added basic string manipulation functions (#777)

* added basic string manipulation functions * fixed build tag * fixed edge case in substring where regex can be used * added tests for variable args
kwilteam · May 30, 2024 · f8a7dd5 · f8a7dd5
1 parent 533ccf8
commit f8a7dd5
Show file tree

Hide file tree

Showing 3 changed files with 444 additions and 179 deletions.
diff --git a/internal/engine/integration/procedure_test.go b/internal/engine/integration/procedure_test.go
@@ -160,6 +160,86 @@ func Test_Procedures(t *testing.T) {
 				{"satoshi", "hello world"},
 			},
 		},
+		{
+			name: "string functions",
+			procedure: `procedure string_funcs() public view {
+				$val := 'hello world';
+				$val := $val || '!!!';
+				$val := upper($val);
+				if $val != 'HELLO WORLD!!!' {
+					error('upper failed');
+				}
+				$val := lower($val);
+				if $val != 'hello world!!!' {
+					error('lower failed');
+				}
+
+				if bit_length($val) != 112 {
+					error('bit_length failed');
+				}
+				if char_length($val) != 14 or character_length($val) != 14 or length($val) != 14 {
+					error('length failed');
+				}
+				if octet_length($val) != 14 {
+					error('octet_length failed');
+				}
+				$val := rtrim($val, '!');
+				if $val != 'hello world' {
+					error('rtrim failed');
+				}
+				if rtrim($val||' ') != 'hello world' {
+					error('rtrim 2 failed');
+				}
+
+				$val := ltrim($val, 'h');
+				if $val != 'ello world' {
+					error('ltrim failed');
+				}
+				if ltrim(' '||$val) != 'ello world' { // add a space and trim it off
+					error('ltrim 2 failed');
+				}
+
+				$val := lpad($val, 11, 'h');
+				if $val != 'hello world' {
+					error('lpad failed');
+				}
+				if lpad($val, 12) != ' hello world' {
+					error('lpad 2 failed');
+				}
+
+				$val := rpad($val, 12, '!');
+				if $val != 'hello world!' {
+					error('rpad failed');
+				}
+				if rpad($val, 13) != 'hello world! ' {
+					error('rpad 2 failed');
+				}
+
+				if overlay($val, 'xx', 2, 5) != 'hxxworld!' {
+					error('overlay failed');
+				}
+				if overlay($val, 'xx', 2) != 'hxxlo world!' {
+					error('overlay 2 failed');
+				}
+
+				if position('world', $val) != 7 {
+					error('position failed');
+				}
+				if substring($val, 7, 5) != 'world' {
+					error('substring failed');
+				}
+				if substring($val, 7) != 'world!' {
+					error('substring 2 failed');
+				}
+
+				if trim(' ' || $val || ' ') != 'hello world!' {
+					error('trim failed');
+				}
+				if trim('a'||$val||'a', 'a') != 'hello world!' {
+					error('trim 2 failed');
+				}
+			}`,
+		},
 	}
 
 	for _, test := range tests {

diff --git a/parse/analyze.go b/parse/analyze.go
@@ -848,17 +848,25 @@ func (s *sqlAnalyzer) VisitExpressionArithmetic(p0 *ExpressionArithmetic) any {
 		return s.expressionTypeErr(p0.Right)
 	}
 
-	if !left.Equals(right) {
-		return s.typeErr(p0.Right, right, left)
-	}
-
 	// both must be numeric UNLESS it is a concat
 	if p0.Operator == ArithmeticOperatorConcat {
-		s.expect(p0.Left, left, types.TextType)
+		if !left.Equals(types.TextType) || !right.Equals(types.TextType) {
+			// Postgres supports concatenation on non-text types, but we do not,
+			// so we give a more descriptive error here.
+			// see the note at the top of: https://www.postgresql.org/docs/16.1/functions-string.html
+			s.errs.AddErr(p0.Left, ErrType, "concatenation only allowed on text types. received %s and %s", left.String(), right.String())
+			return cast(p0, types.UnknownType)
+		}
 	} else {
 		s.expectedNumeric(p0.Left, left)
 	}
 
+	// we check this after to return a more helpful error message if
+	// the user is not concatenating strings.
+	if !left.Equals(right) {
+		return s.typeErr(p0.Right, right, left)
+	}
+
 	return cast(p0, left)
 }