airbyte-source-s3 4.14.4__py3-none-any.whl → 4.14.5.dev202510081800__py3-none-any.whl

{airbyte_source_s3-4.14.4.dist-info → airbyte_source_s3-4.14.5.dev202510081800.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: airbyte-source-s3
-Version: 4.14.4
+Version: 4.14.5.dev202510081800
 Summary: Source implementation for S3.
 Home-page: https://airbyte.com
 License: ELv2
@@ -15,7 +15,7 @@ Requires-Dist: airbyte-cdk[file-based] (>=7.0.4,<8.0.0)
 Requires-Dist: dill (>=0.3.4,<0.4.0)
 Requires-Dist: pendulum (>=3.0.0,<4.0.0)
 Requires-Dist: pytz (>=2024.2,<2025.0)
-Requires-Dist: smart-open[s3] (==4.14.4)
+Requires-Dist: smart-open[s3] (==4.14.5.dev.202510081800)
 Requires-Dist: transformers (>=4.38.2,<5.0.0)
 Requires-Dist: urllib3 (<2)
 Requires-Dist: wcmatch (>=10.0,<11.0)

{airbyte_source_s3-4.14.4.dist-info → airbyte_source_s3-4.14.5.dev202510081800.dist-info}/RECORD

@@ -18,8 +18,8 @@ source_s3/v4/cursor.py,sha256=kZj_6Wtl2yVAVeKZW67XyYQOk2XtbYazTgswEgGfKnI,7298
 source_s3/v4/legacy_config_transformer.py,sha256=OjKwGBYPHZvhZRKGO1LOAR7-cAT-9KvDRQY-G93eoic,7840
 source_s3/v4/source.py,sha256=jugIY53C_G9QhQRwKWBPcXXUgYKF_RESSaewzF_HXhc,8967
 source_s3/v4/stream_reader.py,sha256=7AhFnOkkynZpAec3SD6c9pT-sCS7eeH8oKpDpgz8sCA,16837
-source_s3/v4/zip_reader.py,sha256=5Atoxu3Si0w5fNT6l_qQ-C1ZlKYt2DgV4Lxf8tTU5hw,14608
-airbyte_source_s3-4.14.4.dist-info/METADATA,sha256=LmVcnYTe5jwLWMT3EnJN_6pQpu3wpZUvWpaq11jXi-E,5365
-airbyte_source_s3-4.14.4.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
-airbyte_source_s3-4.14.4.dist-info/entry_points.txt,sha256=1ndXy1aXftjfCR82yLKg797X8-O5mD95tz6rNKA_1dE,47
-airbyte_source_s3-4.14.4.dist-info/RECORD,,
+source_s3/v4/zip_reader.py,sha256=GFDnqv1tYxiSpVmMUObYX55xXiZ0MNrirbfH83gTFdw,15390
+airbyte_source_s3-4.14.5.dev202510081800.dist-info/METADATA,sha256=djWQoFbhRHT_M3k76MXtUK8vfe7_dZLyek64qx2-soQ,5398
+airbyte_source_s3-4.14.5.dev202510081800.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
+airbyte_source_s3-4.14.5.dev202510081800.dist-info/entry_points.txt,sha256=1ndXy1aXftjfCR82yLKg797X8-O5mD95tz6rNKA_1dE,47
+airbyte_source_s3-4.14.5.dev202510081800.dist-info/RECORD,,

source_s3/v4/zip_reader.py

@@ -366,7 +366,21 @@ class ZipContentReader:
         data = self.buffer[:size]
         self.buffer = self.buffer[size:]
 
-        return data.decode(self.encoding) if self.encoding else bytes(data)
+        try:
+            return data.decode(self.encoding) if self.encoding else bytes(data)
+        except UnicodeDecodeError:
+            if self.encoding == "utf_8_sig":
+                # utf_8_sig considers `\xef\xbb\xbf` as a single character and therefore calling `bytearray(b'\xef').decode("utf_8_sig") will
+                # cause an exception to be raised.
+                number_of_bytes_to_add = size - 1
+                if data.endswith(bytearray(b"\xef")):
+                    number_of_bytes_to_add += 2
+                elif data.endswith(bytearray(b"\xbb")):
+                    number_of_bytes_to_add += 1
+                data = data + self.buffer[:number_of_bytes_to_add]
+                self.buffer = self.buffer[number_of_bytes_to_add:]
+                return data.decode(self.encoding) if self.encoding else bytes(data)
+            raise
 
     def seek(self, offset: int, whence: int = io.SEEK_SET) -> int:
         """