgrade",
        "Employee Experience App",
        "Meeting Room Modernization",
        "Remote Work Toolkit",
        "Knowledge Management System",
        "ITSM Self-Service Portal",
        "Print Management Consolidation",
    ],
    "ERP & Finance Systems": [
        "S/4HANA Migration Wave",
        "Finance Close Automation",
        "Procurement Digitalization",
        "Travel & Expense Modernization",
        "Treasury Management System",
        "Consolidation Tool Upgrade",
        "Intercompany Reconciliation",
        "Fixed Assets Module Revamp",
        "E-Invoicing Compliance",
        "Budget Planning Platform",
    ],
    "Manufacturing & OT": [
        "MES Rollout Plant",
        "SCADA Modernization",
        "Digital Twin Pilot",
        "OT Network Segmentation",
        "Quality Management System",
        "Warehouse Management System",
        "Production Planning Optimization",
        "IoT Sensor Integration",
        "Maintenance Management Upgrade",
        "Shop Floor Visualization",
    ],
    "Customer Experience": [
        "CRM Platform Migration",
        "Customer Portal Redesign",
        "Marketing Automation Suite",
        "E-Commerce Platform",
        "Customer Feedback Analytics",
        "Service Desk Transformation",
        "Chatbot Implementation",
        "Loyalty Program System",
        "Omnichannel Integration",
        "Sales Force Effectiveness",
    ],
    "Integration & API Platform": [
        "API Gateway Modernization",
        "iPaaS Implementation",
        "Legacy System Integration",
        "EDI Modernization",
        "Event-Driven Architecture",
        "Service Mesh Deployment",
        "B2B Integration Hub",
        "Middleware Upgrade",
        "Real-Time Integration Layer",
        "API Lifecycle Management",
    ],
}

PM_NAMES = [
    ("Anna Mueller", "anna.mueller@example.com"),
    ("Thomas Weber", "thomas.weber@example.com"),
    ("Sarah Fischer", "sarah.fischer@example.com"),
    ("Michael Brunner", "michael.brunner@example.com"),
    ("Laura Schneider", "laura.schneider@example.com"),
    ("Daniel Keller", "daniel.keller@example.com"),
    ("Julia Hofmann", "julia.hofmann@example.com"),
    ("Markus Steiner", "markus.steiner@example.com"),
    ("Nina Berger", "nina.berger@example.com"),
    ("Patrick Zimmermann", "patrick.zimmermann@example.com"),
    ("Elena Schwarz", "elena.schwarz@example.com"),
    ("Stefan Baumann", "stefan.baumann@example.com"),
    ("Claudia Gerber", "claudia.gerber@example.com"),
    ("Reto Wyss", "reto.wyss@example.com"),
    ("Sandra Moser", "sandra.moser@example.com"),
    ("Felix Huber", "felix.huber@example.com"),
    ("Monika Graf", "monika.graf@example.com"),
    ("Andreas Leuthard", "andreas.leuthard@example.com"),
]

BPM_NAMES = [
    "Marc Dupont",
    "Christina Meier",
    "Oliver Frey",
    "Isabelle Roth",
    "Luca Bianchi",
    "Simone Kessler",
    "David Hartmann",
    "Katrin Vogt",
    "Philippe Martin",
    "Barbara Fuchs",
    "Roger Ammann",
    "Petra Schmid",
    "Jan de Vries",
    "Monique Favre",
    "Tobias Wenger",
]

EXECUTIVE_SUMMARIES = [
    "Project on track. Sprint velocity has stabilized and the team delivered all planned features this month. Key risk around third-party API stability has been mitigated with a fallback mechanism.",
    "Slight delay due to vendor resource constraints. We have escalated to procurement and expect resolution within two weeks. Budget utilization remains within approved limits.",
    "Successfully completed UAT for module 2. Go-live planned for next quarter. One critical defect found during testing is being addressed by the development team with a hotfix scheduled for next week.",
    "Project is progressing well. Infrastructure provisioning completed ahead of schedule. Data migration scripts are currently being validated in the staging environment.",
    "Scope change request approved by steering committee. Timeline extended by 4 weeks to accommodate additional regulatory requirements. No budget impact expected.",
    "Major milestone achieved: core platform deployment completed across all regions. Remaining work focuses on user training and documentation. Stakeholder satisfaction is high.",
    "Risk materialized: key technical resource resigned. Knowledge transfer is underway with the replacement. Timeline impact being assessed but likely 2-3 week delay on Phase 3.",
    "On track for year-end delivery. Integration testing with upstream systems completed successfully. Performance benchmarks exceeded targets by 15%.",
    "Budget overrun identified in professional services category. Root cause is additional customization requirements discovered during implementation. Change request submitted for CHF 50k additional funding.",
    "Pilot phase completed with positive user feedback (4.2/5 satisfaction score). Full rollout plan being finalized. Change management activities progressing well.",
    "Project temporarily paused pending procurement decision on platform selection. RFP evaluation in progress with shortlist of two vendors. Expected to resume in 3 weeks.",
    "All workstreams progressing according to plan. Security review completed with no critical findings. Two medium-severity items being addressed in current sprint.",
    "Successful go-live of Phase 1 in the EMEA region. Minor post-go-live issues resolved within SLA. Phase 2 planning for APAC region has commenced.",
    "Dependencies on the ERP upgrade project are creating scheduling challenges. Weekly sync meetings established to coordinate timelines. Contingency plan developed.",
    "Positive progress on data migration workstream. 85% of legacy data has been cleansed and mapped. Remaining 15% requires manual review due to data quality issues.",
    "Project completed code freeze milestone. Regression testing in progress with 92% pass rate. Remaining failures are cosmetic UI issues being fixed in the current cycle.",
]

STATUS_OPTIONS = ["green", "green", "green", "green", "yellow", "yellow", "red"]

# ────────────────────────────────────────────────────────────────────
# Helpers
# ────────────────────────────────────────────────────────────────────

def _deterministic_seed(name: str) -> int:
    return int(hashlib.md5(name.encode()).hexdigest()[:8], 16)


def _fake_sysid(idx: int) -> str:
    return hashlib.md5(f"project-{idx}".encode()).hexdigest()


def _rand_date(rng: random.Random, start: date, end: date) -> date:
    delta = (end - start).days
    if delta <= 0:
        return start
    return start + timedelta(days=rng.randint(0, delta))


# ────────────────────────────────────────────────────────────────────
# Timeline helpers (must match app_oauth.py section 4)
# ────────────────────────────────────────────────────────────────────

_today = date.today()
_rng_start = date(_today.year - 1, 4, 1)
_rng_end = date(_today.year + 1, 6, 30)
_rng_days = (_rng_end - _rng_start).days or 1
_min_w = (30 / _rng_days) * 100


def _pos(d: date) -> float:
    return (d - _rng_start).days / _rng_days * 100


def _q(d: date) -> str:
    return f"Q{((d.month - 1) // 3) + 1}'{str(d.year)[2:]}"


pct = lambda a, p: round(100 * a / p, 1) if p else None


# ────────────────────────────────────────────────────────────────────
# Main generator
# ────────────────────────────────────────────────────────────────────

def _generate_projects() -> Tuple[List[Dict], List[Dict]]:
    """Return (active_projects, closed_projects) as flat lists."""
    rng = random.Random(42)
    active: List[Dict] = []
    closed: List[Dict] = []
    proj_counter = 0

    for portfolio in PORTFOLIOS:
        names = PROJECT_PREFIXES[portfolio]
        # 5-8 active projects per portfolio
        n_active = rng.randint(5, 8)
        # 1-3 closed projects per portfolio
        n_closed = rng.randint(1, 3)

        for i in range(n_active + n_closed):
            proj_counter += 1
            is_closed = i >= n_active
            name_base = names[i % len(names)]

            # Some projects are key projects (~20%)
            is_key = rng.random() < 0.20
            suffix = ""
            if i >= len(names):
                suffix = f" Phase {i - len(names) + 2}"

            proj_name = name_base + suffix
            if is_key:
                proj_name += " [Key Project]"

            # Cross-portfolio tagging for some projects (~15%)
            if rng.random() < 0.15 and not is_key:
                other = rng.choice([p for p in PORTFOLIOS if p != portfolio])
                tag = other.split()[0][:2].upper()
                proj_name += f" [{tag}]"

            pm_name, pm_email = rng.choice(PM_NAMES)
            bpm = rng.choice(BPM_NAMES)

            # Financial data (CHF)
            plan_cap = rng.choice([0, 0] + list(range(50_000, 2_500_001, 50_000)))
            plan_ope = rng.choice([0] + list(range(20_000, 800_001, 10_000)))

            # Actual spending: typically 30-110% of plan
            if plan_cap > 0:
                cap_ratio = rng.uniform(0.15, 1.15)
                act_cap = round(plan_cap * cap_ratio, 2)
            else:
                act_cap = 0.0

            if plan_ope > 0:
                ope_ratio = rng.uniform(0.10, 1.20)
                act_ope = round(plan_ope * ope_ratio, 2)
            else:
                act_ope = 0.0

            # For closed projects, spending is 85-105% of plan
            if is_closed:
                if plan_cap > 0:
                    act_cap = round(plan_cap * rng.uniform(0.85, 1.05), 2)
                if plan_ope > 0:
                    act_ope = round(plan_ope * rng.uniform(0.85, 1.05), 2)

            # Dates
            if is_closed:
                end_d = _rand_date(rng, _today - timedelta(days=330),
                                   _today - timedelta(days=30))
                start_d = end_d - timedelta(days=rng.randint(90, 540))
                work_end = end_d
            else:
                start_d = _rand_date(rng, _today - timedelta(days=540),
                                     _today - timedelta(days=30))
                end_d = start_d + timedelta(days=rng.randint(120, 720))
                work_end = None

            # Timeline positioning
            s = _pos(start_d)
            e = _pos(end_d)
            if e < s:
                s, e = e, s
            left = max(min(s, 100), 0)
            right = max(min(e, 100), 0)
            width = max(0, right - left)
            if width < _min_w:
                width = _min_w
                if left + width > 100:
                    left = max(0, 100 - width)

            # Status indicators
            schedule = rng.choice(STATUS_OPTIONS)
            cost_st = rng.choice(STATUS_OPTIONS)
            scope_st = rng.choice(STATUS_OPTIONS)

            # Report date (current month = green, older = stale)
            if rng.random() < 0.65:
                rep_date = _today.replace(day=rng.randint(1, min(28, _today.day or 1)))
                rep_date_str = rep_date.strftime("%Y-%m-%d")
            elif rng.random() < 0.5:
                last_month = _today.replace(day=1) - timedelta(days=1)
                rep_date = last_month.replace(day=rng.randint(1, 28))
                rep_date_str = rep_date.strftime("%Y-%m-%d")
            else:
                rep_date_str = None

            summary = rng.choice(EXECUTIVE_SUMMARIES) if rep_date_str else ""

            pid = _fake_sysid(proj_counter)
            number = f"PRJ{proj_counter:07d}"
            rep_id = _fake_sysid(proj_counter + 10000) if rep_date_str else None

            rec: Dict[str, Any] = {
                "id": pid,
                "number": number,
                "name": proj_name,
                "pm": pm_name,
                "pm_mail": pm_email,
                "bpm": bpm,
                "plan_cap": float(plan_cap),
                "act_cap": act_cap,
                "pct_cap": pct(act_cap, plan_cap),
                "plan_ope": float(plan_ope),
                "act_ope": act_ope,
                "pct_ope": pct(act_ope, plan_ope),
                "st_pct": round(left, 2),
                "width_pct": round(width, 2),
                "start_q": _q(start_d),
                "end_q": _q(end_d),
                "is_key": is_key,
                "_portfolio": portfolio,
            }

            if is_closed:
                rec["status"] = {"schedule": None, "cost": None, "scope": None}
                rec["completion"] = work_end.strftime("%Y-%m-%d")
                closed.append(rec)
            else:
                rec["status"] = {
                    "schedule": schedule,
                    "cost": cost_st,
                    "scope": scope_st,
                }
                rec["rep_date"] = rep_date_str
                rec["rep_id"] = rep_id
                rec["summary"] = summary
                active.append(rec)

    return active, closed


def _build_sections(
    projects: List[Dict],
) -> List[Tuple[str, Dict[str, Any]]]:
    """Group flat project list into portfolio sections."""
    sections: Dict[str, Any] = {}

    for rec in projects:
        port = rec.pop("_portfolio", "(No Portfolio)")
        sec = sections.setdefault(
            port,
            {
                "projs": [],
                "tot": dict(pc=0, pa=0, oc=0, oa=0, cnt=0),
                "ktot": dict(pc=0, pa=0, oc=0, oa=0, cnt=0),
            },
        )
        sec["projs"].append(rec)
        for k, v in (
            ("pc", "plan_cap"),
            ("pa", "act_cap"),
            ("oc", "plan_ope"),
            ("oa", "act_ope"),
        ):
            sec["tot"][k] += rec[v]
            if rec["is_key"]:
                sec["ktot"][k] += rec[v]
        sec["tot"]["cnt"] += 1
        if rec["is_key"]:
            sec["ktot"]["cnt"] += 1

    outs = []
    for name, sec in sections.items():
        sec["projs"].sort(
            key=lambda x: x["plan_cap"] + x["plan_ope"], reverse=True
        )
        outs.append((name, sec))
    return sorted(outs, key=lambda x: x[0].lower())


def _grand_totals(
    sections: List[Tuple[str, Dict]], key_only: bool = False
) -> Dict[str, float]:
    g: Dict[str, float] = dict(pc=0.0, pa=0.0, oc=0.0, oa=0.0, cnt=0)
    seen: set = set()
    for _, sec in sections:
        for p in sec["projs"]:
            if key_only and not p["is_key"]:
                continue
            if p["id"] in seen:
                continue
            seen.add(p["id"])
            g["pc"] += p["plan_cap"]
            g["pa"] += p["act_cap"]
            g["oc"] += p["plan_ope"]
            g["oa"] += p["act_ope"]
            g["cnt"] += 1
    return g


def _make_mail_all(sections):
    """Mailto link for every PM with stale / missing report."""
    today_dt = datetime.today()
    yy, mm = today_dt.year, today_dt.month
    emails: set = set()
    for _, sec in sections:
        for p in sec["projs"]:
            d = p.get("rep_date")
            ok = d and int(d[:4]) == yy and int(d[5:7]) == mm
            if not ok and p.get("pm_mail"):
                emails.add(p["pm_mail"])
    if not emails:
        return None
    TO = ";".join(sorted(emails))
    subj = ul.quote("Reminder: monthly project status report")
    body = ul.quote(
        "Dear Project Manager,\r\n"
        "This is a kind reminder to submit your monthly status report(s) "
        "in ServiceNow.\r\nThanks"
    )
    return f"mailto:{TO}?subject={subj}&body={body}"


# ────────────────────────────────────────────────────────────────────
# Public API
# ────────────────────────────────────────────────────────────────────

def build_demo_data():
    """
    Build complete demo dataset.

    Returns:
        (sections_active, sections_closed, grand_all, grand_key, mail_all)
    """
    active_projs, closed_projs = _generate_projects()
    sections_active = _build_sections(active_projs)
    sections_closed = _build_sections(closed_projs)
    grand_all = _grand_totals(sections_active, False)
    grand_key = _grand_totals(sections_active, True)
    mail_all = _make_mail_all(sections_active)
    return sections_active, sections_closed, grand_all, grand_key, mail_all