start project

app/__init__.py

@@ -0,0 +1,84 @@
+"""Application factory (Chapter 02 / Chapter 04)."""
+from __future__ import annotations
+
+from flask import Flask, jsonify, redirect, request, url_for
+from flask_login import current_user
+from sqlalchemy import text
+
+from app.budget import validate_budget_config
+from app.config import get_config
+from app.extensions import cache, csrf, db, limiter, login_manager, migrate
+from app.logging_setup import configure_logging
+from app.utils.assets import register_asset_helper
+
+
+def create_app(config_name: str | None = None) -> Flask:
+    """Build and configure the Flask application instance."""
+    app = Flask(__name__)
+    app.config.from_object(get_config(config_name))
+    validate_budget_config(app)  # Chapter 03: fail loudly on out-of-budget config
+    configure_logging(app)  # Chapter 02 factor 11 / Chapter 12
+
+    if not app.config.get("SECRET_KEY") and not app.testing:
+        raise RuntimeError("SECRET_KEY must be set via environment variable")
+
+    db.init_app(app)
+    cache.init_app(app)
+    csrf.init_app(app)
+    login_manager.init_app(app)
+    migrate.init_app(app, db)
+    limiter.init_app(app)  # Chapter 12: rate limiting
+
+    register_blueprints(app)
+    register_cli(app)
+    register_asset_helper(app)
+
+    @login_manager.unauthorized_handler
+    def handle_unauthorized():
+        """Chapter 11 envelope compliance: JSON 401 for /api/... paths
+        instead of Flask-Login's default redirect (which would hand back
+        login HTML to a fetch()/HTMX JSON caller). Full-page routes still
+        redirect to the login form as normal.
+        """
+        if request.path.startswith("/api/"):
+            return jsonify(error={"code": "unauthorized", "message": "Authentication required."}), 401
+        return redirect(url_for("auth.login", next=request.path))
+
+    @app.get("/healthz")
+    def healthz():
+        """Liveness/readiness check (Chapter 12) — DB reachable, no long-running work."""
+        db.session.execute(text("SELECT 1"))
+        return {"data": {"status": "ok"}, "meta": {}}
+
+    @app.get("/")
+    def index():
+        """Root URL: authenticated -> Overview, otherwise -> the login page."""
+        if current_user.is_authenticated:
+            return redirect(url_for("overview.overview"))
+        return redirect(url_for("auth.login"))
+
+    return app
+
+
+def register_blueprints(app: Flask) -> None:
+    """Register one blueprint per dashboard section + support area (Ch. 04)."""
+    from app.blueprints.api import bp as api_bp
+    from app.blueprints.auth import bp as auth_bp
+    from app.blueprints.overview import bp as overview_bp
+    from app.blueprints.security import bp as security_bp
+    from app.blueprints.seo import bp as seo_bp
+    from app.blueprints.uploads import bp as uploads_bp
+
+    app.register_blueprint(overview_bp)
+    app.register_blueprint(seo_bp)
+    app.register_blueprint(security_bp)
+    app.register_blueprint(uploads_bp)
+    app.register_blueprint(api_bp)
+    app.register_blueprint(auth_bp)
+
+
+def register_cli(app: Flask) -> None:
+    """Attach `flask <command>` admin processes (factor 12)."""
+    from app.cli import register_commands
+
+    register_commands(app)

app/blueprints/api/__init__.py

@@ -0,0 +1,7 @@
+"""Deliberately minimal — Chapter 02's tree lists a standalone `api`
+blueprint, but Chapter 04/11 route each section's JSON endpoints through
+its own blueprint instead. Left unpopulated per the project owner's
+instruction to defer this branch."""
+from flask import Blueprint
+
+bp = Blueprint("api", __name__)

app/blueprints/auth/__init__.py

@@ -0,0 +1,5 @@
+from flask import Blueprint
+
+bp = Blueprint("auth", __name__, template_folder="templates")
+
+from app.blueprints.auth import routes  # noqa: E402,F401  registers routes

app/blueprints/auth/forms.py

@@ -0,0 +1,18 @@
+"""Login form (Chapter 12). CSRF handled automatically via form.hidden_tag()
+(Flask-WTF, Ch12's CSRF requirement)."""
+from __future__ import annotations
+
+from flask_wtf import FlaskForm
+from wtforms import PasswordField, StringField, SubmitField
+from wtforms.validators import DataRequired
+
+
+class LoginForm(FlaskForm):
+    # NOTE: no Email() validator — that validator requires the extra
+    # `email-validator` package. Login checks the value against a stored
+    # user row anyway, so a malformed email simply fails to match rather
+    # than needing format validation up front; kept simple to avoid a new
+    # dependency.
+    email = StringField("Email", validators=[DataRequired()])
+    password = PasswordField("Password", validators=[DataRequired()])
+    submit = SubmitField("Log in")

app/blueprints/auth/routes.py

@@ -0,0 +1,45 @@
+"""Single-admin auth routes (Chapter 12 / Chapter 11's route table)."""
+from __future__ import annotations
+
+from flask import flash, redirect, render_template, request, url_for
+from flask_login import current_user, login_required, login_user, logout_user
+
+from app.blueprints.auth import bp
+from app.blueprints.auth.forms import LoginForm
+from app.extensions import db, limiter, login_manager
+from app.models.user import User
+
+
+@login_manager.user_loader
+def load_user(user_id: str) -> User | None:
+    return db.session.get(User, int(user_id))
+
+
+@bp.route("/login", methods=["GET", "POST"])
+@limiter.limit("10 per minute")  # Ch12: rate limiting on /login at minimum
+def login():
+    if current_user.is_authenticated:
+        return redirect(url_for("overview.overview"))
+
+    form = LoginForm()
+    if form.validate_on_submit():
+        user = User.query.filter_by(email=form.email.data.strip().lower()).first()
+        if user is not None and user.check_password(form.password.data):
+            login_user(user)
+            next_url = request.args.get("next") or url_for("overview.overview")
+            return redirect(next_url)
+        flash("Invalid email or password.")
+
+    return render_template("auth/login.html", form=form)
+
+
+@bp.post("/logout")
+@login_required
+def logout():
+    # NOTE (flagged): Ch11's route table lists "/login, /logout" under a
+    # shared "GET/POST" column. Logout is POST-only here — a state-changing
+    # action behind a plain GET is a CSRF-adjacent anti-pattern Ch12's own
+    # CSRF requirement argues against; login stays GET (show form) + POST
+    # (submit), matching the table as-is.
+    logout_user()
+    return redirect(url_for("auth.login"))

app/blueprints/auth/templates/auth/login.html

@@ -0,0 +1,27 @@
+{% extends "base.html" %}
+{% block title %}Log in — Kavosh{% endblock %}
+{% block content %}
+<div class="max-w-sm mx-auto mt-16 sm:mt-24">
+  <div class="text-center mb-6">
+    <h1 class="font-display font-bold text-2xl tracking-tight">Kavosh</h1>
+    <p class="text-sm text-muted dark:text-muted-dark mt-1">Sign in to view your site's traffic</p>
+  </div>
+  <div class="bg-surface dark:bg-surface-dark border border-line dark:border-line-dark rounded-xl p-6 shadow-sm">
+    {% for message in get_flashed_messages() %}
+      <p class="text-danger dark:text-danger-dark text-sm mb-3 bg-danger/10 dark:bg-danger-dark/10 rounded-md px-3 py-2">{{ message }}</p>
+    {% endfor %}
+    <form method="post">
+      {{ form.hidden_tag() }}
+      <div class="mb-4">
+        {{ form.email.label(class="block text-sm font-medium text-muted dark:text-muted-dark mb-1") }}
+        {{ form.email(class="w-full border border-line dark:border-line-dark rounded-md px-3 py-2 bg-paper dark:bg-paper-dark text-ink dark:text-ink-dark focus:outline-none focus:ring-2 focus:ring-accent dark:focus:ring-accent-dark", autofocus=true) }}
+      </div>
+      <div class="mb-5">
+        {{ form.password.label(class="block text-sm font-medium text-muted dark:text-muted-dark mb-1") }}
+        {{ form.password(class="w-full border border-line dark:border-line-dark rounded-md px-3 py-2 bg-paper dark:bg-paper-dark text-ink dark:text-ink-dark focus:outline-none focus:ring-2 focus:ring-accent dark:focus:ring-accent-dark") }}
+      </div>
+      {{ form.submit(class="w-full bg-accent dark:bg-accent-dark text-white dark:text-paper-dark font-medium rounded-md px-4 py-2 hover:opacity-90 transition-opacity cursor-pointer") }}
+    </form>
+  </div>
+</div>
+{% endblock %}

app/blueprints/overview/__init__.py

@@ -0,0 +1,13 @@
+from flask import Blueprint
+from flask_login import login_required
+
+bp = Blueprint("overview", __name__, template_folder="templates")
+
+
+@bp.before_request
+@login_required
+def require_login():
+    pass
+
+
+from app.blueprints.overview import routes  # noqa: E402,F401  registers routes

app/blueprints/overview/queries.py

@@ -0,0 +1,167 @@
+"""Context-builder query functions for the Overview tab (Chapter 08).
+
+Every function here reads request_stats_hourly / request_stats_daily /
+referrer_stats_daily / browser_stats_daily — never log_entries — per
+Ch03 rule 6 / Ch08's own data-source rule.
+"""
+from __future__ import annotations
+
+from datetime import date
+
+from sqlalchemy import case, func
+
+from app.extensions import db
+from app.models.browser_stats import BrowserStatsDaily
+from app.models.referrer_stats import ReferrerStatsDaily
+from app.models.request_stats import RequestStatsDaily, RequestStatsHourly
+from app.utils.dates import day_bounds
+
+# ASSUMPTION (flagged in Ch08): Ch08 doesn't give a numeric threshold for
+# "hourly vs daily depending on range width" — picked 3 days.
+HOURLY_GRANULARITY_THRESHOLD_DAYS = 3
+
+
+def get_kpis(from_date: date, to_date: date) -> dict:
+    """All six Ch08 KPI-card fields, plus peak-day (folded in — Ch11 has
+    no dedicated route for it and Ch08 calls for only a 'simple max-lookup').
+    """
+    row = (
+        db.session.query(
+            func.coalesce(func.sum(RequestStatsDaily.count), 0).label("total_requests"),
+            func.coalesce(func.sum(RequestStatsDaily.unique_ips), 0).label("unique_ips_sum"),
+            func.coalesce(func.sum(RequestStatsDaily.bytes_sum), 0).label("total_bandwidth"),
+            func.coalesce(func.sum(RequestStatsDaily.error_count), 0).label("total_errors"),
+        )
+        .filter(RequestStatsDaily.date >= from_date, RequestStatsDaily.date <= to_date)
+        .one()
+    )
+    num_days = (to_date - from_date).days + 1
+    avg_response_size = (row.total_bandwidth / row.total_requests) if row.total_requests else 0.0
+    error_rate_pct = (row.total_errors / row.total_requests * 100) if row.total_requests else 0.0
+    avg_requests_per_day = row.total_requests / num_days if num_days else 0.0
+
+    peak_row = (
+        db.session.query(RequestStatsDaily.date, RequestStatsDaily.count)
+        .filter(RequestStatsDaily.date >= from_date, RequestStatsDaily.date <= to_date)
+        .order_by(RequestStatsDaily.count.desc())
+        .first()
+    )
+
+    return {
+        "total_requests": row.total_requests,
+        # APPROXIMATION (flagged in Ch08): sum of daily unique_ips over-counts
+        # repeat visitors across days.
+        "unique_ips": row.unique_ips_sum,
+        "total_bandwidth_bytes": row.total_bandwidth,
+        "avg_response_size_bytes": round(avg_response_size, 1),
+        "error_rate_pct": round(error_rate_pct, 2),
+        "avg_requests_per_day": round(avg_requests_per_day, 1),
+        "peak_day": {"date": peak_row.date.isoformat(), "count": peak_row.count} if peak_row else None,
+    }
+
+
+def get_traffic_chart_series(from_date: date, to_date: date) -> dict:
+    span_days = (to_date - from_date).days + 1
+    if span_days <= HOURLY_GRANULARITY_THRESHOLD_DAYS:
+        start, end = day_bounds(from_date, to_date)
+        rows = (
+            db.session.query(RequestStatsHourly.date_hour, func.sum(RequestStatsHourly.count).label("count"))
+            .filter(RequestStatsHourly.date_hour >= start, RequestStatsHourly.date_hour < end)
+            .group_by(RequestStatsHourly.date_hour)
+            .order_by(RequestStatsHourly.date_hour)
+            .all()
+        )
+        return {"granularity": "hourly", "series": [{"t": r.date_hour.isoformat(), "count": r.count} for r in rows]}
+
+    rows = (
+        db.session.query(RequestStatsDaily.date, RequestStatsDaily.count)
+        .filter(RequestStatsDaily.date >= from_date, RequestStatsDaily.date <= to_date)
+        .order_by(RequestStatsDaily.date)
+        .all()
+    )
+    return {"granularity": "daily", "series": [{"t": r.date.isoformat(), "count": r.count} for r in rows]}
+
+
+def get_status_code_breakdown(from_date: date, to_date: date) -> dict:
+    start, end = day_bounds(from_date, to_date)
+    bucket = case(
+        (RequestStatsHourly.status_code < 300, "2xx"),
+        (RequestStatsHourly.status_code < 400, "3xx"),
+        (RequestStatsHourly.status_code < 500, "4xx"),
+        else_="5xx",
+    )
+    rows = (
+        db.session.query(bucket.label("bucket"), func.sum(RequestStatsHourly.count).label("count"))
+        .filter(RequestStatsHourly.date_hour >= start, RequestStatsHourly.date_hour < end)
+        .group_by("bucket")
+        .all()
+    )
+    breakdown = {"2xx": 0, "3xx": 0, "4xx": 0, "5xx": 0}
+    for r in rows:
+        breakdown[r.bucket] = r.count
+    return breakdown
+
+
+def get_top_urls(from_date: date, to_date: date, page: int, per_page: int) -> tuple[list[list], int]:
+    """Top URLs by hits. NOTE (flagged in Ch08): only available within the
+    ~90-day hourly retention window (Ch06) — request_stats_daily has no
+    path column, so a wider range returns nothing here.
+    """
+    start, end = day_bounds(from_date, to_date)
+    hits = func.sum(RequestStatsHourly.count)
+    errors = func.sum(case((RequestStatsHourly.status_code >= 400, RequestStatsHourly.count), else_=0))
+    bytes_sum = func.sum(RequestStatsHourly.bytes_sent_sum)
+
+    base_query = (
+        db.session.query(RequestStatsHourly.path, hits.label("hits"), bytes_sum.label("bytes_sum"), errors.label("errors"))
+        .filter(RequestStatsHourly.date_hour >= start, RequestStatsHourly.date_hour < end)
+        .group_by(RequestStatsHourly.path)
+    )
+    total = base_query.count()
+    rows = base_query.order_by(hits.desc()).offset((page - 1) * per_page).limit(per_page).all()
+
+    results = [
+        [
+            r.path,
+            r.hits,
+            round(r.bytes_sum / r.hits, 1) if r.hits else 0.0,
+            round(r.errors / r.hits * 100, 2) if r.hits else 0.0,
+        ]
+        for r in rows
+    ]
+    return results, total
+
+
+def get_top_referrers(from_date: date, to_date: date, page: int, per_page: int) -> tuple[list[list], int]:
+    """Domain-bucketed referrers (Method A, Ch08 follow-up)."""
+    hits = func.sum(ReferrerStatsDaily.count)
+    base_query = (
+        db.session.query(ReferrerStatsDaily.referrer_domain, hits.label("hits"))
+        .filter(ReferrerStatsDaily.date >= from_date, ReferrerStatsDaily.date <= to_date)
+        .group_by(ReferrerStatsDaily.referrer_domain)
+    )
+    total = base_query.count()
+    rows = base_query.order_by(hits.desc()).offset((page - 1) * per_page).limit(per_page).all()
+    return [[r.referrer_domain, r.hits] for r in rows], total
+
+
+def get_browser_breakdown(from_date: date, to_date: date) -> dict:
+    """Human-only (bots excluded at rollup-write time, Ch08)."""
+    browser_rows = (
+        db.session.query(BrowserStatsDaily.browser, func.sum(BrowserStatsDaily.count).label("count"))
+        .filter(BrowserStatsDaily.date >= from_date, BrowserStatsDaily.date <= to_date)
+        .group_by(BrowserStatsDaily.browser)
+        .order_by(func.sum(BrowserStatsDaily.count).desc())
+        .all()
+    )
+    os_rows = (
+        db.session.query(BrowserStatsDaily.os, func.sum(BrowserStatsDaily.count).label("count"))
+        .filter(BrowserStatsDaily.date >= from_date, BrowserStatsDaily.date <= to_date)
+        .group_by(BrowserStatsDaily.os)
+        .order_by(func.sum(BrowserStatsDaily.count).desc())
+        .all()
+    )
+    return {
+        "by_browser": [{"name": r.browser, "count": r.count} for r in browser_rows],
+        "by_os": [{"name": r.os, "count": r.count} for r in os_rows],
+    }

app/blueprints/overview/routes.py

@@ -0,0 +1,94 @@
+"""Overview blueprint routes (Chapter 08 / Chapter 11's route table)."""
+from __future__ import annotations
+
+from flask import current_app, jsonify, request
+
+from app.blueprints.overview import bp
+from app.blueprints.overview.queries import (
+    get_browser_breakdown,
+    get_kpis,
+    get_status_code_breakdown,
+    get_top_referrers,
+    get_top_urls,
+    get_traffic_chart_series,
+)
+from app.services.background import resume_incomplete_files
+from app.utils.dates import parse_date_range
+from app.utils.htmx import render_htmx_aware
+from app.utils.pagination import parse_pagination
+
+
+@bp.route("/overview")
+def overview():
+    """Full page on first load, HTMX partial on tab switch / range change (Ch04).
+
+    Also opportunistically resumes any log_files stuck in queued/processing
+    (Ch12 simplification: the replacement for cron's "there's always a
+    next tick" guarantee — see app/services/background.py).
+    """
+    resume_incomplete_files(current_app._get_current_object())
+    from_date, to_date = parse_date_range(request)
+    return render_htmx_aware(
+        request,
+        full_template="overview/index.html",
+        partial_template="overview/_content.html",
+        from_date=from_date,
+        to_date=to_date,
+    )
+
+
+@bp.get("/api/overview/kpis")
+def api_kpis():
+    from_date, to_date = parse_date_range(request)
+    return jsonify(data=get_kpis(from_date, to_date), meta={"from": from_date.isoformat(), "to": to_date.isoformat()})
+
+
+@bp.get("/api/overview/traffic-chart")
+def api_traffic_chart():
+    from_date, to_date = parse_date_range(request)
+    return jsonify(
+        data=get_traffic_chart_series(from_date, to_date),
+        meta={"from": from_date.isoformat(), "to": to_date.isoformat()},
+    )
+
+
+@bp.get("/api/overview/status-codes")
+def api_status_codes():
+    from_date, to_date = parse_date_range(request)
+    return jsonify(
+        data=get_status_code_breakdown(from_date, to_date),
+        meta={"from": from_date.isoformat(), "to": to_date.isoformat()},
+    )
+
+
+@bp.get("/api/overview/top-urls")
+def api_top_urls():
+    from_date, to_date = parse_date_range(request)
+    page, per_page = parse_pagination(request)
+    rows, total = get_top_urls(from_date, to_date, page, per_page)
+    return jsonify(
+        data={"rows": rows, "total": total},
+        meta={"from": from_date.isoformat(), "to": to_date.isoformat(), "page": page, "per_page": per_page},
+    )
+
+
+@bp.get("/api/overview/top-referrers")
+def api_top_referrers():
+    from_date, to_date = parse_date_range(request)
+    page, per_page = parse_pagination(request)
+    rows, total = get_top_referrers(from_date, to_date, page, per_page)
+    return jsonify(
+        data={"rows": rows, "total": total},
+        meta={"from": from_date.isoformat(), "to": to_date.isoformat(), "page": page, "per_page": per_page},
+    )
+
+
+@bp.get("/api/overview/browser-breakdown")
+def api_browser_breakdown():
+    """Chapter 11 addition (Method A, Ch08 follow-up) — not in the
+    original route table; see docs/api-contract-final.md."""
+    from_date, to_date = parse_date_range(request)
+    return jsonify(
+        data=get_browser_breakdown(from_date, to_date),
+        meta={"from": from_date.isoformat(), "to": to_date.isoformat()},
+    )

app/blueprints/overview/templates/overview/_content.html

@@ -0,0 +1,286 @@
+<div id="overview-content"
+     hx-get="{{ url_for('overview.overview') }}"
+     hx-trigger="change from:#date-range-form"
+     hx-include="#date-range-form"
+     hx-target="#overview-content"
+     hx-swap="outerHTML"
+     data-from="{{ from_date.isoformat() }}"
+     data-to="{{ to_date.isoformat() }}">
+
+  <div class="flex flex-wrap items-end justify-between gap-4 mb-6">
+    <div>
+      <h1 class="font-display font-bold text-xl">Overview</h1>
+      <p class="text-sm text-muted dark:text-muted-dark">What happened on your site recently</p>
+    </div>
+    <form id="date-range-form" class="flex gap-3 items-end">
+      <label class="text-sm text-muted dark:text-muted-dark">From
+        <input type="date" name="from" value="{{ from_date.isoformat() }}"
+               class="block border border-line dark:border-line-dark rounded-md px-2 py-1 mt-1 bg-surface dark:bg-surface-dark text-ink dark:text-ink-dark font-data text-sm">
+      </label>
+      <label class="text-sm text-muted dark:text-muted-dark">To
+        <input type="date" name="to" value="{{ to_date.isoformat() }}"
+               class="block border border-line dark:border-line-dark rounded-md px-2 py-1 mt-1 bg-surface dark:bg-surface-dark text-ink dark:text-ink-dark font-data text-sm">
+      </label>
+    </form>
+  </div>
+
+  <!-- Upload panel — real byte-transfer progress via XHR (uploads.js), then
+       a real parse-progress bar (Ch12) once the file is queued. No cron
+       required: processing starts automatically in the background. -->
+  <div class="mb-6 border border-line dark:border-line-dark rounded-xl p-4 bg-surface dark:bg-surface-dark">
+    <h2 class="font-display font-semibold text-sm mb-3">Upload a log file</h2>
+    <form id="upload-form" action="{{ url_for('uploads.upload_log_file') }}"
+          class="flex flex-wrap gap-3 items-center">
+      <input type="file" name="logfile" accept=".log,.txt,.gz" required
+             class="text-sm text-muted dark:text-muted-dark file:mr-3 file:py-1.5 file:px-3 file:rounded-md file:border-0 file:bg-accent/10 file:text-accent dark:file:bg-accent-dark/15 dark:file:text-accent-dark file:text-sm file:font-medium hover:file:bg-accent/20 dark:hover:file:bg-accent-dark/25 file:cursor-pointer cursor-pointer">
+      <select name="server_type" class="border border-line dark:border-line-dark rounded-md px-2 py-1.5 text-sm bg-surface dark:bg-surface-dark text-ink dark:text-ink-dark">
+        <option value="apache">Apache</option>
+        <option value="litespeed">LiteSpeed</option>
+      </select>
+      <input type="text" name="format_string" placeholder="LogFormat (optional — defaults to Combined)"
+             class="border border-line dark:border-line-dark rounded-md px-2 py-1.5 text-sm flex-1 min-w-[220px] bg-surface dark:bg-surface-dark text-ink dark:text-ink-dark placeholder:text-muted dark:placeholder:text-muted-dark">
+      <button type="submit" class="bg-accent dark:bg-accent-dark text-white dark:text-paper-dark text-sm font-medium rounded-md px-4 py-1.5 hover:opacity-90 transition-opacity">
+        Upload
+      </button>
+    </form>
+
+    <div id="upload-progress-wrap" class="hidden mt-3">
+      <div class="flex items-center justify-between text-sm mb-1">
+        <span id="upload-progress-label" class="text-muted dark:text-muted-dark">Uploading…</span>
+      </div>
+      <div class="w-full h-2 rounded-full bg-line dark:bg-line-dark overflow-hidden">
+        <div id="upload-progress-bar" class="h-full rounded-full bg-accent dark:bg-accent-dark transition-all duration-150" style="width: 0%"></div>
+      </div>
+    </div>
+
+    <div id="upload-result" class="mt-3"></div>
+  </div>
+
+  <!-- Uploaded files — select and delete (project-owner follow-up
+       request). Independent of the date-range picker above: this lists
+       uploads by when they arrived, not by which log dates they cover
+       (a single file can span many dates). Selection is intentionally
+       scoped to the currently-rendered page/sort/search view — it
+       doesn't try to persist across a Grid.js re-render, which keeps
+       "what's selected" always visually honest at the cost of losing
+       selection if you page/sort/search mid-selection. -->
+  <div class="mb-6 border border-line dark:border-line-dark rounded-xl bg-surface dark:bg-surface-dark overflow-hidden">
+    <div class="flex items-center justify-between p-4 pb-3">
+      <h2 class="font-display font-semibold text-sm">Uploaded files</h2>
+      <button type="button" id="delete-selected-btn" disabled
+              class="text-sm font-medium rounded-md px-3 py-1.5 border border-danger/40 dark:border-danger-dark/40 text-danger dark:text-danger-dark opacity-40 cursor-not-allowed disabled:opacity-40 enabled:opacity-100 enabled:hover:bg-danger/10 dark:enabled:hover:bg-danger-dark/10 transition-opacity">
+        Delete selected
+      </button>
+    </div>
+    <div id="uploaded-files-grid" class="px-4 pb-4" data-endpoint="{{ url_for('uploads.list_uploads') }}"></div>
+  </div>
+
+  <!-- KPI readout — the one deliberate signature treatment: values set in
+       tabular mono, like numbers straight off the log line, with a thin
+       top rule that switches color when a metric needs attention
+       (structure carries information, not just decoration). -->
+  <div id="kpi-cards" class="grid grid-cols-2 sm:grid-cols-3 gap-3 mb-6" data-endpoint="{{ url_for('overview.api_kpis') }}"></div>
+
+  <div class="grid grid-cols-1 lg:grid-cols-2 gap-4 mb-6">
+    <div class="border border-line dark:border-line-dark rounded-xl p-4 bg-surface dark:bg-surface-dark h-72">
+      <h3 class="text-xs font-medium uppercase tracking-wide text-muted dark:text-muted-dark mb-2">Traffic over time</h3>
+      <div class="h-56"><canvas id="traffic-chart" data-endpoint="{{ url_for('overview.api_traffic_chart') }}"></canvas></div>
+    </div>
+    <div class="border border-line dark:border-line-dark rounded-xl p-4 bg-surface dark:bg-surface-dark h-72">
+      <h3 class="text-xs font-medium uppercase tracking-wide text-muted dark:text-muted-dark mb-2">Status codes</h3>
+      <div class="h-56"><canvas id="status-code-chart" data-endpoint="{{ url_for('overview.api_status_codes') }}"></canvas></div>
+    </div>
+  </div>
+
+  <div class="grid grid-cols-1 lg:grid-cols-2 gap-4">
+    <div>
+      <h3 class="text-xs font-medium uppercase tracking-wide text-muted dark:text-muted-dark mb-2">Top URLs</h3>
+      <div id="top-urls-grid" data-endpoint="{{ url_for('overview.api_top_urls') }}"></div>
+    </div>
+    <div>
+      <h3 class="text-xs font-medium uppercase tracking-wide text-muted dark:text-muted-dark mb-2">Top referrers</h3>
+      <div id="top-referrers-grid" data-endpoint="{{ url_for('overview.api_top_referrers') }}"></div>
+    </div>
+  </div>
+
+  <div class="grid grid-cols-1 lg:grid-cols-2 gap-4 mt-6">
+    <div class="border border-line dark:border-line-dark rounded-xl p-4 bg-surface dark:bg-surface-dark h-64">
+      <h3 class="text-xs font-medium uppercase tracking-wide text-muted dark:text-muted-dark mb-2">Browsers</h3>
+      <div class="h-48"><canvas id="browser-breakdown-chart" data-endpoint="{{ url_for('overview.api_browser_breakdown') }}"></canvas></div>
+    </div>
+    <div class="border border-line dark:border-line-dark rounded-xl p-4 bg-surface dark:bg-surface-dark h-64">
+      <h3 class="text-xs font-medium uppercase tracking-wide text-muted dark:text-muted-dark mb-2">Operating systems</h3>
+      <div class="h-48"><canvas id="os-breakdown-chart"></canvas></div>
+    </div>
+  </div>
+
+  <script>
+    (function initOverviewWidgets() {
+      const root = document.getElementById('overview-content');
+      const from = root.dataset.from, to = root.dataset.to;
+      const withRange = (url) => `${url}?from=${from}&to=${to}`;
+
+      window.initUploadForm('upload-form');
+      initUploadedFilesGrid();
+
+      fetch(withRange(document.getElementById('kpi-cards').dataset.endpoint))
+        .then((r) => r.json())
+        .then(({ data }) => renderKpiCards(data));
+
+      const trafficEl = document.getElementById('traffic-chart');
+      fetch(withRange(trafficEl.dataset.endpoint))
+        .then((r) => r.json())
+        .then(({ data }) => window.initChart('traffic-chart', {
+          type: 'line',
+          data: {
+            labels: data.series.map((p) => p.t),
+            datasets: [{
+              label: 'Requests', data: data.series.map((p) => p.count), tension: 0.3,
+              borderColor: '#0E7C86', backgroundColor: '#0E7C8622', fill: true,
+            }],
+          },
+        }));
+
+      const statusEl = document.getElementById('status-code-chart');
+      fetch(withRange(statusEl.dataset.endpoint))
+        .then((r) => r.json())
+        .then(({ data }) => window.initChart('status-code-chart', {
+          type: 'doughnut',
+          data: { labels: Object.keys(data), datasets: [{ data: Object.values(data), backgroundColor: window.KAVOSH_CHART_PALETTE }] },
+        }));
+
+      window.initGrid(
+        'top-urls-grid',
+        document.getElementById('top-urls-grid').dataset.endpoint,
+        [{ name: 'Path' }, { name: 'Hits' }, { name: 'Avg Size (B)' }, { name: 'Error Rate %' }],
+        { from, to },
+      );
+
+      window.initGrid(
+        'top-referrers-grid',
+        document.getElementById('top-referrers-grid').dataset.endpoint,
+        [{ name: 'Referrer Domain' }, { name: 'Hits' }],
+        { from, to },
+      );
+
+      const browserEl = document.getElementById('browser-breakdown-chart');
+      fetch(withRange(browserEl.dataset.endpoint))
+        .then((r) => r.json())
+        .then(({ data }) => {
+          window.initChart('browser-breakdown-chart', {
+            type: 'bar',
+            data: { labels: data.by_browser.map((r) => r.name), datasets: [{ label: 'Browser', data: data.by_browser.map((r) => r.count), backgroundColor: '#0E7C86' }] },
+          });
+          window.initChart('os-breakdown-chart', {
+            type: 'bar',
+            data: { labels: data.by_os.map((r) => r.name), datasets: [{ label: 'OS', data: data.by_os.map((r) => r.count), backgroundColor: '#B8860B' }] },
+          });
+        });
+
+      function renderKpiCards(kpi) {
+        const errorTone = kpi.error_rate_pct >= 5 ? 'danger' : kpi.error_rate_pct >= 1 ? 'warn' : 'ok';
+        const cards = [
+          ['Total requests', kpi.total_requests.toLocaleString(), 'accent'],
+          ['Unique IPs (approx.)', kpi.unique_ips.toLocaleString(), 'accent'],
+          ['Bandwidth', `${(kpi.total_bandwidth_bytes / 1e6).toFixed(1)} MB`, 'accent'],
+          ['Avg response size', `${kpi.avg_response_size_bytes} B`, 'accent'],
+          ['Error rate', `${kpi.error_rate_pct}%`, errorTone],
+          ['Avg requests / day', kpi.avg_requests_per_day.toLocaleString(), 'accent'],
+        ];
+        const toneBorder = {
+          accent: 'border-t-accent dark:border-t-accent-dark',
+          ok: 'border-t-ok dark:border-t-ok-dark',
+          warn: 'border-t-warn dark:border-t-warn-dark',
+          danger: 'border-t-danger dark:border-t-danger-dark',
+        };
+        const el = document.getElementById('kpi-cards');
+        el.innerHTML = cards.map(([label, value, tone]) => `
+          <div class="bg-surface dark:bg-surface-dark rounded-lg border border-line dark:border-line-dark border-t-2 ${toneBorder[tone]} p-3">
+            <p class="text-xs text-muted dark:text-muted-dark uppercase tracking-wide">${label}</p>
+            <p class="font-data text-2xl font-medium mt-0.5">${value}</p>
+          </div>`).join('');
+        if (kpi.peak_day) {
+          el.innerHTML += `
+            <div class="bg-surface dark:bg-surface-dark rounded-lg border border-line dark:border-line-dark border-t-2 border-t-accent dark:border-t-accent-dark p-3 col-span-2 sm:col-span-3">
+              <p class="text-xs text-muted dark:text-muted-dark uppercase tracking-wide">Peak day</p>
+              <p class="font-data text-xl font-medium mt-0.5">${kpi.peak_day.date} <span class="text-muted dark:text-muted-dark text-sm">— ${kpi.peak_day.count.toLocaleString()} requests</span></p>
+            </div>`;
+        }
+      }
+
+      function initUploadedFilesGrid() {
+        const container = document.getElementById('uploaded-files-grid');
+        const deleteBtn = document.getElementById('delete-selected-btn');
+        const selectedIds = new Set();
+
+        function updateDeleteButton() {
+          deleteBtn.disabled = selectedIds.size === 0;
+          deleteBtn.textContent = selectedIds.size > 0 ? `Delete selected (${selectedIds.size})` : 'Delete selected';
+        }
+
+        // Selection is intentionally NOT restored across a Grid.js
+        // re-render (page/sort/search) — simpler and always visually
+        // honest, at the cost of losing selection if you page away
+        // mid-selection. See the comment above the HTML for this panel.
+        container.addEventListener('change', (e) => {
+          if (!e.target.matches('.file-select-checkbox')) return;
+          const id = parseInt(e.target.value, 10);
+          if (e.target.checked) selectedIds.add(id); else selectedIds.delete(id);
+          updateDeleteButton();
+        });
+
+        const columns = [
+          {
+            name: '',
+            formatter: (cell, row) => {
+              const rawStatus = row.cells[6].data;
+              const disabled = rawStatus === 'processing';
+              return window.gridHtml(
+                `<input type="checkbox" class="file-select-checkbox w-4 h-4 rounded border-line dark:border-line-dark text-accent focus:ring-accent cursor-pointer disabled:cursor-not-allowed disabled:opacity-40"
+                        value="${cell}" ${disabled ? 'disabled title="Still analyzing — wait for it to finish"' : ''}>`
+              );
+            },
+          },
+          { name: 'Filename' },
+          { name: 'Server' },
+          { name: 'Status' },
+          { name: 'Uploaded' },
+          { name: 'Size' },
+          { name: 'Raw Status', hidden: true },
+        ];
+
+        const filesGrid = window.initGrid(
+          'uploaded-files-grid',
+          container.dataset.endpoint,
+          columns,
+          {},
+        );
+
+        deleteBtn.addEventListener('click', () => {
+          if (selectedIds.size === 0) return;
+          if (!confirm(`Delete ${selectedIds.size} file(s) and all data derived from them? This can't be undone.`)) return;
+
+          const token = document.querySelector('meta[name="csrf-token"]')?.content;
+          fetch('{{ url_for("uploads.bulk_delete_uploads") }}', {
+            method: 'DELETE',
+            headers: { 'Content-Type': 'application/json', 'X-CSRFToken': token || '' },
+            body: JSON.stringify({ ids: Array.from(selectedIds) }),
+          })
+            .then((r) => r.json())
+            .then(({ data }) => {
+              selectedIds.clear();
+              updateDeleteButton();
+              filesGrid.forceRender();
+              if (data.skipped && data.skipped.length) {
+                alert('Some files were skipped:\n' + data.skipped.map((s) => `#${s.id} — ${s.reason}`).join('\n'));
+              }
+              // Deleting a file can change rollups for any date it
+              // touched — refresh the whole page's KPIs/charts via the
+              // same mechanism the date-range picker itself uses.
+              window.htmx.trigger(document.getElementById('date-range-form'), 'change');
+            });
+        });
+      }
+    })();
+  </script>
+</div>

app/blueprints/overview/templates/overview/index.html

@@ -0,0 +1,5 @@
+{% extends "base.html" %}
+{% block title %}Overview — Kavosh{% endblock %}
+{% block content %}
+  {% include "overview/_content.html" %}
+{% endblock %}

app/blueprints/security/__init__.py

@@ -0,0 +1,13 @@
+from flask import Blueprint
+from flask_login import login_required
+
+bp = Blueprint("security", __name__, template_folder="templates")
+
+
+@bp.before_request
+@login_required
+def require_login():
+    pass
+
+
+from app.blueprints.security import routes  # noqa: E402,F401  registers routes

app/blueprints/security/queries.py

@@ -0,0 +1,154 @@
+"""Context-builder query functions for the Security tab (Chapter 10),
+with the IP investigation panel's traffic breakdown upgraded to full-
+traffic data (bounded per-IP rollup, added as an explicit follow-up to
+Chapter 10's scope gap).
+"""
+from __future__ import annotations
+
+from collections import defaultdict
+from datetime import date
+
+from sqlalchemy import func
+
+from app.extensions import db
+from app.models.blocklist_suggestion import BlocklistSuggestion
+from app.models.bot_hit import BotHit
+from app.models.ip_registry import IPRegistry
+from app.models.ip_traffic_stats import IpPathStatsDaily, IpStatusStatsDaily
+from app.models.suspicious_event import SuspiciousEvent
+from app.services.severity_scoring import SeverityInputs, compute_effective_severity
+from app.utils.dates import day_bounds
+
+IP_HISTORY_EVENT_LIMIT = 50
+IP_HISTORY_PATH_LIMIT = 20
+
+
+def get_suspicious_events(
+    from_date: date, to_date: date, severity: str | None, rule_type: str | None, page: int, per_page: int,
+) -> tuple[list[list], int]:
+    """suspicious_events is small/indexed/indefinitely-retained (Ch06) —
+    same precedent as Ch09's bot_hits queries, so loading + escalating in
+    Python doesn't violate Ch03 rule 6 (that targets raw per-request rows).
+    """
+    start, end = day_bounds(from_date, to_date)
+    query = db.session.query(SuspiciousEvent).filter(
+        SuspiciousEvent.timestamp >= start, SuspiciousEvent.timestamp < end
+    )
+    if rule_type:
+        query = query.filter(SuspiciousEvent.rule_matched.like(f"{rule_type}:%"))
+    events = query.order_by(SuspiciousEvent.timestamp.desc()).all()
+
+    by_ip: dict[str, list[SuspiciousEvent]] = defaultdict(list)
+    for e in events:
+        by_ip[e.ip].append(e)
+
+    enriched = []
+    for e in events:
+        ip_events = by_ip[e.ip]
+        timestamps = sorted(ev.timestamp for ev in ip_events)
+        avg_interval = (
+            (timestamps[-1] - timestamps[0]).total_seconds() / (len(timestamps) - 1)
+            if len(timestamps) > 1 else None
+        )
+        effective = compute_effective_severity(
+            SeverityInputs(base_severity=e.severity, ip_event_count=len(ip_events), avg_interval_seconds=avg_interval)
+        )
+        if severity and effective != severity:
+            continue
+        enriched.append([e.timestamp.isoformat(), e.ip, e.path, e.rule_matched, effective])
+
+    total = len(enriched)
+    offset = (page - 1) * per_page
+    return enriched[offset : offset + per_page], total
+
+
+def get_sensitive_path_summary(from_date: date, to_date: date) -> list[dict]:
+    """Grouped by request path; filtered to Ch07's sensitive_path rule
+    category. One ranked list, not sub-grouped into config/admin/VCS —
+    Ch07's dictionary has no such taxonomy to reuse.
+    """
+    start, end = day_bounds(from_date, to_date)
+    rows = (
+        db.session.query(
+            SuspiciousEvent.path,
+            func.count().label("hit_count"),
+            func.count(func.distinct(SuspiciousEvent.ip)).label("distinct_ip_count"),
+        )
+        .filter(
+            SuspiciousEvent.timestamp >= start, SuspiciousEvent.timestamp < end,
+            SuspiciousEvent.rule_matched.like("sensitive_path:%"),
+        )
+        .group_by(SuspiciousEvent.path)
+        .order_by(func.count().desc())
+        .all()
+    )
+    return [{"path": r.path, "hit_count": r.hit_count, "distinct_ip_count": r.distinct_ip_count} for r in rows]
+
+
+def get_ip_history(ip: str) -> dict | None:
+    """Pulled from ip_registry (identity + true total_requests), plus
+    bot_hits/suspicious_events (flagged activity), plus the bounded
+    per-IP traffic rollup (top_paths / status_code_distribution — true
+    full-traffic breakdown, added as a follow-up to Ch10's original scope
+    gap). Retention caveat: the per-IP rollup covers roughly the last 30
+    days (see aggregator.py / flask cleanup).
+    """
+    registry = db.session.get(IPRegistry, ip)
+    if registry is None:
+        return None
+
+    path_rows = (
+        db.session.query(IpPathStatsDaily.path, func.sum(IpPathStatsDaily.count).label("count"))
+        .filter(IpPathStatsDaily.ip == ip)
+        .group_by(IpPathStatsDaily.path)
+        .order_by(func.sum(IpPathStatsDaily.count).desc())
+        .limit(IP_HISTORY_PATH_LIMIT)
+        .all()
+    )
+    status_rows = (
+        db.session.query(IpStatusStatsDaily.status_bucket, func.sum(IpStatusStatsDaily.count).label("count"))
+        .filter(IpStatusStatsDaily.ip == ip)
+        .group_by(IpStatusStatsDaily.status_bucket)
+        .all()
+    )
+
+    bot_rows = (
+        db.session.query(BotHit).filter(BotHit.ip == ip)
+        .order_by(BotHit.timestamp.desc()).limit(IP_HISTORY_EVENT_LIMIT).all()
+    )
+    suspicious_rows = (
+        db.session.query(SuspiciousEvent).filter(SuspiciousEvent.ip == ip)
+        .order_by(SuspiciousEvent.timestamp.desc()).limit(IP_HISTORY_EVENT_LIMIT).all()
+    )
+    spoofed_bot_names = sorted({b.bot_name for b in bot_rows if not b.verified})
+
+    return {
+        "ip": ip,
+        "first_seen": registry.first_seen.isoformat(),
+        "last_seen": registry.last_seen.isoformat(),
+        "total_requests": registry.total_requests,
+        "reputation_score": registry.reputation_score,
+        "is_flagged": registry.is_flagged,
+        "spoofed_bot_names": spoofed_bot_names,
+        "top_paths": [[r.path, r.count] for r in path_rows],
+        "status_code_distribution": {r.status_bucket: r.count for r in status_rows},
+        "traffic_window_note": "Path/status breakdown reflects roughly the last 30 days (bounded retention).",
+        "recent_suspicious_events": [
+            {"timestamp": s.timestamp.isoformat(), "path": s.path, "rule_matched": s.rule_matched, "severity": s.severity}
+            for s in suspicious_rows
+        ],
+    }
+
+
+def format_blocklist(suggestions: list[BlocklistSuggestion], fmt: str) -> str:
+    """Ch10: '.htaccess Deny/iptables/fail2ban-style'. 'plain' (a bare IP
+    list) is the most portable interpretation of "fail2ban-style input"
+    without assuming a specific fail2ban jail configuration Ch10 doesn't
+    specify.
+    """
+    ips = [s.ip for s in suggestions]
+    if fmt == "htaccess":
+        return "".join(f"Deny from {ip}\n" for ip in ips)
+    if fmt == "iptables":
+        return "".join(f"iptables -A INPUT -s {ip} -j DROP\n" for ip in ips)
+    return "".join(f"{ip}\n" for ip in ips)

app/blueprints/security/routes.py

@@ -0,0 +1,71 @@
+from __future__ import annotations
+
+from flask import Response, jsonify, render_template, request
+
+from app.blueprints.security import bp
+from app.blueprints.security.queries import (
+    format_blocklist, get_ip_history, get_sensitive_path_summary, get_suspicious_events,
+)
+from app.extensions import db
+from app.models.blocklist_suggestion import BlocklistSuggestion
+from app.utils.dates import parse_date_range
+from app.utils.htmx import render_htmx_aware
+from app.utils.pagination import parse_pagination
+
+
+@bp.route("/security")
+def security():
+    from_date, to_date = parse_date_range(request)
+    severity = request.args.get("severity") or ""
+    rule_type = request.args.get("rule_type") or ""
+    return render_htmx_aware(
+        request, full_template="security/index.html", partial_template="security/_content.html",
+        from_date=from_date, to_date=to_date, severity=severity, rule_type=rule_type,
+    )
+
+
+@bp.get("/api/security/events")
+def api_security_events():
+    from_date, to_date = parse_date_range(request)
+    page, per_page = parse_pagination(request)
+    severity = request.args.get("severity") or None
+    rule_type = request.args.get("rule_type") or None
+    rows, total = get_suspicious_events(from_date, to_date, severity, rule_type, page, per_page)
+    return jsonify(
+        data={"rows": rows, "total": total},
+        meta={"from": from_date.isoformat(), "to": to_date.isoformat(), "page": page, "per_page": per_page},
+    )
+
+
+@bp.get("/api/security/sensitive-paths")
+def api_sensitive_paths():
+    from_date, to_date = parse_date_range(request)
+    return jsonify(data=get_sensitive_path_summary(from_date, to_date), meta={"from": from_date.isoformat(), "to": to_date.isoformat()})
+
+
+@bp.get("/api/security/ip/<ip>")
+def api_ip_history(ip: str):
+    history = get_ip_history(ip)
+    if history is None:
+        return render_template("security/_ip_not_found.html", ip=ip), 404
+    return render_template("security/_ip_history.html", ip_data=history)
+
+
+@bp.get("/api/security/export-blocklist")
+def api_export_blocklist():
+    fmt = request.args.get("format", "plain")
+    include_all = request.args.get("all", "false").lower() == "true"
+    query = BlocklistSuggestion.query
+    if not include_all:
+        query = query.filter_by(exported=False)
+    suggestions = query.order_by(BlocklistSuggestion.created_at).all()
+
+    body = format_blocklist(suggestions, fmt)
+    for s in suggestions:
+        s.exported = True
+    db.session.commit()
+
+    return Response(
+        body, mimetype="text/plain",
+        headers={"Content-Disposition": "attachment; filename=kavosh-blocklist.txt"},
+    )

app/blueprints/security/templates/security/_content.html

@@ -0,0 +1,125 @@
+<div id="security-content"
+     hx-get="{{ url_for('security.security') }}"
+     hx-trigger="change from:#security-filter-form"
+     hx-include="#security-filter-form"
+     hx-target="#security-content"
+     hx-swap="outerHTML"
+     data-from="{{ from_date.isoformat() }}"
+     data-to="{{ to_date.isoformat() }}"
+     data-severity="{{ severity }}"
+     data-rule-type="{{ rule_type }}">
+
+  <div class="flex flex-wrap items-end justify-between gap-4 mb-6">
+    <div>
+      <h1 class="font-display font-bold text-xl">Suspicious Requests &amp; IP History</h1>
+      <p class="text-sm text-muted dark:text-muted-dark">Who's poking at your site, and how hard</p>
+    </div>
+    <form id="security-filter-form" class="flex flex-wrap gap-3 items-end">
+      <label class="text-sm text-muted dark:text-muted-dark">From
+        <input type="date" name="from" value="{{ from_date.isoformat() }}" class="block border border-line dark:border-line-dark rounded-md px-2 py-1 mt-1 bg-surface dark:bg-surface-dark text-ink dark:text-ink-dark font-data text-sm">
+      </label>
+      <label class="text-sm text-muted dark:text-muted-dark">To
+        <input type="date" name="to" value="{{ to_date.isoformat() }}" class="block border border-line dark:border-line-dark rounded-md px-2 py-1 mt-1 bg-surface dark:bg-surface-dark text-ink dark:text-ink-dark font-data text-sm">
+      </label>
+      <label class="text-sm text-muted dark:text-muted-dark">Severity
+        <select name="severity" class="block border border-line dark:border-line-dark rounded-md px-2 py-1 mt-1 bg-surface dark:bg-surface-dark text-ink dark:text-ink-dark text-sm">
+          <option value="" {{ 'selected' if not severity }}>All</option>
+          <option value="low" {{ 'selected' if severity == 'low' }}>Low</option>
+          <option value="medium" {{ 'selected' if severity == 'medium' }}>Medium</option>
+          <option value="high" {{ 'selected' if severity == 'high' }}>High</option>
+        </select>
+      </label>
+      <label class="text-sm text-muted dark:text-muted-dark">Rule Type
+        <select name="rule_type" class="block border border-line dark:border-line-dark rounded-md px-2 py-1 mt-1 bg-surface dark:bg-surface-dark text-ink dark:text-ink-dark text-sm">
+          <option value="" {{ 'selected' if not rule_type }}>All</option>
+          <option value="sensitive_path" {{ 'selected' if rule_type == 'sensitive_path' }}>Sensitive Path</option>
+          <option value="injection" {{ 'selected' if rule_type == 'injection' }}>Injection</option>
+          <option value="scanner_ua" {{ 'selected' if rule_type == 'scanner_ua' }}>Scanner UA</option>
+          <option value="spoofed_bot" {{ 'selected' if rule_type == 'spoofed_bot' }}>Spoofed Bot</option>
+        </select>
+      </label>
+    </form>
+  </div>
+
+  <div class="mb-6">
+    <h3 class="text-xs font-medium uppercase tracking-wide text-muted dark:text-muted-dark mb-2">Suspicious events</h3>
+    <div id="suspicious-events-grid" data-endpoint="{{ url_for('security.api_security_events') }}"></div>
+  </div>
+
+  <div class="mb-6">
+    <h3 class="text-xs font-medium uppercase tracking-wide text-muted dark:text-muted-dark mb-2">Sensitive-path probes</h3>
+    <div id="sensitive-paths-panel" class="border border-line dark:border-line-dark rounded-xl bg-surface dark:bg-surface-dark overflow-hidden"
+         data-endpoint="{{ url_for('security.api_sensitive_paths') }}"></div>
+  </div>
+
+  <div class="mb-6 border border-line dark:border-line-dark rounded-xl p-4 bg-surface dark:bg-surface-dark">
+    <h3 class="font-display font-semibold text-sm mb-3">Export blocklist</h3>
+    <div class="flex flex-wrap gap-2 items-center">
+      <a href="{{ url_for('security.api_export_blocklist') }}"
+         class="bg-accent dark:bg-accent-dark text-white dark:text-paper-dark rounded-md px-3 py-1.5 text-sm font-medium hover:opacity-90 transition-opacity">Download new (.txt)</a>
+      <a href="{{ url_for('security.api_export_blocklist', all='true') }}"
+         class="border border-line dark:border-line-dark rounded-md px-3 py-1.5 text-sm hover:bg-paper dark:hover:bg-paper-dark transition-colors">Re-export all</a>
+      <select id="blocklist-format" class="border border-line dark:border-line-dark rounded-md px-2 py-1.5 text-sm bg-surface dark:bg-surface-dark text-ink dark:text-ink-dark" onchange="updateBlocklistLinks(this.value)">
+        <option value="plain">Plain IP list</option>
+        <option value="htaccess">.htaccess Deny</option>
+        <option value="iptables">iptables</option>
+      </select>
+    </div>
+  </div>
+
+  <div id="ip-history-modal" class="fixed inset-0 bg-ink/40 dark:bg-ink-dark/60 items-center justify-center empty:hidden flex z-50"></div>
+
+  <script>
+    (function initSecurityWidgets() {
+      const root = document.getElementById('security-content');
+      const from = root.dataset.from, to = root.dataset.to;
+      const severity = root.dataset.severity, ruleType = root.dataset.ruleType;
+
+      window.initGrid(
+        'suspicious-events-grid',
+        document.getElementById('suspicious-events-grid').dataset.endpoint,
+        [
+          { name: 'Timestamp' },
+          {
+            name: 'IP',
+            formatter: (cell) => window.gridHtml(
+              `<button class="text-accent dark:text-accent-dark underline" hx-get="/api/security/ip/${cell}" hx-target="#ip-history-modal" hx-swap="innerHTML">${cell}</button>`
+            ),
+          },
+          { name: 'Path' }, { name: 'Rule Matched' },
+          {
+            name: 'Severity',
+            formatter: (cell) => {
+              const tone = { low: 'text-muted dark:text-muted-dark', medium: 'text-warn dark:text-warn-dark', high: 'text-danger dark:text-danger-dark' }[cell] || '';
+              return window.gridHtml(`<span class="font-medium ${tone}">${cell}</span>`);
+            },
+          },
+        ],
+        { from, to, severity, rule_type: ruleType },
+      );
+
+      const pathsEl = document.getElementById('sensitive-paths-panel');
+      fetch(`${pathsEl.dataset.endpoint}?from=${from}&to=${to}`)
+        .then((r) => r.json())
+        .then(({ data }) => {
+          pathsEl.innerHTML = data.length
+            ? `<table class="w-full text-sm font-data">
+                 <thead><tr class="text-left text-muted dark:text-muted-dark text-xs uppercase tracking-wide bg-surface-raised dark:bg-surface-raised-dark font-sans">
+                   <th class="px-3 py-2">Path</th><th class="px-3 py-2">Hits</th><th class="px-3 py-2">Distinct IPs</th>
+                 </tr></thead>
+                 <tbody class="divide-y divide-line dark:divide-line-dark">${
+                   data.map((r) => `<tr><td class="px-3 py-2">${r.path}</td><td class="px-3 py-2">${r.hit_count}</td><td class="px-3 py-2">${r.distinct_ip_count}</td></tr>`).join('')
+                 }</tbody></table>`
+            : `<p class="text-sm text-muted dark:text-muted-dark p-4">No sensitive-path probes in range.</p>`;
+        });
+
+      window.updateBlocklistLinks = (fmt) => {
+        document.querySelectorAll('a[href*="export-blocklist"]').forEach((a) => {
+          const url = new URL(a.href, window.location.origin);
+          url.searchParams.set('format', fmt);
+          a.href = url.toString();
+        });
+      };
+    })();
+  </script>
+</div>

app/blueprints/security/templates/security/_ip_history.html

@@ -0,0 +1,26 @@
+<div class="bg-surface dark:bg-surface-dark rounded-xl p-6 max-w-lg w-full relative border border-line dark:border-line-dark">
+  <button class="absolute top-3 right-3 text-muted dark:text-muted-dark hover:text-ink dark:hover:text-ink-dark" onclick="document.getElementById('ip-history-modal').innerHTML=''">
+    <svg class="w-4 h-4"><use href="/static/dist/icons.svg#x"/></svg>
+  </button>
+  <h3 class="font-display font-semibold text-lg mb-3 font-data">{{ ip_data.ip }}</h3>
+  <dl class="text-sm grid grid-cols-2 gap-y-1.5 mb-4 font-data">
+    <dt class="text-muted dark:text-muted-dark font-sans">First seen</dt><dd>{{ ip_data.first_seen }}</dd>
+    <dt class="text-muted dark:text-muted-dark font-sans">Last seen</dt><dd>{{ ip_data.last_seen }}</dd>
+    <dt class="text-muted dark:text-muted-dark font-sans">Total requests</dt><dd>{{ ip_data.total_requests }}</dd>
+    <dt class="text-muted dark:text-muted-dark font-sans">Reputation score</dt><dd>{{ ip_data.reputation_score }}</dd>
+    <dt class="text-muted dark:text-muted-dark font-sans">Flagged</dt>
+    <dd class="{{ 'text-danger dark:text-danger-dark font-medium' if ip_data.is_flagged else '' }}">{{ 'Yes' if ip_data.is_flagged else 'No' }}</dd>
+    {% if ip_data.spoofed_bot_names %}
+    <dt class="text-muted dark:text-muted-dark font-sans">Spoofed bot claims</dt><dd class="text-danger dark:text-danger-dark">{{ ip_data.spoofed_bot_names | join(', ') }}</dd>
+    {% endif %}
+  </dl>
+  <p class="text-xs text-muted dark:text-muted-dark mb-3">{{ ip_data.traffic_window_note }}</p>
+  <h4 class="font-medium text-sm mb-1">Top paths</h4>
+  <ul class="text-sm mb-3 font-data text-ink dark:text-ink-dark space-y-0.5">{% for path, count in ip_data.top_paths %}<li>{{ path }} <span class="text-muted dark:text-muted-dark">— {{ count }}</span></li>{% endfor %}</ul>
+  <h4 class="font-medium text-sm mb-1">Status codes</h4>
+  <ul class="text-sm mb-3 font-data text-ink dark:text-ink-dark space-y-0.5">{% for code, count in ip_data.status_code_distribution.items() %}<li>{{ code }} <span class="text-muted dark:text-muted-dark">— {{ count }}</span></li>{% endfor %}</ul>
+  {% if ip_data.recent_suspicious_events %}
+  <h4 class="font-medium text-sm mb-1">Recent flagged events</h4>
+  <ul class="text-sm font-data text-ink dark:text-ink-dark space-y-0.5">{% for e in ip_data.recent_suspicious_events %}<li>{{ e.timestamp }} — {{ e.path }} <span class="text-muted dark:text-muted-dark">({{ e.rule_matched }}, {{ e.severity }})</span></li>{% endfor %}</ul>
+  {% endif %}
+</div>

app/blueprints/security/templates/security/_ip_not_found.html

@@ -0,0 +1,4 @@
+<div class="bg-surface dark:bg-surface-dark rounded-xl p-6 max-w-sm w-full border border-line dark:border-line-dark">
+  <p class="text-sm text-ink dark:text-ink-dark">No history found for <span class="font-data">{{ ip }}</span> — it hasn't been seen yet.</p>
+  <button onclick="document.getElementById('ip-history-modal').innerHTML=''" class="mt-3 text-sm text-accent dark:text-accent-dark hover:underline">Close</button>
+</div>

app/blueprints/security/templates/security/index.html

@@ -0,0 +1,5 @@
+{% extends "base.html" %}
+{% block title %}Security — Kavosh{% endblock %}
+{% block content %}
+  {% include "security/_content.html" %}
+{% endblock %}

app/blueprints/seo/__init__.py

@@ -0,0 +1,13 @@
+from flask import Blueprint
+from flask_login import login_required
+
+bp = Blueprint("seo", __name__, template_folder="templates")
+
+
+@bp.before_request
+@login_required
+def require_login():
+    pass
+
+
+from app.blueprints.seo import routes  # noqa: E402,F401  registers routes

app/blueprints/seo/queries.py

@@ -0,0 +1,140 @@
+"""Context-builder query functions for the SEO tab (Chapter 09).
+
+Per Ch09's own Output instruction, bot-related widgets query bot_hits
+directly (small, indefinitely-retained, indexed — Ch06) rather than a new
+rollup; only the human-side of the crawled-vs-visited comparison needs
+the new human_path_stats_daily table.
+"""
+from __future__ import annotations
+
+from collections import defaultdict
+from datetime import date
+
+from sqlalchemy import case, func
+
+from app.extensions import db
+from app.models.bot_hit import BotHit
+from app.models.human_path_stats import HumanPathStatsDaily
+from app.utils.dates import day_bounds
+
+# ASSUMPTION (flagged): Ch09 doesn't define "major crawler" for the
+# crawl-frequency chart's series cap.
+TOP_N_BOTS_FOR_CHART = 6
+
+
+def get_bot_summary(from_date: date, to_date: date) -> list[dict]:
+    start, end = day_bounds(from_date, to_date)
+    rows = (
+        db.session.query(
+            BotHit.bot_name,
+            func.count().label("hits"),
+            func.sum(case((BotHit.verified.is_(True), 1), else_=0)).label("verified_hits"),
+            func.max(BotHit.timestamp).label("last_seen"),
+        )
+        .filter(BotHit.timestamp >= start, BotHit.timestamp < end)
+        .group_by(BotHit.bot_name)
+        .order_by(func.count().desc())
+        .all()
+    )
+    return [
+        {
+            "bot_name": r.bot_name,
+            "hits": r.hits,
+            "verified_pct": round(r.verified_hits / r.hits * 100, 1) if r.hits else 0.0,
+            "last_seen": r.last_seen.isoformat() if r.last_seen else None,
+        }
+        for r in rows
+    ]
+
+
+def get_crawl_chart_data(from_date: date, to_date: date, bot_name: str | None) -> dict:
+    start, end = day_bounds(from_date, to_date)
+    base_filters = [BotHit.timestamp >= start, BotHit.timestamp < end]
+
+    if bot_name:
+        allowed_bots = [bot_name]
+    else:
+        top_rows = (
+            db.session.query(BotHit.bot_name, func.count().label("hits"))
+            .filter(*base_filters)
+            .group_by(BotHit.bot_name)
+            .order_by(func.count().desc())
+            .limit(TOP_N_BOTS_FOR_CHART)
+            .all()
+        )
+        allowed_bots = [r.bot_name for r in top_rows]
+
+    if not allowed_bots:
+        return {"series": []}
+
+    rows = (
+        db.session.query(func.date(BotHit.timestamp).label("day"), BotHit.bot_name, func.count().label("count"))
+        .filter(*base_filters, BotHit.bot_name.in_(allowed_bots))
+        .group_by("day", BotHit.bot_name)
+        .order_by("day")
+        .all()
+    )
+    points_by_bot: dict[str, list[dict]] = defaultdict(list)
+    for r in rows:
+        points_by_bot[r.bot_name].append({"t": r.day, "count": r.count})
+
+    return {"series": [{"bot_name": b, "points": points_by_bot.get(b, [])} for b in allowed_bots]}
+
+
+def get_bot_status_codes(from_date: date, to_date: date) -> dict:
+    start, end = day_bounds(from_date, to_date)
+    bucket = case(
+        (BotHit.status_code < 300, "2xx"),
+        (BotHit.status_code < 400, "3xx"),
+        (BotHit.status_code < 500, "4xx"),
+        else_="5xx",
+    )
+    rows = (
+        db.session.query(bucket.label("bucket"), func.count().label("count"))
+        .filter(BotHit.timestamp >= start, BotHit.timestamp < end)
+        .group_by("bucket")
+        .all()
+    )
+    breakdown = {"2xx": 0, "3xx": 0, "4xx": 0, "5xx": 0}
+    for r in rows:
+        breakdown[r.bucket] = r.count
+
+    # Ch09 calls out 404 by name specifically, not just the 4xx bucket.
+    not_found_404 = (
+        db.session.query(func.count())
+        .filter(BotHit.timestamp >= start, BotHit.timestamp < end, BotHit.status_code == 404)
+        .scalar()
+    )
+    return {"breakdown": breakdown, "not_found_404": not_found_404 or 0}
+
+
+def get_crawled_vs_visited(from_date: date, to_date: date, page: int, per_page: int) -> tuple[list[list], int]:
+    """Diffed table: one row per path, bot hits vs. human hits."""
+    start, end = day_bounds(from_date, to_date)
+    bot_rows = (
+        db.session.query(BotHit.path, func.count().label("hits"))
+        .filter(BotHit.timestamp >= start, BotHit.timestamp < end)
+        .group_by(BotHit.path)
+        .all()
+    )
+    human_rows = (
+        db.session.query(HumanPathStatsDaily.path, func.sum(HumanPathStatsDaily.count).label("hits"))
+        .filter(HumanPathStatsDaily.date >= from_date, HumanPathStatsDaily.date <= to_date)
+        .group_by(HumanPathStatsDaily.path)
+        .all()
+    )
+    bot_counts = {r.path: r.hits for r in bot_rows}
+    human_counts = {r.path: r.hits for r in human_rows}
+
+    combined = []
+    for path in set(bot_counts) | set(human_counts):
+        b, h = bot_counts.get(path, 0), human_counts.get(path, 0)
+        total = b + h
+        combined.append([path, b, h, round(b / total * 100, 1) if total else 0.0])
+
+    # ASSUMPTION (flagged): sorted by bot hits desc — Ch09 doesn't specify.
+    combined.sort(key=lambda row: row[1], reverse=True)
+
+    total_count = len(combined)
+    offset = (page - 1) * per_page
+    return combined[offset : offset + per_page], total_count

app/blueprints/seo/routes.py

@@ -0,0 +1,56 @@
+from __future__ import annotations
+
+from flask import jsonify, request
+
+from app.blueprints.seo import bp
+from app.blueprints.seo.queries import (
+    get_bot_status_codes,
+    get_bot_summary,
+    get_crawl_chart_data,
+    get_crawled_vs_visited,
+)
+from app.utils.dates import parse_date_range
+from app.utils.htmx import render_htmx_aware
+from app.utils.pagination import parse_pagination
+
+
+@bp.route("/seo")
+def seo():
+    from_date, to_date = parse_date_range(request)
+    return render_htmx_aware(
+        request, full_template="seo/index.html", partial_template="seo/_content.html",
+        from_date=from_date, to_date=to_date,
+    )
+
+
+@bp.get("/api/seo/bot-summary")
+def api_bot_summary():
+    from_date, to_date = parse_date_range(request)
+    return jsonify(data=get_bot_summary(from_date, to_date), meta={"from": from_date.isoformat(), "to": to_date.isoformat()})
+
+
+@bp.get("/api/seo/crawl-chart-data")
+def api_crawl_chart_data():
+    from_date, to_date = parse_date_range(request)
+    bot_name = request.args.get("bot")
+    return jsonify(
+        data=get_crawl_chart_data(from_date, to_date, bot_name),
+        meta={"from": from_date.isoformat(), "to": to_date.isoformat(), "bot": bot_name},
+    )
+
+
+@bp.get("/api/seo/bot-status-codes")
+def api_bot_status_codes():
+    from_date, to_date = parse_date_range(request)
+    return jsonify(data=get_bot_status_codes(from_date, to_date), meta={"from": from_date.isoformat(), "to": to_date.isoformat()})
+
+
+@bp.get("/api/seo/crawled-vs-visited")
+def api_crawled_vs_visited():
+    from_date, to_date = parse_date_range(request)
+    page, per_page = parse_pagination(request)
+    rows, total = get_crawled_vs_visited(from_date, to_date, page, per_page)
+    return jsonify(
+        data={"rows": rows, "total": total},
+        meta={"from": from_date.isoformat(), "to": to_date.isoformat(), "page": page, "per_page": per_page},
+    )

app/blueprints/seo/templates/seo/_content.html

@@ -0,0 +1,102 @@
+<div id="seo-content"
+     hx-get="{{ url_for('seo.seo') }}"
+     hx-trigger="change from:#seo-date-range-form"
+     hx-include="#seo-date-range-form"
+     hx-target="#seo-content"
+     hx-swap="outerHTML"
+     data-from="{{ from_date.isoformat() }}"
+     data-to="{{ to_date.isoformat() }}">
+
+  <div class="flex flex-wrap items-end justify-between gap-4 mb-6">
+    <div>
+      <h1 class="font-display font-bold text-xl">SEO &amp; Bot Behavior</h1>
+      <p class="text-sm text-muted dark:text-muted-dark">How search engines are crawling your site</p>
+    </div>
+    <form id="seo-date-range-form" class="flex gap-3 items-end">
+      <label class="text-sm text-muted dark:text-muted-dark">From
+        <input type="date" name="from" value="{{ from_date.isoformat() }}"
+               class="block border border-line dark:border-line-dark rounded-md px-2 py-1 mt-1 bg-surface dark:bg-surface-dark text-ink dark:text-ink-dark font-data text-sm">
+      </label>
+      <label class="text-sm text-muted dark:text-muted-dark">To
+        <input type="date" name="to" value="{{ to_date.isoformat() }}"
+               class="block border border-line dark:border-line-dark rounded-md px-2 py-1 mt-1 bg-surface dark:bg-surface-dark text-ink dark:text-ink-dark font-data text-sm">
+      </label>
+    </form>
+  </div>
+
+  <div id="bot-summary-cards" class="grid grid-cols-2 sm:grid-cols-3 gap-3 mb-6"
+       data-endpoint="{{ url_for('seo.api_bot_summary') }}"></div>
+
+  <div class="grid grid-cols-1 lg:grid-cols-2 gap-4 mb-6">
+    <div class="border border-line dark:border-line-dark rounded-xl p-4 bg-surface dark:bg-surface-dark h-72">
+      <h3 class="text-xs font-medium uppercase tracking-wide text-muted dark:text-muted-dark mb-2">Crawl frequency</h3>
+      <div class="h-56"><canvas id="crawl-frequency-chart" data-endpoint="{{ url_for('seo.api_crawl_chart_data') }}"></canvas></div>
+    </div>
+    <div class="border border-line dark:border-line-dark rounded-xl p-4 bg-surface dark:bg-surface-dark h-72">
+      <h3 class="text-xs font-medium uppercase tracking-wide text-muted dark:text-muted-dark mb-2">Status codes served to bots</h3>
+      <div class="h-44"><canvas id="bot-status-codes-chart" data-endpoint="{{ url_for('seo.api_bot_status_codes') }}"></canvas></div>
+      <p id="bot-404-note" class="text-xs text-muted dark:text-muted-dark mt-2"></p>
+    </div>
+  </div>
+
+  <div>
+    <h3 class="text-xs font-medium uppercase tracking-wide text-muted dark:text-muted-dark mb-2">Most-crawled vs. most-visited URLs</h3>
+    <div id="crawled-vs-visited-grid" data-endpoint="{{ url_for('seo.api_crawled_vs_visited') }}"></div>
+  </div>
+
+  <script>
+    (function initSeoWidgets() {
+      const root = document.getElementById('seo-content');
+      const from = root.dataset.from, to = root.dataset.to;
+      const withRange = (url) => `${url}?from=${from}&to=${to}`;
+
+      fetch(withRange(document.getElementById('bot-summary-cards').dataset.endpoint))
+        .then((r) => r.json())
+        .then(({ data }) => {
+          document.getElementById('bot-summary-cards').innerHTML = data.length ? data.map((bot) => `
+            <div class="bg-surface dark:bg-surface-dark rounded-lg border border-line dark:border-line-dark border-t-2 ${bot.verified_pct < 100 ? 'border-t-warn dark:border-t-warn-dark' : 'border-t-ok dark:border-t-ok-dark'} p-3">
+              <p class="text-xs text-muted dark:text-muted-dark uppercase tracking-wide">${bot.bot_name}</p>
+              <p class="font-data text-xl font-medium mt-0.5">${bot.hits.toLocaleString()} <span class="text-sm text-muted dark:text-muted-dark font-sans">hits</span></p>
+              <p class="text-xs mt-1 ${bot.verified_pct < 100 ? 'text-warn dark:text-warn-dark' : 'text-muted dark:text-muted-dark'}">
+                ${bot.verified_pct}% verified${bot.verified_pct < 100 ? ' — some spoofed' : ''}
+              </p>
+              <p class="text-xs text-muted dark:text-muted-dark font-data mt-0.5">Last seen: ${bot.last_seen ?? '—'}</p>
+            </div>`).join('') : `<p class="text-sm text-muted dark:text-muted-dark col-span-full">No bot activity in this range.</p>`;
+        });
+
+      const crawlEl = document.getElementById('crawl-frequency-chart');
+      fetch(withRange(crawlEl.dataset.endpoint))
+        .then((r) => r.json())
+        .then(({ data }) => window.initChart('crawl-frequency-chart', {
+          type: 'line',
+          data: {
+            labels: [...new Set(data.series.flatMap((s) => s.points.map((p) => p.t)))].sort(),
+            datasets: data.series.map((s, i) => ({
+              label: s.bot_name,
+              data: s.points.map((p) => ({ x: p.t, y: p.count })),
+              tension: 0.3,
+              borderColor: window.KAVOSH_CHART_PALETTE[i % window.KAVOSH_CHART_PALETTE.length],
+            })),
+          },
+        }));
+
+      const statusEl = document.getElementById('bot-status-codes-chart');
+      fetch(withRange(statusEl.dataset.endpoint))
+        .then((r) => r.json())
+        .then(({ data }) => {
+          window.initChart('bot-status-codes-chart', {
+            type: 'bar',
+            data: { labels: Object.keys(data.breakdown), datasets: [{ label: 'Bot Requests', data: Object.values(data.breakdown), backgroundColor: '#0E7C86' }] },
+          });
+          document.getElementById('bot-404-note').textContent = `${data.not_found_404} 404s served to bots in range — wasted crawl budget.`;
+        });
+
+      window.initGrid(
+        'crawled-vs-visited-grid',
+        document.getElementById('crawled-vs-visited-grid').dataset.endpoint,
+        [{ name: 'Path' }, { name: 'Bot Hits' }, { name: 'Human Hits' }, { name: 'Bot Share %' }],
+        { from, to },
+      );
+    })();
+  </script>
+</div>

app/blueprints/seo/templates/seo/index.html

@@ -0,0 +1,5 @@
+{% extends "base.html" %}
+{% block title %}SEO & Bots — Kavosh{% endblock %}
+{% block content %}
+  {% include "seo/_content.html" %}
+{% endblock %}

app/blueprints/uploads/__init__.py

@@ -0,0 +1,15 @@
+from flask import Blueprint
+from flask_login import login_required
+
+bp = Blueprint("uploads", __name__, template_folder="templates")
+
+
+@bp.before_request
+@login_required
+def require_login():
+    """Ch01: dashboard reachable from one AUTHENTICATED shell; Ch12:
+    single-admin login. All routes on this blueprint require a session."""
+    pass
+
+
+from app.blueprints.uploads import routes  # noqa: E402,F401  registers routes

app/blueprints/uploads/queries.py

@@ -0,0 +1,51 @@
+"""Query/formatting helpers for the "Uploaded files" list (project-owner
+follow-up request). Kept separate from routes.py to match this project's
+established per-blueprint queries.py convention (Ch08/09/10).
+"""
+from __future__ import annotations
+
+from app.models.log_file import LogFile
+
+
+def get_uploaded_files(page: int, per_page: int) -> tuple[list[list], int]:
+    """Newest first, independent of the dashboard date-range picker —
+    this lists uploads by when they arrived, not by which log dates they
+    contain (a single file can span many dates).
+    """
+    query = LogFile.query.order_by(LogFile.uploaded_at.desc())
+    total = query.count()
+    rows = query.offset((page - 1) * per_page).limit(per_page).all()
+
+    result = []
+    for lf in rows:
+        result.append([
+            lf.id,
+            lf.filename,
+            lf.server_type,
+            _status_display(lf),
+            lf.uploaded_at.strftime("%Y-%m-%d %H:%M"),
+            _human_size(lf.size_bytes),
+            lf.status,  # raw status (hidden column) — lets the client disable
+                        # the select checkbox for files still "processing"
+        ])
+    return result, total
+
+
+def _status_display(lf: LogFile) -> str:
+    if lf.status == "processing":
+        if lf.total_lines:
+            pct = round(lf.processed_lines / lf.total_lines * 100)
+            return f"processing ({pct}%)"
+        return "processing"
+    if lf.status == "error":
+        return f"error: {lf.error_message}" if lf.error_message else "error"
+    return lf.status
+
+
+def _human_size(num_bytes: int) -> str:
+    size = float(num_bytes)
+    for unit in ("B", "KB", "MB", "GB"):
+        if size < 1024 or unit == "GB":
+            return f"{size:.0f} {unit}" if unit == "B" else f"{size:.1f} {unit}"
+        size /= 1024
+    return f"{size:.1f} GB"

app/blueprints/uploads/routes.py

@@ -0,0 +1,188 @@
+"""Upload endpoint (Chapter 04): validated, streamed-to-disk save.
+
+Parsing happens in a background thread triggered right after this request
+completes (app/services/background.py) — never synchronously inside this
+request (Chapter 03, rule 5 still holds: the response returns immediately
+regardless of file size).
+"""
+from __future__ import annotations
+
+import hashlib
+import uuid
+from pathlib import Path
+
+from flask import current_app, jsonify, render_template, request
+from werkzeug.utils import secure_filename
+
+from app.blueprints.uploads import bp
+from app.blueprints.uploads.queries import get_uploaded_files
+from app.extensions import db, limiter
+from app.models.log_file import LogFile
+from app.services.background import trigger_processing
+from app.services.file_deletion import delete_log_files
+from app.utils.pagination import parse_pagination
+from app.utils.upload_paths import upload_path_for
+
+_ALLOWED_EXTENSIONS = {".log", ".txt", ".gz"}
+_CHUNK_SIZE = 64 * 1024  # 64 KB per read — never buffer the whole upload
+_GZIP_MAGIC = b"\x1f\x8b"
+
+
+class UploadRejected(Exception):
+    """Raised when an upload fails extension/content validation."""
+
+
+def _validate_extension(filename: str) -> str:
+    ext = Path(filename).suffix.lower()
+    if ext not in _ALLOWED_EXTENSIONS:
+        raise UploadRejected(f"Unsupported extension {ext!r}; allowed: {_ALLOWED_EXTENSIONS}")
+    return ext
+
+
+def _sniff_content(first_chunk: bytes, ext: str) -> None:
+    """Light content sniff — don't just trust the client-supplied MIME type."""
+    if ext == ".gz":
+        if not first_chunk.startswith(_GZIP_MAGIC):
+            raise UploadRejected("File has a .gz extension but isn't gzip-magic-prefixed.")
+        return
+    if b"\x00" in first_chunk:
+        raise UploadRejected("File extension claims text but content looks binary.")
+
+
+def _stream_to_temp(file_storage, tmp_path: Path) -> tuple[int, str, bytes]:
+    """Stream the upload to disk in bounded chunks; return (size, sha256_hex, first_chunk).
+
+    Never calls file.read() on the whole stream (Chapter 03, rule 1).
+    """
+    sha256 = hashlib.sha256()
+    size = 0
+    first_chunk: bytes | None = None
+    with tmp_path.open("wb") as out:
+        while True:
+            chunk = file_storage.stream.read(_CHUNK_SIZE)
+            if not chunk:
+                break
+            if first_chunk is None:
+                first_chunk = chunk
+            sha256.update(chunk)
+            size += len(chunk)
+            out.write(chunk)
+    if first_chunk is None:
+        raise UploadRejected("Uploaded file is empty.")
+    return size, sha256.hexdigest(), first_chunk
+
+
+@bp.post("/uploads")
+@limiter.limit("20 per minute")  # Chapter 12: rate limiting on /uploads at minimum
+def upload_log_file():
+    """Validate, stream, and register an uploaded access log (Ch04/Ch11)."""
+    file_storage = request.files.get("logfile")
+    if file_storage is None or not file_storage.filename:
+        return render_template("uploads/_error.html", message="No file provided."), 400
+
+    upload_dir = Path(current_app.config["UPLOAD_DIR"])
+    (upload_dir / "tmp").mkdir(parents=True, exist_ok=True)
+    tmp_path = upload_dir / "tmp" / f"{uuid.uuid4().hex}.part"
+
+    try:
+        ext = _validate_extension(file_storage.filename)
+        size_bytes, checksum, first_chunk = _stream_to_temp(file_storage, tmp_path)
+        _sniff_content(first_chunk, ext)
+
+        max_bytes = current_app.config["UPLOAD_MAX_SIZE_MB"] * 1024 * 1024
+        if size_bytes > max_bytes:
+            raise UploadRejected(f"File exceeds {current_app.config['UPLOAD_MAX_SIZE_MB']}MB limit.")
+    except UploadRejected as exc:
+        tmp_path.unlink(missing_ok=True)
+        return render_template("uploads/_error.html", message=str(exc)), 400
+
+    existing = LogFile.query.filter_by(checksum=checksum).first()
+    if existing is not None:
+        tmp_path.unlink(missing_ok=True)
+        return render_template("uploads/_duplicate.html", log_file=existing)
+
+    log_file = LogFile(
+        filename=secure_filename(file_storage.filename),
+        server_type=request.form.get("server_type", "apache"),
+        format_string=request.form.get("format_string", ""),
+        status="queued",
+        size_bytes=size_bytes,
+        checksum=checksum,
+    )
+    db.session.add(log_file)
+    db.session.commit()  # need the assigned id before the final rename
+
+    tmp_path.rename(upload_path_for(log_file))
+
+    # Chapter 12 simplification: no cron required — kick off processing
+    # immediately in a background thread. The response below returns as
+    # soon as the file is queued (Ch03 rule 5 still holds: this request
+    # never blocks on parsing), while the thread runs independently.
+    trigger_processing(current_app._get_current_object(), log_file.id)
+
+    return render_template("uploads/_queued.html", log_file=log_file)
+
+
+@bp.get("/api/uploads/<int:log_file_id>/status")
+def upload_status(log_file_id: int):
+    """Polled every 3s by the browser (hx-trigger) until done/error (Ch04)."""
+    log_file = db.get_or_404(LogFile, log_file_id)
+
+    if request.headers.get("Accept") == "application/json":
+        return jsonify(
+            data={
+                "id": log_file.id,
+                "status": log_file.status,
+                "processed_lines": log_file.processed_lines,
+                "total_lines": log_file.total_lines,
+            },
+            meta={},
+        )
+
+    template = {
+        "done": "uploads/_status_done.html",
+        "error": "uploads/_status_error.html",
+        # BUG FIX: this key was missing, so "queued" fell through to the
+        # "processing" fallback below — a file that hadn't been picked up
+        # by `flask process-logs` yet displayed as "Processing X: 0 lines"
+        # instead of "Queued — waiting for the next parse cycle", making a
+        # cron job that simply hasn't run yet indistinguishable from one
+        # that's actually hung mid-parse.
+        "queued": "uploads/_queued.html",
+    }.get(log_file.status, "uploads/_status_processing.html")
+    return render_template(template, log_file=log_file)
+
+
+@bp.get("/api/uploads")
+def list_uploads():
+    """Uploaded-files list (project-owner follow-up request) — Grid.js-
+    backed, same page/per_page convention as every other table (Ch11).
+    Sorted by upload recency, independent of the dashboard date-range
+    picker (a single file can span many log dates).
+    """
+    page, per_page = parse_pagination(request)
+    rows, total = get_uploaded_files(page, per_page)
+    return jsonify(
+        data={"rows": rows, "total": total},
+        meta={"page": page, "per_page": per_page},
+    )
+
+
+@bp.delete("/api/uploads")
+def bulk_delete_uploads():
+    """Delete one or more uploaded files and everything derived from them
+    (log_entries/bot_hits/suspicious_events, the raw file on disk, and a
+    correct rollup recompute for the affected dates — see
+    app/services/file_deletion.py for why a rollup recompute is needed
+    rather than a simple per-file delete).
+
+    Body: {"ids": [1, 2, 3]}. Files currently "processing" are skipped,
+    not force-deleted, to avoid racing the background parse thread.
+    """
+    body = request.get_json(silent=True) or {}
+    ids = body.get("ids")
+    if not isinstance(ids, list) or not ids or not all(isinstance(i, int) for i in ids):
+        return jsonify(error={"code": "invalid_request", "message": "Expected {\"ids\": [int, ...]}."}), 400
+
+    result = delete_log_files(ids)
+    return jsonify(data={"deleted": result.deleted, "skipped": result.skipped}, meta={})

app/blueprints/uploads/templates/uploads/_duplicate.html

@@ -0,0 +1,4 @@
+<div class="border border-warn/30 dark:border-warn-dark/30 bg-warn/5 dark:bg-warn-dark/10 rounded-lg p-3 text-sm">
+  <span class="font-data">{{ log_file.filename }}</span>
+  <span class="text-muted dark:text-muted-dark">matches an already-uploaded file (status: {{ log_file.status }}); skipped re-upload.</span>
+</div>

app/blueprints/uploads/templates/uploads/_error.html

@@ -0,0 +1,3 @@
+<div class="border border-danger/30 dark:border-danger-dark/30 bg-danger/5 dark:bg-danger-dark/10 rounded-lg p-3 text-sm text-danger dark:text-danger-dark">
+  {{ message }}
+</div>

app/blueprints/uploads/templates/uploads/_queued.html

@@ -0,0 +1,10 @@
+<div id="upload-status-{{ log_file.id }}"
+     hx-get="{{ url_for('uploads.upload_status', log_file_id=log_file.id) }}"
+     hx-trigger="every 1s" hx-swap="outerHTML"
+     class="border border-line dark:border-line-dark rounded-lg p-3 bg-paper dark:bg-paper-dark">
+  <div class="flex items-center gap-2 text-sm">
+    <svg class="w-4 h-4 text-muted dark:text-muted-dark animate-spin"><use href="/static/dist/icons.svg#loader-circle"/></svg>
+    <span class="font-data">{{ log_file.filename }}</span>
+    <span class="text-muted dark:text-muted-dark">— queued, starting shortly…</span>
+  </div>
+</div>

app/blueprints/uploads/templates/uploads/_status_done.html

@@ -0,0 +1,7 @@
+<div id="upload-status-{{ log_file.id }}" class="border border-ok/30 dark:border-ok-dark/30 bg-ok/5 dark:bg-ok-dark/10 rounded-lg p-3">
+  <div class="flex items-center gap-2 text-sm">
+    <span class="w-2 h-2 rounded-full bg-ok dark:bg-ok-dark shrink-0"></span>
+    <span class="font-data">{{ log_file.filename }}</span>
+    <span class="text-ok dark:text-ok-dark">— done, {{ "{:,}".format(log_file.processed_lines) }} lines analyzed</span>
+  </div>
+</div>

app/blueprints/uploads/templates/uploads/_status_error.html

@@ -0,0 +1,7 @@
+<div id="upload-status-{{ log_file.id }}" class="border border-danger/30 dark:border-danger-dark/30 bg-danger/5 dark:bg-danger-dark/10 rounded-lg p-3">
+  <div class="flex items-center gap-2 text-sm">
+    <span class="w-2 h-2 rounded-full bg-danger dark:bg-danger-dark shrink-0"></span>
+    <span class="font-data">{{ log_file.filename }}</span>
+  </div>
+  <p class="text-danger dark:text-danger-dark text-xs mt-1">{{ log_file.error_message }}</p>
+</div>

app/blueprints/uploads/templates/uploads/_status_processing.html

@@ -0,0 +1,19 @@
+{% set pct = ((log_file.processed_lines / log_file.total_lines) * 100) if log_file.total_lines else None %}
+<div id="upload-status-{{ log_file.id }}"
+     hx-get="{{ url_for('uploads.upload_status', log_file_id=log_file.id) }}"
+     hx-trigger="every 1s" hx-swap="outerHTML"
+     class="border border-line dark:border-line-dark rounded-lg p-3 bg-paper dark:bg-paper-dark">
+  <div class="flex items-center justify-between text-sm mb-2">
+    <span class="font-data">{{ log_file.filename }}</span>
+    <span class="font-data text-muted dark:text-muted-dark">
+      {% if pct is not none %}{{ pct | round(0) | int }}%{% else %}analyzing…{% endif %}
+    </span>
+  </div>
+  <div class="w-full h-2 rounded-full bg-line dark:bg-line-dark overflow-hidden">
+    <div class="h-full rounded-full bg-accent dark:bg-accent-dark transition-all duration-500 ease-out"
+         style="width: {{ pct | round(1) if pct is not none else 8 }}%"></div>
+  </div>
+  <p class="text-xs text-muted dark:text-muted-dark mt-1.5 font-data">
+    {{ "{:,}".format(log_file.processed_lines) }}{% if log_file.total_lines %} / {{ "{:,}".format(log_file.total_lines) }}{% endif %} lines
+  </p>
+</div>

app/budget.py

@@ -0,0 +1,82 @@
+"""Concrete, checkable translation of the cPanel account ceiling (Chapter 03).
+
+Single source of truth for the account's literal resource limits. Other
+chapters should import BUDGET rather than re-hardcoding these numbers.
+validate_budget_config() is called from create_app() so an out-of-budget
+deployment fails loudly at startup instead of silently degrading under load
+— this is the concrete mechanism behind Chapter 03's "flag, don't silently
+accept" requirement.
+"""
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+from flask import Flask
+
+
+@dataclass(frozen=True)
+class AccountBudget:
+    """The hosting account's literal ceiling (Chapter 03).
+
+    iops/io_throughput are informational only here — Python config can't
+    enforce them directly; they constrain how Ch06/07 batch reads/writes.
+    """
+
+    cpu_cores: int = 4
+    max_entry_processes: int = 60
+    memory_mb: int = 2048
+    iops: int = 1024
+    io_throughput_mb_s: int = 16
+    max_total_processes: int = 150
+    max_db_connections: int = 150
+
+    # Derived engineering targets from Chapter 03's budget table.
+    db_pool_size_min: int = 2
+    db_pool_size_max: int = 5
+
+
+BUDGET = AccountBudget()
+
+_DISALLOWED_CACHE_BACKENDS = {"redis", "rediscache", "memcached", "memcachedcache"}
+
+
+def validate_budget_config(app: Flask) -> None:
+    """Raise at startup if config violates a Chapter 03 rule.
+
+    Cheap checks only (string/int comparisons) since this runs on every
+    process boot — Passenger may recycle processes frequently (factor 9).
+    """
+    # Validated as its own config key, not read out of
+    # SQLALCHEMY_ENGINE_OPTIONS — that dict is empty for SQLite (its pool
+    # classes reject pool_size/pool_recycle outright; see app/config.py's
+    # _engine_options_for), so the *intended* setting must be checked
+    # independently of whether the active engine actually consumes it.
+    pool_size = app.config.get("DB_POOL_SIZE")
+    if pool_size is None or not (BUDGET.db_pool_size_min <= pool_size <= BUDGET.db_pool_size_max):
+        raise RuntimeError(
+            f"DB_POOL_SIZE={pool_size} is outside the Chapter 03 budget "
+            f"({BUDGET.db_pool_size_min}-{BUDGET.db_pool_size_max} per process; "
+            f"{BUDGET.max_db_connections} total connections are shared across "
+            f"up to {BUDGET.max_entry_processes} entry processes)."
+        )
+
+    cache_type = str(app.config.get("CACHE_TYPE", "")).lower()
+    if any(name in cache_type for name in _DISALLOWED_CACHE_BACKENDS):
+        raise RuntimeError(
+            f"CACHE_TYPE={app.config.get('CACHE_TYPE')!r} assumes a backend "
+            "(Redis/Memcached) Chapter 03 says not to assume is available. "
+            "Use FileSystemCache or a dashboard_cache DB table (Ch06)."
+        )
+
+    batch_size = app.config.get("PARSE_BATCH_SIZE")
+    if not batch_size or batch_size <= 0:
+        raise RuntimeError(
+            "PARSE_BATCH_SIZE must be a positive integer — unbounded/whole-file "
+            "parsing per invocation violates the Chapter 03 memory budget."
+        )
+
+    max_upload_mb = app.config.get("UPLOAD_MAX_SIZE_MB")
+    if not max_upload_mb or max_upload_mb <= 0:
+        raise RuntimeError(
+            "UPLOAD_MAX_SIZE_MB must be a positive integer to bound disk/IOPS per upload."
+        )

app/cli.py

@@ -0,0 +1,163 @@
+"""Flask CLI admin commands (factor 12).
+
+process-logs is now OPTIONAL (Chapter 12 simplification, per project
+owner request): file processing is triggered automatically in-app right
+after upload (app/services/background.py), so cron is no longer required.
+This command still exists for anyone who'd rather run it manually or via
+cron — it shares the exact same processing code (app/services/
+log_processor.py) as the automatic background trigger, so both paths
+behave identically. cleanup (Chapter 06/12) enforces the retention
+policy. create-admin (Chapter 12) bootstraps the single admin account.
+"""
+from __future__ import annotations
+
+from datetime import date, datetime, timedelta
+
+import click
+from flask import Flask, current_app
+
+from app.extensions import db
+from app.models.ip_traffic_stats import IpPathStatsDaily, IpStatusStatsDaily
+from app.models.log_entry import LogEntry
+from app.models.log_file import LogFile
+from app.models.request_stats import RequestStatsHourly
+from app.models.user import User
+from app.services import aggregator
+from app.services.log_processor import process_one_batch
+from app.utils.upload_paths import upload_path_for
+
+# Chapter 06 retention policy — the constants `flask cleanup` enforces.
+LOG_ENTRIES_RETENTION_DAYS = 30
+HOURLY_STATS_RETENTION_DAYS = 90
+IP_TRAFFIC_STATS_RETENTION_DAYS = 30
+
+
+def register_commands(app: Flask) -> None:
+    app.cli.add_command(process_logs)
+    app.cli.add_command(rollup)
+    app.cli.add_command(cleanup)
+    app.cli.add_command(create_admin)
+
+
+@click.command("process-logs")
+@click.option("--batch-size", default=None, type=int, help="Override PARSE_BATCH_SIZE.")
+def process_logs(batch_size: int | None) -> None:
+    """Optional manual/cron fallback — processing now also runs
+    automatically in-app after upload. Parses queued/processing
+    log_files in bounded, checkpointed batches; one batch per file per
+    invocation, same as before.
+    """
+    batch = batch_size or current_app.config["PARSE_BATCH_SIZE"]
+    pending = LogFile.query.filter(LogFile.status.in_(["queued", "processing"])).all()
+    for log_file in pending:
+        process_one_batch(log_file, batch)
+
+
+@click.command("rollup")
+@click.option("--from", "from_date", required=True, help="ISO date, e.g. 2026-07-01")
+@click.option("--to", "to_date", required=True, help="ISO date, e.g. 2026-07-26")
+def rollup(from_date: str, to_date: str) -> None:
+    """Manual rollup recompute for an explicit range (e.g. after a backfill).
+
+    Requires an explicit range — no "all time" default, mirroring Ch03 rule 7
+    even for an admin command, to avoid an unbounded scan on constrained hosting.
+    """
+    start = date.fromisoformat(from_date)
+    end = date.fromisoformat(to_date)
+    aggregator.compute_rollups_for_range(start, end)
+    click.echo(f"Rolled up {start} .. {end}")
+
+
+@click.command("cleanup")
+def cleanup() -> None:
+    """Enforce the Chapter 06 retention policy (Chapter 12).
+
+    Previously a stub through every earlier chapter — implemented here as
+    Chapter 12's non-functional/deployment concern. Cron-invoked (e.g.
+    daily, off-peak), never a long-running daemon.
+    """
+    now = datetime.utcnow()
+    deleted_entries = _delete_old_log_entries(now)
+    deleted_hourly = _collapse_old_hourly_stats(now)
+    deleted_ip_stats = _delete_old_ip_traffic_stats(now)
+    archived_files = _delete_parsed_upload_files()
+    click.echo(
+        f"Cleanup complete: {deleted_entries} log_entries, {deleted_hourly} "
+        f"request_stats_hourly, {deleted_ip_stats} ip traffic-rollup rows "
+        f"deleted; {archived_files} parsed upload file(s) removed from disk."
+    )
+
+
+def _delete_old_log_entries(now: datetime) -> int:
+    """Ch06: raw log_entries retained ~30 days, then deleted."""
+    cutoff = now - timedelta(days=LOG_ENTRIES_RETENTION_DAYS)
+    count = db.session.query(LogEntry).filter(LogEntry.timestamp < cutoff).delete(synchronize_session=False)
+    db.session.commit()
+    return count
+
+
+def _collapse_old_hourly_stats(now: datetime) -> int:
+    """Ch06: request_stats_hourly retained ~90 days, then collapsed into
+    request_stats_daily only. request_stats_daily is already computed
+    independently by the aggregator straight from raw log_entries, so
+    "collapsing" here just means deleting the now-redundant hourly rows
+    once the retention window passes — no data is lost, since the daily
+    rollup for that period was already written when the file was parsed.
+    """
+    cutoff = now - timedelta(days=HOURLY_STATS_RETENTION_DAYS)
+    count = db.session.query(RequestStatsHourly).filter(
+        RequestStatsHourly.date_hour < cutoff
+    ).delete(synchronize_session=False)
+    db.session.commit()
+    return count
+
+
+def _delete_old_ip_traffic_stats(now: datetime) -> int:
+    """The bounded per-IP rollup (added as a Ch10 follow-up) was designed
+    for ~30-day retention, matching log_entries — see aggregator.py.
+    """
+    cutoff_date = (now - timedelta(days=IP_TRAFFIC_STATS_RETENTION_DAYS)).date()
+    count = db.session.query(IpPathStatsDaily).filter(IpPathStatsDaily.date < cutoff_date).delete(synchronize_session=False)
+    count += db.session.query(IpStatusStatsDaily).filter(IpStatusStatsDaily.date < cutoff_date).delete(synchronize_session=False)
+    db.session.commit()
+    return count
+
+
+def _delete_parsed_upload_files() -> int:
+    """Ch06: 'compress or delete after successful parse + rollup, rather
+    than keeping both the raw file and a full raw-row copy.' Deletes
+    (rather than compresses) — simpler, and avoids spending extra CPU/IOPS
+    gzip-ing data that's already been fully parsed into the database.
+    """
+    done_files = LogFile.query.filter_by(status="done").all()
+    removed = 0
+    for log_file in done_files:
+        path = upload_path_for(log_file)
+        if path.exists():
+            path.unlink()
+            removed += 1
+    return removed
+
+
+@click.command("create-admin")
+@click.option("--email", default=None, help="Defaults to the ADMIN_EMAIL env var.")
+@click.password_option()
+def create_admin(email: str | None, password: str) -> None:
+    """Bootstrap the single admin account (Chapter 12).
+
+    Interactive password prompt (via --password-option's confirmation
+    prompt) keeps the raw credential out of process env/config, unlike an
+    ADMIN_PASSWORD env var would — Chapter 12 doesn't specify a bootstrap
+    mechanism beyond documenting ADMIN_EMAIL, so this is a flagged addition.
+    """
+    resolved_email = (email or current_app.config.get("ADMIN_EMAIL") or "").strip().lower()
+    if not resolved_email:
+        raise click.UsageError("No --email given and ADMIN_EMAIL is not set.")
+    if User.query.filter_by(email=resolved_email).first():
+        raise click.UsageError(f"User {resolved_email} already exists.")
+
+    user = User(email=resolved_email)
+    user.set_password(password)
+    db.session.add(user)
+    db.session.commit()
+    click.echo(f"Created admin user {resolved_email}")

app/config.py

@@ -0,0 +1,109 @@
+"""Environment-sourced configuration classes (12-factor factor 3).
+
+Every value comes from os.environ. No secret, DB URL, or filesystem path
+is ever hardcoded; .env.example documents every variable a deployment
+must set.
+"""
+from __future__ import annotations
+
+import os
+from datetime import timedelta
+
+
+def _bool_env(name: str, default: bool = False) -> bool:
+    """Parse a boolean-ish environment variable."""
+    val = os.environ.get(name)
+    return default if val is None else val.strip().lower() in {"1", "true", "yes", "on"}
+
+
+def _engine_options_for(database_url: str) -> dict:
+    """pool_size/pool_recycle are QueuePool-only kwargs.
+
+    BUG FIX (caught by actually booting the app): SQLite's default pool
+    classes (NullPool for file DBs, StaticPool for :memory:) raise
+    TypeError if handed pool_size/pool_recycle at all — they're not
+    silently ignored. Chapter 06 makes SQLite the default engine and
+    MySQL the opt-in fallback, so these kwargs are only meaningful (and
+    only passed) when DATABASE_URL actually points at a non-SQLite engine.
+    """
+    if database_url.startswith("sqlite"):
+        return {}
+    return {
+        "pool_size": int(os.environ.get("DB_POOL_SIZE", "3")),
+        "pool_recycle": int(os.environ.get("DB_POOL_RECYCLE_SECONDS", "280")),
+        "pool_pre_ping": True,
+    }
+
+
+class BaseConfig:
+    """Shared config. Subclasses override only what differs per environment."""
+
+    SECRET_KEY: str | None = os.environ.get("SECRET_KEY")
+
+    # Chapter 06: SQLite/WAL default; swappable via DATABASE_URL without code changes.
+    SQLALCHEMY_DATABASE_URI: str = os.environ.get("DATABASE_URL", "sqlite:///kavosh.db")
+    SQLALCHEMY_ENGINE_OPTIONS: dict = _engine_options_for(SQLALCHEMY_DATABASE_URI)
+    SQLALCHEMY_TRACK_MODIFICATIONS = False
+
+    # Chapter 03: 150 max DB connections shared across up to 60 entry
+    # processes -> keep each process's pool small. Exposed as its own
+    # config key (not just buried inside SQLALCHEMY_ENGINE_OPTIONS) so
+    # budget.py can validate the *intended* setting regardless of whether
+    # the active engine (SQLite) actually consumes it.
+    DB_POOL_SIZE: int = int(os.environ.get("DB_POOL_SIZE", "3"))
+
+    # Chapter 03: filesystem cache, not Redis.
+    CACHE_TYPE = os.environ.get("CACHE_TYPE", "FileSystemCache")
+    CACHE_DIR = os.environ.get("CACHE_DIR", "/tmp/kavosh-cache")
+    CACHE_DEFAULT_TIMEOUT = int(os.environ.get("CACHE_DEFAULT_TIMEOUT", "60"))
+
+    # Chapter 12: secure session cookie flags.
+    SESSION_COOKIE_SECURE = _bool_env("SESSION_COOKIE_SECURE", True)
+    SESSION_COOKIE_HTTPONLY = True
+    SESSION_COOKIE_SAMESITE = "Lax"
+    PERMANENT_SESSION_LIFETIME = timedelta(
+        hours=int(os.environ.get("SESSION_LIFETIME_HOURS", "12"))
+    )
+
+    WTF_CSRF_ENABLED = True
+
+    # Chapter 04/12: upload constraints.
+    UPLOAD_MAX_SIZE_MB = int(os.environ.get("UPLOAD_MAX_SIZE_MB", "500"))
+    MAX_CONTENT_LENGTH = UPLOAD_MAX_SIZE_MB * 1024 * 1024
+    UPLOAD_DIR = os.environ.get("UPLOAD_DIR", "/home/kavosh/uploads")
+
+    PARSE_BATCH_SIZE = int(os.environ.get("PARSE_BATCH_SIZE", "5000"))
+    ADMIN_EMAIL = os.environ.get("ADMIN_EMAIL")
+
+
+class DevConfig(BaseConfig):
+    DEBUG = True
+    SESSION_COOKIE_SECURE = False  # allow plain-http local dev
+
+
+class ProdConfig(BaseConfig):
+    DEBUG = False
+
+
+class TestConfig(BaseConfig):
+    TESTING = True
+    DEBUG = True  # lets asset() tolerate a missing Vite manifest during tests
+    SQLALCHEMY_DATABASE_URI = "sqlite:///:memory:"
+    SQLALCHEMY_ENGINE_OPTIONS = {}  # always in-memory SQLite regardless of DATABASE_URL
+    WTF_CSRF_ENABLED = False
+
+
+_CONFIGS = {"development": DevConfig, "production": ProdConfig, "testing": TestConfig}
+
+
+def get_config(config_name: str | None = None):
+    """Resolve a config class from FLASK_ENV or an explicit name.
+
+    Defaults to `production` if unset, so an unconfigured deployment never
+    silently runs with DEBUG on.
+    """
+    name = config_name or os.environ.get("FLASK_ENV", "production")
+    try:
+        return _CONFIGS[name]
+    except KeyError as exc:
+        raise ValueError(f"Unknown config_name {name!r}; expected one of {list(_CONFIGS)}") from exc

app/extensions.py

@@ -0,0 +1,28 @@
+"""Singleton Flask extension instances — initialized, not configured, here.
+
+Configuration happens in create_app() via .init_app(), so nothing here
+holds app- or request-scoped state that must survive a process restart
+(factor 6: stateless processes).
+"""
+from flask_caching import Cache
+from flask_limiter import Limiter
+from flask_limiter.util import get_remote_address
+from flask_login import LoginManager
+from flask_migrate import Migrate
+from flask_sqlalchemy import SQLAlchemy
+from flask_wtf import CSRFProtect
+
+db = SQLAlchemy()
+cache = Cache()
+csrf = CSRFProtect()
+login_manager = LoginManager()
+login_manager.login_view = "auth.login"
+migrate = Migrate()
+
+# In-memory storage (Chapter 12: "in-memory or DB-backed... do not require
+# Redis"). CAVEAT (flagged): each of up to 60 entry processes (Ch03) keeps
+# its own counters, so the effective ceiling across the whole app is up to
+# (per-process limit x concurrent processes hit), not one hard global cap.
+# Acceptable for this app's threat model (slowing down /login and /uploads
+# brute-forcing), but not a strict global rate guarantee.
+limiter = Limiter(key_func=get_remote_address, storage_uri="memory://")

app/logging_setup.py

@@ -0,0 +1,34 @@
+"""Structured stdout/stderr logging (Chapter 02 factor 11 / Chapter 12).
+
+Writes structured (key=value) lines to stdout so the host's log capture
+picks them up, per 12-factor logs. Falls back to a size-capped rotating
+file only if LOG_FALLBACK_FILE is explicitly set (Ch12: "if stdout capture
+is unavailable on the specific hosting setup").
+"""
+from __future__ import annotations
+
+import logging
+import os
+import sys
+from logging.handlers import RotatingFileHandler
+
+from flask import Flask
+
+_FORMAT = "%(asctime)s level=%(levelname)s logger=%(name)s msg=%(message)s"
+
+
+def configure_logging(app: Flask) -> None:
+    formatter = logging.Formatter(_FORMAT)
+
+    stdout_handler = logging.StreamHandler(sys.stdout)
+    stdout_handler.setFormatter(formatter)
+
+    app.logger.handlers = [stdout_handler]
+    app.logger.setLevel(logging.INFO if not app.debug else logging.DEBUG)
+    app.logger.propagate = False
+
+    fallback_path = os.environ.get("LOG_FALLBACK_FILE")
+    if fallback_path:
+        file_handler = RotatingFileHandler(fallback_path, maxBytes=5 * 1024 * 1024, backupCount=3)
+        file_handler.setFormatter(formatter)
+        app.logger.addHandler(file_handler)

app/models/__init__.py

@@ -0,0 +1,19 @@
+from app.models.blocklist_suggestion import BlocklistSuggestion
+from app.models.bot_hit import BotHit
+from app.models.browser_stats import BrowserStatsDaily
+from app.models.human_path_stats import HumanPathStatsDaily
+from app.models.ip_registry import IPRegistry
+from app.models.ip_traffic_stats import IpPathStatsDaily, IpStatusStatsDaily
+from app.models.log_entry import LogEntry
+from app.models.log_file import LogFile
+from app.models.referrer_stats import ReferrerStatsDaily
+from app.models.request_stats import RequestStatsDaily, RequestStatsHourly
+from app.models.suspicious_event import SuspiciousEvent
+from app.models.user import User
+
+__all__ = [
+    "LogFile", "LogEntry", "RequestStatsHourly", "RequestStatsDaily",
+    "BotHit", "IPRegistry", "SuspiciousEvent", "BlocklistSuggestion",
+    "ReferrerStatsDaily", "BrowserStatsDaily", "HumanPathStatsDaily",
+    "IpPathStatsDaily", "IpStatusStatsDaily", "User",
+]

app/models/blocklist_suggestion.py

@@ -0,0 +1,18 @@
+"""blocklist_suggestions table (Chapter 06). `id` isn't in Ch06's column
+list — added because `ip` alone can't be the key (the same IP may be
+re-flagged with a different reason later)."""
+from __future__ import annotations
+
+from datetime import datetime
+
+from app.extensions import db
+
+
+class BlocklistSuggestion(db.Model):
+    __tablename__ = "blocklist_suggestions"
+
+    id: int = db.Column(db.Integer, primary_key=True)
+    ip: str = db.Column(db.String(45), nullable=False, index=True)
+    reason: str = db.Column(db.String(255), nullable=False)
+    created_at = db.Column(db.DateTime, nullable=False, default=datetime.utcnow)
+    exported: bool = db.Column(db.Boolean, nullable=False, default=False)

app/models/bot_hit.py

@@ -0,0 +1,22 @@
+"""bot_hits table (Chapter 06) — SEO section (Ch09)."""
+from __future__ import annotations
+
+from app.extensions import db
+
+
+class BotHit(db.Model):
+    __tablename__ = "bot_hits"
+
+    id: int = db.Column(db.Integer, primary_key=True)
+    log_file_id: int = db.Column(db.Integer, db.ForeignKey("log_files.id", ondelete="CASCADE"), nullable=False)
+    timestamp = db.Column(db.DateTime, nullable=False)
+    ip: str = db.Column(db.String(45), nullable=False)
+    bot_name: str = db.Column(db.String(64), nullable=False)
+    verified: bool = db.Column(db.Boolean, nullable=False, default=False)
+    path: str = db.Column(db.Text, nullable=False)
+    status_code: int = db.Column(db.SmallInteger, nullable=False)
+
+    __table_args__ = (
+        db.Index("ix_bot_hits_timestamp", "timestamp"),
+        db.Index("ix_bot_hits_bot_name", "bot_name"),
+    )

app/models/browser_stats.py

@@ -0,0 +1,14 @@
+"""NEW table (Method A, Ch08 follow-up) — not in Chapter 06's original
+list. Human-only browser/OS breakdown widget (Ch08)."""
+from __future__ import annotations
+
+from app.extensions import db
+
+
+class BrowserStatsDaily(db.Model):
+    __tablename__ = "browser_stats_daily"
+
+    date = db.Column(db.Date, primary_key=True)
+    browser: str = db.Column(db.String(64), primary_key=True)
+    os: str = db.Column(db.String(64), primary_key=True)
+    count: int = db.Column(db.Integer, nullable=False, default=0)

app/models/human_path_stats.py

@@ -0,0 +1,15 @@
+"""NEW table (Method A, Ch09 follow-up) — not in Chapter 06's original
+list. request_stats_hourly/_daily aggregate ALL traffic with no is_bot
+split, so Ch09's "most-visited-by-humans" comparison had no rollup to
+read. Bot hits excluded at rollup-write time (Ch07's is_bot flag)."""
+from __future__ import annotations
+
+from app.extensions import db
+
+
+class HumanPathStatsDaily(db.Model):
+    __tablename__ = "human_path_stats_daily"
+
+    date = db.Column(db.Date, primary_key=True)
+    path: str = db.Column(db.String(2048), primary_key=True)
+    count: int = db.Column(db.Integer, nullable=False, default=0)

app/models/ip_registry.py

@@ -0,0 +1,22 @@
+"""ip_registry table (Chapter 06 + Chapter 07 extension).
+
+last_verified_bot_result: NOT in Ch06's literal column list — added in
+Chapter 07. last_verified_at alone can't tell a cache hit *what* was
+verified, only *when*; this stores the outcome.
+"""
+from __future__ import annotations
+
+from app.extensions import db
+
+
+class IPRegistry(db.Model):
+    __tablename__ = "ip_registry"
+
+    ip: str = db.Column(db.String(45), primary_key=True)
+    first_seen = db.Column(db.DateTime, nullable=False)
+    last_seen = db.Column(db.DateTime, nullable=False)
+    total_requests: int = db.Column(db.Integer, nullable=False, default=0)
+    reputation_score: int = db.Column(db.Integer, nullable=False, default=0)
+    is_flagged: bool = db.Column(db.Boolean, nullable=False, default=False)
+    last_verified_at = db.Column(db.DateTime, nullable=True)
+    last_verified_bot_result: bool | None = db.Column(db.Boolean, nullable=True)

app/models/ip_traffic_stats.py

@@ -0,0 +1,35 @@
+"""NEW tables (explicit follow-up to Ch10's flagged scope gap): bounded
+per-IP traffic breakdown so the IP investigation panel reflects TRUE
+full traffic, not just bot_hits/suspicious_events activity.
+
+CARDINALITY NOTE: unlike the other Method-A tables, this one's row count
+scales with distinct IPs per day, which is unbounded for a probed site.
+Two mitigations: ip_path_stats_daily keeps only the top N paths per IP
+per day (not every ip x path pair); both tables use log_entries' ~30-day
+retention (Ch06), enforced by `flask cleanup` (Chapter 12).
+"""
+from __future__ import annotations
+
+from app.extensions import db
+
+
+class IpPathStatsDaily(db.Model):
+    __tablename__ = "ip_path_stats_daily"
+
+    date = db.Column(db.Date, primary_key=True)
+    ip: str = db.Column(db.String(45), primary_key=True)
+    path: str = db.Column(db.String(2048), primary_key=True)
+    count: int = db.Column(db.Integer, nullable=False, default=0)
+
+    __table_args__ = (db.Index("ix_ip_path_stats_ip", "ip"),)
+
+
+class IpStatusStatsDaily(db.Model):
+    __tablename__ = "ip_status_stats_daily"
+
+    date = db.Column(db.Date, primary_key=True)
+    ip: str = db.Column(db.String(45), primary_key=True)
+    status_bucket: str = db.Column(db.String(8), primary_key=True)  # 2xx|3xx|4xx|5xx
+    count: int = db.Column(db.Integer, nullable=False, default=0)
+
+    __table_args__ = (db.Index("ix_ip_status_stats_ip", "ip"),)

app/models/log_entry.py

@@ -0,0 +1,29 @@
+"""Optional, time-boxed raw storage (Chapter 06) — retention (~30 days)
+enforced by `flask cleanup` (Chapter 12)."""
+from __future__ import annotations
+
+from app.extensions import db
+
+
+class LogEntry(db.Model):
+    __tablename__ = "log_entries"
+
+    id: int = db.Column(db.Integer, primary_key=True)
+    log_file_id: int = db.Column(db.Integer, db.ForeignKey("log_files.id", ondelete="CASCADE"), nullable=False)
+    timestamp = db.Column(db.DateTime, nullable=False)  # normalized to UTC (Ch07)
+    ip: str = db.Column(db.String(45), nullable=False)  # IPv4 or IPv6
+    method: str = db.Column(db.String(10), nullable=False)
+    path: str = db.Column(db.Text, nullable=False)
+    status_code: int = db.Column(db.SmallInteger, nullable=False)
+    bytes_sent: int = db.Column(db.Integer, nullable=False, default=0)
+    referrer: str | None = db.Column(db.Text, nullable=True)
+    user_agent: str | None = db.Column(db.Text, nullable=True)
+    is_bot: bool = db.Column(db.Boolean, nullable=False, default=False)
+    flagged: bool = db.Column(db.Boolean, nullable=False, default=False)
+
+    __table_args__ = (
+        db.Index("ix_log_entries_file_ts", "log_file_id", "timestamp"),
+        db.Index("ix_log_entries_ip", "ip"),
+        db.Index("ix_log_entries_path", "path"),
+        db.Index("ix_log_entries_timestamp", "timestamp"),
+    )

app/models/log_file.py

@@ -0,0 +1,22 @@
+"""LogFile model (Chapter 06)."""
+from __future__ import annotations
+
+from datetime import datetime
+
+from app.extensions import db
+
+
+class LogFile(db.Model):
+    __tablename__ = "log_files"
+
+    id: int = db.Column(db.Integer, primary_key=True)
+    filename: str = db.Column(db.String(255), nullable=False)
+    server_type: str = db.Column(db.String(16), nullable=False)  # apache|litespeed
+    format_string: str = db.Column(db.Text, nullable=False)
+    uploaded_at: datetime = db.Column(db.DateTime, nullable=False, default=datetime.utcnow)
+    status: str = db.Column(db.String(16), nullable=False, default="queued")
+    total_lines: int | None = db.Column(db.Integer, nullable=True)
+    processed_lines: int = db.Column(db.Integer, nullable=False, default=0)
+    size_bytes: int = db.Column(db.BigInteger, nullable=False)
+    checksum: str = db.Column(db.String(64), nullable=False, index=True)  # sha256 hex
+    error_message: str | None = db.Column(db.Text, nullable=True)

app/models/referrer_stats.py

@@ -0,0 +1,14 @@
+"""NEW table (Method A, Ch08 follow-up) — not in Chapter 06's original
+list. Gives Top Referrers a rollup data source instead of scanning
+log_entries live. Domain-bucketed to keep cardinality bounded."""
+from __future__ import annotations
+
+from app.extensions import db
+
+
+class ReferrerStatsDaily(db.Model):
+    __tablename__ = "referrer_stats_daily"
+
+    date = db.Column(db.Date, primary_key=True)
+    referrer_domain: str = db.Column(db.String(255), primary_key=True)
+    count: int = db.Column(db.Integer, nullable=False, default=0)

app/models/request_stats.py

@@ -0,0 +1,26 @@
+"""Rollup tables (Chapter 06) — primary source for Overview widgets (Ch08).
+Both are site-wide (no log_file_id), consistent with Ch01's single-site scope.
+"""
+from __future__ import annotations
+
+from app.extensions import db
+
+
+class RequestStatsHourly(db.Model):
+    __tablename__ = "request_stats_hourly"
+
+    date_hour = db.Column(db.DateTime, primary_key=True)
+    path: str = db.Column(db.String(2048), primary_key=True)
+    status_code: int = db.Column(db.SmallInteger, primary_key=True)
+    count: int = db.Column(db.Integer, nullable=False, default=0)
+    bytes_sent_sum: int = db.Column(db.BigInteger, nullable=False, default=0)
+
+
+class RequestStatsDaily(db.Model):
+    __tablename__ = "request_stats_daily"
+
+    date = db.Column(db.Date, primary_key=True)
+    count: int = db.Column(db.Integer, nullable=False, default=0)
+    unique_ips: int = db.Column(db.Integer, nullable=False, default=0)
+    bytes_sum: int = db.Column(db.BigInteger, nullable=False, default=0)
+    error_count: int = db.Column(db.Integer, nullable=False, default=0)

app/models/suspicious_event.py

@@ -0,0 +1,22 @@
+"""suspicious_events table (Chapter 06) — Security section (Ch10)."""
+from __future__ import annotations
+
+from app.extensions import db
+
+
+class SuspiciousEvent(db.Model):
+    __tablename__ = "suspicious_events"
+
+    id: int = db.Column(db.Integer, primary_key=True)
+    log_file_id: int = db.Column(db.Integer, db.ForeignKey("log_files.id", ondelete="CASCADE"), nullable=False)
+    ip: str = db.Column(db.String(45), nullable=False)
+    timestamp = db.Column(db.DateTime, nullable=False)
+    path: str = db.Column(db.Text, nullable=False)
+    rule_matched: str = db.Column(db.String(128), nullable=False)
+    severity: str = db.Column(db.String(8), nullable=False)  # low|medium|high (provisional; Ch10 escalates)
+
+    __table_args__ = (
+        db.Index("ix_suspicious_events_timestamp", "timestamp"),
+        db.Index("ix_suspicious_events_ip", "ip"),
+        db.Index("ix_suspicious_events_severity", "severity"),
+    )

app/models/user.py

@@ -0,0 +1,27 @@
+"""Single-admin user model (Chapter 12). Not in Chapter 06's table list —
+that chapter scopes log-analytics tables; auth is a separate concern this
+chapter owns. Chapter 01 confirms single-admin, no self-registration.
+"""
+from __future__ import annotations
+
+from datetime import datetime
+
+import bcrypt
+from flask_login import UserMixin
+
+from app.extensions import db
+
+
+class User(db.Model, UserMixin):
+    __tablename__ = "users"
+
+    id: int = db.Column(db.Integer, primary_key=True)
+    email: str = db.Column(db.String(255), unique=True, nullable=False, index=True)
+    password_hash: str = db.Column(db.String(255), nullable=False)
+    created_at = db.Column(db.DateTime, nullable=False, default=datetime.utcnow)
+
+    def set_password(self, raw_password: str) -> None:
+        self.password_hash = bcrypt.hashpw(raw_password.encode("utf-8"), bcrypt.gensalt()).decode("utf-8")
+
+    def check_password(self, raw_password: str) -> bool:
+        return bcrypt.checkpw(raw_password.encode("utf-8"), self.password_hash.encode("utf-8"))

app/services/aggregator.py

@@ -0,0 +1,199 @@
+"""Rollup computation (Chapter 06 + Method A extensions from Ch08/09/10).
+
+Called once per parsed file (app/cli.py::process_logs) for the dates it
+touched, ad hoc via `flask rollup` for a manual recompute, and now also
+after a file deletion (app/services/file_deletion.py) for whatever dates
+the deleted file touched. Every _upsert_* function scans log_entries
+inside this background batch job, never at request time — that's what
+makes Ch03 rule 6 compliance possible.
+
+CORRECTNESS FIX: every rollup writer below now deletes a day's existing
+rows before writing whatever the fresh scan finds (including writing
+nothing, if a day now has zero data). Three of the five writers
+previously only ever upserted-when-present and silently left stale rows
+behind when a day's data disappeared — unreachable before file deletion
+existed (rollups only ever grew), but a real correctness bug once
+deletion makes "this day now has less data than before" possible. Only
+the two per-IP writers already had this right (Ch10 follow-up); the
+other three are fixed here to match.
+"""
+from __future__ import annotations
+
+from collections import defaultdict
+from datetime import date, datetime, time, timedelta
+
+from sqlalchemy import case, func
+
+from app.extensions import db
+from app.models.browser_stats import BrowserStatsDaily
+from app.models.human_path_stats import HumanPathStatsDaily
+from app.models.ip_traffic_stats import IpPathStatsDaily, IpStatusStatsDaily
+from app.models.log_entry import LogEntry
+from app.models.referrer_stats import ReferrerStatsDaily
+from app.models.request_stats import RequestStatsDaily, RequestStatsHourly
+from app.services.blocklist import refresh_blocklist_suggestions
+from app.services.referrer import referrer_domain
+from app.services.ua_classifier import classify_browser, classify_os
+from app.utils.http_status import status_bucket
+
+TOP_PATHS_PER_IP_PER_DAY = 15  # bounds ip_path_stats_daily row growth (Ch10 follow-up)
+
+
+def compute_rollups_for_range(start: date, end: date) -> None:
+    """Recompute every rollup for each day in [start, end]. Site-wide,
+    not per-file (Ch01: single site) — recomputing from scratch per day
+    avoids double-counting when two uploads cover the same period, and
+    correctly shrinks a day's numbers back down when a file covering
+    that day is deleted.
+    """
+    current = start
+    while current <= end:
+        _upsert_hourly(current)
+        _upsert_daily(current)
+        _upsert_per_line_derived_stats(current)
+        refresh_blocklist_suggestions(current)
+        current += timedelta(days=1)
+
+
+def _day_bounds(day: date) -> tuple[datetime, datetime]:
+    start = datetime.combine(day, time.min)
+    return start, start + timedelta(days=1)
+
+
+def _upsert_hourly(day: date) -> None:
+    start, end = _day_bounds(day)
+    rows = (
+        db.session.query(
+            func.strftime("%Y-%m-%d %H:00:00", LogEntry.timestamp).label("date_hour"),
+            LogEntry.path,
+            LogEntry.status_code,
+            func.count().label("count"),
+            func.coalesce(func.sum(LogEntry.bytes_sent), 0).label("bytes_sent_sum"),
+        )
+        .filter(LogEntry.timestamp >= start, LogEntry.timestamp < end)
+        .group_by("date_hour", LogEntry.path, LogEntry.status_code)
+        .all()
+    )
+
+    # Delete-then-insert: replaces the day's hourly rows entirely,
+    # including leaving none behind if `rows` is now empty (e.g. the
+    # only file covering this day was just deleted).
+    db.session.query(RequestStatsHourly).filter(
+        RequestStatsHourly.date_hour >= start, RequestStatsHourly.date_hour < end
+    ).delete()
+
+    if rows:
+        payload = [
+            {
+                "date_hour": datetime.strptime(r.date_hour, "%Y-%m-%d %H:%M:%S"),
+                "path": r.path,
+                "status_code": r.status_code,
+                "count": r.count,
+                "bytes_sent_sum": r.bytes_sent_sum,
+            }
+            for r in rows
+        ]
+        db.session.execute(RequestStatsHourly.__table__.insert(), payload)
+
+    db.session.commit()
+
+
+def _upsert_daily(day: date) -> None:
+    start, end = _day_bounds(day)
+    result = (
+        db.session.query(
+            func.count().label("count"),
+            func.count(func.distinct(LogEntry.ip)).label("unique_ips"),
+            func.coalesce(func.sum(LogEntry.bytes_sent), 0).label("bytes_sum"),
+            func.coalesce(func.sum(case((LogEntry.status_code >= 400, 1), else_=0)), 0).label("error_count"),
+        )
+        .filter(LogEntry.timestamp >= start, LogEntry.timestamp < end)
+        .one()
+    )
+
+    # Delete-then-insert: if this day now has zero entries (its only
+    # contributing file was deleted), the stale row is removed rather
+    # than left behind — no rollup row is better than a wrong one.
+    db.session.query(RequestStatsDaily).filter(RequestStatsDaily.date == day).delete()
+
+    if result.count > 0:
+        db.session.execute(
+            RequestStatsDaily.__table__.insert(),
+            {
+                "date": day,
+                "count": result.count,
+                "unique_ips": result.unique_ips,
+                "bytes_sum": result.bytes_sum,
+                "error_count": result.error_count,
+            },
+        )
+
+    db.session.commit()
+
+
+def _upsert_per_line_derived_stats(day: date) -> None:
+    """Referrer domain, browser/OS, human-only path counts (Ch08/09), and
+    per-IP path/status counts (Ch10 follow-up) — one streamed pass over
+    log_entries (Ch03 rule 1: bounded per-chunk memory via yield_per,
+    never the whole day loaded at once). Every table here uses the same
+    delete-then-insert pattern so a day's rows are fully replaced by
+    whatever the fresh scan finds, including nothing.
+    """
+    start, end = _day_bounds(day)
+    referrer_counts: dict[str, int] = defaultdict(int)
+    browser_counts: dict[tuple[str, str], int] = defaultdict(int)
+    human_path_counts: dict[str, int] = defaultdict(int)
+    ip_path_counts: dict[str, dict[str, int]] = defaultdict(lambda: defaultdict(int))
+    ip_status_counts: dict[str, dict[str, int]] = defaultdict(lambda: defaultdict(int))
+
+    query = (
+        db.session.query(
+            LogEntry.referrer, LogEntry.user_agent, LogEntry.is_bot,
+            LogEntry.path, LogEntry.ip, LogEntry.status_code,
+        )
+        .filter(LogEntry.timestamp >= start, LogEntry.timestamp < end)
+    )
+    for referrer, user_agent, is_bot, path, ip, status_code in query.yield_per(1000):
+        domain = referrer_domain(referrer)
+        if domain:
+            referrer_counts[domain] += 1
+        if not is_bot:  # Ch08: bot traffic excluded from human browser/OS breakdown
+            browser_counts[(classify_browser(user_agent), classify_os(user_agent))] += 1
+            human_path_counts[path] += 1
+        ip_path_counts[ip][path] += 1
+        ip_status_counts[ip][status_bucket(status_code)] += 1
+
+    db.session.query(ReferrerStatsDaily).filter(ReferrerStatsDaily.date == day).delete()
+    if referrer_counts:
+        payload = [{"date": day, "referrer_domain": d, "count": c} for d, c in referrer_counts.items()]
+        db.session.execute(ReferrerStatsDaily.__table__.insert(), payload)
+
+    db.session.query(BrowserStatsDaily).filter(BrowserStatsDaily.date == day).delete()
+    if browser_counts:
+        payload = [{"date": day, "browser": b, "os": o, "count": c} for (b, o), c in browser_counts.items()]
+        db.session.execute(BrowserStatsDaily.__table__.insert(), payload)
+
+    db.session.query(HumanPathStatsDaily).filter(HumanPathStatsDaily.date == day).delete()
+    if human_path_counts:
+        payload = [{"date": day, "path": p, "count": c} for p, c in human_path_counts.items()]
+        db.session.execute(HumanPathStatsDaily.__table__.insert(), payload)
+
+    db.session.query(IpPathStatsDaily).filter(IpPathStatsDaily.date == day).delete()
+    if ip_path_counts:
+        payload = []
+        for ip, paths in ip_path_counts.items():
+            top_paths = sorted(paths.items(), key=lambda kv: kv[1], reverse=True)[:TOP_PATHS_PER_IP_PER_DAY]
+            payload.extend({"date": day, "ip": ip, "path": p, "count": c} for p, c in top_paths)
+        if payload:
+            db.session.execute(IpPathStatsDaily.__table__.insert(), payload)
+
+    db.session.query(IpStatusStatsDaily).filter(IpStatusStatsDaily.date == day).delete()
+    if ip_status_counts:
+        payload = [
+            {"date": day, "ip": ip, "status_bucket": bucket, "count": c}
+            for ip, buckets in ip_status_counts.items()
+            for bucket, c in buckets.items()
+        ]
+        db.session.execute(IpStatusStatsDaily.__table__.insert(), payload)
+
+    db.session.commit()

app/services/background.py

@@ -0,0 +1,97 @@
+"""No-cron automatic processing (Chapter 12 simplification, per project
+owner request): triggers file parsing immediately in a background thread
+right after upload, and opportunistically resumes any incomplete files
+when the Overview page loads — replacing the cron-triggered model.
+`flask process-logs` still exists in app/cli.py for anyone who'd rather
+use cron, but nothing requires it anymore.
+
+TRADEOFF (flagged, deviating from Chapter 02/03's "no persistent
+background workers, cron-triggered CLI only" stance): a background thread
+lives inside the same worker process that handled the upload request. If
+Passenger recycles that process mid-parse, the thread dies with it —
+progress up to the last commit is still safely checkpointed (same bounded-
+batch model as before), but nothing will automatically resume it without
+either cron or a page visit. The "resume on page load" hook below is the
+deliberate replacement for that guarantee: visiting the Overview tab
+re-triggers processing for anything left incomplete, so in the worst case
+a stuck file resumes the next time the admin looks at the dashboard,
+rather than never.
+
+This is not a long-lived daemon: each thread terminates once its file
+reaches "done"/"error" (or the process is killed), and no thread survives
+a process restart — it just gets re-triggered fresh next time.
+"""
+from __future__ import annotations
+
+import threading
+
+from flask import Flask
+
+from app.extensions import db
+from app.models.log_file import LogFile
+from app.services.log_processor import process_one_batch
+
+# In-process guard against launching two threads for the same file at
+# once (e.g. the upload trigger and a page-load resume firing close
+# together). Per-worker-process only — a different entry process picking
+# up the same file concurrently is a low-probability edge case accepted
+# for this simplification; each write is still a small checkpointed
+# commit, not a giant one, which limits how bad a collision could be.
+_active_file_ids: set[int] = set()
+_lock = threading.Lock()
+
+
+def _claim(log_file_id: int) -> bool:
+    with _lock:
+        if log_file_id in _active_file_ids:
+            return False
+        _active_file_ids.add(log_file_id)
+        return True
+
+
+def _release(log_file_id: int) -> None:
+    with _lock:
+        _active_file_ids.discard(log_file_id)
+
+
+def _run_to_completion(app: Flask, log_file_id: int, batch_size: int) -> None:
+    with app.app_context():
+        try:
+            log_file = db.session.get(LogFile, log_file_id)
+            if log_file is None:
+                return
+            while log_file.status in ("queued", "processing"):
+                process_one_batch(log_file, batch_size)
+                db.session.refresh(log_file)
+        except Exception:
+            app.logger.exception("Background processing failed for log_file_id=%s", log_file_id)
+            log_file = db.session.get(LogFile, log_file_id)
+            if log_file is not None and log_file.status != "done":
+                log_file.status = "error"
+                log_file.error_message = "Processing failed unexpectedly; see server logs."
+                db.session.commit()
+        finally:
+            _release(log_file_id)
+
+
+def trigger_processing(app: Flask, log_file_id: int) -> None:
+    """Start background processing for one file; no-ops if already running."""
+    if not _claim(log_file_id):
+        return
+    batch_size = app.config["PARSE_BATCH_SIZE"]
+    thread = threading.Thread(
+        target=_run_to_completion, args=(app, log_file_id, batch_size), daemon=True
+    )
+    thread.start()
+
+
+def resume_incomplete_files(app: Flask) -> None:
+    """Opportunistic resume hook, called from the Overview page load —
+    the deliberate replacement for cron's "there's always a next tick"
+    guarantee. Cheap: one indexed status-filtered query.
+    """
+    incomplete_ids = [
+        lf.id for lf in LogFile.query.filter(LogFile.status.in_(["queued", "processing"])).all()
+    ]
+    for log_file_id in incomplete_ids:
+        trigger_processing(app, log_file_id)

app/services/blocklist.py

@@ -0,0 +1,67 @@
+"""Blocklist-suggestion generation (Chapter 10 follow-up): no earlier
+chapter assigned ownership of populating blocklist_suggestions or setting
+ip_registry.is_flagged. Runs in the background aggregator pass (batch, not
+request-time, per Ch02/03), reusing severity_scoring.py so there's exactly
+one scoring model between the dashboard display and the flagging decision.
+"""
+from __future__ import annotations
+
+from collections import defaultdict
+from datetime import date, datetime, timedelta
+
+from app.extensions import db
+from app.models.blocklist_suggestion import BlocklistSuggestion
+from app.models.ip_registry import IPRegistry
+from app.models.suspicious_event import SuspiciousEvent
+from app.services.severity_scoring import SeverityInputs, compute_effective_severity
+
+_RANK = {"low": 0, "medium": 1, "high": 2}
+
+
+def refresh_blocklist_suggestions(day: date) -> None:
+    """Flag an IP (is_flagged + a suggestion row) if its escalated severity
+    for `day` reaches 'high'. Idempotent — skips IPs already suggested.
+    """
+    start = datetime.combine(day, datetime.min.time())
+    end = start + timedelta(days=1)
+
+    events = (
+        db.session.query(SuspiciousEvent.ip, SuspiciousEvent.timestamp, SuspiciousEvent.severity)
+        .filter(SuspiciousEvent.timestamp >= start, SuspiciousEvent.timestamp < end)
+        .all()
+    )
+    if not events:
+        return
+
+    by_ip: dict[str, list] = defaultdict(list)
+    for ip, ts, sev in events:
+        by_ip[ip].append((ts, sev))
+
+    already_suggested = {ip for (ip,) in db.session.query(BlocklistSuggestion.ip).distinct().all()}
+
+    for ip, ip_events in by_ip.items():
+        if ip in already_suggested:
+            continue
+        timestamps = sorted(ts for ts, _ in ip_events)
+        avg_interval = (
+            (timestamps[-1] - timestamps[0]).total_seconds() / (len(timestamps) - 1)
+            if len(timestamps) > 1 else None
+        )
+        worst_base = max((sev for _, sev in ip_events), key=lambda s: _RANK.get(s, 0))
+        effective = compute_effective_severity(
+            SeverityInputs(base_severity=worst_base, ip_event_count=len(ip_events), avg_interval_seconds=avg_interval)
+        )
+        if effective != "high":
+            continue
+
+        db.session.add(BlocklistSuggestion(
+            ip=ip,
+            reason=f"{len(ip_events)} suspicious event(s) on {day.isoformat()}, escalated to high severity",
+            created_at=datetime.utcnow(),
+            exported=False,
+        ))
+        ip_row = db.session.get(IPRegistry, ip)
+        if ip_row is not None:
+            ip_row.is_flagged = True
+
+    db.session.commit()

app/services/bot_identifier.py

@@ -0,0 +1,78 @@
+"""Bot signature matching + reverse-DNS verification (Chapter 07).
+
+Signatures are data (JSON), not hardcoded logic, so the list grows without
+a code change. Verification does real DNS I/O — only ever called from the
+background process-logs batch job (app/services/classification.py), never
+synchronously inside a dashboard request, per Chapter 07's explicit rule.
+"""
+from __future__ import annotations
+
+import json
+import socket
+from dataclasses import dataclass
+from datetime import datetime, timedelta
+from pathlib import Path
+
+SIGNATURES_PATH = Path(__file__).parent / "data" / "bot_signatures.json"
+
+# Skip re-verifying the same IP more often than this (Ch07: DNS latency is
+# a real cost on constrained hosting).
+VERIFICATION_TTL = timedelta(days=7)
+
+
+@dataclass(frozen=True)
+class BotSignature:
+    name: str
+    ua_substrings: tuple[str, ...]
+    verify_suffixes: tuple[str, ...]  # PTR hostname must end in one of these
+
+
+def _load_signatures() -> list[BotSignature]:
+    raw = json.loads(SIGNATURES_PATH.read_text())
+    return [
+        BotSignature(name=e["name"], ua_substrings=tuple(e["ua_substrings"]), verify_suffixes=tuple(e["verify_suffixes"]))
+        for e in raw
+    ]
+
+
+_SIGNATURES = _load_signatures()
+
+
+def classify_bot(user_agent: str | None) -> str | None:
+    """Return the claimed bot name via UA substring match, or None."""
+    if not user_agent:
+        return None
+    ua_lower = user_agent.lower()
+    for sig in _SIGNATURES:
+        if any(sub.lower() in ua_lower for sub in sig.ua_substrings):
+            return sig.name
+    return None
+
+
+def _signature_for(bot_name: str) -> BotSignature | None:
+    return next((s for s in _SIGNATURES if s.name == bot_name), None)
+
+
+def verify_bot_ip(ip: str, bot_name: str) -> bool:
+    """Reverse-DNS + forward-confirm that `ip` really belongs to `bot_name`."""
+    sig = _signature_for(bot_name)
+    if sig is None:
+        return False
+    try:
+        hostname, _, _ = socket.gethostbyaddr(ip)
+    except (socket.herror, socket.gaierror, OSError):
+        return False
+    if not any(hostname.lower().endswith(suffix) for suffix in sig.verify_suffixes):
+        return False
+    try:
+        forward_ips = socket.gethostbyname_ex(hostname)[2]
+    except (socket.herror, socket.gaierror, OSError):
+        return False
+    return ip in forward_ips
+
+
+def is_verification_stale(last_verified_at: datetime | None) -> bool:
+    """True if this IP needs a fresh DNS check (Ch07 caching rule)."""
+    if last_verified_at is None:
+        return True
+    return datetime.utcnow() - last_verified_at > VERIFICATION_TTL

app/services/classification.py

@@ -0,0 +1,122 @@
+"""Per-batch classification pipeline (Chapter 07): ties bot_identifier and
+threat_scanner into the bot_hits/suspicious_events/ip_registry write path.
+Called once per bulk-insert chunk from app/cli.py — DNS-based verification
+belongs here (background batch job), never in a dashboard request.
+"""
+from __future__ import annotations
+
+from dataclasses import dataclass
+from datetime import datetime
+
+from sqlalchemy import func, insert
+from sqlalchemy.dialects.sqlite import insert as sqlite_insert
+
+from app.extensions import db
+from app.models.bot_hit import BotHit
+from app.models.ip_registry import IPRegistry
+from app.models.suspicious_event import SuspiciousEvent
+from app.services import bot_identifier, threat_scanner
+from app.services.log_parser import ParsedEntry
+
+
+@dataclass
+class EntryClassification:
+    """Cheap, no-I/O flags for the log_entries row itself."""
+    is_bot: bool
+    flagged: bool
+
+
+def classify_entry(entry: ParsedEntry) -> EntryClassification:
+    """No DNS I/O here — bot *verification* is batched separately below,
+    since it's stateful (cached per IP) and only worth doing once per IP
+    per batch, not once per line.
+    """
+    return EntryClassification(
+        is_bot=bot_identifier.classify_bot(entry.user_agent) is not None,
+        flagged=threat_scanner.scan(entry) is not None,
+    )
+
+
+def write_batch_side_effects(log_file_id: int, entries: list[ParsedEntry]) -> None:
+    """Derive and bulk-write bot_hits, suspicious_events, ip_registry upserts."""
+    if not entries:
+        return
+
+    bot_hit_rows: list[dict] = []
+    suspicious_rows: list[dict] = []
+    ip_agg: dict[str, dict] = {}
+    verified_this_batch: dict[str, bool] = {}  # avoid repeat DNS for the same IP in one batch
+
+    for entry in entries:
+        agg = ip_agg.setdefault(entry.ip, {"first": entry.timestamp, "last": entry.timestamp, "count": 0})
+        agg["count"] += 1
+        agg["first"] = min(agg["first"], entry.timestamp)
+        agg["last"] = max(agg["last"], entry.timestamp)
+
+        bot_name = bot_identifier.classify_bot(entry.user_agent)
+        if bot_name is not None:
+            verified = verified_this_batch.get(entry.ip)
+            if verified is None:
+                verified = _verify_with_cache(entry.ip, bot_name)
+                verified_this_batch[entry.ip] = verified
+            bot_hit_rows.append({
+                "log_file_id": log_file_id, "timestamp": entry.timestamp, "ip": entry.ip,
+                "bot_name": bot_name, "verified": verified, "path": entry.path,
+                "status_code": entry.status_code,
+            })
+            if not verified:
+                # Single detection, two dashboard consumers (Ch07): spoofed
+                # bot also surfaces as a suspicious_events row.
+                suspicious_rows.append({
+                    "log_file_id": log_file_id, "ip": entry.ip, "timestamp": entry.timestamp,
+                    "path": entry.path, "rule_matched": f"spoofed_bot:{bot_name}", "severity": "medium",
+                })
+
+        threat = threat_scanner.scan(entry)
+        if threat is not None:
+            suspicious_rows.append({
+                "log_file_id": log_file_id, "ip": entry.ip, "timestamp": entry.timestamp,
+                "path": entry.path, "rule_matched": threat.rule_matched, "severity": threat.severity,
+            })
+
+    if bot_hit_rows:
+        db.session.execute(insert(BotHit.__table__), bot_hit_rows)
+    if suspicious_rows:
+        db.session.execute(insert(SuspiciousEvent.__table__), suspicious_rows)
+
+    _upsert_ip_registry(ip_agg, verified_this_batch)
+    db.session.commit()
+
+
+def _verify_with_cache(ip: str, bot_name: str) -> bool:
+    """TTL-gated reverse/forward DNS check, cached via
+    ip_registry.last_verified_bot_result (Ch07 schema addition).
+    """
+    row = db.session.get(IPRegistry, ip)
+    if row is not None and not bot_identifier.is_verification_stale(row.last_verified_at):
+        return bool(row.last_verified_bot_result)
+    return bot_identifier.verify_bot_ip(ip, bot_name)
+
+
+def _upsert_ip_registry(ip_agg: dict[str, dict], verified_this_batch: dict[str, bool]) -> None:
+    now = datetime.utcnow()
+    for ip, agg in ip_agg.items():
+        values = {
+            "ip": ip, "first_seen": agg["first"], "last_seen": agg["last"],
+            "total_requests": agg["count"], "reputation_score": 0, "is_flagged": False,
+        }
+        if ip in verified_this_batch:
+            values["last_verified_at"] = now
+            values["last_verified_bot_result"] = verified_this_batch[ip]
+
+        stmt = sqlite_insert(IPRegistry.__table__).values(**values)
+        update_set = {
+            "last_seen": func.max(IPRegistry.last_seen, stmt.excluded.last_seen),
+            "first_seen": func.min(IPRegistry.first_seen, stmt.excluded.first_seen),
+            "total_requests": IPRegistry.total_requests + stmt.excluded.total_requests,
+        }
+        if ip in verified_this_batch:
+            update_set["last_verified_at"] = stmt.excluded.last_verified_at
+            update_set["last_verified_bot_result"] = stmt.excluded.last_verified_bot_result
+        stmt = stmt.on_conflict_do_update(index_elements=["ip"], set_=update_set)
+        db.session.execute(stmt)

app/services/data/bot_signatures.json

@@ -0,0 +1,9 @@
+[
+  {"name": "Googlebot", "ua_substrings": ["Googlebot"], "verify_suffixes": [".googlebot.com", ".google.com"]},
+  {"name": "Bingbot", "ua_substrings": ["bingbot"], "verify_suffixes": [".search.msn.com"]},
+  {"name": "Yandex", "ua_substrings": ["YandexBot"], "verify_suffixes": [".yandex.ru", ".yandex.com", ".yandex.net"]},
+  {"name": "Baidu", "ua_substrings": ["Baiduspider"], "verify_suffixes": [".baidu.com", ".baidu.jp"]},
+  {"name": "DuckDuckBot", "ua_substrings": ["DuckDuckBot"], "verify_suffixes": [".duckduckgo.com"]},
+  {"name": "AhrefsBot", "ua_substrings": ["AhrefsBot"], "verify_suffixes": [".ahrefs.com"]},
+  {"name": "SemrushBot", "ua_substrings": ["SemrushBot"], "verify_suffixes": [".semrush.com"]}
+]

app/services/data/browser_signatures.json

@@ -0,0 +1,8 @@
+[
+  {"name": "Edge", "ua_substrings": ["Edg/", "EdgA/", "EdgiOS/"]},
+  {"name": "Opera", "ua_substrings": ["OPR/", "Opera"]},
+  {"name": "Chrome", "ua_substrings": ["Chrome/", "CriOS/"]},
+  {"name": "Firefox", "ua_substrings": ["Firefox/", "FxiOS/"]},
+  {"name": "Safari", "ua_substrings": ["Safari/"]},
+  {"name": "Internet Explorer", "ua_substrings": ["MSIE ", "Trident/"]}
+]

app/services/data/os_signatures.json

@@ -0,0 +1,7 @@
+[
+  {"name": "Windows", "ua_substrings": ["Windows NT"]},
+  {"name": "iOS", "ua_substrings": ["iPhone", "iPad", "iPod"]},
+  {"name": "macOS", "ua_substrings": ["Mac OS X", "Macintosh"]},
+  {"name": "Android", "ua_substrings": ["Android"]},
+  {"name": "Linux", "ua_substrings": ["Linux"]}
+]

app/services/data/threat_patterns.json

@@ -0,0 +1,11 @@
+{
+  "sensitive_paths": [
+    "/.env", "/.git/config", "wp-config.php", "/.htpasswd", "/phpmyadmin",
+    "/xmlrpc.php", ".sql.gz", ".sql.bak", ".zip", ".bak", ".old"
+  ],
+  "injection_markers": [
+    "union select", "' or '1'='1", "<script>", "javascript:",
+    "../", "..%2f", "%2e%2e%2f", "%252e%252e%252f"
+  ],
+  "scanner_user_agents": ["sqlmap", "nikto", "nmap", "acunetix", "nessus"]
+}

app/services/file_deletion.py

@@ -0,0 +1,93 @@
+"""Uploaded-file deletion (project-owner follow-up request).
+
+Deleting a LogFile is more than one DELETE statement: log_entries,
+bot_hits, and suspicious_events all reference log_file_id and must be
+removed explicitly — this project's SQLite connections don't have
+`PRAGMA foreign_keys=ON`, so the `ondelete="CASCADE"` in the migrations
+(Ch06) is declarative documentation only, not an enforced behavior.
+Relying on it would silently leave orphaned rows behind.
+
+The harder part is the rollup tables (request_stats_hourly/daily,
+referrer/browser/human-path/ip-path/ip-status): none of them have a
+log_file_id column (Ch06: they're site-wide, since two files can share a
+date). So instead of deleting rollup rows for the deleted file's dates
+directly (which could wipe out another file's contribution to the same
+date), this captures the affected dates BEFORE deleting, then calls
+aggregator.compute_rollups_for_range() AFTER deleting — which recomputes
+each affected day from whatever log_entries remain, correctly handling
+both "another file still covers this day" and "no file covers this day
+anymore" (the latter now handled correctly by the aggregator.py fix that
+shipped alongside this feature).
+
+KNOWN LIMITATION (flagged, not fixed): ip_registry (total_requests,
+first_seen, last_seen, verification cache) and blocklist_suggestions are
+NOT per-file and are NOT recomputed on deletion — doing so correctly
+would mean re-scanning all remaining log_entries for every affected IP,
+which is unbounded work for a single delete action and would violate the
+same "no raw-row scan on a hot path" reasoning Ch03 applies elsewhere.
+After deleting a file, IP History numbers may include a deleted file's
+historical contribution until a full site-wide recompute is added as a
+separate feature. Doesn't affect correctness of Overview/SEO's date-
+scoped numbers, which is what this feature was actually asked to fix.
+"""
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+
+from app.extensions import db
+from app.models.bot_hit import BotHit
+from app.models.log_entry import LogEntry
+from app.models.log_file import LogFile
+from app.models.suspicious_event import SuspiciousEvent
+from app.services import aggregator
+from app.utils.upload_paths import upload_path_for
+
+
+@dataclass
+class BulkDeleteResult:
+    deleted: list[int] = field(default_factory=list)
+    skipped: list[dict] = field(default_factory=list)  # [{"id": ..., "reason": ...}]
+
+
+def delete_log_files(log_file_ids: list[int]) -> BulkDeleteResult:
+    """Delete each id in `log_file_ids`; recompute rollups once at the end
+    for the full span of dates any deleted file touched (cheaper than
+    recomputing per-file when multiple files share dates).
+    """
+    result = BulkDeleteResult()
+    all_touched_dates: set = set()
+
+    for log_file_id in log_file_ids:
+        log_file = db.session.get(LogFile, log_file_id)
+        if log_file is None:
+            result.skipped.append({"id": log_file_id, "reason": "not found"})
+            continue
+        if log_file.status == "processing":
+            result.skipped.append({"id": log_file_id, "reason": "currently being analyzed"})
+            continue
+
+        touched_dates = {
+            row.date() for row, in db.session.query(LogEntry.timestamp)
+            .filter(LogEntry.log_file_id == log_file_id).distinct()
+        }
+        # distinct() on a full timestamp rarely collapses much; reduce to
+        # calendar dates in Python since SQLite's DATE() in a DISTINCT
+        # clause is a bit more awkward to express portably here.
+        all_touched_dates |= touched_dates
+
+        db.session.query(SuspiciousEvent).filter(SuspiciousEvent.log_file_id == log_file_id).delete()
+        db.session.query(BotHit).filter(BotHit.log_file_id == log_file_id).delete()
+        db.session.query(LogEntry).filter(LogEntry.log_file_id == log_file_id).delete()
+
+        raw_path = upload_path_for(log_file)
+        if raw_path.exists():
+            raw_path.unlink()
+
+        db.session.delete(log_file)
+        db.session.commit()
+        result.deleted.append(log_file_id)
+
+    if all_touched_dates:
+        aggregator.compute_rollups_for_range(min(all_touched_dates), max(all_touched_dates))
+
+    return result

app/services/log_parser/__init__.py

@@ -0,0 +1,47 @@
+"""Log parser abstraction (Chapter 07)."""
+from __future__ import annotations
+
+from dataclasses import dataclass
+from datetime import datetime
+from typing import Protocol, TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from app.models.log_file import LogFile
+
+
+@dataclass
+class ParsedEntry:
+    timestamp: datetime
+    ip: str
+    method: str
+    path: str
+    status_code: int
+    bytes_sent: int
+    referrer: str | None
+    user_agent: str | None
+
+
+class LogParser(Protocol):
+    def parse_line(self, line: str) -> ParsedEntry | None: ...
+
+
+def get_parser_for(log_file: "LogFile") -> LogParser:
+    """Build the right LogParser for a given upload's format_string.
+
+    ASSUMPTION (flagged): a blank format_string — which the upload form
+    allows — falls back to Combined, since Chapter 07 states no default
+    for that case. Anything else is compiled as-is; a real custom
+    LiteSpeed format is never silently replaced by a preset.
+    """
+    # Imported lazily to avoid a circular import (apache.py imports
+    # ParsedEntry from this module).
+    from app.services.log_parser.apache import FormatCompiledParser, combined_parser, common_parser
+    from app.services.log_parser.format_compiler import compile_format
+
+    fmt = (log_file.format_string or "").strip()
+    key = fmt.lower()
+    if key in ("", "combined"):
+        return combined_parser()
+    if key == "common":
+        return common_parser()
+    return FormatCompiledParser(compile_format(fmt))

app/services/log_parser/apache.py

@@ -0,0 +1,57 @@
+"""Apache Common/Combined presets + the generic parser built from
+format_compiler (Chapter 07)."""
+from __future__ import annotations
+
+from app.services.log_parser import ParsedEntry
+from app.services.log_parser.format_compiler import (
+    CompiledFormat, compile_format, parse_apache_timestamp, parse_request_line, validate_ip,
+)
+
+COMMON_LOG_FORMAT = '%h %l %u %t "%r" %>s %b'
+COMBINED_LOG_FORMAT = '%h %l %u %t "%r" %>s %b "%{Referer}i" "%{User-agent}i"'
+
+
+class FormatCompiledParser:
+    """LogParser implementation driven by any CompiledFormat (Ch04/07 protocol)."""
+
+    def __init__(self, compiled: CompiledFormat):
+        self._compiled = compiled
+
+    def parse_line(self, line: str) -> ParsedEntry | None:
+        match = self._compiled.pattern.match(line.rstrip("\n"))
+        if match is None:
+            return None  # malformed line — skip, never raise (Ch07)
+        fields = match.groupdict()
+
+        ip = validate_ip(fields["ip"])
+        if ip is None:
+            return None
+
+        req = parse_request_line(fields["request"])
+        if req is None:
+            return None
+        method, path = req
+
+        try:
+            timestamp = parse_apache_timestamp(fields["timestamp"])
+            status_code = int(fields["status"])
+            bytes_sent = 0 if fields["bytes"] == "-" else int(fields["bytes"])
+        except (ValueError, KeyError):
+            return None
+
+        referrer = fields.get("referrer")
+        user_agent = fields.get("user_agent")
+        return ParsedEntry(
+            timestamp=timestamp, ip=ip, method=method, path=path,
+            status_code=status_code, bytes_sent=bytes_sent,
+            referrer=None if referrer in (None, "-") else referrer,
+            user_agent=None if user_agent in (None, "-") else user_agent,
+        )
+
+
+def common_parser() -> FormatCompiledParser:
+    return FormatCompiledParser(compile_format(COMMON_LOG_FORMAT))
+
+
+def combined_parser() -> FormatCompiledParser:
+    return FormatCompiledParser(compile_format(COMBINED_LOG_FORMAT))

app/services/log_parser/format_compiler.py

@@ -0,0 +1,95 @@
+"""LogFormat directive -> compiled regex + named groups (Chapter 07).
+
+One code path for Common, Combined, and arbitrary custom formats — presets
+below are just specific format strings run through this same compiler.
+"""
+from __future__ import annotations
+
+import re
+from dataclasses import dataclass
+from datetime import datetime, timezone
+from ipaddress import ip_address
+
+# Directive -> (group name, regex pattern). NOTE: %r and %{...}i patterns
+# deliberately do NOT include quote characters — the quotes around them
+# are literal text already present in the format string (e.g. `"%r"`),
+# and get regex-escaped by the literal-text path below. %t is different:
+# its brackets are part of what %t itself produces, not literal format-
+# string text, so they belong in this directive's own pattern.
+_SIMPLE_DIRECTIVES: dict[str, tuple[str, str]] = {
+    "%h": ("ip", r"(?P<ip>\S+)"),
+    "%l": ("ident", r"(?P<ident>\S+)"),
+    "%u": ("user", r"(?P<user>\S+)"),
+    "%t": ("timestamp", r"\[(?P<timestamp>[^\]]+)\]"),
+    "%r": ("request", r'(?P<request>[^"]*)'),
+    "%>s": ("status", r"(?P<status>\d{3})"),
+    "%s": ("status", r"(?P<status>\d{3})"),
+    "%b": ("bytes", r"(?P<bytes>\d+|-)"),
+}
+
+_HEADER_DIRECTIVE_RE = re.compile(r'%\{([^}]+)\}i')
+_KNOWN_HEADER_GROUPS = {"referer": "referrer", "user-agent": "user_agent"}
+_DIRECTIVE_TOKEN_RE = re.compile(r"%>?\{[^}]+\}i|%>?[a-zA-Z]")
+
+
+@dataclass(frozen=True)
+class CompiledFormat:
+    pattern: re.Pattern
+    group_names: frozenset[str]
+
+
+def compile_format(format_string: str) -> CompiledFormat:
+    """Compile a LogFormat directive (Common/Combined preset or a pasted
+    custom vhost format) into one regex."""
+    regex_parts: list[str] = []
+    group_names: set[str] = set()
+    pos = 0
+    for match in _DIRECTIVE_TOKEN_RE.finditer(format_string):
+        literal = format_string[pos:match.start()]
+        if literal:
+            regex_parts.append(re.escape(literal))
+        regex_parts.append(_directive_to_pattern(match.group(0), group_names))
+        pos = match.end()
+    regex_parts.append(re.escape(format_string[pos:]))
+
+    pattern = re.compile("^" + "".join(regex_parts) + r"\s*$")
+    return CompiledFormat(pattern=pattern, group_names=frozenset(group_names))
+
+
+def _directive_to_pattern(token: str, group_names: set[str]) -> str:
+    header_match = _HEADER_DIRECTIVE_RE.fullmatch(token)
+    if header_match:
+        header_name = header_match.group(1).lower()
+        group = _KNOWN_HEADER_GROUPS.get(header_name, re.sub(r"[^a-z0-9]+", "_", header_name))
+        group_names.add(group)
+        return f'(?P<{group}>[^"]*)'
+
+    if token not in _SIMPLE_DIRECTIVES:
+        raise ValueError(f"Unsupported LogFormat directive: {token!r}")
+    group, pattern = _SIMPLE_DIRECTIVES[token]
+    group_names.add(group)
+    return pattern
+
+
+def parse_apache_timestamp(raw: str) -> datetime:
+    """'10/Oct/2026:13:55:36 -0700' -> naive UTC datetime (Ch07: store normalized to UTC)."""
+    dt = datetime.strptime(raw, "%d/%b/%Y:%H:%M:%S %z")
+    return dt.astimezone(timezone.utc).replace(tzinfo=None)
+
+
+def parse_request_line(raw: str) -> tuple[str, str] | None:
+    """Split '%r' ("GET /path HTTP/1.1") into (method, path); None if malformed."""
+    parts = raw.split()
+    if len(parts) != 3:
+        return None
+    method, path, _protocol = parts
+    return method, path
+
+
+def validate_ip(raw: str) -> str | None:
+    """Return `raw` if a valid IPv4/IPv6 address, else None (Ch07)."""
+    try:
+        ip_address(raw)
+    except ValueError:
+        return None
+    return raw

app/services/log_parser/litespeed.py

@@ -0,0 +1,10 @@
+"""LiteSpeed uses the same LogFormat directives as Apache (Chapter 07):
+Combined generally works unmodified. get_parser_for() still always
+compiles the vhost's real format_string though — this module is just the
+named default, never a silent override of a custom format.
+"""
+from __future__ import annotations
+
+from app.services.log_parser.apache import combined_parser as default_litespeed_parser
+
+__all__ = ["default_litespeed_parser"]

app/services/log_processor.py

@@ -0,0 +1,148 @@
+"""Core log-file processing pipeline (Chapter 07), factored out of
+app/cli.py so it has exactly one implementation shared by:
+- the optional `flask process-logs` CLI command (for anyone who still
+  wants cron), and
+- the automatic background-thread trigger fired right after upload and
+  opportunistically on page load (app/services/background.py) — the
+  no-cron-required simplification.
+
+process_one_batch() keeps the same bounded-batch, checkpointed-commit
+contract as the original cron design (Ch03 rule 1/4/9): it still never
+loads a whole file into memory, still commits progress incrementally, and
+still stops after `batch_size` lines so a very large file doesn't hold one
+enormous open transaction — the only thing that changed is *who* calls it
+again for the next batch.
+"""
+from __future__ import annotations
+
+import gzip
+import itertools
+from datetime import date
+from pathlib import Path
+
+from sqlalchemy import insert
+
+from app.extensions import db
+from app.models.log_entry import LogEntry
+from app.models.log_file import LogFile
+from app.services import aggregator
+from app.services.classification import classify_entry, write_batch_side_effects
+from app.services.log_parser import ParsedEntry, get_parser_for
+from app.utils.upload_paths import upload_path_for
+
+
+def open_log_stream(path: Path):
+    """Open a log file, transparently decompressing .gz, one line at a time."""
+    if path.suffix == ".gz":
+        return gzip.open(path, mode="rt", encoding="utf-8", errors="replace")
+    return path.open("r", encoding="utf-8", errors="replace")
+
+
+def count_total_lines(path: Path) -> int:
+    """One streaming pass to count lines (bounded memory — never the whole
+    file at once, Ch03 rule 1), so the UI can show a real
+    processed/total percentage instead of an indeterminate spinner.
+
+    This is an extra sequential read of the file beyond the parse pass
+    itself. That cost is now worth paying: processing used to be silently
+    triggered by cron with no live audience watching, but now it runs
+    automatically right after upload while the admin is looking at a
+    progress bar — the UX value of a real percentage justifies the extra
+    I/O pass.
+    """
+    with open_log_stream(path) as fh:
+        return sum(1 for _ in fh)
+
+
+def process_one_batch(log_file: LogFile, batch_size: int) -> None:
+    """Parse up to `batch_size` lines from `log_file`'s current checkpoint.
+
+    Leaves status as "processing" (with progress already committed) if
+    the file isn't finished yet — the caller decides whether to invoke
+    this again: the CLI calls it once per pending file per invocation
+    (unchanged cron-tick semantics); the background thread loops it until
+    the file reaches "done"/"error".
+    """
+    path = upload_path_for(log_file)
+    if not path.exists():
+        log_file.status = "error"
+        log_file.error_message = f"Upload file missing on disk: {path}"
+        db.session.commit()
+        return
+
+    if log_file.total_lines is None:
+        # First pickup of this file — count once, not on every batch.
+        log_file.total_lines = count_total_lines(path)
+
+    log_file.status = "processing"
+    db.session.commit()
+    parser = get_parser_for(log_file)
+
+    lines_seen_this_run = 0
+    skipped_this_run = 0
+    dates_touched: set[date] = set()
+
+    with open_log_stream(path) as fh:
+        remainder = itertools.islice(fh, log_file.processed_lines, None)
+        batch_entries: list[ParsedEntry] = []
+
+        for line in remainder:
+            entry = parser.parse_line(line)
+            if entry is not None:
+                batch_entries.append(entry)
+                dates_touched.add(entry.timestamp.date())
+            else:
+                skipped_this_run += 1  # malformed line — skip, don't abort the batch (Ch07)
+
+            log_file.processed_lines += 1
+            lines_seen_this_run += 1
+
+            if len(batch_entries) >= 500:
+                _bulk_insert_entries(log_file.id, batch_entries)
+                batch_entries = []
+
+            if lines_seen_this_run >= batch_size:
+                _record_skip_count(log_file, skipped_this_run)
+                db.session.commit()  # persist checkpoint; resumable if interrupted
+                return
+
+        if batch_entries:
+            _bulk_insert_entries(log_file.id, batch_entries)
+
+    if dates_touched:
+        aggregator.compute_rollups_for_range(min(dates_touched), max(dates_touched))
+
+    _record_skip_count(log_file, skipped_this_run)
+    log_file.status = "done"
+    db.session.commit()
+
+
+def _record_skip_count(log_file: LogFile, skipped_this_run: int) -> None:
+    """Informational only — doesn't touch `status`.
+
+    STOPGAP (flagged in Ch07): log_files has no dedicated skip-counter
+    column, so this overwrites error_message with the latest run's count
+    rather than accumulating across runs.
+    """
+    if skipped_this_run:
+        log_file.error_message = f"{skipped_this_run} unparsable line(s) skipped in the most recent parse run."
+
+
+def _bulk_insert_entries(log_file_id: int, entries: list[ParsedEntry]) -> None:
+    """Bulk-insert log_entries (Ch03 rule 4), classifying is_bot/flagged
+    per line, then derive bot_hits/suspicious_events/ip_registry (Ch07).
+    """
+    if not entries:
+        return
+    payload = []
+    for e in entries:
+        classification = classify_entry(e)
+        payload.append({
+            "log_file_id": log_file_id, "timestamp": e.timestamp, "ip": e.ip,
+            "method": e.method, "path": e.path, "status_code": e.status_code,
+            "bytes_sent": e.bytes_sent, "referrer": e.referrer, "user_agent": e.user_agent,
+            "is_bot": classification.is_bot, "flagged": classification.flagged,
+        })
+    db.session.execute(insert(LogEntry.__table__), payload)
+    db.session.commit()
+    write_batch_side_effects(log_file_id, entries)

app/services/referrer.py

@@ -0,0 +1,19 @@
+"""Referrer -> domain bucketing (Chapter 08 rollup, Method A)."""
+from __future__ import annotations
+
+from urllib.parse import urlparse
+
+
+def referrer_domain(referrer: str | None) -> str | None:
+    """Bucket a raw referrer URL to its host, stripping a leading 'www.'.
+
+    Full-URL cardinality would make the daily rollup unbounded in row
+    count; domain bucketing keeps it small (Ch03's bounded-resource bias).
+    Empty/unparsable referrers return None and are excluded from the rollup.
+    """
+    if not referrer:
+        return None
+    host = urlparse(referrer).netloc.lower()
+    if not host:
+        return None
+    return host[4:] if host.startswith("www.") else host

app/services/severity_scoring.py

@@ -0,0 +1,49 @@
+"""Severity scoring (Chapter 10): a simple, transparent rank-escalation
+model — no ML, easy to explain to a non-expert user (Ch10's own requirement).
+
+Chapter 07 assigns a PROVISIONAL severity per rule type at parse time.
+This module computes an EFFECTIVE severity by escalating that base rank
+for repeated/rapid hits from the same IP — the exact rule Ch10 names.
+
+Escalation is rank-based and strictly non-decreasing: it starts at the
+base severity's rank and only ever moves up (repeat-count / hit-rate
+bonuses), clamped at "high". An earlier weighted-score-vs-fixed-threshold
+version could silently *downgrade* an isolated high-severity event (e.g.
+a single sqlmap hit) to "medium" purely because the thresholds weren't
+calibrated to the base weights — caught by running the pipeline against
+real sample data. A single dangerous event must never end up rated below
+its own base severity; only repetition/rate should push it higher.
+"""
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+_SEVERITY_RANKS = ["low", "medium", "high"]
+_RANK_BY_SEVERITY = {name: rank for rank, name in enumerate(_SEVERITY_RANKS)}
+
+_REPEAT_COUNT_HIGH = 20
+_REPEAT_COUNT_MEDIUM = 5
+_RAPID_AVG_INTERVAL_SECONDS = 60  # >1 event/minute from one IP suggests automation
+
+
+@dataclass(frozen=True)
+class SeverityInputs:
+    base_severity: str
+    ip_event_count: int
+    avg_interval_seconds: float | None  # None if this IP has < 2 events in the window
+
+
+def compute_effective_severity(inputs: SeverityInputs) -> str:
+    """Two named, inspectable escalation rules: repeat-count and hit-rate."""
+    rank = _RANK_BY_SEVERITY.get(inputs.base_severity, 0)
+
+    if inputs.ip_event_count >= _REPEAT_COUNT_HIGH:
+        rank += 2
+    elif inputs.ip_event_count >= _REPEAT_COUNT_MEDIUM:
+        rank += 1
+
+    if inputs.avg_interval_seconds is not None and inputs.avg_interval_seconds < _RAPID_AVG_INTERVAL_SECONDS:
+        rank += 1
+
+    rank = min(rank, len(_SEVERITY_RANKS) - 1)
+    return _SEVERITY_RANKS[rank]

app/services/threat_scanner.py

@@ -0,0 +1,54 @@
+"""Suspicious-pattern matching (Chapter 07), reused as-is by Chapter 10.
+
+Pattern dictionary is data (JSON), same reasoning as the bot signature
+table — editable without a code change. Severity here is a provisional
+per-rule-type default; Chapter 10 owns the full weighted/escalating score.
+"""
+from __future__ import annotations
+
+import json
+from dataclasses import dataclass
+from pathlib import Path
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from app.services.log_parser import ParsedEntry
+
+PATTERNS_PATH = Path(__file__).parent / "data" / "threat_patterns.json"
+
+
+@dataclass(frozen=True)
+class ThreatMatch:
+    rule_matched: str
+    severity: str  # low|medium|high — provisional; Ch10 may escalate
+
+
+def _load_patterns() -> dict:
+    return json.loads(PATTERNS_PATH.read_text())
+
+
+_PATTERNS = _load_patterns()
+
+
+def scan(entry: "ParsedEntry") -> ThreatMatch | None:
+    """Check one parsed entry against the pattern dictionary.
+
+    Cheap substring checks only — no regex backtracking risk, no network
+    calls (Ch03: CPU-cheap, explainable rule matching).
+    """
+    path_lower = entry.path.lower()
+    ua_lower = (entry.user_agent or "").lower()
+
+    for scanner_ua in _PATTERNS["scanner_user_agents"]:
+        if scanner_ua in ua_lower:
+            return ThreatMatch(rule_matched=f"scanner_ua:{scanner_ua}", severity="high")
+
+    for sensitive in _PATTERNS["sensitive_paths"]:
+        if sensitive.lower() in path_lower:
+            return ThreatMatch(rule_matched=f"sensitive_path:{sensitive}", severity="medium")
+
+    for marker in _PATTERNS["injection_markers"]:
+        if marker.lower() in path_lower:
+            return ThreatMatch(rule_matched=f"injection:{marker}", severity="high")
+
+    return None

app/services/ua_classifier.py

@@ -0,0 +1,54 @@
+"""Lightweight browser/OS classification for human traffic (Chapter 08).
+
+Separate from Chapter 07's bot_identifier — that identifies automated
+crawlers via UA substring + DNS verification. This classifies ordinary
+browsers/OSes, same ordered-substring-match pattern, signatures as JSON
+data (not hardcoded), no third-party UA-parsing dependency — consistent
+with the low-footprint bias in Chapter 03.
+
+Order matters within each signature list: e.g. Edge/Opera must be checked
+before Chrome (their UAs also contain "Chrome/"); iOS before macOS (iPhone
+UAs also contain "like Mac OS X"); Android before Linux (Android UAs also
+contain "Linux").
+"""
+from __future__ import annotations
+
+import json
+from dataclasses import dataclass
+from pathlib import Path
+
+_DATA_DIR = Path(__file__).parent / "data"
+
+
+@dataclass(frozen=True)
+class UASignature:
+    name: str
+    ua_substrings: tuple[str, ...]
+
+
+def _load(filename: str) -> list[UASignature]:
+    raw = json.loads((_DATA_DIR / filename).read_text())
+    return [UASignature(name=e["name"], ua_substrings=tuple(e["ua_substrings"])) for e in raw]
+
+
+_BROWSER_SIGNATURES = _load("browser_signatures.json")
+_OS_SIGNATURES = _load("os_signatures.json")
+
+
+def _match(user_agent: str, signatures: list[UASignature], default: str) -> str:
+    for sig in signatures:
+        if any(sub in user_agent for sub in sig.ua_substrings):
+            return sig.name
+    return default
+
+
+def classify_browser(user_agent: str | None) -> str:
+    if not user_agent:
+        return "Unknown"
+    return _match(user_agent, _BROWSER_SIGNATURES, "Other")
+
+
+def classify_os(user_agent: str | None) -> str:
+    if not user_agent:
+        return "Unknown"
+    return _match(user_agent, _OS_SIGNATURES, "Other")

app/static/src/css/main.css

@@ -0,0 +1,67 @@
+/* app/static/src/css/main.css — Tailwind entry point (Chapter 05) */
+
+/* Self-hosted fonts (Ch05: build-time only, no runtime CDN dependency).
+   Display face used sparingly for headings/labels; body face for UI
+   chrome; mono face for every data readout (IPs, counts, paths,
+   timestamps) — reinforcing that this app's whole job is turning raw
+   log lines into a readable instrument panel. */
+@import "@fontsource/space-grotesk/500.css";
+@import "@fontsource/space-grotesk/700.css";
+@import "@fontsource/public-sans/400.css";
+@import "@fontsource/public-sans/500.css";
+@import "@fontsource/public-sans/600.css";
+@import "@fontsource/jetbrains-mono/400.css";
+@import "@fontsource/jetbrains-mono/500.css";
+
+@import "gridjs/dist/theme/mermaid.css";
+
+@tailwind base;
+@tailwind components;
+@tailwind utilities;
+
+@layer base {
+  html {
+    color-scheme: light;
+  }
+  html.dark {
+    color-scheme: dark;
+  }
+  body {
+    font-family: theme('fontFamily.sans');
+  }
+  h1, h2, h3, h4, .font-display {
+    font-family: theme('fontFamily.display');
+  }
+  /* Every number/IP/path/timestamp reads like it came straight off the
+     log line — the app's one signature typographic choice. */
+  .font-data {
+    font-family: theme('fontFamily.mono');
+    font-variant-numeric: tabular-nums;
+  }
+}
+
+@layer components {
+  /* Grid.js theme override to match the token system in both modes,
+     without forking the whole "mermaid" stylesheet. */
+  .gridjs-wrapper, .gridjs-container {
+    @apply !border-line dark:!border-line-dark !rounded-lg;
+  }
+  .gridjs-table {
+    @apply font-data text-sm;
+  }
+  .gridjs-th {
+    @apply !bg-surface-raised dark:!bg-surface-raised-dark !text-muted dark:!text-muted-dark font-sans !font-medium !text-xs !uppercase !tracking-wide;
+  }
+  .gridjs-td {
+    @apply !bg-surface dark:!bg-surface-dark !text-ink dark:!text-ink-dark !border-line dark:!border-line-dark;
+  }
+  .gridjs-footer {
+    @apply !bg-surface dark:!bg-surface-dark !border-line dark:!border-line-dark;
+  }
+  .gridjs-pagination .gridjs-pages button {
+    @apply !text-ink dark:!text-ink-dark;
+  }
+  .gridjs-search-input {
+    @apply !bg-surface dark:!bg-surface-dark !text-ink dark:!text-ink-dark !border-line dark:!border-line-dark;
+  }
+}

app/static/src/js/charts.js

@@ -0,0 +1,62 @@
+/**
+ * Shared Chart.js setup (Chapter 05) — avoids duplicating config per chart.
+ * Callers must pass pre-aggregated series only (≤~200 points, per Ch05);
+ * this file never fetches or aggregates raw rows itself.
+ *
+ * Dark-mode aware: reads the current theme at init time and colors grid
+ * lines/ticks/legend accordingly. Charts drawn before a theme toggle
+ * don't live-repaint — they pick up the new theme on their next redraw
+ * (date-range change, tab revisit), which keeps this simple.
+ */
+import { Chart, registerables } from 'chart.js';
+
+Chart.register(...registerables);
+
+// Shared categorical palette, drawn from the Kavosh token system rather
+// than Chart.js's stock colors.
+export const CHART_PALETTE = ['#0E7C86', '#B8860B', '#C4372F', '#3F7D5C', '#5B6663', '#8A6FBE'];
+
+function isDarkMode() {
+  return document.documentElement.classList.contains('dark');
+}
+
+function defaultOptions() {
+  const dark = isDarkMode();
+  const textColor = dark ? '#96A19D' : '#5B6663';
+  const gridColor = dark ? '#2C3336' : '#E1E5E2';
+  const tickFont = { family: '"JetBrains Mono", monospace', size: 11 };
+
+  return {
+    responsive: true,
+    maintainAspectRatio: false,
+    animation: { duration: 200 },
+    plugins: {
+      legend: {
+        display: true,
+        position: 'bottom',
+        labels: { color: textColor, font: { family: '"Public Sans", sans-serif', size: 12 }, boxWidth: 12, padding: 12 },
+      },
+    },
+    scales: {
+      x: { ticks: { color: textColor, font: tickFont }, grid: { color: gridColor } },
+      y: { ticks: { color: textColor, font: tickFont }, grid: { color: gridColor } },
+    },
+  };
+}
+
+/** @param {string} canvasId  @param {object} config  @returns {Chart} */
+export function initChart(canvasId, config) {
+  const ctx = document.getElementById(canvasId);
+  if (!ctx) throw new Error(`initChart: no element #${canvasId}`);
+  const base = defaultOptions();
+  const callerOptions = config.options ?? {};
+  return new Chart(ctx, {
+    ...config,
+    options: {
+      ...base,
+      ...callerOptions,
+      scales: config.type === 'doughnut' || config.type === 'pie' ? undefined : { ...base.scales, ...callerOptions.scales },
+      plugins: { ...base.plugins, ...callerOptions.plugins },
+    },
+  });
+}

app/static/src/js/grids.js

@@ -0,0 +1,38 @@
+/**
+ * Shared Grid.js setup (Chapter 05) — every table is server-side
+ * paginated/sorted/searched against `?page=&per_page=&sort=&q=`.
+ */
+import { Grid } from 'gridjs';
+
+/**
+ * @param {string} elementId   container element id
+ * @param {string} endpoint    JSON endpoint, e.g. '/api/overview/top-urls'
+ * @param {Array}  columns     Grid.js column defs
+ * @param {object} extraParams static extra query params (e.g. {from, to})
+ */
+export function initGrid(elementId, endpoint, columns, extraParams = {}) {
+  const buildUrl = (prev, params) => {
+    const url = new URL(prev, window.location.origin);
+    Object.entries({ ...extraParams, ...params }).forEach(([k, v]) => url.searchParams.set(k, v));
+    return url.toString();
+  };
+
+  return new Grid({
+    columns,
+    server: {
+      url: endpoint,
+      then: (res) => res.data.rows,
+      total: (res) => res.data.total,
+    },
+    pagination: {
+      server: { url: (prev, page, per_page) => buildUrl(prev, { page: page + 1, per_page }) },
+      limit: 20,
+    },
+    sort: {
+      server: {
+        url: (prev, cols) => (cols.length ? buildUrl(prev, { sort: cols[0].direction === 1 ? cols[0].index : `-${cols[0].index}` }) : prev),
+      },
+    },
+    search: { server: { url: (prev, keyword) => buildUrl(prev, { q: keyword }) } },
+  }).render(document.getElementById(elementId));
+}

app/static/src/js/main.js

@@ -0,0 +1,36 @@
+/**
+ * Vite entry point: HTMX + Alpine init + Chart.js/Grid.js glue (Chapter 05).
+ * Bundled at build time — no CDN scripts, nothing compiled at request time.
+ */
+import htmx from 'htmx.org';
+import Alpine from 'alpinejs';
+import { html as gridHtml } from 'gridjs';
+
+import { initChart, CHART_PALETTE } from './charts.js';
+import { initGrid } from './grids.js';
+import { initUploadForm } from './uploads.js';
+
+window.htmx = htmx;
+window.Alpine = Alpine;
+window.initChart = initChart;
+window.initGrid = initGrid;
+window.gridHtml = gridHtml;
+window.initUploadForm = initUploadForm;
+window.KAVOSH_CHART_PALETTE = CHART_PALETTE;
+
+// Chapter 12: HTMX requests must carry the CSRF token as a header, not
+// rely on the cookie alone.
+document.body.addEventListener('htmx:configRequest', (event) => {
+  const token = document.querySelector('meta[name="csrf-token"]')?.content;
+  if (token) event.detail.headers['X-CSRFToken'] = token;
+});
+
+// Dark mode toggle (persisted; the initial class is set synchronously in
+// an inline <head> script in base.html to avoid a flash of the wrong
+// theme before this bundle loads).
+window.toggleTheme = function toggleTheme() {
+  const isDark = document.documentElement.classList.toggle('dark');
+  localStorage.setItem('kavosh-theme', isDark ? 'dark' : 'light');
+};
+
+Alpine.start();

app/static/src/js/uploads.js

@@ -0,0 +1,62 @@
+/**
+ * Upload with real byte-transfer progress (Chapter 08/12).
+ *
+ * fetch() can't reliably report upload progress across browsers, so this
+ * uses XMLHttpRequest directly rather than depending on htmx's internal
+ * transport. On completion, the returned HTML fragment (queued/duplicate/
+ * error) is injected manually, then handed to htmx.process() so any
+ * hx-* polling attributes inside it (e.g. the processing-status poll)
+ * activate normally, exactly as if htmx had performed the swap itself.
+ */
+export function initUploadForm(formId) {
+  const form = document.getElementById(formId);
+  if (!form) return;
+
+  const fileInput = form.querySelector('input[type="file"]');
+  const progressWrap = document.getElementById('upload-progress-wrap');
+  const progressBar = document.getElementById('upload-progress-bar');
+  const progressLabel = document.getElementById('upload-progress-label');
+  const resultEl = document.getElementById('upload-result');
+  const submitBtn = form.querySelector('button[type="submit"]');
+
+  form.addEventListener('submit', (event) => {
+    event.preventDefault();
+    if (!fileInput.files.length) return;
+
+    const formData = new FormData(form);
+    const xhr = new XMLHttpRequest();
+
+    submitBtn.disabled = true;
+    resultEl.innerHTML = '';
+    progressWrap.classList.remove('hidden');
+    progressBar.style.width = '0%';
+    progressLabel.textContent = 'Uploading… 0%';
+
+    xhr.upload.addEventListener('progress', (e) => {
+      if (!e.lengthComputable) return;
+      const pct = Math.round((e.loaded / e.total) * 100);
+      progressBar.style.width = pct + '%';
+      progressLabel.textContent = `Uploading… ${pct}%`;
+    });
+
+    xhr.addEventListener('load', () => {
+      submitBtn.disabled = false;
+      progressWrap.classList.add('hidden');
+      resultEl.innerHTML = xhr.responseText;
+      if (window.htmx) window.htmx.process(resultEl);
+      form.reset();
+    });
+
+    xhr.addEventListener('error', () => {
+      submitBtn.disabled = false;
+      progressWrap.classList.add('hidden');
+      resultEl.innerHTML =
+        '<p class="text-danger dark:text-danger-dark text-sm">Upload failed — check your connection and try again.</p>';
+    });
+
+    xhr.open('POST', form.getAttribute('action') || '/uploads');
+    const token = document.querySelector('meta[name="csrf-token"]')?.content;
+    if (token) xhr.setRequestHeader('X-CSRFToken', token);
+    xhr.send(formData);
+  });
+}

app/templates/base.html

@@ -0,0 +1,85 @@
+<!doctype html>
+<html lang="en" class="h-full">
+<head>
+  <meta charset="utf-8">
+  <meta name="viewport" content="width=device-width, initial-scale=1">
+  <meta name="csrf-token" content="{{ csrf_token() }}">
+  <title>{% block title %}Kavosh{% endblock %}</title>
+  <!--
+    Anti-FOUC: set the theme class synchronously, before the stylesheet
+    and before Alpine/htmx load, so there's no flash of the wrong theme
+    on first paint. Kept as a tiny inline script rather than waiting for
+    the bundled JS, which only arrives after the page has started
+    rendering.
+  -->
+  <script>
+    (function () {
+      var stored = localStorage.getItem('kavosh-theme');
+      var isDark = stored ? stored === 'dark' : window.matchMedia('(prefers-color-scheme: dark)').matches;
+      document.documentElement.classList.toggle('dark', isDark);
+    })();
+  </script>
+  <link rel="stylesheet" href="{{ asset('app/static/src/css/main.css') }}">
+</head>
+<body class="h-full bg-paper dark:bg-paper-dark text-ink dark:text-ink-dark antialiased">
+  <div id="app-shell" x-data="{ mobileNavOpen: false }">
+    <header class="border-b border-line dark:border-line-dark bg-surface dark:bg-surface-dark">
+      <nav class="flex gap-1 px-4 py-3 items-center max-w-6xl mx-auto">
+        <span class="font-display font-bold text-lg tracking-tight mr-4">Kavosh</span>
+
+        {% if current_user.is_authenticated %}
+        {% set tabs = [
+          ('overview.overview', '/overview', 'house', 'Overview'),
+          ('seo.seo', '/seo', 'search', 'SEO & Bots'),
+          ('security.security', '/security', 'shield-alert', 'Security'),
+        ] %}
+        {% for endpoint, path, icon, label in tabs %}
+        <a href="{{ path }}" hx-get="{{ path }}" hx-target="#main-content" hx-push-url="true"
+           class="flex items-center gap-1.5 px-3 py-1.5 rounded-md text-sm font-medium transition-colors
+                  {% if request.endpoint == endpoint %}
+                    bg-accent/10 text-accent dark:bg-accent-dark/15 dark:text-accent-dark
+                  {% else %}
+                    text-muted dark:text-muted-dark hover:text-ink dark:hover:text-ink-dark hover:bg-paper dark:hover:bg-paper-dark
+                  {% endif %}">
+          <svg class="w-4 h-4 shrink-0"><use href="/static/dist/icons.svg#{{ icon }}"/></svg>
+          <span class="hidden sm:inline">{{ label }}</span>
+        </a>
+        {% endfor %}
+
+        <div class="ml-auto flex items-center gap-2">
+          <button type="button" onclick="toggleTheme()"
+                  aria-label="Toggle dark mode"
+                  class="p-2 rounded-md text-muted dark:text-muted-dark hover:text-ink dark:hover:text-ink-dark hover:bg-paper dark:hover:bg-paper-dark transition-colors">
+            <svg class="w-4 h-4 block dark:hidden"><use href="/static/dist/icons.svg#moon"/></svg>
+            <svg class="w-4 h-4 hidden dark:block"><use href="/static/dist/icons.svg#sun"/></svg>
+          </button>
+          <span class="text-sm text-muted dark:text-muted-dark font-data hidden md:inline">{{ current_user.email }}</span>
+          <form method="post" action="{{ url_for('auth.logout') }}">
+            <input type="hidden" name="csrf_token" value="{{ csrf_token() }}">
+            <button type="submit" title="Log out"
+                    class="p-2 rounded-md text-muted dark:text-muted-dark hover:text-danger dark:hover:text-danger-dark hover:bg-paper dark:hover:bg-paper-dark transition-colors">
+              <svg class="w-4 h-4"><use href="/static/dist/icons.svg#log-out"/></svg>
+            </button>
+          </form>
+        </div>
+        {% else %}
+        <div class="ml-auto">
+          <button type="button" onclick="toggleTheme()" aria-label="Toggle dark mode"
+                  class="p-2 rounded-md text-muted dark:text-muted-dark hover:text-ink dark:hover:text-ink-dark hover:bg-paper dark:hover:bg-paper-dark transition-colors">
+            <svg class="w-4 h-4 block dark:hidden"><use href="/static/dist/icons.svg#moon"/></svg>
+            <svg class="w-4 h-4 hidden dark:block"><use href="/static/dist/icons.svg#sun"/></svg>
+          </button>
+        </div>
+        {% endif %}
+      </nav>
+    </header>
+
+    <!-- HTMX target swapped by the nav above (Ch01: single-page shell, no full reloads) -->
+    <main id="main-content" class="max-w-6xl mx-auto p-4 sm:p-6">
+      {% block content %}{% endblock %}
+    </main>
+  </div>
+
+  <script type="module" src="{{ asset('app/static/src/js/main.js') }}"></script>
+</body>
+</html>

app/utils/assets.py

@@ -0,0 +1,39 @@
+"""Vite manifest reader — maps a source entry to its hashed dist path (Ch05)."""
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+from flask import Flask
+
+
+class ManifestNotFound(RuntimeError):
+    """app/static/dist/manifest.json is missing — `npm run build` hasn't been run."""
+
+
+def _load_manifest(static_dist_dir: Path) -> dict:
+    for candidate in (static_dist_dir / ".vite" / "manifest.json", static_dist_dir / "manifest.json"):
+        if candidate.exists():
+            return json.loads(candidate.read_text())
+    raise ManifestNotFound(f"No Vite manifest under {static_dist_dir} — run `npm run build`.")
+
+
+def register_asset_helper(app: Flask) -> None:
+    """Register `asset(name)` as a Jinja global (Chapter 05)."""
+    static_dist_dir = Path(app.static_folder) / "dist"
+
+    @app.context_processor
+    def inject_asset_helper():
+        def asset(name: str) -> str:
+            try:
+                manifest = _load_manifest(static_dist_dir)
+            except ManifestNotFound:
+                if app.debug:
+                    return f"/static/dist/{name}"  # tolerate an unbuilt dev checkout
+                raise
+            entry = manifest.get(name)
+            if entry is None:
+                raise KeyError(f"{name!r} not in Vite manifest.")
+            return f"/static/dist/{entry['file']}"
+
+        return {"asset": asset}

app/utils/dates.py

@@ -0,0 +1,32 @@
+"""Shared from/to date-range parsing (Chapter 11: consistent across every
+chart/KPI endpoint; never defaults to "all time" per Chapter 03 rule 7).
+"""
+from __future__ import annotations
+
+from datetime import date, datetime, timedelta
+
+from flask import Request
+
+DEFAULT_RANGE_DAYS = 7  # within Ch11's stated "7 or 30 day" default allowance
+
+
+def parse_date_range(request: Request) -> tuple[date, date]:
+    """Parse `?from=&to=` (ISO 8601 dates), defaulting to the last 7 days."""
+    to_raw = request.args.get("to")
+    from_raw = request.args.get("from")
+    to_date = date.fromisoformat(to_raw) if to_raw else date.today()
+    from_date = date.fromisoformat(from_raw) if from_raw else to_date - timedelta(days=DEFAULT_RANGE_DAYS - 1)
+    if from_date > to_date:
+        from_date, to_date = to_date, from_date
+    return from_date, to_date
+
+
+def day_bounds(from_date: date, to_date: date) -> tuple[datetime, datetime]:
+    """Inclusive [from_date, to_date] -> half-open [start, end) datetime range.
+
+    Shared by overview/queries.py, seo/queries.py, and security/queries.py
+    (consolidated here rather than each keeping its own local copy).
+    """
+    start = datetime.combine(from_date, datetime.min.time())
+    end = datetime.combine(to_date, datetime.min.time()) + timedelta(days=1)
+    return start, end

app/utils/envelope.py

@@ -0,0 +1,18 @@
+"""Shared JSON envelope helpers (Chapter 11).
+
+Every existing /api/... endpoint already hand-builds this exact shape via
+jsonify(data=..., meta=...) — audited for compliance in docs/api-contract-
+final.md. This module exists so NEW endpoints (e.g. this chapter's
+unauthorized_handler) have one call site instead of re-typing the shape.
+"""
+from __future__ import annotations
+
+from flask import jsonify
+
+
+def api_ok(data, meta: dict | None = None):
+    return jsonify(data=data, meta=meta or {})
+
+
+def api_error(code: str, message: str, status: int):
+    return jsonify(error={"code": code, "message": message}), status

app/utils/htmx.py

@@ -0,0 +1,20 @@
+"""Shared HTMX full-page-vs-fragment response helper (Chapter 04).
+
+Every blueprint serving a dashboard tab (Ch08/09/10) uses this instead of
+duplicating the `if request.headers.get("HX-Request")` check, so the
+convention is identical everywhere.
+"""
+from __future__ import annotations
+
+from flask import Request, render_template
+
+
+def is_htmx(request: Request) -> bool:
+    """True if this request was triggered by an hx-* attribute."""
+    return request.headers.get("HX-Request", "").lower() == "true"
+
+
+def render_htmx_aware(request: Request, *, full_template: str, partial_template: str, **ctx):
+    """Render `full_template` on first load, `partial_template` on HTMX swaps."""
+    template = partial_template if is_htmx(request) else full_template
+    return render_template(template, **ctx)

app/utils/http_status.py

@@ -0,0 +1,14 @@
+"""Shared HTTP status-code bucketing — used by the aggregator (per-IP
+rollup) and available for any endpoint needing the same 2xx/3xx/4xx/5xx
+buckets, so there's one bucketing rule, not several copies."""
+from __future__ import annotations
+
+
+def status_bucket(status_code: int) -> str:
+    if status_code < 300:
+        return "2xx"
+    if status_code < 400:
+        return "3xx"
+    if status_code < 500:
+        return "4xx"
+    return "5xx"

app/utils/pagination.py

@@ -0,0 +1,14 @@
+"""Shared page/per_page parsing (Chapter 11: consistent across every
+Grid.js-backed endpoint)."""
+from __future__ import annotations
+
+from flask import Request
+
+DEFAULT_PER_PAGE = 20
+MAX_PER_PAGE = 100
+
+
+def parse_pagination(request: Request) -> tuple[int, int]:
+    page = max(1, request.args.get("page", 1, type=int))
+    per_page = request.args.get("per_page", DEFAULT_PER_PAGE, type=int)
+    return page, max(1, min(per_page, MAX_PER_PAGE))

app/utils/upload_paths.py

@@ -0,0 +1,20 @@
+"""Shared upload-path helper (Chapter 04) — used by the uploads blueprint,
+the optional CLI processor, and the automatic background-thread processor,
+so there's one definition of where a given LogFile's raw upload lives on
+disk. Pulled out of the uploads blueprint so services/ doesn't have to
+import from blueprints/ (the wrong direction) to reach it.
+"""
+from __future__ import annotations
+
+from pathlib import Path
+
+from flask import current_app
+
+from app.models.log_file import LogFile
+
+
+def upload_path_for(log_file: LogFile) -> Path:
+    """Deterministic on-disk path for a LogFile row — avoids a new DB column."""
+    ext = Path(log_file.filename).suffix.lower()
+    upload_dir = Path(current_app.config["UPLOAD_DIR"])
+    return upload_dir / f"{log_file.id}{ext}"