Fixed release sites for profile scraping.

This commit is contained in:
2020-03-11 03:01:37 +01:00
parent 9c8d914b75
commit 0d0acb6f3c
6 changed files with 41 additions and 19 deletions

View File

@@ -8,7 +8,7 @@ const { JSDOM } = require('jsdom');
const moment = require('moment');
const logger = require('../logger')(__filename);
const { get, geta, ctxa } = require('../utils/q');
const { get, geta, ctxa, parseDate } = require('../utils/q');
const { heightToCm } = require('../utils/convert');
const slugify = require('../utils/slugify');
@@ -304,27 +304,37 @@ function scrapeProfile(html, url, actorName) {
const { document } = new JSDOM(html).window;
const bio = document.querySelector('.model_bio').textContent;
const avatarEl = document.querySelector('.model_bio_pic');
const avatarEl = document.querySelector('.model_bio_pic img');
const profile = {
name: actorName,
};
const heightString = bio.match(/\d+ feet \d+ inches/);
const ageString = bio.match(/Age:\s*\d{2}/);
const ageString = bio.match(/Age:\s*(\d{2})/);
const birthDateString = bio.match(/Age:\s*(\w+ \d{1,2}, \d{4})/);
const measurementsString = bio.match(/\w+-\d+-\d+/);
if (birthDateString) profile.birthdate = parseDate(birthDateString[1], 'MMMM D, YYYY');
if (ageString) profile.age = Number(ageString[1]);
if (heightString) profile.height = heightToCm(heightString[0]);
if (ageString) profile.age = Number(ageString[0].match(/\d{2}/)[0]);
if (measurementsString) [profile.bust, profile.waist, profile.hip] = measurementsString[0].split('-');
if (measurementsString) {
const [bust, waist, hip] = measurementsString[0].split('-');
if (bust) profile.bust = bust;
if (waist) profile.waist = Number(waist);
if (hip) profile.hip = Number(hip);
}
if (avatarEl) {
const avatarSources = [
avatarEl.innerHTML.slice(avatarEl.innerHTML.indexOf('src0_3x') + 9, avatarEl.innerHTML.indexOf('3x.jpg') + 6).trim(),
avatarEl.innerHTML.slice(avatarEl.innerHTML.indexOf('src0_2x') + 9, avatarEl.innerHTML.indexOf('2x.jpg') + 6).trim(),
avatarEl.innerHTML.slice(avatarEl.innerHTML.indexOf('src0_1x') + 9, avatarEl.innerHTML.indexOf('1x.jpg') + 6).trim(),
avatarEl.innerHTML.slice(avatarEl.innerHTML.indexOf('src0') + 6, avatarEl.innerHTML.indexOf('set.jpg') + 7).trim(),
avatarEl.innerHTML.slice(avatarEl.innerHTML.indexOf('src') + 5, avatarEl.innerHTML.indexOf('set.jpg') + 7).trim(),
avatarEl.getAttribute('src0_3x'),
avatarEl.getAttribute('src0_2x'),
avatarEl.getAttribute('src0_1x'),
avatarEl.getAttribute('src0'),
avatarEl.getAttribute('src'),
].filter(Boolean);
if (avatarSources.length) profile.avatar = avatarSources;
@@ -332,6 +342,8 @@ function scrapeProfile(html, url, actorName) {
profile.releases = Array.from(document.querySelectorAll('.category_listing_block .update_details > a:first-child'), el => el.href);
console.log(profile);
return profile;
}