v.scanner: fix ambiguity of two-level generics and shift-right (#11540)

vlib/builtin/string.v

@@ -1274,6 +1274,13 @@ pub fn (c byte) is_letter() bool {
 	return (c >= `a` && c <= `z`) || (c >= `A` && c <= `Z`)
 }
 
+// is_alnum returns `true` if the byte is in range a-z, A-Z, 0-9 and `false` otherwise.
+// Example: assert byte(`V`) == true
+[inline]
+pub fn (c byte) is_alnum() bool {
+	return c.is_letter() || c.is_digit()
+}
+
 // free allows for manually freeing the memory occupied by the string
 [manualfree; unsafe]
 pub fn (s &string) free() {

vlib/v/fmt/fmt.v

@@ -1534,7 +1534,7 @@ pub fn (mut f Fmt) call_expr(node ast.CallExpr) {
 		f.comments(arg.comments)
 	}
 	if node.is_method {
-		if node.name in ['map', 'filter'] {
+		if node.name in ['map', 'filter', 'all', 'any'] {
 			f.inside_lambda = true
 			defer {
 				f.inside_lambda = false

vlib/v/scanner/scanner.v

@@ -11,6 +11,7 @@ import v.pref
 import v.util
 import v.vet
 import v.errors
+import v.ast
 
 const (
 	single_quote = `'`
@@ -36,6 +37,7 @@ pub mut:
 	is_inter_end      bool
 	is_enclosed_inter bool
 	line_comment      string
+	last_lt           int = -1 // position of latest <
 	// prev_tok                 TokenKind
 	is_started                  bool
 	is_print_line_on_error      bool
@@ -917,12 +919,49 @@ fn (mut s Scanner) text_scan() token.Token {
 					return s.new_token(.ge, '', 2)
 				} else if nextc == `>` {
 					if s.pos + 2 < s.text.len {
-						if s.text[s.pos + 2] == `=` {
+						// first eat the possible spaces eg `>> (` => `>>(`
-							s.pos += 2
+						mut non_space_pos := s.pos + 2
-							return s.new_token(.right_shift_assign, '', 3)
+						for non_space_pos < s.text.len && s.text[non_space_pos].is_space() {
-						} else if s.text[s.pos + 2] in [`(`, `)`, `{`, `>`, `,`] {
+							non_space_pos++
-							// multi-level generics such as Foo<Bar<baz>>{ }, func<Bar<baz>>( ), etc
+						}
-							return s.new_token(.gt, '', 1)
+						match s.text[non_space_pos] {
+							`=` {
+								s.pos += 2
+								return s.new_token(.right_shift_assign, '', 3)
+							}
+							// definite generic cases such as Foo<Bar<int>>{}
+							`)`, `{`, `}`, `,`, `>`, `[`, `]` {
+								return s.new_token(.gt, '', 1)
+							}
+							// notice two-level generic call and shift-right share the rest patterns
+							// such as `foo<Baz, Bar<int>>(a)` vs `a, b := Foo{}<Foo{}, bar>>(baz)`
+							// which is hard but could be discriminated by my following algorithm
+							// @SleepyRoy if you have smarter algorithm :-)
+							else {
+								// almost correct heuristics: 2-level generic call's last <T> cannot be extremely long
+								// here we set the limit 100 which should be nice for real cases
+								if s.last_lt >= 0 && s.pos - s.last_lt < 100 {
+									// ...Bar<int, []Foo, [20]f64, map[string][]bool>> =>
+									// int, []Foo, [20]f64, map[string][]bool =>
+									// int, Foo, f64, bool
+									typs := s.text[s.last_lt + 1..s.pos].trim_right('>').split(',').map(it.trim_space().trim_right('>').after(']'))
+									// if any typ is neither builtin nor Type, then the case is not generics
+									for typ in typs {
+										if typ.len == 0 {
+											s.pos++
+											return s.new_token(.right_shift, '', 2)
+										}
+										if typ !in ast.builtin_type_names && !(typ[0].is_capital()
+											&& typ[1..].bytes().all(it.is_alnum())) {
+											s.pos++
+											return s.new_token(.right_shift, '', 2)
+										}
+									}
+									return s.new_token(.gt, '', 1)
+								}
+								s.pos++
+								return s.new_token(.right_shift, '', 2)
+							}
 						}
 					}
 					s.pos++
@@ -946,6 +985,7 @@ fn (mut s Scanner) text_scan() token.Token {
 					s.pos++
 					return s.new_token(.arrow, '', 2)
 				} else {
+					s.last_lt = s.pos
 					return s.new_token(.lt, '', 1)
 				}
 			}

vlib/v/tests/generics_test.v

@@ -510,6 +510,21 @@ fn test_multi_level_generics() {
 		two) == 20
 }
 
+struct Empty {}
+
+fn (e1 Empty) < (e2 Empty) bool {
+	return true
+}
+
+struct TandU<T, U> {
+	t T
+	u U
+}
+
+fn boring_function<T>(t T) bool {
+	return true
+}
+
 fn test_generic_detection() {
 	v1, v2 := -1, 1
 
@@ -530,4 +545,43 @@ fn test_generic_detection() {
 	assert multi_generic_args<[]int, int>([]int{}, 0)
 	assert multi_generic_args<map[int]int, int>(map[int]int{}, 0)
 	assert 0 < return_one<int>(10, 0)
+
+	// "the hardest cases"
+	foo, bar, baz := 1, 2, 16
+	res1, res2 := foo < bar, baz >> (foo + 1 - 1)
+	assert res1
+	assert res2 == 8
+	res3, res4 := Empty{} < Empty{}, baz >> (foo + 1 - 1)
+	assert res3
+	assert res4 == 8
+	assert boring_function<TandU<Empty, int>>(TandU<Empty, int>{
+		t: Empty{}
+		u: 10
+	})
+
+	assert boring_function<MultiLevel<MultiLevel<int>>>(MultiLevel<MultiLevel<int>>{
+		foo: MultiLevel<int>{
+			foo: 10
+		}
+	})
+
+	assert boring_function<TandU<MultiLevel<int>, []int>>(TandU<MultiLevel<int>, []int>{
+		t: MultiLevel<int>{
+			foo: 10
+		}
+		u: [10]
+	})
+
+	// this final case challenges your scanner :-)
+	assert boring_function<TandU<TandU<int,MultiLevel<Empty>>, map[string][]int>>(TandU<TandU<int,MultiLevel<Empty>>, map[string][]int>{
+		t: TandU<int,MultiLevel<Empty>>{
+			t: 20
+			u: MultiLevel<Empty>{
+				foo: Empty{}
+			}
+		}
+		u: {
+			'bar': [40]
+		}
+	})
 }