Merge branch 'release-7.0.2'

CHANGELOG.md

@@ -1,3 +1,9 @@
+# 7.0.2, 2024-03-21
+
+* Add `__next__` method to FileLikeProxy (PR [#811](https://github.com/piskvorky/smart_open/pull/811), [@ddelange](https://github.com/ddelange))
+* Fix python_requires minimum python version in setup.py (PR [#807](https://github.com/piskvorky/smart_open/pull/807), [@pressler-vsc](https://github.com/pressler-vsc))
+* Add activity check to cached sftp connections (PR [#808](https://github.com/piskvorky/smart_open/pull/808), [@greg-offerfit](https://github.com/greg-offerfit))
+
 # 7.0.1, 2024-02-26
 
 * Do not touch botocore unless it is installed (PR [#803](https://github.com/piskvorky/smart_open/pull/803), [@ddelange](https://github.com/ddelange))

setup.py

@@ -85,7 +85,7 @@ def read(fname):
         'ssh': ssh_deps,
         'zst': zst_deps,
     },
-    python_requires=">=3.6,<4.0",
+    python_requires=">=3.7,<4.0",
 
     test_suite="smart_open.tests",
 

smart_open/ssh.py

@@ -111,7 +111,10 @@ def parse_uri(uri_as_string):
 
 
 def open_uri(uri, mode, transport_params):
-    smart_open.utils.check_kwargs(open, transport_params)
+    # `connect_kwargs` is a legitimate param *only* for sftp, so this filters it out of validation
+    #   (otherwise every call with this present complains it's not valid)
+    params_to_validate = {k: v for k, v in transport_params.items() if k != 'connect_kwargs'}
+    smart_open.utils.check_kwargs(open, params_to_validate)
     parsed_uri = parse_uri(uri)
     uri_path = parsed_uri.pop('uri_path')
     parsed_uri.pop('scheme')
@@ -266,6 +269,11 @@ def open(path, mode='r', host=None, user=None, password=None, port=None, transpo
     for attempt in range(attempts):
         try:
             ssh = _SSH[key]
+            # Validate that the cached connection is still an active connection
+            #   and if not, refresh the connection
+            if not ssh.get_transport().active:
+                ssh.close()
+                ssh = _SSH[key] = _connect_ssh(host, user, port, password, transport_params)
         except KeyError:
             ssh = _SSH[key] = _connect_ssh(host, user, port, password, transport_params)
 

smart_open/tests/test_s3.py

@@ -379,11 +379,37 @@ def test_binary_iterator(self):
         expected = u"выйду ночью в поле с конём".encode('utf-8').split(b' ')
         _resource('s3').Object(BUCKET_NAME, KEY_NAME).put(Body=b'\n'.join(expected))
 
+        # test the __iter__ method
         with self.assertApiCalls(GetObject=1):
             with smart_open.s3.open(BUCKET_NAME, KEY_NAME, 'rb') as fin:
                 actual = [line.rstrip() for line in fin]
         self.assertEqual(expected, actual)
 
+        # test the __next__ method
+        with self.assertApiCalls(GetObject=1):
+            with smart_open.s3.open(BUCKET_NAME, KEY_NAME, 'rb') as fin:
+                first = next(fin).rstrip()
+        self.assertEqual(expected[0], first)
+
+    def test_text_iterator(self):
+        expected = u"выйду ночью в поле с конём".split(' ')
+        uri = f's3://{BUCKET_NAME}/{KEY_NAME}.gz'
+
+        with smart_open.open(uri, 'w', encoding='utf-8') as fout:
+            fout.write('\n'.join(expected))
+
+        # test the __iter__ method
+        with self.assertApiCalls(GetObject=1):
+            with smart_open.open(uri, 'r', encoding='utf-8') as fin:
+                actual = [line.rstrip() for line in fin]
+        self.assertEqual(expected, actual)
+
+        # test the __next__ method
+        with self.assertApiCalls(GetObject=1):
+            with smart_open.open(uri, 'r', encoding='utf-8') as fin:
+                first = next(fin).rstrip()
+        self.assertEqual(expected[0], first)
+
     def test_defer_seek(self):
         content = b'englishman\nin\nnew\nyork\n'
         _resource('s3').Object(BUCKET_NAME, KEY_NAME).put(Body=content)

smart_open/utils.py

@@ -215,6 +215,9 @@ def __init__(self, outer, inner):
     def __exit__(self, *args, **kwargs):
         """Exit inner after exiting outer."""
         try:
-            super().__exit__(*args, **kwargs)
+            return super().__exit__(*args, **kwargs)
         finally:
             self.__inner.__exit__(*args, **kwargs)
+
+    def __next__(self):
+        return self.__wrapped__.__next__()

smart_open/version.py

@@ -1,4 +1,4 @@
-__version__ = '7.0.1'
+__version__ = '7.0.2'
 
 
 if __name__ == '__main__':