1: <?php
  2: /**
  3:  * This is the main backend page for the linkchecker plugin.
  4:  *
  5:  * @package Plugin
  6:  * @subpackage Linkchecker
  7:  * @version SVN Revision $Rev:$
  8:  *
  9:  * @author Frederic Schneider
 10:  * @copyright four for business AG <www.4fb.de>
 11:  * @license http://www.contenido.org/license/LIZENZ.txt
 12:  * @link http://www.4fb.de
 13:  * @link http://www.contenido.org
 14:  */
 15: 
 16: defined('CON_FRAMEWORK') || die('Illegal call: Missing framework initialization - request aborted.');
 17: 
 18: $plugin_name = "linkchecker";
 19: $cfg = cRegistry::getConfig();
 20: 
 21: if (!$cronjob) {
 22: 
 23:     // Check permissions for linkchecker action
 24:     if (!$perm->have_perm_area_action($plugin_name, "linkchecker") && $cronjob != true) {
 25:         cRegistry::addErrorMessage(i18n("No permissions"));
 26:         $page = new cGuiPage('generic_page');
 27:         $page->abortRendering();
 28:         $page->render();
 29:         exit();
 30:     }
 31: 
 32:     if (cRegistry::getClientId() == 0 && $cronjob != true) {
 33:         $notification->displayNotification("error", i18n("No Client selected"));
 34:         exit();
 35:     }
 36: }
 37: 
 38: // If no mode defined, use mode three
 39: if (empty($_GET['mode'])) {
 40:     $_GET['mode'] = 3;
 41: }
 42: 
 43: // If no action definied
 44: if (empty($_GET['action'])) {
 45:     $_GET['action'] = 'linkchecker';
 46:     $action = "linkchecker";
 47: }
 48: 
 49: plugin_include('linkchecker', 'includes/config.plugin.php');
 50: plugin_include('linkchecker', 'includes/include.checkperms.php');
 51: plugin_include('linkchecker', 'includes/include.linkchecker_tests.php');
 52: 
 53: // Initialization
 54: $aCats = array();
 55: $aSearchIDInfosArt = array();
 56: $aSearchIDInfosCatArt = array();
 57: $aSearchIDInfosNonID = array();
 58: 
 59: // Var initialization
 60: $aUrl = array(
 61:     'cms' => cRegistry::getFrontendUrl(),
 62:     'contenido' => cRegistry::getBackendUrl()
 63: );
 64: 
 65: // Template- and languagevars
 66: if ($cronjob != true) {
 67:     $tpl->set('s', 'MODE', cSecurity::toInteger($_GET['mode']));
 68: }
 69: 
 70: // Fill Subnav I
 71: if (!$cronjob) {
 72:     $sLink = $sess->url("main.php?area=linkchecker&frame=4&action=linkchecker") . '&mode=';
 73: 
 74:     // Fill Subnav II
 75:     $tpl->set('s', 'INTERNS_HREF', $sLink . '1');
 76:     $tpl->set('s', 'INTERNS_LABEL', i18n("Interns", $plugin_name));
 77:     $tpl->set('s', 'EXTERNS_HREF', $sLink . '2');
 78:     $tpl->set('s', 'EXTERNS_LABEL', i18n("Externs", $plugin_name));
 79:     $tpl->set('s', 'INTERNS_EXTERNS_HREF', $sLink . '3');
 80:     $tpl->set('s', 'INTERNS_EXTERNS_LABEL', i18n("Intern/extern Links", $plugin_name));
 81: 
 82:     // Fill Subnav III
 83:     $tpl->set('s', 'UPDATE_HREF', $sLink . cSecurity::toInteger($_GET['mode']) . '&live=1');
 84: }
 85: 
 86: // Cache options
 87: $aCacheName = array(
 88:     'errors' => $sess->id,
 89:     'errorscount' => $aCacheName['errors'] . "ErrorsCountChecked"
 90: );
 91: $oCache = new cFileCache(array(
 92:     'cacheDir' => $cfgClient[$client]['cache']['path'],
 93:     'lifeTime' => $iCacheLifeTime
 94: ));
 95: 
 96: /*
 97:  * ******** Program code ********
 98:  */
 99: 
100: // function linksort
101: function linksort($sErrors) {
102:     if ($_GET['sort'] == "nameart") {
103: 
104:         foreach ($sErrors as $key => $aRow) {
105:             $aNameart[$key] = $aRow['nameart'];
106:         }
107: 
108:         array_multisort($sErrors, SORT_ASC, SORT_STRING, $aNameart);
109:     } elseif ($_GET['sort'] == "namecat") {
110: 
111:         foreach ($sErrors as $key => $aRow) {
112:             $aNamecat[$key] = $aRow['namecat'];
113:         }
114: 
115:         array_multisort($sErrors, SORT_ASC, SORT_STRING, $aNamecat);
116:     } elseif ($_GET['sort'] == "wronglink") {
117: 
118:         foreach ($sErrors as $key => $aRow) {
119:             $aWronglink[$key] = $aRow['url'];
120:         }
121: 
122:         array_multisort($sErrors, SORT_ASC, SORT_STRING, $aWronglink);
123:     } elseif ($_GET['sort'] == "error_type") {
124: 
125:         foreach ($sErrors as $key => $aRow) {
126:             $aError_type[$key] = $aRow['error_type'];
127:         }
128: 
129:         array_multisort($sErrors, SORT_ASC, SORT_STRING, $aError_type);
130:     }
131: 
132:     return $sErrors;
133: }
134: 
135: // function url_is_image
136: function url_is_image($sUrl) {
137:     if (substr($sUrl, -3, 3) == "gif" || substr($sUrl, -3, 3) == "jpg" || substr($sUrl, -4, 4) == "jpeg" || substr($sUrl, -3, 3) == "png" || substr($sUrl, -3, 3) == "tif" || substr($sUrl, -3, 3) == "psd" || substr($sUrl, -3, 3) == "bmp") {
138:         return true;
139:     } else {
140:         return false;
141:     }
142: }
143: 
144: // function url_is_uri
145: function url_is_uri($sUrl) {
146:     if (substr($sUrl, 0, 4) == "file" || substr($sUrl, 0, 3) == "ftp" || substr($sUrl, 0, 4) == "http" || substr($sUrl, 0, 2) == "ww") {
147:         return true;
148:     } else {
149:         return false;
150:     }
151: }
152: 
153: /* Repaire some selected link */
154: if (!empty($_GET['idartlang']) && !empty($_GET['oldlink']) && !empty($_GET['repairedlink'])) {
155: 
156:     if ($_GET['redirect'] == true) { // Update redirect
157:         $sql = "UPDATE " . $cfg['tab']['art_lang'] . " SET redirect_url = '" . $db->escape(base64_decode($_GET['repairedlink'])) . "' WHERE idartlang = '" . cSecurity::toInteger($_GET['idartlang']) . "'";
158:         $db->query($sql);
159:     } else { // Update content
160: 
161:         // Get old value
162:         $sql = "SELECT value FROM " . $cfg['tab']['content'] . " WHERE idartlang = '" . cSecurity::toInteger($_GET['idartlang']) . "'";
163:         $db->query($sql);
164:         $db->next_record();
165: 
166:         // Generate new value
167:         $newvalue = str_replace($db->escape(base64_decode($_GET['oldlink'])), $db->escape(base64_decode($_GET['repairedlink'])), $db->f("value"));
168: 
169:         // Update database table with new value
170:         $sql = "UPDATE " . $cfg['tab']['content'] . " SET value = '" . $newvalue . "' WHERE idartlang = '" . cSecurity::toInteger($_GET['idartlang']) . "'";
171:         $db->query($sql);
172:     }
173: 
174:     // Reset cache
175:     $oCache->remove($aCacheName['errors'], cSecurity::toInteger($_GET['mode']));
176: }
177: 
178: /* Whitelist: Add */
179: if (!empty($_GET['whitelist'])) {
180:     $sql = "REPLACE INTO " . $cfg['tab']['whitelist'] . " VALUES ('" . $db->escape(base64_decode($_GET['whitelist'])) . "', '" . time() . "')";
181:     $db->query($sql);
182: 
183:     $oCache->remove($aCacheName['errors'], cSecurity::toInteger($_GET['mode']));
184: }
185: 
186: /* Whitelist: Get */
187: $sql = "SELECT url FROM " . $cfg['tab']['whitelist'] . " WHERE lastview < " . (time() + $iWhitelistTimeout) . "
188:         AND lastview > " . (time() - $iWhitelistTimeout);
189: $db->query($sql);
190: 
191: $aWhitelist = array();
192: while ($db->nextRecord()) {
193:     $aWhitelist[] = $db->f("url");
194: }
195: 
196: /* Get all links */
197: // Cache errors
198: $sCache_errors = $oCache->get($aCacheName['errors'], cSecurity::toInteger($_GET['mode']));
199: 
200: // Search if cache doesn't exist or we're in live mode
201: if ($sCache_errors && $_GET['live'] != 1) {
202:     $aErrors = unserialize($sCache_errors);
203: } else { // If no cache exists
204: 
205:     // Select all categorys
206:     $sql = "SELECT idcat FROM " . $cfg['tab']['cat'] . " GROUP BY idcat";
207:     $db->query($sql);
208: 
209:     while ($db->nextRecord()) {
210:         if ($cronjob != true) { // Check userrights, if no cronjob
211:             $iCheck = cCatPerm($db->f("idcat"), $db2);
212: 
213:             if ($iCheck == true) {
214:                 $aCats[] = cSecurity::toInteger($db->f("idcat"));
215:             }
216:         } else {
217:             $aCats[] = cSecurity::toInteger($db->f("idcat"));
218:         }
219:     }
220: 
221:     // Build $aCats-Statement
222:     if (count($aCats) == 0) {
223:         $aCats_Sql = "";
224:     } else {
225:         $aCats_Sql = "AND cat.idcat IN (0, " . join(", ", $aCats) . ")";
226:     }
227: 
228:     // Get languageId
229:     $languageId = cRegistry::getLanguageId();
230: 
231:     // How many articles exist? [Text]
232:     $sql = "SELECT art.title, art.idartlang, art.idlang, cat.idart, cat.idcat, catName.name AS namecat, con.value FROM " . $cfg['tab']['cat_art'] . " cat
233:             LEFT JOIN " . $cfg['tab']['art_lang'] . " art ON (art.idart = cat.idart)
234:             LEFT JOIN " . $cfg['tab']['cat_lang'] . " catName ON (catName.idcat = cat.idcat)
235:             LEFT JOIN " . $cfg['tab']['content'] . " con ON (con.idartlang = art.idartlang)
236:             WHERE (con.value LIKE '%action%' OR con.value LIKE '%data%' OR con.value LIKE '%href%' OR con.value LIKE '%src%')
237:             " . $aCats_Sql . " AND cat.idcat != '0'
238:             AND art.idlang = '" . cSecurity::toInteger($languageId) . "' AND catName.idlang = '" . cSecurity::toInteger($languageId) . "'
239:             AND art.online = '1' AND art.redirect = '0'";
240:     $db->query($sql);
241: 
242:     while ($db->nextRecord()) {
243:         // Text decode
244:         $value = $db->f("value");
245: 
246:         // Search the text
247:         searchLinks($value, $db->f("idart"), $db->f("title"), $db->f("idcat"), $db->f("namecat"), $db->f("idartlang"), $db->f("idlang"));
248: 
249:         // Search front_content.php-links
250:         if ($_GET['mode'] != 2) {
251:             searchFrontContentLinks($value, $db->f("idart"), $db->f("title"), $db->f("idcat"), $db->f("namecat"));
252:         }
253:     }
254: 
255:     // How many articles exist? [Redirects]
256:     $sql = "SELECT art.title, art.redirect_url, art.idartlang, art.idlang, cat.idart, cat.idcat, catName.name AS namecat FROM " . $cfg['tab']['cat_art'] . " cat
257:             LEFT JOIN " . $cfg['tab']['art_lang'] . " art ON (art.idart = cat.idart)
258:             LEFT JOIN " . $cfg['tab']['cat_lang'] . " catName ON (catName.idcat = cat.idcat)
259:             WHERE art.online = '1' AND art.redirect = '1' " . $aCats_Sql . "
260:             AND art.idlang = '" . cSecurity::toInteger($languageId) . "' AND catName.idlang = '" . cSecurity::toInteger($languageId) . "'
261:             AND cat.idcat != '0'";
262:     $db->query($sql);
263: 
264:     while ($db->nextRecord()) {
265:         // Search links
266:         searchLinks($db->f("redirect_url"), $db->f("idart"), $db->f("title"), $db->f("idcat"), $db->f("namecat"), $db->f("idartlang"), $db->f("idlang"), "Redirect");
267: 
268:         // Search front_content.php-links
269:         if ($_GET['mode'] != 2) {
270:             searchFrontContentLinks($db->f("redirect_url"), $db->f("idart"), $db->f("title"), $db->f("idcat"), $db->f("namecat"));
271:         }
272:     }
273: 
274:     // Check the links
275:     checkLinks();
276: }
277: 
278: /* Analysis of the errors */
279: // Templateset
280: if ($cronjob != true) {
281:     $tpl->set('s', 'TITLE', i18n('Link analysis from ', $plugin_name) . strftime(i18n('%Y-%m-%d', $plugin_name), time()));
282: }
283: 
284: // If no errors found, say that
285: if (empty($aErrors) && $cronjob != true) {
286:     // Reset cache
287:     $oCache->remove($aCacheName['errors'], cSecurity::toInteger($_GET['mode']));
288: 
289:     $tpl->set('s', 'NO_ERRORS', i18n("<strong>No errors</strong> were found.", $plugin_name));
290:     $tpl->generate($cfg['templates']['linkchecker_noerrors']);
291: } elseif (!empty($aErrors) && $cronjob != true) {
292: 
293:     $tpl->set('s', 'ERRORS_HEADLINE', i18n("Total checked links", $plugin_name));
294:     $tpl->set('s', 'ERRORS_HEADLINE_ARTID', i18n("idart", $plugin_name));
295:     $tpl->set('s', 'ERRORS_HEADLINE_ARTICLE', i18n("Article", $plugin_name));
296:     $tpl->set('s', 'ERRORS_HEADLINE_CATID', i18n("idcat", $plugin_name));
297:     $tpl->set('s', 'ERRORS_HEADLINE_CATNAME', i18n("Category", $plugin_name));
298:     $tpl->set('s', 'ERRORS_HEADLINE_DESCRIPTION', i18n("Description", $plugin_name));
299:     $tpl->set('s', 'ERRORS_HEADLINE_REPAIRED', i18n("Repair", $plugin_name));
300:     $tpl->set('s', 'ERRORS_HEADLINE_LINK', i18n("Linkerror", $plugin_name));
301:     $tpl->set('s', 'ERRORS_HEADLINE_LINKS_ARTICLES', i18n("Links to articles", $plugin_name));
302:     $tpl->set('s', 'ERRORS_HEADLINE_LINKS_CATEGORYS', i18n("Links to categories", $plugin_name));
303:     $tpl->set('s', 'ERRORS_HEADLINE_LINKS_DOCIMAGES', i18n("Links to documents and images", $plugin_name));
304:     $tpl->set('s', 'ERRORS_HEADLINE_OTHERS', i18n("Links to extern sites and not defined links", $plugin_name));
305:     $tpl->set('s', 'ERRORS_HEADLINE_WHITELIST', "Whitelist");
306:     $tpl->set('s', 'ERRORS_HELP_ERRORS', i18n("Wrong links", $plugin_name));
307: 
308:     // error_output initialization
309:     $aError_output = array(
310:         'art' => '',
311:         'cat' => '',
312:         'docimages' => '',
313:         'others' => ''
314:     );
315: 
316:     // Initializing repair class
317:     $repair = new LinkcheckerRepair();
318: 
319:     foreach ($aErrors as $sKey => $aRow) {
320: 
321:         $aRow = linksort($aRow);
322: 
323:         for ($i = 0; $i < count($aRow); $i++) {
324: 
325:             $tpl2 = new cTemplate();
326:             $tpl2->reset();
327: 
328:             // html entities for artname and catname
329:             $aRow[$i]['nameart'] = conHtmlentities($aRow[$i]['nameart']);
330:             $aRow[$i]['namecat'] = conHtmlentities($aRow[$i]['namecat']);
331: 
332:             // set template variables
333:             $tpl2->set('s', 'ERRORS_ARTID', $aRow[$i]['idart']);
334:             $tpl2->set('s', 'ERRORS_ARTICLE', $aRow[$i]['nameart']);
335:             $tpl2->set('s', 'ERRORS_ARTICLE_SHORT', substr($aRow[$i]['nameart'], 0, 20) . ((strlen($aRow[$i]['nameart']) > 20) ? ' ...' : ''));
336:             $tpl2->set('s', 'ERRORS_CATID', $aRow[$i]['idcat']);
337:             $tpl2->set('s', 'ERRORS_LANGARTID', $aRow[$i]['idartlang']);
338:             $tpl2->set('s', 'ERRORS_LINK', $aRow[$i]['url']);
339:             $tpl2->set('s', 'ERRORS_LINK_ENCODE', base64_encode($aRow[$i]['url']));
340:             $tpl2->set('s', 'ERRORS_LINK_SHORT', substr($aRow[$i]['url'], 0, 45) . ((strlen($aRow[$i]['url']) > 45) ? ' ...' : ''));
341:             $tpl2->set('s', 'ERRORS_CATNAME', $aRow[$i]['namecat']);
342:             $tpl2->set('s', 'ERRORS_CATNAME_SHORT', substr($aRow[$i]['namecat'], 0, 20) . ((strlen($aRow[$i]['namecat']) > 20) ? ' ...' : ''));
343:             $tpl2->set('s', 'MODE', $_GET['mode']);
344:             $tpl2->set('s', 'URL_FRONTEND', $aUrl['cms']);
345: 
346:             if ($aRow[$i]['error_type'] == "unknown") {
347:                 $tpl2->set('s', 'ERRORS_ERROR_TYPE', i18n("Unknown", $plugin_name));
348:                 $tpl2->set('s', 'ERRORS_ERROR_TYPE_HELP', i18n("Unknown: articles, documents etc. do not exist.", $plugin_name));
349:             } elseif ($aRow[$i]['error_type'] == "offline") {
350:                 $tpl2->set('s', 'ERRORS_ERROR_TYPE', i18n("Offline", $plugin_name));
351:                 $tpl2->set('s', 'ERRORS_ERROR_TYPE_HELP', i18n("Offline: article or category is offline.", $plugin_name));
352:             } elseif ($aRow[$i]['error_type'] == "startart") {
353:                 $tpl2->set('s', 'ERRORS_ERROR_TYPE', i18n("Offline startarticle", $plugin_name));
354:                 $tpl2->set('s', 'ERRORS_ERROR_TYPE_HELP', i18n("Offline: article or category is offline.", $plugin_name));
355:             } elseif ($aRow[$i]['error_type'] == "dbfs") {
356:                 $tpl2->set('s', 'ERRORS_ERROR_TYPE', i18n("Filemanager", $plugin_name));
357:                 $tpl2->set('s', 'ERRORS_ERROR_TYPE_HELP', i18n("dbfs: no matches found in the dbfs database.", $plugin_name));
358:             } elseif ($aRow[$i]['error_type'] == "invalidurl") {
359:                 $tpl2->set('s', 'ERRORS_ERROR_TYPE', i18n("Invalid url", $plugin_name));
360:                 $tpl2->set('s', 'ERRORS_ERROR_TYPE_HELP', i18n("Invalid url, i. e. syntax error.", $plugin_name));
361: 
362:                 // Try to repair this link misstage
363:                 $repaired_link = $repair->checkLink($aRow[$i]['url']);
364:             }
365: 
366:             // Generate repaired link variables
367:             if ($aRow[$i]['error_type'] != "invalidurl") { // No invalid url
368:                                                            // case
369:                 $tpl2->set('s', 'ERRORS_REPAIRED_LINK', '-');
370:             } elseif ($repaired_link == false) { // Linkchecker can not repaire
371:                                                  // this link
372:                 $tpl2->set('s', 'ERRORS_REPAIRED_LINK', i18n("No repaired link", $plugin_name));
373:             } else { // Yeah, we have an repaired link!
374: 
375:                 // Repaired question
376:                 $repaired_question = i18n("Linkchecker has found a way to repair your wrong link. Do you want to automatically repair the link to the URL below?", $plugin_name);
377: 
378:                 $tpl2->set('s', 'ERRORS_REPAIRED_LINK', '<a href="javascript:void(0)" onclick="javascript:Con.showConfirmation(\'' . $repaired_question . '<br /><br /><strong>' . $repaired_link . '</strong>\', function() { window.location.href=\'' . $aUrl['contenido'] . 'main.php?area=linkchecker&frame=4&contenido=' . $sess->id . '&action=linkchecker&mode=' . $_GET['mode'] . '&idartlang=' . $aRow[$i]['idartlang'] . '&oldlink=' . base64_encode($aRow[$i]['url']) . '&repairedlink=' . base64_encode($repaired_link) . '&redirect=' . $aRow[$i]['redirect'] . '\';})"><img src="' . $aUrl['contenido'] . 'images/but_editlink.gif" alt="" border="0"></a>');
379:             }
380: 
381:             if ($sKey != "cat") {
382:                 $aError_output[$sKey] .= $tpl2->generate($cfg['templates']['linkchecker_test_errors'], 1);
383:             } else {
384:                 // special template for idcats
385:                 $aError_output[$sKey] .= $tpl2->generate($cfg['templates']['linkchecker_test_errors_cat'], 1);
386:             }
387:         }
388:     }
389: 
390:     // Counter
391:     if ($iCounter = $oCache->get($aCacheName['errorscount'], cSecurity::toInteger($_GET['mode']))) {
392:         // Cache exists?
393:         $iErrorsCountChecked = $iCounter;
394:     } else {
395:         // Count searched links: idarts + idcats + idcatarts + others
396:         $iErrorsCountChecked = count($aSearchIDInfosArt) + count($aSearchIDInfosCat) + count($aSearchIDInfosCatArt) + count($aSearchIDInfosNonID);
397:     }
398: 
399:     // Count errors
400:     foreach ($aErrors as $sKey => $aRow) {
401:         $iErrorsCounted += count($aErrors[$sKey]);
402:     }
403: 
404:     $tpl->set('s', 'ERRORS_COUNT_CHECKED', $iErrorsCountChecked);
405:     $tpl->set('s', 'ERRORS_COUNT_ERRORS', $iErrorsCounted);
406:     $tpl->set('s', 'ERRORS_COUNT_ERRORS_PERCENT', round(($iErrorsCounted * 100) / $iErrorsCountChecked, 2));
407: 
408:     /* Template output */
409:     foreach ($aError_output as $sKey => $sValue) {
410: 
411:         if (empty($aError_output[$sKey])) { // Errors for this type?
412:             $tpl2->set('s', 'ERRORS_NOTHING', i18n("No errors for this type.", $plugin_name));
413:             $aError_output[$sKey] = $tpl2->generate($cfg['templates']['linkchecker_test_nothing'], 1);
414:         }
415: 
416:         $tpl->set('s', 'ERRORS_SHOW_' . strtoupper($sKey), $aError_output[$sKey]);
417: 
418:         if (count($aErrors[$sKey]) > 0) {
419:             $tpl->set('s', 'ERRORS_COUNT_ERRORS_' . strtoupper($sKey), '<span class="settingWrong">' . count($aErrors[$sKey]) . '</span>');
420:         } else {
421:             $tpl->set('s', 'ERRORS_COUNT_ERRORS_' . strtoupper($sKey), count($aErrors[$key]));
422:         }
423:     }
424: 
425:     $tpl->generate($cfg['templates']['linkchecker_test']);
426: 
427:     /* Cache */
428:     // Reset cache
429:     $oCache->remove($aCacheName['errors'], cSecurity::toInteger($_GET['mode']));
430: 
431:     // Build new cache
432:     $oCache->save(serialize($aErrors), $aCacheName['errors'], cSecurity::toInteger($_GET['mode']));
433:     $oCache->save($iErrorsCountChecked, $aCacheName['errorscount'], cSecurity::toInteger($_GET['mode']));
434: }
435: 
436: // Log
437: if ($cronjob != true) {
438:     $backend->log(0, 0, cRegistry::getClientId(), cRegistry::getLanguageId(), $action);
439: }
440: 
441: ?>