├── README.md
├── ta-article-setting.php
├── ta-functions.php
└── targetany.php


/README.md:
--------------------------------------------------------------------------------
 1 | # newcrawler-wordpress
 2 | NewCrawler Wordpress Plugin
 3 | 
 4 | 1.Download Plugin
 5 | 
 6 | >Clone or download > Download ZIP
 7 | 
 8 | 2.Install Plugin
 9 | 
10 | >Wordpress > Plugins > Add New > Upload Plugin > (Select file)Install Now
11 | 


--------------------------------------------------------------------------------
/ta-article-setting.php:
--------------------------------------------------------------------------------
 1 | <style>
 2 |     .postbox h3 {
 3 |         font-family: Georgia, "Times New Roman", "Bitstream Charter", Times, serif;
 4 |         font-size: 15px;
 5 |         padding: 10px 10px;
 6 |         margin: 0;
 7 |         line-height: 1;
 8 |     }
 9 |     #ta-table td{
10 |         height:30px;
11 |     }
12 | </style>
13 | <script type="text/javascript">
14 |     jQuery(function ($) {
15 |         $("h3.hndle").click(function () {
16 |             $(this).next(".inside").slideToggle('fast');
17 |         });
18 |     });
19 | </script>
20 | 
21 | <?php
22 | $basicWebAddress = str_replace('\\','/',$_SERVER['HTTP_HOST'].str_replace('/wp-admin', '', dirname($_SERVER['SCRIPT_NAME'])));
23 | 
24 | if (isset($_POST['submit1']) && $_POST['submit1'] != '') {
25 |     update_option('ta_password', $_POST['ta_password']);
26 | 
27 |     echo '<div id="message" class="updated fade"><p>更新成功！</p></div>';
28 | }
29 | ?>
30 | <div class="wrap">
31 |     <h2>鸟巢采集器</h2>
32 |     <br/>
33 | 
34 | 
35 |     <form id="myform" method="post" action="admin.php?page=newcrawler-wordpress-master/ta-article-setting.php">
36 |         <div class="postbox">
37 |             <h3 class="hndle" style="cursor:pointer;">发布设置</h3>
38 |             <div class="inside">
39 |                 <table width="100%" id="ta-table">
40 |                     <tr>
41 |                         <td width="15%">插件版本:</td>
42 |                         <td>鸟巢采集器WordPress发布插件 v3.1.3</td>
43 | 
44 |                     <span class="gray">( https://github.com/speed/newcrawler-wordpress )</span>
45 |                     </tr>
46 |                     <tr>
47 |                         <td width="10%">官网地址:</td>
48 |                         <td><a href="http://www.newcrawler.com" target="_blank">www.newcrawler.com</a></td>
49 |                     </tr>
50 | 
51 |                     <?php $ta_password = get_option('ta_password', "NewCrawler"); ?>
52 |                     <tr>
53 |                         <td width="10%">发布密码:</td>
54 |                         <td><input type="text" name="ta_password" style="color:black;width:300px" value="<?php echo $ta_password; ?>" />
55 |                             <span class="gray">( 建议修改成自己的常用密码，并填入鸟巢采集器导出到 WordPress 选项中的发布密码输入框内 )</span>
56 |                         </td>
57 |                     </tr>
58 |                     <tr>
59 |                         <td width="10%">您网站访问地址为:</td>
60 |                         <td><input type="text" name="ta_web" disabled="disabled" readonly="readonly" style="width:300px" value="<?php echo $basicWebAddress;?>" />
61 |                             <span class="gray">( 此地址为您网站访问地址，您可以复制该链接并填入鸟巢采集器导出到 WordPress 发布接口配置中的发布网址输入框内)</span>
62 |                         </td>
63 |                     </tr>
64 |                     <tr>
65 |                         <td>
66 |                             <input type="submit" class="button-primary"  name="submit1"  value="保存更改" />
67 |                         </td>
68 |                         <td>
69 |                            
70 |                         </td>
71 |                     </tr>
72 |                 </table>
73 |             </div>
74 |     </form>
75 | </div>
76 | </div>


--------------------------------------------------------------------------------
/ta-functions.php:
--------------------------------------------------------------------------------
  1 | <?php
  2 | 
  3 | define('TA_ERROR_NONE', 1);
  4 | define('TA_ERROR_ERROR', 2);
  5 | define('TA_ERROR_PLUGIN_ERROR', 3);
  6 | define('TA_ERROR_INVALID_PWD', 100);
  7 | define('TA_ERROR_MISSING_FIELD', 101);
  8 | 
  9 | function ta_success($data = "", $message = "") {
 10 |     ta_result(1, $data, $message);
 11 | }
 12 | 
 13 | function ta_fail($code = 2, $data = "", $message = "") {
 14 |     ta_result($code, $data, $message);
 15 | }
 16 | 
 17 | function ta_result($result = 1, $data = "", $message = "") {
 18 |     die(json_encode(array("result" => $result, "data" => $data, "message" => urlencode($message))));
 19 | }
 20 | 
 21 | // Get Real Url for 302 URL
 22 | function ta_redirect_url($url) {
 23 |     if (empty($url)) {
 24 |         return false;
 25 |     }
 26 |     if(stripos($url, "static.newcrawler.com") === false){
 27 |     	//if not hosted by newcrawler
 28 |     	return array('realurl' => $url, 'referer' => "");
 29 |     }
 30 |     $result = ta_curl_headers($url.'-dl');
 31 |     if ($result !== false && strpos($result, "302 Moved Temporarily")) {
 32 |         $headers = preg_split("/\r\n+/", $result);
 33 |         if (is_array($headers)) {
 34 |             $real_url = null;
 35 |             $referer = '';
 36 |             $suffix = '';
 37 |             foreach ($headers as $header) {
 38 |                 $header = trim($header);
 39 |                 $locpos = stripos($header, "location");
 40 |                 $refererpos = stripos($header, "X-Referer");
 41 |                 if ($locpos === 0) {
 42 |                   $pp = strpos($header, ":");
 43 |                   $real_url = trim(substr($header, $pp + 1));
 44 |                 }else if ($refererpos === 0) {
 45 |                   $pp = strpos($header, ":");
 46 |                   $referer = trim(substr($header, $pp + 1));
 47 |                 }
 48 |             }
 49 |             if (!empty($real_url) && stripos($real_url, "http") === 0) {
 50 |                 return array('realurl' => $real_url, 'referer' => $referer);
 51 |             }
 52 |         }
 53 |     }
 54 |     return false;
 55 | }
 56 | 
 57 | function ta_curl_headers($url) {
 58 |     // 初始化Curl
 59 |     $ch = curl_init();
 60 |     // 开启header显示
 61 |     curl_setopt($ch, CURLOPT_HEADER, true);
 62 |     // 不输出网页内容
 63 |     curl_setopt($ch, CURLOPT_NOBODY, true);
 64 |     // 禁止自动输出内容
 65 |     curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 66 |     // 自动跳转
 67 |     curl_setopt($ch, CURLOPT_FOLLOWLOCATION, false);
 68 |     // 跳转时自动设置来源地址
 69 |     curl_setopt($ch, CURLOPT_AUTOREFERER, true);
 70 |     // 超时时间
 71 |     curl_setopt($ch, CURLOPT_TIMEOUT, 60);
 72 |     // 设置URL
 73 |     curl_setopt($ch, CURLOPT_URL, $url);
 74 |     // 关闭SSL证书验证
 75 |     curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
 76 |     // 返回结果
 77 |     return curl_exec($ch);
 78 | }
 79 | 
 80 | function ta_log($data) {
 81 |     if ($data && (is_array($data) || is_object($data))) {
 82 |         if (method_exists($data, 'jsonSerialize')) {
 83 |             $data = $data->jsonSerialize();
 84 |         }
 85 |         $str = json_encode($data);
 86 |     } else {
 87 |         $str = $data;
 88 |     }
 89 |     $myfile = fopen("ta_log.txt", "a") or die("Unable to open file!");
 90 |     fwrite($myfile, $str);
 91 |     fclose($myfile);
 92 | }
 93 | 
 94 | function ta_random_ip(){
 95 | 	$ip_long = array(
 96 | 		array('607649792', '608174079'), //36.56.0.0-36.63.255.255
 97 | 		array('1038614528', '1039007743'), //61.232.0.0-61.237.255.255
 98 | 		array('1783627776', '1784676351'), //106.80.0.0-106.95.255.255
 99 | 		array('2035023872', '2035154943'), //121.76.0.0-121.77.255.255
100 | 		array('2078801920', '2079064063'), //123.232.0.0-123.235.255.255
101 | 		array('-1950089216', '-1948778497'), //139.196.0.0-139.215.255.255
102 | 		array('-1425539072', '-1425014785'), //171.8.0.0-171.15.255.255
103 | 		array('-1236271104', '-1235419137'), //182.80.0.0-182.92.255.255
104 | 		array('-770113536', '-768606209'), //210.25.0.0-210.47.255.255
105 | 		array('-569376768', '-564133889'), //222.16.0.0-222.95.255.255
106 | 	);
107 | 	$rand_key = mt_rand(0, 9);
108 | 	$ip = long2ip(mt_rand($ip_long[$rand_key][0], $ip_long[$rand_key][1]));
109 | 	return $ip;	
110 | }
111 | 


--------------------------------------------------------------------------------
/targetany.php:
--------------------------------------------------------------------------------
  1 | <?php
  2 | 
  3 | /*
  4 |   Plugin Name: 鸟巢采集器
  5 |   Plugin URI: https://github.com/speed/newcrawler-wordpress
  6 |   Description: 鸟巢采集器WordPress发布插件，实现采集数据自动发布到WordPress。
  7 |   Version: 3.1.3
  8 |   Author: NewCrawler
  9 |   Author URI: http://www.newcrawler.com
 10 |   License: GPL
 11 |  */
 12 | define('TA_PATH', WP_PLUGIN_DIR . '/newcrawler-wordpress-master');
 13 | 
 14 | include TA_PATH . '/ta-functions.php';
 15 | 
 16 | global $table_prefix, $ta_tbl_setting;
 17 | 
 18 | $ta_tbl_setting = $table_prefix . 'ta_publish_setting';
 19 | 
 20 | global $ta_db_version;
 21 | $ta_db_version = '3.1.3';
 22 | 
 23 | if (is_admin()) {
 24 |     /*  利用 admin_menu 钩子，添加菜单 */
 25 |     add_action('admin_menu', 'ta_menu');
 26 | }
 27 | 
 28 | function ta_menu() {
 29 |     if (function_exists('add_menu_page')) {
 30 |         add_menu_page('鸟巢采集器', '鸟巢采集器', 'administrator', 'newcrawler-wordpress-master/ta-article-setting.php', '', 'dashicons-share-alt');
 31 |     }
 32 | }
 33 | 
 34 | add_action('init', 'ta_auto_post');
 35 | 
 36 | function ta_install() {
 37 |     $old_db_version = get_option('ta_db_version');
 38 |     update_option("ta_db_version", $ta_db_version);
 39 | }
 40 | 
 41 | register_activation_hook(__FILE__, 'ta_install');
 42 | 
 43 | function ta_auto_post() {
 44 |     $ta_password = get_option('ta_password', "NewCrawler");
 45 |     if ($_GET["__ta"] == "post") {
 46 | 
 47 |         if (empty($_POST['__sign']) || $_POST['__sign'] != $ta_password) {
 48 |             ta_fail(TA_ERROR_INVALID_PWD, "password is wrong", "发布密码填写错误");
 49 |         }
 50 |         $title = $_POST["article_title"];
 51 |         $content = $_POST["article_content"];
 52 | 
 53 |         if (empty($content) && empty($title)) {
 54 |             ta_fail(TA_ERROR_MISSING_FIELD, "article_content and article_title are both empty", "文章内容和标题不能都为空！");
 55 |         }
 56 | 
 57 |         $postStatus = 'publish';
 58 |         if (isset($_POST["postStatus"]) && in_array($_POST["postStatus"], array('publish', 'draft'))) {
 59 |             $postStatus = $_POST["postStatus"];
 60 |         }
 61 | 
 62 |         $postPassword = '';
 63 |         if (isset($_POST["accessword"]) && $_POST["accessword"]) {
 64 |             $postPassword = $_POST["accessword"];
 65 |         }
 66 | 
 67 |         $my_post = array(
 68 |             'post_password' => $postPassword,
 69 |             'post_status' => $postStatus,
 70 |             'post_author' => 1
 71 |         );
 72 |         if (!empty($title)) {
 73 |             $my_post['post_title'] = htmlspecialchars_decode($title);
 74 |         }
 75 |         if (!empty($content)) {
 76 |             $my_post['post_content'] = htmlspecialchars_decode($content);
 77 |         }
 78 | 
 79 | 
 80 |         $publish_time = intval($_POST["article_publish_time"]);
 81 |         if (!empty($publish_time)) {
 82 |             $my_post['post_date'] = date("Y-m-d", $publish_time);
 83 |         } else {
 84 |             $my_post['post_date'] = date("Y-m-d", time());
 85 |         }
 86 | 
 87 |         $author = htmlspecialchars_decode($_POST["article_author"]);
 88 | 
 89 |         if (!empty($author)) {
 90 |             $md5author = substr(md5($author), 8, 16);
 91 |             $user_id = username_exists($md5author);
 92 |             if (!$user_id) {
 93 |                 $random_password = wp_generate_password();
 94 |                 $userdata = array(
 95 |                     'user_login' => $md5author,
 96 |                     'user_pass' => $random_password,
 97 |                     'display_name' => $author,
 98 |                 );
 99 | 
100 |                 $user_id = wp_insert_user($userdata);
101 |                 if (is_wp_error($user_id)) {
102 |                     $user_id = 0;
103 |                 }
104 |             }
105 |             if ($user_id) {
106 |                 $my_post['post_author'] = $user_id;
107 |             }
108 |         }
109 |         $article_categories = $_POST["article_categories"];
110 |         if (!empty($article_categories)) {
111 |             $rawCates = stripslashes($article_categories);
112 |             $cates = json_decode($rawCates);
113 |             if (is_array($cates)) {
114 |                 $post_cates = array();
115 |                 foreach ($cates as $cate) {
116 |                     $term = term_exists($cate, "category");
117 | 
118 |                     if ($term === 0 || $term === null) {
119 |                         $term = wp_insert_term($cate, "category");
120 |                     }
121 |                     if ($term !== 0 && $term !== null) {
122 |                         array_push($post_cates, intval($term["term_id"]));
123 |                     }
124 |                 }
125 |                 if (count($post_cates) > 0) {
126 |                     $my_post['post_category'] = $post_cates;
127 |                 }
128 |             }
129 |         }
130 | 
131 |         $article_topics = $_POST["article_topics"];
132 |         if (!empty($article_topics)) {
133 |             $rawTags = stripslashes($article_topics);
134 |             $tags = json_decode($rawTags);
135 | 
136 |             if (is_array($tags)) {
137 |                 $post_tags = array();
138 |                 $term = null;
139 |                 foreach ($tags as $tag) {
140 |                     $term = term_exists($tag, "post_tag");
141 | 
142 |                     if ($term === 0 || $term === null) {
143 |                         $term = wp_insert_term($tag, "post_tag");
144 |                     }
145 |                     if ($term !== 0 && $term !== null) {
146 |                         array_push($post_tags, intval($term["term_id"]));
147 |                     }
148 |                 }
149 |                 if (count($post_tags) > 0) {
150 |                     $my_post['tags_input'] = $post_tags;
151 |                 }
152 |             }
153 |         }
154 | 
155 |         // Insert the post into the database
156 |         kses_remove_filters();
157 |         $post_id = wp_insert_post($my_post); //wp_insert_user wp_insert_comment wp_insert_category
158 |         kses_init_filters();
159 | 
160 |         if (empty($post_id)) {
161 |             ta_fail(TA_ERROR_ERROR, "Empty Post ID", "插入系统失败");
162 |         }
163 | 
164 | 				if(!empty($_POST["article_thumbnail"])){
165 | 	        $image_url = ta_redirect_url($_POST["article_thumbnail"]); // Define the image URL here
166 | 	        //创建thumbnail
167 | 	        // Add Featured Image to Post
168 | 	        if ($image_url !== false && !empty($post_id)) {
169 | 	            $upload_dir = wp_upload_dir(); // Set upload folder
170 | 	            $image_data = file_get_contents($image_url['realurl']); // Get image data
171 | 	            $suffix = "jpg";
172 | 	            $filename = md5($image_url['realurl']) . ".". $suffix; // Create image file name
173 | 	            // Check folder permission and define file location
174 | 	            if (wp_mkdir_p($upload_dir['path'])) {
175 | 	                $file = $upload_dir['path'] . '/' . $filename;
176 | 	            } else {
177 | 	                $file = $upload_dir['basedir'] . '/' . $filename;
178 | 	            }
179 | 
180 | 	            // Create the image  file on the server
181 | 	            file_put_contents($file, $image_data);
182 | 	            if (file_exists($file)) {
183 | 	                //文件存在 在做插入
184 | 	                // Check image file type
185 | 	                $wp_filetype = wp_check_filetype($filename, null);
186 | 
187 | 	                // Set attachment data
188 | 	                $attachment = array(
189 | 	                    'post_mime_type' => $wp_filetype['type'],
190 | 	                    'post_title' => sanitize_file_name($filename),
191 | 	                    'post_content' => '',
192 | 	                    'post_status' => 'inherit'
193 | 	                );
194 | 
195 | 	                // Create the attachment
196 | 	                $attach_id = wp_insert_attachment($attachment, $file, $post_id);
197 | 
198 | 	                // Include image.php
199 | 	                require_once(ABSPATH . 'wp-admin/includes/image.php');
200 | 
201 | 	                // Define attachment metadata
202 | 	                $attach_data = wp_generate_attachment_metadata($attach_id, $file);
203 | 
204 | 	                // Assign metadata to attachment
205 | 	                wp_update_attachment_metadata($attach_id, $attach_data);
206 | 
207 | 	                // And finally assign featured image to post
208 | 	                set_post_thumbnail($post_id, $attach_id);
209 | 	            }
210 | 	        }
211 | 	      }
212 | 
213 | 				//插入评论
214 | 				$comment_json = preg_replace("/[\r\n\t]/", '', $_POST['article_comment']);
215 | 				//ta_fail(TA_ERROR_ERROR, "DEBUG", "".$comment_json);
216 | 				$article_comment = json_decode(stripslashes($comment_json), true);
217 | 
218 | 				if($article_comment != null && is_array($article_comment)){
219 | 					foreach($article_comment as $comment){
220 | 						$content = $comment["article_comment_content"];
221 | 						if(!empty($content)){
222 | 							//内容不是空
223 | 
224 | 							$pub_time = $comment["article_comment_publish_time"];
225 | 							if(empty($pub_time)){
226 | 									$pub_time = time();
227 | 							}
228 | 
229 | 							$cdata = array(
230 | 							    'comment_post_ID' => $post_id,
231 | 							    'comment_content' => $content,
232 | 							    'comment_type' => '',
233 | 							    'comment_parent' => 0,
234 | 							    'comment_agent' => 'Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.0.10) Gecko/2009042316 Firefox/3.0.10 (.NET CLR 3.5.30729)',
235 | 							    'comment_date' => date("Y-m-d H:i:s",$pub_time),
236 | 							    'comment_approved' => 1,
237 | 							);
238 | 
239 | 							$cauthor = $comment["article_comment_author"];
240 | 
241 | 			        if (!empty($cauthor)) {
242 | 			            $cmd5author = substr(md5($cauthor), 8, 16);
243 | 			            $cuser_id = username_exists($cmd5author);
244 | 			            if (!$cuser_id) {
245 | 			                $random_password = wp_generate_password();
246 | 			                $cuserdata = array(
247 | 			                    'user_login' => $cmd5author,
248 | 			                    'user_pass' => $random_password,
249 | 			                    'display_name' => $cauthor,
250 | 			                );
251 | 
252 | 			                $cuser_id = wp_insert_user($cuserdata);
253 | 			                if (is_wp_error($cuser_id)) {
254 | 			                    $cuser_id = 0;
255 | 			                }
256 | 			            }
257 | 			            if($cuser_id != 0){
258 | 			            	$cdata["user_id"] = $cuser_id;
259 | 			            }
260 | 			            $cdata["comment_author"] = $cauthor;
261 | 			            $cdata['comment_author_IP'] = ta_random_ip();
262 | 			        }
263 | 
264 | 							wp_insert_comment($cdata);
265 | 						}
266 | 					}
267 | 				}
268 | 
269 | 
270 | 
271 |         ta_success(array("url" => get_home_url() . "/?p=" . $post_id));
272 |     } else if ($_GET["__ta"] == "details") {
273 |         if (empty($_POST['__sign']) || $_POST['__sign'] != $ta_password) {
274 |             ta_fail(TA_ERROR_INVALID_PWD, "password is wrong", "发布密码填写错误");
275 |         }
276 |         $ret = array(array("value" => "", "text" => urlencode("爬取到的分类")));
277 | 
278 |         if ($_POST["type"] === "cate") {
279 |             $cates = get_terms('category', 'orderby=count&hide_empty=0');
280 | 
281 |             foreach ($cates as $cate) {
282 |                 array_push($ret, array("value" => urlencode($cate->name), "text" => urlencode($cate->name)));
283 |             }
284 |         }
285 | 
286 |         ta_success($ret);
287 |     } else if ($_GET["__ta"] == "version") {
288 |         global $wp_version;
289 |         if (empty($_POST['__sign']) || $_POST['__sign'] != $ta_password) {
290 |             ta_fail(TA_ERROR_INVALID_PWD, "password is wrong", "发布密码填写错误");
291 |         }
292 |         $info = array(
293 |             'protocol' => '1',
294 |             'protocolVersion' => '1',
295 |             'supportStdVersion' => array(
296 |                 'article' => '1.0.0',
297 |                 'question' => '1.0.0'
298 |             ),
299 |             'php' => PHP_VERSION,
300 |             'supportVersion' => 'wp 4.5.1',
301 |             'version' => '3.1.3',
302 |             'pubVersion' => '3.1.3',
303 |             'versionDetail' => array('wp' => $wp_version),
304 |             'otherInfo' => array()
305 |         );
306 |         ta_success($info);
307 |     }
308 | }


--------------------------------------------------------------------------------