fix: allow unmatched quotes in script tags, fixes #1110 (#1111)

Author: a-h

.version

@@ -1 +1 @@
-0.3.856
+0.3.857

parser/v2/scriptparser.go

@@ -8,6 +8,15 @@ import (
 
 var scriptElement = scriptElementParser{}
 
+type jsQuote string
+
+const (
+	jsQuoteNone     jsQuote = ""
+	jsQuoteSingle   jsQuote = `'`
+	jsQuoteDouble   jsQuote = `"`
+	jsQuoteBacktick jsQuote = "`"
+)
+
 type scriptElementParser struct{}
 
 func (p scriptElementParser) Parse(pi *parse.Input) (n Node, ok bool, err error) {
@@ -51,7 +60,7 @@ func (p scriptElementParser) Parse(pi *parse.Input) (n Node, ok bool, err error)
 
 	// Parse the contents, we should get script text or Go expressions up until the closing tag.
 	var sb strings.Builder
-	var isInsideStringLiteral bool
+	var stringLiteralDelimiter jsQuote
 
 loop:
 	for {
@@ -82,7 +91,7 @@ loop:
 			return nil, false, err
 		}
 		if ok {
-			e.Contents = append(e.Contents, NewScriptContentsGo(code.(GoCode), isInsideStringLiteral))
+			e.Contents = append(e.Contents, NewScriptContentsGo(code.(GoCode), stringLiteralDelimiter != jsQuoteNone))
 			continue loop
 		}
 
@@ -108,13 +117,18 @@ loop:
 			if ok {
 				_, isEOF, _ := parse.EOF[string]().Parse(pi)
 				if c == `"` || c == "'" || c == "`" {
-					isInsideStringLiteral = !isInsideStringLiteral
+					// Start or exit a string literal.
+					if stringLiteralDelimiter == jsQuoteNone {
+						stringLiteralDelimiter = jsQuote(c)
+					} else if stringLiteralDelimiter == jsQuote(c) {
+						stringLiteralDelimiter = jsQuoteNone
+					}
 				}
 				peeked, _ := pi.Peek(1)
 				peeked = c + peeked
 
 				breakForGo := peeked == "{{"
-				breakForHTML := !isInsideStringLiteral && (peeked == "</" || peeked == "//" || peeked == "/*")
+				breakForHTML := stringLiteralDelimiter == jsQuoteNone && (peeked == "</" || peeked == "//" || peeked == "/*")
 
 				if isEOF || breakForGo || breakForHTML {
 					if sb.Len() > 0 {
@@ -143,7 +157,21 @@ var endTagStart = parse.String("</")
 
 var jsCharacter = parse.Any(jsEscapedCharacter, parse.AnyRune)
 
-var jsEscapedCharacter = parse.StringFrom(parse.String("\\"), parse.AnyRune)
+// \uXXXX	Unicode code point escape	'\u0061' = 'a'
+var hexDigit = parse.Any(parse.ZeroToNine, parse.RuneIn("abcdef"), parse.RuneIn("ABCDEF"))
+var jsUnicodeEscape = parse.StringFrom(parse.String("\\u"), hexDigit, hexDigit, hexDigit, hexDigit)
+
+// \u{X...}	ES6+ extended Unicode escape	'\u{1F600}' = '😀'
+var jsExtendedUnicodeEscape = parse.StringFrom(parse.String("\\u{"), hexDigit, parse.StringFrom(parse.AtLeast(1, parse.ZeroOrMore(hexDigit))), parse.String("}"))
+
+// \xXX	Hex code (2-digit)	'\x41' = 'A'
+var jsHexEscape = parse.StringFrom(parse.String("\\x"), hexDigit, hexDigit)
+
+// \x Backslash escape	'\\' = '\'
+var jsBackslashEscape = parse.StringFrom(parse.String("\\"), parse.AnyRune)
+
+// All escapes.
+var jsEscapedCharacter = parse.Any(jsBackslashEscape, jsUnicodeEscape, jsHexEscape, jsExtendedUnicodeEscape)
 
 var jsComment = parse.Any(jsSingleLineComment, jsMultiLineComment)
 

parser/v2/scriptparser_test.go

@@ -2,6 +2,7 @@ package parser
 
 import (
 	"path/filepath"
+	"strings"
 	"testing"
 
 	_ "embed"
@@ -39,11 +40,20 @@ func TestScriptElementParserPlain(t *testing.T) {
 			if !isScriptElement {
 				t.Fatalf("expected ScriptElement, got %T", result)
 			}
-			if len(se.Contents) != 1 {
-				t.Fatalf("expected 1 content, got %d", len(se.Contents))
+
+			var actual strings.Builder
+			for _, content := range se.Contents {
+				if content.GoCode != nil {
+					t.Fatalf("expected plain text, got GoCode")
+				}
+				if content.Value == nil {
+					t.Fatalf("expected plain text, got nil")
+				}
+				actual.WriteString(*content.Value)
 			}
+
 			expected := clean(a.Files[1].Data)
-			if diff := cmp.Diff(*se.Contents[0].Value, string(expected)); diff != "" {
+			if diff := cmp.Diff(actual.String(), string(expected)); diff != "" {
 				t.Fatalf("%s:\n%s", file, diff)
 			}
 		})

parser/v2/scriptparsertestdata/backtickquote.txt

@@ -0,0 +1,8 @@
+-- in --
+<script>
+window.alert(`This is 'single quoted' and this is "double quoted"`);
+</script>
+-- out --
+
+window.alert(`This is 'single quoted' and this is "double quoted"`);
+

parser/v2/scriptparsertestdata/backtickquote_apostrophe.txt

@@ -0,0 +1,8 @@
+-- in --
+<script>
+window.alert(`You can use isn't and other text`);
+</script>
+-- out --
+
+window.alert(`You can use isn't and other text`);
+

parser/v2/scriptparsertestdata/doublequote.txt

@@ -0,0 +1,8 @@
+-- in --
+<script>
+window.alert("This is 'quoted'");
+</script>
+-- out --
+
+window.alert("This is 'quoted'");
+

parser/v2/scriptparsertestdata/doublequote_apostrophe.txt

@@ -0,0 +1,8 @@
+-- in --
+<script>
+window.alert("You can use isn't and other text");
+</script>
+-- out --
+
+window.alert("You can use isn't and other text");
+

parser/v2/scriptparsertestdata/escapechars.txt

@@ -0,0 +1,34 @@
+-- in --
+<script>
+const singleQuotedBackslashEscape = '\x61'; // a
+const singleQuotedHexEscape = '\x61';
+const singleQuotedUnicodeEscape = '\u0061';
+const singleQuotedExtendedUnicodeEscape = '\u{61}';
+
+const doubleQuotedBackslashEscape = "\x61"; // a
+const doubleQuotedHexEscape = "\x61";
+const doubleQuotedUnicodeEscape = "\u0061";
+const doubleQuotedExtendedUnicodeEscape = "\u{61}";
+
+const backtickQuotedBackslashEscape = `\x61`; // a
+const backtickQuotedHexEscape = `\x61`;
+const backtickQuotedUnicodeEscape = `\u0061`;
+const backtickQuotedExtendedUnicodeEscape = `\u{61}`;
+</script>
+-- out --
+
+const singleQuotedBackslashEscape = '\x61'; // a
+const singleQuotedHexEscape = '\x61';
+const singleQuotedUnicodeEscape = '\u0061';
+const singleQuotedExtendedUnicodeEscape = '\u{61}';
+
+const doubleQuotedBackslashEscape = "\x61"; // a
+const doubleQuotedHexEscape = "\x61";
+const doubleQuotedUnicodeEscape = "\u0061";
+const doubleQuotedExtendedUnicodeEscape = "\u{61}";
+
+const backtickQuotedBackslashEscape = `\x61`; // a
+const backtickQuotedHexEscape = `\x61`;
+const backtickQuotedUnicodeEscape = `\u0061`;
+const backtickQuotedExtendedUnicodeEscape = `\u{61}`;
+

parser/v2/scriptparsertestdata/singlequote.txt

@@ -0,0 +1,8 @@
+-- in --
+<script>
+window.alert('This is "quoted"');
+</script>
+-- out --
+
+window.alert('This is "quoted"');
+

parser/v2/scriptparsertestdata/singlequote_apostrophe.txt

@@ -0,0 +1,8 @@
+-- in --
+<script>
+window.alert('An unclosed " is allowed in single quotes');
+</script>
+-- out --
+
+window.alert('An unclosed " is allowed in single quotes');
+