ExportProfileData_Background::exportXml()

Complexity

Conditions	49
Paths	> 20000

Size

Total Lines	304
Code Lines	168

Duplication

Lines	0
Ratio	0 %

Importance

Changes	15
Bugs	0	Features	0

<?php
	protected function exportXml($member_info)
	{
		global $smcFunc, $sourcedir, $context, $modSettings, $settings, $user_info, $mbname;
		global $user_profile, $txt, $scripturl, $query_this_board;

		// For convenience...
		$uid = $this->_details['uid'];
		$lang = $this->_details['lang'];
		$included = $this->_details['included'];
		$start = $this->_details['start'];
		$latest = $this->_details['latest'];
		$datatype = $this->_details['datatype'];

		if (!isset($included[$datatype]['func']) || !isset($included[$datatype]['langfile']))
			return;

		require_once($sourcedir . DIRECTORY_SEPARATOR . 'News.php');
		require_once($sourcedir . DIRECTORY_SEPARATOR . 'ScheduledTasks.php');

		// Setup.
		$done = false;
		$delay = 0;
		$func = $included[$datatype]['func'];

The assignment to $func is dead and can be removed.

		$context['xmlnews_uid'] = $uid;
		$context['xmlnews_limit'] = !empty($modSettings['export_rate']) ? $modSettings['export_rate'] : 250;
		$context[$datatype . '_start'] = $start[$datatype];
		$datatypes = array_keys($included);

		// Fake a wee bit of $user_info so that loading the member data & language doesn't choke.
		$user_info = $member_info;

		loadEssentialThemeData();
		$settings['actual_theme_dir'] = $settings['theme_dir'];
		$context['user']['id'] = $uid;
		$context['user']['language'] = $lang;
		loadMemberData($uid);
		loadLanguage(implode('+', array_unique(array('index', 'Modifications', 'Stats', 'Profile', $included[$datatype]['langfile']))), $lang);

		// @todo Ask lawyers whether the GDPR requires us to include posts in the recycle bin.
		$query_this_board = '{query_see_message_board}' . (!empty($modSettings['recycle_enable']) && $modSettings['recycle_board'] > 0 ? ' AND m.id_board != ' . $modSettings['recycle_board'] : '');

		// We need a valid export directory.
		if (empty($modSettings['export_dir']) || !file_exists($modSettings['export_dir']))
		{
			require_once($sourcedir . DIRECTORY_SEPARATOR . 'Profile-Export.php');
			if (create_export_dir() === false)
				return;
		}

		$export_dir_slash = $modSettings['export_dir'] . DIRECTORY_SEPARATOR;

		$idhash = hash_hmac('sha1', $uid, get_auth_secret());
		$idhash_ext = $idhash . '.' . $this->_details['format_settings']['extension'];

		// Increment the file number until we reach one that doesn't exist.
		$filenum = 1;
		$realfile = $export_dir_slash . $filenum . '_' . $idhash_ext;
		while (file_exists($realfile))
			$realfile = $export_dir_slash . ++$filenum . '_' . $idhash_ext;

		$tempfile = $export_dir_slash . $idhash_ext . '.tmp';
		$progressfile = $export_dir_slash . $idhash_ext . '.progress.json';

		$feed_meta = array(
			'title' => sprintf($txt['profile_of_username'], $user_profile[$uid]['real_name']),
			'desc' => sentence_list(array_map(function ($datatype) use ($txt) { return $txt[$datatype]; }, array_keys($included))),
			'author' => $mbname,
			'source' => $scripturl . '?action=profile;u=' . $uid,
			'self' => '', // Unused, but can't be null.
			'page' => &$filenum,
		);

		// Some paranoid hosts disable or hamstring the disk space functions in an attempt at security via obscurity.
		$check_diskspace = !empty($modSettings['export_min_diskspace_pct']) && function_exists('disk_free_space') && function_exists('disk_total_space') && intval(@disk_total_space($modSettings['export_dir']) >= 1440);
		$minspace = $check_diskspace ? ceil(disk_total_space($modSettings['export_dir']) * $modSettings['export_min_diskspace_pct'] / 100) : 0;

		// If a necessary file is missing, we need to start over.
		if (!file_exists($tempfile) || !file_exists($progressfile) || filesize($progressfile) == 0)
		{
			foreach (array_merge(array($tempfile, $progressfile), glob($export_dir_slash . '*_' . $idhash_ext)) as $fpath)

It seems like glob($export_dir_slash . '*_' . $idhash_ext) can also be of type false; however, parameter $array2 of array_merge() does only seem to accept array|null, maybe add an additional type check?

				@unlink($fpath);

			$filenum = 1;
			$realfile = $export_dir_slash . $filenum . '_' . $idhash_ext;

			buildXmlFeed('smf', array(), $feed_meta, 'profile');
			file_put_contents($tempfile, implode('', $context['feed']));

			$progress = array_fill_keys($datatypes, 0);
			file_put_contents($progressfile, $smcFunc['json_encode']($progress));
		}
		else
			$progress = $smcFunc['json_decode'](file_get_contents($progressfile), true);

		// Get the data, always in ascending order.
		$xml_data = call_user_func($included[$datatype]['func'], 'smf', true);

		// No data retrived? Just move on then.
		if (empty($xml_data))
			$datatype_done = true;

		// Basic profile data is quick and easy.
		elseif ($datatype == 'profile')
		{
			buildXmlFeed('smf', $xml_data, $feed_meta, 'profile');
			file_put_contents($tempfile, implode('', $context['feed']));

			$progress[$datatype] = time();
			$datatype_done = true;

			// Cache for subsequent reuse.
			$profile_basic_items = $context['feed']['items'];
			cache_put_data('export_profile_basic-' . $uid, $profile_basic_items, MAX_CLAIM_THRESHOLD);
		}

		// Posts and PMs...
		else
		{
			// We need the basic profile data in every export file.
			$profile_basic_items = cache_get_data('export_profile_basic-' . $uid, MAX_CLAIM_THRESHOLD);
			if (empty($profile_basic_items))
			{
				$profile_data = call_user_func($included['profile']['func'], 'smf', true);
				buildXmlFeed('smf', $profile_data, $feed_meta, 'profile');
				$profile_basic_items = $context['feed']['items'];
				cache_put_data('export_profile_basic-' . $uid, $profile_basic_items, MAX_CLAIM_THRESHOLD);
				unset($context['feed']);
			}

			$per_page = $this->_details['format_settings']['per_page'];
			$prev_item_count = empty($this->_details['item_count']) ? 0 : $this->_details['item_count'];

			// If the temp file has grown enormous, save it so we can start a new one.
			clearstatcache();
			if (file_exists($tempfile) && filesize($tempfile) >= 1024 * 1024 * 250)
			{
				rename($tempfile, $realfile);
				$realfile = $export_dir_slash . ++$filenum . '_' . $idhash_ext;

				if (empty($context['feed']['header']))
					buildXmlFeed('smf', array(), $feed_meta, 'profile');

				file_put_contents($tempfile, implode('', array($context['feed']['header'], $profile_basic_items, $context['feed']['footer'])));

				$prev_item_count = 0;
			}

			// Split $xml_data into reasonably sized chunks.
			if (empty($prev_item_count))
			{
				$xml_data = array_chunk($xml_data, $per_page);
			}
			else
			{
				$first_chunk = array_splice($xml_data, 0, $per_page - $prev_item_count);
				$xml_data = array_merge(array($first_chunk), array_chunk($xml_data, $per_page));
				unset($first_chunk);
			}

			foreach ($xml_data as $chunk => $items)
			{
				unset($new_item_count, $last_id);

				// Remember the last item so we know where to start next time.
				$last_item = end($items);
				if (isset($last_item['content'][0]['content']) && $last_item['content'][0]['tag'] === 'id')
					$last_id = $last_item['content'][0]['content'];

				// Build the XML string from the data.
				buildXmlFeed('smf', $items, $feed_meta, 'profile');

				// If disk space is insufficient, pause for a day so the admin can fix it.
				if ($check_diskspace && disk_free_space($modSettings['export_dir']) - $minspace <= strlen(implode('', $context['feed']) . self::$xslt_info['stylesheet']))
				{
					loadLanguage('Errors');
					log_error(sprintf($txt['export_low_diskspace'], $modSettings['export_min_diskspace_pct']));

					$delay = 86400;
				}
				else
				{
					// We need a file to write to, of course.
					if (!file_exists($tempfile))
						file_put_contents($tempfile, implode('', array($context['feed']['header'], $profile_basic_items, $context['feed']['footer'])));

					// Insert the new data before the feed footer.
					$handle = fopen($tempfile, 'r+');
					if (is_resource($handle))
					{
						fseek($handle, strlen($context['feed']['footer']) * -1, SEEK_END);

						$bytes_written = fwrite($handle, $context['feed']['items'] . $context['feed']['footer']);

						// If we couldn't write everything, revert the changes and consider the write to have failed.
						if ($bytes_written > 0 && $bytes_written < strlen($context['feed']['items'] . $context['feed']['footer']))
						{
							fseek($handle, $bytes_written * -1, SEEK_END);
							$pointer_pos = ftell($handle);
							ftruncate($handle, $pointer_pos);
							rewind($handle);
							fseek($handle, 0, SEEK_END);
							fwrite($handle, $context['feed']['footer']);

							$bytes_written = false;
						}

						fclose($handle);
					}

					// Write failed. We'll try again next time.
					if (empty($bytes_written))
					{
						$delay = MAX_CLAIM_THRESHOLD;
						break;
					}

					// All went well.
					else
					{
						// Track progress by ID where appropriate, and by time otherwise.
						$progress[$datatype] = !isset($last_id) ? time() : $last_id;
						file_put_contents($progressfile, $smcFunc['json_encode']($progress));

						// Are we done with this datatype yet?
						if (!isset($last_id) || (count($items) < $per_page && $last_id >= $latest[$datatype]))
							$datatype_done = true;

						// Finished the file for this chunk, so move on to the next one.
						if (count($items) >= $per_page - $prev_item_count)
						{
							rename($tempfile, $realfile);
							$realfile = $export_dir_slash . ++$filenum . '_' . $idhash_ext;

							file_put_contents($tempfile, implode('', array($context['feed']['header'], $profile_basic_items, $context['feed']['footer'])));

							$prev_item_count = $new_item_count = 0;
						}
						// This was the last chunk.
						else
						{
							// Should we append more items to this file next time?
							$new_item_count = isset($last_id) ? $prev_item_count + count($items) : 0;
						}
					}
				}
			}
		}

		if (!empty($datatype_done))
		{
			$datatype_key = array_search($datatype, $datatypes);
			$done = !isset($datatypes[$datatype_key + 1]);

			if (!$done)
				$datatype = $datatypes[$datatype_key + 1];
		}

		// Remove the .tmp extension from the final tempfile so the system knows it's done.
		if (!empty($done))
		{
			rename($tempfile, $realfile);
		}

		// Oops. Apparently some sneaky monkey cancelled the export while we weren't looking.
		elseif (!file_exists($progressfile))
		{
			@unlink($tempfile);
			return;
		}

		// We have more work to do again later.
		else
		{
			$start[$datatype] = $progress[$datatype];

			$new_details = array(
				'format' => $this->_details['format'],
				'uid' => $uid,
				'lang' => $lang,
				'included' => $included,
				'start' => $start,
				'latest' => $latest,
				'datatype' => $datatype,
				'format_settings' => $this->_details['format_settings'],
				'last_page' => $this->_details['last_page'],
				'dlfilename' => $this->_details['dlfilename'],
			);
			if (!empty($new_item_count))
				$new_details['item_count'] = $new_item_count;

			$smcFunc['db_insert']('insert', '{db_prefix}background_tasks',
				array('task_file' => 'string-255', 'task_class' => 'string-255', 'task_data' => 'string', 'claimed_time' => 'int'),
				array('$sourcedir/tasks/ExportProfileData.php', 'ExportProfileData_Background', $smcFunc['json_encode']($new_details), time() - MAX_CLAIM_THRESHOLD + $delay),
				array()
			);

			if (!file_exists($tempfile))
			{
				buildXmlFeed('smf', array(), $feed_meta, 'profile');
				file_put_contents($tempfile, implode('', array($context['feed']['header'], !empty($profile_basic_items) ? $profile_basic_items : '', $context['feed']['footer'])));
			}
		}

		file_put_contents($progressfile, $smcFunc['json_encode']($progress));
	}
?>