summaryrefslogtreecommitdiff
path: root/sqlparse/lexer.py
diff options
context:
space:
mode:
authorAndi Albrecht <albrecht.andi@gmail.com>2013-04-04 05:54:43 +0200
committerAndi Albrecht <albrecht.andi@gmail.com>2013-04-04 05:54:43 +0200
commit081b23757c1a534baf42f7c099bab471bf20fe83 (patch)
treef751d0eb5f754117bcdf6767cf2600b1f67def62 /sqlparse/lexer.py
parente664ae1da02f87f720878c7699cc26d0a8e9e659 (diff)
downloadsqlparse-081b23757c1a534baf42f7c099bab471bf20fe83.tar.gz
Add encoding parameter to top-level functions (fixes issue20).
Diffstat (limited to 'sqlparse/lexer.py')
-rw-r--r--sqlparse/lexer.py7
1 files changed, 5 insertions, 2 deletions
diff --git a/sqlparse/lexer.py b/sqlparse/lexer.py
index e769d7b..4d200a6 100644
--- a/sqlparse/lexer.py
+++ b/sqlparse/lexer.py
@@ -224,7 +224,8 @@ class Lexer(object):
def _decode(self, text):
if sys.version_info[0] == 3:
- return text
+ if isinstance(text, str):
+ return text
if self.encoding == 'guess':
try:
text = text.decode('utf-8')
@@ -355,11 +356,13 @@ class Lexer(object):
break
-def tokenize(sql):
+def tokenize(sql, encoding=None):
"""Tokenize sql.
Tokenize *sql* using the :class:`Lexer` and return a 2-tuple stream
of ``(token type, value)`` items.
"""
lexer = Lexer()
+ if encoding is not None:
+ lexer.encoding = encoding
return lexer.get_tokens(sql)