Phase 4: anonymous page caching via AnonCachePageMixin

Add AnonCachePageMixin (sapl/middleware/page_cache.py) that stores full
view responses in the default Redis cache for anonymous (unauthenticated)
GET requests only.  Authenticated users always bypass the cache so CSRF
tokens and user-specific UI controls are never served stale.

Applied to:
- ParlamentarCrud.ListView / DetailView — TTL 600 s (changes each term)
- AudienciaCrud.ListView              — TTL 120 s (hearings added infrequently)
- ComissaoCrud.ListView               — TTL 300 s (committees change rarely)

Also:
- Add PAGE_CACHE_TTL_LIST/DETAIL/STABLE settings (env-configurable)
- Add bingbot + SERankingBacklinksBot to nginx UA blocklist (were already
  in BOT_UA_FRAGMENTS / robots.txt; nginx map was the only gap)
- Remove unused ratelimit/method_decorator/RATE_LIMITER_RATE imports from
  audiencia/views.py that crept in during Phase 2

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

docker/config/nginx/nginx.conf

@@ -96,6 +96,8 @@ http {
         "~*DotBot"                 1;
         "~*meta-externalagent"     1;
         "~*OAI-SearchBot"          1;
+        "~*bingbot"                1;
+        "~*SERankingBacklinksBot"  1;
         "~*Chrome/98\.0\.4758"     1;
     }
 

sapl/audiencia/views.py

@@ -9,11 +9,7 @@ from sapl.crud.base import RP_DETAIL, RP_LIST, Crud, MasterDetailCrud
 from .forms import AudienciaForm, AnexoAudienciaPublicaForm
 from .models import AudienciaPublica, AnexoAudienciaPublica
 
-from ratelimit.decorators import ratelimit
-from django.utils.decorators import method_decorator
-
-from ..settings import RATE_LIMITER_RATE
-from sapl.middleware.ratelimit import ratelimit_ip
+from sapl.middleware.page_cache import AnonCachePageMixin
 
 
 def index(request):
@@ -28,8 +24,9 @@ class AudienciaCrud(Crud):
         list_field_names = ['numero', 'nome', 'tipo', 'materia', 'data']
         ordering = '-ano', '-numero', '-data', 'nome', 'tipo'
 
-    class ListView(Crud.ListView):
+    class ListView(AnonCachePageMixin, Crud.ListView):
         paginate_by = 10
+        anon_cache_ttl = 120  # PAGE_CACHE_TTL_LIST — hearings are added infrequently
 
         def get_context_data(self, **kwargs):
             context = super().get_context_data(**kwargs)

sapl/comissoes/views.py

@@ -28,6 +28,7 @@ from sapl.crud.base import (Crud, CrudAux, MasterDetailCrud,
                             RP_LIST)
 from sapl.materia.models import (MateriaEmTramitacao, MateriaLegislativa,
                                  PautaReuniao, Tramitacao)
+from sapl.middleware.page_cache import AnonCachePageMixin
 from sapl.middleware.ratelimit import ratelimit_ip
 from sapl.utils import show_results_filter_set
 
@@ -174,6 +175,10 @@ class ComissaoCrud(Crud):
                             'data_criacao', 'data_extincao', 'ativa']
         ordering = '-ativa', 'sigla'
 
+    class ListView(AnonCachePageMixin, Crud.ListView):
+        # Committee lists change rarely; 5-minute cache is conservative.
+        anon_cache_ttl = 300  # PAGE_CACHE_TTL_DETAIL
+
     class CreateView(Crud.CreateView):
         form_class = ComissaoForm
 

sapl/middleware/page_cache.py

@@ -0,0 +1,75 @@
+"""
+AnonCachePageMixin — anonymous-only Django view response caching.
+
+Why anonymous-only?
+  - Authenticated responses include CSRF tokens and user-specific UI fragments
+    (edit/delete action buttons injected by SAPL's CRUD framework).  Caching
+    those would serve stale or wrong data to other users.
+  - Bot traffic is entirely anonymous.  A 2-minute cache converts hundreds of
+    identical list-view DB queries into a single one — exactly the workload
+    that triggers OOM in the fleet.
+
+How it works:
+  - `dispatch()` short-circuits to the normal (uncached) path for any
+    authenticated request.
+  - For anonymous GET/HEAD requests, the response is stored in the 'default'
+    Redis cache under a key that includes the full URL (scheme + host + path +
+    query string).  The Django cache framework handles key construction and
+    TTL expiry automatically.
+  - HTTPS and HTTP requests are stored under separate keys (Django default).
+
+Usage:
+
+    from sapl.middleware.page_cache import AnonCachePageMixin
+
+    class MyListView(AnonCachePageMixin, ListView):
+        anon_cache_ttl = settings.PAGE_CACHE_TTL_LIST   # 120 s
+
+    class MyDetailView(AnonCachePageMixin, DetailView):
+        anon_cache_ttl = settings.PAGE_CACHE_TTL_DETAIL  # 300 s
+
+TTL reference (see settings.PAGE_CACHE_TTL_*):
+
+    View type                                 Default TTL
+    ─────────────────────────────────────────────────────
+    Public list   (norma, materia, sessao…)   120 s  (PAGE_CACHE_TTL_LIST)
+    Public detail (norma, materia, sessao…)   300 s  (PAGE_CACHE_TTL_DETAIL)
+    Stable detail (parlamentar, comissão)     600 s  (PAGE_CACHE_TTL_STABLE)
+
+Invalidation:
+  The cache TTL is intentionally short (≤ 10 min) so stale content expires
+  on its own.  Explicit invalidation is not implemented — legislative data
+  changes infrequently and short TTLs are acceptable.
+"""
+
+from django.conf import settings
+from django.views.decorators.cache import cache_page, never_cache
+from django.utils.decorators import method_decorator
+
+
+class AnonCachePageMixin:
+    """
+    Cache the full view response for anonymous (unauthenticated) requests.
+
+    Set `anon_cache_ttl` on the subclass to override the default TTL.
+    Authenticated requests always bypass the cache.
+    """
+
+    # Override per view class.  Use settings.PAGE_CACHE_TTL_* for consistency.
+    anon_cache_ttl = getattr(settings, 'PAGE_CACHE_TTL_LIST', 120)
+
+    def dispatch(self, request, *args, **kwargs):
+        if getattr(request, 'user', None) and request.user.is_authenticated:
+            # Authenticated: skip cache entirely — response may contain
+            # user-specific controls (CSRF token, edit/delete buttons).
+            handler = never_cache(
+                lambda req, *a, **kw: super(AnonCachePageMixin, self).dispatch(req, *a, **kw)
+            )
+            return handler(request, *args, **kwargs)
+
+        # Anonymous: wrap the parent dispatch in cache_page so Django stores
+        # the rendered response in the 'default' cache for anon_cache_ttl seconds.
+        handler = cache_page(self.anon_cache_ttl)(
+            lambda req, *a, **kw: super(AnonCachePageMixin, self).dispatch(req, *a, **kw)
+        )
+        return handler(request, *args, **kwargs)

sapl/parlamentares/views.py

@@ -33,6 +33,7 @@ from sapl.materia.models import Autoria, Proposicao, Relatoria
 from sapl.norma.models import AutoriaNorma, NormaJuridica
 from sapl.parlamentares.apps import AppConfig
 from sapl.rules import SAPL_GROUP_VOTANTE
+from sapl.middleware.page_cache import AnonCachePageMixin
 from sapl.middleware.ratelimit import ratelimit_ip
 from sapl.utils import (parlamentares_ativos, show_results_filter_set)
 
@@ -750,7 +751,9 @@ class ParlamentarCrud(Crud):
             'filiacao_atual',
             'ativo']
 
-    class DetailView(Crud.DetailView):
+    class DetailView(AnonCachePageMixin, Crud.DetailView):
+        # Parlamentar profiles change only at term boundaries — 10-minute cache.
+        anon_cache_ttl = 600  # PAGE_CACHE_TTL_STABLE
 
         def get_template_names(self):
             if self.request.user.has_perm(self.permission(RP_CHANGE)):
@@ -789,10 +792,12 @@ class ParlamentarCrud(Crud):
             """
             return super(Crud.CreateView, self).form_valid(form)
 
-    class ListView(Crud.ListView):
+    class ListView(AnonCachePageMixin, Crud.ListView):
         template_name = "parlamentares/parlamentares_list.html"
         paginate_by = None
         logger = logging.getLogger(__name__)
+        # Full list changes only when a mandato starts/ends — 10-minute cache.
+        anon_cache_ttl = 600  # PAGE_CACHE_TTL_STABLE
 
         @xframe_options_exempt
         def get(self, request, *args, **kwargs):

sapl/settings.py

@@ -404,6 +404,18 @@ RATE_LIMIT_WHITELIST_IPS = config(
     cast=lambda v: [x.strip() for x in v.split(',') if x.strip()],
 )
 
+# ---------------------------------------------------------------------------
+# Anonymous page caching — AnonCachePageMixin (sapl/middleware/page_cache.py)
+# TTLs apply only to anonymous (unauthenticated) GET responses.
+# Authenticated users always bypass the cache (see AnonCachePageMixin).
+# ---------------------------------------------------------------------------
+# Public list views (norma, materia, sessao, parlamentares…)
+PAGE_CACHE_TTL_LIST = config('PAGE_CACHE_TTL_LIST', default=120, cast=int)
+# Public detail views — rarely mutated once published
+PAGE_CACHE_TTL_DETAIL = config('PAGE_CACHE_TTL_DETAIL', default=300, cast=int)
+# High-stability detail views (parlamentar, comissão) — change only each term
+PAGE_CACHE_TTL_STABLE = config('PAGE_CACHE_TTL_STABLE', default=600, cast=int)
+
 # Internationalization
 # https://docs.djangoproject.com/en/1.8/topics/i18n/
 LANGUAGE_CODE = 'pt-br'