From ebc1a573dc9b92c365a1fba08b84a215c9bdb038 Mon Sep 17 00:00:00 2001
From: ghost <localhost>
Date: Sun, 25 Jun 2023 22:11:49 +0300
Subject: [PATCH] initiate CLI tool

---
 cli/yggo.php | 124 +++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 124 insertions(+)
 create mode 100644 cli/yggo.php

diff --git a/cli/yggo.php b/cli/yggo.php
new file mode 100644
index 0000000..0e304ce
--- /dev/null
+++ b/cli/yggo.php
@@ -0,0 +1,124 @@
+<?php
+
+// CLI only to prevent https server connection timeout
+if (php_sapi_name() != 'cli') {
+  echo sprintf(_('supported command line interface only'), PHP_EOL);
+  exit;
+}
+
+// Lock multi-thread execution
+$semaphore = sem_get(crc32('crontab.cleaner'), 1);
+
+if (false === sem_acquire($semaphore, true)) {
+
+  echo _('Process locked by another thread.') . PHP_EOL;
+  exit;
+}
+
+// Load system dependencies
+require_once(__DIR__ . '/../config/app.php');
+require_once(__DIR__ . '/../library/mysql.php');
+require_once(__DIR__ . '/../library/filter.php');
+require_once(__DIR__ . '/../library/vendor/simple_html_dom.php');
+
+// Connect database
+$db = new MySQL(DB_HOST, DB_PORT, DB_NAME, DB_USERNAME, DB_PASSWORD);
+
+// CLI begin
+if (empty($argv[1])) $argv[1] = 'help';
+
+switch ($argv[1]) {
+
+  case 'hostPageDom':
+
+    if (empty($argv[2])) {
+      echo PHP_EOL . _('hostPageDom method requires action argument') . PHP_EOL;
+    }
+
+    switch ($argv[2]) {
+
+      case 'generate':
+
+        if (CRAWL_HOST_PAGE_DOM_SELECTORS) {
+
+          // Init variables
+          $hostPagesProcessedTotal = 0;
+          $hostPageDOMAddedTotal   = 0;
+
+          // Begin selectors extraction
+          foreach ($db->getHostPagesByIndexed() as $hostPage) {
+
+            if (false !== stripos(Filter::mime($hostPage->mime), 'text/html')) {
+
+              if ($hostPageDescription = $db->getLastPageDescription($hostPage->hostPageId)) {
+
+                $hostPagesProcessedTotal++;
+
+                if (!empty($hostPageDescription->data)) {
+
+                  $html = str_get_html(base64_decode($hostPageDescription->data));
+
+                  foreach ((array) explode(',', CRAWL_HOST_PAGE_DOM_SELECTORS) as $selector) {
+
+                    foreach($html->find($selector) as $element) {
+
+                      if (!empty($element->innertext)) {
+
+                        $hostPageDOMAddedTotal++;
+
+                        $db->addHostPageDom($hostPage->hostPageId,
+                                            time(),
+                                            $selector,
+                                            trim(CRAWL_HOST_PAGE_DOM_STRIP_TAGS ? strip_tags(
+                                                                                  preg_replace('/[\s]+/',
+                                                                                                ' ',
+                                                                                                str_replace(['<br />', '<br/>', '<br>', '</'],
+                                                                                                            [' ', ' ', ' ', ' </'],
+                                                                                                            $element->innertext))) : $element->innertext));
+                      }
+                    }
+                  }
+                }
+              }
+            }
+          }
+
+          echo sprintf(_('Host pages processed: %s'), $hostPagesProcessedTotal) . PHP_EOL;
+          echo sprintf(_('Host page DOM elements added: %s'), $hostPageDOMAddedTotal) . PHP_EOL;
+          exit;
+        }
+
+        echo PHP_EOL . _('CRAWL_HOST_PAGE_DOM_SELECTORS not provided in the configuration file') . PHP_EOL;
+        exit;
+
+      break;
+      case 'truncate':
+
+        $db->truncateHostPageDom();
+
+        echo _('hostPageDom table successfully truncated') . PHP_EOL;
+        exit;
+
+      break;
+      default:
+
+        echo PHP_EOL . _('undefined action argument') . PHP_EOL;
+    }
+
+  break;
+}
+
+// Default message
+echo '__  ______________      __' . PHP_EOL;
+echo '\ \/ / ____/ ____/___  / /' . PHP_EOL;
+echo ' \  / / __/ / __/ __ \/ /'  . PHP_EOL;
+echo ' / / /_/ / /_/ / /_/ /_/'   . PHP_EOL;
+echo '/_/\____/\____/\____(_)'    . PHP_EOL;
+
+echo PHP_EOL . _('available options:') . PHP_EOL . PHP_EOL;
+
+echo _('  help                 - this message') . PHP_EOL;
+echo _('  hostPageDom generate - make hostPageDom index based on related hostPage.data field') . PHP_EOL;
+echo _('  hostPageDom truncate - flush hostPageDom table') . PHP_EOL . PHP_EOL;
+
+echo _('get support: https://github.com/YGGverse/YGGo/issues') . PHP_EOL . PHP_EOL;