AOE\Crawler\Controller\CrawlerController::getUrlsForPageId()

Complexity

Conditions	27
Paths	84

Size

Total Lines	138
Code Lines	72

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	64
CRAP Score	27.2778

Importance

Changes

0

<?php
    protected function getUrlsForPageId($id, $forceSsl = false)
    {

        /**
         * Get configuration from tsConfig
         */

        // Get page TSconfig for page ID:
        $pageTSconfig = $this->getPageTSconfigForId($id);

        $res = [];

        if (is_array($pageTSconfig) && is_array($pageTSconfig['tx_crawler.']['crawlerCfg.'])) {
            $crawlerCfg = $pageTSconfig['tx_crawler.']['crawlerCfg.'];

            if (is_array($crawlerCfg['paramSets.'])) {
                foreach ($crawlerCfg['paramSets.'] as $key => $values) {
                    if (is_array($values)) {
                        $key = str_replace('.', '', $key);
                        // Sub configuration for a single configuration string:
                        $subCfg = (array)$crawlerCfg['paramSets.'][$key . '.'];
                        $subCfg['key'] = $key;

                        if (strcmp($subCfg['procInstrFilter'], '')) {
                            $subCfg['procInstrFilter'] = implode(',', GeneralUtility::trimExplode(',', $subCfg['procInstrFilter']));
                        }
                        $pidOnlyList = implode(',', GeneralUtility::trimExplode(',', $subCfg['pidsOnly'], true));

                        // process configuration if it is not page-specific or if the specific page is the current page:
                        if (!strcmp($subCfg['pidsOnly'], '') || GeneralUtility::inList($pidOnlyList, $id)) {

                                // add trailing slash if not present
                            if (!empty($subCfg['baseUrl']) && substr($subCfg['baseUrl'], -1) != '/') {
                                $subCfg['baseUrl'] .= '/';
                            }

                            // Explode, process etc.:
                            $res[$key] = [];
                            $res[$key]['subCfg'] = $subCfg;
                            $res[$key]['paramParsed'] = $this->parseParams($values);

$values is of type array, but the function expects a string.

                            $res[$key]['paramExpanded'] = $this->expandParameters($res[$key]['paramParsed'], $id);
                            $res[$key]['origin'] = 'pagets';

                            // recognize MP value
                            if (!$this->MP) {
                                $res[$key]['URLs'] = $this->compileUrls($res[$key]['paramExpanded'], ['?id=' . $id]);
                            } else {
                                $res[$key]['URLs'] = $this->compileUrls($res[$key]['paramExpanded'], ['?id=' . $id . '&MP=' . $this->MP]);
                            }
                        }
                    }
                }
            }
        }

        /**
         * Get configuration from tx_crawler_configuration records
         */

        // get records along the rootline
        $rootLine = BackendUtility::BEgetRootLine($id);

        foreach ($rootLine as $page) {
            $configurationRecordsForCurrentPage = BackendUtility::getRecordsByField(
                'tx_crawler_configuration',
                'pid',
                intval($page['uid']),
                BackendUtility::BEenableFields('tx_crawler_configuration') . BackendUtility::deleteClause('tx_crawler_configuration')
            );

            if (is_array($configurationRecordsForCurrentPage)) {
                foreach ($configurationRecordsForCurrentPage as $configurationRecord) {

                        // check access to the configuration record
                    if (empty($configurationRecord['begroups']) || $GLOBALS['BE_USER']->isAdmin() || $this->hasGroupAccess($GLOBALS['BE_USER']->user['usergroup_cached_list'], $configurationRecord['begroups'])) {
                        $pidOnlyList = implode(',', GeneralUtility::trimExplode(',', $configurationRecord['pidsonly'], true));

                        // process configuration if it is not page-specific or if the specific page is the current page:
                        if (!strcmp($configurationRecord['pidsonly'], '') || GeneralUtility::inList($pidOnlyList, $id)) {
                            $key = $configurationRecord['name'];

                            // don't overwrite previously defined paramSets
                            if (!isset($res[$key])) {

                                    /* @var $TSparserObject \TYPO3\CMS\Core\TypoScript\Parser\TypoScriptParser */
                                $TSparserObject = GeneralUtility::makeInstance('TYPO3\CMS\Core\TypoScript\Parser\TypoScriptParser');
                                $TSparserObject->parse($configurationRecord['processing_instruction_parameters_ts']);

                                // Check Crawler Configurations forceSsl if false use the Page url_scheme.
                                $crawlerWithSsl = $configurationRecord['force_ssl'] ?: $forceSsl;

                                $subCfg = [
                                    'procInstrFilter' => $configurationRecord['processing_instruction_filter'],
                                    'procInstrParams.' => $TSparserObject->setup,
                                    'baseUrl' => $this->getBaseUrlForConfigurationRecord(
                                        $configurationRecord['base_url'],
                                        $configurationRecord['sys_domain_base_url'],
                                        $crawlerWithSsl
                                    ),
                                    'realurl' => $configurationRecord['realurl'],
                                    'cHash' => $configurationRecord['chash'],
                                    'userGroups' => $configurationRecord['fegroups'],
                                    'exclude' => $configurationRecord['exclude'],
                                    'rootTemplatePid' => (int) $configurationRecord['root_template_pid'],
                                    'key' => $key,
                                    'force_ssl' => $configurationRecord['force_ssl'],
                                ];

                                // add trailing slash if not present
                                if (!empty($subCfg['baseUrl']) && substr($subCfg['baseUrl'], -1) != '/') {
                                    $subCfg['baseUrl'] .= '/';
                                }
                                if (!in_array($id, $this->expandExcludeString($subCfg['exclude']))) {
                                    $res[$key] = [];
                                    $res[$key]['subCfg'] = $subCfg;
                                    $res[$key]['paramParsed'] = $this->parseParams($configurationRecord['configuration']);
                                    $res[$key]['paramExpanded'] = $this->expandParameters($res[$key]['paramParsed'], $id);
                                    $res[$key]['URLs'] = $this->compileUrls($res[$key]['paramExpanded'], ['?id=' . $id]);
                                    $res[$key]['origin'] = 'tx_crawler_configuration_' . $configurationRecord['uid'];
                                }
                            }
                        }
                    }
                }
            }
        }

        if (is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['processUrls'])) {
            foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['crawler']['processUrls'] as $func) {
                $params = [
                    'res' => &$res,
                ];
                GeneralUtility::callUserFunction($func, $params, $this);
            }
        }

        return $res;
    }