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 testEncodeFail() {
 113                 // Set up a recursive object that can't be encoded.
 114                 $a = new stdClass;
 115                 $b = new stdClass;
 116                 $a->b = $b;
 117                 $b->a = $a;
 118                 $this->assertFalse( FormatJson::encode( $a ) );
 119         }
 120
 121         public function testDecodeReturnType() {
 122                 $this->assertInternalType(
 123                         'object',
 124                         FormatJson::decode( '{"Name": "Cheeso", "Rank": 7}' ),
 125                         'Default to object'
 126                 );
 127
 128                 $this->assertInternalType(
 129                         'array',
 130                         FormatJson::decode( '{"Name": "Cheeso", "Rank": 7}', true ),
 131                         'Optional array'
 132                 );
 133         }
 134
 135         public static function provideParse() {
 136                 return [
 137                         [ null ],
 138                         [ true ],
 139                         [ false ],
 140                         [ 0 ],
 141                         [ 1 ],
 142                         [ 1.2 ],
 143                         [ '' ],
 144                         [ 'str' ],
 145                         [ [ 0, 1, 2 ] ],
 146                         [ [ 'a' => 'b' ] ],
 147                         [ [ 'a' => 'b' ] ],
 148                         [ [ 'a' => 'b', 'x' => [ 'c' => 'd' ] ] ],
 149                 ];
 150         }
 151
 152         /**
 153          * Recursively convert arrays into stdClass
 154          * @param array|string|bool|int|float|null $value
 155          * @return stdClass|string|bool|int|float|null
 156          */
 157         public static function toObject( $value ) {
 158                 return !is_array( $value ) ? $value : (object)array_map( __METHOD__, $value );
 159         }
 160
 161         /**
 162          * @dataProvider provideParse
 163          * @param mixed $value
 164          */
 165         public function testParse( $value ) {
 166                 $expected = self::toObject( $value );
 167                 $json = FormatJson::encode( $expected, false, FormatJson::ALL_OK );
 168                 $this->assertJson( $json );
 169
 170                 $st = FormatJson::parse( $json );
 171                 $this->assertInstanceOf( Status::class, $st );
 172                 $this->assertTrue( $st->isGood() );
 173                 $this->assertEquals( $expected, $st->getValue() );
 174
 175                 $st = FormatJson::parse( $json, FormatJson::FORCE_ASSOC );
 176                 $this->assertInstanceOf( Status::class, $st );
 177                 $this->assertTrue( $st->isGood() );
 178                 $this->assertEquals( $value, $st->getValue() );
 179         }
 180
 181         /**
 182          * Test data for testParseTryFixing.
 183          *
 184          * Some PHP interpreters use json-c rather than the JSON.org canonical
 185          * parser to avoid being encumbered by the "shall be used for Good, not
 186          * Evil" clause of the JSON.org parser's license. By default, json-c
 187          * parses in a non-strict mode which allows trailing commas for array and
 188          * object delarations among other things, so our JSON_ERROR_SYNTAX rescue
 189          * block is not always triggered. It however isn't lenient in exactly the
 190          * same ways as our TRY_FIXING mode, so the assertions in this test are
 191          * a bit more complicated than they ideally would be:
 192          *
 193          * Optional third argument: true if json-c parses the value without
 194          * intervention, false otherwise. Defaults to true.
 195          *
 196          * Optional fourth argument: expected cannonical JSON serialization of
 197          * json-c parsed result. Defaults to the second argument's value.
 198          */
 199         public static function provideParseTryFixing() {
 200                 return [
 201                         [ "[,]", '[]', false ],
 202                         [ "[ , ]", '[]', false ],
 203                         [ "[ , }", false ],
 204                         [ '[1],', false, true, '[1]' ],
 205                         [ "[1,]", '[1]' ],
 206                         [ "[1\n,]", '[1]' ],
 207                         [ "[1,\n]", '[1]' ],
 208                         [ "[1,]\n", '[1]' ],
 209                         [ "[1\n,\n]\n", '[1]' ],
 210                         [ '["a,",]', '["a,"]' ],
 211                         [ "[[1,]\n,[2,\n],[3\n,]]", '[[1],[2],[3]]' ],
 212                         // I wish we could parse this, but would need quote parsing
 213                         [ '[[1,],[2,],[3,]]', false, true, '[[1],[2],[3]]' ],
 214                         [ '[1,,]', false, false, '[1]' ],
 215                 ];
 216         }
 217
 218         /**
 219          * @dataProvider provideParseTryFixing
 220          * @param string $value
 221          * @param string|bool $expected Expected result with strict parser
 222          * @param bool $jsoncParses Will json-c parse this value without TRY_FIXING?
 223          * @param string|bool $expectedJsonc Expected result with lenient parser
 224          * if different from the strict expectation
 225          */
 226         public function testParseTryFixing(
 227                 $value, $expected,
 228                 $jsoncParses = true, $expectedJsonc = null
 229         ) {
 230                 // PHP5 results are always expected to have isGood() === false
 231                 $expectedGoodStatus = false;
 232
 233                 // Check to see if json parser allows trailing commas
 234                 if ( json_decode( '[1,]' ) !== null ) {
 235                         // Use json-c specific expected result if provided
 236                         $expected = ( $expectedJsonc === null ) ? $expected : $expectedJsonc;
 237                         // If json-c parses the value natively, expect isGood() === true
 238                         $expectedGoodStatus = $jsoncParses;
 239                 }
 240
 241                 $st = FormatJson::parse( $value, FormatJson::TRY_FIXING );
 242                 $this->assertInstanceOf( Status::class, $st );
 243                 if ( $expected === false ) {
 244                         $this->assertFalse( $st->isOK(), 'Expected isOK() == false' );
 245                 } else {
 246                         $this->assertSame( $expectedGoodStatus, $st->isGood(),
 247                                 'Expected isGood() == ' . ( $expectedGoodStatus ? 'true' : 'false' )
 248                         );
 249                         $this->assertTrue( $st->isOK(), 'Expected isOK == true' );
 250                         $val = FormatJson::encode( $st->getValue(), false, FormatJson::ALL_OK );
 251                         $this->assertEquals( $expected, $val );
 252                 }
 253         }
 254
 255         public static function provideParseErrors() {
 256                 return [
 257                         [ 'aaa' ],
 258                         [ '{"j": 1 ] }' ],
 259                 ];
 260         }
 261
 262         /**
 263          * @dataProvider provideParseErrors
 264          * @param mixed $value
 265          */
 266         public function testParseErrors( $value ) {
 267                 $st = FormatJson::parse( $value );
 268                 $this->assertInstanceOf( Status::class, $st );
 269                 $this->assertFalse( $st->isOK() );
 270         }
 271
 272         public function provideStripComments() {
 273                 return [
 274                         [ '{"a":"b"}', '{"a":"b"}' ],
 275                         [ "{\"a\":\"b\"}\n", "{\"a\":\"b\"}\n" ],
 276                         [ '/*c*/{"c":"b"}', '{"c":"b"}' ],
 277                         [ '{"a":"c"}/*c*/', '{"a":"c"}' ],
 278                         [ '/*c//d*/{"c":"b"}', '{"c":"b"}' ],
 279                         [ '{/*c*/"c":"b"}', '{"c":"b"}' ],
 280                         [ "/*\nc\r\n*/{\"c\":\"b\"}", '{"c":"b"}' ],
 281                         [ "//c\n{\"c\":\"b\"}", '{"c":"b"}' ],
 282                         [ "//c\r\n{\"c\":\"b\"}", '{"c":"b"}' ],
 283                         [ '{"a":"c"}//c', '{"a":"c"}' ],
 284                         [ "{\"a-c\"://c\n\"b\"}", '{"a-c":"b"}' ],
 285                         [ '{"/*a":"b"}', '{"/*a":"b"}' ],
 286                         [ '{"a":"//b"}', '{"a":"//b"}' ],
 287                         [ '{"a":"b/*c*/"}', '{"a":"b/*c*/"}' ],
 288                         [ "{\"\\\"/*a\":\"b\"}", "{\"\\\"/*a\":\"b\"}" ],
 289                         [ '', '' ],
 290                         [ '/*c', '' ],
 291                         [ '//c', '' ],
 292                         [ '"http://example.com"', '"http://example.com"' ],
 293                         [ "\0", "\0" ],
 294                         [ '"Blåbærsyltetøy"', '"Blåbærsyltetøy"' ],
 295                 ];
 296         }
 297
 298         /**
 299          * @covers FormatJson::stripComments
 300          * @dataProvider provideStripComments
 301          * @param string $json
 302          * @param string $expect
 303          */
 304         public function testStripComments( $json, $expect ) {
 305                 $this->assertSame( $expect, FormatJson::stripComments( $json ) );
 306         }
 307
 308         public function provideParseStripComments() {
 309                 return [
 310                         [ '/* blah */true', true ],
 311                         [ "// blah \ntrue", true ],
 312                         [ '[ "a" , /* blah */ "b" ]', [ 'a', 'b' ] ],
 313                 ];
 314         }
 315
 316         /**
 317          * @covers FormatJson::parse
 318          * @covers FormatJson::stripComments
 319          * @dataProvider provideParseStripComments
 320          * @param string $json
 321          * @param mixed $expect
 322          */
 323         public function testParseStripComments( $json, $expect ) {
 324                 $st = FormatJson::parse( $json, FormatJson::STRIP_COMMENTS );
 325                 $this->assertInstanceOf( Status::class, $st );
 326                 $this->assertTrue( $st->isGood() );
 327                 $this->assertEquals( $expect, $st->getValue() );
 328         }
 329
 330         /**
 331          * Generate a set of test cases for a particular combination of encoder options.
 332          *
 333          * @param array $unescapedGroups List of character groups to leave unescaped
 334          * @return array Arrays of unencoded strings and corresponding encoded strings
 335          */
 336         private static function getEncodeTestCases( array $unescapedGroups ) {
 337                 $groups = [
 338                         'always' => [
 339                                 // Forward slash (always unescaped)
 340                                 '/' => '/',
 341
 342                                 // Control characters
 343                                 "\0" => '\u0000',
 344                                 "\x08" => '\b',
 345                                 "\t" => '\t',
 346                                 "\n" => '\n',
 347                                 "\r" => '\r',
 348                                 "\f" => '\f',
 349                                 "\x1f" => '\u001f', // representative example
 350
 351                                 // Double quotes
 352                                 '"' => '\"',
 353
 354                                 // Backslashes
 355                                 '\\' => '\\\\',
 356                                 '\\\\' => '\\\\\\\\',
 357                                 '\\u00e9' => '\\\u00e9', // security check for Unicode unescaping
 358
 359                                 // Line terminators
 360                                 "\xe2\x80\xa8" => '\u2028',
 361                                 "\xe2\x80\xa9" => '\u2029',
 362                         ],
 363                         'unicode' => [
 364                                 "\xc3\xa9" => '\u00e9',
 365                                 "\xf0\x9d\x92\x9e" => '\ud835\udc9e', // U+1D49E, outside the BMP
 366                         ],
 367                         'xmlmeta' => [
 368                                 '<' => '\u003C', // JSON_HEX_TAG uses uppercase hex digits
 369                                 '>' => '\u003E',
 370                                 '&' => '\u0026',
 371                         ],
 372                 ];
 373
 374                 $cases = [];
 375                 foreach ( $groups as $name => $rules ) {
 376                         $leaveUnescaped = in_array( $name, $unescapedGroups );
 377                         foreach ( $rules as $from => $to ) {
 378                                 $cases[] = [ $from, '"' . ( $leaveUnescaped ? $from : $to ) . '"' ];
 379                         }
 380                 }
 381
 382                 return $cases;
 383         }
 384
 385         public function provideEmptyJsonKeyStrings() {
 386                 return [
 387                         [
 388                                 '{"":"foo"}',
 389                                 '{"":"foo"}',
 390                                 ''
 391                         ],
 392                         [
 393                                 '{"_empty_":"foo"}',
 394                                 '{"_empty_":"foo"}',
 395                                 '_empty_' ],
 396                         [
 397                                 '{"\u005F\u0065\u006D\u0070\u0074\u0079\u005F":"foo"}',
 398                                 '{"_empty_":"foo"}',
 399                                 '_empty_'
 400                         ],
 401                         [
 402                                 '{"_empty_":"bar","":"foo"}',
 403                                 '{"_empty_":"bar","":"foo"}',
 404                                 ''
 405                         ],
 406                         [
 407                                 '{"":"bar","_empty_":"foo"}',
 408                                 '{"":"bar","_empty_":"foo"}',
 409                                 '_empty_'
 410                         ]
 411                 ];
 412         }
 413
 414         /**
 415          * @covers FormatJson::encode
 416          * @covers FormatJson::decode
 417          * @dataProvider provideEmptyJsonKeyStrings
 418          * @param string $json
 419          *
 420          * Decoding behavior with empty keys can be surprising.
 421          * See https://phabricator.wikimedia.org/T206411
 422          */
 423         public function testEmptyJsonKeyArray( $json, $expect, $php71Name ) {
 424                 // Decoding to array is consistent across supported PHP versions
 425                 $this->assertSame( $expect, FormatJson::encode(
 426                         FormatJson::decode( $json, true ) ) );
 427
 428                 // Decoding to object differs between supported PHP versions
 429                 $obj = FormatJson::decode( $json );
 430                 if ( version_compare( PHP_VERSION, '7.1', '<' ) ) {
 431                         $this->assertEquals( 'foo', $obj->_empty_ );
 432                 } else {
 433                         $this->assertEquals( 'foo', $obj->{$php71Name} );
 434                 }
 435         }
 436 }