includes/Revision/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\Revision;
  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  * @since 1.32 Renamed from MediaWiki\Storage\SlotRecord
  38  */
  39 class SlotRecord {
  40
  41         const MAIN = 'main';
  42
  43         /**
  44          * @var object database result row, as a raw object. Callbacks are supported for field values,
  45          *      to enable on-demand emulation of these values. This is primarily intended for use
  46          *      during schema migration.
  47          */
  48         private $row;
  49
  50         /**
  51          * @var Content|callable
  52          */
  53         private $content;
  54
  55         /**
  56          * Returns a new SlotRecord just like the given $slot, except that calling getContent()
  57          * will fail with an exception.
  58          *
  59          * @param SlotRecord $slot
  60          *
  61          * @return SlotRecord
  62          */
  63         public static function newWithSuppressedContent( SlotRecord $slot ) {
  64                 $row = $slot->row;
  65
  66                 return new SlotRecord( $row, function () {
  67                         throw new SuppressedDataException( 'Content suppressed!' );
  68                 } );
  69         }
  70
  71         /**
  72          * Constructs a new SlotRecord from an existing SlotRecord, overriding some fields.
  73          * The slot's content cannot be overwritten.
  74          *
  75          * @param SlotRecord $slot
  76          * @param array $overrides
  77          *
  78          * @return SlotRecord
  79          */
  80         private static function newDerived( SlotRecord $slot, array $overrides = [] ) {
  81                 $row = clone $slot->row;
  82                 $row->slot_id = null; // never copy the row ID!
  83
  84                 foreach ( $overrides as $key => $value ) {
  85                         $row->$key = $value;
  86                 }
  87
  88                 return new SlotRecord( $row, $slot->content );
  89         }
  90
  91         /**
  92          * Constructs a new SlotRecord for a new revision, inheriting the content of the given SlotRecord
  93          * of a previous revision.
  94          *
  95          * Note that a SlotRecord constructed this way are intended as prototypes,
  96          * to be used wit newSaved(). They are incomplete, so some getters such as
  97          * getRevision() will fail.
  98          *
  99          * @param SlotRecord $slot
 100          *
 101          * @return SlotRecord
 102          */
 103         public static function newInherited( SlotRecord $slot ) {
 104                 // Sanity check - we can't inherit from a Slot that's not attached to a revision.
 105                 $slot->getRevision();
 106                 $slot->getOrigin();
 107                 $slot->getAddress();
 108
 109                 // NOTE: slot_origin and content_address are copied from $slot.
 110                 return self::newDerived( $slot, [
 111                         'slot_revision_id' => null,
 112                 ] );
 113         }
 114
 115         /**
 116          * Constructs a new Slot from a Content object for a new revision.
 117          * This is the preferred way to construct a slot for storing Content that
 118          * resulted from a user edit. The slot is assumed to be not inherited.
 119          *
 120          * Note that a SlotRecord constructed this way are intended as prototypes,
 121          * to be used wit newSaved(). They are incomplete, so some getters such as
 122          * getAddress() will fail.
 123          *
 124          * @param string $role
 125          * @param Content $content
 126          *
 127          * @return SlotRecord An incomplete proto-slot object, to be used with newSaved() later.
 128          */
 129         public static function newUnsaved( $role, Content $content ) {
 130                 Assert::parameterType( 'string', $role, '$role' );
 131
 132                 $row = [
 133                         'slot_id' => null, // not yet known
 134                         'slot_revision_id' => null, // not yet known
 135                         'slot_origin' => null, // not yet known, will be set in newSaved()
 136                         'content_size' => null, // compute later
 137                         'content_sha1' => null, // compute later
 138                         'slot_content_id' => null, // not yet known, will be set in newSaved()
 139                         'content_address' => null, // not yet known, will be set in newSaved()
 140                         'role_name' => $role,
 141                         'model_name' => $content->getModel(),
 142                 ];
 143
 144                 return new SlotRecord( (object)$row, $content );
 145         }
 146
 147         /**
 148          * Constructs a complete SlotRecord for a newly saved revision, based on the incomplete
 149          * proto-slot. This adds information that has only become available during saving,
 150          * particularly the revision ID, content ID and content address.
 151          *
 152          * @param int $revisionId the revision the slot is to be associated with (field slot_revision_id).
 153          *        If $protoSlot already has a revision, it must be the same.
 154          * @param int|null $contentId the ID of the row in the content table describing the content
 155          *        referenced by $contentAddress (field slot_content_id).
 156          *        If $protoSlot already has a content ID, it must be the same.
 157          * @param string $contentAddress the slot's content address (field content_address).
 158          *        If $protoSlot already has an address, it must be the same.
 159          * @param SlotRecord $protoSlot The proto-slot that was provided as input for creating a new
 160          *        revision. $protoSlot must have a content address if inherited.
 161          *
 162          * @return SlotRecord If the state of $protoSlot is inappropriate for saving a new revision.
 163          */
 164         public static function newSaved(
 165                 $revisionId,
 166                 $contentId,
 167                 $contentAddress,
 168                 SlotRecord $protoSlot
 169         ) {
 170                 Assert::parameterType( 'integer', $revisionId, '$revisionId' );
 171                 // TODO once migration is over $contentId must be an integer
 172                 Assert::parameterType( 'integer|null', $contentId, '$contentId' );
 173                 Assert::parameterType( 'string', $contentAddress, '$contentAddress' );
 174
 175                 if ( $protoSlot->hasRevision() && $protoSlot->getRevision() !== $revisionId ) {
 176                         throw new LogicException(
 177                                 "Mismatching revision ID $revisionId: "
 178                                 . "The slot already belongs to revision {$protoSlot->getRevision()}. "
 179                                 . "Use SlotRecord::newInherited() to re-use content between revisions."
 180                         );
 181                 }
 182
 183                 if ( $protoSlot->hasAddress() && $protoSlot->getAddress() !== $contentAddress ) {
 184                         throw new LogicException(
 185                                 "Mismatching blob address $contentAddress: "
 186                                 . "The slot already has content at {$protoSlot->getAddress()}."
 187                         );
 188                 }
 189
 190                 if ( $protoSlot->hasContentId() && $protoSlot->getContentId() !== $contentId ) {
 191                         throw new LogicException(
 192                                 "Mismatching content ID $contentId: "
 193                                 . "The slot already has content row {$protoSlot->getContentId()} associated."
 194                         );
 195                 }
 196
 197                 if ( $protoSlot->isInherited() ) {
 198                         if ( !$protoSlot->hasAddress() ) {
 199                                 throw new InvalidArgumentException(
 200                                         "An inherited blob should have a content address!"
 201                                 );
 202                         }
 203                         if ( !$protoSlot->hasField( 'slot_origin' ) ) {
 204                                 throw new InvalidArgumentException(
 205                                         "A saved inherited slot should have an origin set!"
 206                                 );
 207                         }
 208                         $origin = $protoSlot->getOrigin();
 209                 } else {
 210                         $origin = $revisionId;
 211                 }
 212
 213                 return self::newDerived( $protoSlot, [
 214                         'slot_revision_id' => $revisionId,
 215                         'slot_content_id' => $contentId,
 216                         'slot_origin' => $origin,
 217                         'content_address' => $contentAddress,
 218                 ] );
 219         }
 220
 221         /**
 222          * SlotRecord constructor.
 223          *
 224          * The following fields are supported by the $row parameter:
 225          *
 226          *   $row->blob_data
 227          *   $row->blob_address
 228          *
 229          * @param object $row A database row composed of fields of the slot and content tables,
 230          *        as a raw object. Any field value can be a callback that produces the field value
 231          *        given this SlotRecord as a parameter. However, plain strings cannot be used as
 232          *        callbacks here, for security reasons.
 233          * @param Content|callable $content The content object associated with the slot, or a
 234          *        callback that will return that Content object, given this SlotRecord as a parameter.
 235          */
 236         public function __construct( $row, $content ) {
 237                 Assert::parameterType( 'object', $row, '$row' );
 238                 Assert::parameterType( 'Content|callable', $content, '$content' );
 239
 240                 Assert::parameter(
 241                         property_exists( $row, 'slot_revision_id' ),
 242                         '$row->slot_revision_id',
 243                         'must exist'
 244                 );
 245                 Assert::parameter(
 246                         property_exists( $row, 'slot_content_id' ),
 247                         '$row->slot_content_id',
 248                         'must exist'
 249                 );
 250                 Assert::parameter(
 251                         property_exists( $row, 'content_address' ),
 252                         '$row->content_address',
 253                         'must exist'
 254                 );
 255                 Assert::parameter(
 256                         property_exists( $row, 'model_name' ),
 257                         '$row->model_name',
 258                         'must exist'
 259                 );
 260                 Assert::parameter(
 261                         property_exists( $row, 'slot_origin' ),
 262                         '$row->slot_origin',
 263                         'must exist'
 264                 );
 265                 Assert::parameter(
 266                         !property_exists( $row, 'slot_inherited' ),
 267                         '$row->slot_inherited',
 268                         'must not exist'
 269                 );
 270                 Assert::parameter(
 271                         !property_exists( $row, 'slot_revision' ),
 272                         '$row->slot_revision',
 273                         'must not exist'
 274                 );
 275
 276                 $this->row = $row;
 277                 $this->content = $content;
 278         }
 279
 280         /**
 281          * Implemented to defy serialization.
 282          *
 283          * @throws LogicException always
 284          */
 285         public function __sleep() {
 286                 throw new LogicException( __CLASS__ . ' is not serializable.' );
 287         }
 288
 289         /**
 290          * Returns the Content of the given slot.
 291          *
 292          * @note This is free to load Content from whatever subsystem is necessary,
 293          * performing potentially expensive operations and triggering I/O-related
 294          * failure modes.
 295          *
 296          * @note This method does not apply audience filtering.
 297          *
 298          * @throws SuppressedDataException if access to the content is not allowed according
 299          * to the audience check performed by RevisionRecord::getSlot().
 300          *
 301          * @return Content The slot's content. This is a direct reference to the internal instance,
 302          * copy before exposing to application logic!
 303          */
 304         public function getContent() {
 305                 if ( $this->content instanceof Content ) {
 306                         return $this->content;
 307                 }
 308
 309                 $obj = call_user_func( $this->content, $this );
 310
 311                 Assert::postcondition(
 312                         $obj instanceof Content,
 313                         'Slot content callback should return a Content object'
 314                 );
 315
 316                 $this->content = $obj;
 317
 318                 return $this->content;
 319         }
 320
 321         /**
 322          * Returns the string value of a data field from the database row supplied to the constructor.
 323          * If the field was set to a callback, that callback is invoked and the result returned.
 324          *
 325          * @param string $name
 326          *
 327          * @throws OutOfBoundsException
 328          * @throws IncompleteRevisionException
 329          * @return mixed Returns the field's value, never null.
 330          */
 331         private function getField( $name ) {
 332                 if ( !isset( $this->row->$name ) ) {
 333                         // distinguish between unknown and uninitialized fields
 334                         if ( property_exists( $this->row, $name ) ) {
 335                                 throw new IncompleteRevisionException( 'Uninitialized field: ' . $name );
 336                         } else {
 337                                 throw new OutOfBoundsException( 'No such field: ' . $name );
 338                         }
 339                 }
 340
 341                 $value = $this->row->$name;
 342
 343                 // NOTE: allow callbacks, but don't trust plain string callables from the database!
 344                 if ( !is_string( $value ) && is_callable( $value ) ) {
 345                         $value = call_user_func( $value, $this );
 346                         $this->setField( $name, $value );
 347                 }
 348
 349                 return $value;
 350         }
 351
 352         /**
 353          * Returns the string value of a data field from the database row supplied to the constructor.
 354          *
 355          * @param string $name
 356          *
 357          * @throws OutOfBoundsException
 358          * @throws IncompleteRevisionException
 359          * @return string Returns the string value
 360          */
 361         private function getStringField( $name ) {
 362                 return strval( $this->getField( $name ) );
 363         }
 364
 365         /**
 366          * Returns the int value of a data field from the database row supplied to the constructor.
 367          *
 368          * @param string $name
 369          *
 370          * @throws OutOfBoundsException
 371          * @throws IncompleteRevisionException
 372          * @return int Returns the int value
 373          */
 374         private function getIntField( $name ) {
 375                 return intval( $this->getField( $name ) );
 376         }
 377
 378         /**
 379          * @param string $name
 380          * @return bool whether this record contains the given field
 381          */
 382         private function hasField( $name ) {
 383                 if ( isset( $this->row->$name ) ) {
 384                         // if the field is a callback, resolve first, then re-check
 385                         if ( !is_string( $this->row->$name ) && is_callable( $this->row->$name ) ) {
 386                                 $this->getField( $name );
 387                         }
 388                 }
 389
 390                 return isset( $this->row->$name );
 391         }
 392
 393         /**
 394          * Returns the ID of the revision this slot is associated with.
 395          *
 396          * @return int
 397          */
 398         public function getRevision() {
 399                 return $this->getIntField( 'slot_revision_id' );
 400         }
 401
 402         /**
 403          * Returns the revision ID of the revision that originated the slot's content.
 404          *
 405          * @return int
 406          */
 407         public function getOrigin() {
 408                 return $this->getIntField( 'slot_origin' );
 409         }
 410
 411         /**
 412          * Whether this slot was inherited from an older revision.
 413          *
 414          * If this SlotRecord is already attached to a revision, this returns true
 415          * if the slot's revision of origin is the same as the revision it belongs to.
 416          *
 417          * If this SlotRecord is not yet attached to a revision, this returns true
 418          * if the slot already has an address.
 419          *
 420          * @return bool
 421          */
 422         public function isInherited() {
 423                 if ( $this->hasRevision() ) {
 424                         return $this->getRevision() !== $this->getOrigin();
 425                 } else {
 426                         return $this->hasAddress();
 427                 }
 428         }
 429
 430         /**
 431          * Whether this slot has an address. Slots will have an address if their
 432          * content has been stored. While building a new revision,
 433          * SlotRecords will not have an address associated.
 434          *
 435          * @return bool
 436          */
 437         public function hasAddress() {
 438                 return $this->hasField( 'content_address' );
 439         }
 440
 441         /**
 442          * Whether this slot has an origin (revision ID that originated the slot's content.
 443          *
 444          * @since 1.32
 445          *
 446          * @return bool
 447          */
 448         public function hasOrigin() {
 449                 return $this->hasField( 'slot_origin' );
 450         }
 451
 452         /**
 453          * Whether this slot has a content ID. Slots will have a content ID if their
 454          * content has been stored in the content table. While building a new revision,
 455          * SlotRecords will not have an ID associated.
 456          *
 457          * Also, during schema migration, hasContentId() may return false when encountering an
 458          * un-migrated database entry in SCHEMA_COMPAT_WRITE_BOTH mode.
 459          * It will however always return true for saved revisions on SCHEMA_COMPAT_READ_NEW mode,
 460          * or without SCHEMA_COMPAT_WRITE_NEW mode. In the latter case, an emulated content ID
 461          * is used, derived from the revision's text ID.
 462          *
 463          * Note that hasContentId() returning false while hasRevision() returns true always
 464          * indicates an unmigrated row in SCHEMA_COMPAT_WRITE_BOTH mode, as described above.
 465          * For an unsaved slot, both these methods would return false.
 466          *
 467          * @since 1.32
 468          *
 469          * @return bool
 470          */
 471         public function hasContentId() {
 472                 return $this->hasField( 'slot_content_id' );
 473         }
 474
 475         /**
 476          * Whether this slot has revision ID associated. Slots will have a revision ID associated
 477          * only if they were loaded as part of an existing revision. While building a new revision,
 478          * Slotrecords will not have a revision ID associated.
 479          *
 480          * @return bool
 481          */
 482         public function hasRevision() {
 483                 return $this->hasField( 'slot_revision_id' );
 484         }
 485
 486         /**
 487          * Returns the role of the slot.
 488          *
 489          * @return string
 490          */
 491         public function getRole() {
 492                 return $this->getStringField( 'role_name' );
 493         }
 494
 495         /**
 496          * Returns the address of this slot's content.
 497          * This address can be used with BlobStore to load the Content object.
 498          *
 499          * @return string
 500          */
 501         public function getAddress() {
 502                 return $this->getStringField( 'content_address' );
 503         }
 504
 505         /**
 506          * Returns the ID of the content meta data row associated with the slot.
 507          * This information should be irrelevant to application logic, it is here to allow
 508          * the construction of a full row for the revision table.
 509          *
 510          * Note that this method may return an emulated value during schema migration in
 511          * SCHEMA_COMPAT_WRITE_OLD mode. See RevisionStore::emulateContentId for more information.
 512          *
 513          * @return int
 514          */
 515         public function getContentId() {
 516                 return $this->getIntField( 'slot_content_id' );
 517         }
 518
 519         /**
 520          * Returns the content size
 521          *
 522          * @return int size of the content, in bogo-bytes, as reported by Content::getSize.
 523          */
 524         public function getSize() {
 525                 try {
 526                         $size = $this->getIntField( 'content_size' );
 527                 } catch ( IncompleteRevisionException $ex ) {
 528                         $size = $this->getContent()->getSize();
 529                         $this->setField( 'content_size', $size );
 530                 }
 531
 532                 return $size;
 533         }
 534
 535         /**
 536          * Returns the content size
 537          *
 538          * @return string hash of the content.
 539          */
 540         public function getSha1() {
 541                 try {
 542                         $sha1 = $this->getStringField( 'content_sha1' );
 543                 } catch ( IncompleteRevisionException $ex ) {
 544                         $format = $this->hasField( 'format_name' )
 545                                 ? $this->getStringField( 'format_name' )
 546                                 : null;
 547
 548                         $data = $this->getContent()->serialize( $format );
 549                         $sha1 = self::base36Sha1( $data );
 550                         $this->setField( 'content_sha1', $sha1 );
 551                 }
 552
 553                 return $sha1;
 554         }
 555
 556         /**
 557          * Returns the content model. This is the model name that decides
 558          * which ContentHandler is appropriate for interpreting the
 559          * data of the blob referenced by the address returned by getAddress().
 560          *
 561          * @return string the content model of the content
 562          */
 563         public function getModel() {
 564                 try {
 565                         $model = $this->getStringField( 'model_name' );
 566                 } catch ( IncompleteRevisionException $ex ) {
 567                         $model = $this->getContent()->getModel();
 568                         $this->setField( 'model_name', $model );
 569                 }
 570
 571                 return $model;
 572         }
 573
 574         /**
 575          * Returns the blob serialization format as a MIME type.
 576          *
 577          * @note When this method returns null, the caller is expected
 578          * to auto-detect the serialization format, or to rely on
 579          * the default format associated with the content model.
 580          *
 581          * @return string|null
 582          */
 583         public function getFormat() {
 584                 // XXX: we currently do not plan to store the format for each slot!
 585
 586                 if ( $this->hasField( 'format_name' ) ) {
 587                         return $this->getStringField( 'format_name' );
 588                 }
 589
 590                 return null;
 591         }
 592
 593         /**
 594          * @param string $name
 595          * @param string|int|null $value
 596          */
 597         private function setField( $name, $value ) {
 598                 $this->row->$name = $value;
 599         }
 600
 601         /**
 602          * Get the base 36 SHA-1 value for a string of text
 603          *
 604          * MCR migration note: this replaces Revision::base36Sha1
 605          *
 606          * @param string $blob
 607          * @return string
 608          */
 609         public static function base36Sha1( $blob ) {
 610                 return \Wikimedia\base_convert( sha1( $blob ), 16, 36, 31 );
 611         }
 612
 613         /**
 614          * Returns true if $other has the same content as this slot.
 615          * The check is performed based on the model, address size, and hash.
 616          * Two slots can have the same content if they use different content addresses,
 617          * but if they have the same address and the same model, they have the same content.
 618          * Two slots can have the same content if they belong to different
 619          * revisions or pages.
 620          *
 621          * Note that hasSameContent() may return false even if Content::equals returns true for
 622          * the content of two slots. This may happen if the two slots have different serializations
 623          * representing equivalent Content. Such false negatives are considered acceptable. Code
 624          * that has to be absolutely sure the Content is really not the same if hasSameContent()
 625          * returns false should call getContent() and compare the Content objects directly.
 626          *
 627          * @since 1.32
 628          *
 629          * @param SlotRecord $other
 630          * @return bool
 631          */
 632         public function hasSameContent( SlotRecord $other ) {
 633                 if ( $other === $this ) {
 634                         return true;
 635                 }
 636
 637                 if ( $this->getModel() !== $other->getModel() ) {
 638                         return false;
 639                 }
 640
 641                 if ( $this->hasAddress()
 642                         && $other->hasAddress()
 643                         && $this->getAddress() == $other->getAddress()
 644                 ) {
 645                         return true;
 646                 }
 647
 648                 if ( $this->getSize() !== $other->getSize() ) {
 649                         return false;
 650                 }
 651
 652                 if ( $this->getSha1() !== $other->getSha1() ) {
 653                         return false;
 654                 }
 655
 656                 return true;
 657         }
 658
 659 }
 660
 661 /**
 662  * Retain the old class name for backwards compatibility.
 663  * @deprecated since 1.32
 664  */
 665 class_alias( SlotRecord::class, 'MediaWiki\Storage\SlotRecord' );