treat empty-string cursors the same as nonexistent

src/millipds/atproto_repo.py

@@ -220,7 +220,9 @@ async def repo_list_records(request: web.Request):
 	if limit < 1 or limit > 100:
 		raise web.HTTPBadRequest(text="limit out of range")
 	reverse = request.query.get("reverse") == "true"
-	cursor = request.query.get("cursor", "" if reverse else "\xff")
+	cursor = request.query.get("cursor", "")
+	if not reverse and not cursor:
+		cursor = "\xff"  # compares greater than all valid rkeys
 	did_or_handle = request.query["repo"]
 	collection = request.query["collection"]
 	records = []

src/millipds/atproto_sync.py

@@ -190,7 +190,7 @@ async def sync_list_blobs(request: web.Request):
 	limit = int(request.query.get("limit", 500))
 	if limit < 1 or limit > 1000:
 		raise web.HTTPBadRequest(text="limit out of range")
-	cursor = int(request.query.get("cursor", 0))
+	cursor = int(request.query.get("cursor") or 0)
 
 	cids = []
 	for id_, cid in get_db(request).con.execute(
@@ -246,9 +246,9 @@ async def sync_subscribe_repos(request: web.Request):
 		await ws.prepare(request)  # TODO: should this be outside of try?
 
 		last_sent_seq = None
-		if "cursor" in request.query:
+		if cursor_str := request.query.get("cursor"):
 			# TODO: try to limit number of concurrent backfillers? (it's more expensive than livetailing)
-			cursor = int(request.query["cursor"])
+			cursor = int(cursor_str)
 			db = get_db(request)
 			while True:
 				# we read one at a time to force gaps between reads - could be a perf win to read in small batches

src/millipds/database.py

@@ -150,6 +150,7 @@ def _init_tables(self):
 			) STRICT
 			"""
 		)
+		# TODO: index on timestamp for efficient purging of old events.
 
 		# repo storage stuff
 		self.con.execute(

tests/integration_test.py

@@ -494,17 +494,18 @@ async def test_deleteSession(s, pds_host):
 	) as r:
 		assert r.status != 200
 
+
 async def test_updateHandle(s, pds_host, auth_headers):
 	async with s.post(
 		pds_host + "/xrpc/com.atproto.identity.updateHandle",
 		headers=auth_headers,
-		json={ "handle": "juliet.test" },
+		json={"handle": "juliet.test"},
 	) as r:
 		assert r.status == 200
 
 	async with s.get(
 		pds_host + "/xrpc/com.atproto.repo.describeRepo",
-		params={ "repo":  TEST_DID },
+		params={"repo": TEST_DID},
 	) as r:
 		assert r.status == 200
 		r = await r.json()