app/analytics.py

"""
analytics.py — deterministic, grounded computation over a project's milestones.

Everything the model must NOT guess is computed here in Python: today's date,
days-to-due, what's overdue, what's blocked, the project's final deadline, and a
plain-text status digest. The chat layer injects these facts so answers about
deadlines / blockers / status are read off the sheet, not invented.
"""
import datetime
import os
import re

RISK_HINTS = ("may extend", "may vary", "depending on", "lead time", "to be ordered",
              "issues", "delay", "risk")
DONE = {"completed", "done", "closed"}
BLOCKED = {"delayed"}


def today() -> datetime.date:
    """Reference 'now'. Override with PORTAL_TODAY=YYYY-MM-DD for testing."""
    override = os.environ.get("PORTAL_TODAY", "").strip()
    if override:
        try:
            return datetime.date.fromisoformat(override)
        except ValueError:
            pass
    return datetime.date.today()


def _date(s):
    if not s:
        return None
    try:
        return datetime.date.fromisoformat(str(s)[:10])
    except ValueError:
        return None


def enrich(milestone: dict, ref: datetime.date) -> dict:
    """Return the milestone plus derived fields (non-destructive copy)."""
    m = dict(milestone)
    status = (m.get("status") or "").strip().lower()
    target = _date(m.get("target_end"))
    m["_done"] = status in DONE
    m["_blocked"] = status in BLOCKED
    remarks = (m.get("remarks") or "").lower()
    m["_at_risk"] = any(h in remarks for h in RISK_HINTS)
    if target and not m["_done"]:
        m["_days_to_due"] = (target - ref).days
        m["_overdue"] = m["_days_to_due"] < 0
    else:
        m["_days_to_due"] = None
        m["_overdue"] = False
    return m


def snapshot(project: dict, ref: datetime.date) -> dict:
    rows = [enrich(m, ref) for m in (project.get("milestones") or [])]
    by_status = {}
    for m in rows:
        key = (m.get("status") or "Unknown").strip() or "Unknown"
        by_status[key] = by_status.get(key, 0) + 1

    overdue = sorted(
        [m for m in rows if m["_overdue"]],
        key=lambda m: m["_days_to_due"],
    )
    blocked = [m for m in rows if m["_blocked"]]
    at_risk = [m for m in rows if m["_at_risk"] and not m["_done"]]
    due_soon = sorted(
        [m for m in rows if m["_days_to_due"] is not None and 0 <= m["_days_to_due"] <= 14],
        key=lambda m: m["_days_to_due"],
    )

    open_targets = [(_date(m.get("target_end")), m) for m in rows if not m["_done"]]
    open_targets = [(d, m) for d, m in open_targets if d]
    next_deadline = min(open_targets, key=lambda x: x[0]) if open_targets else None

    all_targets = [(_date(m.get("target_end")), m) for m in rows]
    all_targets = [(d, m) for d, m in all_targets if d]
    final_deadline = max(all_targets, key=lambda x: x[0]) if all_targets else None

    total = len(rows)
    done = sum(1 for m in rows if m["_done"])
    return {
        "today": ref.isoformat(),
        "total": total,
        "done": done,
        "percent_complete": round(100 * done / total) if total else 0,
        "by_status": by_status,
        "overdue": overdue,
        "blocked": blocked,
        "at_risk": at_risk,
        "due_soon": due_soon,
        "next_deadline": next_deadline,   # (date, milestone) or None
        "final_deadline": final_deadline,
    }


def _label(m):
    sno = (m.get("sno") or "").strip()
    return (sno + " " if sno else "") + (m.get("milestone") or "").strip()


def _section_key(sno: str) -> str:
    m = re.match(r"(\d+)", (sno or "").strip())
    return m.group(1) if m else "?"


def _row_view(m: dict, section_name: str) -> dict:
    """Serializable milestone row with derived flags for the dashboard/table."""
    return {
        "sno": m.get("sno", ""),
        "milestone": m.get("milestone", ""),
        "owner": m.get("owner", ""),
        "status": m.get("status", ""),
        "start": m.get("start", ""),
        "target_end": m.get("target_end", ""),
        "actual_end": m.get("actual_end", ""),
        "duration": m.get("duration", ""),
        "depends": m.get("depends", ""),
        "remarks": m.get("remarks", ""),
        "section": section_name,
        "done": m["_done"],
        "overdue": m["_overdue"],
        "blocked": m["_blocked"],
        "at_risk": m["_at_risk"],
        "days_to_due": m["_days_to_due"],
    }


def dashboard(project: dict, ref: datetime.date) -> dict:
    """Everything the executive dashboard needs in one call (no LLM, grounded)."""
    rows = [enrich(m, ref) for m in (project.get("milestones") or [])]

    # section (in-project "domain") names from the x.0 rows
    topname = {}
    for m in rows:
        sno = (m.get("sno") or "").strip()
        if re.match(r"^\d+\.0$", sno):
            topname[_section_key(sno)] = (m.get("milestone") or "").strip()

    sec_rows = {}
    for m in rows:
        k = _section_key(m.get("sno"))
        sec_rows.setdefault(k, []).append(m)

    def sec_name(k):
        return topname.get(k, f"Section {k}")

    sections = []
    for k in sorted(sec_rows, key=lambda x: (int(x) if x.isdigit() else 999)):
        items = sec_rows[k]
        total = len(items)
        done = sum(1 for m in items if m["_done"])
        starts = [_date(m.get("start")) for m in items if _date(m.get("start"))]
        ends = [_date(m.get("target_end")) for m in items if _date(m.get("target_end"))]
        sections.append({
            "key": k, "name": sec_name(k), "total": total, "done": done,
            "in_progress": sum(1 for m in items if (m.get("status") or "").lower() == "in progress"),
            "overdue": sum(1 for m in items if m["_overdue"]),
            "blocked": sum(1 for m in items if m["_blocked"]),
            "percent": round(100 * done / total) if total else 0,
            "start": min(starts).isoformat() if starts else None,
            "end": max(ends).isoformat() if ends else None,
        })

    # owner workload
    owners = {}
    for m in rows:
        o = (m.get("owner") or "—").strip() or "—"
        owners.setdefault(o, {"owner": o, "total": 0, "overdue": 0, "blocked": 0})
        owners[o]["total"] += 1
        if m["_overdue"]:
            owners[o]["overdue"] += 1
        if m["_blocked"]:
            owners[o]["blocked"] += 1
    owners_list = sorted(owners.values(), key=lambda x: x["total"], reverse=True)

    # status breakdown for the ring
    status_counts = {}
    for m in rows:
        s = (m.get("status") or "Unknown").strip() or "Unknown"
        status_counts[s] = status_counts.get(s, 0) + 1
    status_breakdown = [{"status": s, "count": c} for s, c in
                        sorted(status_counts.items(), key=lambda x: -x[1])]

    milestones = [_row_view(m, sec_name(_section_key(m.get("sno")))) for m in rows]
    blockers = [r for r in milestones if r["blocked"]]
    at_risk = [r for r in milestones if r["at_risk"] and not r["done"]]

    all_starts = [_date(m.get("start")) for m in rows if _date(m.get("start"))]
    all_ends = [_date(m.get("target_end")) for m in rows if _date(m.get("target_end"))]

    return {
        "summary": snapshot_summary(project, ref),
        "status_breakdown": status_breakdown,
        "sections": sections,
        "owners": owners_list[:8],
        "milestones": milestones,
        "blockers": blockers,
        "at_risk": at_risk,
        "timeline": {
            "start": min(all_starts).isoformat() if all_starts else None,
            "end": max(all_ends).isoformat() if all_ends else None,
            "today": ref.isoformat(),
        },
        "facets": {
            "statuses": sorted({(m.get("status") or "").strip() for m in rows if (m.get("status") or "").strip()}),
            "owners": sorted({(m.get("owner") or "").strip() for m in rows if (m.get("owner") or "").strip()}),
            "sections": [s["name"] for s in sections],
        },
    }


def snapshot_summary(project: dict, ref: datetime.date) -> dict:
    """JSON-safe headline numbers for the UI stat cards (no LLM, fully grounded)."""
    s = snapshot(project, ref)

    def deadline(t):
        if not t:
            return None
        d, m = t
        return {"date": d.isoformat(), "label": _label(m), "days": (d - ref).days}

    bs = s["by_status"]
    return {
        "today": s["today"],
        "total": s["total"],
        "done": s["done"],
        "in_progress": bs.get("In Progress", 0),
        "not_started": bs.get("Not Started", 0),
        "delayed": bs.get("Delayed", 0),
        "overdue": len(s["overdue"]),
        "blocked": len(s["blocked"]),
        "at_risk": len(s["at_risk"]),
        "due_soon": len(s["due_soon"]),
        "percent_complete": s["percent_complete"],
        "next_deadline": deadline(s["next_deadline"]),
        "final_deadline": deadline(s["final_deadline"]),
    }


def status_digest_text(project: dict, ref: datetime.date) -> str:
    """A fully grounded, no-LLM status summary — the backbone of alerting."""
    s = snapshot(project, ref)
    out = [f"PROJECT STATUS — {project.get('name','(unnamed)')}  (as of {s['today']})",
           f"Progress: {s['done']}/{s['total']} milestones completed ({s['percent_complete']}%).",
           "By status: " + ", ".join(f"{k}: {v}" for k, v in sorted(s['by_status'].items())) or "n/a"]

    if s["final_deadline"]:
        d, m = s["final_deadline"]
        out.append(f"Final milestone deadline: {d.isoformat()} — {_label(m)}.")
    if s["next_deadline"]:
        d, m = s["next_deadline"]
        days = (d - ref).days
        when = f"in {days} day(s)" if days >= 0 else f"{abs(days)} day(s) ago"
        out.append(f"Next open deadline: {d.isoformat()} ({when}) — {_label(m)}.")

    def block(title, items, render):
        if not items:
            return f"\n{title}: none."
        return f"\n{title} ({len(items)}):\n" + "\n".join("  - " + render(m) for m in items)

    out.append(block("OVERDUE", s["overdue"],
        lambda m: f"{_label(m)} — owner {m.get('owner') or '?'}, due {m.get('target_end') or '?'}, "
                  f"{abs(m['_days_to_due'])} day(s) late, status {m.get('status') or '?'}"))
    out.append(block("BLOCKED (status = Delayed)", s["blocked"],
        lambda m: f"{_label(m)} — owner {m.get('owner') or '?'}"
                  + (f"; remark: {m.get('remarks')}" if m.get('remarks') else "")
                  + (f"; dependency: {m.get('depends')}" if m.get('depends') else "")))
    out.append(block("AT-RISK (remarks flag schedule risk)", s["at_risk"],
        lambda m: f"{_label(m)} — {m.get('remarks')}"))
    out.append(block("DUE WITHIN 14 DAYS", s["due_soon"],
        lambda m: f"{_label(m)} — due {m.get('target_end')}, in {m['_days_to_due']} day(s), "
                  f"owner {m.get('owner') or '?'}, status {m.get('status') or '?'}"))
    return "\n".join(out)