feat: SubscribeStar joins the membership roster (387 D1)

The second platform through the seam note 3970 contracted, characterized first from a live capture of the account's /subscriptions page (note 3989). The capture lives in the gitignored captures dir; the committed fixture is hand-built with invented values and was verified tag-for-tag against it - card wrappers, both table heads, and every distinct row shape - before any code depended on it.

What the page is, and the three decisions it forced:

The table IS the status. SubscribeStar has no per-row status word: a creator is either in the active_subscriptions card or the cancelled_subscriptions one. The card's data-identifier is stored verbatim as Membership.status and mapped in MEMBERSHIP_STATUS, keyed on the identifier rather than the table class because the cancelled table's class names the same list differently (for-unsubscribed_users).

The creator's numeric data-user-id is the key, not the slug. A slug re-keys when a creator renames; the old row stops appearing; and a disappearance is exactly what reconciliation reads as a lapse. Keyed on the slug, a rename would have told a paying subscriber they had cancelled. The slug rides as vanity, where the identity join already looks for a handle.

Price is kept as text, never parsed into amount_cents. A bare $ names no currency and a page price is not proven to be the charge - 3970 finding 4. Tier names live behind a per-row modal and are not fetched.

Refusals, because SubscribeStar offers nothing like Patreon's meta.pagination.total and every conclusion downstream is drawn from absence. The parser raises when: the active card is missing (auth error on a login/age wall, drift otherwise); a row lacks a numeric creator id or a creator link; anything renders after a card's table; or the page carries a page= link. Both cards are paginatable (app#embed_pagination) and the captured account was too small to show what pagination looks like, so possible pagination is a roster FC cannot prove complete. A loud error on a larger account beats a quiet half-list. A missing cancelled card is not drift, and a creator in both tables is reported once, as active.

Fetched from subscribestar.adult, not the .art the capture came from: FC's requests never clear the .art age wall with the 18+ cookie (1259, 1284). Whether /subscriptions on .adult authenticates exactly as .art did in the browser is untested - if not, the sweep records a visible error and C6 shows its unavailable rung.

The seam leak D1 found. Note 3970 promised a second platform would be one builders line plus the client method. The sweep instead called current_user_id() on every client, which only Patreon's has, so SubscribeStar would have raised AttributeError on the first sweep. roster_user_id probes it with getattr, the same way the sweep already probes iter_memberships.

Two existing tests were passing for the wrong reason and now can fail:
- "a platform that has never been characterised says nothing" named SubscribeStar, and stayed green only because active_patron is not a SubscribeStar word. Now uses hentaifoundry, with a positive SubscribeStar test beside it.
- the freshness test gave SubscribeStar a Patreon word, so the vocabulary excluded it and deleting the freshness gate outright would have left it green. It now uses cancelled_subscriptions, making the gate the only thing that excludes it.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SHQB1YukL3VyvMK8rcbmV9

backend/app/services/subscribestar_client.py

@@ -43,6 +43,7 @@ import requests
 from ..utils.paths import filehash_from_url
 from .native_ingest_common import (
     _MAX_429_RETRIES,
+    Membership,
     NativeAuthError,
     NativeDriftError,
     NativeIngestError,
@@ -297,6 +298,196 @@ def _extract_creator_name(html: str) -> str | None:
     return name or None
 
 
+# -- membership roster (#387 D1) ------------------------------------------
+#
+# Characterized from a live operator capture of the account's /subscriptions
+# page, 2026-09-13 — Scribe note #3989. Read that note before changing any of
+# this; each constant below is a finding from it, not a guess.
+
+# The account page is fetched from `.adult`. The `.art` age wall never clears
+# with the 18+ cookie for FC's requests (see _normalize_ss_host, issues #1259 /
+# #1284). The capture itself came from `.art` only because a human had clicked
+# through the gate in the browser.
+_ROSTER_BASE = "https://subscribestar.adult"
+_ROSTER_URL = f"{_ROSTER_BASE}/subscriptions"
+
+# Two tables, and WHICH table a creator sits in is the only status the page
+# gives — there is no per-row status word. Keyed on each card's
+# `data-identifier`, the one vocabulary that names a state: the table class
+# inside the cancelled card says `for-unsubscribed_users`, a different word for
+# the same list (note #3989, CORRECTION 1). The identifier is stored verbatim as
+# Membership.status and mapped in membership_roster.MEMBERSHIP_STATUS.
+_ROSTER_ACTIVE = "active_subscriptions"
+_ROSTER_CANCELLED = "cancelled_subscriptions"
+
+_ROSTER_ROW_OPEN = '<td class="for-name">'
+# Active rows nest a second `<tr class="for-actions">` INSIDE the row's own
+# <tr> — a narrow-screen duplicate of the actions cell. Its <td>s are not
+# columns, so every row is cut here before its cells are read.
+_ROSTER_NESTED_ROW = '<tr class="for-actions"'
+_ROSTER_HREF_RE = re.compile(r'<a href="/([^"/?#]+)"')
+_ROSTER_USER_ID_RE = re.compile(r'data-user-id="([^"]*)"')
+_ROSTER_NAME_RE = re.compile(r"<img [^>]*>([^<]*)</div>")
+_ROSTER_HEAD_RE = re.compile(r'<th class="[^"]*"[^>]*>(.*?)</th>', re.DOTALL)
+_ROSTER_CELL_RE = re.compile(r'<td class="[^"]*"[^>]*>(.*?)</td>', re.DOTALL)
+_ROSTER_PAGE_LINK_RE = re.compile(r'href="[^"]*[?&]page=\d')
+_TAG_RE = re.compile(r"<[^>]+>")
+
+# Columns that hold identity or controls rather than facts about the
+# subscription, so they stay out of `details`. Matched on the header's own text,
+# lowercased — the page's words, not ours.
+_ROSTER_SKIP_COLUMNS = frozenset({"profile", "updates", "actions"})
+
+
+def _cell_text(fragment: str) -> str:
+    """Visible text of a cell: tags dropped, entities decoded, whitespace folded.
+
+    Decoding matters here specifically: an active row with no Discord link
+    renders its cell as the entity `&mdash;`, not as an empty cell.
+    """
+    return " ".join(unescape(_TAG_RE.sub(" ", fragment)).split())
+
+
+def _roster_table(html: str, identifier: str) -> tuple[str, str] | None:
+    """One roster card: (its table markup, whatever trails `</table>` inside it).
+
+    None when the card is absent. The trailing part is returned rather than
+    discarded because it is the pagination check: in the characterized page a
+    card closes the moment its table does.
+    """
+    start = html.find(f'data-identifier="{identifier}"')
+    if start < 0:
+        return None
+    end = html.find("</table>", start)
+    if end < 0:
+        raise SubscribeStarDriftError(
+            f"SubscribeStar roster card {identifier!r} has no table"
+        )
+    close = html.find("</div>", end)
+    trailing = html[end + len("</table>"): close if close >= 0 else len(html)]
+    return html[start:end], trailing
+
+
+def _roster_rows(table: str, identifier: str, base: str) -> list[Membership]:
+    labels = [_cell_text(h).lower() for h in _ROSTER_HEAD_RE.findall(table)]
+    body = table[table.find("<tbody>"):] if "<tbody>" in table else ""
+    starts = [m.start() for m in re.finditer(re.escape(_ROSTER_ROW_OPEN), body)]
+    rows = []
+    for n, start in enumerate(starts):
+        row = body[start: starts[n + 1] if n + 1 < len(starts) else len(body)]
+        row = row.split(_ROSTER_NESTED_ROW, 1)[0]
+
+        href = _ROSTER_HREF_RE.search(row)
+        if href is None:
+            raise SubscribeStarDriftError(
+                f"SubscribeStar roster row in {identifier!r} has no creator link"
+            )
+        # The creator's numeric id, NOT the slug, is the key (note #3989,
+        # CORRECTION 2). A slug re-keys when a creator renames; the old row then
+        # stops appearing, and a disappearance is exactly what reconciliation
+        # reads as a lapse. The id survives a rename.
+        user_id = _ROSTER_USER_ID_RE.search(row)
+        if user_id is None or not user_id.group(1).isdigit():
+            raise SubscribeStarDriftError(
+                f"SubscribeStar roster row in {identifier!r} has no numeric "
+                f"data-user-id — a membership that cannot be attributed to a "
+                f"creator is not usable"
+            )
+        name = _ROSTER_NAME_RE.search(row)
+        slug = unescape(href.group(1))
+        cells = _ROSTER_CELL_RE.findall(row)
+        rows.append(Membership(
+            campaign_id=user_id.group(1),
+            display_name=(_cell_text(name.group(1)) if name else "") or None,
+            url=f"{base}/{slug}",
+            vanity=slug,
+            status=identifier,
+            # No free-follow concept on this page (#3970 §2: False when a
+            # platform has none).
+            is_free_member=False,
+            # Tier names live behind a per-row modal, not inline. Fetching every
+            # modal would be N authenticated requests for a field nothing reads.
+            tier_names=[],
+            # Deliberately NOT parsed from the price cell: a bare `$` names no
+            # currency, and a page price is not proven to be the charge (#3970
+            # finding 4). None keeps "unknown" distinct from zero. The raw text
+            # is kept in `details`.
+            amount_cents=None,
+            currency=None,
+            details={
+                # Paired with the header text by POSITION: two columns share the
+                # `for-date` class, and the updates column's <td> does not carry
+                # its <th>'s class at all.
+                "columns": {
+                    label: _cell_text(cell)
+                    for label, cell in zip(labels, cells)
+                    if label not in _ROSTER_SKIP_COLUMNS
+                },
+            },
+        ))
+    return rows
+
+
+def parse_subscriptions_page(html: str, *, base: str = _ROSTER_BASE) -> list[Membership]:
+    """Every membership on the account's /subscriptions page.
+
+    Refuses rather than guessing, because every conclusion downstream is drawn
+    from ABSENCE — a roster that comes back short reads as "you cancelled
+    those". So this raises when:
+
+    * the active card is missing — as SubscribeStarAuthError if the page is a
+      login or age wall (the fix is credentials), otherwise as drift;
+    * a row has no creator link or no numeric creator id;
+    * anything renders after a card's table, or the page carries a `page=` link.
+      Both cards are paginatable (`data-view="app#embed_pagination"`), and the
+      characterized account was too small to show what pagination looks like —
+      so possible pagination is treated as a roster FC cannot prove complete.
+
+    A missing cancelled card is NOT drift: an account that has never cancelled
+    plausibly has no such table. A creator present in both tables is reported
+    once, as active — a current subscription is the fact that matters.
+    """
+    active = _roster_table(html, _ROSTER_ACTIVE)
+    if active is None:
+        if any(marker in html for marker in _LOGIN_MARKERS):
+            raise SubscribeStarAuthError(
+                "SubscribeStar served a login/age wall instead of the "
+                "subscriptions page (cookies expired or age cookie missing)"
+            )
+        raise SubscribeStarDriftError(
+            f"SubscribeStar subscriptions page has no {_ROSTER_ACTIVE!r} card "
+            f"— {_describe_page(html)}"
+        )
+    roster_region = html[html.find(f'data-identifier="{_ROSTER_ACTIVE}"'):]
+    if _ROSTER_PAGE_LINK_RE.search(roster_region):
+        raise SubscribeStarDriftError(
+            "SubscribeStar subscriptions page carries a page= link — the roster "
+            "may be paginated, and FC cannot prove it is complete (note #3989)"
+        )
+
+    memberships: list[Membership] = []
+    seen: set[str] = set()
+    for identifier, found in (
+        (_ROSTER_ACTIVE, active),
+        (_ROSTER_CANCELLED, _roster_table(html, _ROSTER_CANCELLED)),
+    ):
+        if found is None:
+            continue
+        table, trailing = found
+        if trailing.strip():
+            raise SubscribeStarDriftError(
+                f"SubscribeStar roster card {identifier!r} renders content after "
+                f"its table — possibly pagination, so the roster cannot be "
+                f"proven complete (note #3989)"
+            )
+        for membership in _roster_rows(table, identifier, base):
+            if membership.campaign_id in seen:
+                continue
+            seen.add(membership.campaign_id)
+            memberships.append(membership)
+    return memberships
+
+
 class SubscribeStarClient:
     """Synchronous SubscribeStar HTML-scrape read client. Construct with a path
     to a Netscape cookies.txt (the same file CredentialService.get_cookies_path
@@ -645,6 +836,23 @@ class SubscribeStarClient:
             return None
         return _extract_creator_name(html)
 
+    # -- membership roster (#387 D1) ----------------------------------------
+
+    def iter_memberships(self, user_id: str | None = None) -> Iterator[Membership]:
+        """Yield every subscription the account holds (note #3989).
+
+        `user_id` exists for the seam's signature (note #3970) and is ignored:
+        the page is the logged-in account's own, so there is nothing to resolve.
+        The sweep only resolves an id for a client that exposes
+        `current_user_id`, which this one does not.
+
+        One request, and the whole page is parsed before anything is yielded, so
+        a drift error can never leave a caller holding part of a roster.
+        """
+        self._session.headers["Referer"] = f"{_ROSTER_BASE}/"
+        resp = self._get(_ROSTER_URL)
+        yield from parse_subscriptions_page(resp.text or "", base=_ROSTER_BASE)
+
     # -- verify ------------------------------------------------------------
 
     def verify_auth(self, campaign_id: str) -> tuple[bool | None, str]: