I am using pupeteer to profile the performance of a webpage initialization:
import puppeteer from "puppeteer";
const browser = await puppeteer.launch({ headless: false });
const page = await browser.newPage();
await page.tracing.start({ path: './profile.json', screenshots: true});
await page.goto("https://stackoverflow.com/", {
waitUntil: 'domcontentloaded',
timeout: 120000
});
try {
await page.waitForNetworkIdle({ idleTime: 2000, timeout: 120000 });
} catch(e) {
console.warn("Network didn't idle");
}
await page.tracing.stop();
await browser.close();
Running that code generates a "profile.json" (in the Trace Event Format).
Importing this profile.json into Chrome shows a nice aggregate time calculation for each profiled function execution:
Inside the JSON there is a ProfileChunk containing multiple CPU nodes which include the very same function call (the third entry inside args.data.cpuProfile.nodes):
"args": {
"data": {
"cpuProfile": {
"nodes": [
{
"callFrame": {
"codeType": "JS",
"columnNumber": 29925,
"functionName": "ready",
"lineNumber": 1,
"scriptId": 146,
"url": "https://ajax.googleapis.com/ajax/libs/jquery/1.12.4/jquery.min.js"
},
"id": 638,
"parent": 637
},
{
"callFrame": {
"codeType": "JS",
"columnNumber": 28148,
"functionName": "fireWith",
"lineNumber": 1,
"scriptId": 146,
"url": "https://ajax.googleapis.com/ajax/libs/jquery/1.12.4/jquery.min.js"
},
"id": 639,
"parent": 638
},
{
"callFrame": {
"codeType": "JS",
"columnNumber": 27374,
"functionName": "i",
"lineNumber": 1,
"scriptId": 146,
"url": "https://ajax.googleapis.com/ajax/libs/jquery/1.12.4/jquery.min.js"
},
"id": 640,
"parent": 639
},
{
"callFrame": {
"codeType": "JS",
"columnNumber": 31004,
"functionName": "",
"lineNumber": 1,
"scriptId": 146,
"url": "https://ajax.googleapis.com/ajax/libs/jquery/1.12.4/jquery.min.js"
},
"id": 641,
"parent": 640
}
],
"samples": [
637,
639,
640,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641,
641
]
},
"lines": [
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2
],
"timeDeltas": [
116,
135,
128,
264,
222,
1450,
136,
388,
380,
144,
147,
129,
143,
131,
166,
140,
214,
166,
179,
182,
174,
128,
139,
121,
131,
163,
138,
141,
117,
144,
135,
316,
126,
163,
163,
123,
128,
133,
166,
147,
182,
138,
1084,
202,
147,
126,
1624,
105,
150,
145,
103,
140,
130,
131,
169,
165,
108,
204,
72,
109,
128,
142,
113,
140,
114,
138,
129,
145,
137,
163,
141,
146,
154,
151,
131,
133,
130,
132,
131,
131,
130,
131,
131,
131,
130,
133,
133,
155,
137,
160,
116,
119,
164,
127,
281,
133,
120,
152,
128,
129
]
}
},
"cat": "disabled-by-default-v8.cpu_profiler",
"id": "0x1",
"name": "ProfileChunk",
"ph": "P",
"pid": 90837,
"tid": 71195,
"ts": 795527225633,
"tts": 96468
Is there any way to calculate the duration programmatically from the profile.json?