admin/SEONexus
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

debug-novel-task

Browse Source
This commit is contained in:
make
2025-06-19 18:45:46 +08:00
parent 321f5eda01
commit c24882beab
1 changed files with 23 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
code/app/task/crawler/zw/page/Site.php
View File

@@ -200,6 +200,7 @@ class Site
*/ */
public function getNovelPageList(array $arrFilter): array public function getNovelPageList(array $arrFilter): array
{ {
$QueueManage = QueueManage::getInstance();
/** @var NovelPage[] */ /** @var NovelPage[] */
$arrNovelPage = []; $arrNovelPage = [];
@@ -229,10 +230,19 @@ class Site
continue; continue;
} }
// 判断是否包含 $strContent <body class="novel_info">，不包含跳过 // 判断是否包含 $strContent <body class="novel_info">，不包含跳过 Gateway Timeout
if (strpos($strContent, '<body class="novel_info">') === false) { if (strpos($strContent, '<body class="novel_info">') === false) {
// 不包含该字符串，跳过 // 不包含该字符串，跳过
var_dump(mb_substr($strContent, 0, 300)); var_dump(mb_substr($strContent, 0, 300));
if (strpos($strContent, '超过并发限制') === true || strpos($strContent, 'Timeout') === true) {
// 重新推送任务
$arrTask = [
'callback' => [Zw74Scheduler::class, 'getNovelInfo'],
'data' => $arrFilter,
];
$QueueManage->set($arrTask);
}
continue; continue;
} }
@@ -251,7 +261,6 @@ class Site
'callback' => [Zw74Scheduler::class, 'getNovelInfo'], 'callback' => [Zw74Scheduler::class, 'getNovelInfo'],
'data' => $arrFilter, 'data' => $arrFilter,
]; ];
$QueueManage = QueueManage::getInstance();
$QueueManage->set($arrTask); $QueueManage->set($arrTask);
$incKey = config('task.app_name').'_errer_getNovelPageInfo:' . date('Y-m-d'); $incKey = config('task.app_name').'_errer_getNovelPageInfo:' . date('Y-m-d');
@@ -278,6 +287,8 @@ class Site
*/ */
public function getChapterPageList(array $arrData): array public function getChapterPageList(array $arrData): array
{ {
$QueueManage = QueueManage::getInstance();
$arrFilter = $arrData['filter']; $arrFilter = $arrData['filter'];
/** @var ChapterPage[] */ /** @var ChapterPage[] */
@@ -304,6 +315,16 @@ class Site
$strPattern = "/lastread\.set\((.*?)\);/"; $strPattern = "/lastread\.set\((.*?)\);/";
if (!preg_match($strPattern, $strContent, $arrMatches)) { if (!preg_match($strPattern, $strContent, $arrMatches)) {
var_dump(mb_substr($strContent, 0, 300)); // 防止输出太长 var_dump(mb_substr($strContent, 0, 300)); // 防止输出太长
if (strpos($strContent, '超过并发限制') === true || strpos($strContent, 'Timeout') === true) {
// 重新推送任务
$arrTask = [
'callback' => [Zw74Scheduler::class, 'getChapterInfo'],
'data' => $arrData,
];
$QueueManage->set($arrTask);
}
continue; continue;
} }
@@ -323,7 +344,6 @@ class Site
'callback' => [Zw74Scheduler::class, 'getChapterInfo'], 'callback' => [Zw74Scheduler::class, 'getChapterInfo'],
'data' => $arrData, 'data' => $arrData,
]; ];
$QueueManage = QueueManage::getInstance();
$QueueManage->set($arrTask); $QueueManage->set($arrTask);
$this->redisDailyInc(config('task.app_name') . '_errer_getChapterPageList:' . date('Y-m-d')); $this->redisDailyInc(config('task.app_name') . '_errer_getChapterPageList:' . date('Y-m-d'));