Handle binary files and utf8

 * Binary files are now explicitly detected if we're unable to read them
   as utf-8 text files. In this case the last file will be used.

 * As a side affect utf-8 related things are now correctly handled
   across python 2/3.

Thank you to @unbrice as this work was based off of GH-25.

bin/dots

@@ -73,6 +73,7 @@
 
 from __future__ import print_function
 
+import codecs
 import sys
 import os
 import shutil
@@ -337,6 +338,7 @@ class ConfigFile(object):
         self.path = relative_path
         self.config = config
         self.compiled = None
+        self.compiled_binary = None
 
     def name(self):
         """Get the name of the file"""
@@ -418,11 +420,21 @@ class ConfigFile(object):
         self.compiled = None
         compiling_files = []
 
+        binary = False
+
         # Read all file paths into arrays
         # This may be memory inefficient and may have to be changed later
         for path in self.real_paths():
-            with open(path, 'r') as file:
-                contents = file.readlines()
+            with open(path, 'rb') as file:
+                contents_binary = file.read()
+
+                # Attempt to read the lines as uft-8 strings. Files content that
+                # can't be decoded as utf-8 will be considered a binary file.
+                try:
+                    contents = codecs.decode(contents_binary, 'utf-8').splitlines(True)
+                except UnicodeDecodeError:
+                    binary = True
+                    contents = ["<binary file: {}>\n".format(path)]
 
             # Replace front and back empty lines with None
             contents = self.trim_file_whitespace(contents)
@@ -442,6 +454,15 @@ class ConfigFile(object):
             self.compiled_file = ""
             return self
 
+
+        # If we have multiple binary files use the file in the last group. This
+        # will be the last file read into contents_binary
+        if binary:
+            self.compiled = ''.join(contents)
+            self.compiled_binary = contents_binary
+
+            return self
+
         compiled = compiling_files[0]
 
         # Handle merging the file_data into one file. This will look for default
@@ -470,18 +491,19 @@ class ConfigFile(object):
 
         # Remove unused append point identifiers
         self.compiled = [l for l in compiled if not l.strip().startswith(AP_IDENTIFIER)]
+        self.compiled_binary = ''.join(self.compiled).encode('utf-8')
 
         return self
 
     def diff(self, git_args=[], against=INSTALL_DIR):
         """Display a diff of the compiled file contest against the currently
         installed version of this file. This uses git as an external command to
         do the diffing"""
-        if self.compiled is None: self.compile()
+        if self.compiled_binary is None: self.compile()
 
         # Create a named temporary file to allow git to diff
-        with tempfile.NamedTemporaryFile('w+') as file:
-            file.writelines(self.compiled)
+        with tempfile.NamedTemporaryFile('wb+') as file:
+            file.write(self.compiled_binary)
             file.flush()
 
             git_diff(git_args, os.path.join(against, self.path), file.name)
@@ -507,7 +529,7 @@ class ConfigFile(object):
 
     def install(self, to, exec_install_scripts=False):
         """Install this file to a location"""
-        if not self.compiled: self.compile()
+        if not self.compiled_binary: self.compile()
 
         # Ensure directory exists for file
         path = os.path.join(to, self.path)
@@ -518,19 +540,17 @@ class ConfigFile(object):
 
         # Test that the file hasn't changed
         if os.path.exists(path):
-            compiled = ''.join(self.compiled)
-            compiled = hashlib.md5(compiled).digest()
+            compiled_hash = hashlib.md5(self.compiled_binary).digest()
 
-            with open(path, 'r') as installed_file:
-                installed = installed_file.read()
-                installed = hashlib.md5(installed).digest()
+            with open(path, 'rb') as installed_file:
+                installed_hash = hashlib.md5(installed_file.read()).digest()
 
-            if installed == compiled and os.stat(path).st_mode == self.mode():
+            if installed_hash == compiled_hash and os.stat(path).st_mode == self.mode():
                 return
 
         # Write the file
-        with open(path, 'w') as file:
-            file.writelines(self.compiled)
+        with open(path, 'wb') as file:
+            file.write(self.compiled_binary)
 
         # Set the file's mode
         os.chmod(path, self.mode())