unpkg/modules/utils/ingestLogs.js

193 lines
5.0 KiB
JavaScript
Raw Normal View History

2019-01-06 00:50:05 +00:00
import url from 'url';
import { startOfDay, addDays } from 'date-fns';
2018-04-04 05:32:32 +00:00
import data from './data';
import isValidPackageName from './isValidPackageName';
import parsePackageURL from './parsePackageURL';
import * as cloudflare from './cloudflare';
import * as stats from './stats';
2017-11-08 19:07:48 +00:00
2017-05-23 22:00:09 +00:00
/**
* Domains we want to analyze.
*/
2018-04-04 05:32:32 +00:00
const domainNames = [
2018-12-17 17:38:05 +00:00
'unpkg.com'
2018-02-18 04:21:19 +00:00
//"npmcdn.com" // We don't have log data on npmcdn.com yet :/
2018-02-18 02:00:56 +00:00
];
2017-05-23 22:00:09 +00:00
2018-08-26 03:14:49 +00:00
let cachedZones;
function getSeconds(date) {
2018-02-18 02:00:56 +00:00
return Math.floor(date.getTime() / 1000);
}
2017-05-20 05:41:24 +00:00
function stringifySeconds(seconds) {
2018-12-17 17:38:05 +00:00
return new Date(seconds * 1000).toISOString().replace(/\.0+Z$/, 'Z');
}
2017-05-20 05:41:24 +00:00
function toSeconds(ms) {
return Math.floor(ms / 1000);
2017-05-20 05:41:24 +00:00
}
function computeCounters(stream) {
2017-11-08 18:14:46 +00:00
return new Promise((resolve, reject) => {
2018-02-18 02:00:56 +00:00
const counters = {};
const expireat = {};
let totalEntries = 0;
function incr(key, member, by, expiry) {
2018-02-18 02:00:56 +00:00
counters[key] = counters[key] || {};
counters[key][member] = (counters[key][member] || 0) + by;
expireat[key] = expiry;
}
2017-05-23 22:00:09 +00:00
stream
2018-12-17 17:38:05 +00:00
.on('error', reject)
.on('data', entry => {
totalEntries += 1;
const date = new Date(Math.round(entry.EdgeStartTimestamp / 1000000));
2018-02-18 02:00:56 +00:00
const nextDay = startOfDay(addDays(date, 1));
const sevenDaysLater = getSeconds(addDays(nextDay, 7));
const thirtyDaysLater = getSeconds(addDays(nextDay, 30));
2019-01-06 00:50:05 +00:00
const dayKey = stats.createDayKey(date);
2017-05-23 22:00:09 +00:00
if (entry.EdgeResponseStatus === 200) {
// Q: How many requests do we serve for a package per day?
// Q: How many bytes do we serve for a package per day?
2019-01-06 00:50:05 +00:00
const parsed = parsePackageURL(entry.ClientRequestURI);
const packageName = parsed && parsed.packageName;
2017-05-29 05:41:01 +00:00
2018-05-26 00:30:53 +00:00
if (packageName && isValidPackageName(packageName)) {
2018-02-18 02:00:56 +00:00
incr(
`stats-packageRequests-${dayKey}`,
packageName,
1,
thirtyDaysLater
);
incr(
`stats-packageBytes-${dayKey}`,
packageName,
entry.EdgeResponseBytes,
2018-02-18 02:00:56 +00:00
thirtyDaysLater
);
}
}
2017-05-23 22:00:09 +00:00
// Q: How many requests per day do we receive via a protocol?
const protocol = entry.ClientRequestProtocol;
2017-05-23 22:00:09 +00:00
if (protocol) {
2018-02-18 02:00:56 +00:00
incr(
`stats-protocolRequests-${dayKey}`,
protocol,
1,
thirtyDaysLater
);
}
2017-05-23 22:00:09 +00:00
// Q: How many requests do we receive from a hostname per day?
// Q: How many bytes do we serve to a hostname per day?
const referer = entry.ClientRequestReferer;
2019-01-06 00:50:05 +00:00
const hostname = referer && url.parse(referer).hostname;
2017-05-23 22:00:09 +00:00
if (hostname) {
2018-02-18 02:00:56 +00:00
incr(`stats-hostnameRequests-${dayKey}`, hostname, 1, sevenDaysLater);
incr(
`stats-hostnameBytes-${dayKey}`,
hostname,
entry.EdgeResponseBytes,
2018-02-18 02:00:56 +00:00
sevenDaysLater
);
}
2017-05-23 22:00:09 +00:00
})
2018-12-17 17:38:05 +00:00
.on('end', () => {
resolve({ counters, expireat, totalEntries });
2018-02-18 02:00:56 +00:00
});
});
}
2017-05-23 22:00:09 +00:00
function processLogs(stream) {
return computeCounters(stream).then(
({ counters, expireat, totalEntries }) => {
Object.keys(counters).forEach(key => {
const values = counters[key];
Object.keys(values).forEach(member => {
2019-01-06 00:50:05 +00:00
data.zincrby(key, values[member], member);
});
if (expireat[key]) {
2019-01-06 00:50:05 +00:00
data.expireat(key, expireat[key]);
}
2018-02-18 02:00:56 +00:00
});
2017-05-29 05:41:01 +00:00
return totalEntries;
}
);
}
2017-05-23 22:00:09 +00:00
2018-08-26 03:14:49 +00:00
function ingestLogsForZone(zone, startDate, endDate) {
const startSeconds = toSeconds(startDate);
const endSeconds = toSeconds(endDate);
const startFetchTime = Date.now();
const fields = [
2018-12-17 17:38:05 +00:00
'EdgeStartTimestamp',
'EdgeResponseStatus',
'EdgeResponseBytes',
'ClientRequestProtocol',
'ClientRequestURI',
'ClientRequestReferer'
];
2019-01-06 00:50:05 +00:00
return cloudflare
.getLogs(
zone.id,
2017-05-20 05:41:24 +00:00
stringifySeconds(startSeconds),
stringifySeconds(endSeconds),
2019-01-06 00:50:05 +00:00
fields
)
.then(stream => {
const endFetchTime = Date.now();
2017-05-20 05:41:24 +00:00
2019-01-06 00:50:05 +00:00
console.log(
'Fetched logs for %s from %s to %s (%dms)',
zone.name,
2019-01-06 00:50:05 +00:00
stringifySeconds(startSeconds),
stringifySeconds(endSeconds),
endFetchTime - startFetchTime
);
2019-01-06 00:50:05 +00:00
const startProcessTime = Date.now();
return processLogs(stream).then(totalEntries => {
const endProcessTime = Date.now();
console.log(
'Processed %d log entries for %s (%dms)',
totalEntries,
zone.name,
endProcessTime - startProcessTime
);
});
});
}
2017-05-20 05:41:24 +00:00
2018-08-26 03:14:49 +00:00
function getZones(domainNames) {
2019-01-06 00:50:05 +00:00
return Promise.all(domainNames.map(cloudflare.getZones)).then(results =>
2018-08-26 03:14:49 +00:00
results.reduce((memo, zones) => memo.concat(zones))
);
}
2019-01-06 00:50:05 +00:00
export default function ingestLogs(startDate, endDate) {
2018-08-26 03:14:49 +00:00
return Promise.resolve(cachedZones || getZones(domainNames)).then(zones => {
if (!cachedZones) cachedZones = zones;
2017-05-20 05:41:24 +00:00
2018-08-26 03:14:49 +00:00
return Promise.all(
zones.map(zone => ingestLogsForZone(zone, startDate, endDate))
);
});
2017-05-20 05:41:24 +00:00
}