LinkAnalyzer.php

Go to the documentation of this file.

<?php
namespace TYPO3\CMS\Linkvalidator;

use TYPO3\CMS\Backend\Utility\BackendUtility;

class LinkAnalyzer {

    protected $searchFields = array();

    protected $pidList = '';

    protected $linkCounts = array();

    protected $brokenLinkCounts = array();

    protected $recordsWithBrokenLinks = array();

    protected $hookObjectsArr = array();

    protected $extPageInTreeInfo = array();

    protected $recordReference = '';

    protected $pageWithAnchor = '';

    public function __construct() {
        $GLOBALS['LANG']->includeLLFile('EXT:linkvalidator/Resources/Private/Language/Module/locallang.xlf');
        // Hook to handle own checks
        if (is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['linkvalidator']['checkLinks'])) {
            foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['linkvalidator']['checkLinks'] as $key => $classRef) {
                $this->hookObjectsArr[$key] = \TYPO3\CMS\Core\Utility\GeneralUtility::getUserObj($classRef);
            }
        }
    }

    public function init(array $searchField, $pid) {
        $this->searchFields = $searchField;
        $this->pidList = $pid;
    }

    public function getLinkStatistics($checkOptions = array(), $considerHidden = FALSE) {
        $results = array();
        $pidList = implode(',', \TYPO3\CMS\Core\Utility\GeneralUtility::intExplode(',', $this->pidList, true));
        if (count($checkOptions) > 0 && !empty($pidList)) {
            $checkKeys = array_keys($checkOptions);
            $checkLinkTypeCondition = ' AND link_type IN (\'' . implode('\',\'', $checkKeys) . '\')';
            $GLOBALS['TYPO3_DB']->exec_DELETEquery(
                'tx_linkvalidator_link',
                '(record_pid IN (' . $pidList . ')' .
                    ' OR ( record_uid IN (' . $pidList . ') AND table_name like \'pages\'))' .
                    $checkLinkTypeCondition
            );
            // Traverse all configured tables
            foreach ($this->searchFields as $table => $fields) {
                if ($table === 'pages') {
                    $where = 'deleted = 0 AND uid IN (' . $pidList . ')';
                } else {
                    $where = 'deleted = 0 AND pid IN (' . $pidList . ')';
                }
                if (!$considerHidden) {
                    $where .= BackendUtility::BEenableFields($table);
                }
                // If table is not configured, assume the extension is not installed
                // and therefore no need to check it
                if (!is_array($GLOBALS['TCA'][$table])) {
                    continue;
                }
                // Re-init selectFields for table
                $selectFields = 'uid, pid';
                $selectFields .= ', ' . $GLOBALS['TCA'][$table]['ctrl']['label'] . ', ' . implode(', ', $fields);
                // TODO: only select rows that have content in at least one of the relevant fields (via OR)
                $res = $GLOBALS['TYPO3_DB']->exec_SELECTquery($selectFields, $table, $where);
                // Get record rows of table
                while (($row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($res)) !== FALSE) {
                    // Analyse each record
                    $this->analyzeRecord($results, $table, $fields, $row);
                }
                $GLOBALS['TYPO3_DB']->sql_free_result($res);
            }
            foreach ($this->hookObjectsArr as $key => $hookObj) {
                if (is_array($results[$key]) && empty($checkOptions) || is_array($results[$key]) && $checkOptions[$key]) {
                    //  Check them
                    foreach ($results[$key] as $entryKey => $entryValue) {
                        $table = $entryValue['table'];
                        $record = array();
                        $record['headline'] = BackendUtility::getRecordTitle($table, $entryValue['row']);
                        $record['record_pid'] = $entryValue['row']['pid'];
                        $record['record_uid'] = $entryValue['uid'];
                        $record['table_name'] = $table;
                        $record['link_title'] = $entryValue['link_title'];
                        $record['field'] = $entryValue['field'];
                        $record['last_check'] = time();
                        $this->recordReference = $entryValue['substr']['recordRef'];
                        $this->pageWithAnchor = $entryValue['pageAndAnchor'];
                        if (!empty($this->pageWithAnchor)) {
                            // Page with anchor, e.g. 18#1580
                            $url = $this->pageWithAnchor;
                        } else {
                            $url = $entryValue['substr']['tokenValue'];
                        }
                        $this->linkCounts[$table]++;
                        $checkUrl = $hookObj->checkLink($url, $entryValue, $this);
                        // Broken link found
                        if (!$checkUrl) {
                            $response = array();
                            $response['valid'] = FALSE;
                            $response['errorParams'] = $hookObj->getErrorParams();
                            $this->brokenLinkCounts[$table]++;
                            $record['link_type'] = $key;
                            $record['url'] = $url;
                            $record['url_response'] = serialize($response);
                            $GLOBALS['TYPO3_DB']->exec_INSERTquery('tx_linkvalidator_link', $record);
                        } elseif (\TYPO3\CMS\Core\Utility\GeneralUtility::_GP('showalllinks')) {
                            $response = array();
                            $response['valid'] = TRUE;
                            $this->brokenLinkCounts[$table]++;
                            $record['url'] = $url;
                            $record['link_type'] = $key;
                            $record['url_response'] = serialize($response);
                            $GLOBALS['TYPO3_DB']->exec_INSERTquery('tx_linkvalidator_link', $record);
                        }
                    }
                }
            }
        }
    }

    public function analyzeRecord(array &$results, $table, array $fields, array $record) {
        // Put together content of all relevant fields
        $haystack = '';
        $htmlParser = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('TYPO3\\CMS\\Core\\Html\\HtmlParser');
        $idRecord = $record['uid'];
        // Get all references
        foreach ($fields as $field) {
            $haystack .= $record[$field] . ' --- ';
            $conf = $GLOBALS['TCA'][$table]['columns'][$field]['config'];
            $valueField = $record[$field];
            // Check if a TCA configured field has soft references defined (see TYPO3 Core API document)
            if ($conf['softref'] && strlen($valueField)) {
                // Explode the list of soft references/parameters
                $softRefs = BackendUtility::explodeSoftRefParserList($conf['softref']);
                if ($softRefs !== FALSE) {
                    // Traverse soft references
                    foreach ($softRefs as $spKey => $spParams) {
                        $softRefObj = BackendUtility::softRefParserObj($spKey);
                        // If there is an object returned...
                        if (is_object($softRefObj)) {
                            // Do processing
                            $resultArray = $softRefObj->findRef($table, $field, $idRecord, $valueField, $spKey, $spParams);
                            if (!empty($resultArray['elements'])) {
                                if ($spKey == 'typolink_tag') {
                                    $this->analyseTypoLinks($resultArray, $results, $htmlParser, $record, $field, $table);
                                } else {
                                    $this->analyseLinks($resultArray, $results, $record, $field, $table);
                                }
                            }
                        }
                    }
                }
            }
        }
    }

    protected function analyseLinks(array $resultArray, array &$results, array $record, $field, $table) {
        foreach ($resultArray['elements'] as $element) {
            $r = $element['subst'];
            $type = '';
            $idRecord = $record['uid'];
            if (!empty($r)) {
                foreach ($this->hookObjectsArr as $keyArr => $hookObj) {
                    $type = $hookObj->fetchType($r, $type, $keyArr);
                    // Store the type that was found
                    // This prevents overriding by internal validator
                    if (!empty($type)) {
                        $r['type'] = $type;
                    }
                }
                $results[$type][$table . ':' . $field . ':' . $idRecord . ':' . $r['tokenID']]['substr'] = $r;
                $results[$type][$table . ':' . $field . ':' . $idRecord . ':' . $r['tokenID']]['row'] = $record;
                $results[$type][$table . ':' . $field . ':' . $idRecord . ':' . $r['tokenID']]['table'] = $table;
                $results[$type][$table . ':' . $field . ':' . $idRecord . ':' . $r['tokenID']]['field'] = $field;
                $results[$type][$table . ':' . $field . ':' . $idRecord . ':' . $r['tokenID']]['uid'] = $idRecord;
            }
        }
    }

    protected function analyseTypoLinks(array $resultArray, array &$results, $htmlParser, array $record, $field, $table) {
        $currentR = array();
        $linkTags = $htmlParser->splitIntoBlock('link', $resultArray['content']);
        $idRecord = $record['uid'];
        $type = '';
        $title = '';
        for ($i = 1; $i < count($linkTags); $i += 2) {
            $referencedRecordType = '';
            foreach ($resultArray['elements'] as $element) {
                $type = '';
                $r = $element['subst'];
                if (!empty($r['tokenID'])) {
                    if (substr_count($linkTags[$i], $r['tokenID'])) {
                        // Type of referenced record
                        if (strpos($r['recordRef'], 'pages') !== FALSE) {
                            $currentR = $r;
                            // Contains number of the page
                            $referencedRecordType = $r['tokenValue'];
                            $wasPage = TRUE;
                        } elseif (strpos($r['recordRef'], 'tt_content') !== FALSE && (isset($wasPage) && $wasPage === TRUE)) {
                            $referencedRecordType = $referencedRecordType . '#c' . $r['tokenValue'];
                            $wasPage = FALSE;
                        } else {
                            $currentR = $r;
                        }
                        $title = strip_tags($linkTags[$i]);
                    }
                }
            }
            foreach ($this->hookObjectsArr as $keyArr => $hookObj) {
                $type = $hookObj->fetchType($currentR, $type, $keyArr);
                // Store the type that was found
                // This prevents overriding by internal validator
                if (!empty($type)) {
                    $currentR['type'] = $type;
                }
            }
            $results[$type][$table . ':' . $field . ':' . $idRecord . ':' . $currentR['tokenID']]['substr'] = $currentR;
            $results[$type][$table . ':' . $field . ':' . $idRecord . ':' . $currentR['tokenID']]['row'] = $record;
            $results[$type][$table . ':' . $field . ':' . $idRecord . ':' . $currentR['tokenID']]['table'] = $table;
            $results[$type][$table . ':' . $field . ':' . $idRecord . ':' . $currentR['tokenID']]['field'] = $field;
            $results[$type][$table . ':' . $field . ':' . $idRecord . ':' . $currentR['tokenID']]['uid'] = $idRecord;
            $results[$type][$table . ':' . $field . ':' . $idRecord . ':' . $currentR['tokenID']]['link_title'] = $title;
            $results[$type][$table . ':' . $field . ':' . $idRecord . ':' . $currentR['tokenID']]['pageAndAnchor'] = $referencedRecordType;
        }
    }

    public function getLinkCounts($curPage) {
        $markerArray = array();
        if (empty($this->pidList)) {
            $this->pidList = $curPage;
        }
        $this->pidList = rtrim($this->pidList, ',');
        if (($res = $GLOBALS['TYPO3_DB']->exec_SELECTquery(
            'count(uid) as nbBrokenLinks,link_type',
            'tx_linkvalidator_link',
            'record_pid in (' . $this->pidList . ')', 'link_type')
        )) {
            while (($row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($res)) !== FALSE) {
                $markerArray[$row['link_type']] = $row['nbBrokenLinks'];
                $markerArray['brokenlinkCount'] += $row['nbBrokenLinks'];
            }
        }
        $GLOBALS['TYPO3_DB']->sql_free_result($res);
        return $markerArray;
    }

    public function extGetTreeList($id, $depth, $begin = 0, $permsClause, $considerHidden = FALSE) {
        $depth = (int)$depth;
        $begin = (int)$begin;
        $id = (int)$id;
        $theList = '';
        if ($depth > 0) {
            $res = $GLOBALS['TYPO3_DB']->exec_SELECTquery(
                'uid,title,hidden,extendToSubpages',
                'pages', 'pid=' . $id . ' AND deleted=0 AND ' . $permsClause
            );
            while (($row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($res)) !== FALSE) {
                if ($begin <= 0 && ($row['hidden'] == 0 || $considerHidden == 1)) {
                    $theList .= $row['uid'] . ',';
                    $this->extPageInTreeInfo[] = array($row['uid'], htmlspecialchars($row['title'], $depth));
                }
                if ($depth > 1 && (!($row['hidden'] == 1 && $row['extendToSubpages'] == 1) || $considerHidden == 1)) {
                    $theList .= $this->extGetTreeList($row['uid'], $depth - 1, $begin - 1, $permsClause, $considerHidden);
                }
            }
            $GLOBALS['TYPO3_DB']->sql_free_result($res);
        }
        return $theList;
    }

    public function getRootLineIsHidden(array $pageInfo) {
        $hidden = FALSE;
        if ($pageInfo['extendToSubpages'] == 1 && $pageInfo['hidden'] == 1) {
            $hidden = TRUE;
        } else {
            if ($pageInfo['pid'] > 0) {
                $res = $GLOBALS['TYPO3_DB']->exec_SELECTquery('uid,title,hidden,extendToSubpages', 'pages', 'uid=' . $pageInfo['pid']);
                while (($row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($res)) !== FALSE) {
                    $hidden = $this->getRootLineIsHidden($row);
                }
                $GLOBALS['TYPO3_DB']->sql_free_result($res);
            } else {
                $hidden = FALSE;
            }
        }
        return $hidden;
    }

}