2022年1月

PHP 分析 Nginx 日志

作者: Chuwen
时间: 2022-01-18
分类: PHP
评论

源地址：https://gist.github.com/zwjzxh520/4444e276db0db5423dfc3dd0e437408d

```php
<?php 
/**
 * nginx日志文件分析。
 * 根据配置的nginx日志格式，分析对应的日志内容。各变量
 */

class NginxLog {
    protected static $br = "\n";
    /**
     * 根据日志格式分析nginx访问日志。
     * 格式示例："$remote_addr" $request_time - $remote_user d [$time_local] "$request" $status $body_bytes_sent "$http_referer" "$http_user_agent" "$http_x_forwarded_for"
     * @param  array $log    访问日志数组
     * @param  string $format 访问日志格式
     * @return array         分析后的访问日志数组。key是访问日志变量名， value是该变量对应的日志信息
     */
    public function analysisAccess($log, $format)
    {
        $logVars = $this->parseFormat($format);

$log = explode(self::$br, $log);

$formatLen = strlen($format);
        $logArr = [];
        foreach ($log as $lineNum=>$line) {
            $line = trim($line);
            if (empty($line)) {
                continue;
            }
            //日志内容偏移量
            $linePos = 0;
            //日志格式偏移量
            $formatPos = 0;

foreach ($logVars as $var => $info) {
                //本变量内容的起始位置。
                $linePos += $info['prevLen'];

//本变量日志内容的结束位置
                if ($info['endChar']) {
                    $varLogEnd = strpos($line, $info['endChar'], $linePos);
                } else {
                    $varLogEnd = 0;
                }

$logArr[$lineNum][ $var ] = substr($line, $linePos, $varLogEnd ? $varLogEnd - $linePos : null);
                //下一个变量的起始位置
                if ($varLogEnd) {
                    $linePos += $varLogEnd - $linePos;
                }
            }
        }
        return $logArr;
    }
    /**
     * 分析nginx错误日志。
     * 由于有些错误信息，被分割成多行（例如php fatal error的堆栈错误），因此默认情况下，
     * 利用$mergeLog变量保留错误日志，将这些错误日志合并成一行进行分析。
     * 如果是不完整的错误日志，将不会进行分析。
     * 如果得到完整的错误日志， $mergeLog 会被重置。
     * 提供 $reset 参数，用于清空 $mergeLog 变量。
     * @param  array  $logLines 错误日志数组
     * @param  boolean $reset   是否重置上次分析保留不完整错误日志
     * @return array            分析后的错误日志数组
     */
    public function analysisError($logLines, $reset = false)
    {
        $result = [];
        $yinhaoCount = 2;
        static $mergeLog = '';
        if ($reset) {
            $mergeLog = '';
        }
        foreach ($logLines as $lineNum => $log) {
            if (substr_count($log, '"') >= $yinhaoCount) {
                if ($mergeLog) {
                    $log = $mergeLog.$log;
                    $mergeLog = '';
                }
                $logResult = $this->parseLineError($log);
                if (!empty($logResult)) {
                    $result[] = $logResult;
                }
            } else {
                $mergeLog .= $log."\n";
            }
        }
        return $result;
    }
    /**
     * 对一行错误日志进行分析。
     * 完整的日志，将会以时间开头，如果不是，将不会分析。
     * 返回结果：
     * [
     *     'datetime'   =>  '错误发生的时间戳'
     *     'type'   =>  'nginx错误类型'
     *     'fd'   =>  'nginx线程id'
     *     'msg'   =>  'nginx错误类型说明'
     *     'desc'   =>  'nginx错误详细提示'
     *     'http'   =>  ['http 请求相关信息']
     *     'phperr'   =>  'php错误类型，如果不是php错误，则为空'
     * ]
     * @param  string $log 一行完整的错误日志
     * @return array      [description]
     */
    protected function parseLineError($log)
    {
        $datetimeLen = 19;  //日期字符串长度
        $metaPos = $datetimeLen + 1; //meta字符截取起始位置
        $datetime = substr($log, 0, 19);
        $result = [];
        if ($this->isDateTimeFormat($datetime)) {
            //日期时间
            $result['datetime'] = strtotime($datetime);
            $metaEndPos = strpos($log, '"', $metaPos);
            $errorDescEndPos = strpos($log, '"', $metaEndPos + 1);
            $metaStr = trim(substr($log, $metaPos, $metaEndPos - $metaPos));

$metaArr = explode(' ', $metaStr);

//错误类型
            $result['type'] = substr($metaArr[0], 1, -1);
            //文件描述符。nginx进程相关
            $result['fd']   = $metaArr[1].' '.$metaArr[2];
            //错误类型说明
            $result['msg'] = rtrim(implode(' ', array_slice($metaArr, 3)), ':');
            //错误提示说明
            $result['desc'] = substr($log, $metaEndPos +1, $errorDescEndPos - $metaEndPos - 1);
            //http相关信息
            $result['http'] = $this->parseErrorHTTP(trim(substr($log, $errorDescEndPos + 1)));
            //php错误，如果是php错误的话
            $result['phperr'] = $this->getPHPErrorType($result['desc']);
            
        }
        return $result;
    }
    /**
     * 分析php错误类型，如果不是php错误，则返回空
     * @param  string $str 
     * @return string      
     */
    protected function getPHPErrorType($str) 
    {
        //PHP message:
        $type = '';
        if ('PHP message:' === substr($str, 0, 12)) {
            //为什么是17，因为PHP message: PHP Waring， 要加' PHP '的长度
            $type = substr($str, 17, strpos($str, ':', 12) - 17);
        }
        return $type;
    }
    /**
     * 解析http请求的相关信息。一般是从client: 字符串一直到日志结束
     * 返回结果(以下字段不一定都有，根据日志而定)：
     * [
     *     '0'   =>  'nginx错误详细提示（desc）后，client关键词前的一段提示信息'
     *     'client'   =>  '客户端ip'
     *     'server'   =>  '域名'
     *     'method'   =>  '请求方法'
     *     'uri'   =>  '请求的uri'
     *     'httpver'   =>  'http版本'
     *     'upstream'   =>  'php-fpm地址'
     *     'host'   =>  'host头'
     *     'referer'   =>  '引用url地址'
     * ]
     * @param  string $http 
     * @return array       
     */
    protected function parseErrorHTTP($http)
    {
        //client:
        $split = strpos($http, 'client:');
        $result[0] = substr($http, 0, $split - 2);

$other = explode(',', substr($http, $split));

foreach ($other as $value) {
            $pos = strpos($value, ': ');
            $key = trim(substr($value, 0, $pos));
            $val = trim(substr($value, $pos + 2), '"');
            if ('request' === $key) {
                $result = array_merge($result, $this->parseRequest($val));
            } else {
                $result[$key] = $val;
            }
        }
        return $result;
    }
    /**
     * 解析http请求信息。是对"GET http:/www.baidu.com HTTP/1.1"的解析。
     * 返回结果：
     * [
     *     'method'   =>  '请求方法'
     *     'uri'   =>  '请求的uri'
     *     'httpver'   =>  'http版本'
     * ]
     * @param  string $request 示例："GET http:/www.baidu.com HTTP/1.1"
     * @return array          
     */
    protected function parseRequest($request)
    {
        $result = [];
        $split = ' ';
        $firstPos = strpos($request, $split);
        $lastPos = strrpos($request, $split);

return [
            'method'    =>  substr($request, 0, $firstPos),
            'uri'       => trim(substr($request, $firstPos, $lastPos - $firstPos)),
            'httpver'   =>  trim(substr($request, $lastPos)),
        ];
    }
    /**
     * 检查字符串是否是日期格式。固定格式：yyyy/mm/dd hh:mm:ss
     * @param  string  $str 待检查的字符串
     * @return boolean      
     */
    protected function isDateTimeFormat($str)
    {
        //2016/09/06 00:32:19
        $number = '0123456789';
        $format = '0000/00/00 00:00:00';

$strlen = strlen($str);
        $result = true;
        for($i=0; $i<$strlen; $i++) {
            if ($format{$i} === '0') {
                if (false === strpos($number, $str{$i})) {
                    $result = false;
                    break;
                }
            } elseif ($format{$i} !== $str{$i}) {
                $result = false;
                break;
            }
        }
        return $result;
    }

/**
     * 分析访问日志格式，得出变量名称，并且保存变量在日志内容中的起始位置。
     * @param  string $format nginx配置文件中的日志格式
     * @return array         
     */
    public function parseFormat($format) 
    {
        $varCharList = 'abcdefghijklmnopqrstuvwxyz_$ABCDEFGHIJKLMNOPQRSTUVWXYZ';
        
        $return = [];

$formatLen = strlen($format);

$logVar = '';
        $prevStrLen = 0;
        for ($i=0; $i < $formatLen; $i++) {
            $char = $format{$i};
            if (strpos($varCharList, $char) === false || (empty($logVar) && $char != '$')) {
                if ($logVar) {
                    $return[$logVar]['endChar'] = $char; //变量后面紧挨着的字符，日志内容中，该字符的位置，即表示本变量的内容结束位置
                    $return[$logVar]['prevLen'] = $prevStrLen; //变量前面忽略的字符长度
                    $prevStrLen = 1;
                    $logVar = '';
                } else {
                    $prevStrLen++;
                }
            } else {
                $logVar .= $char;
            }
        }
        if (!empty($logVar)) {
            $return[$logVar]['endChar'] = '';
            $return[$logVar]['prevLen'] = $prevStrLen;
            $logVar = '';
            $prevStrLen = 0;
        }
        return $return;
    }

/**
     * 读取大文件。
     * $pos参数和$startLine参数，只设置其中一个，否则结果可能与预期不符。
     * 指定$pos时，为保证读取到的是完整的一行，会将该位置所在行的数据全部读出。
     * @param  string  $filepath  文件绝对路径
     * @param  integer $pos        读取的起始位置
     * @param  integer $startLine 读取的起始行
     * @return [type]             [description]
     */
    public function readBigFile($filepath, $pos = 0, $startLine = 1)
    {
        $handle = fopen($filepath, 'rb');
        $br = self::$br;
        $brLen = strlen($br);
        $lastShortLine = '';
        $blockLen = 20480;  //每次读取的数据块大小
        if ($handle) {
            if ($pos > 0) {
                //移动到合适的位置，保证读取的是完整的一行数据
                $movePos = 0;
                while($pos - $movePos >= 0) {
                    fseek($handle, $pos - $movePos);
                    $c = fread($handle, 1);
                    if ($c === "\n" || $c === "\r") {
                        break;
                    }
                    $movePos ++;
                }
                if ($c !== "\n" && $c !== "\r") {
                    fseek($handle, -1, SEEK_CUR);
                }
                unset($movePos);
            }
            while ( ($data = fread($handle, $blockLen)) ) {
                if (!feof($handle)) {
                    $dataSize = strlen($data);
                    //补上上次读取的，不完整的行数据
                    if ($lastShortLine){
                        $data = $lastShortLine.$data;
                        $dataSize = strlen($data);
                    }

$lastBrPos = strrpos($data, $br);
                    if ($lastBrPos !== false && $lastBrPos != $dataSize+$brLen) {
                        $lastShortLine = substr($data, $lastBrPos+$brLen);
                    } else  {
                        $lastShortLine = '';
                    }
                    $result = $lastBrPos !== false ? substr($data, 0, $lastBrPos) : $data;
                    //定位起始行
                    if ($startLine > 1) {
                        $rows = substr_count($result, $br);
                        $startLine = $startLine - $rows;
                        if ($startLine > 1) {
                            continue;
                        } else {
                            $startpos = 0;
                            $startLine = abs($startLine);
                            while ($rows - $startLine > 1) {
                                $startpos += strpos($result, $br, $startpos) + $brLen;
                                $startLine++;
                            }
                            $result = substr($result, $startpos);
                            unset($rows, $startLine, $startpos);
                        }
                    }
                } else {
                    $result = $lastShortLine.$data;
                }
                yield $result;
            }
            fclose($handle);
        }
    }

}

$format = '"$remote_addr" $request_time - $remote_user d [$time_local] "$request" $status $body_bytes_sent "$http_referer" "$http_user_agent" "$http_x_forwarded_for"';

$log = <<<LOG
"192.168.110.1" 0.040 - - d [05/Aug/2016:08:22:54 +0800] "GET /uc_server/data/avatar/006/40/73/36_real_avatar_small.jpg HTTP/1.0" 404 6664 "-" "-" "-"
"36.149.107.72" 3.016 - - d [05/Aug/2016:08:22:54 +0800] "GET /cardniu/api/api_splashinterface.php?udid=deviceId-866968026335845-generate-cardniu&systemName=android+OS&systemVersion=4.4.4&productName=Cardniu&productVersion=4&position=KNSQTZXQB&positionList=1&chanelSys=shequ HTTP/1.1" 200 5 "http://www.baidu.com/cardniu/detail.php?tid=552221&utm_source=552221&utm_medium=ribao&utm_campaign=xiaoxi" "Mozilla/5.0 (Linux; Android 4.4.4; A31 Build/KTU84P) AppleWebKit/537.36 (KHTML, like Gecko) Version/4.0 Chrome/33.0.0.0 Mobile Safari/537.36 feideeAndroid-V4 MymoneySms7.3.2-oppo" "-"
"192.168.110.1" 0.040 - - d [05/Aug/2016:08:22:54 +0800] "GET //uc_server/data/avatar/005/86/04/56_real_avatar_small.jpg HTTP/1.0" 404 6664 "-" "-" "-"
LOG;

// $log = file_get_contents('test.txt');
//测试访问日志
$startTime = microtime(true);
// var_export( (new NginxLog) -> parseFormat($format) );
var_export( (new NginxLog) -> analysisAccess($log, $format) );
echo microtime(true) - $startTime;

//测试错误日志
// $logFile = __DIR__.'/../testdata/nginx_access.log';

$errorLogFile = __DIR__.'/../testdata/www.baidu.com_error_2016-09-05.log';

$value = <<<LOG
2016/09/05 14:52:52 [error] 17890#0: *1424957037 FastCGI sent in stderr: "PHP message: PHP Fatal error:  Uncaught exception 'DbException' with message 'Duplicate entry '240319164' for key 'sid'' in /var/www/html/bbs/source/class/db/db_driver_mysql.php:218
Stack trace:
#0 /var/www/html/bbs/source/class/db/db_driver_mysql.php(151): db_driver_mysql->halt('Duplicate entry...', 1062, 'UPDATE  pre_com...')
#1 /var/www/html/bbs/source/class/db/db_driver_mysql_slave.php(62): db_driver_mysql->query('UPDATE  pre_com...', false, false)
#2 /var/www/html/bbs/source/class/discuz/discuz_database.php(179): db_driver_mysql_slave->query('UPDATE  pre_com...', false, false)
#3 /var/www/html/bbs/source/class/discuz/discuz_database.php(102): discuz_database::query('UPDATE  pre_com...', '')
#4 /var/www/html/bbs/source/class/discuz/discuz_table.php(52): discuz_database::update('common_member', Array, '`uid`='10929277...', false, false)
#5 /var/www/html/bbs/m/register_by_email.php(72): discuz_table->update(10929277, Array)
#6 /var/www/html/bbs/m/function/ssj_function.php(788): register_by_ssjuser('', '" while reading response header from upstream, client: 192.168.241.97, server: www.baidu.com, request: "POST /m/api/credit.php HTTP/1.1", upstream: "fastcgi://127.0.0.1:9000", host: "www.baidu.com"
2016/09/06 05:11:54 [error] 10527#0: *915961403 open() "/var/www/html/cardniu/thread-544114" failed (2: No such file or directory), client: 180.153.205.253, server: www.baidu.com, request: "GET /thread-544114?10000skip=true-1-1.html HTTP/1.1", host: "www.baidu.com", referrer: "http://www.baidu.com/cardniu/detail.php?tid=544114?10000skip=true"
2016/09/06 06:22:15 [error] 10535#0: *916177977 FastCGI sent in stderr: "Primary script unknown" while reading response header from upstream, client: 111.127.121.50, server: www.baidu.com, request: "GET /cardniu/api/home.php?mod=spacecp&ac=usergroup&do=expiry HTTP/1.1", upstream: "fastcgi://127.0.0.1:9000", host: "www.baidu.com", referrer: "http://www.baidu.com/cardniu/kn_daily.php?stamptime=2016-09-05&udid=deviceId-868979027115530-generate-cardniu&bankcode=SPD,CMB,CEB"
LOG;

$startTime = microtime(true);
$nginxLog = new NginxLog;
$logNum = 0;

$phperrs = [];
foreach ($nginxLog -> readBigFile($errorLogFile) as $value) {
    $logString = trim($value);
    $logArr = explode("\n", $logString);

$logNum += count($logArr);

foreach ($nginxLog->analysisError($logArr) as $log) {
        if (!empty($log['phperr'])) {
            $phperrs[] = $log;
        }
    }
}
echo 'finish. cost time:'.(microtime(true) - $startTime)."\n";
echo '日志数量：'.$logNum."\n";
echo 'php错误日志数量：'.count($phperrs)."\n";
var_export($phperrs);
```

MboX + ReactECharts 不更新数据

作者: Chuwen
时间: 2022-01-12
分类: React
评论

React + Vite + TypeScript 读取环境变量报错: Uncaught ReferenceError: process is not defined

作者: Chuwen
时间: 2022-01-05
分类: React
评论

PHP 分析 Nginx 日志

MboX + ReactECharts 不更新数据

React + Vite + TypeScript 读取环境变量报错: Uncaught ReferenceError: process is not defined

最新文章

最近回复

分类

归档

其它