0.6.26

mabel/adapters/google/google_cloud_storage_writer.py

@@ -46,6 +46,8 @@ def commit(self, byte_data, override_blob_name=None):
         # name from the path builder
         if override_blob_name:
             blob_name = override_blob_name
+            if blob_name.startswith(self.bucket + "/"):
+                blob_name = blob_name[len(self.bucket) + 1 :]
         else:
             blob_name = self._build_path()
 

mabel/data/internals/group_by.py

@@ -73,8 +73,7 @@ def _map(self, collect_columns):
         for record in self._dictset:
             try:
                 group_key: int = xxh3_64_intdigest(
-                    "".join([str(record[column]) for column in self._columns]),
-                    HASH_SEED
+                    "".join([str(record[column]) for column in self._columns]), HASH_SEED
                 )
             except KeyError:
                 group_key: int = xxh3_64_intdigest(

mabel/data/writers/internals/blob_writer.py

@@ -23,7 +23,6 @@ class BlobWriter(object):
     # this variable outside the __init__.
     buffer = bytearray()
     byte_count = 0
-    manifest = {}
 
     def __init__(
         self,
@@ -141,7 +140,6 @@ def commit(self):
         if self.records_in_buffer > 0:
             lock = threading.Lock()
 
-            summary = None
             try:
                 lock.acquire(blocking=True, timeout=10)
 
@@ -155,10 +153,6 @@ def commit(self):
                         )
 
                     pytable = self.wal.arrow()
-                    try:
-                        summary = self.wal.profile.to_dicts()
-                    except Exception as e:
-                        print(f"[MABEL] Unable to profile morsel - {type(e).__name__} - {e}")
 
                     # if we have a schema, make effort to align the parquet file to it
                     if self.schema:
@@ -179,7 +173,6 @@ def commit(self):
                 committed_blob_name = self.inner_writer.commit(
                     byte_data=write_buffer, override_blob_name=None
                 )
-                self.manifest[committed_blob_name] = summary
 
                 if "BACKOUT" in committed_blob_name:
                     get_logger().warning(

tests/test_data_group_by.py

@@ -130,7 +130,9 @@ def test_gappy_set():
         {"AVG(key)": 1.0, "value": "one"},
     ]
 
-    assert set(tuple(sorted(d.items())) for d in gs) == set(tuple(sorted(d.items())) for d in expected)
+    assert set(tuple(sorted(d.items())) for d in gs) == set(
+        tuple(sorted(d.items())) for d in expected
+    )
 
 
 if __name__ == "__main__":  # pragma: no cover

tests/test_writer_batch_writer.py

@@ -6,6 +6,7 @@
 import pytest
 
 sys.path.insert(1, os.path.join(sys.path[0], ".."))
+
 from mabel.adapters.disk import DiskReader, DiskWriter
 from mabel.data import BatchWriter
 from mabel.data import Reader
@@ -215,7 +216,7 @@ def get_data():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    from tests.helpers.runner import run_tests
+    from helpers.runner import run_tests
 
     test_writer_without_schema_parquet()
     run_tests()