def get(self):
code = int(self.request.get('code', '0'))
q = Department.gql('WHERE dptCode >= :1 ORDER BY dptCode', code)
dpts = q.fetch(limit=2)
nowDpt = dpts[0]
soup = BeautifulSoup( urllib2.urlopen( nowDpt.dptLink ) )
list = soup.table.findAll('a')
for one in list:
text = one.text;
name = re.split('[0-9]', text)[0]
code = text[ len(name):].split(' ')[0].split('(')[0] # Dealing w/ the special cases
doc = Doctor.all().filter('docCode =', code).get()
if not doc and len(code) != 0:
doc = Doctor()
doc.docName = name
doc.docCode = code
doc.put()
if( len(dpts) > 1):
nextDpt = dpts[1]
nextUrl = '/parse/doctor?code=%d' % nextDpt.dptCode
nextName = nextDpt.dptName
else:
nextUrl = '/'
nextName = 'END OF PARSING'
context = {
'type' : 'Doctor',
'nextUrl' : nextUrl,
'nextName': nextName,
}
path = os.path.join( os.path.dirname('__file__'), 'templates', 'parser.html')
self.response.out.write( template.render( path, context) )
