mirror of
https://github.com/superdesigndev/treg.git
synced 2026-10-02 03:24:35 +08:00
SToneX
e0c20961bb
/media was a plain StaticFiles mount with no Cache-Control, unlike /logos and /vendor. Its names are unversioned, and without the header a browser applies a heuristic lifetime and never revalidates, so an edited landing script could keep running against new HTML. Scripts, stylesheets and text now answer no-cache (the existing ETag makes that a 304); images, video and fonts, which a page only swaps by renaming, keep a day's cache like the logos.
97 lines
4.4 KiB
Python
97 lines
4.4 KiB
Python
"""The hand-built pages under src/treg/web: what they ship, how they are cached, and the copy rule
|
|
that the catalog's size is read from the catalog, never typed."""
|
|
|
|
from __future__ import annotations
|
|
|
|
import re
|
|
from pathlib import Path
|
|
from urllib.parse import urlsplit
|
|
|
|
import pytest
|
|
from httpx import AsyncClient
|
|
|
|
from treg.domain.catalog import store as catalog_store
|
|
|
|
WEB = Path(__file__).resolve().parents[1] / "src" / "treg" / "web"
|
|
ROOT = WEB.parents[2]
|
|
|
|
|
|
async def test_grokbot_icons_are_served_files(clients: AsyncClient):
|
|
"""The page used to inline every icon as base64 (half its weight, re-sent on every visit).
|
|
Each image it names now has to resolve, or the icon silently disappears."""
|
|
page = await clients.get("/grokbot")
|
|
assert page.status_code == 200
|
|
assert "data:image" not in page.text
|
|
srcs = set(re.findall(r'<img[^>]*\ssrc="(/[^"]+)"', page.text))
|
|
assert len(srcs) > 20
|
|
for src in sorted(srcs):
|
|
r = await clients.get(urlsplit(src).path)
|
|
assert r.status_code == 200, src
|
|
assert r.headers["content-type"].startswith("image/"), src
|
|
|
|
|
|
async def test_static_pages_fill_the_catalog_size(clients: AsyncClient):
|
|
endpoints, providers = catalog_store.headline_counts(catalog_store.load())
|
|
for path in ("/grokbot", "/fable", "/people-search", "/ugc", "/jev", "/gpt6", "/resources"):
|
|
text = (await clients.get(path)).text
|
|
assert f"{endpoints} " in text, path
|
|
assert f"{providers} providers" in (await clients.get("/grokbot")).text
|
|
|
|
|
|
async def test_static_page_revalidates_to_a_304(clients: AsyncClient):
|
|
"""Filling placeholders turned these pages from FileResponses into rendered HTML; the ETag
|
|
keeps a browser's `no-cache` revalidation cheap, including an edge-weakened tag."""
|
|
first = await clients.get("/grokbot")
|
|
etag = first.headers["etag"]
|
|
assert first.headers["cache-control"] == "no-cache"
|
|
for sent in (etag, f"W/{etag}", f'"other", {etag}'):
|
|
again = await clients.get("/grokbot", headers={"If-None-Match": sent})
|
|
assert again.status_code == 304, sent
|
|
assert again.content == b""
|
|
assert (await clients.get("/grokbot", headers={"If-None-Match": '"other"'})).status_code == 200
|
|
|
|
|
|
@pytest.mark.parametrize(("path", "policy"), [
|
|
("/media/landing/gateway.js", "no-cache"),
|
|
("/media/landing/gateway.css", "no-cache"),
|
|
("/media/grokbot/grokbot.png", "public, max-age=86400"),
|
|
("/media/og.png", "public, max-age=86400"),
|
|
])
|
|
async def test_media_cache_policy(clients: AsyncClient, path: str, policy: str):
|
|
"""Unversioned page media: code revalidates (a heuristic lifetime would pair stale scripts
|
|
with new HTML), images keep a day like /logos."""
|
|
r = await clients.get(path)
|
|
assert r.status_code == 200, path
|
|
assert r.headers["cache-control"] == policy
|
|
if policy == "no-cache":
|
|
again = await clients.get(path, headers={"If-None-Match": r.headers["etag"]})
|
|
assert again.status_code == 304
|
|
|
|
|
|
# A typed catalog size: "2,896 tools", "2,600+ data endpoints", "+ 2,867 more</b> tools", and any
|
|
# whole-catalog provider count ("60 providers", "60+ providers", "~80 providers"). Per-job figures
|
|
# ("3 providers, ranked", "16 providers") are facts about one capability and stay allowed.
|
|
_TYPED_ENDPOINTS = re.compile(r"\b\d{1,2},\d{3}\+?\s+(?:more\s+)?(?:[\w&;]+\s+){0,3}(?:tools|endpoints)\b", re.I)
|
|
_TYPED_PROVIDERS = re.compile(r"(?:~\s*\d+|\b\d+\+|\b[3-9]\d|\b\d{3,})\s+providers\b", re.I)
|
|
|
|
|
|
def _front_door_files() -> list[Path]:
|
|
skip = {"dashboard", "vendor", "media"}
|
|
files = [p for p in WEB.rglob("*") if p.suffix in {".html", ".js", ".md", ".txt"}
|
|
and not skip & set(p.relative_to(WEB).parts)]
|
|
return files + [ROOT / "README.md", ROOT / "docs" / "TUTORIAL.md"]
|
|
|
|
|
|
def test_no_page_types_the_catalog_size():
|
|
"""AGENTS.md: the catalog changes weekly and every stale number is a lie. Six pages once
|
|
disagreed with each other (and with the catalog) on how many tools there are. A served page
|
|
says `{ENDPOINTS}` / `{PROVIDERS}` and the route fills it; copy that is not templated states
|
|
no count at all."""
|
|
hits = []
|
|
for path in _front_door_files():
|
|
for n, line in enumerate(path.read_text(encoding="utf-8").splitlines(), 1):
|
|
plain = re.sub(r"<[^>]+>", "", line)
|
|
for rx in (_TYPED_ENDPOINTS, _TYPED_PROVIDERS):
|
|
hits += [f"{path.relative_to(ROOT)}:{n}: {m.group(0)}" for m in rx.finditer(plain)]
|
|
assert not hits, "\n".join(hits)
|