vlr-stats-scraper 1.2.0 → 1.4.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/main.mjs +91 -86
- package/package.json +1 -1
- package/scripts/player-info-scraper.mjs +16 -13
package/main.mjs
CHANGED
|
@@ -1,86 +1,91 @@
|
|
|
1
|
-
#!/usr/bin/env node
|
|
2
|
-
import promptSync from 'prompt-sync';
|
|
3
|
-
import { scrapeWebsite } from './scripts/player-link-scraper.mjs';
|
|
4
|
-
import scrapePlayer from './scripts/player-info-scraper.mjs';
|
|
5
|
-
import {
|
|
6
|
-
savePlayer,
|
|
7
|
-
saveEvent,
|
|
8
|
-
addPlayerToEvent,
|
|
9
|
-
getEventByName,
|
|
10
|
-
emptyEvent,
|
|
11
|
-
clearEvent,
|
|
12
|
-
clearDatabase,
|
|
13
|
-
} from './scripts/db.mjs';
|
|
14
|
-
|
|
15
|
-
const args = process.argv.slice(2);
|
|
16
|
-
|
|
17
|
-
if (args[0] === '--clear') {
|
|
18
|
-
const prompt = promptSync();
|
|
19
|
-
const answer = prompt('This deletes every event and player in players.db. Type "yes" to continue: ');
|
|
20
|
-
if (answer?.trim().toLowerCase() === 'yes') {
|
|
21
|
-
clearDatabase();
|
|
22
|
-
console.log('Cleared players.db');
|
|
23
|
-
} else {
|
|
24
|
-
console.log('Cancelled');
|
|
25
|
-
}
|
|
26
|
-
process.exit(0);
|
|
27
|
-
}
|
|
28
|
-
|
|
29
|
-
if (args[0] === '--clear-event') {
|
|
30
|
-
const name = args[1];
|
|
31
|
-
if (!name) {
|
|
32
|
-
console.error('Usage: valstats --clear-event <event-name>');
|
|
33
|
-
process.exit(1);
|
|
34
|
-
}
|
|
35
|
-
if (clearEvent(name)) {
|
|
36
|
-
console.log(`Removed "${name}" and its players from players.db`);
|
|
37
|
-
} else {
|
|
38
|
-
console.error(`No event named "${name}" in players.db`);
|
|
39
|
-
process.exit(1);
|
|
40
|
-
}
|
|
41
|
-
process.exit(0);
|
|
42
|
-
}
|
|
43
|
-
|
|
44
|
-
const [link, year, eventName] = args;
|
|
45
|
-
|
|
46
|
-
if (!link || !year || !eventName) {
|
|
47
|
-
console.error('Usage: valstats <link> <year> <event-name>');
|
|
48
|
-
console.error(' valstats --clear-event <event-name>');
|
|
49
|
-
console.error(' valstats --clear');
|
|
50
|
-
process.exit(1);
|
|
51
|
-
}
|
|
52
|
-
|
|
53
|
-
const isStage = eventName.toLowerCase().includes('stage');
|
|
54
|
-
|
|
55
|
-
let region = null;
|
|
56
|
-
if (isStage) {
|
|
57
|
-
const prompt = promptSync();
|
|
58
|
-
region = prompt('Enter the region: ');
|
|
59
|
-
}
|
|
60
|
-
|
|
61
|
-
// Re-adding an event + year that's already saved replaces its old players.
|
|
62
|
-
const existing = getEventByName(eventName);
|
|
63
|
-
if (existing && existing.year === parseInt(year, 10)) {
|
|
64
|
-
emptyEvent(eventName);
|
|
65
|
-
console.log(`"${eventName}" (${year}) is already in players.db, replacing it`);
|
|
66
|
-
}
|
|
67
|
-
|
|
68
|
-
const eventId = saveEvent({
|
|
69
|
-
name: eventName,
|
|
70
|
-
url: link,
|
|
71
|
-
region,
|
|
72
|
-
year: parseInt(year, 10),
|
|
73
|
-
stage: null,
|
|
74
|
-
});
|
|
75
|
-
|
|
76
|
-
const links = await scrapeWebsite(link);
|
|
77
|
-
|
|
78
|
-
|
|
79
|
-
|
|
80
|
-
|
|
81
|
-
|
|
82
|
-
|
|
83
|
-
|
|
84
|
-
|
|
85
|
-
|
|
86
|
-
|
|
1
|
+
#!/usr/bin/env node
|
|
2
|
+
import promptSync from 'prompt-sync';
|
|
3
|
+
import { scrapeWebsite } from './scripts/player-link-scraper.mjs';
|
|
4
|
+
import scrapePlayer from './scripts/player-info-scraper.mjs';
|
|
5
|
+
import {
|
|
6
|
+
savePlayer,
|
|
7
|
+
saveEvent,
|
|
8
|
+
addPlayerToEvent,
|
|
9
|
+
getEventByName,
|
|
10
|
+
emptyEvent,
|
|
11
|
+
clearEvent,
|
|
12
|
+
clearDatabase,
|
|
13
|
+
} from './scripts/db.mjs';
|
|
14
|
+
|
|
15
|
+
const args = process.argv.slice(2);
|
|
16
|
+
|
|
17
|
+
if (args[0] === '--clear') {
|
|
18
|
+
const prompt = promptSync();
|
|
19
|
+
const answer = prompt('This deletes every event and player in players.db. Type "yes" to continue: ');
|
|
20
|
+
if (answer?.trim().toLowerCase() === 'yes') {
|
|
21
|
+
clearDatabase();
|
|
22
|
+
console.log('Cleared players.db');
|
|
23
|
+
} else {
|
|
24
|
+
console.log('Cancelled');
|
|
25
|
+
}
|
|
26
|
+
process.exit(0);
|
|
27
|
+
}
|
|
28
|
+
|
|
29
|
+
if (args[0] === '--clear-event') {
|
|
30
|
+
const name = args[1];
|
|
31
|
+
if (!name) {
|
|
32
|
+
console.error('Usage: valstats --clear-event <event-name>');
|
|
33
|
+
process.exit(1);
|
|
34
|
+
}
|
|
35
|
+
if (clearEvent(name)) {
|
|
36
|
+
console.log(`Removed "${name}" and its players from players.db`);
|
|
37
|
+
} else {
|
|
38
|
+
console.error(`No event named "${name}" in players.db`);
|
|
39
|
+
process.exit(1);
|
|
40
|
+
}
|
|
41
|
+
process.exit(0);
|
|
42
|
+
}
|
|
43
|
+
|
|
44
|
+
const [link, year, eventName] = args;
|
|
45
|
+
|
|
46
|
+
if (!link || !year || !eventName) {
|
|
47
|
+
console.error('Usage: valstats <link> <year> <event-name>');
|
|
48
|
+
console.error(' valstats --clear-event <event-name>');
|
|
49
|
+
console.error(' valstats --clear');
|
|
50
|
+
process.exit(1);
|
|
51
|
+
}
|
|
52
|
+
|
|
53
|
+
const isStage = eventName.toLowerCase().includes('stage');
|
|
54
|
+
|
|
55
|
+
let region = null;
|
|
56
|
+
if (isStage) {
|
|
57
|
+
const prompt = promptSync();
|
|
58
|
+
region = prompt('Enter the region: ');
|
|
59
|
+
}
|
|
60
|
+
|
|
61
|
+
// Re-adding an event + year that's already saved replaces its old players.
|
|
62
|
+
const existing = getEventByName(eventName);
|
|
63
|
+
if (existing && existing.year === parseInt(year, 10)) {
|
|
64
|
+
emptyEvent(eventName);
|
|
65
|
+
console.log(`"${eventName}" (${year}) is already in players.db, replacing it`);
|
|
66
|
+
}
|
|
67
|
+
|
|
68
|
+
const eventId = saveEvent({
|
|
69
|
+
name: eventName,
|
|
70
|
+
url: link,
|
|
71
|
+
region,
|
|
72
|
+
year: parseInt(year, 10),
|
|
73
|
+
stage: null,
|
|
74
|
+
});
|
|
75
|
+
|
|
76
|
+
const links = await scrapeWebsite(link);
|
|
77
|
+
|
|
78
|
+
if (!links) {
|
|
79
|
+
console.error(`Couldn't get the player list from ${link}`);
|
|
80
|
+
process.exit(1);
|
|
81
|
+
}
|
|
82
|
+
|
|
83
|
+
for (const playerLink of links) {
|
|
84
|
+
const player = await scrapePlayer(playerLink, region);
|
|
85
|
+
if (player) {
|
|
86
|
+
savePlayer(player);
|
|
87
|
+
addPlayerToEvent(eventId, player.name);
|
|
88
|
+
console.log(`Saved ${player.name} to players.db under "${eventName}"`);
|
|
89
|
+
console.log(player);
|
|
90
|
+
}
|
|
91
|
+
}
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "vlr-stats-scraper",
|
|
3
|
-
"version": "1.
|
|
3
|
+
"version": "1.4.0",
|
|
4
4
|
"description": "Scrapes player stats from vlr.gg and stores them in a local SQLite database, grouped by tournament or stage.",
|
|
5
5
|
"homepage": "https://github.com/OverYander/VLR-Webscraper#readme",
|
|
6
6
|
"bugs": {
|
|
@@ -15,7 +15,7 @@ function getClient() {
|
|
|
15
15
|
}
|
|
16
16
|
|
|
17
17
|
async function scrapePlayer(url, region) {
|
|
18
|
-
console.log('scrapePlayer ran')
|
|
18
|
+
// console.log('scrapePlayer ran')
|
|
19
19
|
try {
|
|
20
20
|
const scrapeData = await getClient().scrape.startAndWait({
|
|
21
21
|
url: url,
|
|
@@ -51,12 +51,14 @@ node scripts/player-info-scraper.mjs
|
|
|
51
51
|
*/
|
|
52
52
|
|
|
53
53
|
function getPlayerName(link) {
|
|
54
|
-
|
|
55
|
-
|
|
54
|
+
// Links look like .../player/123/name/?timespan=all, so take the last path segment
|
|
55
|
+
let name = new URL(link).pathname.split('/').filter(ele => ele !== '').pop()
|
|
56
|
+
console.log(name)
|
|
57
|
+
return name
|
|
56
58
|
}
|
|
57
59
|
|
|
58
60
|
function findAgents($) {
|
|
59
|
-
console.log('findAgents ran')
|
|
61
|
+
// console.log('findAgents ran')
|
|
60
62
|
const header = $('.mod-agent-rows').first().children().first()
|
|
61
63
|
|
|
62
64
|
const statCategories = header.text().split('\n').map(ele => ele.trim()).filter(ele => ele !== '')
|
|
@@ -66,7 +68,7 @@ function findAgents($) {
|
|
|
66
68
|
const agents = agentStats.map((i, el) => {
|
|
67
69
|
el = $(el)
|
|
68
70
|
return {
|
|
69
|
-
agent: el.children().first().
|
|
71
|
+
agent: el.children().first().find('img').attr('alt')?.trim(),
|
|
70
72
|
stats: el.text().split('\t').map(ele => ele.trim()).filter(ele => ele !== '')
|
|
71
73
|
}
|
|
72
74
|
}).get()
|
|
@@ -87,14 +89,15 @@ function findAgents($) {
|
|
|
87
89
|
}
|
|
88
90
|
|
|
89
91
|
function getCurTeam($) {
|
|
90
|
-
console.log('getCurTeam ran')
|
|
92
|
+
// console.log('getCurTeam ran')
|
|
91
93
|
const currentTeam = $('h2:contains("Current Teams")').first().next().text().trim()
|
|
92
94
|
const currentTeamName = currentTeam.split('\t').filter(ele => ele.trim() !== '')
|
|
93
|
-
|
|
95
|
+
// Free agents have no current team
|
|
96
|
+
return currentTeamName.length ? currentTeamName[0].trim() : null
|
|
94
97
|
}
|
|
95
98
|
|
|
96
99
|
function getPastTeam($) {
|
|
97
|
-
console.log('getPastTeam ran')
|
|
100
|
+
// console.log('getPastTeam ran')
|
|
98
101
|
const pastTeams = $('h2:contains("Past Teams")').first().next().text().trim()
|
|
99
102
|
const pastTeamNames = pastTeams.split('\t').filter(ele => ele.trim() !== ''
|
|
100
103
|
&& !ele.includes('joined')
|
|
@@ -105,18 +108,18 @@ function getPastTeam($) {
|
|
|
105
108
|
}
|
|
106
109
|
|
|
107
110
|
function getWinnings($) {
|
|
108
|
-
console.log('getWinnings ran')
|
|
111
|
+
// console.log('getWinnings ran')
|
|
109
112
|
const events = $('h2:contains("Event Placements")').first().next()
|
|
110
113
|
const winnings = events.children().first().text().trim()
|
|
111
|
-
// console.log(winnings.split('\t').map(ele => ele.trim()).filter(ele => ele !== '')[1])
|
|
114
|
+
// // console.log(winnings.split('\t').map(ele => ele.trim()).filter(ele => ele !== '')[1])
|
|
112
115
|
return winnings.split('\t').map(ele => ele.trim()).filter(ele => ele !== '')[1]
|
|
113
116
|
|
|
114
117
|
}
|
|
115
118
|
|
|
116
119
|
function getNationality($) {
|
|
117
|
-
console.log('getNationality ran')
|
|
120
|
+
// console.log('getNationality ran')
|
|
118
121
|
const header = $('.flag').first()
|
|
119
|
-
console.log('flags found:', $('.flag').length);
|
|
120
|
-
console.log(header.parent().text().trim())
|
|
122
|
+
// console.log('flags found:', $('.flag').length);
|
|
123
|
+
// console.log(header.parent().text().trim())
|
|
121
124
|
return header.parent().text().trim()
|
|
122
125
|
}
|