fix(es/lexer): Fix lexing of hexadecimal escape sequences (#2838)

swc-project · Nov 24, 2021 · cf777c5 · cf777c5 · github-actions · Nov 24, 2021
1 parent a8bda3b
commit cf777c5
Show file tree

Hide file tree

Showing 20 changed files with 49 additions and 13 deletions.
diff --git a/crates/swc/tests/fixture/issue-2822/1/input/.swcrc b/crates/swc/tests/fixture/issue-2822/1/input/.swcrc
@@ -0,0 +1,5 @@
+{
+  "jsc": {
+    "target": "es5"
+  }
+}
diff --git a/crates/swc/tests/fixture/issue-2822/1/input/index.js b/crates/swc/tests/fixture/issue-2822/1/input/index.js
@@ -0,0 +1 @@
+console.log`\x22\x21\x224`
diff --git a/crates/swc/tests/fixture/issue-2822/1/output/index.js b/crates/swc/tests/fixture/issue-2822/1/output/index.js
@@ -0,0 +1,22 @@
+function _taggedTemplateLiteral(strings, raw) {
+    if (!raw) {
+        raw = strings.slice(0);
+    }
+    return Object.freeze(Object.defineProperties(strings, {
+        raw: {
+            value: Object.freeze(raw)
+        }
+    }));
+}
+function _templateObject() {
+    var data = _taggedTemplateLiteral([
+        "\"!\"4"
+    ], [
+        "\\x22\\x21\\x224"
+    ]);
+    _templateObject = function _templateObject() {
+        return data;
+    };
+    return data;
+}
+console.log(_templateObject());
diff --git a/crates/swc/tests/fixture/issue-2822/2/input/.swcrc b/crates/swc/tests/fixture/issue-2822/2/input/.swcrc
@@ -0,0 +1,5 @@
+{
+  "jsc": {
+    "target": "es2015"
+  }
+}
diff --git a/crates/swc/tests/fixture/issue-2822/2/input/index.js b/crates/swc/tests/fixture/issue-2822/2/input/index.js
@@ -0,0 +1 @@
+`\x22\x21\x224`
diff --git a/crates/swc/tests/fixture/issue-2822/2/output/index.js b/crates/swc/tests/fixture/issue-2822/2/output/index.js
@@ -0,0 +1 @@
+`\x22\x21\x224`;
diff --git a/.../es6/templates/templateStringControlCharacterEscapes01/input.ts/es2015.1.normal/output.js b/.../es6/templates/templateStringControlCharacterEscapes01/input.ts/es2015.1.normal/output.js
@@ -1 +1 @@
-var x = `\0\0x00\u0000 0 00 0000`;
+var x = `\0\x00\u0000 0 00 0000`;
diff --git a/.../templates/templateStringControlCharacterEscapes01_ES6/input.ts/es2015.1.normal/output.js b/.../templates/templateStringControlCharacterEscapes01_ES6/input.ts/es2015.1.normal/output.js
@@ -1,2 +1,2 @@
 // @target: es6
-var x = `\0\0x00\u0000 0 00 0000`;
+var x = `\0\x00\u0000 0 00 0000`;
diff --git a/.../es6/templates/templateStringControlCharacterEscapes03/input.ts/es2015.1.normal/output.js b/.../es6/templates/templateStringControlCharacterEscapes03/input.ts/es2015.1.normal/output.js
@@ -1 +1 @@
-var x = `\x1f\u001f 1F 1f`;
+var x = `\x1F\u001f 1F 1f`;
diff --git a/.../templates/templateStringControlCharacterEscapes03_ES6/input.ts/es2015.1.normal/output.js b/.../templates/templateStringControlCharacterEscapes03_ES6/input.ts/es2015.1.normal/output.js
@@ -1,2 +1,2 @@
 // @target: es6
-var x = `\x1f\u001f 1F 1f`;
+var x = `\x1F\u001f 1F 1f`;
diff --git a/crates/swc_ecma_minifier/src/compress/pure/strings.rs b/crates/swc_ecma_minifier/src/compress/pure/strings.rs
@@ -23,6 +23,7 @@ where
                 }) && (M::force_str_for_tpl()
                     || (!c.value.contains("\\n") && !c.value.contains("\\r")))
                     && !c.value.contains("\\0")
+                    && !c.value.contains("\\x")
                 {
                     *e = Expr::Lit(Lit::Str(Str {
                         value: c

diff --git a/crates/swc_ecma_parser/src/lexer/mod.rs b/crates/swc_ecma_parser/src/lexer/mod.rs
@@ -530,7 +530,7 @@ impl<'a, I: Input> Lexer<'a, I> {
 
             // read hexadecimal escape sequences
             'x' => {
-                raw.push_str("0x");
+                raw.push_str("x");
                 self.bump(); // 'x'
                 return self.read_hex_char(start, 2, raw).map(Some);
             }

diff --git a/.../tests/typescript/tsc/es6/templates/templateStringControlCharacterEscapes01/input.ts.json b/.../tests/typescript/tsc/es6/templates/templateStringControlCharacterEscapes01/input.ts.json
@@ -72,7 +72,7 @@
                     "end": 33,
                     "ctxt": 0
                   },
-                  "value": "\\0\\0x00\\u0000 0 00 0000",
+                  "value": "\\0\\x00\\u0000 0 00 0000",
                   "hasEscape": true,
                   "kind": {
                     "type": "normal",

diff --git a/...ts/typescript/tsc/es6/templates/templateStringControlCharacterEscapes01_ES6/input.ts.json b/...ts/typescript/tsc/es6/templates/templateStringControlCharacterEscapes01_ES6/input.ts.json
@@ -72,7 +72,7 @@
                     "end": 48,
                     "ctxt": 0
                   },
-                  "value": "\\0\\0x00\\u0000 0 00 0000",
+                  "value": "\\0\\x00\\u0000 0 00 0000",
                   "hasEscape": true,
                   "kind": {
                     "type": "normal",

diff --git a/.../tests/typescript/tsc/es6/templates/templateStringControlCharacterEscapes02/input.ts.json b/.../tests/typescript/tsc/es6/templates/templateStringControlCharacterEscapes02/input.ts.json
@@ -72,7 +72,7 @@
                     "end": 24,
                     "ctxt": 0
                   },
-                  "value": "\\0x19\\u0019 19",
+                  "value": "\\x19\\u0019 19",
                   "hasEscape": true,
                   "kind": {
                     "type": "normal",

diff --git a/...ts/typescript/tsc/es6/templates/templateStringControlCharacterEscapes02_ES6/input.ts.json b/...ts/typescript/tsc/es6/templates/templateStringControlCharacterEscapes02_ES6/input.ts.json
@@ -72,7 +72,7 @@
                     "end": 39,
                     "ctxt": 0
                   },
-                  "value": "\\0x19\\u0019 19",
+                  "value": "\\x19\\u0019 19",
                   "hasEscape": true,
                   "kind": {
                     "type": "normal",

diff --git a/.../tests/typescript/tsc/es6/templates/templateStringControlCharacterEscapes03/input.ts.json b/.../tests/typescript/tsc/es6/templates/templateStringControlCharacterEscapes03/input.ts.json
@@ -72,7 +72,7 @@
                     "end": 27,
                     "ctxt": 0
                   },
-                  "value": "\\0x1F\\u001f 1F 1f",
+                  "value": "\\x1F\\u001f 1F 1f",
                   "hasEscape": true,
                   "kind": {
                     "type": "normal",

diff --git a/...ts/typescript/tsc/es6/templates/templateStringControlCharacterEscapes03_ES6/input.ts.json b/...ts/typescript/tsc/es6/templates/templateStringControlCharacterEscapes03_ES6/input.ts.json
@@ -72,7 +72,7 @@
                     "end": 42,
                     "ctxt": 0
                   },
-                  "value": "\\0x1F\\u001f 1F 1f",
+                  "value": "\\x1F\\u001f 1F 1f",
                   "hasEscape": true,
                   "kind": {
                     "type": "normal",

diff --git a/.../tests/typescript/tsc/es6/templates/templateStringControlCharacterEscapes04/input.ts.json b/.../tests/typescript/tsc/es6/templates/templateStringControlCharacterEscapes04/input.ts.json
@@ -72,7 +72,7 @@
                     "end": 24,
                     "ctxt": 0
                   },
-                  "value": "\\0x20\\u0020 20",
+                  "value": "\\x20\\u0020 20",
                   "hasEscape": true,
                   "kind": {
                     "type": "normal",

diff --git a/...ts/typescript/tsc/es6/templates/templateStringControlCharacterEscapes04_ES6/input.ts.json b/...ts/typescript/tsc/es6/templates/templateStringControlCharacterEscapes04_ES6/input.ts.json
@@ -72,7 +72,7 @@
                     "end": 39,
                     "ctxt": 0
                   },
-                  "value": "\\0x20\\u0020 20",
+                  "value": "\\x20\\u0020 20",
                   "hasEscape": true,
                   "kind": {
                     "type": "normal",
Original file line number	Diff line number	Diff line change
		@@ -1 +1 @@
		var x = `\0\0x00\u0000 0 00 0000`;
		var x = `\0\x00\u0000 0 00 0000`;
Original file line number	Diff line number	Diff line change
		@@ -1 +1 @@
		var x = `\x1f\u001f 1F 1f`;
		var x = `\x1F\u001f 1F 1f`;
Benchmark suite	Current: `cf777c5`	Previous: `a8bda3b`	Ratio
`base_tr_fixer`	`27641` ns/iter (`± 323`)	`23642` ns/iter (`± 276`)	`1.17`
`base_tr_resolver_and_hygiene`	`153930` ns/iter (`± 28932`)	`130288` ns/iter (`± 24860`)	`1.18`
`codegen_es2015`	`61030` ns/iter (`± 372`)	`51393` ns/iter (`± 431`)	`1.19`
`codegen_es2016`	`61298` ns/iter (`± 408`)	`51605` ns/iter (`± 383`)	`1.19`
`codegen_es2017`	`61375` ns/iter (`± 450`)	`51568` ns/iter (`± 398`)	`1.19`
`codegen_es2018`	`61112` ns/iter (`± 440`)	`51556` ns/iter (`± 320`)	`1.19`
`codegen_es2019`	`60915` ns/iter (`± 376`)	`51626` ns/iter (`± 414`)	`1.18`
`codegen_es2020`	`61612` ns/iter (`± 10449`)	`51585` ns/iter (`± 365`)	`1.19`
`codegen_es3`	`61173` ns/iter (`± 330`)	`51499` ns/iter (`± 328`)	`1.19`
`codegen_es5`	`61345` ns/iter (`± 376`)	`51668` ns/iter (`± 377`)	`1.19`
`full_es2015`	`206584311` ns/iter (`± 8019457`)	`181224590` ns/iter (`± 8385947`)	`1.14`
`full_es2016`	`163413114` ns/iter (`± 5959340`)	`145550053` ns/iter (`± 8288243`)	`1.12`
`full_es2017`	`174419093` ns/iter (`± 11947824`)	`157853695` ns/iter (`± 9749004`)	`1.10`
`full_es2018`	`172437147` ns/iter (`± 8294328`)	`156687310` ns/iter (`± 6160645`)	`1.10`
`full_es2019`	`170150641` ns/iter (`± 8229759`)	`154457376` ns/iter (`± 8865541`)	`1.10`
`full_es2020`	`171622876` ns/iter (`± 10600575`)	`153694169` ns/iter (`± 5611120`)	`1.12`
`full_es3`	`246404415` ns/iter (`± 16128581`)	`210457792` ns/iter (`± 17128898`)	`1.17`
`full_es5`	`225233025` ns/iter (`± 11944865`)	`197997281` ns/iter (`± 12623345`)	`1.14`
`parser`	`758544` ns/iter (`± 26780`)	`654591` ns/iter (`± 17106`)	`1.16`
`ser_ast_node`	`179` ns/iter (`± 9`)	`147` ns/iter (`± 9`)	`1.22`
`ser_serde`	`192` ns/iter (`± 5`)	`159` ns/iter (`± 4`)	`1.21`
`emit_colors`	`21076733` ns/iter (`± 26854946`)	`18576217` ns/iter (`± 24517337`)	`1.13`
`emit_large`	`110393160` ns/iter (`± 168478079`)	`119598771` ns/iter (`± 182837125`)	`0.92`
`base_clone`	`2706277` ns/iter (`± 249446`)	`2331803` ns/iter (`± 399503`)	`1.16`
`fold_span`	`4540883` ns/iter (`± 399162`)	`3808626` ns/iter (`± 203931`)	`1.19`
`fold_span_panic`	`4915342` ns/iter (`± 421866`)	`4032999` ns/iter (`± 146948`)	`1.22`
`visit_mut_span`	`3276820` ns/iter (`± 286126`)	`2793427` ns/iter (`± 79052`)	`1.17`
`visit_mut_span_panic`	`3324890` ns/iter (`± 197198`)	`3028097` ns/iter (`± 246525`)	`1.10`
`ast_clone`	`19558` ns/iter (`± 253`)	`16744` ns/iter (`± 394`)	`1.17`
`ast_clone_to_stable`	`59694` ns/iter (`± 483`)	`51250` ns/iter (`± 344`)	`1.16`
`ast_clone_to_stable_then_to_unstable`	`108503` ns/iter (`± 645`)	`92981` ns/iter (`± 495`)	`1.17`
`json_deserialize`	`2319353` ns/iter (`± 8069`)	`1932153` ns/iter (`± 5292`)	`1.20`
`json_serialize`	`101037` ns/iter (`± 413`)	`85263` ns/iter (`± 292`)	`1.19`
`boxing_boxed`	`170` ns/iter (`± 0`)	`151` ns/iter (`± 2`)	`1.13`
`boxing_boxed_clone`	`81` ns/iter (`± 0`)	`68` ns/iter (`± 1`)	`1.19`
`boxing_unboxed`	`193` ns/iter (`± 1`)	`134` ns/iter (`± 0`)	`1.44`
`boxing_unboxed_clone`	`77` ns/iter (`± 0`)	`63` ns/iter (`± 0`)	`1.22`
`time_10`	`371` ns/iter (`± 4`)	`309` ns/iter (`± 3`)	`1.20`
`time_15`	`765` ns/iter (`± 6`)	`665` ns/iter (`± 2`)	`1.15`
`time_20`	`1520` ns/iter (`± 8`)	`1259` ns/iter (`± 2`)	`1.21`
`time_40`	`5155` ns/iter (`± 21`)	`7169` ns/iter (`± 20`)	`0.72`
`time_5`	`121` ns/iter (`± 2`)	`102` ns/iter (`± 1`)	`1.19`
`time_60`	`10200` ns/iter (`± 40`)	`16334` ns/iter (`± 68`)	`0.62`