def _basic_init(self):
self._page = rget(self._url)
if self._page.status_code == 404:
raise ParserFailure(
"Unrecognized fund, please check fund code you input.")
if self._page.text[:800].find("Data_millionCopiesIncome") >= 0:
raise FundTypeError(
"This code seems to be a mfund, use mfundinfo instead")
l = re.match(r"[\s\S]*Data_netWorthTrend = ([^;]*);[\s\S]*",
self._page.text).groups()[0]
l = l.replace("null", "None") # 暂未发现基金净值有 null 的基金，若有，其他地方也很可能出问题！
l = eval(l)
ltot = re.match(r"[\s\S]*Data_ACWorthTrend = ([^;]*);[\s\S]*",
self._page.text).groups()[0] # .* doesn't match \n
ltot = ltot.replace("null", "None") ## 096001 总值数据中有 null！
ltot = eval(ltot)
## timestamp transform tzinfo must be taken into consideration
tz_bj = dt.timezone(dt.timedelta(hours=8))
infodict = {
"date": [
dt.datetime.fromtimestamp(int(d["x"]) / 1e3,
tz=tz_bj).replace(tzinfo=None)
for d in l
],
"netvalue": [float(d["y"]) for d in l],
"comment": [_nfloat(d["unitMoney"]) for d in l],
}
if len(l) == len(ltot): # 防止总值和净值数据量不匹配，已知有该问题的基金：502010
infodict["totvalue"] = [d[1] for d in ltot]
try:
rate = float(
eval(
re.match(r"[\s\S]*fund_Rate=([^;]*);[\s\S]*",
self._page.text).groups()[0]))
except ValueError:
rate = 0
logger.info(
"warning: this fund has no data for rate") # know cases: ETF
name = eval(
re.match(r"[\s\S]*fS_name = ([^;]*);[\s\S]*",
self._page.text).groups()[0])
self.rate = rate
# shengou rate in tiantianjijin, daeshengou rate discount is not considered
self.name = name # the name of the fund
df = pd.DataFrame(data=infodict)
df = df[df["date"].isin(opendate)]
df = df.reset_index(drop=True)
if len(df) == 0:
raise ParserFailure("no price table found for this fund %s" %
self.code)
self.price = df[df["date"] <= yesterdaydash()]
# deal with the redemption fee attrs finally
if not self.priceonly:
self._feepreprocess()
def _basic_init(self):
self._page = rget(self._url)
if self._page.text[:800].find("Data_fundSharesPositions") >= 0:
raise FundTypeError(
"This code seems to be a fund, use fundinfo instead")
l = eval(
re.match(r"[\s\S]*Data_millionCopiesIncome = ([^;]*);[\s\S]*",
self._page.text).groups()[0])
self.name = re.match(r"[\s\S]*fS_name = \"([^;]*)\";[\s\S]*",
self._page.text).groups()[0]
tz_bj = dt.timezone(dt.timedelta(hours=8))
datel = [
dt.datetime.fromtimestamp(int(d[0]) / 1e3,
tz=tz_bj).replace(tzinfo=None) for d in l
]
ratel = [float(d[1]) for d in l]
netvalue = [1]
for dailyrate in ratel:
netvalue.append(netvalue[-1] * (1 + dailyrate * 1e-4))
netvalue.remove(1)
df = pd.DataFrame(
data={
"date": datel,
"netvalue": netvalue,
"totvalue": netvalue,
"comment": [0 for _ in datel],
})
df = df[df["date"].isin(opendate)]
if len(df) == 0:
raise ParserFailure("no price table for %s" % self.code)
df = df.reset_index(drop=True)
self.price = df[df["date"] <= yesterdaydash()]
def vtradevolume(cftable, freq="D", rendered=True):
"""
aid function on visualization of trade summary
:param cftable: cftable (pandas.DataFrame) with at least date and cash columns
:param freq: one character string, frequency label, now supporting D for date,
W for week and M for month, namely the trade volume is shown based on the time unit
:returns: the Bar object
"""
### WARN: datazoom and time conflict, sliding till 1970..., need further look into pyeacharts
### very unsatisfied about current visualize effect, and it seems the order of add and set option matters a lot
if freq == "D":
datedata = [d.to_pydatetime() for d in cftable["date"]]
selldata = [[row["date"].to_pydatetime(), row["cash"]]
for _, row in cftable.iterrows() if row["cash"] > 0]
buydata = [[row["date"].to_pydatetime(), row["cash"]]
for _, row in cftable.iterrows() if row["cash"] < 0]
elif freq == "W":
cfmerge = cftable.groupby(
[cftable["date"].dt.year, cftable["date"].dt.week])["cash"].sum()
datedata = [
dt.datetime.strptime(str(a) + "4", "(%Y, %W)%w")
for a, _ in cfmerge.iteritems()
]
selldata = [[dt.datetime.strptime(str(a) + "4", "(%Y, %W)%w"), b]
for a, b in cfmerge.iteritems() if b > 0]
buydata = [[dt.datetime.strptime(str(a) + "4", "(%Y, %W)%w"), b]
for a, b in cfmerge.iteritems() if b < 0]
elif freq == "M":
cfmerge = cftable.groupby(
[cftable["date"].dt.year, cftable["date"].dt.month])["cash"].sum()
datedata = [
dt.datetime.strptime(str(a) + "15", "(%Y, %m)%d")
for a, _ in cfmerge.iteritems()
]
selldata = [[dt.datetime.strptime(str(a) + "15", "(%Y, %m)%d"), b]
for a, b in cfmerge.iteritems() if b > 0]
buydata = [[dt.datetime.strptime(str(a) + "15", "(%Y, %m)%d"), b]
for a, b in cfmerge.iteritems() if b < 0]
else:
raise ParserFailure("no such freq tag supporting")
buydata = [[d, round(x, 1)] for d, x in buydata]
selldata = [[d, round(x, 1)] for d, x in selldata]
bar = Bar()
bar.add_xaxis(datedata)
# buydata should before selldata, since emptylist in the first line would make the output fig empty: may be bug in pyecharts
bar.add_yaxis(series_name="买入", yaxis_data=buydata, category_gap="90%")
bar.add_yaxis(series_name="卖出", yaxis_data=selldata, category_gap="90%")
bar.set_global_opts(
xaxis_opts=opts.AxisOpts(type_="time"),
datazoom_opts=[opts.DataZoomOpts(range_start=99, range_end=100)],
)
if rendered:
return bar.render_notebook()
else:
return bar
def set_display(env=""):
"""
开关 DataFrame 的显示模式，仅 Jupyter Notebook 有效。
:param env: str, default "". If env="notebook", pd.DataFrame will be shown in fantastic web language
:return:
"""
if not env:
try:
delattr(pd.DataFrame, "_repr_javascript_")
except AttributeError:
pass
elif env in ["notebook", "jupyter", "ipython"]:
_set_display_notebook()
else:
raise ParserFailure("unknown env %s" % env)
def get_sh_status(category="cb", date=None):
url = "http://query.sse.com.cn/commonQuery.do?jsonCallBack=&"
if category in ["cb", "kzz"]:
url += "isPagination=false&sqlId=COMMON_BOND_KZZFLZ_ALL&KZZ=1"
elif category in ["fund", "fs"]:
if not date:
date = today_obj().strftime("%Y%m%d")
date = date.replace("/", "").replace("-", "")
url += "&sqlId=COMMON_SSE_FUND_LOF_SCALE_CX_S&pageHelp.pageSize=10000&FILEDATE={date}".format(
date=date)
else:
raise ParserFailure("unrecoginzed category %s" % category)
r = rget_json(
url,
headers={
"user-agent": "Mozilla/5.0",
"Host": "query.sse.com.cn",
"Referer":
"http://www.sse.com.cn/market/bonddata/data/convertible/",
},
)
return pd.DataFrame(r["result"])
def v(self, y="lne"):
"""
总资产或总利润与拟合曲线的可视化
:param y: str. one of lne, lnb, e, b, roe
:return:
"""
df = self.df
if y == "roe":
return df.plot(x="date", y="roe")
fitx = np.arange(0, df.iloc[-1]["date_count"], 10)
if y == "lne":
fity = self.intercept_e + self.slope_e * fitx
elif y == "lnb":
fity = self.intercept_b + self.slope_b * fitx
elif y == "e":
fity = np.exp(self.intercept_e + self.slope_e * fitx)
elif y == "b":
fity = np.exp(self.intercept_b + self.slope_b * fitx)
else:
raise ParserFailure("Unrecogized y %s" % y)
ax = df.plot(x="date_count", y=y)
ax.plot(fitx, fity)
return ax
def _addrow(self):
"""
Return cashflow table with one more line or raise an exception if there is no more line to add
The same logic also applies to rem table
关于对于一个基金多个操作存在于同一交易日的说明：无法处理历史买入第一笔同时是分红日的情形, 事实上也不存在这种情形。无法处理一日多笔买卖的情形。
同一日既有卖也有买不现实，多笔买入只能在 csv 上合并记录，由此可能引起份额计算 0.01 的误差。可以处理分红日买入卖出的情形。
分级份额折算日封闭无法买入，所以程序直接忽略当天的买卖。因此不会出现多个操作共存的情形。
"""
# the design on data remtable is disaster, it is very dangerous though works now
code = self.aim.code
if len(self.cftable) == 0:
if len(self.status[self.status[code] != 0]) == 0:
raise Exception("no other info to be add into cashflow table")
i = 0
while self.status.iloc[i].loc[code] == 0:
i += 1
value = self.status.iloc[i].loc[code]
date = self.status.iloc[i].date
if value > 0:
rdate, cash, share = self.aim.shengou(value, date)
rem = rm.buy([], share, rdate)
else:
raise TradeBehaviorError("You cannot sell first when you never buy")
elif len(self.cftable) > 0:
recorddate = list(self.status.date)
lastdate = self.cftable.iloc[-1].date + pd.Timedelta(1, unit="d")
while (lastdate not in self.aim.specialdate) and (
(lastdate not in recorddate)
or (
(lastdate in recorddate)
and (
self.status[self.status["date"] == lastdate].loc[:, code].any()
== 0
)
)
):
lastdate += pd.Timedelta(1, unit="d")
if (lastdate - yesterdayobj()).days >= 1:
raise Exception("no other info to be add into cashflow table")
date = lastdate
label = self.aim.dividend_label # 现金分红 0, 红利再投 1
cash = 0
share = 0
rem = self.remtable.iloc[-1].rem
rdate = date
if (date in recorddate) and (date not in self.aim.zhesuandate):
# deal with buy and sell and label the fenhongzaitouru, namely one label a 0.05 in the original table to label fenhongzaitouru
value = self.status[self.status["date"] == date].iloc[0].loc[code]
fenhongmark = round(10 * value - int(10 * value), 1)
if fenhongmark == 0.5 and label == 0:
label = 1 # fenhong reinvest
value = round(value, 1)
elif fenhongmark == 0.5 and label == 1:
label = 0
value = round(value, 1)
if value > 0: # value stands for purchase money
rdate, dcash, dshare = self.aim.shengou(value, date)
rem = rm.buy(rem, dshare, rdate)
elif value < -0.005: # value stands for redemp share
rdate, dcash, dshare = self.aim.shuhui(
-value, date, self.remtable.iloc[-1].rem
)
_, rem = rm.sell(rem, -dshare, rdate)
elif value >= -0.005 and value < 0:
# value now stands for the ratio to be sold in terms of remain positions, -0.005 stand for sell 100%
remainshare = sum(self.cftable.loc[:, "share"])
ratio = -value / 0.005
rdate, dcash, dshare = self.aim.shuhui(
remainshare * ratio, date, self.remtable.iloc[-1].rem
)
_, rem = rm.sell(rem, -dshare, rdate)
else: # in case value=0, when specialday is in record day
rdate, dcash, dshare = date, 0, 0
cash += dcash
share += dshare
if date in self.aim.specialdate: # deal with fenhong and xiazhe
comment = (
self.aim.price[self.aim.price["date"] == date]
.iloc[0]
.loc["comment"]
)
if isinstance(comment, float):
if comment < 0:
dcash2, dshare2 = (
0,
sum([myround(sh * (-comment - 1)) for _, sh in rem]),
) # xiazhe are seperately carried out based on different purchase date
rem = rm.trans(rem, -comment, date)
# myround(sum(cftable.loc[:,'share'])*(-comment-1))
elif comment > 0 and label == 0:
dcash2, dshare2 = (
myround(sum(self.cftable.loc[:, "share"]) * comment),
0,
)
rem = rm.copy(rem)
elif comment > 0 and label == 1:
dcash2, dshare2 = (
0,
myround(
sum(self.cftable.loc[:, "share"])
* (
comment
/ self.aim.price[self.aim.price["date"] == date]
.iloc[0]
.netvalue
)
),
)
rem = rm.buy(rem, dshare2, date)
cash += dcash2
share += dshare2
else:
raise ParserFailure("comments not recoginized")
self.cftable = self.cftable.append(
pd.DataFrame([[rdate, cash, share]], columns=["date", "cash", "share"]),
ignore_index=True,
)
self.remtable = self.remtable.append(
pd.DataFrame([[rdate, rem]], columns=["date", "rem"]), ignore_index=True
)
def _addrow(self):
"""
Return cashflow table with one more line or raise an exception if there is no more line to add
The same logic also applies to rem table
关于对于一个基金多个操作存在于同一交易日的说明：无法处理历史买入第一笔同时是分红日的情形, 事实上也不存在这种情形。无法处理一日多笔买卖的情形。
同一日既有卖也有买不现实，多笔买入只能在 csv 上合并记录，由此可能引起份额计算 0.01 的误差。可以处理分红日买入卖出的情形。
分级份额折算日封闭无法买入，所以程序直接忽略当天的买卖。因此不会出现多个操作共存的情形。
"""
# the design on data remtable is disaster, it is very dangerous though works now
# possibly failing cases include:
# 买卖日记录是节假日，而顺延的日期恰好是折算日（理论上无法申赎）或分红日（可能由于 date 和 rdate 的错位而没有考虑到），
# 又比如周日申购记录，周一申购记录，那么周日记录会现金流记在周一，继续现金流标更新将从周二开始，周一数据被丢弃
code = self.aim.code
if len(self.cftable) == 0:
if len(self.status[self.status[code] != 0]) == 0:
raise Exception("no other info to be add into cashflow table")
i = 0
while self.status.iloc[i].loc[code] == 0:
i += 1
value = self.status.iloc[i].loc[code]
date = self.status.iloc[i].date
self.lastdate = date
if len(self.price[self.price["date"] >= date]) > 0:
date = self.price[self.price["date"] >= date].iloc[0]["date"]
else:
date = self.price[self.price["date"] <= date].iloc[-1]["date"]
# 这里没有像下边部分一样仔细处理单独的 lastdate，hopefully 不会出现其他奇怪的问题，有 case 再说
# https://github.com/refraction-ray/xalpha/issues/47
# 凭直觉这个地方的处理很可能还有其他 issue
if value > 0:
feelabel = 100 * value - int(100 * value)
if round(feelabel, 1) == 0.5:
# binary encoding, 10000.005 is actually 10000.0050...1, see issue #59
feelabel = feelabel - 0.5
if abs(feelabel) < 1e-4:
feelabel = 0
else:
feelabel *= 100
else:
feelabel = None
value = int(value * 100) / 100
assert feelabel is None or feelabel >= 0.0, "自定义申购费必须为正值"
rdate, cash, share = self.aim.shengou(value,
date,
fee=feelabel)
rem = rm.buy([], share, rdate)
else:
raise TradeBehaviorError(
"You cannot sell first when you never buy")
elif len(self.cftable) > 0:
# recorddate = list(self.status.date)
if not getattr(self, "lastdate", None):
lastdate = self.cftable.iloc[-1].date + pd.Timedelta(1,
unit="d")
else:
lastdate = self.lastdate + pd.Timedelta(1, unit="d")
while (lastdate not in self.aim.specialdate) and (
(lastdate not in self.recorddate_set) or
((lastdate in self.recorddate_set) and
(self.status[self.status["date"] ==
lastdate].loc[:, code].any() == 0))):
lastdate += pd.Timedelta(1, unit="d")
if (lastdate - yesterdayobj()).days >= 1:
raise Exception(
"no other info to be add into cashflow table")
if (lastdate - yesterdayobj()).days >= 1:
raise Exception("no other info to be add into cashflow table")
date = lastdate
# 无净值日优先后移，无法后移则前移
# 还是建议日期记录准确，不然可能有无法完美兼容的错误出现
if len(self.price[self.price["date"] >= date]) > 0:
date = self.price[self.price["date"] >= date].iloc[0]["date"]
else:
date = self.price[self.price["date"] <= date].iloc[-1]["date"]
if date != lastdate and date in list(self.status.date):
# 日期平移到了其他记录日，很可能出现问题!
logger.warning(
"账单日期 %s 非 %s 的净值记录日期，日期智能平移后 %s 与账单其他日期重合！交易处理极可能出现问题！！ "
"靠后日期的记录被覆盖" % (lastdate, self.code, date))
self.lastdate = lastdate
if date > lastdate:
self.lastdate = date
# see https://github.com/refraction-ray/xalpha/issues/27, begin new date from last one in df is not reliable
label = self.aim.dividend_label # 现金分红 0, 红利再投 1
cash = 0
share = 0
rem = self.remtable.iloc[-1].rem
rdate = date
if (lastdate
in self.recorddate_set) and (date
not in self.aim.zhesuandate):
# deal with buy and sell and label the fenhongzaitouru, namely one label a 0.05 in the original table to label fenhongzaitouru
value = self.status[
self.status["date"] <= lastdate].iloc[-1].loc[code]
if date in self.aim.fenhongdate: # 0.05 的分红行为标记，只有分红日才有效
fenhongmark = round(10 * value - int(10 * value), 1)
if fenhongmark == 0.5 and label == 0:
label = 1 # fenhong reinvest
value = value - math.copysign(0.05, value)
elif fenhongmark == 0.5 and label == 1:
label = 0
value = value - math.copysign(0.05, value)
if value > 0: # value stands for purchase money
feelabel = 100 * value - int(100 * value)
if int(10 * feelabel) == 5:
feelabel = (feelabel - 0.5) * 100
else:
feelabel = None
value = int(value * 100) / 100
rdate, dcash, dshare = self.aim.shengou(
value, date, fee=feelabel
) # shengou fee is in the unit of percent, different than shuhui case
rem = rm.buy(rem, dshare, rdate)
elif value < -0.005: # value stands for redemp share
feelabel = int(100 * value) - 100 * value
if int(10 * feelabel) == 5:
feelabel = feelabel - 0.5
else:
feelabel = None
value = int(value * 100) / 100
rdate, dcash, dshare = self.aim.shuhui(
-value, date, self.remtable.iloc[-1].rem, fee=feelabel)
_, rem = rm.sell(rem, -dshare, rdate)
elif value >= -0.005 and value < 0:
# value now stands for the ratio to be sold in terms of remain positions, -0.005 stand for sell 100%
remainshare = sum(self.cftable[
self.cftable["date"] <= date].loc[:, "share"])
ratio = -value / 0.005
rdate, dcash, dshare = self.aim.shuhui(
remainshare * ratio, date, self.remtable.iloc[-1].rem,
0)
_, rem = rm.sell(rem, -dshare, rdate)
else: # in case value=0, when specialday is in record day
rdate, dcash, dshare = date, 0, 0
cash += dcash
share += dshare
if date in self.aim.specialdate: # deal with fenhong and xiazhe
comment = self.price[self.price["date"] ==
date].iloc[0].loc["comment"]
if isinstance(comment, float):
if comment < 0:
dcash2, dshare2 = (
0,
sum([
myround(sh * (-comment - 1)) for _, sh in rem
]),
) # xiazhe are seperately carried out based on different purchase date
rem = rm.trans(rem, -comment, date)
# myround(sum(cftable.loc[:,'share'])*(-comment-1))
elif comment > 0 and label == 0:
dcash2, dshare2 = (
myround(
sum(self.cftable.loc[:, "share"]) * comment),
0,
)
rem = rm.copy(rem)
elif comment > 0 and label == 1:
dcash2, dshare2 = (
0,
myround(
sum(self.cftable.loc[:, "share"]) *
(comment / self.price[self.price["date"] ==
date].iloc[0].netvalue)),
)
rem = rm.buy(rem, dshare2, date)
cash += dcash2
share += dshare2
else:
raise ParserFailure("comments not recognized")
self.cftable = self.cftable.append(
pd.DataFrame([[rdate, cash, share]],
columns=["date", "cash", "share"]),
ignore_index=True,
)
self.remtable = self.remtable.append(pd.DataFrame(
[[rdate, rem]], columns=["date", "rem"]),
ignore_index=True)
def vtradevolume(cftable, freq="D", rendered=True):
"""
aid function on visualization of trade summary
:param cftable: cftable (pandas.DataFrame) with at least date and cash columns
:param freq: one character string, frequency label, now supporting D for date,
W for week and M for month, namely the trade volume is shown based on the time unit
:returns: the Bar object
"""
### WARN: datazoom and time conflict, sliding till 1970..., need further look into pyeacharts
startdate = cftable.iloc[0]["date"]
if freq == "D":
# datedata = [d.to_pydatetime() for d in cftable["date"]]
datedata = pd.date_range(startdate, yesterdayobj(), freq="D")
selldata = [[row["date"].to_pydatetime(), row["cash"]]
for _, row in cftable.iterrows() if row["cash"] > 0]
buydata = [[row["date"].to_pydatetime(), row["cash"]]
for _, row in cftable.iterrows() if row["cash"] < 0]
elif freq == "W":
cfmerge = cftable.groupby(
[cftable["date"].dt.year, cftable["date"].dt.week])["cash"].sum()
# datedata = [
# dt.datetime.strptime(str(a) + "4", "(%Y, %W)%w")
# for a, _ in cfmerge.iteritems()
# ]
datedata = pd.date_range(startdate,
yesterdayobj() + pd.Timedelta(days=7),
freq="W-THU")
selldata = [[dt.datetime.strptime(str(a) + "4", "(%G, %V)%w"), b]
for a, b in cfmerge.iteritems() if b > 0]
buydata = [[dt.datetime.strptime(str(a) + "4", "(%G, %V)%w"), b]
for a, b in cfmerge.iteritems() if b < 0]
# %V pandas gives iso weeknumber which is different from python original %W or %U,
# see https://stackoverflow.com/questions/5882405/get-date-from-iso-week-number-in-python for more details
# python3.6+ required for %G and %V
# but now seems no equal distance between sell and buy data, no idea why
elif freq == "M":
cfmerge = cftable.groupby(
[cftable["date"].dt.year, cftable["date"].dt.month])["cash"].sum()
# datedata = [
# dt.datetime.strptime(str(a) + "15", "(%Y, %m)%d")
# for a, _ in cfmerge.iteritems()
# ]
datedata = pd.date_range(startdate,
yesterdayobj() + pd.Timedelta(days=31),
freq="MS")
selldata = [[dt.datetime.strptime(str(a) + "1", "(%Y, %m)%d"), b]
for a, b in cfmerge.iteritems() if b > 0]
buydata = [[dt.datetime.strptime(str(a) + "1", "(%Y, %m)%d"), b]
for a, b in cfmerge.iteritems() if b < 0]
else:
raise ParserFailure("no such freq tag supporting")
buydata = [[d, round(x, 1)] for d, x in buydata]
selldata = [[d, round(x, 1)] for d, x in selldata]
bar = Bar()
datedata = list(datedata)
bar.add_xaxis(xaxis_data=datedata)
# buydata should before selldata, since emptylist in the first line would make the output fig empty: may be bug in pyecharts
bar.add_yaxis(series_name="买入", yaxis_data=buydata)
bar.add_yaxis(series_name="卖出", yaxis_data=selldata)
bar.set_global_opts(
tooltip_opts=opts.TooltipOpts(
is_show=True,
trigger="axis",
trigger_on="mousemove",
axis_pointer_type="cross",
),
datazoom_opts=[opts.DataZoomOpts(range_start=90, range_end=100)],
)
if rendered:
return bar.render_notebook()
else:
return bar
def get_fund_holdings(code, year="", season="", month="", category="jjcc"):
"""
获取基金详细的底层持仓信息
:param code: str. 6 位基金代码
:param year: int. eg. 2019
:param season: int, 1,2,3,4
:param month: Optional[int]. 指定 season 即可，一般不需理会
:param category: str. stock 股票持仓， bond 债券持仓，天天基金无法自动处理海外基金持仓，暂未兼容 FOF 的国内基金持仓
:return: pd.DataFrame or None. 没有对应持仓时返回 None。
"""
if not month and season:
month = 3 * int(season)
if category in ["stock", "stocks", "jjcc", "", "gp", "s"]:
category = "jjcc"
elif category in ["bond", "bonds", "zq", "zqcc", "b"]:
category = "zqcc"
else:
raise ParserFailure("unrecognized category %s" % category)
if code.startswith("F"):
code = code[1:]
r = rget(
"http://fundf10.eastmoney.com/FundArchivesDatas.aspx?type={category}&code={code}&topline=10&\
year={year}&month={month}".format(year=str(year),
month=str(month),
code=code,
category=category),
headers={
"Host":
"fundf10.eastmoney.com",
"Referer":
"http://fundf10.eastmoney.com/ccmx_{code}.html".format(code=code),
},
)
if len(r.text) < 50:
return
# raise ParserFailure(
# "This fund has no holdings on stock or bonds in this period"
# )
s = BeautifulSoup(
re.match("[\s\S]*apidata={ content:(.*),arryear:", r.text).groups()[0],
"lxml")
if len(s.text) < 30:
return
# raise ParserFailure(
# "This fund has no holdings on stock or bonds in this period"
# )
timeline = [
i.string for i in s.findAll("font", class_="px12")
if i.text.startswith("2")
]
ind = 0
if month:
for i, d in enumerate(timeline):
if d.split("-")[1][-1] == str(month)[-1]: # avoid 09 compare to 9
ind = i
break
else:
return # not update to this month
t1 = s.findAll("table")[ind]
main = [[j.text for j in i.contents] for i in t1.findAll("tr")[1:]]
cols = [j.text for j in t1.findAll("tr")[0].contents if j.text.strip()]
icode = 1
iname = 2
iratio = 4
ishare = 5
ivalue = 6
for j, col in enumerate(cols):
if col.endswith("代码"):
icode = j
elif col.endswith("名称"):
iname = j
elif col.endswith("比例"):
iratio = j
elif col.startswith("持股数"):
ishare = j
elif col.startswith("持仓市值"):
ivalue = j
if category == "jjcc":
result = {
"code": [],
"name": [],
"ratio": [],
"share": [],
"value": []
}
for l in main:
result["code"].append(l[icode])
result["name"].append(l[iname])
result["ratio"].append(float(l[iratio][:-1]))
result["share"].append(_float(l[ishare]))
result["value"].append(_float(l[ivalue]))
elif category == "zqcc":
result = {"code": [], "name": [], "ratio": [], "value": []}
for l in main:
result["code"].append(l[1])
result["name"].append(l[2])
result["ratio"].append(float(l[3][:-1]))
result["value"].append(_float(l[4]))
return pd.DataFrame(result)