generate shorter unicode escape sequence (mishoo#5914)

- fix input option mutation
alexlamsl · Aug 7, 2024 · 80ebacb · 80ebacb
1 parent 69f4e3a
commit 80ebacb
Show file tree

Hide file tree

Showing 7 changed files with 143 additions and 11 deletions.
diff --git a/bin/uglifyjs b/bin/uglifyjs
@@ -467,11 +467,14 @@ function run() {
     } else if (output) {
         var code;
         if (result.ast) {
-            var opts = {};
+            var output_options = {};
+            for (var name in UglifyJS.default_options("output")) {
+                if (name in options) output_options[name] = options[name];
+            }
             for (var name in options.output) {
-                if (!/^ast|code$/.test(name)) opts[name] = options.output[name];
+                if (!/^ast|code$/.test(name)) output_options[name] = options.output[name];
             }
-            code = UglifyJS.AST_Node.from_mozilla_ast(result.ast.to_mozilla_ast()).print_to_string(opts);
+            code = UglifyJS.AST_Node.from_mozilla_ast(result.ast.to_mozilla_ast()).print_to_string(output_options);
         } else {
             code = result.code;
         }

diff --git a/lib/minify.js b/lib/minify.js
@@ -49,8 +49,9 @@ function parse_source_map(content) {
 function set_shorthand(name, options, keys) {
     keys.forEach(function(key) {
         if (options[key]) {
-            if (typeof options[key] != "object") options[key] = {};
-            if (!(name in options[key])) options[key][name] = options[name];
+            var defs = {};
+            defs[name] = options[name];
+            options[key] = defaults(options[key], defs);
         }
     });
 }
@@ -105,7 +106,7 @@ function minify(files, options) {
         if (options.keep_fargs) set_shorthand("keep_fargs", options, [ "compress", "mangle", "rename" ]);
         if (options.keep_fnames) set_shorthand("keep_fnames", options, [ "compress", "mangle", "rename" ]);
         if (options.module === undefined && !options.ie) options.module = true;
-        if (options.module) set_shorthand("module", options, [ "compress", "parse" ]);
+        if (options.module) set_shorthand("module", options, [ "compress", "output", "parse" ]);
         if (options.toplevel !== undefined) set_shorthand("toplevel", options, [ "compress", "mangle", "rename" ]);
         if (options.v8) set_shorthand("v8", options, [ "mangle", "output", "rename" ]);
         if (options.webkit) set_shorthand("webkit", options, [ "compress", "mangle", "output", "rename" ]);

diff --git a/lib/output.js b/lib/output.js
@@ -63,6 +63,7 @@ function OutputStream(options) {
         inline_script    : true,
         keep_quoted_props: false,
         max_line_len     : false,
+        module           : false,
         preamble         : null,
         preserve_line    : false,
         quote_keys       : false,
@@ -140,12 +141,22 @@ function OutputStream(options) {
 
     reset();
     var to_utf8 = options.ascii_only ? function(str, identifier) {
-        if (identifier) str = str.replace(/[\ud800-\udbff][\udc00-\udfff]/g, function(ch) {
+        if (identifier || options.module) str = str.replace(/[\ud800-\udbff][\udc00-\udfff]/g, function(ch) {
             return "\\u{" + (ch.charCodeAt(0) - 0xd7c0 << 10 | ch.charCodeAt(1) - 0xdc00).toString(16) + "}";
         });
-        return str.replace(/[\u0000-\u001f\u007f-\uffff]/g, function(ch) {
-            var code = ch.charCodeAt(0).toString(16);
+        return str.replace(/[\u0000-\u001f\u007f-\uffff]/g, function(s, i) {
+            var code = s.charCodeAt(0).toString(16);
             if (code.length <= 2 && !identifier) {
+                switch (s) {
+                  case "\n": return "\\n";
+                  case "\r": return "\\r";
+                  case "\t": return "\\t";
+                  case "\b": return "\\b";
+                  case "\f": return "\\f";
+                  case "\x0B": return options.ie ? "\\x0B" : "\\v";
+                  case "\0":
+                    return /[0-9]/.test(str.charAt(i+1)) ? "\\x00" : "\\0";
+                }
                 while (code.length < 2) code = "0" + code;
                 return "\\x" + code;
             } else {

diff --git a/test/compress/templates.js b/test/compress/templates.js
@@ -432,6 +432,7 @@ pure_funcs: {
 ascii_only: {
     beautify = {
         ascii_only: true,
+        module: false,
     }
     options = {
         templates: false,
@@ -445,9 +446,27 @@ ascii_only: {
     node_version: ">=6"
 }
 
+ascii_only_ecma: {
+    beautify = {
+        ascii_only: true,
+        module: true,
+    }
+    options = {
+        templates: false,
+    }
+    input: {
+        console.log(`\ud801\udc37\ud801𐐷${42}\u{10437}`);
+    }
+    expect_exact: "console.log(`\\ud801\\udc37\\ud801\\u{10437}${42}\\u{10437}`);"
+    expect_stdout: "𐐷\ud801𐐷42𐐷"
+    // non-BMP support is platform-dependent on Node.js v4
+    node_version: ">=6"
+}
+
 ascii_only_templates: {
     beautify = {
         ascii_only: true,
+        module: false,
     }
     options = {
         templates: true,
@@ -461,9 +480,44 @@ ascii_only_templates: {
     node_version: ">=6"
 }
 
+ascii_only_templates_ecma: {
+    beautify = {
+        ascii_only: true,
+        module: true,
+    }
+    options = {
+        templates: true,
+    }
+    input: {
+        console.log(`\ud801\udc37\ud801𐐷${42}\u{10437}`);
+    }
+    expect_exact: "console.log(`\\u{10437}\\ud801\\u{10437}${42}\\u{10437}`);"
+    expect_stdout: "𐐷\ud801𐐷42𐐷"
+    // non-BMP support is platform-dependent on Node.js v4
+    node_version: ">=6"
+}
+
 unicode: {
     beautify = {
         ascii_only: false,
+        module: false,
+    }
+    options = {
+        templates: false,
+    }
+    input: {
+        console.log(`\ud801\udc37\ud801𐐷${42}\u{10437}`);
+    }
+    expect_exact: "console.log(`\\ud801\\udc37\\ud801𐐷${42}\\u{10437}`);"
+    expect_stdout: "𐐷\ud801𐐷42𐐷"
+    // non-BMP support is platform-dependent on Node.js v4
+    node_version: ">=6"
+}
+
+unicode_ecma: {
+    beautify = {
+        ascii_only: false,
+        module: true,
     }
     options = {
         templates: false,
@@ -480,6 +534,24 @@ unicode: {
 unicode_templates: {
     beautify = {
         ascii_only: false,
+        module: false,
+    }
+    options = {
+        templates: true,
+    }
+    input: {
+        console.log(`\ud801\udc37\ud801𐐷${42}\u{10437}`);
+    }
+    expect_exact: "console.log(`𐐷\\ud801𐐷${42}𐐷`);"
+    expect_stdout: "𐐷\ud801𐐷42𐐷"
+    // non-BMP support is platform-dependent on Node.js v4
+    node_version: ">=6"
+}
+
+unicode_templates_ecma: {
+    beautify = {
+        ascii_only: false,
+        module: true,
     }
     options = {
         templates: true,

diff --git a/test/compress/unicode.js b/test/compress/unicode.js
@@ -191,6 +191,24 @@ issue_2569: {
 surrogate_pair: {
     beautify = {
         ascii_only: false,
+        module: false,
+    }
+    input: {
+        var \u{2f800} = {
+            \u{2f801}: "\u{100000}",
+        };
+        \u{2f800}.\u{2f802} = "\u{100001}";
+        console.log(typeof \u{2f800}, \u{2f800}.\u{2f801}, \u{2f800}["\u{2f802}"]);
+    }
+    expect_exact: 'var \ud87e\udc00={"\ud87e\udc01":"\udbc0\udc00"};\ud87e\udc00.\ud87e\udc02="\udbc0\udc01";console.log(typeof \ud87e\udc00,\ud87e\udc00.\ud87e\udc01,\ud87e\udc00["\ud87e\udc02"]);'
+    expect_stdout: "object \udbc0\udc00 \udbc0\udc01"
+    // non-BMP support is platform-dependent on Node.js v4
+    node_version: ">=6"
+}
+surrogate_pair_ecma: {
+    beautify = {
+        ascii_only: false,
+        module: true,
     }
     input: {
         var \u{2f800} = {
@@ -208,6 +226,7 @@ surrogate_pair: {
 surrogate_pair_ascii: {
     beautify = {
         ascii_only: true,
+        module: false,
     }
     input: {
         var \u{2f800} = {
@@ -221,3 +240,21 @@ surrogate_pair_ascii: {
     // non-BMP support is platform-dependent on Node.js v4
     node_version: ">=6"
 }
+
+surrogate_pair_ascii_ecma: {
+    beautify = {
+        ascii_only: true,
+        module: true,
+    }
+    input: {
+        var \u{2f800} = {
+            \u{2f801}: "\u{100000}",
+        };
+        \u{2f800}.\u{2f802} = "\u{100001}";
+        console.log(typeof \u{2f800}, \u{2f800}.\u{2f801}, \u{2f800}["\u{2f802}"]);
+    }
+    expect_exact: 'var \\u{2f800}={"\\u{2f801}":"\\u{100000}"};\\u{2f800}.\\u{2f802}="\\u{100001}";console.log(typeof \\u{2f800},\\u{2f800}.\\u{2f801},\\u{2f800}["\\u{2f802}"]);'
+    expect_stdout: "object \udbc0\udc00 \udbc0\udc01"
+    // non-BMP support is platform-dependent on Node.js v4
+    node_version: ">=6"
+}
diff --git a/test/mocha/minify.js b/test/mocha/minify.js
@@ -19,7 +19,10 @@ describe("minify", function() {
         var options = {
             compress: true,
             mangle: false,
-            output: {},
+            output: {
+                v8: false,
+            },
+            webkit: true,
         };
         var value = JSON.stringify(options);
         var result = UglifyJS.minify("print(6 * 7);", options);

diff --git a/tools/node.js b/tools/node.js
@@ -97,7 +97,12 @@ function infer_options(options) {
     return result.error && result.error.defs;
 }
 
-exports.default_options = function() {
+exports.default_options = function(component) {
+    if (component) {
+        var options = { module: false };
+        options[component] = { 0: 0 };
+        return infer_options(options);
+    }
     var defs = infer_options({ 0: 0 });
     Object.keys(defs).forEach(function(component) {
         var options = { module: false };