js_test.go

     1  // Copyright 2011 The Go Authors. All rights reserved.
     2  // Use of this source code is governed by a BSD-style
     3  // license that can be found in the LICENSE file.
     4  
     5  package template
     6  
     7  import (
     8  	"math"
     9  	"strings"
    10  	"testing"
    11  )
    12  
    13  func TestNextJsCtx(t *testing.T) {
    14  	tests := []struct {
    15  		jsCtx jsCtx
    16  		s     string
    17  	}{
    18  		// Statement terminators precede regexps.
    19  		{jsCtxRegexp, ";"},
    20  		// This is not airtight.
    21  		//     ({ valueOf: function () { return 1 } } / 2)
    22  		// is valid JavaScript but in practice, devs do not do this.
    23  		// A block followed by a statement starting with a RegExp is
    24  		// much more common:
    25  		//     while (x) {...} /foo/.test(x) || panic()
    26  		{jsCtxRegexp, "}"},
    27  		// But member, call, grouping, and array expression terminators
    28  		// precede div ops.
    29  		{jsCtxDivOp, ")"},
    30  		{jsCtxDivOp, "]"},
    31  		// At the start of a primary expression, array, or expression
    32  		// statement, expect a regexp.
    33  		{jsCtxRegexp, "("},
    34  		{jsCtxRegexp, "["},
    35  		{jsCtxRegexp, "{"},
    36  		// Assignment operators precede regexps as do all exclusively
    37  		// prefix and binary operators.
    38  		{jsCtxRegexp, "="},
    39  		{jsCtxRegexp, "+="},
    40  		{jsCtxRegexp, "*="},
    41  		{jsCtxRegexp, "*"},
    42  		{jsCtxRegexp, "!"},
    43  		// Whether the + or - is infix or prefix, it cannot precede a
    44  		// div op.
    45  		{jsCtxRegexp, "+"},
    46  		{jsCtxRegexp, "-"},
    47  		// An incr/decr op precedes a div operator.
    48  		// This is not airtight. In (g = ++/h/i) a regexp follows a
    49  		// pre-increment operator, but in practice devs do not try to
    50  		// increment or decrement regular expressions.
    51  		// (g++/h/i) where ++ is a postfix operator on g is much more
    52  		// common.
    53  		{jsCtxDivOp, "--"},
    54  		{jsCtxDivOp, "++"},
    55  		{jsCtxDivOp, "x--"},
    56  		// When we have many dashes or pluses, then they are grouped
    57  		// left to right.
    58  		{jsCtxRegexp, "x---"}, // A postfix -- then a -.
    59  		// return followed by a slash returns the regexp literal or the
    60  		// slash starts a regexp literal in an expression statement that
    61  		// is dead code.
    62  		{jsCtxRegexp, "return"},
    63  		{jsCtxRegexp, "return "},
    64  		{jsCtxRegexp, "return\t"},
    65  		{jsCtxRegexp, "return\n"},
    66  		{jsCtxRegexp, "return\u2028"},
    67  		// Identifiers can be divided and cannot validly be preceded by
    68  		// a regular expressions. Semicolon insertion cannot happen
    69  		// between an identifier and a regular expression on a new line
    70  		// because the one token lookahead for semicolon insertion has
    71  		// to conclude that it could be a div binary op and treat it as
    72  		// such.
    73  		{jsCtxDivOp, "x"},
    74  		{jsCtxDivOp, "x "},
    75  		{jsCtxDivOp, "x\t"},
    76  		{jsCtxDivOp, "x\n"},
    77  		{jsCtxDivOp, "x\u2028"},
    78  		{jsCtxDivOp, "preturn"},
    79  		// Numbers precede div ops.
    80  		{jsCtxDivOp, "0"},
    81  		// Dots that are part of a number are div preceders.
    82  		{jsCtxDivOp, "0."},
    83  		// Some JS interpreters treat NBSP as a normal space, so
    84  		// we must too in order to properly escape things.
    85  		{jsCtxRegexp, "=\u00A0"},
    86  	}
    87  
    88  	for _, test := range tests {
    89  		if ctx := nextJSCtx([]byte(test.s), jsCtxRegexp); ctx != test.jsCtx {
    90  			t.Errorf("%q: want %s got %s", test.s, test.jsCtx, ctx)
    91  		}
    92  		if ctx := nextJSCtx([]byte(test.s), jsCtxDivOp); ctx != test.jsCtx {
    93  			t.Errorf("%q: want %s got %s", test.s, test.jsCtx, ctx)
    94  		}
    95  	}
    96  
    97  	if nextJSCtx([]byte("   "), jsCtxRegexp) != jsCtxRegexp {
    98  		t.Error("Blank tokens")
    99  	}
   100  
   101  	if nextJSCtx([]byte("   "), jsCtxDivOp) != jsCtxDivOp {
   102  		t.Error("Blank tokens")
   103  	}
   104  }
   105  
   106  func TestJSValEscaper(t *testing.T) {
   107  	tests := []struct {
   108  		x  any
   109  		js string
   110  	}{
   111  		{int(42), " 42 "},
   112  		{uint(42), " 42 "},
   113  		{int16(42), " 42 "},
   114  		{uint16(42), " 42 "},
   115  		{int32(-42), " -42 "},
   116  		{uint32(42), " 42 "},
   117  		{int16(-42), " -42 "},
   118  		{uint16(42), " 42 "},
   119  		{int64(-42), " -42 "},
   120  		{uint64(42), " 42 "},
   121  		{uint64(1) << 53, " 9007199254740992 "},
   122  		// ulp(1 << 53) > 1 so this loses precision in JS
   123  		// but it is still a representable integer literal.
   124  		{uint64(1)<<53 + 1, " 9007199254740993 "},
   125  		{float32(1.0), " 1 "},
   126  		{float32(-1.0), " -1 "},
   127  		{float32(0.5), " 0.5 "},
   128  		{float32(-0.5), " -0.5 "},
   129  		{float32(1.0) / float32(256), " 0.00390625 "},
   130  		{float32(0), " 0 "},
   131  		{math.Copysign(0, -1), " -0 "},
   132  		{float64(1.0), " 1 "},
   133  		{float64(-1.0), " -1 "},
   134  		{float64(0.5), " 0.5 "},
   135  		{float64(-0.5), " -0.5 "},
   136  		{float64(0), " 0 "},
   137  		{math.Copysign(0, -1), " -0 "},
   138  		{"", `""`},
   139  		{"foo", `"foo"`},
   140  		// Newlines.
   141  		{"\r\n\u2028\u2029", `"\r\n\u2028\u2029"`},
   142  		// "\v" == "v" on IE 6 so use "\u000b" instead.
   143  		{"\t\x0b", `"\t\u000b"`},
   144  		{struct{ X, Y int }{1, 2}, `{"X":1,"Y":2}`},
   145  		{[]any{}, "[]"},
   146  		{[]any{42, "foo", nil}, `[42,"foo",null]`},
   147  		{[]string{"<!--", "</script>", "-->"}, `["\u003c!--","\u003c/script\u003e","--\u003e"]`},
   148  		{"<!--", `"\u003c!--"`},
   149  		{"-->", `"--\u003e"`},
   150  		{"<![CDATA[", `"\u003c![CDATA["`},
   151  		{"]]>", `"]]\u003e"`},
   152  		{"</script", `"\u003c/script"`},
   153  		{"\U0001D11E", "\"\U0001D11E\""}, // or "\uD834\uDD1E"
   154  		{nil, " null "},
   155  	}
   156  
   157  	for _, test := range tests {
   158  		if js := jsValEscaper(test.x); js != test.js {
   159  			t.Errorf("%+v: want\n\t%q\ngot\n\t%q", test.x, test.js, js)
   160  		}
   161  		// Make sure that escaping corner cases are not broken
   162  		// by nesting.
   163  		a := []any{test.x}
   164  		want := "[" + strings.TrimSpace(test.js) + "]"
   165  		if js := jsValEscaper(a); js != want {
   166  			t.Errorf("%+v: want\n\t%q\ngot\n\t%q", a, want, js)
   167  		}
   168  	}
   169  }
   170  
   171  func TestJSStrEscaper(t *testing.T) {
   172  	tests := []struct {
   173  		x   any
   174  		esc string
   175  	}{
   176  		{"", ``},
   177  		{"foo", `foo`},
   178  		{"\u0000", `\u0000`},
   179  		{"\t", `\t`},
   180  		{"\n", `\n`},
   181  		{"\r", `\r`},
   182  		{"\u2028", `\u2028`},
   183  		{"\u2029", `\u2029`},
   184  		{"\\", `\\`},
   185  		{"\\n", `\\n`},
   186  		{"foo\r\nbar", `foo\r\nbar`},
   187  		// Preserve attribute boundaries.
   188  		{`"`, `\u0022`},
   189  		{`'`, `\u0027`},
   190  		// Allow embedding in HTML without further escaping.
   191  		{`&amp;`, `\u0026amp;`},
   192  		// Prevent breaking out of text node and element boundaries.
   193  		{"</script>", `\u003c\/script\u003e`},
   194  		{"<![CDATA[", `\u003c![CDATA[`},
   195  		{"]]>", `]]\u003e`},
   196  		// https://dev.w3.org/html5/markup/aria/syntax.html#escaping-text-span
   197  		//   "The text in style, script, title, and textarea elements
   198  		//   must not have an escaping text span start that is not
   199  		//   followed by an escaping text span end."
   200  		// Furthermore, spoofing an escaping text span end could lead
   201  		// to different interpretation of a </script> sequence otherwise
   202  		// masked by the escaping text span, and spoofing a start could
   203  		// allow regular text content to be interpreted as script
   204  		// allowing script execution via a combination of a JS string
   205  		// injection followed by an HTML text injection.
   206  		{"<!--", `\u003c!--`},
   207  		{"-->", `--\u003e`},
   208  		// From https://code.google.com/p/doctype/wiki/ArticleUtf7
   209  		{"+ADw-script+AD4-alert(1)+ADw-/script+AD4-",
   210  			`\u002bADw-script\u002bAD4-alert(1)\u002bADw-\/script\u002bAD4-`,
   211  		},
   212  		// Invalid UTF-8 sequence
   213  		{"foo\xA0bar", "foo\xA0bar"},
   214  		// Invalid unicode scalar value.
   215  		{"foo\xed\xa0\x80bar", "foo\xed\xa0\x80bar"},
   216  	}
   217  
   218  	for _, test := range tests {
   219  		esc := jsStrEscaper(test.x)
   220  		if esc != test.esc {
   221  			t.Errorf("%q: want %q got %q", test.x, test.esc, esc)
   222  		}
   223  	}
   224  }
   225  
   226  func TestJSRegexpEscaper(t *testing.T) {
   227  	tests := []struct {
   228  		x   any
   229  		esc string
   230  	}{
   231  		{"", `(?:)`},
   232  		{"foo", `foo`},
   233  		{"\u0000", `\u0000`},
   234  		{"\t", `\t`},
   235  		{"\n", `\n`},
   236  		{"\r", `\r`},
   237  		{"\u2028", `\u2028`},
   238  		{"\u2029", `\u2029`},
   239  		{"\\", `\\`},
   240  		{"\\n", `\\n`},
   241  		{"foo\r\nbar", `foo\r\nbar`},
   242  		// Preserve attribute boundaries.
   243  		{`"`, `\u0022`},
   244  		{`'`, `\u0027`},
   245  		// Allow embedding in HTML without further escaping.
   246  		{`&amp;`, `\u0026amp;`},
   247  		// Prevent breaking out of text node and element boundaries.
   248  		{"</script>", `\u003c\/script\u003e`},
   249  		{"<![CDATA[", `\u003c!\[CDATA\[`},
   250  		{"]]>", `\]\]\u003e`},
   251  		// Escaping text spans.
   252  		{"<!--", `\u003c!\-\-`},
   253  		{"-->", `\-\-\u003e`},
   254  		{"*", `\*`},
   255  		{"+", `\u002b`},
   256  		{"?", `\?`},
   257  		{"[](){}", `\[\]\(\)\{\}`},
   258  		{"$foo|x.y", `\$foo\|x\.y`},
   259  		{"x^y", `x\^y`},
   260  	}
   261  
   262  	for _, test := range tests {
   263  		esc := jsRegexpEscaper(test.x)
   264  		if esc != test.esc {
   265  			t.Errorf("%q: want %q got %q", test.x, test.esc, esc)
   266  		}
   267  	}
   268  }
   269  
   270  func TestEscapersOnLower7AndSelectHighCodepoints(t *testing.T) {
   271  	input := ("\x00\x01\x02\x03\x04\x05\x06\x07\x08\t\n\x0b\x0c\r\x0e\x0f" +
   272  		"\x10\x11\x12\x13\x14\x15\x16\x17\x18\x19\x1a\x1b\x1c\x1d\x1e\x1f" +
   273  		` !"#$%&'()*+,-./` +
   274  		`0123456789:;<=>?` +
   275  		`@ABCDEFGHIJKLMNO` +
   276  		`PQRSTUVWXYZ[\]^_` +
   277  		"`abcdefghijklmno" +
   278  		"pqrstuvwxyz{|}~\x7f" +
   279  		"\u00A0\u0100\u2028\u2029\ufeff\U0001D11E")
   280  
   281  	tests := []struct {
   282  		name    string
   283  		escaper func(...any) string
   284  		escaped string
   285  	}{
   286  		{
   287  			"jsStrEscaper",
   288  			jsStrEscaper,
   289  			`\u0000\u0001\u0002\u0003\u0004\u0005\u0006\u0007` +
   290  				`\u0008\t\n\u000b\f\r\u000e\u000f` +
   291  				`\u0010\u0011\u0012\u0013\u0014\u0015\u0016\u0017` +
   292  				`\u0018\u0019\u001a\u001b\u001c\u001d\u001e\u001f` +
   293  				` !\u0022#$%\u0026\u0027()*\u002b,-.\/` +
   294  				`0123456789:;\u003c=\u003e?` +
   295  				`@ABCDEFGHIJKLMNO` +
   296  				`PQRSTUVWXYZ[\\]^_` +
   297  				"\\u0060abcdefghijklmno" +
   298  				"pqrstuvwxyz{|}~\u007f" +
   299  				"\u00A0\u0100\\u2028\\u2029\ufeff\U0001D11E",
   300  		},
   301  		{
   302  			"jsRegexpEscaper",
   303  			jsRegexpEscaper,
   304  			`\u0000\u0001\u0002\u0003\u0004\u0005\u0006\u0007` +
   305  				`\u0008\t\n\u000b\f\r\u000e\u000f` +
   306  				`\u0010\u0011\u0012\u0013\u0014\u0015\u0016\u0017` +
   307  				`\u0018\u0019\u001a\u001b\u001c\u001d\u001e\u001f` +
   308  				` !\u0022#\$%\u0026\u0027\(\)\*\u002b,\-\.\/` +
   309  				`0123456789:;\u003c=\u003e\?` +
   310  				`@ABCDEFGHIJKLMNO` +
   311  				`PQRSTUVWXYZ\[\\\]\^_` +
   312  				"`abcdefghijklmno" +
   313  				`pqrstuvwxyz\{\|\}~` + "\u007f" +
   314  				"\u00A0\u0100\\u2028\\u2029\ufeff\U0001D11E",
   315  		},
   316  	}
   317  
   318  	for _, test := range tests {
   319  		if s := test.escaper(input); s != test.escaped {
   320  			t.Errorf("%s once: want\n\t%q\ngot\n\t%q", test.name, test.escaped, s)
   321  			continue
   322  		}
   323  
   324  		// Escape it rune by rune to make sure that any
   325  		// fast-path checking does not break escaping.
   326  		var buf strings.Builder
   327  		for _, c := range input {
   328  			buf.WriteString(test.escaper(string(c)))
   329  		}
   330  
   331  		if s := buf.String(); s != test.escaped {
   332  			t.Errorf("%s rune-wise: want\n\t%q\ngot\n\t%q", test.name, test.escaped, s)
   333  			continue
   334  		}
   335  	}
   336  }
   337  
   338  func TestIsJsMimeType(t *testing.T) {
   339  	tests := []struct {
   340  		in  string
   341  		out bool
   342  	}{
   343  		{"application/javascript;version=1.8", true},
   344  		{"application/javascript;version=1.8;foo=bar", true},
   345  		{"application/javascript/version=1.8", false},
   346  		{"text/javascript", true},
   347  		{"application/json", true},
   348  		{"application/ld+json", true},
   349  		{"module", true},
   350  	}
   351  
   352  	for _, test := range tests {
   353  		if isJSType(test.in) != test.out {
   354  			t.Errorf("isJSType(%q) = %v, want %v", test.in, !test.out, test.out)
   355  		}
   356  	}
   357  }
   358  
   359  func BenchmarkJSValEscaperWithNum(b *testing.B) {
   360  	for i := 0; i < b.N; i++ {
   361  		jsValEscaper(3.141592654)
   362  	}
   363  }
   364  
   365  func BenchmarkJSValEscaperWithStr(b *testing.B) {
   366  	for i := 0; i < b.N; i++ {
   367  		jsValEscaper("The <i>quick</i>,\r\n<span style='color:brown'>brown</span> fox jumps\u2028over the <canine class=\"lazy\">dog</canine>")
   368  	}
   369  }
   370  
   371  func BenchmarkJSValEscaperWithStrNoSpecials(b *testing.B) {
   372  	for i := 0; i < b.N; i++ {
   373  		jsValEscaper("The quick, brown fox jumps over the lazy dog")
   374  	}
   375  }
   376  
   377  func BenchmarkJSValEscaperWithObj(b *testing.B) {
   378  	o := struct {
   379  		S string
   380  		N int
   381  	}{
   382  		"The <i>quick</i>,\r\n<span style='color:brown'>brown</span> fox jumps\u2028over the <canine class=\"lazy\">dog</canine>\u2028",
   383  		42,
   384  	}
   385  	for i := 0; i < b.N; i++ {
   386  		jsValEscaper(o)
   387  	}
   388  }
   389  
   390  func BenchmarkJSValEscaperWithObjNoSpecials(b *testing.B) {
   391  	o := struct {
   392  		S string
   393  		N int
   394  	}{
   395  		"The quick, brown fox jumps over the lazy dog",
   396  		42,
   397  	}
   398  	for i := 0; i < b.N; i++ {
   399  		jsValEscaper(o)
   400  	}
   401  }
   402  
   403  func BenchmarkJSStrEscaperNoSpecials(b *testing.B) {
   404  	for i := 0; i < b.N; i++ {
   405  		jsStrEscaper("The quick, brown fox jumps over the lazy dog.")
   406  	}
   407  }
   408  
   409  func BenchmarkJSStrEscaper(b *testing.B) {
   410  	for i := 0; i < b.N; i++ {
   411  		jsStrEscaper("The <i>quick</i>,\r\n<span style='color:brown'>brown</span> fox jumps\u2028over the <canine class=\"lazy\">dog</canine>")
   412  	}
   413  }
   414  
   415  func BenchmarkJSRegexpEscaperNoSpecials(b *testing.B) {
   416  	for i := 0; i < b.N; i++ {
   417  		jsRegexpEscaper("The quick, brown fox jumps over the lazy dog")
   418  	}
   419  }
   420  
   421  func BenchmarkJSRegexpEscaper(b *testing.B) {
   422  	for i := 0; i < b.N; i++ {
   423  		jsRegexpEscaper("The <i>quick</i>,\r\n<span style='color:brown'>brown</span> fox jumps\u2028over the <canine class=\"lazy\">dog</canine>")
   424  	}
   425  }
   426
View as plain text