diff --git a/core/modules/migrate/src/Row.php b/core/modules/migrate/src/Row.php index 4be9712..7b3669e 100644 --- a/core/modules/migrate/src/Row.php +++ b/core/modules/migrate/src/Row.php @@ -176,9 +176,12 @@ public function setSourceProperty($property, $data) { /** * Freezes the source. + * + * @return $this */ public function freezeSource() { $this->frozen = TRUE; + return $this; } /** diff --git a/core/modules/migrate/src/Source.php b/core/modules/migrate/src/Source.php index be4368b..7f30557 100644 --- a/core/modules/migrate/src/Source.php +++ b/core/modules/migrate/src/Source.php @@ -32,7 +32,7 @@ class Source implements \Iterator, \Countable { * * @var array */ - protected $currentIds; + protected $currentSourceIds; /** * Number of rows intentionally ignored (prepareRow() returned FALSE) @@ -122,10 +122,10 @@ class Source implements \Iterator, \Countable { protected $highWaterProperty; /** - * Getter for currentIds data member. + * Getter for currentSourceIds data member. */ public function getCurrentIds() { - return $this->currentIds; + return $this->currentSourceIds; } /** @@ -262,7 +262,7 @@ public function current() { * preferable. */ public function key() { - return serialize($this->currentIds); + return serialize($this->currentSourceIds); } /** @@ -297,118 +297,79 @@ public function rewind() { /** * {@inheritdoc} + * + * The migration iterates over a couple of source rows. This method determines + * the next row, which will be processed and imported into the system. + * + * The method tracks the source and destination IDs using the ID map. + * + * The also takes care about highwater support. Highwater allows to reimport + * rows from a previous migration run, which got changed in the meantime. + * This is done by specifying a highwater field, which is compared with the + * last time, the migration got executed (originalHighWater). */ public function next() { - $this->currentIds = NULL; + $this->currentSourceIds = NULL; $this->currentRow = NULL; $source_configuration = $this->migration->get('source'); + // In order to find the next row we want to process, we ask the source + // plugin for the next possible row. while ($this->getIterator()->valid()) { $row_data = $this->getIterator()->current() + $source_configuration; $this->getIterator()->next(); $row = new Row($row_data, $this->migration->getSourcePlugin()->getIds(), $this->migration->get('destinationIds')); // Populate the source key for this row. - $this->currentIds = $row->getSourceIdValues(); + $this->currentSourceIds = $row->getSourceIdValues(); // Pick up the existing map row, if any, unless getNextRow() did it. - if (!$this->mapRowAdded && ($id_map = $this->idMap->getRowBySource($this->currentIds))) { + if (!$this->mapRowAdded && ($id_map = $this->idMap->getRowBySource($this->currentSourceIds))) { $row->setIdMap($id_map); } - // First, determine if this row should be passed to prepareRow(), or - // skipped entirely. The rules are: - // 1. If there's an explicit idlist, that's all we care about (ignore - // high waters and map rows). - $prepared = FALSE; - if (!empty($this->idList)) { - if (in_array(reset($this->currentIds), $this->idList)) { - // In the list, fall through. - } - else { - // Not in the list, skip it. - continue; - } - } - // 2. If the row is not in the map (we have never tried to import it - // before), we always want to try it. - elseif (!$row->getIdMap()) { - // Fall through - } - // 3. If the row is marked as needing update, pass it. - elseif ($row->needsUpdate()) { - // Fall through. - } - // 4. At this point, we have a row which has previously been imported and - // not marked for update. If we're not using high water marks, then we - // will not take this row. Except, if we're looking for changes in the - // data, we need to go through prepareRow() before we can decide to - // skip it. - elseif (!empty($this->highWaterProperty['field'])) { - if ($this->trackChanges) { - if ($this->prepareRow($row) !== FALSE) { - if ($row->changed()) { - // This is a keeper - $this->currentRow = $row; - break; - } - else { - // No change, skip it. - continue; - } - } - else { - // prepareRow() told us to skip it. - continue; - } - } - else { - // No high water and not tracking changes, skip. - continue; - } - } - // 5. The initial high water mark, before anything is migrated, is ''. We - // want to make sure we don't mistakenly skip rows with a high water - // field value of 0, so explicitly handle '' here. - elseif ($this->originalHighWater === '') { - // Fall through + // If there is an explicit ID list, we know that we want to process the + // row. + $id_in_the_list = $this->idList && in_array(reset($this->currentSourceIds), $this->idList); + // In case we have specified an ID list, but the ID given by the source is + // not in there, we skip the row. + if ($this->idList && !$id_in_the_list) { + continue; } - // 6. So, we are using high water marks. Take the row if its high water - // field value is greater than the saved mark, otherwise skip it. - else { - // Call prepareRow() here, in case the highWaterField needs preparation. - if ($this->prepareRow($row) !== FALSE) { - if ($row->getSourceProperty($this->highWaterProperty['name']) > $this->originalHighWater) { - $this->currentRow = $row; - break; - } - else { - // Skip. - continue; - } - } - $prepared = TRUE; - } - - // Allow the Migration to prepare this row. prepareRow() can return - // boolean FALSE to ignore this row. - if (!$prepared) { + if ( + // Allow explicit specified IDs. + $id_in_the_list || + // We haven't imported the row yet, so the ID map is empty. + !$row->getIdMap() || + // All checks below are just applied, if the row got imported before. + + // Rows can be marked as required to be updated (MigrateIdMapInterface::STATUS_NEEDS_UPDATE). + $row->needsUpdate() || + // For migrations with highwater support, + ($this->usesHighWater() && $row->getSourceProperty($this->highWaterProperty['name']) > $this->originalHighWater) || // now let's try to see whether the source has changed, either via highwater which is like node.changed + // Without highwater, we figure out a change by using a hash of the row. + (!$this->usesHighwater() && $this->trackedRowChanged($row)) ) { + // Migrate fires a hook for every row to allow skipping them. On top of + // do general house keeping by preparing the row. if ($this->prepareRow($row) !== FALSE) { - // Finally, we've got a keeper. - $this->currentRow = $row; + // Stop the searching for the next row. + $this->currentRow = $row->freezeSource(); break; } else { + // The preparation failed, let's try the next one. $this->currentRow = NULL; } } } - if ($this->currentRow) { - $this->currentRow->freezeSource(); - } - else { - $this->currentIds = NULL; - } + } + + protected function trackedRowChanged($row) { + return $this->trackChanges && $this->prepareRow($row) !== FALSE && $row->changed(); + } + + protected function usesHighwater() { + return $this->highWaterProperty && $this->originalHighWater !== ''; } /** @@ -426,12 +387,12 @@ protected function prepareRow(Row $row) { // Make sure we replace any previous messages for this item with any // new ones. $id_map = $this->migration->getIdMap(); - $id_map->delete($this->currentIds, TRUE); + $id_map->delete($this->currentSourceIds, TRUE); $this->migrateExecutable->saveQueuedMessages(); $id_map->saveIdMapping($row, array(), MigrateIdMapInterface::STATUS_IGNORED, $this->migrateExecutable->rollbackAction); $this->numIgnored++; $this->currentRow = NULL; - $this->currentIds = NULL; + $this->currentSourceIds = NULL; } else { // When tracking changed data, We want to quietly skip (rather than