From 4bbe1d967baf93842bce8556cab44ad3a5506b80 Mon Sep 17 00:00:00 2001
From: Brian Huisman <bhuisman@greywyvern.com>
Date: Tue, 17 Oct 2023 10:36:34 -0400
Subject: [PATCH] Misc fixes

Save the process id of the crawler in the sp_crawling DB value instead of just a flag; we can use it to compare and further prevent race conditions which still seem to happen occasionally.
---
 orcinus/admin.php   |  2 +-
 orcinus/config.php  |  2 +-
 orcinus/crawler.php | 20 +++++++++++++-------
 orcinus/search.php  |  2 +-
 4 files changed, 16 insertions(+), 10 deletions(-)

diff --git a/orcinus/admin.php b/orcinus/admin.php
index 47a4053..0dae315 100644
--- a/orcinus/admin.php
+++ b/orcinus/admin.php
@@ -2673,7 +2673,7 @@ ORCINUS;
                           </label>
                         </td>
                         <th class="fs-5 text-center os_sorting os_desc" scope="col">
-                          <span data-bs-toggle="tooltip" data-bs-placement="top" title="The number of times this query has been searched for with (unique users / IP addresses) in brackets"
+                          <span data-bs-toggle="tooltip" data-bs-placement="top" title="The number of times this query has been searched for with unique users by IP address in (brackets)"
                             role="button" id="os_queries_hits">Hits</span>
                           <img src="img/arrow-down.svg" alt="Sort" title="Sort order" class="align-middle svg-icon-sm mb-1">
                         </th>
diff --git a/orcinus/config.php b/orcinus/config.php
index 52fb2d4..1e1ce8c 100644
--- a/orcinus/config.php
+++ b/orcinus/config.php
@@ -101,7 +101,7 @@ if (!in_array($_DDATA['tbprefix'].'config', $_DDATA['tables'], true)) {
       `sp_sitemap_file` TINYTEXT NOT NULL,
       `sp_sitemap_hostname` TINYTEXT NOT NULL,
       `sp_useragent` TINYTEXT NOT NULL,
-      `sp_crawling` BOOLEAN NOT NULL,
+      `sp_crawling` INT UNSIGNED NOT NULL,
       `sp_cancel` BOOLEAN NOT NULL,
       `sp_progress` TINYTEXT NOT NULL,
       `sp_email_success` BOOLEAN NOT NULL,
diff --git a/orcinus/crawler.php b/orcinus/crawler.php
index f74c4d7..b8e560f 100644
--- a/orcinus/crawler.php
+++ b/orcinus/crawler.php
@@ -474,19 +474,16 @@ switch ($_SERVER['REQUEST_METHOD']) {
       if (empty($_POST->action)) $_POST->action = '';
       switch ($_POST->action) {
         case 'crawl':
-          if (!empty($_POST->sp_key) &&
-              $_ODATA['sp_key'] &&
+          if (!empty($_POST->sp_key) && OS_getValue('sp_key') &&
               $_POST->sp_key == $_ODATA['sp_key']) {
             if (OS_getValue('sp_crawling')) {
               $response = array(
                 'status' => 'Error',
                 'message' => 'Crawler is already running; current progress: '.$_ODATA['sp_progress'][0].'/'.$_ODATA['sp_progress'][1]
               );
-            }
 
             // Go crawl!
-            OS_setValue('sp_crawling', 1);
-            OS_setValue('sp_key', '');
+            } else OS_setValue('sp_crawling', getmypid());
 
           } else {
             $response = array(
@@ -494,6 +491,8 @@ switch ($_SERVER['REQUEST_METHOD']) {
               'message' => 'Incorrect key to initiate crawler'
             );
           }
+
+          OS_setValue('sp_key', '');
           break;
 
         case 'progress':
@@ -513,6 +512,7 @@ switch ($_SERVER['REQUEST_METHOD']) {
             }
           }
 
+          // If crawl is in progress, return just the last 15 lines
           if ($_ODATA['sp_crawling']) $lines = array_slice($lines, -15);
 
           $response = array(
@@ -603,7 +603,7 @@ switch ($_SERVER['REQUEST_METHOD']) {
         } else $_RDATA['sp_log_clilevel'] = 2;
 
         // Start a crawl
-        OS_setValue('sp_crawling', 1);
+        OS_setValue('sp_crawling', getmypid());
 
       } else die('Crawler is already running; exiting...');
     } else die($_ODATA['sp_useragent']);
@@ -619,7 +619,7 @@ switch ($_SERVER['REQUEST_METHOD']) {
         die('Crawler is already running; exiting...');
 
       // Start a crawl
-      OS_setValue('sp_crawling', 1);
+      OS_setValue('sp_crawling', getmypid());
 
     } else die($_ODATA['sp_useragent']);
     break;
@@ -632,6 +632,12 @@ switch ($_SERVER['REQUEST_METHOD']) {
 }
 
 
+// One last check for a race condition
+sleep(1);
+if (OS_getValue('sp_crawling') != getmypid())
+  die('Crawler is already running; exiting...');
+
+
 /* ***** Begin Crawl Execution ************************************* */
 register_shutdown_function('OS_crawlCleanUp');
 ignore_user_abort(true);
diff --git a/orcinus/search.php b/orcinus/search.php
index 3087568..715e69c 100644
--- a/orcinus/search.php
+++ b/orcinus/search.php
@@ -495,7 +495,7 @@ if ($_RDATA['s_searchable_pages']) {
                       if (count($splitter) == 1) {
                         // Grab some random content if there were no
                         // matches in the content
-                        $offset = mt_rand(0, mb_strlen($row['content'], 'UTF-8') - $_ODATA['s_limit_matchtext']);
+                        $offset = mt_rand(0, max(0, mb_strlen($row['content'], 'UTF-8') - $_ODATA['s_limit_matchtext']));
                       } else {
                         $_SDATA['results'][$key]['fragment'][] = $split[0];
                         $offset = floor(max(0, $split[1] - (mb_strlen($term, 'UTF-8') + $_ODATA['s_limit_matchtext']) / 2));