def base = 'http://www.the-asci.org/core/asci' (1969..2011).each{ def m = "${base}/list.php?type=year&key=$it".toURL().text =~ /<a href=profile\.php\?pid=(\d*)>/ m.each{ def text = "${base}/profile.php?pid=${it[1]}".toURL().text, data = [:] def n = text =~ /<div class=member_name>(.*?)<\/div>/ data.name = n[0][1] n = text =~ /<div class=member_address>(.*?)<\/div>/ def address = [] n.each{address << it[1].replaceAll(/<\/?.*?>/,'')} data.contactInfo = address.join('\n') def l = text =~ /<div width='100%'><div class=rh>Specialties<\/div>(.*?)<\/div>(<div width='100%'>|\s{3})/, specialties = [] if(l) (l[0][1] =~ /<div class=r\d>(\w*)<\/div>/).each{specialties << it[1]} data.specialties = specialties.join(', ') new Physician(data).save() } }
asci
猜你喜欢
转载自sam-ds-chen.iteye.com/blog/1271250
今日推荐
周排行