html导入信息平台

application/controllers/import.php

@@ -1,27 +1,28 @@
 <?php
 
-if (!defined('BASEPATH'))
+if (!defined('BASEPATH')) {
     exit('No direct script access allowed');
+}
 
 class Import extends CI_Controller
 {
-
+    public function __construct()
-    function __construct()
     {
         parent::__construct();
-        $this->permission->is_admin();
+        //$this->permission->is_admin();
         //$this->output->enable_profiler(TRUE);
-        $this->load->model('Area_model');
+        // $this->load->model('Area_model');
         $this->load->model('Import_model');
-        $this->load->model('InfoContents_model');
+        // $this->load->model('InfoContents_model');
-        $this->load->model('InfoStructures_model');
+        // $this->load->model('InfoStructures_model');
+        $this->load->library('simple_html_dom_lib');
     }
 
     public function index()
     {
         //@set_time_limit(0);
         ini_set('max_execution_time', '0');
-       // $this->Import_model->import_from_other_database(69690,63564);
+        // $this->Import_model->import_from_other_database(69690,63564);
         //$this->Import_model->import_clear();
         //$this->Import_model->import_info();
         //$this->Import_model->import_xxoo();
@@ -36,9 +37,106 @@ class Import extends CI_Controller
              echo $item->ic_url_title . '<br/>';
         }*/
     }
+    
+    public function html_to_database($sitecode)
+    {
+        ini_set('max_execution_time', '0');
+        $file_array=$this->Import_model->not_done_html_info($sitecode);
+        if (empty($file_array)) {
+            echo 'HTML文件都处理完毕!';
+            return;
+        }
+        
+        foreach ($file_array as $item) {
+            $h_id=$item->h_id;
+            $file=$item->h_file;
+            $h_done=1;
+            //$file='D:\wwwroot\gh-html\chinahighlights\xian\tour-planning-md.htm';
+            if (!is_file($file)) {
+                echo '文件不存在:'.$file;
+                continue;
+            }
+            $html_object=file_get_html($file);
+            if (empty($html_object)) {
+                echo '文件加载失败:'.$file;
+                continue;
+            }
+            $h_content=$html_object->find('div[id=mainContentRight]', 0);
+            if (empty($h_content)) {
+                $h_content=$html_object->find('div[id=content]', 0);
+            }
+            if (empty($h_content)) {
+                $h_content=$html_object->find('div[id=mainContent]', 0);
+            }
+            if (empty($h_content)) {
+                $h_content=$html_object->find('div[id=contentRight]', 0);
+            }
+            if (empty($h_content)) {
+                $h_content=$html_object->find('div[id=MainhotelContent]', 0);
+            }
+            if (empty($h_content)) {
+                $h_content=$html_object->find('div.col-md-19', 0);
+            }
+            if (empty($h_content)) {
+                $h_content=$html_object->find('div[id=historyLine]', 0);
+            }
+            if (empty($h_content)) {
+                $h_content=$html_object->find('div.container', 0);
+            }
+            if (empty($h_content)) {
+                $h_content='';
+                $h_done=2;
+            } else {
+                $h_content=$h_content->innertext;
+            }
+
+            $h_url=str_replace('\\', '/', $file);
+            $h_url=str_replace('D:/wwwroot/gh-html/chinahighlights/', 'https://www.chinahighlights.com/', $h_url);
+            $h_url=str_replace('D:/wwwroot/gh-html/asiahighlights/', 'https://www.asiahighlights.com/', $h_url);
+            $h_url=str_replace('index-md.htm', '', $h_url);
+            $h_url=str_replace('-md.htm', '.htm', $h_url);
+            
+            foreach ($html_object->find('meta') as $item) {
+                if ($item->name == 'description' && !empty($item->content)) {
+                    $h_description = $item->content;
+                    break;
+                }
+            }
+            if (empty($h_description)) {
+                $h_description='';
+                $h_done=2;
+            }
+            
+            $h_title=$html_object->find('title', 0);
+            if (empty($h_title)) {
+                $h_title='';
+                $h_done=2;
+            } else {
+                $h_title=$h_title->innertext;
+            }
+            
+            $h_h1=$html_object->find('h1', 0);
+            if (empty($h_h1)) {
+                $h_h1='';
+                $h_done=2;
+            } else {
+                $h_h1=$h_h1->innertext;
+            }
 
-    public function clear_site(){
+            echo 'file:'.$file.'</br>';
-        $this->Import_model->import_clear_site();echo 'clear_site';
+            echo 'h_title:'.$h_title.'</br>';
+            echo 'h_url:'.$h_url.'</br>';
+            echo 'h_description:'.$h_description.'</br>';
+            echo '<hr/>';
+            //echo $h_content;
+            $this->Import_model->update_html_info($h_id, $h_h1, $h_title, $h_url, $h_description, $h_content, $h_done);
+        }
+    }
+
+    public function clear_site()
+    {
+        $this->Import_model->import_clear_site();
+        echo 'clear_site';
     }
 
     public function reset_path()
@@ -51,7 +149,7 @@ class Import extends CI_Controller
     public function to_sqlite()
     {
         $this->load->driver('cache', array('adapter' => 'file', 'backup' => 'file'));
-        $this->HT = $this->load->database('HT', TRUE);
+        $this->HT = $this->load->database('HT', true);
         $sql_array = array();
         $sql_ht = "SELECT ic.ic_id, \n"
                 . "       ic.ic_url, \n"
@@ -91,8 +189,7 @@ class Import extends CI_Controller
                 . "WHERE  ic.ic_ht_area_type = 'c' \n"
                 . "       AND ic.ic_ht_area_id IN (177, 54, 1, 3, 7)";
         $sql_ht_query = $this->HT->query($sql_ht);
-        foreach ($sql_ht_query->result() as $item)
+        foreach ($sql_ht_query->result() as $item) {
-        {
             $sql_content = "INSERT INTO information \n"
                     . "  ( \n"
                     . "	ic_id, ic_url_title, ic_type,ic_content, ic_title,  \n"
@@ -123,12 +220,10 @@ class Import extends CI_Controller
                     . $this->HT->escape($item->is_sort)
                     . "  );";
             $sql_array[] = $sql_content;
-           // echo $sql_content.';';
+            // echo $sql_content.';';
             //$this->cache->save(md5($item->ic_id), $item->ic_content, 60 * 60 * 24 * 30);
         }
-        file_put_contents('import.sql',  implode("\n", $sql_array));
+        file_put_contents('import.sql', implode("\n", $sql_array));
         //print_r($sql_array);
     }
-
 }
-

application/models/import_model.php

@@ -6,9 +6,20 @@ class Import_model extends CI_Model
     function __construct()
     {
         parent::__construct();
-        $this->HT = $this->load->database('HT', TRUE);
+        $this->HT = $this->load->database('INFO', TRUE);
     }
 
+    function not_done_html_info($h_sitecode){
+        $sql="select top 1000 * from html_infos where h_sitecode=? and h_done=2 order by h_id desc";
+        $query=$this->HT->query($sql,$h_sitecode);
+        return $query->result();
+    }
+	
+	function update_html_info($h_id,$h_h1,$h_title,$h_url,$h_description,$h_content,$h_done){
+		$sql=' update html_infos set h_done=? ,h_url=N?, h_description=N? ,h_h1=N? ,h_title=N?,h_content=N? where h_id=? ';
+		return $this->HT->query($sql,array($h_done,$h_url,$h_description,$h_h1,$h_title,$h_content,$h_id));
+	}
+
     function import_clear()
     {
         $this->HT->query('TRUNCATE TABLE infoContents');

lib/simple_html_dom.php

@@ -1,5 +1,5 @@
 <?php
-/**
+/**  https://simplehtmldom.sourceforge.io/manual.htm
  * Website: http://sourceforge.net/projects/simplehtmldom/
  * Acknowledge: Jose Solorzano (https://sourceforge.net/projects/php-html/)
  * Contributions by: