Merge pull request #488 from tkmtnt7000/fix-python3-ros-google-cloud

[ros_google_cloud_language] Fix for python3; Avoid attribute error

ros_google_cloud_language/node_scripts/analyze_text.py

@@ -8,6 +8,11 @@
 import actionlib
 import rospy
 import os
+import sys
+try:
+    from unidecode import unidecode
+except ImportError:
+    pass
 
 from ros_google_cloud_language.msg import AnalyzeTextAction
 from ros_google_cloud_language.msg import AnalyzeTextResult
@@ -50,13 +55,21 @@ def execute_cb(self, goal):
                  encoding_type='UTF32',
             )
             for entity in response.entities:
+                if sys.version_info.major < 3:
+                    # To avoid AttributeError 'bytes' object has no attribute 'encode' in python3 #NOQA
+                    entity.name = entity.name.encode('utf-8')
+                else:
+                    # Forcefully changed to ascii string because ros message only supports ascii character #NOQA
+                    entity.name = unidecode(entity.name)
                 result.entities.append(TextEntity(
-                    name=entity.name.encode('utf-8'),
+                    name=entity.name,
                     type=entity.type,
-                    metadata=map(lambda kv: KeyValue(kv[0], kv[1]),
-                                 entity.metadata.items()),
+                    # 'map' function is different from python2 and python3
+                    metadata=[
+                        KeyValue(kv[0], kv[1])
+                        for kv in entity.metadata.items()
+                    ],
                     salience=entity.salience))
-
             # Detects the sentiment of the text
             sentiment = self._client.analyze_sentiment(
                 document=document,
@@ -74,9 +87,17 @@ def execute_cb(self, goal):
 
             for token in syntax.tokens:
                 # print("{} {} {}".format(token.part_of_speech.tag, token.text.content.encode('utf-8'), token.dependency_edge.head_token_index)
+                if sys.version_info.major < 3:
+                    # To avoid AttributeError 'bytes' object has no attribute 'encode' in python3 #NOQA
+                    token.text.content = token.text.content.encode('utf-8')
+                    token.lemma = token.lemma.encode('utf-8')
+                else:
+                    # Forcefully changed to ascii string because ros message only supports ascii character #NOQA
+                    token.text.content = unidecode(token.text.content)
+                    token.lemma = unidecode(token.lemma)
                 result.syntaxes.append(TextSyntax(
-                    name=token.text.content.encode('utf-8'),
-                    lemma=token.lemma.encode('utf-8'),
+                    name=token.text.content,
+                    lemma=token.lemma,
                     dependency_edge=token.dependency_edge.head_token_index,
                     part_of_speech=token.part_of_speech.tag,
                     parse_label=token.dependency_edge.label

ros_google_cloud_language/package.xml

@@ -15,6 +15,7 @@
 
   <exec_depend>message_runtime</exec_depend>
   <exec_depend>rospy</exec_depend>
+  <exec_depend condition="$ROS_PYTHON_VERSION == 3">python3-unidecode</exec_depend>
 
   <depend>actionlib_msgs</depend>
   <depend>diagnostic_msgs</depend>