Merge pull request #1572 from ssanderson/more-unicode-fun

BUG: Fix unicode errors for py2 client and py3 server.

blaze/compatibility.py

@@ -4,6 +4,7 @@
 from types import MethodType
 
 import pandas.util.testing as tm
+from toolz import identity
 
 PY3 = sys.version_info[0] == 3
 PY2 = sys.version_info[0] == 2
@@ -135,3 +136,10 @@ def assert_series_equal(left, right, check_names=True, **kwargs):
         if check_names:
             assert left.name == right.name
         return tm.assert_series_equal(left, right, **kwargs)
+
+
+if PY2:
+    def u8(cs):
+        return cs.decode('utf-8')
+else:
+    u8 = identity

blaze/server/serialization/json_dumps.py

@@ -27,27 +27,27 @@ def json_dumps(dt):
         if not dt.tzname():
             s += u'Z'
 
-    return {'__!datetime': s}
+    return {u'__!datetime': s}
 
 
 @dispatch(frozenset)
 def json_dumps(ds):
-    return {'__!frozenset': list(ds)}
+    return {u'__!frozenset': list(ds)}
 
 
 @dispatch(datetime.timedelta)
 def json_dumps(ds):
-    return {'__!timedelta': ds.total_seconds()}
+    return {u'__!timedelta': ds.total_seconds()}
 
 
 @dispatch(Mono)
 def json_dumps(m):
-    return {'__!mono': unicode(m)}
+    return {u'__!mono': unicode(m)}
 
 
 @dispatch(DataShape)
 def json_dumps(ds):
-    return {'__!datashape': unicode(ds)}
+    return {u'__!datashape': unicode(ds)}
 
 
 @dispatch(object)

blaze/server/serialization/json_dumps_trusted.py

@@ -3,6 +3,7 @@
 from collections import Callable
 
 from blaze.dispatch import dispatch
+from blaze.compatibility import u8
 from functools import partial
 
 from .json_dumps import json_dumps
@@ -17,8 +18,8 @@ def json_dumps_trusted(f):
     # let the server serialize any callable - this is only used for testing
     # at present - do the error handling when json comes from client so in
     # object_hook, catch anything that is not pandas_numpy
-    fcn = ".".join([f.__module__, f.__name__])
-    return {'__!callable': fcn}
+    fcn = u8(".".join([f.__module__, f.__name__]))
+    return {u'__!callable': fcn}
 
 
 for tps, func in json_dumps.funcs.items():

blaze/server/server.py

@@ -21,11 +21,11 @@
 from flask import Blueprint, Flask, Response
 from flask_cors import cross_origin
 from werkzeug.http import parse_options_header
-from toolz import valmap, compose, identity
+from toolz import valmap, compose
 
 import blaze
 from blaze import compute, resource
-from blaze.compatibility import ExitStack, PY2
+from blaze.compatibility import ExitStack, u8
 from blaze.compute import compute_up
 from .serialization import json, all_formats
 from ..interactive import _Data
@@ -401,13 +401,6 @@ def shape():
     return pprint(discover(_get_data()), width=0)
 
 
-if PY2:
-    def _u8(cs):
-        return cs.decode('utf-8')
-else:
-    _u8 = identity
-
-
 def to_tree(expr, names=None):
     """ Represent Blaze expression with core data structures
 
@@ -466,12 +459,12 @@ def to_tree(expr, names=None):
     if isinstance(expr, expr_utils._slice):
         return to_tree(expr.as_slice(), names=names)
     elif isinstance(expr, _Data):
-        return to_tree(symbol(_u8(expr._name), expr.dshape), names)
+        return to_tree(symbol(u8(expr._name), expr.dshape), names)
     elif isinstance(expr, Expr):
-        return {u'op': _u8(type(expr).__name__),
+        return {u'op': u8(type(expr).__name__),
                 u'args': [to_tree(arg, names) for arg in expr._args]}
     elif isinstance(expr, str):
-        return _u8(expr)
+        return u8(expr)
     else:
         return expr
 

docs/source/whatsnew/0.11.1.txt

@@ -41,6 +41,9 @@ Bug Fixes
 * Fixed a testing regression introduced by the latest pymysql version
 (:issue:`1571`).
 
+* Fixed unicode conversion issues when using a Python 3 blaze server and a
+  Python 2 client (:issue:`1572`), (:issue:`1566`).
+
 Miscellaneous
 ~~~~~~~~~~~~~