Don't change file encoding

Author: GeorgianaElena

repo2docker/buildpacks/conda/__init__.py

@@ -6,7 +6,7 @@
 from ruamel.yaml import YAML
 
 from ..base import BaseImage
-from ...utils import is_local_pip_requirement, open_utf8convert_read
+from ...utils import is_local_pip_requirement
 
 # pattern for parsing conda dependency line
 PYTHON_REGEX = re.compile(r"python\s*=+\s*([\d\.]*)")
@@ -140,7 +140,7 @@ def environment_yaml(self):
             self._environment_yaml = {}
             return self._environment_yaml
 
-        with open_utf8convert_read(environment_yml) as f:
+        with open(environment_yml) as f:
             env = YAML().load(f)
             # check if the env file is empty, if so instantiate an empty dictionary.
             if env is None:

repo2docker/buildpacks/python/__init__.py

@@ -2,7 +2,7 @@
 import os
 
 from ..conda import CondaBuildPack
-from ...utils import is_local_pip_requirement, open_utf8convert_read
+from ...utils import is_local_pip_requirement, open_guess_encoding
 
 
 class PythonBuildPack(CondaBuildPack):
@@ -86,7 +86,7 @@ def _should_preassemble_pip(self):
             requirements_txt = self.binder_path(name)
             if not os.path.exists(requirements_txt):
                 continue
-            with open_utf8convert_read(requirements_txt) as f:
+            with open_guess_encoding(requirements_txt) as f:
                 for line in f:
                     if is_local_pip_requirement(line):
                         return False

repo2docker/utils.py

@@ -70,17 +70,23 @@ def chdir(path):
     finally:
         os.chdir(old_dir)
 
+
 @contextmanager
-def open_utf8convert_read(path):
+def open_guess_encoding(path):
+    """
+    Open a file in text mode, specifying its encoding,
+    that we guess using chardet.
+    """
+    detector = chardet.universaldetector.UniversalDetector()
     with open(path, "rb") as f:
-        file_to_encode = f.read()
-
-    encoding_detection_result = chardet.detect(file_to_encode)
-    if not "utf-8" in encoding_detection_result:
-        with open(path, "wb") as f:
-            f.write(file_to_encode.decode(encoding_detection_result["encoding"]).encode("utf-8"))
-
-    file = open(path)
+        for line in f.readlines():
+            detector.feed(line)
+            print(str(i) + str(detector.done))
+            if detector.done:
+                break
+    detector.close()
+
+    file = open(path, encoding=detector.result["encoding"])
     try:
         yield file
     finally: