def extract(al,progress=True):
seq=[]#
for a in al:
rd=parse(a['path'])
if progress:
print 'load '+a['path']+' succ'
main_keys=['type','religion']
wait_keys=['size','money']
weight_keys=['literacy']
#dd=defaultdict(lambda:{'size':0.0,'money':0.0})
sum_dict={mk:{} for mk in main_keys}
#sum_dict['sum']={wk:0.0 for wk in wait_keys}#sum是直接和，不分类
record={'pop_number':len(rd['pop'].values()),'sum':sum_dict,'main_number':copy.deepcopy(sum_dict)}
for pop in rd['pop'].values():
for mk in main_keys:
tk=pop[mk]
if not(record['sum'][mk].has_key(tk)):
record['sum'][mk][tk]={k:0.0 for k in wait_keys+weight_keys}
for wk in wait_keys:
record['sum'][mk][tk][wk]+=float(pop[wk])
for wk in weight_keys:
record['sum'][mk][tk][wk]+=float(pop[wk])*float(pop['size'])
#record['sum'][mk][wk]+=float(pop[wk])
record['main_number'][mk][tk]=record['main_number'][mk].get(tk,0)+1
sur_l=record['sum']['type'].values()
record['sum']['sum']={sk:sum([sd[sk] for sd in sur_l]) for sk in sur_l[0].keys()}
record['date']=a['date']
seq.append(record)
if progress:
print 'deal '+a['path']+' succ'
return seq
def to_sql_all(al,cache_path='cache'):
#接收路径词典列表，使用日期给cache编号
for a in al:
date,path=a['date'],a['path']
year,month,day=date
rd=parse(path)
save_path=cache_path+'/'+year+'.db'
to_sql(rd,connect_path=save_path)
print 'clear',date
print 'end'
def test(al):
global money_seq,size_seq
money_seq=[]
size_seq=[]
for a in al:
rd=parse(a['path'])
print 'load '+a['path']+' succ'
money_t=0.0
size_t=0.0
for pop in rd['pop'].values():
money_t+=float(pop['money'])
size_t+=float(pop['size'])
money_seq.append(money_t)
size_seq.append(size_t)
def sample():
return select2.get_building_price(select2.parse(path))
def sample(path="../SPA1901_09_28.v2"):
return parse(path)
def test2():
rd=parse('SPA1901_09_28.v2')
to_sql(rd,connect_path='bignews.db')
country_l=RD_list('bignews.db','country')
country_d=RD_dict(country_l)
country_d['ENG']