Added proper error messages when having duplicate column names

2 files changed, 24 insertions, 1 deletions

diff --git a/bin/q b/bin/q
index b794ef3..c3b0182 100755
--- a/bin/q
+++ b/bin/q
@@ -525,6 +525,13 @@ class TableColumnInferer(object):
                 column_name_errors.append(
                     (v, "Column name must be UTF-8 Compatible"))
                 continue
+            # We're checking for column duplication for each field in order to be able to still provide it along with other errors
+            if len(filter(lambda x: x == v,value_list)) > 1:
+                entry = (v, "Column name is duplicated")
+                # Don't duplicate the error report itself
+                if entry not in column_name_errors:
+                    column_name_errors.append(entry)
+                continue
             nul_index = v.find("\x00")
             if nul_index >= 0:
                 column_name_errors.append(
diff --git a/test/test-suite b/test/test-suite
index c6d9344..38dc0ac 100755
--- a/test/test-suite
+++ b/test/test-suite
@@ -1338,6 +1338,23 @@ class BasicTests(AbstractQTestCase):
 
         self.cleanup(tmpfile)
 
+    def test_duplicate_column_name_detection(self):
+        file_data = "a,b,a\n10,20,30\n30,40,50"
+        tmpfile = self.create_file_with_data(file_data)
+
+        cmd = '../bin/q -H -d , "select a from %s"' % tmpfile.name
+        retcode, o, e = run_command(cmd)
+
+        self.assertEquals(retcode, 35)
+        self.assertEquals(len(o), 0)
+        self.assertEquals(len(e), 2)
+
+        self.assertTrue(e[0].startswith('Bad header row:'))
+        self.assertEquals(e[1],"'a': Column name is duplicated")
+
+        self.cleanup(tmpfile)
+
+
 class ParsingModeTests(AbstractQTestCase):
 
     def test_strict_mode_column_count_mismatch_error(self):
@@ -1727,7 +1744,6 @@ class FormattingTests(AbstractQTestCase):
         self.cleanup(tmp_data_file)
 
 
-
 class SqlTests(AbstractQTestCase):
 
     def test_find_example(self):