feat: add trim/ltrim/rtrim with single input argument to remove spaces

extensions/functions_string.yaml

@@ -1316,6 +1316,16 @@ scalar_functions:
             name: "characters"
             description: "The set of characters to remove."
         return: "string"
+      - args:
+          - value: "varchar<L1>"
+            name: "input"
+            description: "The string to remove characters from."
+        return: "varchar<L1>"
+      - args:
+          - value: "string"
+            name: "input"
+            description: "The string to remove characters from."
+        return: "string"
   -
     name: rtrim
     description: >-
@@ -1338,6 +1348,16 @@ scalar_functions:
             name: "characters"
             description: "The set of characters to remove."
         return: "string"
+      - args:
+          - value: "varchar<L1>"
+            name: "input"
+            description: "The string to remove characters from."
+        return: "varchar<L1>"
+      - args:
+          - value: "string"
+            name: "input"
+            description: "The string to remove characters from."
+        return: "string"
   -
     name: trim
     description: >-
@@ -1360,6 +1380,16 @@ scalar_functions:
             name: "characters"
             description: "The set of characters to remove."
         return: "string"
+      - args:
+          - value: "varchar<L1>"
+            name: "input"
+            description: "The string to remove characters from."
+        return: "varchar<L1>"
+      - args:
+          - value: "string"
+            name: "input"
+            description: "The string to remove characters from."
+        return: "string"
   -
     name: lpad
     description: >-

tests/cases/string/ltrim.test

@@ -10,7 +10,32 @@ ltrim(''::str, ' '::str) = ''::str
 ltrim(' '::str, ' '::str) = ''::str
 ltrim(null::str, ' '::str) = null::str
 
+# spaces_only: Examples with only spaces to trim off
+ltrim('abc'::str) = 'abc'::str
+ltrim('   abc'::str) = 'abc'::str
+ltrim('abc   '::str) = 'abc   '::str
+ltrim('   abc   '::str) = 'abc   '::str
+ltrim(''::str) = ''::str
+ltrim(' '::str) = ''::str
+ltrim(null::str) = null::str
+
 # two_inputs: Examples with character input to trim off
 ltrim('aaaaabc'::str, 'a'::str) [spaces_only:FALSE] = 'bc'::str
 ltrim('abcabcdef'::str, 'abc'::str) [spaces_only:FALSE] = 'def'::str
 ltrim('abccbadef'::str, 'abc'::str) [spaces_only:FALSE] = 'def'::str
+
+# varchar
+ltrim('abc'::vchar<20>, ' '::vchar<5>) = 'abc'::vchar<20>
+ltrim('   abc'::vchar<20>, ' '::vchar<5>) = 'abc'::vchar<20>
+ltrim('abc   '::vchar<20>, ' '::vchar<5>) = 'abc   '::vchar<20>
+ltrim('   abc   '::vchar<20>, ' '::vchar<5>) = 'abc   '::vchar<20>
+ltrim('abc'::vchar<20>) = 'abc'::vchar<20>
+ltrim('   abc'::vchar<20>) = 'abc'::vchar<20>
+ltrim('abc   '::vchar<20>) = 'abc   '::vchar<20>
+ltrim('   abc   '::vchar<20>) = 'abc   '::vchar<20>
+ltrim('aaaaabc'::vchar<20>, 'a'::vchar<9>) [spaces_only:False] = 'bc'::vchar<20>
+ltrim('abcabcdef'::vchar<20>, 'abc'::vchar<9>) [spaces_only:False] = 'def'::vchar<20>
+ltrim('   \t\tHello World'::vchar<30>) = '\t\tHello World'::vchar<30>
+ltrim('   \n\nHello World'::vchar<30>) = '\n\nHello World'::vchar<30>
+ltrim('   \r\rHello World'::vchar<30>) = '\r\rHello World'::vchar<30>
+ltrim('   \u2003Hello World'::vchar<30>) = '\u2003Hello World'::vchar<30>

tests/cases/string/rtrim.test

@@ -10,7 +10,32 @@ rtrim(''::str, ' '::str) = ''::str
 rtrim(' '::str, ' '::str) = ''::str
 rtrim(null::str, ' '::str) = null::str
 
+# spaces_only: Examples with only spaces to trim off
+rtrim('abc'::str) = 'abc'::str
+rtrim('   abc'::str) = '   abc'::str
+rtrim('abc   '::str) = 'abc'::str
+rtrim('   abc   '::str) = '   abc'::str
+rtrim(''::str) = ''::str
+rtrim(' '::str) = ''::str
+rtrim(null::str) = null::str
+
 # two_inputs: Examples with character input to trim off
 rtrim('aaaaabccccc'::str, 'c'::str) [spaces_only:FALSE] = 'aaaaab'::str
 rtrim('abcabcdef'::str, 'def'::str) [spaces_only:FALSE] = 'abcabc'::str
 rtrim('defabccba'::str, 'abc'::str) [spaces_only:FALSE] = 'def'::str
+
+# varchar
+rtrim('abc'::vchar<20>, ' '::vchar<5>) = 'abc'::vchar<20>
+rtrim('   abc'::vchar<20>, ' '::vchar<5>) = '   abc'::vchar<20>
+rtrim('abc   '::vchar<20>, ' '::vchar<5>) = 'abc'::vchar<20>
+rtrim('   abc   '::vchar<20>, ' '::vchar<5>) = '   abc'::vchar<20>
+rtrim('abc'::vchar<20>) = 'abc'::vchar<20>
+rtrim('   abc'::vchar<20>) = '   abc'::vchar<20>
+rtrim('abc   '::vchar<20>) = 'abc'::vchar<20>
+rtrim('   abc   '::vchar<20>) = '   abc'::vchar<20>
+rtrim('aaaaabccccc'::vchar<20>, 'c'::vchar<9>) [spaces_only:False] = 'aaaaab'::vchar<20>
+rtrim('abcabcdef'::vchar<20>, 'def'::vchar<9>) [spaces_only:False] = 'abcabc'::vchar<20>
+rtrim('Hello World\t\t   '::vchar<30>) = 'Hello World\t\t'::vchar<30>
+rtrim('Hello World\n\n   '::vchar<30>) = 'Hello World\n\n'::vchar<30>
+rtrim('Hello World\r\r   '::vchar<30>) = 'Hello World\r\r'::vchar<30>
+rtrim('Hello World\u2003  '::vchar<30>) = 'Hello World\u2003'::vchar<30>

tests/cases/string/trim.test

@@ -10,7 +10,32 @@ trim(''::str, ' '::str) = ''::str
 trim(' '::str, ' '::str) = ''::str
 trim(null::str, ' '::str) = null::str
 
+# spaces_only: Examples with only spaces to trim off
+trim('abc'::str) = 'abc'::str
+trim('   abc'::str) = 'abc'::str
+trim('abc   '::str) = 'abc'::str
+trim('   abc   '::str) = 'abc'::str
+trim(''::str) = ''::str
+trim(' '::str) = ''::str
+trim(null::str) = null::str
+
 # two_inputs: Examples with character input to trim off
 trim('aaaaabcccccaaa'::str, 'a'::str) [spaces_only:False] = 'bccccc'::str
 trim('defabcabcdef'::str, 'def'::str) [spaces_only:False] = 'abcabc'::str
 trim('abcdefcbaa'::str, 'abc'::str) [spaces_only:False] = 'def'::str
+
+# varchar
+trim('abc'::vchar<20>, ' '::vchar<5>) = 'abc'::vchar<20>
+trim('   abc'::vchar<20>, ' '::vchar<5>) = 'abc'::vchar<20>
+trim('abc   '::vchar<20>, ' '::vchar<5>) = 'abc'::vchar<20>
+trim('   abc   '::vchar<20>, ' '::vchar<5>) = 'abc'::vchar<20>
+trim('abc'::vchar<20>) = 'abc'::vchar<20>
+trim('   abc'::vchar<20>) = 'abc'::vchar<20>
+trim('abc   '::vchar<20>) = 'abc'::vchar<20>
+trim('   abc   '::vchar<20>) = 'abc'::vchar<20>
+trim('aaaaabcccccaaa'::vchar<20>, 'a'::vchar<9>) [spaces_only:False] = 'bccccc'::vchar<20>
+trim('defabcabcdef'::vchar<20>, 'def'::vchar<9>) [spaces_only:False] = 'abcabc'::vchar<20>
+trim('   \tHello World\t   '::vchar<30>) = '\tHello World\t'::vchar<30>
+trim('   \nHello World\n   '::vchar<30>) = '\nHello World\n'::vchar<30>
+trim('   \rHello World\r   '::vchar<30>) = '\rHello World\r'::vchar<30>
+trim('   \u2003Hello World\u2003  '::vchar<30>) = '\u2003Hello World\u2003'::vchar<30>

tests/coverage/visitor.py

@@ -268,6 +268,8 @@ def visitArgument(self, ctx: FuncTestCaseParser.ArgumentContext):
             return self.visitBooleanArg(ctx.booleanArg())
         if ctx.stringArg() is not None:
             return self.visitStringArg(ctx.stringArg())
+        if ctx.varCharArg() is not None:
+            return self.visitVarCharArg(ctx.varCharArg())
         if ctx.decimalArg() is not None:
             return self.visitDecimalArg(ctx.decimalArg())
         if ctx.dateArg() is not None:
@@ -330,6 +332,12 @@ def visitBooleanArg(self, ctx: FuncTestCaseParser.BooleanArgContext):
     def visitStringArg(self, ctx: FuncTestCaseParser.StringArgContext):
         return CaseLiteral(value=ctx.StringLiteral().getText(), type="str")
 
+    def visitVarCharArg(self, ctx: FuncTestCaseParser.VarCharArgContext):
+        return CaseLiteral(
+            value=ctx.StringLiteral().getText(),
+            type=ctx.varCharType().getText().lower(),
+        )
+
     def visitDecimalArg(self, ctx: FuncTestCaseParser.DecimalArgContext):
         return CaseLiteral(
             value=self.visitNumericLiteral(ctx.numericLiteral()),

tests/test_extensions.py

@@ -24,15 +24,15 @@ def test_substrait_extension_coverage():
     all_test_files = load_all_testcases(test_case_dir)
     coverage = get_test_coverage(all_test_files, registry)
 
-    assert coverage.test_count >= 1077
+    assert coverage.test_count >= 1140
     assert (
         coverage.num_tests_with_no_matching_function == 0
     ), f"{coverage.num_tests_with_no_matching_function} tests with no matching function"
-    assert coverage.num_covered_function_variants >= 226
-    assert coverage.total_function_variants >= 513
+    assert coverage.num_covered_function_variants >= 235
+    assert coverage.total_function_variants >= 519
     assert (
         coverage.total_function_variants - coverage.num_covered_function_variants
-    ) <= 287, (
+    ) <= 284, (
         f"Coverage gap too large: {coverage.total_function_variants - coverage.num_covered_function_variants} "
         f"function variants with no tests, out of {coverage.total_function_variants} total function variants."
     )