保存RSS源中的所有图像

Posted 2021-02-23

Gets "src" attributes from all "img" tags in RSS entries descriptions

RSS_URL = 'http://35photo.ru/rss/photo_day.xml'
OUT_NAM = "/var/tmp/${new Date().format('YYYYMMddHHmmss')}"
 
new XmlParser().parse(RSS_URL).channel.item.description*.text()*.findAll(/(?i)<imgs[^>]+>/) {im ->
	im.findAll(/(?i)src=['"]([^'"]+.([^.'"]+))['"]/) {ma, sr, ex ->
		new FileOutputStream(OUT_NAM + "${sr.hashCode()}.$ex").write(new URL(sr).getBytes())
	}
}