(An Untitled Masterwork)
ActivePublic
Actions

Authored by Urbanecm_WMF on Mar 30 2021, 9:42 PM.

Tags

None

Referenced Files

	F34198274: raw-paste-data.txt
	Mar 30 2021, 9:42 PM

Subscribers

None

	#!/usr/bin/env python3

	import requests
	import sys

	LANG = 'vi'
	API_URL = 'https://%s.wikipedia.org/w/api.php' % LANG
	ARTICLE_PATH_BASE = 'https://@@LANG@@.wikipedia.org/wiki/%s'.replace('@@LANG@@', LANG)
	LINK_REC_API = 'https://api.wikimedia.org/service/linkrecommendation/v1/linkrecommendations/wikipedia/@@LANG@@/%s'.replace('@@LANG@@', LANG)
	USER_AGENT = 'Urbanecm\'s script to generate a spreadsheet for link recommendation algorithm review (urbanecm@tools.wmflabs.org)'

	s = requests.Session()
	s.headers.update({'User-Agent': USER_AGENT})

	def make_request(payload):
	return s.post(API_URL, data=payload)

	def get_links_api(article_title):
	return LINK_REC_API % article_title

	def get_recommendations(article_title):
	r = s.get(get_links_api(article_title))
	try:
	data = r.json()
	except:
	print(r.content, file=sys.stderr)
	raise
	return data.get('links', [])

	def random_articles(number=150):
	payload = {
	"action": "query",
	"format": "json",
	"list": "random",
	"rnnamespace": "0",
	"rnfilterredir": "nonredirects",
	"rnlimit": number
	}
	r = make_request(payload)
	data = r.json()
	randomPages = data.get('query', {}).get('random', [])
	for record in randomPages:
	title = record.get('title', None)
	if title is not None:
	yield title

	def get_page_length(page_title):
	r = make_request({
	"action": "query",
	"format": "json",
	"prop": "revisions",
	"titles": page_title,
	"rvprop": "ids\|user\|size",
	"rvslots": "main",
	"rvlimit": "1",
	"rvdir": "older"
	})
	data = r.json().get('query', {}).get('pages', {})
	return data[list(data.keys())[0]].get('revisions', [])[0].get('size')

	article_suggested = 0
	for random_article in random_articles():
	recs = get_recommendations(random_article)
	if len(recs) == 0:
	print('Skipping %s, no links recommended' % random_article, file=sys.stderr)
	continue # skip article, 0 recs
	#print(random_article)
	links_api = get_links_api(random_article)
	print("\t".join([
	LANG,
	'=HYPERLINK("%s", "%s")' % (ARTICLE_PATH_BASE % random_article, random_article),
	str(get_page_length(random_article)),
	'=HYPERLINK("%s", "%s")' % (links_api, "API"),
	str(len(recs)),
	"FILL IN",
	"FILL IN"
	]))
	article_suggested += 1
	if article_suggested >= 25:
	break

Event Timeline

Urbanecm_WMF created this paste.Mar 30 2021, 9:42 PM

Urbanecm_WMF mentioned this in T278864: Add a link: evaluate link recommendation (Mar 30 2021).Mar 31 2021, 12:14 AM

(An Untitled Masterwork)ActivePublicActions

Event Timeline

(An Untitled Masterwork)
ActivePublic
Actions