From 1ac9d2bfe5783840351e2e184545f61dfc547418 Mon Sep 17 00:00:00 2001
From: Phillip Cloud <cpcloud@gmail.com>
Date: Tue, 28 Jan 2014 21:26:48 -0500
Subject: [PATCH] BUG: allow lex string comparisons

---
 doc/source/enhancingperf.rst | 34 ++++++++++++++++++++++++++--------
 doc/source/release.rst       |  2 ++
 pandas/computation/expr.py   |  3 ++-
 pandas/tests/test_frame.py   | 19 +++++++++++++++++++
 4 files changed, 49 insertions(+), 9 deletions(-)

diff --git a/doc/source/enhancingperf.rst b/doc/source/enhancingperf.rst
index ddf1fa62b2d61..066fcce64c5ac 100644
--- a/doc/source/enhancingperf.rst
+++ b/doc/source/enhancingperf.rst
@@ -464,19 +464,20 @@ evaluate an expression in the "context" of a ``DataFrame``.
 
 Any expression that is a valid :func:`~pandas.eval` expression is also a valid
 ``DataFrame.eval`` expression, with the added benefit that *you don't have to
-prefix the name of the* ``DataFrame`` *to the column you're interested in
+prefix the name of the* ``DataFrame`` *to the column(s) you're interested in
 evaluating*.
 
-In addition, you can perform in-line assignment of columns within an expression.
-This can allow for *formulaic evaluation*. Only a signle assignement is permitted.
-It can be a new column name or an existing column name. It must be a string-like.
+In addition, you can perform assignment of columns within an expression.
+This allows for *formulaic evaluation*. Only a single assignment is permitted.
+The assignment target can be a new column name or an existing column name, and
+it must be a valid Python identifier.
 
 .. ipython:: python
 
-   df = DataFrame(dict(a = range(5), b = range(5,10)))
-   df.eval('c=a+b')
-   df.eval('d=a+b+c')
-   df.eval('a=1')
+   df = DataFrame(dict(a=range(5), b=range(5, 10)))
+   df.eval('c = a + b')
+   df.eval('d = a + b + c')
+   df.eval('a = 1')
    df
 
 Local Variables
@@ -616,3 +617,20 @@ different engines.
 
 This plot was created using a ``DataFrame`` with 3 columns each containing
 floating point values generated using ``numpy.random.randn()``.
+
+Technical Minutia
+~~~~~~~~~~~~~~~~~
+- Expressions that would result in an object dtype (including simple
+  variable evaluation) have to be evaluated in Python space. The main reason
+  for this behavior is to maintain backwards compatbility with versions of
+  numpy < 1.7. In those versions of ``numpy`` a call to ``ndarray.astype(str)``
+  will truncate any strings that are more than 60 characters in length. Second,
+  we can't pass ``object`` arrays to ``numexpr`` thus string comparisons must
+  be evaluated in Python space.
+- The upshot is that this *only* applies to object-dtype'd expressions. So,
+  if you have an expression--for example--that's a string comparison
+  ``and``-ed together with another boolean expression that's from a numeric
+  comparison, the numeric comparison will be evaluated by ``numexpr``. In fact,
+  in general, :func:`~pandas.query`/:func:`~pandas.eval` will "pick out" the
+  subexpressions that are ``eval``-able by ``numexpr`` and those that must be
+  evaluated in Python space transparently to the user.
diff --git a/doc/source/release.rst b/doc/source/release.rst
index 2caad982f044b..a79182af13955 100644
--- a/doc/source/release.rst
+++ b/doc/source/release.rst
@@ -168,6 +168,8 @@ Bug Fixes
   - Bug in DataFrame construction with recarray and non-ns datetime dtype (:issue:`6140`)
   - Bug in ``.loc`` setitem indexing with a datafrme on rhs, multiple item setting, and
     a datetimelike (:issue:`6152`)
+  - Fixed a stack overflow bug in ``query``/``eval`` during lexicographic
+    string comparisons (:issue:`6155`).
 
 pandas 0.13.0
 -------------
diff --git a/pandas/computation/expr.py b/pandas/computation/expr.py
index c16205ff34b1f..7ebffef7368a0 100644
--- a/pandas/computation/expr.py
+++ b/pandas/computation/expr.py
@@ -508,7 +508,8 @@ def _possibly_eval(self, binop, eval_in_python):
 
     def _possibly_evaluate_binop(self, op, op_class, lhs, rhs,
                                  eval_in_python=('in', 'not in'),
-                                 maybe_eval_in_python=('==', '!=')):
+                                 maybe_eval_in_python=('==', '!=', '<', '>',
+                                                       '<=', '>=')):
         res = op(lhs, rhs)
 
         if self.engine != 'pytables':
diff --git a/pandas/tests/test_frame.py b/pandas/tests/test_frame.py
index 50c5cca935553..4578375ab7dad 100644
--- a/pandas/tests/test_frame.py
+++ b/pandas/tests/test_frame.py
@@ -12841,6 +12841,25 @@ def test_query_with_nested_string(self):
         for parser, engine in product(PARSERS, ENGINES):
             yield self.check_query_with_nested_strings, parser, engine
 
+    def check_query_lex_compare_strings(self, parser, engine):
+        tm.skip_if_no_ne(engine=engine)
+        import operator as opr
+
+        a = Series(tm.choice(list('abcde'), 20))
+        b = Series(np.arange(a.size))
+        df = DataFrame({'X': a, 'Y': b})
+
+        ops = {'<': opr.lt, '>': opr.gt, '<=': opr.le, '>=': opr.ge}
+
+        for op, func in ops.items():
+            res = df.query('X %s "d"' % op, engine=engine, parser=parser)
+            expected = df[func(df.X, 'd')]
+            assert_frame_equal(res, expected)
+
+    def test_query_lex_compare_strings(self):
+        for parser, engine in product(PARSERS, ENGINES):
+            yield self.check_query_lex_compare_strings, parser, engine
+
 class TestDataFrameEvalNumExprPandas(tm.TestCase):
 
     @classmethod