From a8ffe143493730fb0b45b19506a5e0915c30540f Mon Sep 17 00:00:00 2001
From: ghost <localhost>
Date: Thu, 17 Aug 2023 14:58:06 +0300
Subject: [PATCH] implement 'hostPage add' CLI method

---
 README.md        |  2 ++
 src/cli/yggo.php | 86 ++++++++++++++++++++++++++++++++++++++++++++++--
 2 files changed, 86 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 1a558a7..c56a0c3 100644
--- a/README.md
+++ b/README.md
@@ -49,6 +49,7 @@ sphinxsearch
 * Configuration examples presented at `/config` folder
 * Make sure `/src/storage/cache`, `/src/storage/tmp`, `/src/storage/snap` folders are writable
 * Set up the `/src/crontab` by following [example](https://github.com/YGGverse/YGGo/blob/main/config/crontab.txt)
+* To start crawler, add at least one initial URL using search form or CLI
 
 #### JSON API
 
@@ -241,6 +242,7 @@ _*CLI interface still under construction, use it for your own risk!_
   + [ ] delete
   + [ ] flush
 * [x] hostPage
+  + [x] add
   + [x] rank
     + [x] reindex
 * [x] hostPageSnap
diff --git a/src/cli/yggo.php b/src/cli/yggo.php
index 23928cb..3d6dd60 100644
--- a/src/cli/yggo.php
+++ b/src/cli/yggo.php
@@ -6,6 +6,7 @@ require_once(__DIR__ . '/../library/cli.php');
 require_once(__DIR__ . '/../library/mysql.php');
 require_once(__DIR__ . '/../library/filter.php');
 require_once(__DIR__ . '/../library/ftp.php');
+require_once(__DIR__ . '/../library/helper.php');
 require_once __DIR__ . '/../../vendor/autoload.php';
 
 // CLI only to prevent https server connection timeout
@@ -42,8 +43,29 @@ if (false === sem_acquire($semaphore, true)) {
   exit;
 }
 
-// Connect database
-$db = new MySQL(DB_HOST, DB_PORT, DB_NAME, DB_USERNAME, DB_PASSWORD);
+// Connect DB
+try {
+
+  $db = new MySQL(DB_HOST, DB_PORT, DB_NAME, DB_USERNAME, DB_PASSWORD);
+
+} catch(Exception $e) {
+
+  var_dump($e);
+
+  exit;
+}
+
+// Connect Yggverse\Cache\Memory
+try {
+
+  $memory = new Yggverse\Cache\Memory(MEMCACHED_HOST, MEMCACHED_PORT, MEMCACHED_NAMESPACE, MEMCACHED_TIMEOUT + time());
+
+} catch(Exception $e) {
+
+  var_dump($e);
+
+  exit;
+}
 
 // CLI begin
 if (!empty($argv[1])) {
@@ -401,6 +423,65 @@ if (!empty($argv[1])) {
 
         switch ($argv[2]) {
 
+          case 'add':
+
+            if (empty($argv[3])) {
+
+              CLI::danger('URL required');
+
+              exit;
+            }
+
+            if (false === Yggverse\Parser\Url::is($argv[3])) {
+
+              CLI::danger('URL address invalid');
+
+              exit;
+            }
+
+            try {
+
+              $db->beginTransaction();
+
+              if ($linkToDBresult = Helper::addLinkToDB($db, $memory, $argv[3])) {
+
+                if (count($linkToDBresult->new->hostPageId)) {
+
+                  CLI::success('URL successfully registered in the crawler queue!');
+
+                  $db->commit();
+
+                  exit;
+
+                } else {
+
+                  CLI::warning('URL already registered in the crawler queue!');
+
+                   $db->rollBack();
+
+                  exit;
+                }
+
+              } else {
+
+                CLI::danger('URL address not supported by this host rules!');
+
+                $db->rollBack();
+
+                exit;
+              }
+
+            } catch(Exception $e){
+
+              var_dump($e);
+
+              $db->rollBack();
+
+              exit;
+            }
+
+          break;
+
           case 'rank':
 
             if (!empty($argv[3])) {
@@ -675,6 +756,7 @@ CLI::default('    crawl                        - execute step in crawler queue')
 CLI::default('    clean                        - execute step in cleaner queue');
 CLI::break();
 CLI::default('  hostPage                       ');
+CLI::default('    add [URL]                    - register new address in the crawl queue');
 CLI::default('    rank                         ');
 CLI::default('      reindex                    - reindex hostPage.rank fields');
 CLI::break();