automate sqllogictest for different string types

datafusion/sqllogictest/test_files/string_functions/substr/substr_literal.slt.part

@@ -0,0 +1,119 @@
+query T
+SELECT substr('alphabet', -3)
+----
+alphabet
+
+query T
+SELECT substr('alphabet', 0)
+----
+alphabet
+
+query T
+SELECT substr('alphabet', 1)
+----
+alphabet
+
+query T
+SELECT substr('alphabet', 2)
+----
+lphabet
+
+query T
+SELECT substr('alphabet', 3)
+----
+phabet
+
+query T
+SELECT substr('alphabet', 30)
+----
+(empty)
+
+query T
+SELECT substr('alphabet', 3, 2)
+----
+ph
+
+query T
+SELECT substr('alphabet', 3, 20)
+----
+phabet
+
+query TT
+select 
+    substr(arrow_cast('alphabet', 'LargeUtf8'), 3, 20), 
+    substr(arrow_cast('alphabet', 'Utf8View'), 3, 20);
+----
+phabet phabet
+
+# test range ouside of string length
+query TTTTTTTTTTTT
+SELECT
+    substr('hi🌏', 1, 3),
+    substr('hi🌏', 1, 4),
+    substr('hi🌏', 1, 100),
+    substr('hi🌏', 0, 1),
+    substr('hi🌏', 0, 2),
+    substr('hi🌏', 0, 4),
+    substr('hi🌏', 0, 5),
+    substr('hi🌏', -10, 100),
+    substr('hi🌏', -10, 12),
+    substr('hi🌏', -10, 5),
+    substr('hi🌏', 10, 0),
+    substr('hi🌏', 10, 10);
+----
+hi🌏 hi🌏 hi🌏 (empty) h hi🌏 hi🌏 hi🌏 h (empty) (empty) (empty)
+
+query TTTTTTTTTTTT
+SELECT
+    substr('', 1, 3),
+    substr('', 1, 4),
+    substr('', 1, 100),
+    substr('', 0, 1),
+    substr('', 0, 2),
+    substr('', 0, 4),
+    substr('', 0, 5),
+    substr('', -10, 100),
+    substr('', -10, 12),
+    substr('', -10, 5),
+    substr('', 10, 0),
+    substr('', 10, 10);
+----
+(empty) (empty) (empty) (empty) (empty) (empty) (empty) (empty) (empty) (empty) (empty) (empty)
+
+# Nulls
+query TTTTTTTTTT
+SELECT 
+  substr('alphabet', NULL),
+  substr(NULL, 1),
+  substr(NULL, NULL),
+  substr('alphabet', CAST(NULL AS int), -20),
+  substr('alphabet', 3, CAST(NULL AS int)),
+  substr(NULL, 3, -4),
+  substr(NULL, NULL, 4),
+  substr(NULL, 1, NULL),
+  substr('', NULL, NULL),
+  substr(NULL, NULL, NULL);
+----
+NULL NULL NULL NULL NULL NULL NULL NULL NULL NULL
+
+query T
+SELECT substr('Hello🌏世界', 5)
+----
+o🌏世界
+
+query T
+SELECT substr('Hello🌏世界', 5, 3)
+----
+o🌏世
+
+statement error The SUBSTR function can only accept strings, but got Int64.
+SELECT substr(1, 3)
+
+statement error The SUBSTR function can only accept strings, but got Int64.
+SELECT substr(1, 3, 4)
+
+statement error Execution error: negative substring length not allowed
+select substr(arrow_cast('foo', 'Utf8View'), 1, -1);
+
+statement error Execution error: negative substring length not allowed
+select substr('', 1, -1);

datafusion/sqllogictest/test_files/string_functions/substr/substr_runner.slt

@@ -0,0 +1,55 @@
+# --------------------------------------
+# Test `substr()` with literal arguments
+# --------------------------------------
+include ./substr_literal.slt.part
+
+# --------------------------------------
+# Setup test tables with different physical string types (Utf8/Utf8View/LargeUtf8)
+# and repeat tests in `substr_table.slt.part`
+# --------------------------------------
+statement ok
+create table test_substr_base (
+    col1 VARCHAR
+) as values ('foo'), ('hello🌏世界'), ('💩'), ('ThisIsAVeryLongASCIIString'), (''), (NULL);
+
+#
+# Run1: Utf8
+#
+statement ok
+create table test_substr as
+select arrow_cast(col1, 'Utf8') as c1 from test_substr_base;
+
+include ./substr_table.slt.part
+
+statement ok
+drop table test_substr;
+
+#
+# Run2: Utf8View
+#
+statement ok
+create table test_substr as
+select arrow_cast(col1, 'Utf8View') as c1 from test_substr_base;
+
+include ./substr_table.slt.part
+
+statement ok
+drop table test_substr;
+
+#
+# Run3: LargeUtf8
+#
+statement ok
+create table test_substr as
+select arrow_cast(col1, 'LargeUtf8') as c1 from test_substr_base;
+
+include ./substr_table.slt.part
+
+statement ok
+drop table test_substr;
+
+# --------------------------------------
+# Cleanup
+# --------------------------------------
+statement ok
+drop table test_substr_base;

datafusion/sqllogictest/test_files/string_functions/substr/substr_table.slt.part

@@ -0,0 +1,28 @@
+# This file is intended to be run with tables already defined
+# with standard data but possibly different types in string columns
+# (String, StringView, etc.)
+
+query TTTTTTTTTTTTTT
+select 
+    substr(c1, 1),
+    substr(c1, 3),
+    substr(c1, 100),
+    substr(c1, -1),
+    substr(c1, 0, 0),
+    substr(c1, -1, 2),
+    substr(c1, -2, 10),
+    substr(c1, -100, 200),
+    substr(c1, -10, 10),
+    substr(c1, -100, 10),
+    substr(c1, 1, 100),
+    substr(c1, 5, 3),
+    substr(c1, 100, 200),
+    substr(c1, 8, 0)
+from test_substr;
+----
+foo o (empty) foo (empty) (empty) foo foo (empty) (empty) foo (empty) (empty) (empty)
+hello🌏世界 llo🌏世界 (empty) hello🌏世界 (empty) (empty) hello🌏世 hello🌏世界 (empty) (empty) hello🌏世界 o🌏世 (empty) (empty)
+💩 (empty) (empty) 💩 (empty) (empty) 💩 💩 (empty) (empty) 💩 (empty) (empty) (empty)
+ThisIsAVeryLongASCIIString isIsAVeryLongASCIIString (empty) ThisIsAVeryLongASCIIString (empty) (empty) ThisIsA ThisIsAVeryLongASCIIString (empty) (empty) ThisIsAVeryLongASCIIString IsA (empty) (empty)
+(empty) (empty) (empty) (empty) (empty) (empty) (empty) (empty) (empty) (empty) (empty) (empty) (empty) (empty)
+NULL NULL NULL NULL NULL NULL NULL NULL NULL NULL NULL NULL NULL NULL