无法使用numpy loadtxt用python3.6转换日期 [英] fail to use numpy loadtxt to convert date with python3.6

问题描述

我的代码如下:

import numpy as np
import matplotlib.dates as mdates

stockFile = 'HistoricalQuotes.csv'

dates, close = np.loadtxt(stockFile, delimiter=',', unpack=True, usecols=(0,1), \
                          converters = {0: mdates.strpdate2num('%Y-%m-%d')})
print("Dates = ", dates)

和回溯:

Traceback (most recent call last):


File "/Users/zdlzdlxs/Documents/workspace/pynum_practice/part3/simplestats.py", line 7, in <module>
    converters = {0: mdates.strpdate2num('%Y-%m-%d')})
  File "/Users/zdlzdlxs/Library/Python/3.6/lib/python/site-packages/numpy/lib/npyio.py", line 1024, in loadtxt
    items = [conv(val) for (conv, val) in zip(converters, vals)]
  File "/Users/zdlzdlxs/Library/Python/3.6/lib/python/site-packages/numpy/lib/npyio.py", line 1024, in <listcomp>
    items = [conv(val) for (conv, val) in zip(converters, vals)]
  File "/Users/zdlzdlxs/Library/Python/3.6/lib/python/site-packages/matplotlib/dates.py", line 287, in __call__
    return date2num(datetime.datetime(*time.strptime(s, self.fmt)[:6]))
  File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/_strptime.py", line 559, in _strptime_time
    tt = _strptime(data_string, format)[0]
  File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/_strptime.py", line 329, in _strptime
    raise TypeError(msg.format(index, type(arg)))
TypeError: strptime() argument 0 must be str, not <class 'bytes'>

我试图在stackoverflow中找到解决方案，但失败了. 有人可以帮我正确转换日期时间吗?非常感谢.

I've tried to find solutions in stackoverflow but failed. Could anyone help me to convert datetime correctly? Thanks a lot.

推荐答案

此日期转换器需要一个常规字符串，在Py3中为unicode.但是loadtxt将文本加载为字节串:

This date converter expects a regular string, which in Py3 is unicode. But loadtxt loads the text as bytestrings:

In [12]: mdates.strpdate2num('%Y-%m-%d')(b'1999-01-31')
---------------------------------------------------------------------------
....
TypeError: strptime() argument 0 must be str, not <class 'bytes'>

将转换器修改为decode字节串应解决以下问题:

Modifying the converter to decode the bytestring should take care of the issue:

In [13]: fn = lambda astr: mdates.strpdate2num('%Y-%m-%d')(astr.decode())

In [14]: fn(b'1999-01-01')
Out[14]: 729755.0

在loadtxt中测试:

testing in loadtxt:

In [15]: txt=b"""1999-01-01
    ...: 2000-03-03
    ...: 2015-12-12
    ...: """

In [16]: np.loadtxt(txt.splitlines(), converters={0:fn})
Out[16]: array([ 729755.,  730182.,  735944.])

这篇关于无法使用numpy loadtxt用python3.6转换日期的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！