tests/phpunit/includes/json/FormatJsonTest.php

   1 <?php
   2
   3 /**
   4  * @covers FormatJson
   5  */
   6 class FormatJsonTest extends MediaWikiTestCase {
   7
   8         public static function provideEncoderPrettyPrinting() {
   9                 return [
  10                         // Four spaces
  11                         [ true, '    ' ],
  12                         [ '    ', '    ' ],
  13                         // Two spaces
  14                         [ '  ', '  ' ],
  15                         // One tab
  16                         [ "\t", "\t" ],
  17                 ];
  18         }
  19
  20         /**
  21          * @dataProvider provideEncoderPrettyPrinting
  22          */
  23         public function testEncoderPrettyPrinting( $pretty, $expectedIndent ) {
  24                 $obj = [
  25                         'emptyObject' => new stdClass,
  26                         'emptyArray' => [],
  27                         'string' => 'foobar\\',
  28                         'filledArray' => [
  29                                 [
  30                                         123,
  31                                         456,
  32                                 ],
  33                                 // Nested json works without problems
  34                                 '"7":["8",{"9":"10"}]',
  35                                 // Whitespace clean up doesn't touch strings that look alike
  36                                 "{\n\t\"emptyObject\": {\n\t},\n\t\"emptyArray\": [ ]\n}",
  37                         ],
  38                 ];
  39
  40                 // No trailing whitespace, no trailing linefeed
  41                 $json = '{
  42         "emptyObject": {},
  43         "emptyArray": [],
  44         "string": "foobar\\\\",
  45         "filledArray": [
  46                 [
  47                         123,
  48                         456
  49                 ],
  50                 "\"7\":[\"8\",{\"9\":\"10\"}]",
  51                 "{\n\t\"emptyObject\": {\n\t},\n\t\"emptyArray\": [ ]\n}"
  52         ]
  53 }';
  54
  55                 $json = str_replace( "\r", '', $json ); // Windows compat
  56                 $json = str_replace( "\t", $expectedIndent, $json );
  57                 $this->assertSame( $json, FormatJson::encode( $obj, $pretty ) );
  58         }
  59
  60         public static function provideEncodeDefault() {
  61                 return self::getEncodeTestCases( [] );
  62         }
  63
  64         /**
  65          * @dataProvider provideEncodeDefault
  66          */
  67         public function testEncodeDefault( $from, $to ) {
  68                 $this->assertSame( $to, FormatJson::encode( $from ) );
  69         }
  70
  71         public static function provideEncodeUtf8() {
  72                 return self::getEncodeTestCases( [ 'unicode' ] );
  73         }
  74
  75         /**
  76          * @dataProvider provideEncodeUtf8
  77          */
  78         public function testEncodeUtf8( $from, $to ) {
  79                 $this->assertSame( $to, FormatJson::encode( $from, false, FormatJson::UTF8_OK ) );
  80         }
  81
  82         public static function provideEncodeXmlMeta() {
  83                 return self::getEncodeTestCases( [ 'xmlmeta' ] );
  84         }
  85
  86         /**
  87          * @dataProvider provideEncodeXmlMeta
  88          */
  89         public function testEncodeXmlMeta( $from, $to ) {
  90                 $this->assertSame( $to, FormatJson::encode( $from, false, FormatJson::XMLMETA_OK ) );
  91         }
  92
  93         public static function provideEncodeAllOk() {
  94                 return self::getEncodeTestCases( [ 'unicode', 'xmlmeta' ] );
  95         }
  96
  97         /**
  98          * @dataProvider provideEncodeAllOk
  99          */
 100         public function testEncodeAllOk( $from, $to ) {
 101                 $this->assertSame( $to, FormatJson::encode( $from, false, FormatJson::ALL_OK ) );
 102         }
 103
 104         public function testEncodePhpBug46944() {
 105                 $this->assertNotEquals(
 106                         '\ud840\udc00',
 107                         strtolower( FormatJson::encode( "\xf0\xa0\x80\x80" ) ),
 108                         'Test encoding an broken json_encode character (U+20000)'
 109                 );
 110         }
 111
 112         public function testDecodeReturnType() {
 113                 $this->assertInternalType(
 114                         'object',
 115                         FormatJson::decode( '{"Name": "Cheeso", "Rank": 7}' ),
 116                         'Default to object'
 117                 );
 118
 119                 $this->assertInternalType(
 120                         'array',
 121                         FormatJson::decode( '{"Name": "Cheeso", "Rank": 7}', true ),
 122                         'Optional array'
 123                 );
 124         }
 125
 126         public static function provideParse() {
 127                 return [
 128                         [ null ],
 129                         [ true ],
 130                         [ false ],
 131                         [ 0 ],
 132                         [ 1 ],
 133                         [ 1.2 ],
 134                         [ '' ],
 135                         [ 'str' ],
 136                         [ [ 0, 1, 2 ] ],
 137                         [ [ 'a' => 'b' ] ],
 138                         [ [ 'a' => 'b' ] ],
 139                         [ [ 'a' => 'b', 'x' => [ 'c' => 'd' ] ] ],
 140                 ];
 141         }
 142
 143         /**
 144          * Recursively convert arrays into stdClass
 145          * @param array|string|bool|int|float|null $value
 146          * @return stdClass|string|bool|int|float|null
 147          */
 148         public static function toObject( $value ) {
 149                 return !is_array( $value ) ? $value : (object)array_map( __METHOD__, $value );
 150         }
 151
 152         /**
 153          * @dataProvider provideParse
 154          * @param mixed $value
 155          */
 156         public function testParse( $value ) {
 157                 $expected = self::toObject( $value );
 158                 $json = FormatJson::encode( $expected, false, FormatJson::ALL_OK );
 159                 $this->assertJson( $json );
 160
 161                 $st = FormatJson::parse( $json );
 162                 $this->assertInstanceOf( Status::class, $st );
 163                 $this->assertTrue( $st->isGood() );
 164                 $this->assertEquals( $expected, $st->getValue() );
 165
 166                 $st = FormatJson::parse( $json, FormatJson::FORCE_ASSOC );
 167                 $this->assertInstanceOf( Status::class, $st );
 168                 $this->assertTrue( $st->isGood() );
 169                 $this->assertEquals( $value, $st->getValue() );
 170         }
 171
 172         /**
 173          * Test data for testParseTryFixing.
 174          *
 175          * Some PHP interpreters use json-c rather than the JSON.org cannonical
 176          * parser to avoid being encumbered by the "shall be used for Good, not
 177          * Evil" clause of the JSON.org parser's license. By default, json-c
 178          * parses in a non-strict mode which allows trailing commas for array and
 179          * object delarations among other things, so our JSON_ERROR_SYNTAX rescue
 180          * block is not always triggered. It however isn't lenient in exactly the
 181          * same ways as our TRY_FIXING mode, so the assertions in this test are
 182          * a bit more complicated than they ideally would be:
 183          *
 184          * Optional third argument: true if json-c parses the value without
 185          * intervention, false otherwise. Defaults to true.
 186          *
 187          * Optional fourth argument: expected cannonical JSON serialization of
 188          * json-c parsed result. Defaults to the second argument's value.
 189          */
 190         public static function provideParseTryFixing() {
 191                 return [
 192                         [ "[,]", '[]', false ],
 193                         [ "[ , ]", '[]', false ],
 194                         [ "[ , }", false ],
 195                         [ '[1],', false, true, '[1]' ],
 196                         [ "[1,]", '[1]' ],
 197                         [ "[1\n,]", '[1]' ],
 198                         [ "[1,\n]", '[1]' ],
 199                         [ "[1,]\n", '[1]' ],
 200                         [ "[1\n,\n]\n", '[1]' ],
 201                         [ '["a,",]', '["a,"]' ],
 202                         [ "[[1,]\n,[2,\n],[3\n,]]", '[[1],[2],[3]]' ],
 203                         // I wish we could parse this, but would need quote parsing
 204                         [ '[[1,],[2,],[3,]]', false, true, '[[1],[2],[3]]' ],
 205                         [ '[1,,]', false, false, '[1]' ],
 206                 ];
 207         }
 208
 209         /**
 210          * @dataProvider provideParseTryFixing
 211          * @param string $value
 212          * @param string|bool $expected Expected result with strict parser
 213          * @param bool $jsoncParses Will json-c parse this value without TRY_FIXING?
 214          * @param string|bool $expectedJsonc Expected result with lenient parser
 215          * if different from the strict expectation
 216          */
 217         public function testParseTryFixing(
 218                 $value, $expected,
 219                 $jsoncParses = true, $expectedJsonc = null
 220         ) {
 221                 // PHP5 results are always expected to have isGood() === false
 222                 $expectedGoodStatus = false;
 223
 224                 // Check to see if json parser allows trailing commas
 225                 if ( json_decode( '[1,]' ) !== null ) {
 226                         // Use json-c specific expected result if provided
 227                         $expected = ( $expectedJsonc === null ) ? $expected : $expectedJsonc;
 228                         // If json-c parses the value natively, expect isGood() === true
 229                         $expectedGoodStatus = $jsoncParses;
 230                 }
 231
 232                 $st = FormatJson::parse( $value, FormatJson::TRY_FIXING );
 233                 $this->assertInstanceOf( Status::class, $st );
 234                 if ( $expected === false ) {
 235                         $this->assertFalse( $st->isOK(), 'Expected isOK() == false' );
 236                 } else {
 237                         $this->assertSame( $expectedGoodStatus, $st->isGood(),
 238                                 'Expected isGood() == ' . ( $expectedGoodStatus ? 'true' : 'false' )
 239                         );
 240                         $this->assertTrue( $st->isOK(), 'Expected isOK == true' );
 241                         $val = FormatJson::encode( $st->getValue(), false, FormatJson::ALL_OK );
 242                         $this->assertEquals( $expected, $val );
 243                 }
 244         }
 245
 246         public static function provideParseErrors() {
 247                 return [
 248                         [ 'aaa' ],
 249                         [ '{"j": 1 ] }' ],
 250                 ];
 251         }
 252
 253         /**
 254          * @dataProvider provideParseErrors
 255          * @param mixed $value
 256          */
 257         public function testParseErrors( $value ) {
 258                 $st = FormatJson::parse( $value );
 259                 $this->assertInstanceOf( Status::class, $st );
 260                 $this->assertFalse( $st->isOK() );
 261         }
 262
 263         public function provideStripComments() {
 264                 return [
 265                         [ '{"a":"b"}', '{"a":"b"}' ],
 266                         [ "{\"a\":\"b\"}\n", "{\"a\":\"b\"}\n" ],
 267                         [ '/*c*/{"c":"b"}', '{"c":"b"}' ],
 268                         [ '{"a":"c"}/*c*/', '{"a":"c"}' ],
 269                         [ '/*c//d*/{"c":"b"}', '{"c":"b"}' ],
 270                         [ '{/*c*/"c":"b"}', '{"c":"b"}' ],
 271                         [ "/*\nc\r\n*/{\"c\":\"b\"}", '{"c":"b"}' ],
 272                         [ "//c\n{\"c\":\"b\"}", '{"c":"b"}' ],
 273                         [ "//c\r\n{\"c\":\"b\"}", '{"c":"b"}' ],
 274                         [ '{"a":"c"}//c', '{"a":"c"}' ],
 275                         [ "{\"a-c\"://c\n\"b\"}", '{"a-c":"b"}' ],
 276                         [ '{"/*a":"b"}', '{"/*a":"b"}' ],
 277                         [ '{"a":"//b"}', '{"a":"//b"}' ],
 278                         [ '{"a":"b/*c*/"}', '{"a":"b/*c*/"}' ],
 279                         [ "{\"\\\"/*a\":\"b\"}", "{\"\\\"/*a\":\"b\"}" ],
 280                         [ '', '' ],
 281                         [ '/*c', '' ],
 282                         [ '//c', '' ],
 283                         [ '"http://example.com"', '"http://example.com"' ],
 284                         [ "\0", "\0" ],
 285                         [ '"Blåbærsyltetøy"', '"Blåbærsyltetøy"' ],
 286                 ];
 287         }
 288
 289         /**
 290          * @covers FormatJson::stripComments
 291          * @dataProvider provideStripComments
 292          * @param string $json
 293          * @param string $expect
 294          */
 295         public function testStripComments( $json, $expect ) {
 296                 $this->assertSame( $expect, FormatJson::stripComments( $json ) );
 297         }
 298
 299         public function provideParseStripComments() {
 300                 return [
 301                         [ '/* blah */true', true ],
 302                         [ "// blah \ntrue", true ],
 303                         [ '[ "a" , /* blah */ "b" ]', [ 'a', 'b' ] ],
 304                 ];
 305         }
 306
 307         /**
 308          * @covers FormatJson::parse
 309          * @covers FormatJson::stripComments
 310          * @dataProvider provideParseStripComments
 311          * @param string $json
 312          * @param mixed $expect
 313          */
 314         public function testParseStripComments( $json, $expect ) {
 315                 $st = FormatJson::parse( $json, FormatJson::STRIP_COMMENTS );
 316                 $this->assertInstanceOf( Status::class, $st );
 317                 $this->assertTrue( $st->isGood() );
 318                 $this->assertEquals( $expect, $st->getValue() );
 319         }
 320
 321         /**
 322          * Generate a set of test cases for a particular combination of encoder options.
 323          *
 324          * @param array $unescapedGroups List of character groups to leave unescaped
 325          * @return array Arrays of unencoded strings and corresponding encoded strings
 326          */
 327         private static function getEncodeTestCases( array $unescapedGroups ) {
 328                 $groups = [
 329                         'always' => [
 330                                 // Forward slash (always unescaped)
 331                                 '/' => '/',
 332
 333                                 // Control characters
 334                                 "\0" => '\u0000',
 335                                 "\x08" => '\b',
 336                                 "\t" => '\t',
 337                                 "\n" => '\n',
 338                                 "\r" => '\r',
 339                                 "\f" => '\f',
 340                                 "\x1f" => '\u001f', // representative example
 341
 342                                 // Double quotes
 343                                 '"' => '\"',
 344
 345                                 // Backslashes
 346                                 '\\' => '\\\\',
 347                                 '\\\\' => '\\\\\\\\',
 348                                 '\\u00e9' => '\\\u00e9', // security check for Unicode unescaping
 349
 350                                 // Line terminators
 351                                 "\xe2\x80\xa8" => '\u2028',
 352                                 "\xe2\x80\xa9" => '\u2029',
 353                         ],
 354                         'unicode' => [
 355                                 "\xc3\xa9" => '\u00e9',
 356                                 "\xf0\x9d\x92\x9e" => '\ud835\udc9e', // U+1D49E, outside the BMP
 357                         ],
 358                         'xmlmeta' => [
 359                                 '<' => '\u003C', // JSON_HEX_TAG uses uppercase hex digits
 360                                 '>' => '\u003E',
 361                                 '&' => '\u0026',
 362                         ],
 363                 ];
 364
 365                 $cases = [];
 366                 foreach ( $groups as $name => $rules ) {
 367                         $leaveUnescaped = in_array( $name, $unescapedGroups );
 368                         foreach ( $rules as $from => $to ) {
 369                                 $cases[] = [ $from, '"' . ( $leaveUnescaped ? $from : $to ) . '"' ];
 370                         }
 371                 }
 372
 373                 return $cases;
 374         }
 375 }