convert all strings and code to use unicode in Python client

genie-client/src/main/python/pygenie/jobs/core.py

@@ -15,8 +15,10 @@
 from collections import defaultdict
 
 from ..conf import GenieConf
-from ..utils import (is_str,
+from ..utils import (convert_to_unicode,
+                     is_str,
                      str_to_list,
+                     unicodify,
                      uuid_str)
 from .utils import (add_to_repr,
                     arg_list,
@@ -41,6 +43,12 @@ def __init__(self, class_name=None):
         self.__repr_list = list()
 
     def __repr__(self):
+        return str(self)
+
+    def __str__(self):
+        return unicode(self).encode('utf-8')
+
+    def __unicode__(self):
         return '.'.join(self.repr_list)
 
     @staticmethod
@@ -73,12 +81,15 @@ def append(self, func_name=None, args=None, kwargs=None):
     def args_to_str(self, args):
         """Convert args tuple to string."""
 
-        return ', '.join([
-            '{qu}{val}{qu}' \
-                .format(val=a,
-                        qu=self.__quote(a) if is_str(a) else '')
-            for a in args
-        ]) if args is not None else ''
+        if args is not None:
+            results = list()
+            for arg in [convert_to_unicode(a) for a in args]:
+                results.append('{qu}{val}{qu}' \
+                    .format(val=arg,
+                            qu=self.__quote(arg) if is_str(arg) else ''))
+            return ', '.join(results)
+
+        return ''
 
     def kwargs_to_str(self, kwargs):
         """Convert kwargs dict to string."""
@@ -163,7 +174,13 @@ def __init__(self, conf=None):
         self._cluster_tag_mapping['default'] = self.default_cluster_tags
 
     def __repr__(self):
-        return str(self.repr_obj)
+        return str(self)
+
+    def __str__(self):
+        return unicode(self).encode('utf-8')
+
+    def __unicode__(self):
+        return unicode(self.repr_obj)
 
     def _add_dependency(self, dep):
         """
@@ -174,6 +191,7 @@ def _add_dependency(self, dep):
         if dep not in self._dependencies:
             self._dependencies.append(dep)
 
+    @unicodify
     def _add_cluster_tag(self, tags, priority=0):
         """
         Add a cluster tag to level. The priority is the level of precedence when
@@ -185,6 +203,7 @@ def _add_cluster_tag(self, tags, priority=0):
         # negate priority so can do sorted(self._cluster_tag_mapping.keys())
         self._cluster_tag_mapping[-int(priority)].extend(tags)
 
+    @unicodify
     def _set_command_option(self, flag, name, value=None):
         """
         Convenience method for storing an option which can later be used
@@ -198,6 +217,7 @@ def _set_command_option(self, flag, name, value=None):
 
         self._command_options[flag][name] = value
 
+    @unicodify
     @arg_list
     @add_to_repr('append')
     def applications(self, _application_ids):
@@ -250,6 +270,7 @@ def archive(self, archive):
         self._archive = archive
         return self
 
+    @unicodify
     @add_to_repr('append')
     def cluster_tags(self, cluster_tags):
         """
@@ -290,6 +311,7 @@ def cmd_args(self):
         raise GenieJobError('should not try to access core GenieJob ' \
                             'constructed command arguments')
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def command_arguments(self, _command_arguments):
@@ -307,6 +329,7 @@ def command_arguments(self, _command_arguments):
             :py:class:`GenieJob`: self
         """
 
+    @unicodify
     @arg_list
     @add_to_repr('append')
     def command_tags(self, _command_tags):
@@ -349,6 +372,7 @@ def dependencies(self, _dependencies):
             :py:class:`GenieJob`: self
         """
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def description(self, _description):
@@ -376,6 +400,7 @@ def disable_archive(self):
 
         return self.archive(False)
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def email(self, _email):
@@ -500,6 +525,7 @@ def get(self, attr, default=None):
 
         return self.to_dict().get(attr, default)
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def group(self, _group):
@@ -518,6 +544,7 @@ def group(self, _group):
             :py:class:`GenieJob`: self
         """
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def job_id(self, _job_id):
@@ -540,6 +567,7 @@ def job_id(self, _job_id):
             :py:class:`GenieJob`: self
         """
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def job_name(self, _job_name):
@@ -557,6 +585,7 @@ def job_name(self, _job_name):
             :py:class:`GenieJob`: self
         """
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def job_version(self, _job_version):
@@ -574,6 +603,7 @@ def job_version(self, _job_version):
             :py:class:`GenieJob`: self
         """
 
+    @unicodify
     @add_to_repr('append')
     def parameter(self, name, value):
         """
@@ -625,6 +655,7 @@ def parameters(self, **kwargs):
 
         return self
 
+    @unicodify
     @add_to_repr('overwrite')
     def setup_file(self, setup_file):
         """
@@ -655,6 +686,7 @@ def setup_file(self, setup_file):
 
         return self
 
+    @unicodify
     @arg_list
     @add_to_repr('append')
     def tags(self, _tags):
@@ -737,6 +769,7 @@ def to_json(self):
 
         return json.dumps(self.to_dict(), sort_keys=True, indent=4)
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def username(self, _username):

genie-client/src/main/python/pygenie/jobs/hadoop.py

@@ -11,6 +11,7 @@
 import logging
 import os
 
+from ..utils import unicodify
 from .core import GenieJob
 from .utils import (add_to_repr,
                     arg_string)
@@ -59,6 +60,7 @@ def command(self, script):
 
         return self.script(script)
 
+    @unicodify
     @add_to_repr('append')
     def property(self, name, value):
         """
@@ -85,6 +87,7 @@ def property(self, name, value):
 
         return self
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def property_file(self, _property_file):
@@ -110,6 +113,7 @@ def property_file(self, _property_file):
 
         return self
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def script(self, _script):

genie-client/src/main/python/pygenie/jobs/hive.py

@@ -11,6 +11,7 @@
 import logging
 import os
 
+from ..utils import unicodify
 from .core import GenieJob
 from .utils import (add_to_repr,
                     arg_string,
@@ -64,7 +65,7 @@ def cmd_args(self):
 
         params_str = ' '.join([
             '-d {qu}{name}={value}{qu}' \
-                .format(name=k, value=v, qu='"' if ' ' in str(v) else '') \
+                .format(name=k, value=v, qu='"' if ' ' in unicode(v) else '') \
             for k, v in self._parameters.items()
         ])
 
@@ -100,6 +101,7 @@ def headers(self):
         self.tags('headers')
         return self.hiveconf('hive.cli.print.header', 'true')
 
+    @unicodify
     @add_to_repr('append')
     def hiveconf(self, name, value):
         """
@@ -131,6 +133,7 @@ def property(self, name, value):
         """Alias for :py:meth:`HiveJob.hiveconf`"""
         return self.hiveconf(name, value)
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def property_file(self, _property_file):
@@ -162,6 +165,7 @@ def query(self, script):
 
         return self.script(script)
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def script(self, _script):

genie-client/src/main/python/pygenie/jobs/pig.py

@@ -11,6 +11,7 @@
 import logging
 import os
 
+from ..utils import unicodify
 from .core import GenieJob
 from .utils import (add_to_repr,
                     arg_list,
@@ -71,7 +72,7 @@ def cmd_args(self):
 
         params_str = ' '.join([
             '-p {qu}{name}={value}{qu}' \
-                .format(name=k, value=v, qu='"' if ' ' in str(v) else '') \
+                .format(name=k, value=v, qu='"' if ' ' in unicode(v) else '') \
             for k, v in self._parameters.items()
         ])
 
@@ -92,6 +93,7 @@ def cmd_args(self):
                     params=params_str) \
             .strip()
 
+    @unicodify
     @arg_list
     @add_to_repr('append')
     def parameter_file(self, _parameter_files):
@@ -119,6 +121,7 @@ def parameter_file(self, _parameter_files):
 
         return self
 
+    @unicodify
     @add_to_repr('append')
     def property(self, name, value):
         """
@@ -146,6 +149,7 @@ def property(self, name, value):
 
         return self
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def property_file(self, _property_file):
@@ -172,6 +176,7 @@ def property_file(self, _property_file):
 
         return self
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def script(self, _script):

genie-client/src/main/python/pygenie/jobs/presto.py

@@ -23,6 +23,7 @@
 import logging
 import os
 
+from ..utils import unicodify
 from .core import GenieJob
 from .utils import (add_to_repr,
                     arg_string,
@@ -106,6 +107,7 @@ def headers(self):
 
         return self.option('output-format', 'CSV_HEADER')
 
+    @unicodify
     @add_to_repr('append')
     def option(self, name, value=None):
         """
@@ -139,6 +141,7 @@ def query(self, script):
 
         return self.script(script)
 
+    @unicodify
     @arg_string
     @add_to_repr('overwrite')
     def script(self, _script):
@@ -159,6 +162,7 @@ def script(self, _script):
             :py:class:`PrestoJob`: self
         """
 
+    @unicodify
     @add_to_repr('append')
     def session(self, name, value):
         """

genie-client/src/main/python/pygenie/jobs/sqoop.py

@@ -12,6 +12,7 @@
 
 from collections import defaultdict
 
+from ..utils import unicodify
 from .core import GenieJob
 from .utils import add_to_repr
 
@@ -76,6 +77,7 @@ def cmd_args(self):
                     options=' '.join(opts_list)) \
             .strip()
 
+    @unicodify
     @add_to_repr('overwrite')
     def cmd(self, cmd):
         """
@@ -98,6 +100,7 @@ def cmd(self, cmd):
 
         return self
 
+    @unicodify
     @add_to_repr('append')
     def option(self, name, value=None):
         """
@@ -125,6 +128,7 @@ def option(self, name, value=None):
 
         return self
 
+    @unicodify
     @add_to_repr('append')
     def property(self, name, value, flag='-D'):
         """