fix regexp comment

Signed-off-by: Dmitry Stoletov <[email protected]>
iMega · Feb 3, 2021 · faa229a · faa229a
1 parent f669df1
commit faa229a
Show file tree

Hide file tree

Showing 5 changed files with 93 additions and 14 deletions.
diff --git a/formatter/lexer.go b/formatter/lexer.go
@@ -17,6 +17,7 @@ package formatter
 import (
 	"bytes"
 	"io"
+	"unicode"
 
 	"github.com/timtadh/lexmachine"
 	"github.com/timtadh/lexmachine/machines"
@@ -38,13 +39,16 @@ func newScanner(code []byte) (*scanner, error) {
 	lexer.Add([]byte(`([a-zA-Z_][a-zA-Z0-9_:]*)`), token(nID))
 	lexer.Add([]byte(`\s*\n\s*\n\s*`), token(nLF))
 	lexer.Add([]byte("( |\t|\f|\r|\n)+"), skip)
-	lexer.Add([]byte(`--[^\[\[]( |\S)*`), comment(nComment))
-	lexer.Add([]byte(`--\[\[([^\]\]])*\]\]`), commentLong(nCommentLong))
-	lexer.Add([]byte(`::([^::])*::`), token(nLabel))
+	lexer.Add([]byte(`::[^:]*::`), token(nLabel))
+
+	lexer.Add([]byte(`--[^[\n]+[^\n]*`), comment(nComment))
+	lexer.Add([]byte(`--\[\[[^]]*\]\]`), commentLong(nCommentLong))
+	lexer.Add([]byte(`--\n`), comment(nComment))
+	lexer.Add([]byte(`--\r\n`), comment(nComment))
 
 	lexer.Add([]byte(`"[^"]*"`), token(nString))
 	lexer.Add([]byte(`'[^']*'`), token(nString))
-	lexer.Add([]byte(`\[\[[^\]\]]*\]\]`), token(nString))
+	lexer.Add([]byte(`\[\[[^]]*\]\]`), token(nString))
 
 	if err := lexer.Compile(); err != nil {
 		return nil, err
@@ -127,8 +131,13 @@ func token(nodeID int) lexmachine.Action {
 
 func comment(nodeID int) lexmachine.Action {
 	return func(s *lexmachine.Scanner, m *machines.Match) (interface{}, error) {
-		b := bytes.TrimLeft(m.Bytes, "--")
-		b = bytes.TrimSpace(b)
+		prefix := []byte("-- ")
+		if !bytes.HasPrefix(m.Bytes, prefix) {
+			prefix = []byte("--")
+		}
+
+		b := bytes.TrimPrefix(m.Bytes, prefix)
+		b = bytes.TrimRightFunc(b, unicode.IsSpace)
 		m.Bytes = b
 
 		return s.Token(nodeID, string(b), m), nil
@@ -137,8 +146,8 @@ func comment(nodeID int) lexmachine.Action {
 
 func commentLong(nodeID int) lexmachine.Action {
 	return func(s *lexmachine.Scanner, m *machines.Match) (interface{}, error) {
-		b := bytes.TrimLeft(m.Bytes, "--[[")
-		b = bytes.TrimRight(b, "]]")
+		b := bytes.TrimPrefix(m.Bytes, []byte("--[["))
+		b = bytes.TrimSuffix(b, []byte("]]"))
 		b = bytes.TrimSpace(b)
 		m.Bytes = b
 

diff --git a/formatter/statement_comment.go b/formatter/statement_comment.go
@@ -15,6 +15,7 @@
 package formatter
 
 import (
+	"bytes"
 	"io"
 )
 
@@ -50,7 +51,12 @@ func (s *commentStatement) GetStatement(prev, cur *element) statement {
 }
 
 func (s *commentStatement) Format(c *Config, p printer, w io.Writer) error {
-	if _, err := w.Write([]byte("-- ")); err != nil {
+	prefix := []byte("-- ")
+	if len(s.Element.Token.Lexeme) == 0 || bytes.HasPrefix(s.Element.Token.Lexeme, []byte("---")) {
+		prefix = []byte("--")
+	}
+
+	if _, err := w.Write(prefix); err != nil {
 		return err
 	}
 

diff --git a/formatter/statement_comment_test.go b/formatter/statement_comment_test.go
@@ -81,8 +81,8 @@ func TestParseComment(t *testing.T) {
 							Element: &element{
 								Token: &lexmachine.Token{
 									Type:        nComment,
-									Value:       "comment2",
-									Lexeme:      []byte("comment2"),
+									Value:       "    comment2",
+									Lexeme:      []byte("    comment2"),
 									TC:          12,
 									StartLine:   3,
 									StartColumn: 1,
@@ -98,6 +98,70 @@ func TestParseComment(t *testing.T) {
 			},
 			wantErr: false,
 		},
+		{
+			skip: false,
+			name: "two comment statement",
+			args: args{
+				code: []byte(`
+---------------
+-- comment
+--     comment2
+`,
+				),
+			},
+			want: &document{
+				Body: &body{
+					Blocks: map[uint64]statement{
+						0: &commentStatement{
+							Element: &element{
+								Token: &lexmachine.Token{
+									Type:        nComment,
+									Value:       "-------------",
+									Lexeme:      []byte("-------------"),
+									TC:          1,
+									StartLine:   2,
+									StartColumn: 1,
+									EndLine:     2,
+									EndColumn:   15,
+								},
+							},
+						},
+						1: &commentStatement{
+							Element: &element{
+								Token: &lexmachine.Token{
+									Type:        nComment,
+									Value:       "comment",
+									Lexeme:      []byte("comment"),
+									TC:          17,
+									StartLine:   3,
+									StartColumn: 1,
+									EndLine:     3,
+									EndColumn:   10,
+								},
+							},
+							IsNewline: true,
+						},
+						2: &commentStatement{
+							Element: &element{
+								Token: &lexmachine.Token{
+									Type:        nComment,
+									Value:       "    comment2",
+									Lexeme:      []byte("    comment2"),
+									TC:          28,
+									StartLine:   4,
+									StartColumn: 1,
+									EndLine:     4,
+									EndColumn:   15,
+								},
+							},
+							IsNewline: true,
+						},
+					},
+					Qty: 3,
+				},
+			},
+			wantErr: false,
+		},
 	}
 	for _, tt := range tests {
 		if tt.skip == true {

diff --git a/tests/default/0005.lua b/tests/default/0005.lua
@@ -7,7 +7,7 @@ for i, pkg in ipairs(packages) do
     end
  end
 
---- Load a local or remote manifest describing a repository.
+-- Load a local or remote manifest describing a repository.
 -- All functions that use manifest tables assume they were obtained
 -- through either this function or load_local_manifest.
 -- @param repo_url string: URL or pathname for the repository.

diff --git a/tests/default/0007.lua b/tests/default/0007.lua
@@ -78,7 +78,7 @@ bar.say("hello") -- using the module
 -- bad
 local skt = require("socket")
 
---- @module foo.bar
+-- @module foo.bar
 local bar = {}
 
 function bar.say(greeting)
@@ -99,7 +99,7 @@ local bla = require "bla"
 -- good
 local bla = require("bla")
 
---- @module myproject.myclass
+-- @module myproject.myclass
 local myclass = {}
 
 -- class table