Merge branch 'release-7.0.4'

CHANGELOG.md

@@ -1,3 +1,8 @@
+# 7.0.4, 2024-03-26
+
+* Fix wb mode with zstd compression (PR [#815](https://github.com/piskvorky/smart_open/pull/815), [@djudd](https://github.com/djudd))
+* Remove GCS bucket.exists call to avoid storage.buckets.get permission (PR [#813](https://github.com/piskvorky/smart_open/pull/813), [@ddelange](https://github.com/ddelange))
+
 # 7.0.3, 2024-03-21
 
 * add support for zst writing (PR [#812](https://github.com/piskvorky/smart_open/pull/812), [@mpenkov](https://github.com/mpenkov))

smart_open/compression.py

@@ -6,6 +6,7 @@
 # from the MIT License (MIT).
 #
 """Implements the compression layer of the ``smart_open`` library."""
+import io
 import logging
 import os.path
 
@@ -108,6 +109,17 @@ def _handle_gzip(file_obj, mode):
 def _handle_zstd(file_obj, mode):
     import zstandard  # type: ignore
     result = zstandard.open(filename=file_obj, mode=mode)
+    # zstandard.open returns an io.TextIOWrapper in text mode, but otherwise
+    # returns a raw stream reader/writer, and we need the `io` wrapper
+    # to make FileLikeProxy work correctly.
+    #
+    # See:
+    #
+    # https://github.com/indygreg/python-zstandard/blob/d7d81e79dbe74feb22fb73405ebfb3e20f4c4653/zstandard/__init__.py#L169-L174
+    if "b" in mode and "w" in mode:
+        result = io.BufferedWriter(result)
+    elif "b" in mode and "r" in mode:
+        result = io.BufferedReader(result)
     return result
 
 

smart_open/gcs.py

@@ -152,11 +152,7 @@ def Writer(bucket,
 
     blob_open_kwargs = {**_DEFAULT_WRITE_OPEN_KWARGS, **blob_open_kwargs}
 
-    g_bucket = client.bucket(bucket)
-    if not g_bucket.exists():
-        raise google.cloud.exceptions.NotFound(f'bucket {bucket} not found')
-
-    g_blob = g_bucket.blob(
+    g_blob = client.bucket(bucket).blob(
         blob,
         chunk_size=min_part_size,
     )

smart_open/tests/test_smart_open.py

@@ -89,6 +89,18 @@ def test_zst_write():
     assert got == ["hello world\n", "this is a test\n"]
 
 
+def test_zst_write_binary():
+    with named_temporary_file(suffix=".zst") as tmp:
+        with smart_open.open(tmp.name, "wb") as fout:
+            fout.write(b"hello world\n")
+            fout.write(b"this is a test\n")
+
+        with smart_open.open(tmp.name, "rb") as fin:
+            got = list(fin)
+
+    assert got == [b"hello world\n", b"this is a test\n"]
+
+
 class ParseUriTest(unittest.TestCase):
     """
     Test ParseUri class.

smart_open/version.py

@@ -1,4 +1,4 @@
-__version__ = '7.0.3'
+__version__ = '7.0.4'
 
 
 if __name__ == '__main__':