From e39126d54ca0f0720605490598cd4e76b9aae51e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E6=9D=A8=E5=BF=97?= <vcooli@qq.com>
Date: Mon, 2 Feb 2026 14:53:38 +0800
Subject: [PATCH] up

---
 app/controller/Crawler.php     |  7 +++---
 app/service/CrawlerService.php | 44 +++++++++++++++++++++++++---------
 2 files changed, 36 insertions(+), 15 deletions(-)

diff --git a/app/controller/Crawler.php b/app/controller/Crawler.php
index ba34d08..5edfbc5 100644
--- a/app/controller/Crawler.php
+++ b/app/controller/Crawler.php
@@ -344,12 +344,11 @@ class Crawler extends BaseController
             if (is_array($treeData)) {
                 foreach ($treeData as $item) {
                     if (isset($item['CODE']) && !empty($item['CODE'])) {
-                        $code = $item['CODE'];
-                        // 跳过152开头的职位代码
-                        if (strpos($code, '152') === 0) {
+                        // 跳过带 nocheck: true 的项（如父级/分组节点，不可选）
+                        if (!empty($item['nocheck'])) {
                             continue;
                         }
-                        $zwdmList[] = $code;
+                        $zwdmList[] = $item['CODE'];
                     }
                 }
             }
diff --git a/app/service/CrawlerService.php b/app/service/CrawlerService.php
index 528ebfd..f0e4424 100644
--- a/app/service/CrawlerService.php
+++ b/app/service/CrawlerService.php
@@ -400,28 +400,50 @@ class CrawlerService
     }
 
     /**
-     * 从HTML中提取地区选项
+     * 从HTML中提取地区选项（仅从地市/地区相关的 select 中提取，并过滤占位项与乱码）
      * @param string $html HTML内容
      * @return array
      */
     public function extractDsdmOptions(string $html): array
     {
         $options = [];
-        
-        // 使用正则表达式提取option标签
-        preg_match_all('/<option\s+value="([^"]+)"[^>]*>([^<]+)<\/option>/i', $html, $matches);
-        
+        $placeholderValues = ['0', '-1', ''];
+        $placeholderTexts = ['请选择', '请选择地区', '请选择地市', '全部', ''];
+
+        // 先尝试只匹配「地市」相关 select 内的 option，减少误匹配
+        if (preg_match('/<select[^>]*地市[^>]*>.*?<\/select>/is', $html, $selectBlock)) {
+            $html = $selectBlock[0];
+        }
+
+        preg_match_all('/<option\s+value="([^"]*)"[^>]*>([^<]*)<\/option>/i', $html, $matches);
+
         if (!empty($matches[1])) {
             foreach ($matches[1] as $index => $value) {
-                if (!empty($value)) { // 跳过"请选择"等空值
-                    $options[] = [
-                        'value' => $value,
-                        'text' => $matches[2][$index] ?? $value,
-                    ];
+                $value = trim($value);
+                $text = trim($matches[2][$index] ?? $value);
+
+                if (in_array($value, $placeholderValues, true)) {
+                    continue;
                 }
+                if ($text === '' || in_array($text, $placeholderTexts, true)) {
+                    continue;
+                }
+                // 过滤乱码：含控制字符或非 UTF-8 的视为无效
+                if (preg_match('/[\x00-\x08\x0B\x0C\x0E-\x1F]/', $text) || !mb_check_encoding($text, 'UTF-8')) {
+                    continue;
+                }
+                // 过滤明显非地区名称的短乱码（如单字符乱码）
+                if (mb_strlen($text) < 2 && !preg_match('/^[\x{4e00}-\x{9fa5}A-Za-z0-9]+$/u', $text)) {
+                    continue;
+                }
+
+                $options[] = [
+                    'value' => $value,
+                    'text'  => $text,
+                ];
             }
         }
-        
+
         return $options;
     }
 }