Blink parity: semicolon, bracket-comment, float, destructuring (#130)

Author: TomasVotruba

blink/internal/fixer/rules/array_notation.go

@@ -250,6 +250,17 @@ func topBracketIsArray(stack []string) bool {
 	return len(stack) > 0 && stack[len(stack)-1] == "["
 }
 
+// isDestructuringAssignOpen reports whether the "[" at open is a short-list
+// destructuring target - its matching "]" is directly followed by a single "=".
+func isDestructuringAssignOpen(s *tokens.Stream, open int) bool {
+	c := s.MatchForward(open)
+	if c < 0 {
+		return false
+	}
+	n := nextSignificantIndex(s, c)
+	return n >= 0 && s.At(n).Kind == token.Punct && s.At(n).Value == "="
+}
+
 // PHP-CS-Fixer: https://github.com/PHP-CS-Fixer/PHP-CS-Fixer/blob/master/src/Fixer/ArrayNotation/NoWhitespaceBeforeCommaInArrayFixer.php
 //
 // NoWhitespaceBeforeCommaInArray removes single-line whitespace before a comma
@@ -340,7 +351,15 @@ func (f WhitespaceAfterCommaInArray) Fix(s *tokens.Stream) bool {
 			continue
 		}
 		switch t.Value {
-		case "(", "[", "{":
+		case "[":
+			// a destructuring target ("[$a,$b] = ...", or nested in one) is not an
+			// array literal, so php-cs-fixer does not space its commas
+			if (len(stack) > 0 && stack[len(stack)-1] == "d[") || isDestructuringAssignOpen(s, i) {
+				stack = append(stack, "d[")
+			} else {
+				stack = append(stack, "[")
+			}
+		case "(", "{":
 			stack = append(stack, t.Value)
 		case ")", "]", "}":
 			if len(stack) > 0 {

blink/internal/fixer/rules/constructs.go

@@ -261,7 +261,8 @@ func (f NoSpacesInsideParenthesis) Fix(s *tokens.Stream) bool {
 	for i := 0; i < s.Len(); i++ {
 		t := s.At(i)
 		if t.Kind == token.Punct && t.Value == "(" &&
-			i+1 < s.Len() && s.At(i+1).Kind == token.Whitespace && !hasNewline(s.At(i+1).Value) {
+			i+1 < s.Len() && s.At(i+1).Kind == token.Whitespace && !hasNewline(s.At(i+1).Value) &&
+			!commentAt(s, i+2) { // keep the space before a trailing comment
 			s.RemoveAt(i + 1)
 			changed = true
 		}

blink/internal/fixer/rules/gen_arrays.go

@@ -145,6 +145,11 @@ func isComment(t token.Token) bool {
 	return t.Kind == token.Comment || t.Kind == token.DocComment
 }
 
+// commentAt reports whether index i is in range and holds a comment token.
+func commentAt(s *tokens.Stream, i int) bool {
+	return i >= 0 && i < s.Len() && isComment(s.At(i))
+}
+
 func isLineComment(t token.Token) bool {
 	if t.Kind != token.Comment {
 		return false

blink/internal/fixer/rules/no_singleline_whitespace_before_semicolons.go

@@ -30,6 +30,11 @@ func (NoSinglelineWhitespaceBeforeSemicolons) Fix(s *tokens.Stream) bool {
 		}
 		prev := s.At(i - 1)
 		if prev.Kind == token.Whitespace && !strings.ContainsAny(prev.Value, "\n\r") {
+			// keep the space before a ";" that follows another ";" - the empty
+			// condition of a for() header ("for ($i = 0; ; ++$i)")
+			if p, ok := prevSignificant(s, i-1); ok && p.Kind == token.Punct && p.Value == ";" {
+				continue
+			}
 			s.RemoveAt(i - 1)
 			changed = true
 		}

blink/internal/fixer/rules/rules_test.go

@@ -33,9 +33,10 @@ func TestSpaceAfterSemicolon(t *testing.T) {
 	if want := "<?php $a=1; $b=2;"; !changed || got != want {
 		t.Fatalf("changed=%v got=%q want=%q", changed, got, want)
 	}
-	// empty for-loop head must not gain spaces
-	if _, changed := apply(t, SpaceAfterSemicolon{}, "<?php for(;;){}"); changed {
-		t.Fatal("';' before ')' should be left alone")
+	// a ";" followed by ")" is left alone, but one followed by ";" (empty for
+	// condition) still gains a space, matching php-cs-fixer
+	if got, changed := apply(t, SpaceAfterSemicolon{}, "<?php for(;;){}"); !changed || got != "<?php for(; ;){}" {
+		t.Fatalf("empty for head: changed=%v got=%q", changed, got)
 	}
 }
 

blink/internal/fixer/rules/space_after_semicolon.go

@@ -56,7 +56,9 @@ func (f SpaceAfterSemicolon) Fix(s *tokens.Stream) bool {
 		if next.Kind == token.Whitespace || next.Kind == token.CloseTag {
 			continue
 		}
-		if next.Kind == token.Punct && (next.Value == ")" || next.Value == ";") {
+		// a space is still inserted before a following ";" (empty for condition:
+		// "for (;; )" -> "for (; ; )"); only ")" suppresses it
+		if next.Kind == token.Punct && next.Value == ")" {
 			continue
 		}
 		s.InsertAt(i+1, token.Token{Kind: token.Whitespace, Value: " "})

blink/internal/fixer/rules/token_batch.go

@@ -118,11 +118,14 @@ func (TrimArraySpaces) Fix(s *tokens.Stream) bool {
 		if closeIdx < 0 {
 			continue
 		}
-		if closeIdx-1 > i && s.At(closeIdx-1).Kind == token.Whitespace && !hasNewline(s.At(closeIdx-1).Value) {
+		// a space kept next to a comment is left alone ("[ // note" / "/* note */ ]")
+		if closeIdx-1 > i && s.At(closeIdx-1).Kind == token.Whitespace && !hasNewline(s.At(closeIdx-1).Value) &&
+			!commentAt(s, closeIdx-2) {
 			s.RemoveAt(closeIdx - 1)
 			changed = true
 		}
-		if i+1 < s.Len() && s.At(i+1).Kind == token.Whitespace && !hasNewline(s.At(i+1).Value) {
+		if i+1 < s.Len() && s.At(i+1).Kind == token.Whitespace && !hasNewline(s.At(i+1).Value) &&
+			!commentAt(s, i+2) {
 			s.RemoveAt(i + 1)
 			changed = true
 		}

blink/internal/lexer/lexer.go

@@ -137,7 +137,8 @@ func (l *lexer) lexPHP() {
 		l.lexString(start, '\'')
 	case c == '"':
 		l.lexString(start, '"')
-	case isDigit(c):
+	case isDigit(c) || (c == '.' && l.pos+1 < len(l.src) && isDigit(l.src[l.pos+1])):
+		// a leading-dot float (".1") is one DNUMBER in PHP's context-free scanner
 		for l.pos < len(l.src) && isNumber(l.src[l.pos]) {
 			l.pos++
 		}