includes/Storage/SlotRecord.php

   1 <?php
   2 /**
   3  * Value object representing a content slot associated with a page revision.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 namespace MediaWiki\Storage;
  24
  25 use Content;
  26 use InvalidArgumentException;
  27 use LogicException;
  28 use OutOfBoundsException;
  29 use Wikimedia\Assert\Assert;
  30
  31 /**
  32  * Value object representing a content slot associated with a page revision.
  33  * SlotRecord provides direct access to a Content object.
  34  * That access may be implemented through a callback.
  35  *
  36  * @since 1.31
  37  */
  38 class SlotRecord {
  39
  40         /**
  41          * @var object database result row, as a raw object
  42          */
  43         private $row;
  44
  45         /**
  46          * @var Content|callable
  47          */
  48         private $content;
  49
  50         /**
  51          * Returns a new SlotRecord just like the given $slot, except that calling getContent()
  52          * will fail with an exception.
  53          *
  54          * @param SlotRecord $slot
  55          *
  56          * @return SlotRecord
  57          */
  58         public static function newWithSuppressedContent( SlotRecord $slot ) {
  59                 $row = $slot->row;
  60
  61                 return new SlotRecord( $row, function () {
  62                         throw new SuppressedDataException( 'Content suppressed!' );
  63                 } );
  64         }
  65
  66         /**
  67          * Constructs a new SlotRecord from an existing SlotRecord, overriding some fields.
  68          * The slot's content cannot be overwritten.
  69          *
  70          * @param SlotRecord $slot
  71          * @param array $overrides
  72          *
  73          * @return SlotRecord
  74          */
  75         private static function newDerived( SlotRecord $slot, array $overrides = [] ) {
  76                 $row = clone $slot->row;
  77                 $row->slot_id = null; // never copy the row ID!
  78
  79                 foreach ( $overrides as $key => $value ) {
  80                         $row->$key = $value;
  81                 }
  82
  83                 return new SlotRecord( $row, $slot->content );
  84         }
  85
  86         /**
  87          * Constructs a new SlotRecord for a new revision, inheriting the content of the given SlotRecord
  88          * of a previous revision.
  89          *
  90          * Note that a SlotRecord constructed this way are intended as prototypes,
  91          * to be used wit newSaved(). They are incomplete, so some getters such as
  92          * getRevision() will fail.
  93          *
  94          * @param SlotRecord $slot
  95          *
  96          * @return SlotRecord
  97          */
  98         public static function newInherited( SlotRecord $slot ) {
  99                 // Sanity check - we can't inherit from a Slot that's not attached to a revision.
 100                 $slot->getRevision();
 101                 $slot->getOrigin();
 102                 $slot->getAddress();
 103
 104                 // NOTE: slot_origin and content_address are copied from $slot.
 105                 return self::newDerived( $slot, [
 106                         'slot_revision_id' => null,
 107                 ] );
 108         }
 109
 110         /**
 111          * Constructs a new Slot from a Content object for a new revision.
 112          * This is the preferred way to construct a slot for storing Content that
 113          * resulted from a user edit. The slot is assumed to be not inherited.
 114          *
 115          * Note that a SlotRecord constructed this way are intended as prototypes,
 116          * to be used wit newSaved(). They are incomplete, so some getters such as
 117          * getAddress() will fail.
 118          *
 119          * @param string $role
 120          * @param Content $content
 121          *
 122          * @return SlotRecord An incomplete proto-slot object, to be used with newSaved() later.
 123          */
 124         public static function newUnsaved( $role, Content $content ) {
 125                 Assert::parameterType( 'string', $role, '$role' );
 126
 127                 $row = [
 128                         'slot_id' => null, // not yet known
 129                         'slot_revision_id' => null, // not yet known
 130                         'slot_origin' => null, // not yet known, will be set in newSaved()
 131                         'content_size' => null, // compute later
 132                         'content_sha1' => null, // compute later
 133                         'slot_content_id' => null, // not yet known, will be set in newSaved()
 134                         'content_address' => null, // not yet known, will be set in newSaved()
 135                         'role_name' => $role,
 136                         'model_name' => $content->getModel(),
 137                 ];
 138
 139                 return new SlotRecord( (object)$row, $content );
 140         }
 141
 142         /**
 143          * Constructs a complete SlotRecord for a newly saved revision, based on the incomplete
 144          * proto-slot. This adds information that has only become available during saving,
 145          * particularly the revision ID and content address.
 146          *
 147          * @param int $revisionId the revision the slot is to be associated with (field slot_revision_id).
 148          *        If $protoSlot already has a revision, it must be the same.
 149          * @param int $contentId the ID of the row in the content table describing the content
 150          *        referenced by $contentAddress (field slot_content_id).
 151          *        If $protoSlot already has a content ID, it must be the same.
 152          * @param string $contentAddress the slot's content address (field content_address).
 153          *        If $protoSlot already has an address, it must be the same.
 154          * @param SlotRecord $protoSlot The proto-slot that was provided as input for creating a new
 155          *        revision. $protoSlot must have a content address if inherited.
 156          *
 157          * @return SlotRecord If the state of $protoSlot is inappropriate for saving a new revision.
 158          */
 159         public static function newSaved(
 160                 $revisionId,
 161                 $contentId,
 162                 $contentAddress,
 163                 SlotRecord $protoSlot
 164         ) {
 165                 Assert::parameterType( 'integer', $revisionId, '$revisionId' );
 166                 Assert::parameterType( 'integer', $contentId, '$contentId' );
 167                 Assert::parameterType( 'string', $contentAddress, '$contentAddress' );
 168
 169                 if ( $protoSlot->hasRevision() && $protoSlot->getRevision() !== $revisionId ) {
 170                         throw new LogicException(
 171                                 "Mismatching revision ID $revisionId: "
 172                                 . "The slot already belongs to revision {$protoSlot->getRevision()}. "
 173                                 . "Use SlotRecord::newInherited() to re-use content between revisions."
 174                         );
 175                 }
 176
 177                 if ( $protoSlot->hasAddress() && $protoSlot->getAddress() !== $contentAddress ) {
 178                         throw new LogicException(
 179                                 "Mismatching blob address $contentAddress: "
 180                                 . "The slot already has content at {$protoSlot->getAddress()}."
 181                         );
 182                 }
 183
 184                 if ( $protoSlot->hasAddress() && $protoSlot->getContentId() !== $contentId ) {
 185                         throw new LogicException(
 186                                 "Mismatching content ID $contentId: "
 187                                 . "The slot already has content row {$protoSlot->getContentId()} associated."
 188                         );
 189                 }
 190
 191                 if ( $protoSlot->isInherited() ) {
 192                         if ( !$protoSlot->hasAddress() ) {
 193                                 throw new InvalidArgumentException(
 194                                         "An inherited blob should have a content address!"
 195                                 );
 196                         }
 197                         if ( !$protoSlot->hasField( 'slot_origin' ) ) {
 198                                 throw new InvalidArgumentException(
 199                                         "A saved inherited slot should have an origin set!"
 200                                 );
 201                         }
 202                         $origin = $protoSlot->getOrigin();
 203                 } else {
 204                         $origin = $revisionId;
 205                 }
 206
 207                 return self::newDerived( $protoSlot, [
 208                         'slot_revision_id' => $revisionId,
 209                         'slot_content_id' => $contentId,
 210                         'slot_origin' => $origin,
 211                         'content_address' => $contentAddress,
 212                 ] );
 213         }
 214
 215         /**
 216          * SlotRecord constructor.
 217          *
 218          * The following fields are supported by the $row parameter:
 219          *
 220          *   $row->blob_data
 221          *   $row->blob_address
 222          *
 223          * @param object $row A database row composed of fields of the slot and content tables,
 224          *        as a raw object. Any field value can be a callback that produces the field value
 225          *        given this SlotRecord as a parameter. However, plain strings cannot be used as
 226          *        callbacks here, for security reasons.
 227          * @param Content|callable $content The content object associated with the slot, or a
 228          *        callback that will return that Content object, given this SlotRecord as a parameter.
 229          */
 230         public function __construct( $row, $content ) {
 231                 Assert::parameterType( 'object', $row, '$row' );
 232                 Assert::parameterType( 'Content|callable', $content, '$content' );
 233
 234                 Assert::parameter(
 235                         property_exists( $row, 'slot_id' ),
 236                         '$row->slot_id',
 237                         'must exist'
 238                 );
 239                 Assert::parameter(
 240                         property_exists( $row, 'slot_revision_id' ),
 241                         '$row->slot_revision_id',
 242                         'must exist'
 243                 );
 244                 Assert::parameter(
 245                         property_exists( $row, 'slot_content_id' ),
 246                         '$row->slot_content_id',
 247                         'must exist'
 248                 );
 249                 Assert::parameter(
 250                         property_exists( $row, 'content_address' ),
 251                         '$row->content_address',
 252                         'must exist'
 253                 );
 254                 Assert::parameter(
 255                         property_exists( $row, 'model_name' ),
 256                         '$row->model_name',
 257                         'must exist'
 258                 );
 259                 Assert::parameter(
 260                         property_exists( $row, 'slot_origin' ),
 261                         '$row->slot_origin',
 262                         'must exist'
 263                 );
 264                 Assert::parameter(
 265                         !property_exists( $row, 'slot_inherited' ),
 266                         '$row->slot_inherited',
 267                         'must not exist'
 268                 );
 269                 Assert::parameter(
 270                         !property_exists( $row, 'slot_revision' ),
 271                         '$row->slot_revision',
 272                         'must not exist'
 273                 );
 274
 275                 $this->row = $row;
 276                 $this->content = $content;
 277         }
 278
 279         /**
 280          * Implemented to defy serialization.
 281          *
 282          * @throws LogicException always
 283          */
 284         public function __sleep() {
 285                 throw new LogicException( __CLASS__ . ' is not serializable.' );
 286         }
 287
 288         /**
 289          * Returns the Content of the given slot.
 290          *
 291          * @note This is free to load Content from whatever subsystem is necessary,
 292          * performing potentially expensive operations and triggering I/O-related
 293          * failure modes.
 294          *
 295          * @note This method does not apply audience filtering.
 296          *
 297          * @throws SuppressedDataException if access to the content is not allowed according
 298          * to the audience check performed by RevisionRecord::getSlot().
 299          *
 300          * @return Content The slot's content. This is a direct reference to the internal instance,
 301          * copy before exposing to application logic!
 302          */
 303         public function getContent() {
 304                 if ( $this->content instanceof Content ) {
 305                         return $this->content;
 306                 }
 307
 308                 $obj = call_user_func( $this->content, $this );
 309
 310                 Assert::postcondition(
 311                         $obj instanceof Content,
 312                         'Slot content callback should return a Content object'
 313                 );
 314
 315                 $this->content = $obj;
 316
 317                 return $this->content;
 318         }
 319
 320         /**
 321          * Returns the string value of a data field from the database row supplied to the constructor.
 322          * If the field was set to a callback, that callback is invoked and the result returned.
 323          *
 324          * @param string $name
 325          *
 326          * @throws OutOfBoundsException
 327          * @throws IncompleteRevisionException
 328          * @return mixed Returns the field's value, never null.
 329          */
 330         private function getField( $name ) {
 331                 if ( !isset( $this->row->$name ) ) {
 332                         // distinguish between unknown and uninitialized fields
 333                         if ( property_exists( $this->row, $name ) ) {
 334                                 throw new IncompleteRevisionException( 'Uninitialized field: ' . $name );
 335                         } else {
 336                                 throw new OutOfBoundsException( 'No such field: ' . $name );
 337                         }
 338                 }
 339
 340                 $value = $this->row->$name;
 341
 342                 // NOTE: allow callbacks, but don't trust plain string callables from the database!
 343                 if ( !is_string( $value ) && is_callable( $value ) ) {
 344                         $value = call_user_func( $value, $this );
 345                         $this->setField( $name, $value );
 346                 }
 347
 348                 return $value;
 349         }
 350
 351         /**
 352          * Returns the string value of a data field from the database row supplied to the constructor.
 353          *
 354          * @param string $name
 355          *
 356          * @throws OutOfBoundsException
 357          * @throws IncompleteRevisionException
 358          * @return string Returns the string value
 359          */
 360         private function getStringField( $name ) {
 361                 return strval( $this->getField( $name ) );
 362         }
 363
 364         /**
 365          * Returns the int value of a data field from the database row supplied to the constructor.
 366          *
 367          * @param string $name
 368          *
 369          * @throws OutOfBoundsException
 370          * @throws IncompleteRevisionException
 371          * @return int Returns the int value
 372          */
 373         private function getIntField( $name ) {
 374                 return intval( $this->getField( $name ) );
 375         }
 376
 377         /**
 378          * @param string $name
 379          * @return bool whether this record contains the given field
 380          */
 381         private function hasField( $name ) {
 382                 return isset( $this->row->$name );
 383         }
 384
 385         /**
 386          * Returns the ID of the revision this slot is associated with.
 387          *
 388          * @return int
 389          */
 390         public function getRevision() {
 391                 return $this->getIntField( 'slot_revision_id' );
 392         }
 393
 394         /**
 395          * Returns the revision ID of the revision that originated the slot's content.
 396          *
 397          * @return int
 398          */
 399         public function getOrigin() {
 400                 return $this->getIntField( 'slot_origin' );
 401         }
 402
 403         /**
 404          * Whether this slot was inherited from an older revision.
 405          *
 406          * If this SlotRecord is already attached to a revision, this returns true
 407          * if the slot's revision of origin is the same as the revision it belongs to.
 408          *
 409          * If this SlotRecord is not yet attached to a revision, this returns true
 410          * if the slot already has an address.
 411          *
 412          * @return bool
 413          */
 414         public function isInherited() {
 415                 if ( $this->hasRevision() ) {
 416                         return $this->getRevision() !== $this->getOrigin();
 417                 } else {
 418                         return $this->hasAddress();
 419                 }
 420         }
 421
 422         /**
 423          * Whether this slot has an address. Slots will have an address if their
 424          * content has been stored. While building a new revision,
 425          * SlotRecords will not have an address associated.
 426          *
 427          * @return bool
 428          */
 429         public function hasAddress() {
 430                 return $this->hasField( 'content_address' );
 431         }
 432
 433         /**
 434          * Whether this slot has revision ID associated. Slots will have a revision ID associated
 435          * only if they were loaded as part of an existing revision. While building a new revision,
 436          * Slotrecords will not have a revision ID associated.
 437          *
 438          * @return bool
 439          */
 440         public function hasRevision() {
 441                 return $this->hasField( 'slot_revision_id' );
 442         }
 443
 444         /**
 445          * Returns the role of the slot.
 446          *
 447          * @return string
 448          */
 449         public function getRole() {
 450                 return $this->getStringField( 'role_name' );
 451         }
 452
 453         /**
 454          * Returns the address of this slot's content.
 455          * This address can be used with BlobStore to load the Content object.
 456          *
 457          * @return string
 458          */
 459         public function getAddress() {
 460                 return $this->getStringField( 'content_address' );
 461         }
 462
 463         /**
 464          * Returns the ID of the content meta data row associated with the slot.
 465          * This information should be irrelevant to application logic, it is here to allow
 466          * the construction of a full row for the revision table.
 467          *
 468          * @return int
 469          */
 470         public function getContentId() {
 471                 return $this->getIntField( 'slot_content_id' );
 472         }
 473
 474         /**
 475          * Returns the content size
 476          *
 477          * @return int size of the content, in bogo-bytes, as reported by Content::getSize.
 478          */
 479         public function getSize() {
 480                 try {
 481                         $size = $this->getIntField( 'content_size' );
 482                 } catch ( IncompleteRevisionException $ex ) {
 483                         $size = $this->getContent()->getSize();
 484                         $this->setField( 'content_size', $size );
 485                 }
 486
 487                 return $size;
 488         }
 489
 490         /**
 491          * Returns the content size
 492          *
 493          * @return string hash of the content.
 494          */
 495         public function getSha1() {
 496                 try {
 497                         $sha1 = $this->getStringField( 'content_sha1' );
 498                 } catch ( IncompleteRevisionException $ex ) {
 499                         $format = $this->hasField( 'format_name' )
 500                                 ? $this->getStringField( 'format_name' )
 501                                 : null;
 502
 503                         $data = $this->getContent()->serialize( $format );
 504                         $sha1 = self::base36Sha1( $data );
 505                         $this->setField( 'content_sha1', $sha1 );
 506                 }
 507
 508                 return $sha1;
 509         }
 510
 511         /**
 512          * Returns the content model. This is the model name that decides
 513          * which ContentHandler is appropriate for interpreting the
 514          * data of the blob referenced by the address returned by getAddress().
 515          *
 516          * @return string the content model of the content
 517          */
 518         public function getModel() {
 519                 try {
 520                         $model = $this->getStringField( 'model_name' );
 521                 } catch ( IncompleteRevisionException $ex ) {
 522                         $model = $this->getContent()->getModel();
 523                         $this->setField( 'model_name', $model );
 524                 }
 525
 526                 return $model;
 527         }
 528
 529         /**
 530          * Returns the blob serialization format as a MIME type.
 531          *
 532          * @note When this method returns null, the caller is expected
 533          * to auto-detect the serialization format, or to rely on
 534          * the default format associated with the content model.
 535          *
 536          * @return string|null
 537          */
 538         public function getFormat() {
 539                 // XXX: we currently do not plan to store the format for each slot!
 540
 541                 if ( $this->hasField( 'format_name' ) ) {
 542                         return $this->getStringField( 'format_name' );
 543                 }
 544
 545                 return null;
 546         }
 547
 548         /**
 549          * @param string $name
 550          * @param string|int|null $value
 551          */
 552         private function setField( $name, $value ) {
 553                 $this->row->$name = $value;
 554         }
 555
 556         /**
 557          * Get the base 36 SHA-1 value for a string of text
 558          *
 559          * MCR migration note: this replaces Revision::base36Sha1
 560          *
 561          * @param string $blob
 562          * @return string
 563          */
 564         public static function base36Sha1( $blob ) {
 565                 return \Wikimedia\base_convert( sha1( $blob ), 16, 36, 31 );
 566         }
 567
 568         /**
 569          * Returns true if $other has the same content as this slot.
 570          * The check is performed based on the model, address size, and hash.
 571          * Two slots can have the same content if they use different content addresses,
 572          * but if they have the same address and the same model, they have the same content.
 573          * Two slots can have the same content if they belong to different
 574          * revisions or pages.
 575          *
 576          * Note that hasSameContent() may return false even if Content::equals returns true for
 577          * the content of two slots. This may happen if the two slots have different serializations
 578          * representing equivalent Content. Such false negatives are considered acceptable. Code
 579          * that has to be absolutely sure the Content is really not the same if hasSameContent()
 580          * returns false should call getContent() and compare the Content objects directly.
 581          *
 582          * @since 1.32
 583          *
 584          * @param SlotRecord $other
 585          * @return bool
 586          */
 587         public function hasSameContent( SlotRecord $other ) {
 588                 if ( $other === $this ) {
 589                         return true;
 590                 }
 591
 592                 if ( $this->getModel() !== $other->getModel() ) {
 593                         return false;
 594                 }
 595
 596                 if ( $this->hasAddress()
 597                         && $other->hasAddress()
 598                         && $this->getAddress() == $other->getAddress()
 599                 ) {
 600                         return true;
 601                 }
 602
 603                 if ( $this->getSize() !== $other->getSize() ) {
 604                         return false;
 605                 }
 606
 607                 if ( $this->getSha1() !== $other->getSha1() ) {
 608                         return false;
 609                 }
 610
 611                 return true;
 612         }
 613
 614 }