def main(self):
pages = self._crawl_pages
precision = lambda d: '{0:.2f}'.format(float(d.strip()) * 10000)
for page in range(1, pages + 1):
url = self._base_url + self._query_string.format(page)
dt, code_price_volu_amou, buy_sale = self.parse_web(url)
for i in range(len(dt)):
# Note that this program to securities (stock and fund , fond is None)
# True insert all data at dt[i] date, otherwise insert omission data
rp_flag = self._latest_date < dt[i]
secu, curr = secu_currency(code_price_volu_amou[i][0], 'stock', 'fund')
clo_price = close_price(secu, dt[i], 'mysql')
disc = '' if clo_price is None else ('%.4f' % (float(code_price_volu_amou[i][1]) -
float(clo_price) - 1))
total = total_equity(secu, dt[i], 'vary', 'stock')
volu = float(precision(code_price_volu_amou[i][2]))
ratio = '' if total is None else ('%.4f' % ((volu / total) * 100))
name = ''.join([dt[i]] + code_price_volu_amou[i] + buy_sale[i])
uid = str(uuid.uuid5(uuid.NAMESPACE_DNS, name))
data = {
'secu': secu, 'y': dt[i], 'buy': buy_sale[i][0], 'sale': buy_sale[i][1],
'price': code_price_volu_amou[i][1], 'disc': disc, 'ratio': ratio, 'stat': 2,
'volu': volu, 'amou': precision(code_price_volu_amou[i][3]),
'c': {'cd': curr, 'szh': '人民币' if curr == 'CNY' else '港币', 'en': curr},
'uuid': uid, 'crt': datetime.datetime.now(), 'typ': 'szx_secu','upt': datetime.datetime.now()
}
if not rp_flag and not self._coll_in.get({'uuid': uid}):
self._coll_in.insert(data)
elif rp_flag:
self._coll_in.insert(data)
print u'权益类证券大宗交易 page: %d done!' % page
self._coll_in.disconnect()
def main(self):
pages = self._crawl_pages
for page in range(1, pages + 1):
url = self._base_url + self._query_string.format(page)
for item in self.parse_web(url):
# Note that this program to securities (stock and fund , fond is None)
# True insert all data at item[0] date, otherwise insert omission data
rp_flag = self._latest_date < item[0]
secu, curr = secu_currency(item[1], typ_bond='bond')
clo_price = close_price(secu, item[0], 'mysql')
disc = '' if clo_price is None else ('%.4f' % (float(item[2]) - float(clo_price) - 1))
total = total_equity(secu, item[0], 'vary', 'stock')
ratio = '' if total is None else ('%.4f' % ((float(item[3]) / total) * 100))
uid = str(uuid.uuid5(uuid.NAMESPACE_DNS, ''.join(item)))
data = {
'secu': secu, 'y': item[0], 'buy': item[4], 'sale': item[5],
'price': item[2], 'disc': disc, 'ratio': ratio, 'stat': 2,
'volu': '{0:.2f}'.format(float(item[3])),
'amou': '{0:.2f}'.format(float(item[3]) * float(item[2])),
'c': {'cd': curr, 'szh': '人民币' if curr == 'CNY' else '港币', 'en': curr},
'uuid': uid, 'crt': datetime.datetime.now(), 'typ': 'szx_bond',
'upt': datetime.datetime.now()
}
if not rp_flag and not self._coll_in.get({'uuid': uid}):
self._coll_in.insert(data)
elif rp_flag:
self._coll_in.insert(data)
print u'债券大宗交易 page: %d done!' % page
def main(self):
pages = self._crawl_pages
precision = lambda d: '{0:.2f}'.format(float(d.strip()) * 10000)
for page in range(1, pages + 1):
url = self._base_url + self._query_string.format(page)
dt, code_price_volu_amou, buy_sale = self.parse_web(url)
for i in range(len(dt)):
# Note that this program to securities (stock and fund , fond is None)
# True insert all data at dt[i] date, otherwise insert omission data
rp_flag = self._latest_date < dt[i]
secu, curr = secu_currency(code_price_volu_amou[i][0], 'stock',
'fund')
clo_price = close_price(secu, dt[i], 'mysql')
disc = '' if clo_price is None else (
'%.4f' %
(float(code_price_volu_amou[i][1]) - float(clo_price) - 1))
total = total_equity(secu, dt[i], 'vary', 'stock')
volu = float(precision(code_price_volu_amou[i][2]))
ratio = '' if total is None else ('%.4f' %
((volu / total) * 100))
name = ''.join([dt[i]] + code_price_volu_amou[i] + buy_sale[i])
uid = str(uuid.uuid5(uuid.NAMESPACE_DNS, name))
data = {
'secu': secu,
'y': dt[i],
'buy': buy_sale[i][0],
'sale': buy_sale[i][1],
'price': code_price_volu_amou[i][1],
'disc': disc,
'ratio': ratio,
'stat': 2,
'volu': volu,
'amou': precision(code_price_volu_amou[i][3]),
'c': {
'cd': curr,
'szh': '人民币' if curr == 'CNY' else '港币',
'en': curr
},
'uuid': uid,
'crt': datetime.datetime.now(),
'typ': 'szx_secu',
'upt': datetime.datetime.now()
}
if not rp_flag and not self._coll_in.get({'uuid': uid}):
self._coll_in.insert(data)
elif rp_flag:
self._coll_in.insert(data)
print u'权益类证券大宗交易 page: %d done!' % page
self._coll_in.disconnect()
def insert(typ_key, secu_bond_datas):
assert typ_key == 'sf_data' or typ_key == 'bond_data'
coll_in = query_string.coll_in
data_list = secu_bond_datas[typ_key]
if typ_key == 'sf_data':
typ = 'sha_secu'
keys = ['y', 'secu', 'price', 'amou', 'volu', 'buy', 'sale', 'ot']
else:
typ = 'sha_bond'
keys = ['y', 'secu', 'price', 'volu', 'ot']
print('Now will insert [{0}] data to mongo'.format(typ))
for data in data_list:
to_data = dict(zip(keys, data))
to_data.pop('ot')
if typ_key == 'sf_data':
to_data['amou'] = '{0:.2f}'.format(float(to_data['amou']) * 10000)
to_data['volu'] = '{0:.2f}'.format(float(to_data['volu']) * 10000)
secu, curr = secu_currency(to_data['secu'], 'stock', 'fund')
else:
to_data['buy'], to_data['sale'] = '', ''
to_data['volu'] = '{0:.2f}'.format(float(to_data['volu']) * 10000 * 10)
to_data['amou'] = '{0:.2f}'.format(float(to_data['price']) * float(to_data['volu']))
secu, curr = secu_currency(to_data['secu'], typ_bond='bond')
clo_price = close_price(secu, to_data['y'], 'mysql')
disc = '' if clo_price is None else ('%.4f' % (float(to_data['price']) - float(clo_price) - 1))
total = total_equity(secu, to_data['y'], 'vary', 'stock')
ratio = '' if total is None else ('%.4f' % ((float(to_data['volu']) / total) * 100))
uid = str(uuid.uuid5(uuid.NAMESPACE_DNS, ''.join(data).encode('u8')))
new_dict = {'typ': typ, 'disc': disc, 'ratio': ratio, 'stat': 2,
'c': {'cd': curr, 'szh': '人民币' if curr == 'CNY' else '港币', 'en': curr},
'uuid': uid, 'crt': datetime.now(), 'secu': secu,
'upt': datetime.now()}
to_data.update(new_dict)
if not coll_in.get({'uuid': uid}):
coll_in.insert(to_data)
else:
print '%s uuid:' % typ_key, uid
print('[{0}] data insert ok.'.format(typ))
coll_in.disconnect()