tree-sitter: rewrite comment grammar, detach line- and block comments (…

…#71)

src/analyzer/psi/Comment.v

@@ -1,5 +1,9 @@
 module psi
 
-pub struct Comment {
+pub struct LineComment {
+	PsiElementImpl
+}
+
+pub struct BlockComment {
 	PsiElementImpl
 }

src/analyzer/psi/EnumFieldDeclaration.v

@@ -19,7 +19,7 @@ pub fn (f &EnumFieldDeclaration) doc_comment() string {
 		return stub.comment
 	}
 
-	if comment := f.find_child_by_type(.comment) {
+	if comment := f.find_child_by_type(.line_comment) {
 		return comment.get_text().trim_string_left('//').trim(' \t')
 	}
 

src/analyzer/psi/FieldDeclaration.v

@@ -26,7 +26,7 @@ pub fn (f &FieldDeclaration) doc_comment() string {
 		return stub.comment
 	}
 
-	if comment := f.find_child_by_type(.comment) {
+	if comment := f.find_child_by_type(.line_comment) {
 		return comment.get_text().trim_string_left('//').trim(' \t')
 	}
 

src/analyzer/psi/doc_comment_extractor.v

@@ -5,13 +5,13 @@ import strings
 pub fn extract_doc_comment(el PsiElement) string {
 	el_start_line := el.node.start_point().row
 	mut comment := el.prev_sibling() or { return '' }
-	if comment !is Comment {
+	if comment !is LineComment {
 		comment = comment.prev_sibling() or { return '' }
 	}
 
 	mut comments := []PsiElement{}
 
-	for comment is Comment {
+	for comment is LineComment {
 		comment_start_line := comment.node.start_point().row
 
 		if comment_start_line + 1 + u32(comments.len) != el_start_line {
@@ -29,7 +29,7 @@ pub fn extract_doc_comment(el PsiElement) string {
 	mut field_eol_comment := ''
 	if el is FieldDeclaration {
 		if next := el.next_sibling() {
-			if next is Comment {
+			if next is LineComment {
 				comment_start_line := next.node.start_point().row
 				if comment_start_line == el_start_line {
 					field_eol_comment = next.get_text().trim_string_left('//').trim_space()

src/analyzer/psi/element_factory.v

@@ -194,8 +194,14 @@ pub fn create_element(node AstNode, containing_file &PsiFile) PsiElement {
 		}
 	}
 
-	if node.type_name == .comment {
-		return Comment{
+	if node.type_name == .line_comment {
+		return LineComment{
+			PsiElementImpl: base_node
+		}
+	}
+
+	if node.type_name == .block_comment {
+		return BlockComment{
 			PsiElementImpl: base_node
 		}
 	}

tree_sitter_v/bindings/node_types.v

@@ -31,6 +31,7 @@ pub enum NodeType {
 	attributes
 	binary_expression
 	block
+	block_comment
 	break_statement
 	c_string_literal
 	call_expression
@@ -92,6 +93,7 @@ pub enum NodeType {
 	label_definition
 	label_reference
 	labeled_statement
+	line_comment
 	literal
 	literal_attribute
 	lock_expression
@@ -139,6 +141,7 @@ pub enum NodeType {
 	selector_expression
 	send_statement
 	shared_type
+	shebang
 	short_element_list
 	short_lambda
 	signature
@@ -171,7 +174,6 @@ pub enum NodeType {
 	var_definition_list
 	visibility_modifiers
 	wrong_pointer_type
-	comment
 	escape_sequence
 	false_
 	float_literal
@@ -182,7 +184,6 @@ pub enum NodeType {
 	none_
 	pseudo_compile_time_identifier
 	rune_literal
-	shebang
 	true_
 }
 
@@ -341,6 +342,7 @@ const node_type_name_to_enum = {
 	'attributes':                       NodeType.attributes
 	'binary_expression':                NodeType.binary_expression
 	'block':                            NodeType.block
+	'block_comment':                    NodeType.block_comment
 	'break_statement':                  NodeType.break_statement
 	'c_string_literal':                 NodeType.c_string_literal
 	'call_expression':                  NodeType.call_expression
@@ -402,6 +404,7 @@ const node_type_name_to_enum = {
 	'label_definition':                 NodeType.label_definition
 	'label_reference':                  NodeType.label_reference
 	'labeled_statement':                NodeType.labeled_statement
+	'line_comment':                     NodeType.line_comment
 	'literal':                          NodeType.literal
 	'literal_attribute':                NodeType.literal_attribute
 	'lock_expression':                  NodeType.lock_expression
@@ -449,6 +452,7 @@ const node_type_name_to_enum = {
 	'selector_expression':              NodeType.selector_expression
 	'send_statement':                   NodeType.send_statement
 	'shared_type':                      NodeType.shared_type
+	'shebang':                          NodeType.shebang
 	'short_element_list':               NodeType.short_element_list
 	'short_lambda':                     NodeType.short_lambda
 	'signature':                        NodeType.signature
@@ -481,7 +485,6 @@ const node_type_name_to_enum = {
 	'var_definition_list':              NodeType.var_definition_list
 	'visibility_modifiers':             NodeType.visibility_modifiers
 	'wrong_pointer_type':               NodeType.wrong_pointer_type
-	'comment':                          NodeType.comment
 	'escape_sequence':                  NodeType.escape_sequence
 	'false':                            NodeType.false_
 	'float_literal':                    NodeType.float_literal
@@ -492,6 +495,5 @@ const node_type_name_to_enum = {
 	'none':                             NodeType.none_
 	'pseudo_compile_time_identifier':   NodeType.pseudo_compile_time_identifier
 	'rune_literal':                     NodeType.rune_literal
-	'shebang':                          NodeType.shebang
 	'true':                             NodeType.true_
 }

tree_sitter_v/grammar.js

@@ -84,7 +84,7 @@ const list_separator = choice(semi, ',');
 module.exports = grammar({
 	name: 'v',
 
-	extras: ($) => [$.comment, /\s/],
+	extras: ($) => [/\s/, $.line_comment, $.block_comment],
 
 	word: ($) => $.identifier,
 
@@ -124,18 +124,19 @@ module.exports = grammar({
 				),
 			),
 
-		shebang: (_) => token(/\#\!([^\\\r\n]+)+/),
+		shebang: (_) => seq('#!', /.*/),
 
-		// http://stackoverflow.com/questions/13014947/regex-to-match-a-c-style-multiline-comment/36328890#36328890
-		comment: (_) =>
-			token(
-				choice(
-					/\/\/[^\n\r]*/,
-					/\/\*(?:[^\/][^\*]+\/\*+[^\/][^\*]+)+(?:[^\*][^\/]+\*+\/[^\*][^\/]+)+\//,
-					/\/\*[^\*]*\*\//,
-				),
+		line_comment: (_) => seq('//', /.*/),
+
+		block_comment: (_) =>
+			seq(
+				'/*',
+				token(choice(/(?:[^/][^*]+\/\*+[^/][^*]+)+(?:[^*][^/]+\*+\/[^*][^/]+)+/, /[^*]*\*/)),
+				'/',
 			),
 
+		comment: ($) => choice($.line_comment, $.block_comment),
+
 		module_clause: ($) => seq(optional($.attributes), 'module', $.identifier),
 
 		import_list: ($) => repeat1($.import_declaration),

tree_sitter_v/queries/helix.highlights.scm

@@ -1,5 +1,6 @@
 [
- (comment)
+ (line_comment)
+ (block_comment)
  (shebang)
 ] @comment
 

tree_sitter_v/queries/highlights.scm

@@ -1,7 +1,8 @@
 (ERROR) @error
 
 [
- (comment)
+ (line_comment)
+ (block_comment)
  (shebang)
 ] @comment