diff --git a/README.md b/README.md index 599aea6..f62827d 100644 --- a/README.md +++ b/README.md @@ -1,4 +1,4 @@ -# 検索オンライン 1.0 +# 検索オンライン 1.0.1 日本産検索エンジン。 diff --git a/srv/crontab b/srv/crontab index 267c851..c49068d 100644 --- a/srv/crontab +++ b/srv/crontab @@ -1 +1 @@ - 0 2 * * 0 root php /www/active/kensaku.online/tool/crawler.php +0 2 * * * root cd /www/active/kensaku.online/tool && php crawler.php diff --git a/tool/crawler.php b/tool/crawler.php index 0cead6e..689492d 100644 --- a/tool/crawler.php +++ b/tool/crawler.php @@ -62,6 +62,7 @@ // 内容 $pbody = shell_exec('curl -s -L '.$o.' | pandoc -f html -t plain 2>&1'); + if (strlen($o) > 100 || mb_strlen(htmlentities($ptitle)) > 100) continue; $res[] = [ 'website_id' => $k,