feat: Settings can point backups at an off-site S3 bucket, with a Test destination check

Settings → Maintenance → Storage gains "Off-site backup destination": the
address (with its port), bucket, region, access key and secret key of an S3
bucket such as Fabled Archives. The secret key is sealed with the credential
key, never returned by the API, and kept when the field is left blank. Test
destination lists the bucket with the fields as typed and writes nothing, and
names the cause when it fails ("nothing is listening there (check the port)",
"the secret key is wrong", "answered, but not as S3").

services/backup_s3 is a standard-library SigV4 client (put, get, paged list)
ported from Grimoire's; it is held to AWS's published signing examples and
exercised against a local HTTP server. Migration 0118 adds the columns; no
destination is set up until someone saves one.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

backend/app/api/system_backup.py

@@ -236,6 +236,109 @@ async def get_settings():
     })
 
 
+@system_backup_bp.route("/destination", methods=["GET"])
+async def get_destination():
+    """The off-site destination as the settings screen shows it. The secret
+    key is never returned, only whether one is saved."""
+    from ..services import backup_destination as dest
+
+    async with get_session() as session:
+        row = await ImportSettings.load(session)
+        return jsonify(dest.describe(row))
+
+
+@system_backup_bp.route("/destination", methods=["PATCH"])
+async def patch_destination():
+    """Save the destination. Body {endpoint, bucket, access_key, region,
+    secret_key}; an empty field clears it. A missing or empty secret_key
+    keeps the saved one, so the screen can save the other fields without the
+    secret being typed again."""
+    from ..services import backup_destination as dest
+
+    body = await request.get_json(silent=True)
+    if not isinstance(body, dict):
+        return _bad("invalid_body", detail="body must be a JSON object")
+    fields, why = dest.clean_fields(body)
+    if fields is None:
+        return _bad("invalid_destination", detail=why)
+    secret = body.get("secret_key")
+    if secret is not None and not isinstance(secret, str):
+        return _bad("invalid_destination", detail="secret_key must be text")
+
+    async with get_session() as session:
+        row = await ImportSettings.load(session)
+        row.backup_s3_endpoint = fields.endpoint
+        row.backup_s3_bucket = fields.bucket
+        row.backup_s3_access_key = fields.access_key
+        row.backup_s3_region = fields.region
+        if secret and secret.strip():
+            row.backup_s3_secret_sealed = dest.seal(secret.strip())
+        await session.commit()
+        return jsonify(dest.describe(row))
+
+
+@system_backup_bp.route("/destination", methods=["DELETE"])
+async def delete_destination():
+    """Forget the destination, secret included. What was already sent stays in
+    the bucket: FabledCurator never deletes there."""
+    from ..services import backup_destination as dest
+
+    async with get_session() as session:
+        row = await ImportSettings.load(session)
+        row.backup_s3_endpoint = None
+        row.backup_s3_bucket = None
+        row.backup_s3_access_key = None
+        row.backup_s3_secret_sealed = None
+        row.backup_s3_region = dest.DEFAULT_REGION
+        await session.commit()
+        return jsonify(dest.describe(row))
+
+
+@system_backup_bp.route("/destination/test", methods=["POST"])
+async def try_destination():
+    """Reach the bucket with the fields in the body (the form as typed, saved
+    or not); a missing secret_key uses the saved one. Lists the bucket and
+    writes nothing. Answers {ok, message} either way: a destination that
+    can't be reached is a result to show, not an API error."""
+    import asyncio
+
+    from ..services import backup_destination as dest
+    from ..services.backup_s3 import S3Error
+    from ..services.credential_crypto import InvalidCredentialBlob
+
+    body = await request.get_json(silent=True) or {}
+    if not isinstance(body, dict):
+        return _bad("invalid_body", detail="body must be a JSON object")
+    fields, why = dest.clean_fields(body)
+    if fields is None:
+        return jsonify({"ok": False, "message": f"Not tested: {why}."})
+    secret = body.get("secret_key")
+    secret = secret.strip() if isinstance(secret, str) else ""
+    if not secret:
+        async with get_session() as session:
+            row = await ImportSettings.load(session)
+            sealed = row.backup_s3_secret_sealed
+        if sealed:
+            try:
+                secret = dest.unseal(sealed)
+            except InvalidCredentialBlob:
+                return jsonify({"ok": False, "message": (
+                    "Not tested: the saved secret key can't be opened on this server. "
+                    "Enter the secret key again."
+                )})
+    target = dest.target_for(fields, secret or None)
+    if target is None:
+        missing = [name for name, v in (
+            ("address", fields.endpoint), ("bucket", fields.bucket),
+            ("access key", fields.access_key), ("secret key", secret),
+        ) if not v]
+        return jsonify({"ok": False, "message": f"Not tested: fill in the {', '.join(missing)}."})
+    try:
+        return jsonify(await asyncio.to_thread(dest.check_target, target))
+    except S3Error as exc:
+        return jsonify({"ok": False, "message": str(exc)})
+
+
 @system_backup_bp.route("/settings", methods=["PATCH"])
 async def patch_settings():
     body = await request.get_json(silent=True)

backend/app/models/import_settings.py

@@ -9,6 +9,7 @@ from sqlalchemy import (
     CheckConstraint,
     Float,
     Integer,
+    String,
     Text,
     select,
     text,
@@ -100,6 +101,16 @@ class ImportSettings(Base):
         Integer, nullable=False, default=3,
         server_default="3",
     )
+    # Off-site backup destination: an S3 bucket (services/backup_destination).
+    # The secret key is stored sealed with the credential key, never in clear.
+    backup_s3_endpoint: Mapped[str | None] = mapped_column(Text, nullable=True)
+    backup_s3_bucket: Mapped[str | None] = mapped_column(String(255), nullable=True)
+    backup_s3_access_key: Mapped[str | None] = mapped_column(Text, nullable=True)
+    backup_s3_secret_sealed: Mapped[str | None] = mapped_column(Text, nullable=True)
+    backup_s3_region: Mapped[str] = mapped_column(
+        String(64), nullable=False, default="us-east-1",
+        server_default="us-east-1",
+    )
 
     # FC-6.3 series continuation matcher. enabled gates the rescan; threshold is
     # the weighted-score cut-off (0..1) above which a pending suggestion is made.

backend/app/services/backup_destination.py

@@ -0,0 +1,137 @@
+"""The off-site backup destination: an S3 bucket (Fabled Archives, or any S3
+service) that the database dumps and the media blobs are sent to.
+
+The address, bucket, access key and region are plain settings. The secret key
+is sealed at rest with the credential key (`/images/secrets`), the same key
+that seals account logins, and no API ever returns it: the settings answer
+only whether one is set. Because the credential key is kept out of every
+backup, a database dump carries the secret only as ciphertext nobody can open
+without the original server.
+"""
+from __future__ import annotations
+
+from dataclasses import dataclass
+from pathlib import Path
+from urllib.parse import urlsplit
+
+from .backup_s3 import S3Error, S3Target, list_objects
+from .credential_crypto import CredentialCrypto, InvalidCredentialBlob
+
+KEY_PATH = Path("/images/secrets/credential_key.b64")
+DEFAULT_REGION = "us-east-1"
+_TEST_LISTING = 5
+
+
+def _crypto() -> CredentialCrypto:
+    return CredentialCrypto(KEY_PATH)
+
+
+def seal(secret: str) -> str:
+    return _crypto().encrypt(secret).decode("ascii")
+
+
+def unseal(sealed: str) -> str:
+    return _crypto().decrypt(sealed.encode("ascii"))
+
+
+@dataclass(frozen=True)
+class DestinationFields:
+    endpoint: str | None
+    bucket: str | None
+    access_key: str | None
+    region: str
+
+
+def clean_fields(body: dict) -> tuple[DestinationFields | None, str | None]:
+    """The destination fields from a request body, trimmed, with empty ones as
+    None; or (None, why) when the address can't be an S3 endpoint."""
+    def text(name: str) -> str | None:
+        v = body.get(name)
+        if v is None:
+            return None
+        if not isinstance(v, str):
+            raise TypeError(name)
+        return v.strip() or None
+
+    try:
+        endpoint = text("endpoint")
+        bucket = text("bucket")
+        access_key = text("access_key")
+        region = text("region") or DEFAULT_REGION
+    except TypeError as exc:
+        return None, f"{exc.args[0]} must be text"
+    if endpoint is not None:
+        u = urlsplit(endpoint)
+        if u.scheme not in ("http", "https") or not u.hostname:
+            return None, ("the address must start with http:// or https:// and name "
+                          "a host, e.g. http://archives.example:8081")
+        if u.query or u.fragment:
+            return None, "the address can't carry a query or a fragment"
+        endpoint = endpoint.rstrip("/")
+    if bucket is not None and "/" in bucket:
+        return None, "the bucket is a name, not a path: put any path in the address"
+    return DestinationFields(endpoint, bucket, access_key, region), None
+
+
+def target_for(fields: DestinationFields, secret: str | None) -> S3Target | None:
+    """The S3 target, or None while any part of it is missing."""
+    if not (fields.endpoint and fields.bucket and fields.access_key and secret):
+        return None
+    return S3Target(fields.endpoint, fields.bucket, fields.access_key, secret,
+                    region=fields.region)
+
+
+def saved_target(settings) -> S3Target | None:
+    """The destination saved in ImportSettings, or None when none is set up.
+    Raises S3Error when the secret can't be opened (the credential key
+    changed since it was saved)."""
+    sealed = settings.backup_s3_secret_sealed
+    if not sealed:
+        return None
+    try:
+        secret = unseal(sealed)
+    except InvalidCredentialBlob:
+        raise S3Error(
+            "The saved secret key can't be opened on this server (the credential "
+            "key changed since it was saved). Enter the secret key again."
+        ) from None
+    return target_for(fields_of(settings), secret)
+
+
+def fields_of(settings) -> DestinationFields:
+    return DestinationFields(
+        settings.backup_s3_endpoint, settings.backup_s3_bucket,
+        settings.backup_s3_access_key, settings.backup_s3_region or DEFAULT_REGION,
+    )
+
+
+def describe(settings) -> dict:
+    """The destination as the settings screen shows it: never the secret."""
+    f = fields_of(settings)
+    return {
+        "endpoint": f.endpoint or "",
+        "bucket": f.bucket or "",
+        "access_key": f.access_key or "",
+        "region": f.region,
+        "secret_set": bool(settings.backup_s3_secret_sealed),
+        "configured": bool(
+            f.endpoint and f.bucket and f.access_key and settings.backup_s3_secret_sealed
+        ),
+    }
+
+
+def check_target(target: S3Target) -> dict:
+    """Reach the bucket by listing it. A listing proves the address, the
+    bucket and both keys at once, and writes nothing, because an Archives key
+    is allowed to add objects but a test object could never be removed."""
+    try:
+        page = list_objects(target, limit=_TEST_LISTING)
+    except S3Error as exc:
+        return {"ok": False, "message": str(exc)}
+    if not page.objects:
+        said = f"Reached {target.bucket} at {target.endpoint}. The bucket is empty."
+    else:
+        more = "+" if page.next_token else ""
+        said = (f"Reached {target.bucket} at {target.endpoint}. It holds "
+                f"{len(page.objects)}{more} object(s), e.g. {page.objects[0].key}.")
+    return {"ok": True, "message": said}

backend/app/services/backup_s3.py

@@ -0,0 +1,282 @@
+"""A small S3 client for sending backups off the box: AWS Signature Version 4,
+path-style addressing (endpoint/bucket/key), and the calls a backup needs —
+put an object, read one back, list a prefix.
+
+FabledCurator puts, lists and reads; it never deletes. The bucket keeps the
+history and thins it on its own policy, so a key that may only add and read
+objects is enough, and is the safer key to hand a service: a stolen one cannot
+wipe the history it writes into.
+
+Built on the standard library's http.client rather than boto3: three calls
+don't justify a large dependency, and signing is a pure function the tests
+hold to AWS's own published examples. Every call is synchronous, because the
+uploads run in Celery workers; the API runs the Test destination call in a
+thread.
+"""
+from __future__ import annotations
+
+import hashlib
+import hmac
+import http.client
+import os
+import socket
+import ssl
+import xml.etree.ElementTree as ET
+from dataclasses import dataclass
+from datetime import UTC, datetime
+from pathlib import Path
+from urllib.parse import parse_qsl, quote, unquote, urlencode, urlsplit
+
+EMPTY_SHA256 = hashlib.sha256(b"").hexdigest()
+TIMEOUT_SECONDS = 30.0
+# A single object can be a multi-gigabyte video or archive; a slow link still
+# gets this long between bytes before the call gives up.
+TRANSFER_TIMEOUT_SECONDS = 15 * 60
+CHUNK = 1 << 20
+
+
+class S3Error(Exception):
+    """Why a call to the bucket failed, said to the person."""
+
+
+@dataclass(frozen=True)
+class S3Target:
+    endpoint: str          # http(s)://host[:port][/prefix], no bucket
+    bucket: str
+    access_key: str
+    secret_key: str
+    region: str = "us-east-1"
+    verify_tls: bool = True
+
+    def url(self, key: str = "") -> str:
+        base = f"{self.endpoint.rstrip('/')}/{quote(self.bucket, safe='')}"
+        return f"{base}/{quote(key, safe='/-_.~')}" if key else f"{base}/"
+
+
+@dataclass(frozen=True)
+class StoredObject:
+    key: str
+    size: int
+    last_modified: str
+
+
+@dataclass(frozen=True)
+class ListPage:
+    objects: list[StoredObject]
+    next_token: str | None     # pass back as `continuation` for the next page
+
+
+# ── Signing ──────────────────────────────────────────────────────────────────
+
+def _encode(text: str, keep_slash: bool = False) -> str:
+    return quote(text, safe="-_.~" + ("/" if keep_slash else ""))
+
+
+def _hmac(key: bytes, msg: str) -> bytes:
+    return hmac.new(key, msg.encode(), hashlib.sha256).digest()
+
+
+def sign(method: str, url: str, headers: dict[str, str], payload_sha256: str,
+         access_key: str, secret_key: str, region: str, at: datetime,
+         service: str = "s3") -> dict[str, str]:
+    """The request's headers with Host, x-amz-date, x-amz-content-sha256 and
+    Authorization added. Every header given is signed, Host always."""
+    u = urlsplit(url)
+    stamp = at.astimezone(UTC).strftime("%Y%m%dT%H%M%SZ")
+    day = stamp[:8]
+    out = dict(headers)
+    out["Host"] = u.netloc
+    out["x-amz-date"] = stamp
+    out["x-amz-content-sha256"] = payload_sha256
+    signed = {k.lower(): " ".join(str(v).split()) for k, v in out.items()}
+    names = sorted(signed)
+
+    query = sorted(
+        (_encode(k), _encode(v)) for k, v in parse_qsl(u.query, keep_blank_values=True)
+    )
+    canonical = "\n".join([
+        method.upper(),
+        _encode(unquote(u.path) or "/", keep_slash=True),
+        "&".join(f"{k}={v}" for k, v in query),
+        "".join(f"{n}:{signed[n]}\n" for n in names),
+        ";".join(names),
+        payload_sha256,
+    ])
+    scope = f"{day}/{region}/{service}/aws4_request"
+    to_sign = "\n".join(["AWS4-HMAC-SHA256", stamp, scope,
+                         hashlib.sha256(canonical.encode()).hexdigest()])
+    key = _hmac(_hmac(_hmac(_hmac(f"AWS4{secret_key}".encode(), day), region),
+                      service), "aws4_request")
+    signature = hmac.new(key, to_sign.encode(), hashlib.sha256).hexdigest()
+    out["Authorization"] = (f"AWS4-HMAC-SHA256 Credential={access_key}/{scope}, "
+                            f"SignedHeaders={';'.join(names)}, Signature={signature}")
+    return out
+
+
+# ── Transport ────────────────────────────────────────────────────────────────
+
+def _connect(target: S3Target, url: str, timeout: float) -> http.client.HTTPConnection:
+    u = urlsplit(url)
+    if u.scheme == "https":
+        ctx = ssl.create_default_context()
+        if not target.verify_tls:
+            ctx.check_hostname = False
+            ctx.verify_mode = ssl.CERT_NONE
+        return http.client.HTTPSConnection(
+            u.hostname, u.port, timeout=timeout, context=ctx, blocksize=CHUNK,
+        )
+    if u.scheme != "http":
+        raise S3Error(f"{target.endpoint} isn't an http:// or https:// address.")
+    return http.client.HTTPConnection(u.hostname, u.port, timeout=timeout, blocksize=CHUNK)
+
+
+def _request_target(url: str) -> str:
+    u = urlsplit(url)
+    return (u.path or "/") + (f"?{u.query}" if u.query else "")
+
+
+def _reach_error(target: S3Target, exc: Exception) -> S3Error:
+    if isinstance(exc, TimeoutError):
+        return S3Error(f"{target.endpoint} didn't answer in time.")
+    if isinstance(exc, socket.gaierror):
+        return S3Error(f"Couldn't reach {target.endpoint}: the host name doesn't resolve.")
+    if isinstance(exc, ConnectionRefusedError):
+        return S3Error(
+            f"Couldn't reach {target.endpoint}: nothing is listening there "
+            "(check the port)."
+        )
+    if isinstance(exc, ssl.SSLError):
+        return S3Error(f"Couldn't reach {target.endpoint}: TLS failed ({exc.reason}).")
+    return S3Error(f"Couldn't reach {target.endpoint}: {exc}.")
+
+
+def _says(status: int, reason: str, content: bytes) -> str:
+    """An S3 error response in words: its Code and Message when it has them."""
+    try:
+        root = ET.fromstring(content) if content else None
+        code = (root.findtext("Code") or "") if root is not None else ""
+        message = (root.findtext("Message") or "") if root is not None else ""
+    except ET.ParseError:
+        code, message = "", ""
+    hints = {
+        "SignatureDoesNotMatch": "the secret key is wrong",
+        "InvalidAccessKeyId": "the access key isn't known there",
+        "NoSuchBucket": "there is no such bucket",
+        "AccessDenied": "the key isn't allowed to do that",
+    }
+    said = hints.get(code) or message or reason
+    if not code and 400 <= status < 500:
+        # S3 refuses with an <Error><Code> body. A 4xx without one came from
+        # something else at that address — most often a web server on the
+        # default port, when the S3 service listens on a port of its own.
+        said = ("the address answered, but not as S3 — check the scheme and "
+                "the port (an S3 service usually listens on its own port)")
+    return f"HTTP {status}{f' {code}' if code else ''}: {said}."
+
+
+# ── Calls ────────────────────────────────────────────────────────────────────
+
+def file_sha256(path: Path) -> str:
+    h = hashlib.sha256()
+    with open(path, "rb") as f:
+        while chunk := f.read(CHUNK):
+            h.update(chunk)
+    return h.hexdigest()
+
+
+def put_file(target: S3Target, key: str, path: Path, *,
+             content_type: str = "application/octet-stream",
+             payload_sha256: str | None = None,
+             now: datetime | None = None) -> None:
+    """Upload one file as one object, streamed from disk. `payload_sha256` is
+    the file's sha256 when the caller already knows it (the library stores
+    every file's), which saves reading the file twice."""
+    digest = payload_sha256 or file_sha256(path)
+    size = path.stat().st_size
+    url = target.url(key)
+    headers = sign("PUT", url, {"Content-Length": str(size), "Content-Type": content_type},
+                   digest, target.access_key, target.secret_key, target.region,
+                   now or datetime.now(UTC))
+    conn = _connect(target, url, TRANSFER_TIMEOUT_SECONDS)
+    try:
+        with open(path, "rb") as f:
+            conn.request("PUT", _request_target(url), body=f, headers=headers)
+            response = conn.getresponse()
+            content = response.read()
+    except (OSError, http.client.HTTPException) as exc:
+        raise _reach_error(target, exc) from None
+    finally:
+        conn.close()
+    if response.status // 100 != 2:
+        raise S3Error(f"Upload of {key} refused: "
+                      f"{_says(response.status, response.reason, content)}")
+
+
+def get_file(target: S3Target, key: str, path: Path, *,
+             now: datetime | None = None) -> None:
+    """Download one object to `path`, streamed to disk. The bytes land under a
+    `.partial` name and are renamed into place when complete, so a broken
+    transfer never leaves a truncated file at `path`."""
+    url = target.url(key)
+    headers = sign("GET", url, {}, EMPTY_SHA256, target.access_key, target.secret_key,
+                   target.region, now or datetime.now(UTC))
+    partial = path.with_name(path.name + ".partial")
+    conn = _connect(target, url, TRANSFER_TIMEOUT_SECONDS)
+    try:
+        conn.request("GET", _request_target(url), headers=headers)
+        response = conn.getresponse()
+        if response.status != 200:
+            raise S3Error(f"Reading {key} refused: "
+                          f"{_says(response.status, response.reason, response.read())}")
+        with open(partial, "wb") as f:
+            while chunk := response.read(CHUNK):
+                f.write(chunk)
+        os.replace(partial, path)
+    except (OSError, http.client.HTTPException) as exc:
+        partial.unlink(missing_ok=True)
+        raise _reach_error(target, exc) from None
+    finally:
+        conn.close()
+
+
+def list_objects(target: S3Target, prefix: str = "", limit: int = 1000, *,
+                 continuation: str | None = None,
+                 now: datetime | None = None) -> ListPage:
+    """One page (up to `limit`) of the objects under `prefix`, ListObjectsV2."""
+    params = {"list-type": "2", "max-keys": str(limit)}
+    if prefix:
+        params["prefix"] = prefix
+    if continuation:
+        params["continuation-token"] = continuation
+    url = f"{target.url()}?{urlencode(params, quote_via=quote)}"
+    headers = sign("GET", url, {}, EMPTY_SHA256, target.access_key, target.secret_key,
+                   target.region, now or datetime.now(UTC))
+    conn = _connect(target, url, TIMEOUT_SECONDS)
+    try:
+        conn.request("GET", _request_target(url), headers=headers)
+        response = conn.getresponse()
+        content = response.read()
+    except (OSError, http.client.HTTPException) as exc:
+        raise _reach_error(target, exc) from None
+    finally:
+        conn.close()
+    if response.status != 200:
+        raise S3Error(f"Listing {target.bucket} refused: "
+                      f"{_says(response.status, response.reason, content)}")
+    try:
+        root = ET.fromstring(content)
+    except ET.ParseError:
+        raise S3Error(
+            f"{target.endpoint} answered a listing with something that isn't S3's XML."
+        ) from None
+    objects, next_token = [], None
+    for item in root.iter():
+        tag = item.tag.rsplit("}", 1)[-1]
+        if tag == "NextContinuationToken":
+            next_token = item.text or None
+        if tag != "Contents":
+            continue
+        fields = {child.tag.rsplit("}", 1)[-1]: (child.text or "") for child in item}
+        objects.append(StoredObject(fields.get("Key", ""), int(fields.get("Size") or 0),
+                                    fields.get("LastModified", "")))
+    return ListPage(objects, next_token)