feat(codegen)!: print StringLiteral raw if minify option disabled

crates/oxc_codegen/src/str.rs

@@ -28,6 +28,98 @@ impl Codegen<'_> {
     pub(crate) fn print_string_literal(&mut self, s: &StringLiteral<'_>, allow_backtick: bool) {
         self.add_source_mapping(s.span);
 
+        if !self.options.minify {
+            if let Some(raw) = s.raw {
+                self.print_string_literal_raw(&raw);
+                return;
+            }
+        }
+
+        self.print_string_literal_encode(s, allow_backtick);
+    }
+
+    /// Print a [`StringLiteral`], from its `raw` representation.
+    ///
+    /// Only change made is to change quotes to what's specified in `options.single_quote`.
+    fn print_string_literal_raw(&mut self, raw: &str) {
+        let quote = self.quote;
+        let raw_bytes = raw.as_bytes();
+
+        // Assertion here should remove the bounds checks from `raw_bytes[0]`,
+        // `raw_bytes[raw_bytes.len() - 1]` and `&raw_bytes[1..raw_bytes.len() - 1]` below
+        assert!(raw_bytes.len() >= 2);
+
+        // If `raw` already has desired quotes, print `raw` unchanged
+        if raw_bytes[0] == quote as u8 {
+            self.print_str(raw);
+            return;
+        }
+
+        // Escape quotes
+
+        // Cut off quotes from start and end of `raw`.
+        // Check the last char of `raw` is ASCII. This ensures that trimmed `raw_bytes` is a valid
+        // UTF-8 string, and doesn't end with an unfinished part of a unicode byte sequence.
+        // This ensures the safety of `print_bytes_unchecked` calls below.
+        // We have to do this check because we don't have a static guarantee that `raw` starts and ends
+        // with an ASCII quote. Without this check, a bug in parser could cause undefined behavior here.
+        assert!(raw_bytes[raw_bytes.len() - 1].is_ascii());
+        let raw_bytes = &raw_bytes[1..raw_bytes.len() - 1];
+        let mut bytes = raw_bytes.iter();
+        let mut chunk_start = bytes.as_slice().as_ptr();
+
+        quote.print(self);
+
+        while let Some(&byte) = bytes.clone().next() {
+            if byte == quote as u8 {
+                // Print up to before quote, print slash, and start next chunk on the quote,
+                // so quote gets pushed in next chunk.
+                // Note: `byte` was peeked, not consumed, so `quote_ptr` points to the quote, not after it.
+                let quote_ptr = bytes.as_slice().as_ptr();
+                // SAFETY: `chunk_start` points to either start of string content or an ASCII quote char.
+                // Either way, that's on a UTF-8 char boundary, and in bounds of `raw_bytes`.
+                // `quote_ptr` points to an ASCII quote char, so also on a UTF-8 char boundary.
+                // `quote_ptr >= chunk_start` because `bytes` only gets advanced,
+                // and `chunk_start` is either start of string or a previous `quote_ptr`.
+                unsafe {
+                    let chunk_len = quote_ptr.offset_from(chunk_start);
+                    let chunk_len = usize::try_from(chunk_len).unwrap_unchecked();
+                    let chunk = slice::from_raw_parts(chunk_start, chunk_len);
+                    self.code.print_bytes_unchecked(chunk);
+                }
+                self.print_ascii_byte(b'\\');
+                chunk_start = quote_ptr;
+
+                // Consume the quote
+                bytes.next().unwrap();
+            } else if byte == b'\\' {
+                // Consume slash and next byte.
+                // Next byte might be an escaped quote which don't want to escape again e.g. `\"`.
+                bytes.next().unwrap();
+                bytes.next().unwrap();
+            } else {
+                // Consume the peeked byte
+                bytes.next().unwrap();
+            }
+        }
+
+        // SAFETY: `chunk_start` points to either start of string content or an ASCII quote character.
+        // Either way, that's on a UTF-8 char boundary, and in bounds of `raw_bytes`.
+        // `bytes` is exhausted, so `bytes.as_slice().as_ptr()` points to end of `raw_bytes`.
+        // `chunk_start` must be before it, or (if string is empty) equal to it.
+        unsafe {
+            let end_ptr = bytes.as_slice().as_ptr();
+            let chunk_len = end_ptr.offset_from(chunk_start);
+            let chunk_len = usize::try_from(chunk_len).unwrap_unchecked();
+            let chunk = slice::from_raw_parts(chunk_start, chunk_len);
+            self.code.print_bytes_unchecked(chunk);
+        }
+
+        quote.print(self);
+    }
+
+    /// Print a [`StringLiteral`], re-encoding from its `value`.
+    fn print_string_literal_encode(&mut self, s: &StringLiteral<'_>, allow_backtick: bool) {
         // If `minify` option enabled, quote will be chosen depending on what produces shortest output.
         // What is the best quote to use will be determined when first character needing escape is found.
         // This avoids iterating through the string twice if it contains no quotes (common case).

crates/oxc_codegen/tests/integration/esbuild.rs

@@ -333,52 +333,89 @@ fn test_nullish() {
 
 #[test]
 fn test_string() {
+    // No `minify` option
+
+    // Prints double-quoted strings as in original
+    test("let x = \"\"", "let x = \"\";\n");
+    test("let x = \"abc\"", "let x = \"abc\";\n");
+    test("let x = \"\t\"", "let x = \"\t\";\n");
+    test("let x = \"\\t\"", "let x = \"\\t\";\n");
+
+    // Converts single quote to double
     test("let x = ''", "let x = \"\";\n");
-    test("let x = '\\b'", "let x = \"\\b\";\n");
-    test("let x = '\\f'", "let x = \"\\f\";\n");
-    test("let x = '\t'", "let x = \"\t\";\n");
-    test("let x = '\\v'", "let x = \"\\v\";\n");
-    test("let x = '\\n'", "let x = \"\\n\";\n");
-    test("let x = '\\r'", "let x = \"\\r\";\n");
-    test("let x = '\\r\\n'", "let x = \"\\r\\n\";\n");
-    test("let x = '\\''", "let x = \"'\";\n");
+    test("let x = 'abc'", "let x = \"abc\";\n");
     test("let x = '\"'", "let x = \"\\\"\";\n");
-    test("let x = '`'", "let x = \"`\";\n");
-    test("let x = '\\'\"'", "let x = \"'\\\"\";\n");
-    test("let x = '\\'`'", "let x = \"'`\";\n");
-    test("let x = '\"`'", "let x = \"\\\"`\";\n");
-    test("let x = '\\\\'", "let x = \"\\\\\";\n");
-    test("let x = '\x00'", "let x = \"\\0\";\n");
-    test("let x = '\x00!'", "let x = \"\\0!\";\n");
-    test("let x = '\x001'", "let x = \"\\x001\";\n");
-    test("let x = '\\0'", "let x = \"\\0\";\n");
-    test("let x = '\\0!'", "let x = \"\\0!\";\n");
-    test("let x = '\x07'", "let x = \"\\x07\";\n");
-    test("let x = '\x07!'", "let x = \"\\x07!\";\n");
-    test("let x = '\x071'", "let x = \"\\x071\";\n");
-    test("let x = '\\7'", "let x = \"\\x07\";\n");
-    test("let x = '\\7!'", "let x = \"\\x07!\";\n");
-    test("let x = '\\01'", "let x = \"\x01\";\n");
-    test("let x = '\x10'", "let x = \"\x10\";\n");
-    test("let x = '\\x10'", "let x = \"\x10\";\n");
-    test("let x = '\x1B'", "let x = \"\\x1B\";\n");
-    test("let x = '\\x1B'", "let x = \"\\x1B\";\n");
-    test("let x = '\\x41'", "let x = \"A\";\n");
-    test("let x = '\u{ABCD}'", "let x = \"\u{ABCD}\";\n");
-    test("let x = '\\uABCD'", "let x = \"\u{ABCD}\";\n");
-    test("let x = '\\U000123AB'", "let x = \"U000123AB\";\n");
-    test("let x = '\\u{123AB}'", "let x = \"\u{123ab}\";\n");
-    test("let x = '\\u{41}'", "let x = \"A\";\n");
-    test("let x = '\\uD808\\uDFAB'", "let x = \"\u{123ab}\";\n");
-    test("let x = '\\uD808'", "let x = \"\\ud808\";\n"); // lone surrogate
-    test("let x = '\\uD808X'", "let x = \"\\ud808X\";\n");
-    test("let x = '\\uDFAB'", "let x = \"\\udfab\";\n");
-    test("let x = '\\uDFABX'", "let x = \"\\udfabX\";\n");
-
-    test("let x = '\\x80'", "let x = \"\u{80}\";\n");
-    test("let x = '\\xFF'", "let x = \"ÿ\";\n");
-    test("let x = '\\xF0\\x9F\\x8D\\x95'", "let x = \"ð\u{9f}\u{8d}\u{95}\";\n");
-    test("let x = '\\uD801\\uDC02\\uDC03\\uD804'", "let x = \"𐐂\\udc03\\ud804\";\n"); // surrogates
+    test("let x = 'abc\"'", "let x = \"abc\\\"\";\n");
+    test("let x = 'abc\"\"\"'", "let x = \"abc\\\"\\\"\\\"\";\n");
+    test("let x = '\"def'", "let x = \"\\\"def\";\n");
+    test("let x = '\"\"\"def'", "let x = \"\\\"\\\"\\\"def\";\n");
+    test("let x = 'abc\"def'", "let x = \"abc\\\"def\";\n");
+    test("let x = 'abc\"\"\"def\"\"\"ghi'", "let x = \"abc\\\"\\\"\\\"def\\\"\\\"\\\"ghi\";\n");
+    // Does not double-escape already-escaped quotes
+    test("let x = '\\\"'", "let x = \"\\\"\";\n");
+    test("let x = 'abc\\\"\\\"'", "let x = \"abc\\\"\\\"\";\n");
+    test("let x = '\\\"\\\"def'", "let x = \"\\\"\\\"def\";\n");
+    test("let x = 'abc\\\"\\\"def'", "let x = \"abc\\\"\\\"def\";\n");
+    test("let x = '\\r\\n\"'", "let x = \"\\r\\n\\\"\";\n");
+    test("let x = '\\\\\"'", "let x = \"\\\\\\\"\";\n");
+    test("let x = '\\\\\\\"'", "let x = \"\\\\\\\"\";\n");
+    // Does not escape other characters
+    test("let x = '\t'", "let x = \"\t\";\n");
+    // Prints other escapes as in original
+    test("let x = '\\t'", "let x = \"\\t\";\n");
+    test("let x = '\\x41'", "let x = \"\\x41\";\n");
+    test("let x = '\\u{41}'", "let x = \"\\u{41}\";\n");
+    test("let x = '\\uD800'", "let x = \"\\uD800\";\n");
+    test("let x = '\\uD801\\uDC02'", "let x = \"\\uD801\\uDC02\";\n");
+
+    // `minify` option
+
+    // Escapes characters and chooses best quote character
+    test_minify("let x = ''", "let x=``;");
+    test_minify("let x = '\\b'", "let x=`\\b`;");
+    test_minify("let x = '\\f'", "let x=`\\f`;");
+    test_minify("let x = '\t'", "let x=`\t`;");
+    test_minify("let x = '\\v'", "let x=`\\v`;");
+    test_minify("let x = '\\n'", "let x=`\n`;");
+    test_minify("let x = '\\r'", "let x=`\\r`;");
+    test_minify("let x = '\\r\\n'", "let x=`\\r\n`;");
+    test_minify("let x = '\\''", "let x=`'`;");
+    test_minify("let x = '\"'", "let x=`\"`;");
+    test_minify("let x = '`'", "let x=\"`\";");
+    test_minify("let x = '\\'\"'", "let x=`'\"`;");
+    test_minify("let x = '\\'`'", "let x=\"'`\";");
+    test_minify("let x = '\"`'", "let x='\"`';");
+    test_minify("let x = '\\\\'", "let x=`\\\\`;");
+    test_minify("let x = '\x00'", "let x=`\\0`;");
+    test_minify("let x = '\x00!'", "let x=`\\0!`;");
+    test_minify("let x = '\x001'", "let x=`\\x001`;");
+    test_minify("let x = '\\0'", "let x=`\\0`;");
+    test_minify("let x = '\\0!'", "let x=`\\0!`;");
+    test_minify("let x = '\x07'", "let x=`\\x07`;");
+    test_minify("let x = '\x07!'", "let x=`\\x07!`;");
+    test_minify("let x = '\x071'", "let x=`\\x071`;");
+    test_minify("let x = '\\7'", "let x=`\\x07`;");
+    test_minify("let x = '\\7!'", "let x=`\\x07!`;");
+    test_minify("let x = '\\01'", "let x=`\x01`;");
+    test_minify("let x = '\x10'", "let x=`\x10`;");
+    test_minify("let x = '\\x10'", "let x=`\x10`;");
+    test_minify("let x = '\x1B'", "let x=`\\x1B`;");
+    test_minify("let x = '\\x1B'", "let x=`\\x1B`;");
+    test_minify("let x = '\\x41'", "let x=`A`;");
+    test_minify("let x = '\u{ABCD}'", "let x=`\u{ABCD}`;");
+    test_minify("let x = '\\uABCD'", "let x=`\u{ABCD}`;");
+    test_minify("let x = '\\U000123AB'", "let x=`U000123AB`;");
+    test_minify("let x = '\\u{123AB}'", "let x=`\u{123ab}`;");
+    test_minify("let x = '\\u{41}'", "let x=`A`;");
+    test_minify("let x = '\\uD808\\uDFAB'", "let x=`\u{123ab}`;");
+    test_minify("let x = '\\uD808'", "let x=`\\ud808`;"); // lone surrogate
+    test_minify("let x = '\\uD808X'", "let x=`\\ud808X`;");
+    test_minify("let x = '\\uDFAB'", "let x=`\\udfab`;");
+    test_minify("let x = '\\uDFABX'", "let x=`\\udfabX`;");
+    test_minify("let x = '\\x80'", "let x=`\u{80}`;");
+    test_minify("let x = '\\xFF'", "let x=`ÿ`;");
+    test_minify("let x = '\\xF0\\x9F\\x8D\\x95'", "let x=`ð\u{9f}\u{8d}\u{95}`;");
+    test_minify("let x = '\\uD801\\uDC02\\uDC03\\uD804'", "let x=`𐐂\\udc03\\ud804`;"); // surrogates
 }
 
 #[test]

crates/oxc_codegen/tests/integration/unit.rs

@@ -150,7 +150,13 @@ fn unicode_escape() {
     test("console.log('こんにちは');", "console.log(\"こんにちは\");\n");
     test("console.log('안녕하세요');", "console.log(\"안녕하세요\");\n");
     test("console.log('🧑‍🤝‍🧑');", "console.log(\"🧑‍🤝‍🧑\");\n");
-    test("console.log(\"\\uD800\\uD801\")", "console.log(\"\\ud800\\ud801\");\n");
+    test("console.log(\"\\uD800\\uD801\")", "console.log(\"\\uD800\\uD801\");\n");
+
+    test_minify("console.log('你好');", "console.log(`你好`);");
+    test_minify("console.log('こんにちは');", "console.log(`こんにちは`);");
+    test_minify("console.log('안녕하세요');", "console.log(`안녕하세요`);");
+    test_minify("console.log('🧑‍🤝‍🧑');", "console.log(`🧑‍🤝‍🧑`);");
+    test_minify("console.log(\"\\uD800\\uD801\")", "console.log(`\\ud800\\ud801`);");
 }
 
 #[test]
@@ -511,53 +517,59 @@ fn getter_setter() {
 
 #[test]
 fn string() {
+    // Uses quotes as requested in options
+    let single_quote = CodegenOptions { single_quote: true, ..CodegenOptions::default() };
+    test_options("let x = \"'\";", "let x = '\\'';\n", single_quote);
+    let double_quote = CodegenOptions { single_quote: false, ..CodegenOptions::default() };
+    test_options("let x = '\\\"';", "let x = \"\\\"\";\n", double_quote);
+
     // `${` only escaped when quote is backtick
-    test("let x = \"${}\";", "let x = \"${}\";\n");
-    test_minify("let x = \"${}\";", "let x=\"${}\";");
+    test("let x = '${}';", "let x = \"${}\";\n");
+    test_minify("let x = '${}';", "let x=\"${}\";");
     test("let x = '\"\"${}';", "let x = \"\\\"\\\"${}\";\n");
     test_minify("let x = '\"\"${}';", "let x='\"\"${}';");
-    test("let x = '\"\"\\'\\'${}';", "let x = \"\\\"\\\"''${}\";\n");
+    test("let x = '\"\"\\'\\'${}';", "let x = \"\\\"\\\"\\'\\'${}\";\n");
     test_minify("let x = '\"\"\\'\\'${}';", "let x=`\"\"''\\${}`;");
     test_minify("let x = '\\'\\'\\'\"\"\"${}';", "let x=`'''\"\"\"\\${}`;");
 
     // Lossy replacement character
-    test("let x = \"�\\u{FFFD}\";", "let x = \"��\";\n");
+    test("let x = '�\\u{FFFD}';", "let x = \"�\\u{FFFD}\";\n");
     test_minify("let x = \"�\\u{FFFD}\";", "let x=`��`;");
     test(
-        "let x = \"� ��� \\u{FFFD} \\u{FFFD}\\u{FFFD}\\u{FFFD} �\";",
-        "let x = \"� ��� � ��� �\";\n",
+        "let x = '� ��� \\u{FFFD} \\u{FFFD}\\u{FFFD}\\u{FFFD} �';",
+        "let x = \"� ��� \\u{FFFD} \\u{FFFD}\\u{FFFD}\\u{FFFD} �\";\n",
     );
     test_minify(
-        "let x = \"� ��� \\u{FFFD} \\u{FFFD}\\u{FFFD}\\u{FFFD} �\";",
+        "let x = '� ��� \\u{FFFD} \\u{FFFD}\\u{FFFD}\\u{FFFD} �';",
         "let x=`� ��� � ��� �`;",
     );
     // Lone surrogates
     test(
-        "let x = \"\\uD800 \\uDBFF \\uDC00 \\uDFFF\";",
-        "let x = \"\\ud800 \\udbff \\udc00 \\udfff\";\n",
+        "let x = '\\uD800 \\uDBFF \\uDC00 \\uDFFF';",
+        "let x = \"\\uD800 \\uDBFF \\uDC00 \\uDFFF\";\n",
     );
     test_minify(
-        "let x = \"\\uD800 \\uDBFF \\uDC00 \\uDFFF\";",
+        "let x = '\\uD800 \\uDBFF \\uDC00 \\uDFFF';",
         "let x=`\\ud800 \\udbff \\udc00 \\udfff`;",
     );
-    test("let x = \"\\uD800\u{41}\";", "let x = \"\\ud800A\";\n");
-    test_minify("let x = \"\\uD800\u{41}\";", "let x=`\\ud800A`;");
+    test("let x = '\\uD800\\u{41}';", "let x = \"\\uD800\\u{41}\";\n");
+    test_minify("let x = '\\uD800\\u{41}';", "let x=`\\ud800A`;");
     // Invalid pairs
     test(
-        "let x = \"\\uD800\\uDBFF \\uDC00\\uDFFF\";",
-        "let x = \"\\ud800\\udbff \\udc00\\udfff\";\n",
+        "let x = '\\uD800\\uDBFF \\uDC00\\uDFFF';",
+        "let x = \"\\uD800\\uDBFF \\uDC00\\uDFFF\";\n",
     );
     test_minify(
-        "let x = \"\\uD800\\uDBFF \\uDC00\\uDFFF\";",
+        "let x = '\\uD800\\uDBFF \\uDC00\\uDFFF';",
         "let x=`\\ud800\\udbff \\udc00\\udfff`;",
     );
     // Lone surrogates and lossy replacement characters
     test(
-        "let x = \"��\\u{FFFD}\\u{FFFD}\\uD800\\uDBFF��\\u{FFFD}\\u{FFFD}\\uDC00\\uDFFF��\\u{FFFD}\\u{FFFD}\";",
-        "let x = \"����\\ud800\\udbff����\\udc00\\udfff����\";\n",
+        "let x = '��\\u{FFFD}\\u{FFFD}\\uD800\\uDBFF��\\u{FFFD}\\u{FFFD}\\uDC00\\uDFFF��\\u{FFFD}\\u{FFFD}';",
+        "let x = \"��\\u{FFFD}\\u{FFFD}\\uD800\\uDBFF��\\u{FFFD}\\u{FFFD}\\uDC00\\uDFFF��\\u{FFFD}\\u{FFFD}\";\n",
     );
     test_minify(
-        "let x = \"��\\u{FFFD}\\u{FFFD}\\uD800\\uDBFF��\\u{FFFD}\\u{FFFD}\\uDC00\\uDFFF��\\u{FFFD}\\u{FFFD}\";",
+        "let x = '��\\u{FFFD}\\u{FFFD}\\uD800\\uDBFF��\\u{FFFD}\\u{FFFD}\\uDC00\\uDFFF��\\u{FFFD}\\u{FFFD}';",
         "let x=`����\\ud800\\udbff����\\udc00\\udfff����`;",
     );