Don't group trailing whitepsace in WHERE clauses (fixes issue35).

4 files changed, 21 insertions, 6 deletions

diff --git a/CHANGES b/CHANGES
index c23ae94..37f63f5 100644
--- a/CHANGES
+++ b/CHANGES
@@ -3,6 +3,7 @@ In Development
 
 Bug Fixes
  * Improve parsing of floats (thanks to Kris).
+ * When formatting a statement a space before LIMIT was removed (issue35).
 
 
 Release 0.1.2 (Nov 23, 2010)
diff --git a/sqlparse/engine/grouping.py b/sqlparse/engine/grouping.py
index 3a2e3c3..d233772 100644
--- a/sqlparse/engine/grouping.py
+++ b/sqlparse/engine/grouping.py
@@ -253,7 +253,8 @@ def group_where(tlist):
         else:
             end = tlist.tokens[tlist.token_index(end) - 1]
         group = tlist.group_tokens(sql.Where,
-                                   tlist.tokens_between(token, end))
+                                   tlist.tokens_between(token, end),
+                                   ignore_ws=True)
         idx = tlist.token_index(group)
         token = tlist.token_next_match(idx, T.Keyword, 'WHERE')
 
diff --git a/sqlparse/sql.py b/sqlparse/sql.py
index 8e2e82b..722204d 100644
--- a/sqlparse/sql.py
+++ b/sqlparse/sql.py
@@ -287,16 +287,21 @@ class TokenList(Token):
         If *exclude_end* is ``True`` (default is ``False``) the end token
         is included too.
         """
+        # FIXME(andi): rename exclude_end to inlcude_end
         if exclude_end:
             offset = 0
         else:
             offset = 1
-        return self.tokens[
-            self.token_index(start):self.token_index(end) + offset]
+        end_idx = self.token_index(end) + offset
+        start_idx = self.token_index(start)
+        return self.tokens[start_idx:end_idx]
 
-    def group_tokens(self, grp_cls, tokens):
+    def group_tokens(self, grp_cls, tokens, ignore_ws=False):
         """Replace tokens by an instance of *grp_cls*."""
         idx = self.token_index(tokens[0])
+        if ignore_ws:
+            while tokens and tokens[-1].is_whitespace():
+                tokens = tokens[:-1]
         for t in tokens:
             self.tokens.remove(t)
         grp = grp_cls(tokens)
diff --git a/tests/test_regressions.py b/tests/test_regressions.py
index 3046b97..cbe1234 100644
--- a/tests/test_regressions.py
+++ b/tests/test_regressions.py
@@ -1,13 +1,13 @@
 # -*- coding: utf-8 -*-
 
-import unittest
+from tests.utils import TestCaseBase
 
 import sqlparse
 from sqlparse import sql
 from sqlparse import tokens as T
 
 
-class RegressionTests(unittest.TestCase):
+class RegressionTests(TestCaseBase):
 
     def test_issue9(self):
         # make sure where doesn't consume parenthesis
@@ -31,3 +31,11 @@ class RegressionTests(unittest.TestCase):
         t = sqlparse.parse("create")[0].token_first()
         self.assertEqual(t.match(T.Keyword.DDL, "create"), True)
         self.assertEqual(t.match(T.Keyword.DDL, "CREATE"), True)
+
+    def test_issue35(self):
+        # missing space before LIMIT
+        sql = sqlparse.format("select * from foo where bar = 1 limit 1",
+                              reindent=True)
+        self.ndiffAssertEqual(sql, "\n".join(["select *",
+                                              "from foo",
+                                              "where bar = 1 limit 1"]))