JS: fix escaping of single/double/backtick quotes in strings when enc…

…oded as unicode escape literal, fixes #654
tdewolff · Jan 6, 2024 · 6b45973 · 6b45973
1 parent e00409b
commit 6b45973
Show file tree

Hide file tree

Showing 3 changed files with 80 additions and 13 deletions.
diff --git a/js/js_test.go b/js/js_test.go
@@ -153,7 +153,7 @@ func TestJS(t *testing.T) {
 		{`"string\∀string"`, `"string∀string"`},
 		{`"string\0\uFFFFstring"`, "\"string\\0\uffffstring\""},
 		{`"string\x00\x55\x0A\x0D\x22\x27string"`, "`string\\x00U\n\r\"'string`"},
-		{`"string\000\12\015\042\47\411string"`, "\"string\\0\\n\\r\\\"'!1string\""},
+		{`"string\000\12\015\042\47\411string"`, "`string\\0\n\r\"'!1string`"},
 		{`"\x005"`, `"\x005"`},
 		{"'string\\n\\rstring'", "`string\n\rstring`"},
 		{"'string\\\r\nstring\\\nstring\\\rstring\\\u2028string\\\u2029string'", `"stringstringstringstringstringstring"`},

diff --git a/js/util.go b/js/util.go
@@ -940,7 +940,6 @@ func minifyString(b []byte, allowTemplate bool) []byte {
 	backtickQuotes := 0
 	newlines := 0
 	dollarSigns := 0
-	hasOctals := false
 	for i := 1; i < len(b)-1; i++ {
 		if b[i] == '\'' {
 			singleQuotes++
@@ -953,36 +952,64 @@ func minifyString(b []byte, allowTemplate bool) []byte {
 		} else if b[i] == '\\' && i+1 < len(b) {
 			if b[i+1] == 'n' || b[i+1] == 'r' {
 				newlines++
-			} else if '1' <= b[i+1] && b[i+1] <= '9' || b[i+1] == '0' && i+2 < len(b) && '0' <= b[i+2] && b[i+2] <= '9' {
-				if i+2 < len(b) && b[i+1] == '1' && (b[i+2] == '2' || b[i+2] == '5') {
+			} else if '1' <= b[i+1] && b[i+1] <= '9' && i+2 < len(b) {
+				if b[i+1] == '1' && (b[i+2] == '2' || b[i+2] == '5') {
 					newlines++
-				} else {
-					hasOctals = true
+				} else if b[i+1] == '4' && b[i+2] == '2' {
+					doubleQuotes++
+				} else if b[i+1] == '4' && b[i+2] == '7' {
+					singleQuotes++
+				} else if i+3 < len(b) && b[i+1] == '1' && b[i+2] == '4' && b[i+3] == '0' {
+					backtickQuotes++
+				}
+			} else if b[i+1] == 'x' && i+3 < len(b) {
+				if b[i+2] == '0' && (b[i+3]|0x20 == 'a' || b[i+3]|0x20 == 'd') {
+					newlines++
+				} else if b[i+2] == '2' && b[i+3] == '2' {
+					doubleQuotes++
+				} else if b[i+2] == '2' && b[i+3] == '7' {
+					singleQuotes++
+				} else if b[i+2] == '6' && b[i+3] == '0' {
+					backtickQuotes++
+				}
+			} else if b[i+1] == 'u' && i+5 < len(b) && b[i+2] == '0' && b[i+3] == '0' {
+				if b[i+4] == '0' && (b[i+5]|0x20 == 'a' || b[i+5]|0x20 == 'd') {
+					newlines++
+				} else if b[i+4] == '2' && b[i+5] == '2' {
+					doubleQuotes++
+				} else if b[i+4] == '2' && b[i+5] == '7' {
+					singleQuotes++
+				} else if b[i+4] == '6' && b[i+5] == '0' {
+					backtickQuotes++
 				}
-			} else if b[i+1] == 'x' && i+3 < len(b) && b[i+2] == '0' && (b[i+3]|0x20 == 'a' || b[i+3]|0x20 == 'd') {
-				newlines++
-			} else if b[i+1] == 'u' && i+5 < len(b) && b[i+2] == '0' && b[i+3] == '0' && b[i+4] == '0' && (b[i+5]|0x20 == 'a' || b[i+5]|0x20 == 'd') {
-				newlines++
 			} else if b[i+1] == 'u' && i+4 < len(b) && b[i+2] == '{' {
 				j := i + 3
 				for j < len(b) && b[j] == '0' {
 					j++
 				}
 				if j+1 < len(b) && (b[j]|0x20 == 'a' || b[j]|0x20 == 'd') && b[j+1] == '}' {
 					newlines++
+				} else if j+2 < len(b) && b[j+2] == '}' {
+					if b[j] == '2' && b[j+1] == '2' {
+						doubleQuotes++
+					} else if b[j] == '2' && b[j+1] == '7' {
+						singleQuotes++
+					} else if b[j] == '6' && b[j+1] == '0' {
+						backtickQuotes++
+					}
 				}
 			}
 		}
 	}
 	quote := byte('"') // default to " for better GZIP compression
-	quotes := singleQuotes
+	quotes := doubleQuotes
 	if doubleQuotes < singleQuotes {
 		quote = byte('"')
-		quotes = doubleQuotes
 	} else if singleQuotes < doubleQuotes {
 		quote = byte('\'')
+		quotes = singleQuotes
 	}
-	if allowTemplate && !hasOctals && backtickQuotes+dollarSigns < quotes+newlines {
+	if allowTemplate && backtickQuotes+dollarSigns < quotes+newlines {
 		quote = byte('`')
 	}
 	b[0] = quote
@@ -1084,6 +1111,10 @@ func replaceEscapes(b []byte, quote byte, prefix, suffix int) []byte {
 					if m == -1 {
 						i++
 						continue
+					} else if num < 256 && quote == byte(num) {
+						b[i] = '\\'
+						i++
+						n--
 					}
 					utf8.EncodeRune(b[i:], rune(num))
 					i += m

diff --git a/js/util_test.go b/js/util_test.go
@@ -56,6 +56,42 @@ func TestString(t *testing.T) {
 		{`"\x0A\x0D${${"`, `"\n\r${${"`},
 		{`"\u000A\u000D${${"`, `"\n\r${${"`},
 		{`"\u{A}\u{D}${${"`, `"\n\r${${"`},
+		{`"\42"`, `'"'`},
+		{`"\x22"`, `'"'`},
+		{`"\u0022"`, `'"'`},
+		{`"\u{22}"`, `'"'`},
+		{`"\42''"`, "`\"''`"},
+		{`"\x22''"`, "`\"''`"},
+		{`"\u0022''"`, "`\"''`"},
+		{`"\u{022}''"`, "`\"''`"},
+		{"\"\\42''``\"", "\"\\\"''``\""},
+		{"\"\\x22''``\"", "\"\\\"''``\""},
+		{"\"\\u0022''``\"", "\"\\\"''``\""},
+		{"\"\\u{0022}''``\"", "\"\\\"''``\""},
+		{`'\47'`, `"'"`},
+		{`'\x27'`, `"'"`},
+		{`'\u0027'`, `"'"`},
+		{`'\u{27}'`, `"'"`},
+		{`'\47""'`, "`'\"\"`"},
+		{`'\x27""'`, "`'\"\"`"},
+		{`'\u0027""'`, "`'\"\"`"},
+		{`'\u{027}""'`, "`'\"\"`"},
+		{"'\\47\"\"``'", "'\\'\"\"``'"},
+		{"'\\x27\"\"``'", "'\\'\"\"``'"},
+		{"'\\u0027\"\"``'", "'\\'\"\"``'"},
+		{"'\\u{0027}\"\"``'", "'\\'\"\"``'"},
+		{`'\140'`, "\"`\""},
+		{`'\x60'`, "\"`\""},
+		{`'\u0060'`, "\"`\""},
+		{`'\u{60}'`, "\"`\""},
+		{`'\140""'`, "'`\"\"'"},
+		{`'\x60""'`, "'`\"\"'"},
+		{`'\u0060""'`, "'`\"\"'"},
+		{`'\u{060}""'`, "'`\"\"'"},
+		{`'\140""\'\''`, "`\\`\"\"''`"},
+		{`'\x60""\'\''`, "`\\`\"\"''`"},
+		{`'\u0060""\'\''`, "`\\`\"\"''`"},
+		{`'\u{0060}""\'\''`, "`\\`\"\"''`"},
 	}
 
 	for _, tt := range tests {