作者 刘锟

采集单词替换

@@ -156,6 +156,10 @@ class HtmlCollect extends Command @@ -156,6 +156,10 @@ class HtmlCollect extends Command
156 return true; 156 return true;
157 } 157 }
158 158
  159 + if ($project_id == 437) {
  160 + //437项目单词替换
  161 + $html = str_replace('Forenia', 'Tourle', $html);
  162 + }
159 $collect_info->html = $html; 163 $collect_info->html = $html;
160 $collect_info->status = CollectTask::STATUS_COM; 164 $collect_info->status = CollectTask::STATUS_COM;
161 $collect_info->save(); 165 $collect_info->save();
@@ -123,13 +123,20 @@ class ProjectUpdate extends Command @@ -123,13 +123,20 @@ class ProjectUpdate extends Command
123 $page_list = array_column($data_page, 'path'); 123 $page_list = array_column($data_page, 'path');
124 } 124 }
125 125
  126 + //获取需要替换的内容
  127 + $replace = [];
  128 + if ($project_id == 437) {
  129 + //437项目单词替换
  130 + $replace = ['Forenia' => 'Tourle'];
  131 + }
  132 +
126 //设置数据库 133 //设置数据库
127 $project = ProjectServer::useProject($project_id); 134 $project = ProjectServer::useProject($project_id);
128 if ($project) { 135 if ($project) {
129 if ($api_type == 'category') { 136 if ($api_type == 'category') {
130 //产品分类 137 //产品分类
131 $url = $api_url . '?' . http_build_query(['w' => 'category']); 138 $url = $api_url . '?' . http_build_query(['w' => 'category']);
132 - $data = curl_c($url); 139 + $data = curl_c($url, true, $replace);
133 if (isset($data['code']) && $data['code'] == 200) { 140 if (isset($data['code']) && $data['code'] == 200) {
134 $items = $data['data'] ?? []; 141 $items = $data['data'] ?? [];
135 $this->category_insert($project_id, $items, 0, $domain_arr['host'], $web_url_domain, $home_url); 142 $this->category_insert($project_id, $items, 0, $domain_arr['host'], $web_url_domain, $home_url);
@@ -139,7 +146,7 @@ class ProjectUpdate extends Command @@ -139,7 +146,7 @@ class ProjectUpdate extends Command
139 } elseif ($api_type == 'category_news') { 146 } elseif ($api_type == 'category_news') {
140 // 新闻分类 147 // 新闻分类
141 $url = $api_url . '?' . http_build_query(['w' => 'category_news']); 148 $url = $api_url . '?' . http_build_query(['w' => 'category_news']);
142 - $data = curl_c($url); 149 + $data = curl_c($url, true, $replace);
143 if (isset($data['code']) && $data['code'] == 200) { 150 if (isset($data['code']) && $data['code'] == 200) {
144 $items = $data['data'] ?? []; 151 $items = $data['data'] ?? [];
145 $this->category_news_insert($project_id, $items, 0); 152 $this->category_news_insert($project_id, $items, 0);
@@ -156,7 +163,7 @@ class ProjectUpdate extends Command @@ -156,7 +163,7 @@ class ProjectUpdate extends Command
156 $total_page = ceil($count / $page_size); 163 $total_page = ceil($count / $page_size);
157 for ($page = 1; $page <= $total_page; $page++) { 164 for ($page = 1; $page <= $total_page; $page++) {
158 $url_page = $api_url . '?' . http_build_query(['w' => 'tag', 'page' => $page, 'pagesize' => $page_size]); 165 $url_page = $api_url . '?' . http_build_query(['w' => 'tag', 'page' => $page, 'pagesize' => $page_size]);
159 - $data_page = curl_c($url_page); 166 + $data_page = curl_c($url_page, true, $replace);
160 if (isset($data_page['code']) && $data_page['code'] == 200) { 167 if (isset($data_page['code']) && $data_page['code'] == 200) {
161 $items = $data_page['data']['data'] ?? []; 168 $items = $data_page['data']['data'] ?? [];
162 169
@@ -287,7 +294,7 @@ class ProjectUpdate extends Command @@ -287,7 +294,7 @@ class ProjectUpdate extends Command
287 $total_page = ceil($count / $page_size); 294 $total_page = ceil($count / $page_size);
288 for ($page = 1; $page <= $total_page; $page++) { 295 for ($page = 1; $page <= $total_page; $page++) {
289 $url_page = $api_url . '?' . http_build_query(['w' => 'post', 'page' => $page, 'pagesize' => $page_size]); 296 $url_page = $api_url . '?' . http_build_query(['w' => 'post', 'page' => $page, 'pagesize' => $page_size]);
290 - $data_page = curl_c($url_page); 297 + $data_page = curl_c($url_page, true, $replace);
291 if (isset($data_page['code']) && $data_page['code'] == 200) { 298 if (isset($data_page['code']) && $data_page['code'] == 200) {
292 $items = $data_page['data']['data'] ?? []; 299 $items = $data_page['data']['data'] ?? [];
293 300
@@ -522,7 +529,7 @@ class ProjectUpdate extends Command @@ -522,7 +529,7 @@ class ProjectUpdate extends Command
522 $total_page = ceil($count / $page_size); 529 $total_page = ceil($count / $page_size);
523 for ($page = 1; $page <= $total_page; $page++) { 530 for ($page = 1; $page <= $total_page; $page++) {
524 $url_page = $api_url . '?' . http_build_query(['w' => $api_type, 'page' => $page, 'pagesize' => $page_size]); 531 $url_page = $api_url . '?' . http_build_query(['w' => $api_type, 'page' => $page, 'pagesize' => $page_size]);
525 - $data_page = curl_c($url_page); 532 + $data_page = curl_c($url_page, true, $replace);
526 if (isset($data_page['code']) && $data_page['code'] == 200) { 533 if (isset($data_page['code']) && $data_page['code'] == 200) {
527 $items = $data_page['data']['data'] ?? []; 534 $items = $data_page['data']['data'] ?? [];
528 535
@@ -616,7 +623,7 @@ class ProjectUpdate extends Command @@ -616,7 +623,7 @@ class ProjectUpdate extends Command
616 $total_page = ceil($count / $page_size); 623 $total_page = ceil($count / $page_size);
617 for ($page = 1; $page <= $total_page; $page++) { 624 for ($page = 1; $page <= $total_page; $page++) {
618 $url_page = $api_url . '?' . http_build_query(['w' => 'page', 'page' => $page, 'pagesize' => $page_size]); 625 $url_page = $api_url . '?' . http_build_query(['w' => 'page', 'page' => $page, 'pagesize' => $page_size]);
619 - $data_page = curl_c($url_page); 626 + $data_page = curl_c($url_page, true, $replace);
620 if (isset($data_page['code']) && $data_page['code'] == 200) { 627 if (isset($data_page['code']) && $data_page['code'] == 200) {
621 $items = $data_page['data']['data'] ?? []; 628 $items = $data_page['data']['data'] ?? [];
622 629
@@ -706,7 +713,7 @@ class ProjectUpdate extends Command @@ -706,7 +713,7 @@ class ProjectUpdate extends Command
706 $total_page = ceil($count / $page_size); 713 $total_page = ceil($count / $page_size);
707 for ($page = 1; $page <= $total_page; $page++) { 714 for ($page = 1; $page <= $total_page; $page++) {
708 $url_page = $api_url . '?' . http_build_query(['w' => $api_type, 'page' => $page, 'pagesize' => $page_size]); 715 $url_page = $api_url . '?' . http_build_query(['w' => $api_type, 'page' => $page, 'pagesize' => $page_size]);
709 - $data_page = curl_c($url_page); 716 + $data_page = curl_c($url_page, true, $replace);
710 if (isset($data_page['code']) && $data_page['code'] == 200) { 717 if (isset($data_page['code']) && $data_page['code'] == 200) {
711 $items = $data_page['data']['data'] ?? []; 718 $items = $data_page['data']['data'] ?? [];
712 719
@@ -170,12 +170,13 @@ function contains_russian($text) { @@ -170,12 +170,13 @@ function contains_russian($text) {
170 if (!function_exists('curl_c')) { 170 if (!function_exists('curl_c')) {
171 /** 171 /**
172 * @param $url 172 * @param $url
  173 + * @param array $replace
173 * @param $is_array 174 * @param $is_array
174 * @return [] 175 * @return []
175 * @author Akun 176 * @author Akun
176 * @date 2023/11/22 11:33 177 * @date 2023/11/22 11:33
177 */ 178 */
178 - function curl_c($url,$is_array=true){ 179 + function curl_c($url,$is_array=true,$replace=[]){
179 $header = array( 180 $header = array(
180 'Expect:', 181 'Expect:',
181 'Content-Type: application/json; charset=utf-8' 182 'Content-Type: application/json; charset=utf-8'
@@ -197,6 +198,11 @@ if (!function_exists('curl_c')) { @@ -197,6 +198,11 @@ if (!function_exists('curl_c')) {
197 $http_code = curl_getinfo($ch,CURLINFO_HTTP_CODE); 198 $http_code = curl_getinfo($ch,CURLINFO_HTTP_CODE);
198 curl_close($ch); 199 curl_close($ch);
199 if($http_code == 200){ 200 if($http_code == 200){
  201 + if(!empty($replace)){
  202 + foreach ($replace as $k=>$v){
  203 + $content = str_replace($k,$v,$content);
  204 + }
  205 + }
200 return $is_array ? json_decode($content, true) : $content; 206 return $is_array ? json_decode($content, true) : $content;
201 }else{ 207 }else{
202 return false; 208 return false;