#!/bin/bash
web_site=
"http://网站域名/管理目录/admin_reslib2.php"
web_pwd=
"8888e82e85bd4540f0defa3fb7a8e888"
web_api=(
'?ac=day&rid=1&url=https://api.iokzy.com/inc/ldg_seackm3u8s.php'
'?ac=day&rid=2&url=http://www.zdziyuan.com/inc/s_ldgm3u8_sea.php'
)
web_ua="Mozilla
/5
.0 (Windows NT 10.0; Win64; x64) AppleWebKit
/537
.36 (KHTML, like Gecko)
Chrome
/76
.0.3809.100 Safari
/537
.36 seacmsbot
/1
.2;"
function
get_content() {
echo
"正在采集第$page页..."
cResult=$(curl --connect-timeout 10 -m 20 -k -s -L -A
"$web_ua"
"$1"
)
echo
$cResult |
grep
-q
"采集"
if
[
"$?"
=
"0"
];
then
next_content
"$cResult"
else
echo
-e
"采集失败,请检查设置!\n失败链接-->$1\n返回信息-->$cResult\n采集结束,共0页"
fi
}
function
next_content() {
Result=$(
echo
"$1"
|
tr
"
" "
\n")
a=$(
echo
"$Result"
|
grep
-c
"采集成功"
)
b=$(
echo
"$Result"
|
grep
-c
"更新数据"
)
c=$(
echo
"$Result"
|
grep
-c
"无需更新"
)
d=$(
echo
"$Result"
|
grep
-c
"跳过"
)
echo
"采集成功-->已更$c部,新增$a部,更新$b部,跳过$d部"
let
add+=$a
let
update+=$b
let
none+=$c
let
jmp+=$d
next_url=${1
next_url=${next_url%%\'*}
if
[
"${next_url:0:1}"
=
"?"
]
then
let
page++
get_content
"$web_site$next_url"
else
echo
"采集结束,共$page页"
fi
}
echo
"海洋CMS自动采集脚本开始执行 版本:v1.2"
starttime=$(
date
+%s)
update=0
add=0
none=0
jmp=0
for
url
in
${web_api[@]};
do
if
[[ ! -z $url ]]
then
web_param=
"$web_site$url&password=$web_pwd"
page=1
echo
"开始采集:$url"
get_content $web_param
fi
done
endtime=$(
date
+%s)
echo
"============================"
echo
"入库-->$add部"
echo
"更新-->$update部"
echo
"跳过-->$jmp部(未绑定分类或链接错误)"
echo
"今日-->$[none+add+update]部"
echo
"============================"
echo
"全部采集结束,耗时$[endtime - starttime]秒"