pybiolib 1.1.1542__py3-none-any.whl → 1.1.1564__py3-none-any.whl

biolib/biolib_binary_format/file_in_container.py

@@ -1,6 +1,6 @@
-import io
 import os.path
 import tarfile
+import tempfile
 
 from docker.models.containers import Container  # type: ignore
 
@@ -32,48 +32,74 @@ class FileInContainer:
         if self._path_on_disk:
             return os.path.isfile(self._path_on_disk)
         else:
-            return self._get_file_data_from_container_via_tar() is not None
+            tmp_file = self._get_temp_file_from_container_via_tar()
+            if tmp_file:
+                os.remove(tmp_file)
+                return True
+            else:
+                return False
 
     def get_data_size_in_bytes(self) -> int:
         if self._path_on_disk:
             return os.path.getsize(self._path_on_disk)
         else:
-            file_data = self._get_file_data_from_container_via_tar()
-            return len(file_data) if file_data else 0
+            tmp_file = self._get_temp_file_from_container_via_tar()
+            if tmp_file:
+                file_length = os.path.getsize(tmp_file)
+                os.remove(tmp_file)
+                return file_length
+            else:
+                return 0
 
     def get_data_stream(self) -> Iterable[bytes]:
         if self._path_on_disk:
             with open(self._path_on_disk, mode='rb') as file:
                 while True:
-                    chunk = file.read(1_024)
+                    chunk = file.read(1_000_000)
                     if not chunk:
                         return
 
                     yield chunk
         else:
-            yield self._get_file_data_from_container_via_tar() or bytes()
-
-    def _get_file_data_from_container_via_tar(self) -> Optional[bytes]:
-        if not self._buffered_file_data:
-            with io.BytesIO() as tmp_io:
-                stream, _ = self._container.get_archive(path=self._path_in_container)
-                for chunk in stream:
-                    tmp_io.write(chunk)
-
-                tmp_io.seek(0)
-                with tarfile.open(fileobj=tmp_io) as tar:
-                    members = tar.getmembers()
-                    file_members = [member for member in members if member.isfile()]
-                    if not file_members:
-                        # Path was not a file
-                        return None
-
-                    assert len(file_members) == 1
-                    file_obj = tar.extractfile(member=file_members[0])
-                    if not file_obj:
-                        # Path was not a file
-                        return None
-
-                    self._buffered_file_data = file_obj.read()
-
-        return self._buffered_file_data
+            tmp_file = self._get_temp_file_from_container_via_tar()
+            if not tmp_file:
+                yield bytes()
+                return
+            else:
+                file = open(tmp_file, mode='rb')
+                while True:
+                    chunk = file.read(1_000_000)
+                    if not chunk:
+                        file.close()
+                        os.remove(tmp_file)
+                        return
+
+                    yield chunk
+
+    def _get_temp_file_from_container_via_tar(self) -> Optional[str]:
+        with tempfile.NamedTemporaryFile(mode='w+', delete=True) as tmp_io:
+            stream, _ = self._container.get_archive(path=self._path_in_container)
+            for chunk in stream:
+                tmp_io.write(chunk)
+
+            tmp_io.seek(0)
+            with tarfile.open(tmp_io.name) as tar:
+                members = tar.getmembers()
+                file_members = [member for member in members if member.isfile()]
+                if not file_members:
+                    # Path was not a file
+                    return None
+
+                assert len(file_members) == 1
+                file_obj = tar.extractfile(member=file_members[0])
+                if not file_obj:
+                    # Path was not a file
+                    return None
+                tmp_io_2 = tempfile.NamedTemporaryFile(mode='w+', delete=False)
+                while True:
+                    chunk = file_obj.read(1_000_000)
+                    if not chunk:
+                        break
+                    tmp_io_2.write(chunk)
+                tmp_io_2.close()
+                return tmp_io_2.name

{pybiolib-1.1.1542.dist-info → pybiolib-1.1.1564.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: pybiolib
-Version: 1.1.1542
+Version: 1.1.1564
 Summary: BioLib Python Client
 Home-page: https://github.com/biolib
 License: MIT

{pybiolib-1.1.1542.dist-info → pybiolib-1.1.1564.dist-info}/RECORD

@@ -21,7 +21,7 @@ biolib/biolib_api_client/lfs_types.py,sha256=xaGjE-yUyNVM3LyKdslJn5ZXWp6_kVCd4o-
 biolib/biolib_api_client/user_state.py,sha256=XcgWV-MgVk88mIlMmnu8yHxMu8OCaw8o0tk7TVo5Hcg,637
 biolib/biolib_binary_format/__init__.py,sha256=HMl5SdX_VUWE4OQzi4Jf_yFvC7b0bSPOGPHYi9dWM2Q,185
 biolib/biolib_binary_format/base_bbf_package.py,sha256=vxRV4iKy0dXeDOlFWnMFI0hGnDBYDH5Cgh5gAfuObt8,959
-biolib/biolib_binary_format/file_in_container.py,sha256=sgj0YWPuWhNOiFsVf4X7Sf2AYO4kGCHj4DSUceCifJ0,2753
+biolib/biolib_binary_format/file_in_container.py,sha256=oAXva2pVEwHs175zjnPAibqTCSkseD6ZVxmOwh8pq48,3591
 biolib/biolib_binary_format/module_input.py,sha256=led2QhHeec_ymBPw5uEn3_3vJKI-1T8zrFQGqwEWLMY,2788
 biolib/biolib_binary_format/module_output_v2.py,sha256=J5ZO5gCSeudpE12EVDrjYrNTS2DwgszY-SVXT7Qjuyg,5913
 biolib/biolib_binary_format/remote_endpoints.py,sha256=LgWd_cewZK4ncPR8303GM_PmIEZZJ5RQatJ_NSXUgbU,1306
@@ -99,8 +99,8 @@ biolib/utils/cache_state.py,sha256=BFrZlV4XZIueIFzAFiPidX4hmwADKY5Y5ZuqlerF5l0,3
 biolib/utils/multipart_uploader.py,sha256=PEorMsTNg5f72e3Y-KA3LrI-F6EAnsD9vENV-5EuiX0,9896
 biolib/utils/seq_util.py,sha256=g4PIw9ovxjO0eg85uBSDRzzvxPkxu9pOT5sEryGD4JA,3105
 biolib/utils/zip/remote_zip.py,sha256=NCdUnVbGCv7SfXCI-yVU-is_OnyWmLAnVpIdSvo-W4k,23500
-pybiolib-1.1.1542.dist-info/LICENSE,sha256=F2h7gf8i0agDIeWoBPXDMYScvQOz02pAWkKhTGOHaaw,1067
-pybiolib-1.1.1542.dist-info/METADATA,sha256=wEVBcb7SQUAdsZ8pQSB6kwdVcAvAC5yzkLZncvmf-3Q,1543
-pybiolib-1.1.1542.dist-info/WHEEL,sha256=FMvqSimYX_P7y0a7UY-_Mc83r5zkBZsCYPm7Lr0Bsq4,88
-pybiolib-1.1.1542.dist-info/entry_points.txt,sha256=p6DyaP_2kctxegTX23WBznnrDi4mz6gx04O5uKtRDXg,42
-pybiolib-1.1.1542.dist-info/RECORD,,
+pybiolib-1.1.1564.dist-info/LICENSE,sha256=F2h7gf8i0agDIeWoBPXDMYScvQOz02pAWkKhTGOHaaw,1067
+pybiolib-1.1.1564.dist-info/METADATA,sha256=HnHMpug5D5I6xVtQrZT71yShKima3sLI1ZWGHMI9yBE,1543
+pybiolib-1.1.1564.dist-info/WHEEL,sha256=FMvqSimYX_P7y0a7UY-_Mc83r5zkBZsCYPm7Lr0Bsq4,88
+pybiolib-1.1.1564.dist-info/entry_points.txt,sha256=p6DyaP_2kctxegTX23WBznnrDi4mz6gx04O5uKtRDXg,42
+pybiolib-1.1.1564.dist-info/RECORD,,