Handle records as entities

opencultureconsulting · Jan 6, 2024 · ec7f72c · ec7f72c
1 parent 1e27542
commit ec7f72c
Show file tree

Hide file tree

Showing 3 changed files with 49 additions and 42 deletions.
diff --git a/src/Console/CsvImportCommand.php b/src/Console/CsvImportCommand.php
@@ -95,6 +95,13 @@ function (): array {
             'Name of the CSV column which holds the records\' sets list.',
             'sets'
         );
+        $this->addOption(
+            'noValidation',
+            null,
+            InputOption::VALUE_NONE,
+            'Omit content validation (improves performance for large record sets).',
+            false
+        );
         parent::configure();
     }
 
@@ -117,6 +124,8 @@ protected function execute(InputInterface $input, OutputInterface $output): int
         $arguments = $input->getArguments();
         /** @var Format */
         $format = Database::getInstance()->getEntityManager()->getReference(Format::class, $arguments['format']);
+        /** @var bool */
+        $noValidation = $input->getOption('noValidation');
         /** @var resource */
         $file = fopen($arguments['file'], 'r');
 
@@ -129,23 +138,25 @@ protected function execute(InputInterface $input, OutputInterface $output): int
         $progressIndicator = new ProgressIndicator($output, 'verbose', 200, ['⠏', '⠛', '⠹', '⢸', '⣰', '⣤', '⣆', '⡇']);
         $progressIndicator->start('Importing...');
 
-        while ($record = fgetcsv($file)) {
-            Database::getInstance()->addOrUpdateRecord(
-                $record[$columns['idColumn']],
+        while ($row = fgetcsv($file)) {
+            $record = new Record(
+                $row[$columns['idColumn']],
                 $format,
-                trim($record[$columns['contentColumn']]),
-                new DateTime($record[$columns['dateColumn']] ?? 'now'),
-                // TODO: Complete support for sets.
-                /* $record[$columns['setColumn']] ?? */ null,
-                true
+                null,
+                new DateTime($row[$columns['dateColumn']] ?? 'now')
             );
+            if (strlen(trim($row[$columns['contentColumn']])) > 0) {
+                $record->setContent($row[$columns['contentColumn']], !$noValidation);
+            }
+            // TODO: Complete support for sets.
+            Database::getInstance()->addOrUpdateRecord($record, true);
 
             ++$count;
             $progressIndicator->advance();
             $progressIndicator->setMessage((string) $count . ' done.');
 
             // Flush to database if memory usage reaches 90% of available limit.
-            if (memory_get_usage() / $memoryLimit > 0.9) {
+            if ((memory_get_usage() / $memoryLimit) > 0.9) {
                 Database::getInstance()->flush([Record::class]);
             }
         }

diff --git a/src/Database.php b/src/Database.php
@@ -92,46 +92,32 @@ public function addOrUpdateMetadataFormat(Format $newFormat): void
     /**
      * Add or update record.
      *
-     * @param string $identifier The record identifier
-     * @param Format $format The metadata prefix
-     * @param ?string $data The record's content
-     * @param ?DateTime $lastChanged The date of last change
-     * @param ?array<string, Set> $sets The record's associated sets
+     * @param Record $newRecord The record
      * @param bool $bulkMode Should we operate in bulk mode (no flush)?
      *
      * @return void
      */
-    public function addOrUpdateRecord(
-        string $identifier,
-        Format $format,
-        ?string $data = null,
-        ?DateTime $lastChanged = null,
-        // TODO: Complete support for sets
-        ?array $sets,
-        bool $bulkMode = false
-    ): void
+    public function addOrUpdateRecord(Record $newRecord, bool $bulkMode = false): void
     {
-        $record = $this->entityManager->find(Record::class, ['identifier' => $identifier, 'format' => $format]);
-        if (!isset($data) && Configuration::getInstance()->deletedRecords === 'no') {
-            if (isset($record)) {
-                $this->entityManager->remove($record);
+        $oldRecord = $this->entityManager->find(
+            Record::class,
+            [
+                'identifier' => $newRecord->getIdentifier(),
+                'format' => $newRecord->getFormat()
+            ]
+        );
+        if (isset($oldRecord)) {
+            if ($newRecord->hasContent() || Configuration::getInstance()->deletedRecords !== 'no') {
+                $oldRecord->setContent($newRecord->getContent(), false);
+                $oldRecord->setLastChanged($newRecord->getLastChanged());
+                // TODO: Add full set support.
+            } else {
+                $this->entityManager->remove($oldRecord);
             }
         } else {
-            if (isset($record)) {
-                try {
-                    $record->setContent($data);
-                    $record->setLastChanged($lastChanged);
-                } catch (ValidationFailedException $exception) {
-                    throw $exception;
-                }
-            } else {
-                try {
-                    $record = new Record($identifier, $format, $data, $lastChanged);
-                } catch (ValidationFailedException $exception) {
-                    throw $exception;
-                }
+            if ($newRecord->hasContent() || Configuration::getInstance()->deletedRecords !== 'no') {
+                $this->entityManager->persist($newRecord);
             }
-            $this->entityManager->persist($record);
         }
         if (!$bulkMode) {
             $this->entityManager->flush();

diff --git a/src/Database/Record.php b/src/Database/Record.php
@@ -156,6 +156,16 @@ public function getSets(): array
         return $this->sets->toArray();
     }
 
+    /**
+     * Whether this record has any content.
+     *
+     * @return bool TRUE if content exists, FALSE otherwise
+     */
+    public function hasContent(): bool
+    {
+        return isset($this->content);
+    }
+
     /**
      * Remove record from set.
      *
@@ -184,6 +194,7 @@ public function removeSet(Set $set): void
     public function setContent(?string $data = null, bool $validate = true): void
     {
         if (isset($data)) {
+            $data = trim($data);
             if ($validate) {
                 try {
                     $data = $this->validate($data);
@@ -233,7 +244,6 @@ public function setLastChanged(?DateTime $dateTime = null): void
      */
     protected function validate(string $xml): string
     {
-        $xml = trim($xml);
         $validator = Validation::createValidator();
         $violations = $validator->validate(
             $xml,