对日期序列进行排序的最 Pythonic 方式是什么？

Question

我有一个代表一年中一个月的字符串列表（未排序且不连续）： ['1/2013', '7/2013', '2/2013', '3/2013', '4/2014', '12/2013', '10/2013', '11/2013', '1/2014', '2/2014']

我正在寻找一种 Pythonic 方法来对所有这些进行排序并按照以下建议分隔每个连续序列：

[ ['1/2013', '2/2013', '3/2013', '4/2013'], 
  ['7/2013'], 
  ['10/2013', '11/2013', '12/2013', '1/2014', '2/2014'] 
]

有什么想法吗？

Answer 1

基于the example from the docs that shows how to find runs of consecutive numbers 使用itertools.groupby()：

from itertools import groupby
from pprint import pprint

def month_number(date):
    month, year = date.split('/')
    return int(year) * 12 + int(month)

L = [[date for _, date in run]
     for _, run in groupby(enumerate(sorted(months, key=month_number)),
                           key=lambda (i, date): (i - month_number(date)))]
pprint(L)

解决方案的关键是与enumerate() 生成的范围进行区分，以便连续的月份都出现在同一个组中（运行）。

输出

[['1/2013', '2/2013', '3/2013'],
 ['7/2013'],
 ['10/2013', '11/2013', '12/2013', '1/2014', '2/2014'],
 ['4/2014']]

Answer 2

groupby 示例很可爱，但过于密集，并且会在以下输入上中断：['1/2013', '2/2017']，即当存在不相邻年份的相邻月份时。

from datetime import datetime
from dateutil.relativedelta import relativedelta

def areAdjacent(old, new):
    return old + relativedelta(months=1) == new

def parseDate(s):
    return datetime.strptime(s, '%m/%Y')

def generateGroups(seq):
    group = []
    last = None
    for (current, formatted) in sorted((parseDate(s), s) for s in seq):
        if group and last is not None and not areAdjacent(last, current):
            yield group
            group = []
        group.append(formatted)
        last = current
    if group:
        yield group

结果：

[['1/2013', '2/2013', '3/2013'], 
 ['7/2013'],
 ['10/2013', '11/2013', '12/2013', '1/2014', '2/2014'],
 ['4/2014']]

Answer 3

如果您只想对列表进行排序，请使用 sorted 函数并传递 key 值 = 将日期字符串转换为 Python 的 datetime 对象的函数 lambda d: datetime.strptime(d, '%m/%Y')，请检查以下代码示例为您的列表作为 L：

>>> from datetime import datetime
>>> sorted(L, key = lambda d: datetime.strptime(d, '%m/%Y'))
['1/2013', '2/2013', '3/2013', '7/2013', '10/2013', 
 '11/2013', '12/2013', '1/2014', '2/2014', '4/2014'] # indented by hand

要将“月/年字符串列表”拆分为“连续月份列表列表”，您可以使用以下脚本（读取 cmets），其中，我首先对列表L 进行排序，然后在基础上对字符串进行分组连续月份（检查连续月份我写了一个函数）：

def is_cm(d1, d2):
    """ is consecutive month pair?
        : Assumption d1 is older day's date than d2
    """
    d1 = datetime.strptime(d1, '%m/%Y')
    d2 = datetime.strptime(d2, '%m/%Y') 

    y1, y2 = d1.year, d2.year
    m1, m2 = d1.month, d2.month

    if y1 == y2: # if years are same d2 should be in next month
        return (m2 - m1) == 1
    elif (y2 - y1) == 1: # if years are consecutive
        return (m1 == 12 and m2 == 1)

它的工作原理如下：

>>> is_cm('1/2012', '2/2012')
True # yes, consecutive
>>> is_cm('12/2012', '1/2013')
True # yes, consecutive
>>> is_cm('1/2015', '12/2012') # None --> # not consecutive
>>> is_cm('12/2012', '2/2013')
False # not consecutive

拆分代码的代码：

def result(dl):
    """
    dl: dates list - a iterator of 'month/year' strings
    type: list of strings

    returns: list of lists of strings
    """
    #Sort list:
    s_dl = sorted(dl, key=lambda d: datetime.strptime(d, '%m/%Y'))
    r_dl = [] # list to be return
    # split list into list of lists
    t_dl = [s_dl[0]] # temp list
    for d in s_dl[1:]:
        if not is_cm(t_dl[-1], d): # check if months are not consecutive
            r_dl.append(t_dl)
            t_dl = [d]
        else:
            t_dl.append(d)
    return r_dl

result(L)

不要忘记包含from datetime import datetime，这个技巧我相信你可以很容易地更新一个新的日期列表，其中日期是其他格式的。

在@9000 提示之后，如果您想检查旧脚本检查@codepad，我可以简化我的排序函数并删除旧答案。

Answer 4

在这种特定情况下（元素不多）的简单解决方案就是迭代所有月份：

year = dates[0].split('/')[1]
result = []
current = []
for i in range(1, 13):
    x = "%i/%s" % (i, year)
    if x in dates:
        current.append(x)
        if len(current) == 1:
            result.append(current)
    else:
        current = []

Answer 5

好吧，这是一个没有 itertools 的工具，而且我可以在不影响可读性的情况下尽可能短。诀窍在于使用zip。这基本上是@moe 的答案展开了一点。

def parseAsPair(piece):
  """Transforms things like '7/2014' into (2014, 7) """
  m, y = piece.split('/')
  return (int(y), int(m))

def goesAfter(earlier, later):
  """Returns True iff earlier goes right after later."""
  earlier_y, earlier_m = earlier
  later_y, later_m = later
  if earlier_y == later_y:  # same year?
    return later_m == earlier_m + 1 # next month
  else: # next year? must be Dec -> Jan
    return later_y == earlier_y + 1 and earlier_m == 12 and later_m == 1

def groupSequentially(months):
  result = []  # final result
  if months:
    sorted_months = sorted(months, key=parseAsPair)
    span = [sorted_months[0]]  # current span; has at least the first month
    for earlier, later in zip(sorted_months, sorted_months[1:]):
      if not goesAfter(parseAsPair(earlier), parseAsPair(later)):
        # current span is over
        result.append(span)
        span = []
      span.append(later)
    # last span was not appended because sequence ended without breaking
    result.append(span)
  return result

尝试一下：

months =['1/2013', '7/2013', '2/2013', '3/2013', '4/2014', '12/2013',
         '10/2013', '11/2013', '1/2014', '2/2014']

print groupSequentially(months)  # output wrapped manually

[['1/2013', '2/2013', '3/2013'], 
 ['7/2013'], 
 ['10/2013', '11/2013', '12/2013', '1/2014', '2/2014'], 
 ['4/2014']]

如果我们最后将parseAsPair 映射到列表上，我们可以节省一些性能和认知负担。然后每次调用parseAsPair 都可以从groupSequentially 中删除，但我们必须再次将结果转换为字符串。