def history_sysframe(exchange, url, force):
for c in Collection.query({'exchange': exchange}):
try:
if not force and history_exists(c):
continue
# 拿到数据文件
theurl = ('{}/hqApplet/data/day/00{}.day.zip').format(
url, c.symbol)
log.info('fetching exchange {} url {}'.format(exchange, theurl))
r = session.get(
theurl, timeout=(5, 10), verify=False)
if r.status_code != 200:
log.warning('{}_{}下载失败, 错误码: {}'
''.format(exchange, c.symbol, r.status_code))
continue
content = gzip.decompress(r.content)
# 解析之
num_rows = struct.unpack('>i', content[0:4])[0]
kline_days = []
for i in range(num_rows):
raw_row = content[4 + 40 * i: 4 + 40 * i + 40]
row = struct.unpack('>i5f2ifi', raw_row)
t = row[0]
date = datetime(year=int(str(t)[0:2]) + 1997,
month=int(str(t)[2:4]),
day=int(str(t)[4:6]),
minute=int(str(t)[6:8] or 0),
second=int(str(t)[8:10] or 0))
# row[6]总是0, 不知道是啥
q = {
'exchange': exchange,
'symbol': c.symbol,
'quote_type': '1d',
'quote_at': date,
'open_': row[1],
'high': row[2],
'low': row[3],
'close': row[4],
'mean': row[5],
'volume': row[7],
'amount': row[8],
'quantity': row[9],
}
if kline_days:
q['lclose'] = kline_days[-1]['close']
save_quotes(q, c, first_quote=False)
else:
save_quotes(q, c, first_quote=True)
kline_days.append(q)
except:
log.exception('{}:{} 抓取失败'.format(c.exchange, c.symbol))
def history_sysframe(exchange, url, force):
for c in Collection.query({'exchange': exchange}):
try:
if not force and history_exists(c):
continue
# 拿到数据文件
theurl = ('{}/hqApplet/data/day/00{}.day.zip').format(
url, c.symbol)
log.info('fetching exchange {} url {}'.format(exchange, theurl))
r = session.get(theurl, timeout=(5, 10), verify=False)
if r.status_code != 200:
log.warning('{}_{}下载失败, 错误码: {}'
''.format(exchange, c.symbol, r.status_code))
continue
content = gzip.decompress(r.content)
# 解析之
num_rows = struct.unpack('>i', content[0:4])[0]
kline_days = []
for i in range(num_rows):
raw_row = content[4 + 40 * i:4 + 40 * i + 40]
row = struct.unpack('>i5f2ifi', raw_row)
t = row[0]
date = datetime(year=int(str(t)[0:2]) + 1997,
month=int(str(t)[2:4]),
day=int(str(t)[4:6]),
minute=int(str(t)[6:8] or 0),
second=int(str(t)[8:10] or 0))
# row[6]总是0, 不知道是啥
q = {
'exchange': exchange,
'symbol': c.symbol,
'quote_type': '1d',
'quote_at': date,
'open_': row[1],
'high': row[2],
'low': row[3],
'close': row[4],
'mean': row[5],
'volume': row[7],
'amount': row[8],
'quantity': row[9],
}
if kline_days:
q['lclose'] = kline_days[-1]['close']
save_quotes(q, c, first_quote=False)
else:
save_quotes(q, c, first_quote=True)
kline_days.append(q)
except:
log.exception('{}:{} 抓取失败'.format(c.exchange, c.symbol))
def save_quotes(q, c, first_quote=False):
if first_quote:
# 看一下需不需要补齐申购数据
if not c.name.endswith('指数'):
if not c.offers_at or abs((c.offers_at - q['quote_at']).days) >= 7:
c.offers_at = q['quote_at'] - timedelta(days=2)
c.offer_price = q['high'] / 1.3
if 'quantity' in q:
# 假定申购是全部数量的1/10
c.offer_quantity = q['quantity'] // 10
log.info('补齐数据:{}_{}, 申购价:{}, 申购量:{}, 时间:{}'
''.format(c.exchange, c.symbol, c.offer_price,
c.offer_quantity, c.offers_at))
# 然而实际上不补齐
# c.upsert()
Quote(q).upsert()
def save_quotes(q, c, first_quote=False):
if first_quote:
# 看一下需不需要补齐申购数据
if not c.name.endswith('指数'):
if not c.offers_at or abs((c.offers_at - q['quote_at']).days) >= 7:
c.offers_at = q['quote_at'] - timedelta(days=2)
c.offer_price = q['high'] / 1.3
if 'quantity' in q:
# 假定申购是全部数量的1/10
c.offer_quantity = q['quantity'] // 10
log.info('补齐数据:{}_{}, 申购价:{}, 申购量:{}, 时间:{}'
''.format(c.exchange, c.symbol,
c.offer_price, c.offer_quantity,
c.offers_at))
# 然而实际上不补齐
# c.upsert()
Quote(q).upsert()
def realtime(site):
conf = get_conf(site)
exchange = conf['abbr']
url = conf['quote']['realtime']['url']
type_ = conf['quote']['realtime']['type']
today = datetime.utcnow().replace(
minute=0, second=0, microsecond=0) + timedelta(hours=8)
if not url:
log.warning('{}尚未配置实时行情url'.format(exchange))
return
if today.hour < 9 or today.hour > 22:
log.warning('不在9点到22点之间, 不做解析')
return
today = today.replace(hour=0)
text = session.get(url, timeout=(3, 7)).text
quotes = parse_quotes(type_, text)
saved = 0
for q in quotes:
Collection.update_one({'exchange': exchange,
'symbol': q['symbol'].strip()},
{'$set': {'name': q['name']}},
upsert=True)
q['exchange'] = exchange
q['quote_type'] = '1d'
q['quote_at'] = today
if q['open_'] in ['—', '-', None, '']:
continue
else:
# 找到上一个交易日的数据, 如果和lclose不符则舍弃
# 需要保证数据每天更新/不足时需要把日线补足才能正常显示
lq = Quote.query_one({'exchange': exchange,
'symbol': q['symbol'].strip(),
'quote_type': '1d',
'quote_at': {'$lt': today}},
sort=[('quote_at', -1)])
if not lq or abs(lq.close - q['lclose']) < 0.01:
Quote(q).upsert()
saved += 1
log.info('{} 导入 {}/{} 条实时交易记录'.format(exchange, saved, len(quotes)))
def history_winner(exchange, url, force):
assert url.startswith('tcp://')
host, port = url[6:].split(':')
port = int(port)
s = socket.socket()
s.connect((host, port))
def get_day_data(symbol):
bsymbol = symbol.encode('utf-8')
bizdata = b''.join([
b'\xfe\x8f\x00\x00',
b'\x00' * 12,
b'\x01\x00',
b',\x00',
b'\x02\x04',
b'\x00',
b'\x00',
b'\x00\x00\x00\x00',
b'\x013' + bsymbol,
b'\x03\x00',
b'\x00\x00',
b'\x00\x00\x00\x00\x00\x00\x00\x00',
b'Z\x00\x10\x00\x013' + bsymbol,
b'\x00\x00\x00\x00',
b'\x00',
b'\x00',
])
bdata = b''.join([
b'\x95\x00\x00\x95',
b'11=524\x00',
b'13=8\x00',
b'5=36862\x00',
b'4=109\x00',
b'1=66\x008=',
bizdata,
b'\x00',
b'1=40\x0069=markid=60ba7308cab942ee961536a74ec7c5f9\x00\x00',
])
s.sendall(bdata)
batch = 8192
result = []
while True:
result.append(s.recv(batch))
if len(result[-1]) != batch:
break
return b''.join(result)
def parse_day_data(data):
size = struct.unpack('>i', b'\x00' + data[1:4])[0]
assert size + 4 == len(data)
m = re.compile(b'1=(\d+)\x008=').search(data)
if m:
bizsize = int(m.group(1))
start = m.span()[1]
try:
bbiz = zlib.decompress(data[start:start + bizsize])
except:
bbiz = data[start:start + bizsize]
assert bbiz[:4] == b'\xfe\x8f\x00\x00'
num_packs = struct.unpack('<H', bbiz[16:18])[0]
sizes = [
struct.unpack('<H', bbiz[18 + 2 * i:20 + 2 * i])[0]
for i in range(num_packs)
]
for i in range(num_packs):
start = 18 + 2 * num_packs + 0 if i == 0 else sizes[i - 1]
if bbiz[start:start + 2] == b'\x02\x04':
# K线数据
symbol = bbiz[start + 10:start + 16].decode('utf-8')
count = struct.unpack('<I', bbiz[start + 16:start + 20])[0]
kline_days = []
for i in range(count):
begin = start + 20 + i * 32
row = struct.unpack('<Iiiiiiii',
bbiz[begin:begin + 32])
q = {
'exchange': exchange,
'symbol': symbol,
'quote_type': '1d',
'quote_at':
datetime.strptime(str(row[0]), "%Y%m%d"),
'open_': row[1] / 100,
'high': row[2] / 100,
'low': row[3] / 100,
'close': row[4] / 100,
'amount': row[5] / 1.,
'volume': row[6],
}
if kline_days:
q['lclose'] = kline_days[-1]['close']
save_quotes(q, c, first_quote=False)
else:
save_quotes(q, c, first_quote=True)
kline_days.append(q)
for c in Collection.query({'exchange': exchange}):
try:
if not force and history_exists(c):
continue
if '$' not in c.symbol:
log.info('feching {}_{} on {}'.format(c.exchange, c.symbol,
url))
data = get_day_data(c.symbol)
parse_day_data(data)
except:
log.exception('{}_{}获取失败'.format(exchange, c.symbol))
def history_winner(exchange, url, force):
assert url.startswith('tcp://')
host, port = url[6:].split(':')
port = int(port)
s = socket.socket()
s.connect((host, port))
def get_day_data(symbol):
bsymbol = symbol.encode('utf-8')
bizdata = b''.join([
b'\xfe\x8f\x00\x00', b'\x00' * 12,
b'\x01\x00',
b',\x00',
b'\x02\x04', b'\x00', b'\x00', b'\x00\x00\x00\x00',
b'\x013' + bsymbol,
b'\x03\x00', b'\x00\x00',
b'\x00\x00\x00\x00\x00\x00\x00\x00',
b'Z\x00\x10\x00\x013' + bsymbol, b'\x00\x00\x00\x00',
b'\x00', b'\x00',
])
bdata = b''.join([
b'\x95\x00\x00\x95',
b'11=524\x00',
b'13=8\x00',
b'5=36862\x00',
b'4=109\x00',
b'1=66\x008=', bizdata, b'\x00',
b'1=40\x0069=markid=60ba7308cab942ee961536a74ec7c5f9\x00\x00',
])
s.sendall(bdata)
batch = 8192
result = []
while True:
result.append(s.recv(batch))
if len(result[-1]) != batch:
break
return b''.join(result)
def parse_day_data(data):
size = struct.unpack('>i', b'\x00' + data[1:4])[0]
assert size + 4 == len(data)
m = re.compile(b'1=(\d+)\x008=').search(data)
if m:
bizsize = int(m.group(1))
start = m.span()[1]
try:
bbiz = zlib.decompress(data[start: start + bizsize])
except:
bbiz = data[start: start + bizsize]
assert bbiz[:4] == b'\xfe\x8f\x00\x00'
num_packs = struct.unpack('<H', bbiz[16:18])[0]
sizes = [struct.unpack('<H',
bbiz[18 + 2 * i: 20 + 2 * i])[0]
for i in range(num_packs)]
for i in range(num_packs):
start = 18 + 2 * num_packs + 0 if i == 0 else sizes[i - 1]
if bbiz[start:start + 2] == b'\x02\x04':
# K线数据
symbol = bbiz[start + 10: start + 16].decode('utf-8')
count = struct.unpack(
'<I', bbiz[start + 16: start + 20])[0]
kline_days = []
for i in range(count):
begin = start + 20 + i * 32
row = struct.unpack('<Iiiiiiii',
bbiz[begin: begin + 32])
q = {
'exchange': exchange,
'symbol': symbol,
'quote_type': '1d',
'quote_at': datetime.strptime(str(row[0]),
"%Y%m%d"),
'open_': row[1] / 100,
'high': row[2] / 100,
'low': row[3] / 100,
'close': row[4] / 100,
'amount': row[5] / 1.,
'volume': row[6],
}
if kline_days:
q['lclose'] = kline_days[-1]['close']
save_quotes(q, c, first_quote=False)
else:
save_quotes(q, c, first_quote=True)
kline_days.append(q)
for c in Collection.query({'exchange': exchange}):
try:
if not force and history_exists(c):
continue
if '$' not in c.symbol:
log.info('feching {}_{} on {}'.format(c.exchange, c.symbol, url))
data = get_day_data(c.symbol)
parse_day_data(data)
except:
log.exception('{}_{}获取失败'.format(exchange, c.symbol))
