mirror of
https://gitee.com/ulthon/ulthon_admin.git
synced 2026-09-04 15:11:37 +08:00
feat(tools): XHProf 日志导入三层服务与函数监控物化
- XhprofProfileParserService（Base/App）：纯数组解析 jsonl 行， meta 字段以真实 fixture 为准；simple_url 缺失兜底与采集端回调同算法； watch 正则按 callee 物化（ct/wt_sum/wt_max），非法正则编译计 fail 跳过 - XhprofLogReaderService（Base/App）：增量读取，半行缓冲超限整行丢弃 （不照抄 nginx 8MB 强制切行）；EOF 无换行半行不 yield 不推进 offset； 进度不自动落盘，由导入侧在同一事务内显式 savePosition - XhprofLogImport 定时任务（Base/App）：不受 XHPROF_ENABLE 门控， runs-*.jsonl 按文件名序增量导入；run 逐条 insert + detail chunk 20 insertAll + run_watch insertAll 与 position 同事务落盘；导至 EOF 后 只删已导完文件 - 修复 T1 缺陷：request_time decimal(12,3) 装不下 10 位时间戳， Scheme 与 migration 同步改 decimal(13,3) - 新增 parser 单元测试（真实 fixture 断言，9 用例 42 断言） 与 jsonl fixture；phpunit 注册 unit 测试套件
This commit is contained in:
augushong
204
extend/base/common/service/XhprofProfileParserServiceBase.php
Normal file
204
extend/base/common/service/XhprofProfileParserServiceBase.php
Normal file
@@ -0,0 +1,204 @@
|
||||
<?php
|
||||
|
||||
declare(strict_types=1);
|
||||
|
||||
namespace base\common\service;
|
||||
|
||||
/**
|
||||
* XHProf jsonl 行解析器（Base 层）.
|
||||
*
|
||||
* 输入：php-profiler FileSaver 落盘的 jsonl 单行 {"profile":{...},"meta":{...}}；
|
||||
* 输出：ul_xhprof_run 入库行字段 + 解码后的 profile（供 watch 物化）。
|
||||
*
|
||||
* 纯数组处理，不依赖 Xhgui\Profiler 等镜像内类库（app 侧禁止 use，见计划 guardrails）。
|
||||
*
|
||||
* meta 真实结构（以容器内实采 fixture 为准，tests/fixtures/xhprof-sample.jsonl）：
|
||||
* url = meta.url（原生存在，通常为 path 形如 '/'，无需从 REQUEST_URI 拼）
|
||||
* simple_url = meta.simple_url（bootstrap 的 profiler.simple_url 回调原生产出；
|
||||
* 仅缺失/为空时兜底自算，算法与回调保持一致，勿漂移）
|
||||
* method = meta.SERVER.REQUEST_METHOD
|
||||
* request_time = meta.request_ts_micro.sec + usec/1e6（缺失退化 meta.SERVER.REQUEST_TIME_FLOAT）
|
||||
* wall/cpu/pmu = profile['main()'] 的 wt/cpu/pmu（main() 缺失降级 0 并置 degraded，由上层计 fail）
|
||||
*
|
||||
* 业务侧通过 app/common/service/XhprofProfileParserService 覆盖本类方法.
|
||||
*/
|
||||
class XhprofProfileParserServiceBase
|
||||
{
|
||||
/** run.url 列长（ul_xhprof_run.url varchar(500)，超长截断） */
|
||||
protected const URL_MAX_LENGTH = 500;
|
||||
|
||||
/** run.simple_url 列长 */
|
||||
protected const SIMPLE_URL_MAX_LENGTH = 255;
|
||||
|
||||
/** run.method 列长 */
|
||||
protected const METHOD_MAX_LENGTH = 10;
|
||||
|
||||
/**
|
||||
* 解析单行 jsonl.
|
||||
*
|
||||
* @param string $line 原始行（不含换行符）
|
||||
*
|
||||
* @return array|null 成功返回 run 行字段 + profile；空行/非法 JSON/缺 profile 键返回 null。
|
||||
* main() 缺失时 wall/cpu/memory_peak 降级 0 且 degraded=true（行仍可入库）。
|
||||
*/
|
||||
public function parse(string $line): ?array
|
||||
{
|
||||
$line = trim($line);
|
||||
if ($line === '' || $line[0] !== '{') {
|
||||
return null;
|
||||
}
|
||||
|
||||
$data = json_decode($line, true);
|
||||
if (!is_array($data) || !isset($data['profile']) || !is_array($data['profile'])) {
|
||||
return null;
|
||||
}
|
||||
|
||||
$meta = isset($data['meta']) && is_array($data['meta']) ? $data['meta'] : [];
|
||||
|
||||
// main() 缺失：降级 0 并标记 degraded（导入侧对 degraded 行计 fail 但不丢数据）
|
||||
$main = $data['profile']['main()'] ?? null;
|
||||
$degraded = !is_array($main);
|
||||
$wallTime = $degraded ? 0 : (int) ($main['wt'] ?? 0);
|
||||
$cpuTime = $degraded ? 0 : (int) ($main['cpu'] ?? 0);
|
||||
$memoryPeak = $degraded ? 0 : (int) ($main['pmu'] ?? 0);
|
||||
|
||||
$url = (string) ($meta['url'] ?? '');
|
||||
$simpleUrl = (string) ($meta['simple_url'] ?? '');
|
||||
if ($simpleUrl === '') {
|
||||
$simpleUrl = $this->buildSimpleUrlFallback($url);
|
||||
}
|
||||
|
||||
return [
|
||||
'url' => mb_substr($url, 0, self::URL_MAX_LENGTH),
|
||||
'simple_url' => mb_substr($simpleUrl, 0, self::SIMPLE_URL_MAX_LENGTH),
|
||||
'method' => mb_substr((string) ($meta['SERVER']['REQUEST_METHOD'] ?? ''), 0, self::METHOD_MAX_LENGTH),
|
||||
'wall_time' => $wallTime,
|
||||
'cpu_time' => $cpuTime,
|
||||
'memory_peak' => $memoryPeak,
|
||||
'request_time' => $this->parseRequestTime($meta),
|
||||
'profile' => $data['profile'],
|
||||
'degraded' => $degraded,
|
||||
];
|
||||
}
|
||||
|
||||
/**
|
||||
* simple_url 兜底计算.
|
||||
*
|
||||
* 与采集端 bootstrap 的 profiler.simple_url 回调算法完全一致（唯一实现点在采集端，
|
||||
* 这里仅做缺失兜底）：path 去 query、纯数字段替换 ':id'、'/' 连接（无前导斜杠）。
|
||||
*/
|
||||
public function buildSimpleUrlFallback(string $url): string
|
||||
{
|
||||
$path = parse_url($url, PHP_URL_PATH);
|
||||
if (!is_string($path) || $path === '') {
|
||||
return '';
|
||||
}
|
||||
$parts = explode('/', trim($path, '/'));
|
||||
foreach ($parts as &$part) {
|
||||
if ($part !== '' && ctype_digit($part)) {
|
||||
$part = ':id';
|
||||
}
|
||||
}
|
||||
unset($part);
|
||||
|
||||
return implode('/', $parts);
|
||||
}
|
||||
|
||||
/**
|
||||
* 编译 watch 正则（导入批次调用一次，结果供 materializeWatches 复用，避免逐行重复编译）.
|
||||
*
|
||||
* @param array $watches [['id' => int, 'regex' => string], ...]
|
||||
*
|
||||
* @return array{patterns: array<int, string>, fails: int}
|
||||
* patterns 以 watch_id 为键；id 非法/正则为空/@preg_match 编译失败
|
||||
* 的条目跳过并计入 fails（防非法正则打爆导入任务）
|
||||
*/
|
||||
public function compileWatchRegexes(array $watches): array
|
||||
{
|
||||
$patterns = [];
|
||||
$fails = 0;
|
||||
foreach ($watches as $watch) {
|
||||
$id = (int) ($watch['id'] ?? 0);
|
||||
$regex = (string) ($watch['regex'] ?? '');
|
||||
if ($id <= 0 || $regex === '') {
|
||||
$fails++;
|
||||
continue;
|
||||
}
|
||||
if (@preg_match($regex, '') === false) {
|
||||
$fails++;
|
||||
continue;
|
||||
}
|
||||
$patterns[$id] = $regex;
|
||||
}
|
||||
|
||||
return ['patterns' => $patterns, 'fails' => $fails];
|
||||
}
|
||||
|
||||
/**
|
||||
* watch 物化：拆 "caller==>callee" 边，按 callee（被调用函数名）匹配正则.
|
||||
*
|
||||
* 命中边 ct 累加、wt 累加 wt_sum、wt_max 取最大。
|
||||
* 裸键（如 main()）无 ==> 分隔符，不参与物化。
|
||||
*
|
||||
* @param array $profile 解码后的 profile 边表
|
||||
* @param array $patterns compileWatchRegexes 产出的 [watch_id => regex]
|
||||
*
|
||||
* @return array<int, array{ct: int, wt_sum: int, wt_max: int}> 以 watch_id 为键的聚合行
|
||||
*/
|
||||
public function materializeWatches(array $profile, array $patterns): array
|
||||
{
|
||||
if (empty($patterns)) {
|
||||
return [];
|
||||
}
|
||||
|
||||
$result = [];
|
||||
foreach ($profile as $key => $metrics) {
|
||||
$sep = strpos($key, '==>');
|
||||
if ($sep === false) {
|
||||
continue;
|
||||
}
|
||||
$callee = substr($key, $sep + 3);
|
||||
$ct = (int) ($metrics['ct'] ?? 0);
|
||||
$wt = (int) ($metrics['wt'] ?? 0);
|
||||
|
||||
foreach ($patterns as $watchId => $regex) {
|
||||
if (@preg_match($regex, $callee) !== 1) {
|
||||
continue;
|
||||
}
|
||||
if (!isset($result[$watchId])) {
|
||||
$result[$watchId] = ['ct' => 0, 'wt_sum' => 0, 'wt_max' => 0];
|
||||
}
|
||||
$result[$watchId]['ct'] += $ct;
|
||||
$result[$watchId]['wt_sum'] += $wt;
|
||||
if ($wt > $result[$watchId]['wt_max']) {
|
||||
$result[$watchId]['wt_max'] = $wt;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return $result;
|
||||
}
|
||||
|
||||
/**
|
||||
* request_time：优先 meta.request_ts_micro（sec+usec），缺失退化 meta.SERVER.REQUEST_TIME_FLOAT.
|
||||
*
|
||||
* @return float 浮点秒（round 3 位，与 decimal(12,3) 列精度对齐）
|
||||
*/
|
||||
protected function parseRequestTime(array $meta): float
|
||||
{
|
||||
$ts = $meta['request_ts_micro'] ?? null;
|
||||
if (is_array($ts) && isset($ts['sec'])) {
|
||||
$sec = (int) $ts['sec'];
|
||||
$usec = (int) ($ts['usec'] ?? 0);
|
||||
|
||||
return round($sec + $usec / 1000000, 3);
|
||||
}
|
||||
|
||||
$float = $meta['SERVER']['REQUEST_TIME_FLOAT'] ?? null;
|
||||
if (is_numeric($float)) {
|
||||
return round((float) $float, 3);
|
||||
}
|
||||
|
||||
return 0.0;
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user