Use keyset pagination for the stream (C4)

The stream did a COUNT(*) and OFFSET slice on every request, ordered by
a non-unique (published_at, fetched_at) — deep pages got slower and
articles sharing a timestamp could repeat or vanish across pages.

It now orders by (Coalesce(published_at, fetched_at), id) and pages with
before/after cursors, fetching one extra row to know whether another
page exists instead of counting. Pagination links also keep the feed
scope, which the old links silently dropped.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
This commit is contained in:
ksmithandClaude Sonnet 5.5 committed 2026-11-01 08:51:20 -07:00
1 parent 70b0d4f277
commit 8d2900424a
3 files changed
+156 -16

No files matched your search

+57 -12
View File
@@ -1,8 +1,10 @@
from datetime import datetime
from itertools import groupby
from urllib.parse import urlencode
from django.contrib.auth.decorators import login_required
from django.db.models import Max, Q
from django.db.models.functions import Coalesce
from django.http import Http404
from django.shortcuts import get_object_or_404, redirect, render
from django.urls import reverse
@@ -157,26 +159,68 @@ def _filtered_queryset(request, active_filter):
return _apply_scope(request, queryset)
def _encode_cursor(article) -> str:
return f"{article.sort_at.isoformat()},{article.pk}"
def _decode_cursor(raw):
"""(sort_at, id) from a cursor produced by _encode_cursor, or None for
a missing/garbled one (treated as "no cursor" — the first page)."""
if not raw:
return None
try:
timestamp, pk = raw.rsplit(",", 1)
return datetime.fromisoformat(timestamp), int(pk)
except ValueError:
return None
@login_required
def stream(request):
active_filter = _resolve_active_filter(request)
queryset, active_tag, active_feed = _filtered_queryset(request, active_filter)
try:
page = max(int(request.GET.get("page", 1)), 1)
except ValueError:
page = 1
start, end = (page - 1) * PAGE_SIZE, page * PAGE_SIZE
total_count = queryset.count()
articles = with_user_state(queryset[start:end], request.user)
scope_params = {"filter": active_filter}
if active_tag:
scope_params["tag"] = active_tag.pk
if active_feed:
scope_params["feed"] = active_feed.pk
# Keyset pagination (C4) over (sort_at, id) — sort_at falls back to
# fetched_at for entries a feed gave no published date, and id breaks
# ties so pages neither shift nor repeat articles that share a
# timestamp. No COUNT(*): fetching one row past the page size is
# enough to know whether another page exists in that direction.
ordered = queryset.annotate(sort_at=Coalesce("published_at", "fetched_at"))
after = _decode_cursor(request.GET.get("after"))
before = _decode_cursor(request.GET.get("before"))
if after:
sort_at, pk = after
rows = list(
ordered.filter(Q(sort_at__gt=sort_at) | Q(sort_at=sort_at, id__gt=pk)).order_by(
"sort_at", "id"
)[: PAGE_SIZE + 1]
)
has_newer_page = len(rows) > PAGE_SIZE
has_older_page = True
rows = rows[:PAGE_SIZE][::-1]
else:
if before:
sort_at, pk = before
ordered = ordered.filter(Q(sort_at__lt=sort_at) | Q(sort_at=sort_at, id__lt=pk))
rows = list(ordered.order_by("-sort_at", "-id")[: PAGE_SIZE + 1])
has_older_page = len(rows) > PAGE_SIZE
has_newer_page = bool(before)
rows = rows[:PAGE_SIZE]
articles = with_user_state(rows, request.user)
newer_qs = older_qs = ""
if articles:
newer_qs = urlencode({**scope_params, "after": _encode_cursor(articles[0])})
older_qs = urlencode({**scope_params, "before": _encode_cursor(articles[-1])})
context = {
"date_groups": _group_by_date(articles),
"active_filter": active_filter,
@@ -184,9 +228,10 @@ def stream(request):
"active_tag": active_tag,
"active_feed": active_feed,
"user_tags": Tag.objects.filter(user=request.user),
"page": page,
"has_next_page": end < total_count,
"has_previous_page": page > 1,
"has_newer_page": has_newer_page and bool(articles),
"has_older_page": has_older_page and bool(articles),
"newer_qs": newer_qs,
"older_qs": older_qs,
# Lets the "new articles" banner ask, on this exact same scope,
# whether anything with a higher id has shown up since this page
# was rendered — id (not published_at) because it's monotonic