|
| 1 | +package compiler |
| 2 | + |
| 3 | +import ( |
| 4 | + "strings" |
| 5 | + "testing" |
| 6 | + |
| 7 | + "github.com/gopherjs/gopherjs/internal/srctesting" |
| 8 | +) |
| 9 | + |
| 10 | +func Test_encodeString(t *testing.T) { |
| 11 | + tests := []struct { |
| 12 | + name string |
| 13 | + in string |
| 14 | + wantUtf8 string |
| 15 | + wantUtf16 string |
| 16 | + }{ |
| 17 | + { |
| 18 | + name: `empty`, |
| 19 | + in: ``, |
| 20 | + wantUtf8: `""`, |
| 21 | + wantUtf16: `""`, |
| 22 | + }, |
| 23 | + { |
| 24 | + name: `ascii`, |
| 25 | + in: `hello`, |
| 26 | + wantUtf8: `"hello"`, |
| 27 | + wantUtf16: `"hello"`, |
| 28 | + }, |
| 29 | + { |
| 30 | + name: `escape backslash`, |
| 31 | + in: `a\b`, |
| 32 | + wantUtf8: `"a\\b"`, |
| 33 | + wantUtf16: `"a\\b"`, |
| 34 | + }, |
| 35 | + { |
| 36 | + name: `escape double quote`, |
| 37 | + in: `a"b`, |
| 38 | + wantUtf8: `"a\"b"`, |
| 39 | + wantUtf16: `"a\"b"`, |
| 40 | + }, |
| 41 | + { |
| 42 | + name: `escape control runes`, |
| 43 | + in: "\b\f\n\r\t\v", |
| 44 | + wantUtf8: `"\b\f\n\r\t\v"`, |
| 45 | + wantUtf16: `"\b\f\n\r\t\v"`, |
| 46 | + }, |
| 47 | + { |
| 48 | + name: `escape low non-printable`, |
| 49 | + in: "\x00\x01\x1f", |
| 50 | + wantUtf8: `"\x00\x01\x1F"`, |
| 51 | + wantUtf16: `"\x00\x01\x1F"`, |
| 52 | + }, |
| 53 | + { |
| 54 | + name: `escape DEL (with é for non-ASCII path)`, |
| 55 | + in: "\u00e9\x7f", |
| 56 | + wantUtf8: `"\xC3\xA9\x7F"`, |
| 57 | + wantUtf16: `"\u00E9\u007F"`, |
| 58 | + }, |
| 59 | + { |
| 60 | + name: `printable boundary`, |
| 61 | + in: " ~", |
| 62 | + wantUtf8: `" ~"`, |
| 63 | + wantUtf16: `" ~"`, |
| 64 | + }, |
| 65 | + { |
| 66 | + name: `UTF-8 BMP bytes (é)`, |
| 67 | + in: "h\xc3\xa9llo", |
| 68 | + wantUtf8: `"h\xC3\xA9llo"`, |
| 69 | + wantUtf16: `"h\u00E9llo"`, |
| 70 | + }, |
| 71 | + { |
| 72 | + name: `UTF-8 CJK bytes`, |
| 73 | + in: "日本語", |
| 74 | + wantUtf8: `"\xE6\x97\xA5\xE6\x9C\xAC\xE8\xAA\x9E"`, |
| 75 | + wantUtf16: `"\u65E5\u672C\u8A9E"`, |
| 76 | + }, |
| 77 | + { |
| 78 | + name: `UTF-8 supplementary plane bytes and UTF-16 surrogate pair (😀)`, |
| 79 | + in: "\U0001F600", |
| 80 | + wantUtf8: `"\xF0\x9F\x98\x80"`, |
| 81 | + wantUtf16: `"\uD83D\uDE00"`, |
| 82 | + }, |
| 83 | + { |
| 84 | + name: `control char in non-ASCII for UTF-16 path stays escaped as \n`, |
| 85 | + in: "\u00e9\n", |
| 86 | + wantUtf8: `"\xC3\xA9\n"`, |
| 87 | + wantUtf16: `"\u00E9\n"`, |
| 88 | + }, |
| 89 | + { |
| 90 | + name: `UTF-16 high BMP edge (U+FFFF)`, |
| 91 | + in: "\uffff", |
| 92 | + wantUtf8: `"\xEF\xBF\xBF"`, |
| 93 | + wantUtf16: `"\uFFFF"`, |
| 94 | + }, |
| 95 | + { |
| 96 | + name: `UTF-16 low supplementary edge (U+10000)`, |
| 97 | + in: "\U00010000", |
| 98 | + wantUtf8: `"\xF0\x90\x80\x80"`, |
| 99 | + wantUtf16: `"\uD800\uDC00"`, |
| 100 | + }, |
| 101 | + { |
| 102 | + name: `UTF-16 high supplementary edge (U+10FFFF)`, |
| 103 | + in: "\U0010FFFF", |
| 104 | + wantUtf8: `"\xF4\x8F\xBF\xBF"`, |
| 105 | + wantUtf16: `"\uDBFF\uDFFF"`, |
| 106 | + }, |
| 107 | + } |
| 108 | + for _, tt := range tests { |
| 109 | + t.Run(tt.name, func(t *testing.T) { |
| 110 | + if got := encodeString(tt.in); got != tt.wantUtf8 { |
| 111 | + t.Errorf("UTF-8: encodeString(%q):\n\tgot: %q\n\twant: %q", tt.in, got, tt.wantUtf8) |
| 112 | + } |
| 113 | + if got := preexternalizeString(tt.in); got != tt.wantUtf16 { |
| 114 | + t.Errorf("UTF-16: preexternalizeString(%q):\n\tgot: %q\n\twant: %q", tt.in, got, tt.wantUtf16) |
| 115 | + } |
| 116 | + }) |
| 117 | + } |
| 118 | +} |
| 119 | + |
| 120 | +func TestExternalize_LiteralVsRuntime(t *testing.T) { |
| 121 | + const src = ` |
| 122 | + package main |
| 123 | +
|
| 124 | + import "github.com/gopherjs/gopherjs/js" |
| 125 | +
|
| 126 | + func setStuff(obj *js.Object, dyn string) { |
| 127 | + obj.Set("ascii", "hello") |
| 128 | + obj.Set("unicode", "h` + "\u00e9" + `llo") |
| 129 | + obj.Set("emoji", "` + "\U0001F600" + `") |
| 130 | + obj.Set("dynamic", dyn) |
| 131 | + obj.Set("concat", "pre-" + dyn) |
| 132 | + } |
| 133 | +
|
| 134 | + func main() { |
| 135 | + setStuff(nil, "") |
| 136 | + } |
| 137 | + ` |
| 138 | + srcFiles := []srctesting.Source{{Name: `main.go`, Contents: []byte(src)}} |
| 139 | + out := compile(t, srcFiles, false) |
| 140 | + |
| 141 | + mustContain := []struct { |
| 142 | + name string |
| 143 | + good string |
| 144 | + bad string |
| 145 | + }{ |
| 146 | + { |
| 147 | + name: "ASCII literal pre-externalized as plain JS string", |
| 148 | + good: `"hello"`, |
| 149 | + }, |
| 150 | + { |
| 151 | + name: "BMP literal pre-externalized as UTF-16 escapes", |
| 152 | + good: `"h\u00E9llo"`, |
| 153 | + }, |
| 154 | + { |
| 155 | + name: "Supplementary literal pre-externalized as surrogate pair", |
| 156 | + good: `"\uD83D\uDE00"`, |
| 157 | + }, |
| 158 | + { |
| 159 | + name: "Plain string variable falls back to runtime $externalize", |
| 160 | + good: `$externalize(dyn, $String)`, |
| 161 | + }, |
| 162 | + { |
| 163 | + name: "Computed (non-constant) string falls back to runtime $externalize", |
| 164 | + good: `$externalize("pre-" + dyn, $String)`, |
| 165 | + }, |
| 166 | + { |
| 167 | + name: "ASCII literal must not be wrapped in $externalize", |
| 168 | + bad: `$externalize("hello"`, |
| 169 | + }, |
| 170 | + { |
| 171 | + name: "BMP literal must not be wrapped in $externalize", |
| 172 | + bad: `$externalize("h\u00E9llo"`, |
| 173 | + }, |
| 174 | + { |
| 175 | + name: "Supplementary literal must not be wrapped in $externalize", |
| 176 | + bad: `$externalize("\uD83D\uDE00"`, |
| 177 | + }, |
| 178 | + } |
| 179 | + for _, c := range mustContain { |
| 180 | + t.Run(c.name, func(t *testing.T) { |
| 181 | + if len(c.good) > 0 && !strings.Contains(out, c.good) { |
| 182 | + t.Errorf("compiled output missing %q\n--- output ---\n%s", c.good, out) |
| 183 | + } |
| 184 | + if len(c.bad) > 0 && strings.Contains(out, c.bad) { |
| 185 | + t.Errorf("compiled output unexpectedly contains %q\n--- output ---\n%s", c.bad, out) |
| 186 | + } |
| 187 | + }) |
| 188 | + } |
| 189 | +} |
0 commit comments