From 5d1e5f08040c622af72596335d8e924695349efb Mon Sep 17 00:00:00 2001
From: kznr <who.is.kazunari@gmail.com>
Date: Thu, 20 May 2021 23:23:27 +0900
Subject: [PATCH] Resolve issue 207 and issue 226

It aims to avoid the unnecessary conversion by the decode method.

The original hint was provided by Mr. Lutz K.
Thank you very much
---
 src/harvesters/core.py                     | 32 ++++++++++++----------
 src/harvesters/test/test_harvester_core.py | 17 ++++++++++++
 2 files changed, 34 insertions(+), 15 deletions(-)

diff --git a/src/harvesters/core.py b/src/harvesters/core.py
index 00d3f98..03922c0 100644
--- a/src/harvesters/core.py
+++ b/src/harvesters/core.py
@@ -3026,8 +3026,6 @@ def _release_buffers(self) -> None:
             _ = self._queue.get_nowait()
 
 
-
-
 def _save_file(
         *,
         xml_dir_to_store: Optional[str] = None,
@@ -3057,26 +3055,20 @@ def _save_file(
     file_path = os.path.join(xml_dir_to_store, _file_name)
 
     #
-    mode_base = 'w+'
-    mode = mode_base
-    data_to_write = bytes_content = bytes_io.getvalue()
-    if pathlib.Path(file_path).suffix.lower() == '.zip':
-        mode += 'b'
-    else:
-        data_to_write = bytes_content.decode()
-        pos = data_to_write.find('\x00')
-        if pos != -1:
-            # Found a \x00:
-            data_to_write = data_to_write[:pos]
+    mode = 'w+'
+    data_to_write = bytes_io.getvalue()
+    if pathlib.Path(file_path).suffix.lower() != '.zip':
+        data_to_write = _drop_unnecessary_trailer(data_to_write)
+
     #
     try:
-        with open(file_path, mode) as f:
+        with open(file_path, mode + 'b') as f:
             f.write(data_to_write)
     except UnicodeEncodeError:
         # Probably you've caught "UnicodeEncodeError: 'charmap' codec can't
         # encode characters"; the file must be a text file:
         try:
-            with io.open(file_path, mode_base, encoding="utf-8") as f:
+            with io.open(file_path, mode, encoding="utf-8") as f:
                 f.write(data_to_write)
         except:
             e = sys.exc_info()[0]
@@ -3090,6 +3082,16 @@ def _save_file(
     return file_path
 
 
+def _drop_unnecessary_trailer(data_to_write: bytes):
+    assert data_to_write
+    pos = data_to_write.find(0x00)
+    if pos != -1:
+        # Found a \x00:
+        return data_to_write[:pos]
+    else:
+        return data_to_write
+
+
 class _CallbackDestroyImageAcquirer(Callback):
     def __init__(self, harvester):
         self._harvester = harvester
diff --git a/src/harvesters/test/test_harvester_core.py b/src/harvesters/test/test_harvester_core.py
index b921d6e..d457bdc 100644
--- a/src/harvesters/test/test_harvester_core.py
+++ b/src/harvesters/test/test_harvester_core.py
@@ -40,6 +40,7 @@
 from harvesters.core import Callback
 from harvesters.core import Harvester
 from harvesters.core import ImageAcquirer
+from harvesters.core import _drop_unnecessary_trailer
 from harvesters.test.helper import get_package_dir
 from harvesters.util.pfnc import Dictionary
 from harvesters.core import Component2DImage
@@ -1011,5 +1012,21 @@ def test_issue_238(self):
                 )
 
 
+class TestUtility(unittest.TestCase):
+    def test_issue_207_and_226(self):
+        body = b'\xc2\xb0'  # °
+        padding = b'\x00\x00'
+        data = body + padding
+        data = _drop_unnecessary_trailer(data)
+        self.assertEqual(data, body)
+        self.assertEqual('°', str(data, encoding='utf-8'))
+
+    @unittest.skip  # experimental
+    def test_issue_207_and_226_in_orignal_way(self):
+        data = b'\xc2\xb0'  # °
+        decoded = data.decode()
+        self.assertEqual('°', decoded)
+
+
 if __name__ == '__main__':
     unittest.main()