fix(subscribestar): initial feed GET is a navigation, not XHR (first-run drift)
CI / lint (push) Successful in 2s
CI / frontend-build (push) Successful in 18s
CI / backend-lint-and-test (push) Successful in 33s
CI / integration (push) Successful in 3m14s

First live run (cheunart) tripped the drift guard: "no posts and no recognizable
feed container". The browser-saved page was normal (6 posts + posts_container-list),
so the parser was fine — our live HTTP fetch got a different response. Cause: the
client set X-Requested-With: XMLHttpRequest (+ a JSON Accept) session-wide, so the
initial creator-page GET was sent as an XHR. SubscribeStar (Rails) content-
negotiates an XHR full-page request to a non-HTML body → no container → drift.

Fix: the session now uses browser-like navigation headers (Accept: html, NO
X-Requested-With); the XHR header + JSON Accept are applied PER-REQUEST only on
the "load more" endpoint (which is a genuine XHR). Drift message now reports the
response length + a JSON hint so a recurrence is self-explaining. Regression test
pins the header split (navigation initial GET, XHR load-more).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
2026-06-17 14:25:40 -04:00
parent d526447496
commit 78a3977f8a
2 changed files with 52 additions and 11 deletions
+25 -11
View File
@@ -54,9 +54,16 @@ from .native_ingest_common import (
log = logging.getLogger(__name__)
_TIMEOUT_SECONDS = 30.0
# Accept header for the HTML feed + JSON "load more" + the XHR marker.
_ACCEPT = "text/html,application/xhtml+xml,application/json,*/*"
_XHR_HEADERS = {"X-Requested-With": "XMLHttpRequest"}
# The INITIAL creator-page GET is a browser-like NAVIGATION (Accept: html, NO
# X-Requested-With). SubscribeStar is Rails: an XHR-flagged request to the full
# page content-negotiates to a non-HTML response with no posts_container — which
# tripped the drift guard on the first live run (cheunart, 2026-06-17). The
# "load more" endpoint IS a real XHR and gets these headers per-request.
_NAV_ACCEPT = "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8"
_LOADMORE_HEADERS = {
"X-Requested-With": "XMLHttpRequest",
"Accept": "application/json, text/javascript, */*; q=0.01",
}
# A post block opens with this wrapper; we slice the page between consecutive
# occurrences (regex can't match the balanced close of nested divs, so chunk-
@@ -153,21 +160,21 @@ class SubscribeStarClient:
max_retries: int = _MAX_429_RETRIES,
):
self.cookies_path = str(cookies_path) if cookies_path else None
self._session = make_session(
cookies_path, accept=_ACCEPT, extra_headers=_XHR_HEADERS
)
# Browser-like navigation defaults; the load-more XHR headers are applied
# per-request (NOT session-wide) so the initial page GET isn't flagged XHR.
self._session = make_session(cookies_path, accept=_NAV_ACCEPT)
self._request_sleep = request_sleep or 0.0
self._max_retries = max_retries
# -- request -----------------------------------------------------------
def _get(self, url: str) -> requests.Response:
def _get(self, url: str, *, headers: dict | None = None) -> requests.Response:
if self._request_sleep > 0:
time.sleep(self._request_sleep)
attempt = 0
while True:
try:
resp = self._session.get(url, timeout=_TIMEOUT_SECONDS)
resp = self._session.get(url, timeout=_TIMEOUT_SECONDS, headers=headers)
except requests.RequestException as exc:
raise SubscribeStarAPIError(
f"SubscribeStar request failed ({url}): {exc}"
@@ -216,8 +223,8 @@ class SubscribeStarClient:
return text
def _loadmore_html(self, url: str) -> str:
"""Subsequent pages: GET returns JSON {"html": "<fragment>"}."""
resp = self._get(url)
"""Subsequent pages: a real XHR GET that returns JSON {"html": "..."}."""
resp = self._get(url, headers=_LOADMORE_HEADERS)
try:
payload = resp.json()
except ValueError as exc:
@@ -386,9 +393,16 @@ class SubscribeStarClient:
# body canary catches systematic emptiness across a populated feed;
# here we only raise if the feed container itself is missing.
if 'data-role="posts_container-list"' not in html:
# Surface what we actually got (length + a JSON hint) — the
# first live run tripped this because an XHR-flagged GET
# returned a non-HTML body, which is invisible from a bare
# "markup changed?" message.
looks_json = html.lstrip()[:1] in ("{", "[")
raise SubscribeStarDriftError(
f"SubscribeStar feed for {slug!r} had no posts and no "
"recognizable feed container — markup changed?"
f"recognizable feed container ({len(html)} bytes"
f"{', looks like JSON' if looks_json else ''}) — "
"markup changed, or the feed wasn't served as a full page?"
)
for post in posts:
post["_base"] = base
+27
View File
@@ -77,6 +77,33 @@ def test_parse_ss_datetime_variants():
assert _parse_ss_datetime("nonsense") is None
# -- client: request headers (regression: live-fetch drift) -----------------
def test_initial_fetch_is_navigation_loadmore_is_xhr(monkeypatch):
# The initial creator-page GET must NOT be flagged XHR (Rails content-
# negotiates an XHR full-page request to a non-HTML body → drift; first live
# run, cheunart 2026-06-17). Only the "load more" endpoint is a real XHR.
client = SubscribeStarClient(None)
seen = []
class _R:
status_code = 200
text = '<div data-role="posts_container-list"></div>'
def json(self):
return {"html": ""}
def fake_get(url, *, headers=None):
seen.append(headers)
return _R()
monkeypatch.setattr(client, "_get", fake_get)
client._feed_html("https://subscribestar.adult/x")
client._loadmore_html("https://subscribestar.adult/posts?page=2")
assert seen[0] is None # navigation — no per-request XHR header
assert seen[1]["X-Requested-With"] == "XMLHttpRequest"
# -- client: iteration + pagination -----------------------------------------
def test_iter_posts_parses_and_paginates(monkeypatch):