Add CAR and CID benchmarks (#20)

pytests/conftest.py

@@ -1,5 +1,6 @@
 import json
 import os
+import urllib.request
 from typing import Any, List, Tuple
 
 
@@ -32,3 +33,15 @@ def load_cbor_data_fixtures(dir_path: str) -> List[Tuple[str, Any]]:
             fixtures.append((os.path.basename(root), fixture))
 
     return fixtures
+
+
+def load_car_fixture(did: str, path: str) -> bytes:
+    if os.path.exists(path):
+        with open(path, 'rb') as f:
+            return f.read()
+
+    contents = urllib.request.urlopen(f'https://bsky.network/xrpc/com.atproto.sync.getRepo?did={did}').read()
+    with open(path, 'wb') as f:
+        f.write(contents)
+
+    return contents

pytests/test_dag_cbor.py

@@ -8,6 +8,7 @@
 _ROUNDTRIP_DATA_DIR = os.path.join(os.path.dirname(__file__), '..', 'data', 'roundtrip')
 _REAL_DATA_DIR = os.path.join(os.path.dirname(__file__), '..', 'data')
 _FIXTURES_DATA_DIR = os.path.join(os.path.dirname(__file__), '..', 'data', 'fixtures')
+_CIDS_DAG_CBOR_PATH = os.path.join(os.path.dirname(__file__), '..', 'data', 'torture_cids.dag-cbor')
 
 
 def _dag_cbor_encode(benchmark, data) -> None:
@@ -103,3 +104,8 @@ def test_dag_cbor_decode_real_data(benchmark, data) -> None:
 @pytest.mark.parametrize('data', load_cbor_data_fixtures(_FIXTURES_DATA_DIR), ids=lambda data: data[0])
 def test_dag_cbor_decode_fixtures(benchmark, data) -> None:
     _dag_cbor_decode(benchmark, data)
+
+
+def test_dag_cbor_decode_torture_cids(benchmark) -> None:
+    dag_cbor = open(_CIDS_DAG_CBOR_PATH, 'rb').read()
+    benchmark(libipld.decode_dag_cbor, dag_cbor)

pytests/test_decode_car.py

@@ -0,0 +1,29 @@
+import os
+
+import libipld
+import pytest
+
+from conftest import load_car_fixture
+
+
+_DID = os.environ.get('CAR_REPO_DID', 'did:plc:w4es6sfh43zlht3bgrzi5qzq')  # default is public bot in bsky.app
+_REPO_CAR_PATH = os.path.join(os.path.dirname(__file__), '..', 'data', 'repo.car')
+
+
+@pytest.fixture(scope='session')
+def car() -> bytes:
+    return load_car_fixture(_DID, _REPO_CAR_PATH)
+
+
+def test_decode_car(benchmark, car) -> None:
+    header, blocks = benchmark(libipld.decode_car, car)
+
+    assert 1 == header['version']
+    assert isinstance(header['roots'], list)
+    assert 1 == len(header['roots'])
+
+    assert isinstance(blocks, dict)
+    assert all(isinstance(k, str) for k in blocks.keys())
+    assert all(len(k) == 59 for k in blocks.keys())
+    assert all(isinstance(v, dict) for v in blocks.values())
+    assert all(v for v in blocks.values())  # not empty dict