From 41f6b25f0f27913ca8ef44a989738cea73c4884e Mon Sep 17 00:00:00 2001
From: Brian Huisman <bhuisman@greywyvern.com>
Date: Thu, 27 Apr 2023 13:10:22 -0400
Subject: [PATCH] Allow specifying Default Category

---
 orcinus/admin.php   | 24 ++++++++++++++++++++++++
 orcinus/config.php  |  2 ++
 orcinus/crawler.php |  2 +-
 3 files changed, 27 insertions(+), 1 deletion(-)

diff --git a/orcinus/admin.php b/orcinus/admin.php
index 12e9fed..5e849b4 100644
--- a/orcinus/admin.php
+++ b/orcinus/admin.php
@@ -425,6 +425,14 @@ if (!$_SESSION['admin_username']) {
             OS_setValue('sp_ignore_ext', substr($_POST['os_sp_ignore_ext'], 0, 4095));
           }
 
+          if (isset($_POST['os_sp_category_default'])) {
+            $_POST['os_sp_category_default'] = preg_replace(array('/\s/', '/ {2,}/'), ' ', trim($_POST['os_sp_category_default']));
+            $_POST['os_sp_category_default'] = preg_replace('/[^\w \d-]/', '', $_POST['os_sp_category_default']);
+            if ($_POST['os_sp_category_default']) {
+              OS_setValue('sp_category_default', substr($_POST['os_sp_category_default'], 0, 30));
+            } else $_SESSION['error'][] = 'Category names may only contain letters, numbers, spaces or dashes.';
+          } else $_SESSION['error'][] = 'Please supply a category name.';
+
           if (isset($_POST['os_sp_ignore_css'])) {
             $_POST['os_sp_ignore_css'] = preg_replace(
               array('/[^\w\d\. #_:-]/', '/ {2,}/'),
@@ -2146,6 +2154,22 @@ document.write(mustache.render(
                           ?></textarea>
                         </label>
                       </li>
+                      <li class="list-group-item">
+                        <h4 aria-labelledby="os_sp_category_text">Categories</h4>
+                        <p id="os_sp_category_text" class="form-text">
+                          Usually you'll want all your indexed pages in just one category. In some cases
+                          however, you may want to offer users an additional way to restrict results by
+                          putting groups of pages into multiple categories. You can set page categories
+                          from the <em>Page Index</em>.
+                        </p>
+                        <label class="d-flex lh-lg w-100 mb-2">
+                          <strong class="pe-2">Default Category:</strong>
+                          <span class="flex-grow-1 text-end">
+                            <input type="text" name="os_sp_category_default" value="<?php echo $_ODATA['sp_category_default']; ?>" maxlength="30" class="form-control d-inline-block w-auto mw-10em"
+                              data-bs-toggle="tooltip" data-bs-placement="bottom" title="Default Category value to apply to newly found pages.">
+                          </span>
+                        </label>
+                      </li>
                       <li class="list-group-item">
                         <h4>Content Filters</h4>
                         <div class="row">
diff --git a/orcinus/config.php b/orcinus/config.php
index d309e54..63e2f24 100644
--- a/orcinus/config.php
+++ b/orcinus/config.php
@@ -58,6 +58,7 @@ if (!in_array($_DDATA['tbprefix'].'config', $_DDATA['tables'])) {
       `sp_ignore_css` TEXT NOT NULL,
       `sp_require_url` TEXT NOT NULL,
       `sp_title_strip` TEXT NOT NULL,
+      `sp_category_default` TINYTEXT NOT NULL,
       `sp_timeout_url` TINYINT UNSIGNED NOT NULL,
       `sp_timeout_crawl` SMALLINT UNSIGNED NOT NULL,
       `sp_interval` TINYINT UNSIGNED NOT NULL,
@@ -137,6 +138,7 @@ if (!count($testConf->fetchAll())) {
       `sp_ignore_css`=\'.noindex footer form head nav noscript select style svg textarea\',
       `sp_require_url`=\'\',
       `sp_title_strip`=\'\',
+      `sp_category_default`=\'Main\',
       `sp_timeout_url`=10,
       `sp_timeout_crawl`=300,
       `sp_interval`=24,
diff --git a/orcinus/crawler.php b/orcinus/crawler.php
index 623b285..f7e05d2 100644
--- a/orcinus/crawler.php
+++ b/orcinus/crawler.php
@@ -1432,7 +1432,7 @@ while ($_cURL && count($_RDATA['sp_queue'])) {
           'title' => $data['title'],
           'description' => $data['description'],
           'keywords' => $data['keywords'],
-          'category' => 'Main',
+          'category' => $_ODATA['sp_category_default'],
           'weighted' => $data['weighted'],
           'links' => json_encode($data['links'], JSON_INVALID_UTF8_IGNORE),
           'content' => $data['content'],