fix bugs of issue https://github.com/neopenx/Dragon/issues/5

Dragon/modules/python/dragon.cc

@@ -132,15 +132,14 @@ bool SwitchWorkspaceInternal(const string& name, const bool create_if_missing) {
 }
 }
 
 
 PyObject* SwitchWorkspaceCC(PyObject* self, PyObject *args) {
 PyObject* SwitchWorkspaceCC(PyObject* self, PyObject *args) {
-    PyObject* name = nullptr;
+    char* cname;
     PyObject* create_if_missing = nullptr;
     PyObject* create_if_missing = nullptr;
-    if (!PyArg_ParseTuple(args, "S|O", &name, &create_if_missing)) {
+    if (!PyArg_ParseTuple(args, "s|O", &cname, &create_if_missing)) {
         PyErr_SetString(PyExc_ValueError, "SwitchWorkspaceCC takes a workspace name and a optional "
         PyErr_SetString(PyExc_ValueError, "SwitchWorkspaceCC takes a workspace name and a optional "
                                           "bool value that specific whether to create the workspace if missing.");
                                           "bool value that specific whether to create the workspace if missing.");
         return nullptr;
         return nullptr;
     }
     }
-    bool success = SwitchWorkspaceInternal(PyBytesToStdString(name), 
-                                           PyObject_IsTrue(create_if_missing));
+    bool success = SwitchWorkspaceInternal(string(cname), PyObject_IsTrue(create_if_missing));
     if (!success) {
     if (!success) {
         PyErr_SetString(PyExc_RuntimeError, "workspace of the given name is not exist and "
         PyErr_SetString(PyExc_RuntimeError, "workspace of the given name is not exist and "
                                             "is not allowed to create. (try alllow ?)");
                                             "is not allowed to create. (try alllow ?)");
@@ -162,15 +161,16 @@ PyObject* WorkspacesCC(PyObject* self, PyObject* args) {
 }
 }
 
 
 PyObject* ResetWorkspaceCC(PyObject* self, PyObject* args) {
 PyObject* ResetWorkspaceCC(PyObject* self, PyObject* args) {
-    PyObject* root_folder = nullptr;
-    if (!PyArg_ParseTuple(args, "|S", &root_folder)) {
+    char* cname;
+    if (!PyArg_ParseTuple(args, "|s", &cname)) {
         PyErr_SetString(PyExc_ValueError, "ResetWorkspaceCC takes in either no args or a string "
         PyErr_SetString(PyExc_ValueError, "ResetWorkspaceCC takes in either no args or a string "
-                                          "specifing the root folder of the workspace.");
+                                          "specifing the name of the workspace.");
         return nullptr;
         return nullptr;
     }
     }
     LOG(INFO) << "Reset the Workspace(" << g_current_workspace << ")";
     LOG(INFO) << "Reset the Workspace(" << g_current_workspace << ")";
-    if (root_folder == nullptr) g_workspaces[g_current_workspace].reset(new Workspace());
-    else g_workspaces[g_current_workspace].reset(new Workspace(PyBytesToStdString(root_folder)));
+	string workspace_name = string(cname);
+	if (workspace_name.empty()) g_workspaces[g_current_workspace].reset(new Workspace());
+    else g_workspaces[g_current_workspace].reset(new Workspace(workspace_name));
     g_workspace = g_workspaces[g_current_workspace].get();
     g_workspace = g_workspaces[g_current_workspace].get();
     Py_RETURN_TRUE;
     Py_RETURN_TRUE;
 }
 }
@@ -343,12 +343,12 @@ PyObject* SnapshotCC(PyObject* self, PyObject* args) {
         return nullptr;
         return nullptr;
     }
     }
     switch (format) {
     switch (format) {
-        case 0:     // cPickle
+        case 0:    //  cPickle
             PyErr_SetString(PyExc_NotImplementedError, "format(0) depends on cPickle, should not be used in CC.");
             PyErr_SetString(PyExc_NotImplementedError, "format(0) depends on cPickle, should not be used in CC.");
             break;
             break;
-        case 1:     // caffemodel
+        case 1:    //  caffemodel
             for (int i = 0; i < PyList_Size(names); i++)
             for (int i = 0; i < PyList_Size(names); i++)
-                tensors.push_back(g_workspace->GetTensor(PyBytesToStdString(PyList_GetItem(names, i))));
+                tensors.push_back(g_workspace->GetTensor(PyString_AsString(PyList_GetItem(names, i))));
             SavaCaffeModel(string(cname), tensors);
             SavaCaffeModel(string(cname), tensors);
             break;
             break;
         default: LOG(FATAL) << "Unknwon Restore Format, code: " << format;
         default: LOG(FATAL) << "Unknwon Restore Format, code: " << format;

Dragon/modules/python/dragon.h

@@ -21,6 +21,7 @@
 
 
 #ifdef WITH_PYTHON3
 #ifdef WITH_PYTHON3
 #define PyString_AsString PyUnicode_AsUTF8
 #define PyString_AsString PyUnicode_AsUTF8
+#define PyBytes_FromStringAndSize PyUnicode_FromStringAndSize
 #endif
 #endif
 
 
 using namespace dragon;
 using namespace dragon;

Dragon/python/dragon/config.py

@@ -69,9 +69,4 @@ def SetLoggingLevel(level):
         'WARNING': logging.WARNING,
         'WARNING': logging.WARNING,
         'ERROR': logging.ERROR,
         'ERROR': logging.ERROR,
         'FATAL': logging.CRITICAL
         'FATAL': logging.CRITICAL
-    }[level])
-
+    }[level])
\ No newline at end of file
-
-
-
-

Dragon/python/dragon/core/gradient_maker.py

@@ -17,9 +17,8 @@ class GraphGradientMaker(object):
     def CreateGradientForOp(cls, op_def, g_output):
     def CreateGradientForOp(cls, op_def, g_output):
         """ parse ops from string """
         """ parse ops from string """
         g_ops, g_inputs, defaults = CreateGradientDefsCC(op_def.SerializeToString(), g_output)
         g_ops, g_inputs, defaults = CreateGradientDefsCC(op_def.SerializeToString(), g_output)
-        if sys.version_info >= (3, 0):
-            g_inputs = [g_input.decode('ascii') for g_input in g_inputs]
         for idx, g_op in enumerate(g_ops):
         for idx, g_op in enumerate(g_ops):
+            if sys.version_info >= (3, 0): g_op = g_op.encode()
             new_def = pb.OperatorDef()
             new_def = pb.OperatorDef()
             new_def.ParseFromString(g_op)
             new_def.ParseFromString(g_op)
             _, new_def.name = GetOperatorName()
             _, new_def.name = GetOperatorName()

Dragon/python/dragon/core/scope.py

@@ -36,7 +36,7 @@ def GetTensorName():
 class TensorScope(object):
 class TensorScope(object):
     SEPARATOR = '/'
     SEPARATOR = '/'
     def __init__(self, prefix):
     def __init__(self, prefix):
-        assert isinstance(prefix, basestring), \
+        assert isinstance(prefix, type('str')), \
             "TensorScope takes in a string as its argument."
             "TensorScope takes in a string as its argument."
         self.prefix = prefix + TensorScope.SEPARATOR
         self.prefix = prefix + TensorScope.SEPARATOR
 
 
@@ -51,7 +51,7 @@ class TensorScope(object):
 
 
 class PhaseScope(object):
 class PhaseScope(object):
     def __init__(self, phase):
     def __init__(self, phase):
-        assert isinstance(phase, basestring), \
+        assert isinstance(phase, type('str')), \
             "PhaseScope takes in a string as its argument."
             "PhaseScope takes in a string as its argument."
         self.phase = phase
         self.phase = phase
 
 

Dragon/python/dragon/tools/db.py

@@ -49,7 +49,7 @@ class LMDB(object):
 
 
 
 
     def put(self, key, value):
     def put(self, key, value):
-        self._buffer.append((wrapper_str(key), wrapper_str(value)))
+        self._buffer.append((wrapper_str(key), value))
         self._cur_put += 1
         self._cur_put += 1
         if (self._cur_put >= self._max_commit): self._try_put()
         if (self._cur_put >= self._max_commit): self._try_put()
 
 

Dragon/python/dragon/vm/caffe/layers/common.py

@@ -260,7 +260,7 @@ class BNLayer(Layer):
                        'eps': param.eps}
                        'eps': param.eps}
         mean = Tensor(LayerParameter.name + '@param0').Constant()
         mean = Tensor(LayerParameter.name + '@param0').Constant()
         var = Tensor(LayerParameter.name + '@param1').Constant()
         var = Tensor(LayerParameter.name + '@param1').Constant()
-        scale = Tensor(LayerParameter.name + '@param2').Uniform(low=0.0, high=1.0)
+        scale = Tensor(LayerParameter.name + '@param2').Constant(value=1.0)
         bias = Tensor(LayerParameter.name + '@param3').Constant(value=0.0)
         bias = Tensor(LayerParameter.name + '@param3').Constant(value=0.0)
         self.norm_blobs = [{'data': mean, 'diff': None},
         self.norm_blobs = [{'data': mean, 'diff': None},
                            {'data': var, 'diff': None}]
                            {'data': var, 'diff': None}]