Scrape.py

from mechanize import Browser
from BeautifulSoup import BeautifulSoup

mech = Browser()
years = ['2009','2008']
teams = ['aag','cai','sbr','ial','fau','bal','way','hba','nbh','mbk','bry','sbb','mam','wbg','hac','naj','wai','sad','dau','abh','uvs','scc','dae','hbd','iao','aaj','mbn','wbj','sbe','tap','nam','map','flg','wal','kab','pal','sag','vac','cbm','xaa','dah','tbq','scf','tas','hay','mbq','cao','sbh','wao','lan','oai','gad','tba','fak','kae','saj','rae','bab','scy','pao','tac','vaf','cbp','yaa','sci','nfl','dak','aap','bau','tav','laq','mav','bae','nas','nba','mbd','gag','iae','jab','fan','hal','cab','nac','maf','rah','eaa','par','sam','laa','pab','vai','may','ncb','aas','cau','gaj','cbc','nbd','mbg','lat','nav','jae','eat','sbn','hao','nds','uaf','tai','rak','faa','ead','pau','sap','lad','aac','cbv','lbu','aav','nay','cax','sas','aaf','wax','cah','iak','fat','sbq','oab','nai','mal','tal','pah','qae','sac','wah','dat','cbi','sbt','dad','mbm','ban','sav','ian','tao','nal','wbi','hae','oae','raa','kaa','sds','saf','vab','kns','cbl','dag','can','mbp','baq','tar','iaa','oah','gac','sbg','nao','sai','wan','ccm','mab','rad','baa','pan','vae','mbs','caq','tau','sch','nar','tbc','lap','fam','jaa','hak','oak','sdy','gaf','nab','tae','mae','rag','caa','paq','cbr','vah','mbv','cat','baw','aar','dam','bag','las','iag','wat','nbc','mbf','max','mah','eac','wbb','uae','aab','sao','nae','wad','nax','baz','aca','mbi','nbf','sbp','lav','cag','gal','eaf','pbe','laf','oaa','nah','mak','tak','sab','pag','abf','aax','cbh','aah','lay','dac','bam','caj','sca','iam','fav','oad','sbc','had','lai','tan','man','nak','eai','wbh','sae','cbk','vaa','abi','dav','sax','bap','scd','maq','oag','nan','gab','sbf','wbk','lal','sah','sec','maa','sby','vad','cbn','cap','aan','bas','scg','iac','gae','naq','cdm','oaj','fal','sbi','wap','raf','uaa','tad','kaf','naa','cbq','yab','aaq','dal','bav','taw','wbq','iav','nbb','jac','ear','maw','nat','gah','iaf','fao','sbl','ham','mbe','lab','uad','pas','tag','nad','cbt','wac','maz','aat','taz','bai','gak','far','iai','sbo','lau','abb','nbe','mbh','taj','eae','fab','wbd','nag','maj','waf','cdc','paf','saa','cay','ncf','naz','cbg','dar']
file = open('team_data.csv', 'at')

for team in teams:
  for year in years:
    url = "http://rivals.yahoo.com/ncaa/basketball/teams/"+team+"/stats?year="+year
    page = mech.open(url)
    html = page.read()
    soup = BeautifulSoup(html)
    table = soup.find("table",width="100%", border="0", cellspacing="0", cellpadding="0",attrs={"class" : "yspwhitebg"})
    rows = table.findAll('tr')[1:]
    rows.pop(0)
    for row in rows:
        col = row.findAll('td')
        list = [year,team]
        for x in col:
          if x and x.string:
            list.append(x.string.replace(" ",""))
          else:
            list.append("")
        file.write(",".join(list)+"\n")
print "Finished writing team_data.csv"