Ensure that Python scripts are read in the desired encoding

waflib/Context.py

@@ -248,7 +248,7 @@ class Context(ctx):
 		if self.cur_script:
 			self.path = self.cur_script.parent
 
-	def recurse(self, dirs, name=None, mandatory=True, once=True):
+	def recurse(self, dirs, name=None, mandatory=True, once=True, encoding=None):
 		"""
 		Run user code from the supplied list of directories.
 		The directories can be either absolute, or relative to the directory
@@ -283,7 +283,7 @@ class Context(ctx):
 				cache[node] = True
 				self.pre_recurse(node)
 				try:
-					function_code = node.read('rU')
+					function_code = node.read('rU', encoding)
 					exec(compile(function_code, node.abspath(), 'exec'), self.exec_dict)
 				finally:
 					self.post_recurse(node)
@@ -294,7 +294,7 @@ class Context(ctx):
 					cache[tup] = True
 					self.pre_recurse(node)
 					try:
-						wscript_module = load_module(node.abspath())
+						wscript_module = load_module(node.abspath(), encoding=encoding)
 						user_function = getattr(wscript_module, (name or self.fun), None)
 						if not user_function:
 							if not mandatory:
@@ -597,7 +597,7 @@ Dictionary holding already loaded modules, keyed by their absolute path.
 The modules are added automatically by :py:func:`waflib.Context.load_module`
 """
 
-def load_module(path):
+def load_module(path, encoding=None):
 	"""
 	Load a source file as a python module.
 
@@ -613,7 +613,7 @@ def load_module(path):
 
 	module = imp.new_module(WSCRIPT_FILE)
 	try:
-		code = Utils.readf(path, m='rU')
+		code = Utils.readf(path, m='rU', encoding=encoding)
 	except (IOError, OSError):
 		raise Errors.WafError('Could not read the file %r' % path)
 

waflib/Task.py

@@ -484,15 +484,15 @@ class Task(TaskBase):
 		try:
 			return self.uid_
 		except AttributeError:
-			# this is not a real hot zone, but we want to avoid surprises here
 			m = Utils.md5()
 			up = m.update
-			up(self.__class__.__name__.encode())
+			up(self.__class__.__name__)
 			for x in self.inputs + self.outputs:
-				up(x.abspath().encode())
+				up(x.abspath())
 			self.uid_ = m.digest()
 			return self.uid_
 
+
 	def set_inputs(self, inp):
 		"""
 		Append the nodes to the *inputs*
@@ -824,6 +824,20 @@ class Task(TaskBase):
 				if not tsk.hasrun:
 					#print "task is not ready..."
 					raise Errors.TaskNotReady('not ready')
+if sys.hexversion > 0x3000000:
+	def uid(self):
+		try:
+			return self.uid_
+		except AttributeError:
+			m = Utils.md5()
+			up = m.update
+			up(self.__class__.__name__.encode('iso8859-1'))
+			for x in self.inputs + self.outputs:
+				up(x.abspath().encode('iso8859-1'))
+			self.uid_ = m.digest()
+			return self.uid_
+	uid.__doc__ = Task.uid.__doc__
+	Task.uid = uid
 
 def is_before(t1, t2):
 	"""

waflib/Utils.py

@@ -167,7 +167,10 @@ def readf(fname, m='r', encoding='ISO8859-1'):
 			txt = f.read()
 		finally:
 			f.close()
-		txt = txt.decode(encoding)
+		if encoding:
+			txt = txt.decode(encoding)
+		else:
+			txt = txt.decode()
 	else:
 		f = open(fname, m)
 		try:
@@ -254,7 +257,10 @@ def readf_win32(f, m='r', encoding='ISO8859-1'):
 			txt = f.read()
 		finally:
 			f.close()
-		txt = txt.decode(encoding)
+		if encoding:
+			txt = txt.decode(encoding)
+		else:
+			txt = txt.decode()
 	else:
 		f = os.fdopen(fd, m)
 		try: