FIX: UnicodeEncodeError when importing the case 0809, this case has non-ascii characters

tfmoraes
1 parent 9bc7d016
Showing 4 changed files with 52 additions and 28 deletions Show diff stats
invesalius/data/vtk_utils.py
invesalius/gui/import_panel.py
invesalius/project.py
invesalius/reader/dicom.py
@@ -129,8 +129,10 @@ class Text(object):
             value = str(value)
             if sys.platform == 'win32':
                 value += "" # Otherwise 0 is not shown under win32
-            
-        self.mapper.SetInput(str(value))
+        # With some encoding in some dicom fields (like name) raises a
+        # UnicodeEncodeError because they have non-ascii characters. To avoid
+        # that we encode in utf-8.
+        self.mapper.SetInput(value.encode("utf-8"))
     def SetPosition(self, position):
         self.actor.GetPositionCoordinate().SetValue(position[0],
@@ -160,17 +160,17 @@ class TextPanel(wx.Panel):
                 first += 1
             tree.SetItemPyData(parent, patient)
-            tree.SetItemText(parent, str(dicom.patient.id), 1)
-            tree.SetItemText(parent, str(dicom.patient.age), 2)
-            tree.SetItemText(parent, str(dicom.patient.gender), 3)
-            tree.SetItemText(parent, str(dicom.acquisition.study_description), 4)
-            tree.SetItemText(parent, str(dicom.acquisition.modality), 5)
-            tree.SetItemText(parent, str(date_time), 6)
-            tree.SetItemText(parent, str(patient.nslices), 7)
-            tree.SetItemText(parent, str(dicom.acquisition.institution), 8)
-            tree.SetItemText(parent, str(dicom.patient.birthdate), 9)
-            tree.SetItemText(parent, str(dicom.acquisition.accession_number), 10)
-            tree.SetItemText(parent, str(dicom.patient.physician), 11)
+            tree.SetItemText(parent, "%s" % dicom.patient.id, 1)
+            tree.SetItemText(parent, "%s" % dicom.patient.age, 2)
+            tree.SetItemText(parent, "%s" % dicom.patient.gender, 3)
+            tree.SetItemText(parent, "%s" % dicom.acquisition.study_description, 4)
+            tree.SetItemText(parent, "%s" % dicom.acquisition.modality, 5)
+            tree.SetItemText(parent, "%s" % date_time, 6)
+            tree.SetItemText(parent, "%s" % patient.nslices, 7)
+            tree.SetItemText(parent, "%s" % dicom.acquisition.institution, 8)
+            tree.SetItemText(parent, "%s" % dicom.patient.birthdate, 9)
+            tree.SetItemText(parent, "%s" % dicom.acquisition.accession_number, 10)
+            tree.SetItemText(parent, "%s" % dicom.patient.physician, 11)
             group_list = patient.GetGroups()
             for group in group_list:
@@ -179,11 +179,11 @@ class TextPanel(wx.Panel):
                 child = tree.AppendItem(parent, group.title)
                 tree.SetItemPyData(child, group)
-                tree.SetItemText(child, str(group.title), 0)
-                tree.SetItemText(child, str(dicom.acquisition.protocol_name), 4)
-                tree.SetItemText(child, str(dicom.acquisition.modality), 5)
-                tree.SetItemText(child, str(date_time), 6)
-                tree.SetItemText(child, str(group.nslices), 7)
+                tree.SetItemText(child, "%s" % group.title, 0)
+                tree.SetItemText(child, "%s" % dicom.acquisition.protocol_name, 4)
+                tree.SetItemText(child, "%s" % dicom.acquisition.modality, 5)
+                tree.SetItemText(child, "%s" % date_time, 6)
+                tree.SetItemText(child, "%s" % group.nslices, 7)
         tree.Expand(self.root)
@@ -157,22 +157,27 @@ class Project(object):
         ps.Publisher.sendMessage('Set raycasting preset', preset)
     def SavePlistProject(self, dir_, filename):
-        
+
+        # Some filenames have non-ascii characters and encoded in a strange
+        # encoding, in that cases a UnicodeEncodeError is raised. To avoid
+        # that we encode in utf-8.
+        filename = filename.encode('utf-8')
         dir_temp = tempfile.mkdtemp(filename)
         filename_tmp = os.path.join(dir_temp, filename)
-        
+
         project = {}
-        
+
         for key in self.__dict__:
             if getattr(self.__dict__[key], 'SavePlist', None):
-                project[key] = {'#plist': self.__dict__[key].SavePlist(filename_tmp)}
+                project[key] = {'#plist':
+                                self.__dict__[key].SavePlist(filename_tmp).decode('utf-8')}
             else:
                 project[key] = self.__dict__[key]
         masks = {}
         for index in self.mask_dict:
             masks[str(index)] = {'#mask':\
-                                 self.mask_dict[index].SavePlist(filename_tmp)}
+                                 self.mask_dict[index].SavePlist(filename_tmp).decode('utf-8')}
             print index
         surfaces = {}
@@ -185,9 +190,10 @@ class Project(object):
         project['mask_dict'] = masks
         img_file = '%s_%s.vti' % (filename_tmp, 'imagedata')
         iu.Export(self.imagedata, img_file, bin=True)
-        project['imagedata'] = {'$vti':os.path.split(img_file)[1]}
+        project['imagedata'] = {'$vti':os.path.split(img_file)[1].decode('utf-8')}
+
         plistlib.writePlist(project, filename_tmp + '.plist')
-        
+
         path = os.path.join(dir_,filename)
         Compress(dir_temp, path)#os.path.join("~/Desktop/","teste.inv3"))
         shutil.rmtree(dir_temp)
@@ -90,6 +90,7 @@ class Parser():
     def __init__(self):
         self.filename = ""
+        self.encoding = ""
         self.vtkgdcm_reader = vtkgdcm.vtkGDCMImageReader()
     def GetAcquisitionDate(self):
@@ -1136,7 +1137,8 @@ class Parser():
         data = self.vtkgdcm_reader.GetMedicalImageProperties()\
                                             .GetPatientName()
         if (data):
-            return data.strip()
+            # Returns a unicode decoded in the own dicom encoding
+            return data.strip().decode(self.GetEncoding())
         return ""
     def GetPatientID(self):
@@ -1150,7 +1152,8 @@ class Parser():
         data = self.vtkgdcm_reader.GetMedicalImageProperties()\
                                                 .GetPatientID()
         if (data):
-            return data
+            # Returns a unicode decoded in the own dicom encoding
+            return data.decode(self.GetEncoding())
         return ""
@@ -1486,7 +1489,20 @@ class Parser():
             data = str(ds.GetDataElement(tag).GetValue())
             if (data):
                 return data
-        return ""        
+        return ""
+
+    def GetEncoding(self):
+        """
+        Return the dicom encoding
+        DICOM standard tag (0x0008, 0x0005) was used.
+        """
+        if self.encoding:
+            return self.encoding
+        tag = gdcm.Tag(0x0008, 0x0005)
+        ds = self.gdcm_reader.GetFile().GetDataSet()
+        if ds.FindDataElement(tag):
+            self.encoding = str(ds.GetDataElement(tag).GetValue())
+            return self.encoding
 class DicomWriter: