includes/CommentStore.php

   1 <?php
   2 /**
   3  * Manage storage of comments in the database
   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 use Wikimedia\Rdbms\IDatabase;
  24
  25 /**
  26  * CommentStore handles storage of comments (edit summaries, log reasons, etc)
  27  * in the database.
  28  * @since 1.30
  29  */
  30 class CommentStore {
  31
  32         /**
  33          * Define fields that use temporary tables for transitional purposes
  34          * @var array Keys are '$key', values are arrays with four fields:
  35          *  - table: Temporary table name
  36          *  - pk: Temporary table column referring to the main table's primary key
  37          *  - field: Temporary table column referring comment.comment_id
  38          *  - joinPK: Main table's primary key
  39          */
  40         protected static $tempTables = [
  41                 'rev_comment' => [
  42                         'table' => 'revision_comment_temp',
  43                         'pk' => 'revcomment_rev',
  44                         'field' => 'revcomment_comment_id',
  45                         'joinPK' => 'rev_id',
  46                 ],
  47                 'img_description' => [
  48                         'table' => 'image_comment_temp',
  49                         'pk' => 'imgcomment_name',
  50                         'field' => 'imgcomment_description_id',
  51                         'joinPK' => 'img_name',
  52                 ],
  53         ];
  54
  55         /**
  56          * Fields that formerly used $tempTables
  57          * @var array Key is '$key', value is the MediaWiki version in which it was
  58          *  removed from $tempTables.
  59          */
  60         protected static $formerTempTables = [];
  61
  62         /** @var string */
  63         protected $key;
  64
  65         /** @var int One of the MIGRATION_* constants */
  66         protected $stage;
  67
  68         /** @var array|null Cache for `self::getJoin()` */
  69         protected $joinCache = null;
  70
  71         /**
  72          * @param string $key A key such as "rev_comment" identifying the comment
  73          *  field being fetched.
  74          */
  75         public function __construct( $key ) {
  76                 global $wgCommentTableSchemaMigrationStage;
  77
  78                 $this->key = $key;
  79                 $this->stage = $wgCommentTableSchemaMigrationStage;
  80         }
  81
  82         /**
  83          * Static constructor for easier chaining
  84          * @param string $key A key such as "rev_comment" identifying the comment
  85          *  field being fetched.
  86          * @return CommentStore
  87          */
  88         public static function newKey( $key ) {
  89                 return new CommentStore( $key );
  90         }
  91
  92         /**
  93          * Get SELECT fields for the comment key
  94          *
  95          * Each resulting row should be passed to `self::getCommentLegacy()` to get the
  96          * actual comment.
  97          *
  98          * @note Use of this method may require a subsequent database query to
  99          *  actually fetch the comment. If possible, use `self::getJoin()` instead.
 100          * @return string[] to include in the `$vars` to `IDatabase->select()`. All
 101          *  fields are aliased, so `+` is safe to use.
 102          */
 103         public function getFields() {
 104                 $fields = [];
 105                 if ( $this->stage === MIGRATION_OLD ) {
 106                         $fields["{$this->key}_text"] = $this->key;
 107                         $fields["{$this->key}_data"] = 'NULL';
 108                         $fields["{$this->key}_cid"] = 'NULL';
 109                 } else {
 110                         if ( $this->stage < MIGRATION_NEW ) {
 111                                 $fields["{$this->key}_old"] = $this->key;
 112                         }
 113                         if ( isset( self::$tempTables[$this->key] ) ) {
 114                                 $fields["{$this->key}_pk"] = self::$tempTables[$this->key]['joinPK'];
 115                         } else {
 116                                 $fields["{$this->key}_id"] = "{$this->key}_id";
 117                         }
 118                 }
 119                 return $fields;
 120         }
 121
 122         /**
 123          * Get SELECT fields and joins for the comment key
 124          *
 125          * Each resulting row should be passed to `self::getComment()` to get the
 126          * actual comment.
 127          *
 128          * @return array With three keys:
 129          *   - tables: (string[]) to include in the `$table` to `IDatabase->select()`
 130          *   - fields: (string[]) to include in the `$vars` to `IDatabase->select()`
 131          *   - joins: (array) to include in the `$join_conds` to `IDatabase->select()`
 132          *  All tables, fields, and joins are aliased, so `+` is safe to use.
 133          */
 134         public function getJoin() {
 135                 if ( $this->joinCache === null ) {
 136                         $tables = [];
 137                         $fields = [];
 138                         $joins = [];
 139
 140                         if ( $this->stage === MIGRATION_OLD ) {
 141                                 $fields["{$this->key}_text"] = $this->key;
 142                                 $fields["{$this->key}_data"] = 'NULL';
 143                                 $fields["{$this->key}_cid"] = 'NULL';
 144                         } else {
 145                                 $join = $this->stage === MIGRATION_NEW ? 'JOIN' : 'LEFT JOIN';
 146
 147                                 if ( isset( self::$tempTables[$this->key] ) ) {
 148                                         $t = self::$tempTables[$this->key];
 149                                         $alias = "temp_$this->key";
 150                                         $tables[$alias] = $t['table'];
 151                                         $joins[$alias] = [ $join, "{$alias}.{$t['pk']} = {$t['joinPK']}" ];
 152                                         $joinField = "{$alias}.{$t['field']}";
 153                                 } else {
 154                                         $joinField = "{$this->key}_id";
 155                                 }
 156
 157                                 $alias = "comment_$this->key";
 158                                 $tables[$alias] = 'comment';
 159                                 $joins[$alias] = [ $join, "{$alias}.comment_id = {$joinField}" ];
 160
 161                                 if ( $this->stage === MIGRATION_NEW ) {
 162                                         $fields["{$this->key}_text"] = "{$alias}.comment_text";
 163                                 } else {
 164                                         $fields["{$this->key}_text"] = "COALESCE( {$alias}.comment_text, $this->key )";
 165                                 }
 166                                 $fields["{$this->key}_data"] = "{$alias}.comment_data";
 167                                 $fields["{$this->key}_cid"] = "{$alias}.comment_id";
 168                         }
 169
 170                         $this->joinCache = [
 171                                 'tables' => $tables,
 172                                 'fields' => $fields,
 173                                 'joins' => $joins,
 174                         ];
 175                 }
 176
 177                 return $this->joinCache;
 178         }
 179
 180         /**
 181          * Extract the comment from a row
 182          *
 183          * Shared implementation for getComment() and getCommentLegacy()
 184          *
 185          * @param IDatabase|null $db Database handle for getCommentLegacy(), or null for getComment()
 186          * @param object|array $row
 187          * @param bool $fallback
 188          * @return CommentStoreComment
 189          */
 190         private function getCommentInternal( IDatabase $db = null, $row, $fallback = false ) {
 191                 $key = $this->key;
 192                 $row = (array)$row;
 193                 if ( array_key_exists( "{$key}_text", $row ) && array_key_exists( "{$key}_data", $row ) ) {
 194                         $cid = isset( $row["{$key}_cid"] ) ? $row["{$key}_cid"] : null;
 195                         $text = $row["{$key}_text"];
 196                         $data = $row["{$key}_data"];
 197                 } elseif ( $this->stage === MIGRATION_OLD ) {
 198                         $cid = null;
 199                         if ( $fallback && isset( $row[$key] ) ) {
 200                                 wfLogWarning( "Using deprecated fallback handling for comment $key" );
 201                                 $text = $row[$key];
 202                         } else {
 203                                 wfLogWarning( "Missing {$key}_text and {$key}_data fields in row with MIGRATION_OLD" );
 204                                 $text = '';
 205                         }
 206                         $data = null;
 207                 } else {
 208                         if ( isset( self::$tempTables[$key] ) ) {
 209                                 if ( array_key_exists( "{$key}_pk", $row ) ) {
 210                                         if ( !$db ) {
 211                                                 throw new InvalidArgumentException(
 212                                                         "\$row does not contain fields needed for comment $key and getComment(), but "
 213                                                         . "does have fields for getCommentLegacy()"
 214                                                 );
 215                                         }
 216                                         $t = self::$tempTables[$key];
 217                                         $id = $row["{$key}_pk"];
 218                                         $row2 = $db->selectRow(
 219                                                 [ $t['table'], 'comment' ],
 220                                                 [ 'comment_id', 'comment_text', 'comment_data' ],
 221                                                 [ $t['pk'] => $id ],
 222                                                 __METHOD__,
 223                                                 [],
 224                                                 [ 'comment' => [ 'JOIN', [ "comment_id = {$t['field']}" ] ] ]
 225                                         );
 226                                 } elseif ( $fallback && isset( $row[$key] ) ) {
 227                                         wfLogWarning( "Using deprecated fallback handling for comment $key" );
 228                                         $row2 = (object)[ 'comment_text' => $row[$key], 'comment_data' => null ];
 229                                 } else {
 230                                         throw new InvalidArgumentException( "\$row does not contain fields needed for comment $key" );
 231                                 }
 232                         } else {
 233                                 if ( array_key_exists( "{$key}_id", $row ) ) {
 234                                         if ( !$db ) {
 235                                                 throw new InvalidArgumentException(
 236                                                         "\$row does not contain fields needed for comment $key and getComment(), but "
 237                                                         . "does have fields for getCommentLegacy()"
 238                                                 );
 239                                         }
 240                                         $id = $row["{$key}_id"];
 241                                         $row2 = $db->selectRow(
 242                                                 'comment',
 243                                                 [ 'comment_id', 'comment_text', 'comment_data' ],
 244                                                 [ 'comment_id' => $id ],
 245                                                 __METHOD__
 246                                         );
 247                                 } elseif ( $fallback && isset( $row[$key] ) ) {
 248                                         wfLogWarning( "Using deprecated fallback handling for comment $key" );
 249                                         $row2 = (object)[ 'comment_text' => $row[$key], 'comment_data' => null ];
 250                                 } else {
 251                                         throw new InvalidArgumentException( "\$row does not contain fields needed for comment $key" );
 252                                 }
 253                         }
 254
 255                         if ( $row2 ) {
 256                                 $cid = $row2->comment_id;
 257                                 $text = $row2->comment_text;
 258                                 $data = $row2->comment_data;
 259                         } elseif ( $this->stage < MIGRATION_NEW && array_key_exists( "{$key}_old", $row ) ) {
 260                                 $cid = null;
 261                                 $text = $row["{$key}_old"];
 262                                 $data = null;
 263                         } else {
 264                                 // @codeCoverageIgnoreStart
 265                                 wfLogWarning( "Missing comment row for $key, id=$id" );
 266                                 $cid = null;
 267                                 $text = '';
 268                                 $data = null;
 269                                 // @codeCoverageIgnoreEnd
 270                         }
 271                 }
 272
 273                 $msg = null;
 274                 if ( $data !== null ) {
 275                         $data = FormatJson::decode( $data );
 276                         if ( !is_object( $data ) ) {
 277                                 // @codeCoverageIgnoreStart
 278                                 wfLogWarning( "Invalid JSON object in comment: $data" );
 279                                 $data = null;
 280                                 // @codeCoverageIgnoreEnd
 281                         } else {
 282                                 $data = (array)$data;
 283                                 if ( isset( $data['_message'] ) ) {
 284                                         $msg = self::decodeMessage( $data['_message'] )
 285                                                 ->setInterfaceMessageFlag( true );
 286                                 }
 287                                 if ( !empty( $data['_null'] ) ) {
 288                                         $data = null;
 289                                 } else {
 290                                         foreach ( $data as $k => $v ) {
 291                                                 if ( substr( $k, 0, 1 ) === '_' ) {
 292                                                         unset( $data[$k] );
 293                                                 }
 294                                         }
 295                                 }
 296                         }
 297                 }
 298
 299                 return new CommentStoreComment( $cid, $text, $msg, $data );
 300         }
 301
 302         /**
 303          * Extract the comment from a row
 304          *
 305          * Use `self::getJoin()` to ensure the row contains the needed data.
 306          *
 307          * If you need to fake a comment in a row for some reason, set fields
 308          * `{$key}_text` (string) and `{$key}_data` (JSON string or null).
 309          *
 310          * @param object|array $row Result row.
 311          * @param bool $fallback If true, fall back as well as possible instead of throwing an exception.
 312          * @return CommentStoreComment
 313          */
 314         public function getComment( $row, $fallback = false ) {
 315                 return $this->getCommentInternal( null, $row, $fallback );
 316         }
 317
 318         /**
 319          * Extract the comment from a row, with legacy lookups.
 320          *
 321          * If `$row` might have been generated using `self::getFields()` rather
 322          * than `self::getJoin()`, use this. Prefer `self::getComment()` if you
 323          * know callers used `self::getJoin()` for the row fetch.
 324          *
 325          * If you need to fake a comment in a row for some reason, set fields
 326          * `{$key}_text` (string) and `{$key}_data` (JSON string or null).
 327          *
 328          * @param IDatabase $db Database handle to use for lookup
 329          * @param object|array $row Result row.
 330          * @param bool $fallback If true, fall back as well as possible instead of throwing an exception.
 331          * @return CommentStoreComment
 332          */
 333         public function getCommentLegacy( IDatabase $db, $row, $fallback = false ) {
 334                 return $this->getCommentInternal( $db, $row, $fallback );
 335         }
 336
 337         /**
 338          * Create a new CommentStoreComment, inserting it into the database if necessary
 339          *
 340          * If a comment is going to be passed to `self::insert()` or the like
 341          * multiple times, it will be more efficient to pass a CommentStoreComment
 342          * once rather than making `self::insert()` do it every time through.
 343          *
 344          * @note When passing a CommentStoreComment, this may set `$comment->id` if
 345          *  it's not already set. If `$comment->id` is already set, it will not be
 346          *  verified that the specified comment actually exists or that it
 347          *  corresponds to the comment text, message, and/or data in the
 348          *  CommentStoreComment.
 349          * @param IDatabase $dbw Database handle to insert on. Unused if `$comment`
 350          *  is a CommentStoreComment and `$comment->id` is set.
 351          * @param string|Message|CommentStoreComment $comment Comment text or Message object, or
 352          *  a CommentStoreComment.
 353          * @param array|null $data Structured data to store. Keys beginning with '_' are reserved.
 354          *  Ignored if $comment is a CommentStoreComment.
 355          * @return CommentStoreComment
 356          */
 357         public function createComment( IDatabase $dbw, $comment, array $data = null ) {
 358                 global $wgContLang;
 359
 360                 if ( !$comment instanceof CommentStoreComment ) {
 361                         if ( $data !== null ) {
 362                                 foreach ( $data as $k => $v ) {
 363                                         if ( substr( $k, 0, 1 ) === '_' ) {
 364                                                 throw new InvalidArgumentException( 'Keys in $data beginning with "_" are reserved' );
 365                                         }
 366                                 }
 367                         }
 368                         if ( $comment instanceof Message ) {
 369                                 $message = clone $comment;
 370                                 $text = $message->inLanguage( $wgContLang ) // Avoid $wgForceUIMsgAsContentMsg
 371                                         ->setInterfaceMessageFlag( true )
 372                                         ->text();
 373                                 $comment = new CommentStoreComment( null, $text, $message, $data );
 374                         } else {
 375                                 $comment = new CommentStoreComment( null, $comment, null, $data );
 376                         }
 377                 }
 378
 379                 if ( $this->stage > MIGRATION_OLD && !$comment->id ) {
 380                         $dbData = $comment->data;
 381                         if ( !$comment->message instanceof RawMessage ) {
 382                                 if ( $dbData === null ) {
 383                                         $dbData = [ '_null' => true ];
 384                                 }
 385                                 $dbData['_message'] = self::encodeMessage( $comment->message );
 386                         }
 387                         if ( $dbData !== null ) {
 388                                 $dbData = FormatJson::encode( (object)$dbData, false, FormatJson::ALL_OK );
 389                         }
 390
 391                         $hash = self::hash( $comment->text, $dbData );
 392                         $comment->id = $dbw->selectField(
 393                                 'comment',
 394                                 'comment_id',
 395                                 [
 396                                         'comment_hash' => $hash,
 397                                         'comment_text' => $comment->text,
 398                                         'comment_data' => $dbData,
 399                                 ],
 400                                 __METHOD__
 401                         );
 402                         if ( !$comment->id ) {
 403                                 $comment->id = $dbw->nextSequenceValue( 'comment_comment_id_seq' );
 404                                 $dbw->insert(
 405                                         'comment',
 406                                         [
 407                                                 'comment_id' => $comment->id,
 408                                                 'comment_hash' => $hash,
 409                                                 'comment_text' => $comment->text,
 410                                                 'comment_data' => $dbData,
 411                                         ],
 412                                         __METHOD__
 413                                 );
 414                                 $comment->id = $dbw->insertId();
 415                         }
 416                 }
 417
 418                 return $comment;
 419         }
 420
 421         /**
 422          * Implementation for `self::insert()` and `self::insertWithTempTable()`
 423          * @param IDatabase $dbw
 424          * @param string|Message|CommentStoreComment $comment
 425          * @param array|null $data
 426          * @return array [ array $fields, callable $callback ]
 427          */
 428         private function insertInternal( IDatabase $dbw, $comment, $data ) {
 429                 $fields = [];
 430                 $callback = null;
 431
 432                 $comment = $this->createComment( $dbw, $comment, $data );
 433
 434                 if ( $this->stage <= MIGRATION_WRITE_BOTH ) {
 435                         $fields[$this->key] = $comment->text;
 436                 }
 437
 438                 if ( $this->stage >= MIGRATION_WRITE_BOTH ) {
 439                         if ( isset( self::$tempTables[$this->key] ) ) {
 440                                 $t = self::$tempTables[$this->key];
 441                                 $func = __METHOD__;
 442                                 $commentId = $comment->id;
 443                                 $callback = function ( $id ) use ( $dbw, $commentId, $t, $func ) {
 444                                         $dbw->insert(
 445                                                 $t['table'],
 446                                                 [
 447                                                         $t['pk'] => $id,
 448                                                         $t['field'] => $commentId,
 449                                                 ],
 450                                                 $func
 451                                         );
 452                                 };
 453                         } else {
 454                                 $fields["{$this->key}_id"] = $comment->id;
 455                         }
 456                 }
 457
 458                 return [ $fields, $callback ];
 459         }
 460
 461         /**
 462          * Prepare for the insertion of a row with a comment
 463          *
 464          * @note It's recommended to include both the call to this method and the
 465          *  row insert in the same transaction.
 466          * @param IDatabase $dbw Database handle to insert on
 467          * @param string|Message|CommentStoreComment $comment As for `self::createComment()`
 468          * @param array|null $data As for `self::createComment()`
 469          * @return array Fields for the insert or update
 470          */
 471         public function insert( IDatabase $dbw, $comment, $data = null ) {
 472                 if ( isset( self::$tempTables[$this->key] ) ) {
 473                         throw new InvalidArgumentException( "Must use insertWithTempTable() for $this->key" );
 474                 }
 475
 476                 list( $fields ) = $this->insertInternal( $dbw, $comment, $data );
 477                 return $fields;
 478         }
 479
 480         /**
 481          * Prepare for the insertion of a row with a comment and temporary table
 482          *
 483          * This is currently needed for "rev_comment" and "img_description". In the
 484          * future that requirement will be removed.
 485          *
 486          * @note It's recommended to include both the call to this method and the
 487          *  row insert in the same transaction.
 488          * @param IDatabase $dbw Database handle to insert on
 489          * @param string|Message|CommentStoreComment $comment As for `self::createComment()`
 490          * @param array|null $data As for `self::createComment()`
 491          * @return array Two values:
 492          *  - array Fields for the insert or update
 493          *  - callable Function to call when the primary key of the row being
 494          *    inserted/updated is known. Pass it that primary key.
 495          */
 496         public function insertWithTempTable( IDatabase $dbw, $comment, $data = null ) {
 497                 if ( isset( self::$formerTempTables[$this->key] ) ) {
 498                         wfDeprecated( __METHOD__ . " for $this->key", self::$formerTempTables[$this->key] );
 499                 } elseif ( !isset( self::$tempTables[$this->key] ) ) {
 500                         throw new InvalidArgumentException( "Must use insert() for $this->key" );
 501                 }
 502
 503                 list( $fields, $callback ) = $this->insertInternal( $dbw, $comment, $data );
 504                 if ( !$callback ) {
 505                         $callback = function () {
 506                                 // Do nothing.
 507                         };
 508                 }
 509                 return [ $fields, $callback ];
 510         }
 511
 512         /**
 513          * Encode a Message as a PHP data structure
 514          * @param Message $msg
 515          * @return array
 516          */
 517         protected static function encodeMessage( Message $msg ) {
 518                 $key = count( $msg->getKeysToTry() ) > 1 ? $msg->getKeysToTry() : $msg->getKey();
 519                 $params = $msg->getParams();
 520                 foreach ( $params as &$param ) {
 521                         if ( $param instanceof Message ) {
 522                                 $param = [
 523                                         'message' => self::encodeMessage( $param )
 524                                 ];
 525                         }
 526                 }
 527                 array_unshift( $params, $key );
 528                 return $params;
 529         }
 530
 531         /**
 532          * Decode a message that was encoded by self::encodeMessage()
 533          * @param array $data
 534          * @return Message
 535          */
 536         protected static function decodeMessage( $data ) {
 537                 $key = array_shift( $data );
 538                 foreach ( $data as &$param ) {
 539                         if ( is_object( $param ) ) {
 540                                 $param = (array)$param;
 541                         }
 542                         if ( is_array( $param ) && count( $param ) === 1 && isset( $param['message'] ) ) {
 543                                 $param = self::decodeMessage( $param['message'] );
 544                         }
 545                 }
 546                 return new Message( $key, $data );
 547         }
 548
 549         /**
 550          * Hashing function for comment storage
 551          * @param string $text Comment text
 552          * @param string|null $data Comment data
 553          * @return int 32-bit signed integer
 554          */
 555         public static function hash( $text, $data ) {
 556                 $hash = crc32( $text ) ^ crc32( (string)$data );
 557
 558                 // 64-bit PHP returns an unsigned CRC, change it to signed for
 559                 // insertion into the database.
 560                 if ( $hash >= 0x80000000 ) {
 561                         $hash |= -1 << 32;
 562                 }
 563
 564                 return $hash;
 565         }
 566
 567 }