Python 分类日期
我有一个Excel电子表格,我正准备迁移到Access,日期列有多种格式的条目,如:1963年至1969年、1968年8月至1968年9月、1972年、1973年3月、7月24日、1980年10月2日、1980年8月29日、1946年7月等,以及“未注明日期”。我将作为键(地图编号)和日期列的列拉入csv并写回csv。 我可以去掉4位数的年份,但不能去掉范围。我被难倒了,如何用手工提取缺少重新格式化的天数和2位数的年份。我的代码不是很优雅,可能不是最佳实践:Python 分类日期,python,excel,csv,Python,Excel,Csv,我有一个Excel电子表格,我正准备迁移到Access,日期列有多种格式的条目,如:1963年至1969年、1968年8月至1968年9月、1972年、1973年3月、7月24日、1980年10月2日、1980年8月29日、1946年7月等,以及“未注明日期”。我将作为键(地图编号)和日期列的列拉入csv并写回csv。 我可以去掉4位数的年份,但不能去掉范围。我被难倒了,如何用手工提取缺少重新格式化的天数和2位数的年份。我的代码不是很优雅,可能不是最佳实践: import csv, xlwt,
import csv, xlwt, re
# create new Excel document and add sheet
# from tempfile import TemporaryFile
from xlwt import Workbook
book = Workbook()
sheet1 = book.add_sheet('Sheet 1')
# populate first row with header
sheet1.write(0,0,"Year")
sheet1.write(0,1,"Map")
sheet1.write(0,2,"As Entered")
# count variable for populating sheet
rowCount=0
# open csv file and read
with open('C:\dateTestMSDOs.csv', 'rb') as f:
reader=csv.reader(f)
for row in reader:
map = row[0] # first row is map number
dateRaw = row[1] # second row is raw date as entered
# write undated and blank entries
if dateRaw == 'undated':
yearStr = '0000'
rowCount +=1
sheet1.write(rowCount, 0, yearStr)
sheet1.write(rowCount, 1, map)
sheet1.write(rowCount, 2, dateRaw)
#print rowCount, yearStr, map, dateRaw, '\n'
yearStr=''
if dateRaw == '':
yearStr = 'NoEntry'
rowCount +=1
sheet1.write(rowCount, 0, yearStr)
sheet1.write(rowCount, 1, map)
sheet1.write(rowCount, 2, dateRaw)
#print rowCount, yearStr, map, dateRaw, '\n'
yearStr=''
# search and write instances of four consecutive digits
try:
year = re.search(r'\d\d\d\d', dateRaw)
yearStr= year.group()
#print yearStr, map, dateRaw
rowCount +=1
sheet1.write(rowCount, 0, yearStr)
sheet1.write(rowCount, 1, map)
sheet1.write(rowCount, 2, dateRaw)
#print rowCount, yearStr, map, dateRaw, '\n'
yearStr=''
# if none exist flag for cleaning spreadsheet and print
except:
#print 'Nope', map, dateRaw
rowCount +=1
yearStr='Format'
sheet1.write(rowCount, 0, yearStr)
sheet1.write(rowCount, 1, map)
sheet1.write(rowCount, 2, dateRaw)
#print rowCount, yearStr, map, dateRaw, '\n'
yearStr=''
yearStr=''
dateRaw=''
book.save('D:\dateProperty.xls')
print "Done!"
我想在一个附加列中写入日和月,并提取范围条目的第二个4位日期。您可以尝试使用
dateutil
。我认为您仍然需要以不同的方式处理一些更难的格式。请参见下面的示例实现:
代码:
import dateutil.parser as dateparser
date_list = ['1963 to 1969',
'Aug. 1968 to Sept. 1968',
'Mar-73',
'24-Jul',
'Oct. 2 1980',
'Aug 29, 1980',
'July 1946',
'undated']
for d in date_list:
if 'to' in d:
a, b = d.split('to')
# Get the higher number. Use min to get lower of two.
print max(dateparser.parse(a.strip()).year, dateparser.parse(b.strip()).year)
elif d == 'undated':
print '0000'
else:
yr = dateparser.parse(d).year
print yr
1969
1968
1973
2014
1980
1980
1946
0000
[Finished in 0.4s]
结果:
import dateutil.parser as dateparser
date_list = ['1963 to 1969',
'Aug. 1968 to Sept. 1968',
'Mar-73',
'24-Jul',
'Oct. 2 1980',
'Aug 29, 1980',
'July 1946',
'undated']
for d in date_list:
if 'to' in d:
a, b = d.split('to')
# Get the higher number. Use min to get lower of two.
print max(dateparser.parse(a.strip()).year, dateparser.parse(b.strip()).year)
elif d == 'undated':
print '0000'
else:
yr = dateparser.parse(d).year
print yr
1969
1968
1973
2014
1980
1980
1946
0000
[Finished in 0.4s]
我能看到的唯一突出的问题是
7月24日
返回日期2014
,因为解析器假定当前日期、月份或年份代替缺少的组件,即73年3月
将变为1973-03-20
,如果今天是该月20日,等等。不完全确定这是否是您想要的,但我只是使用了一个“简单”的正则表达式搜索,然后应用定义的给定函数遍历匹配的组集。如果找到匹配项,则调用的函数(在regex\u groups变量中找到)应返回具有以下键的字典:start\u day、start\u month、start\u year、end\u day、end\u month、end\u year
然后你可以用这些值做任何你想做的事情。绝对不是最干净的解决方案,但据我所知,它是有效的
#!/usr/local/bin/python2.7
import re
# Crazy regex
regex_pattern = '(?:(\d{4}) to (\d{4}))|(?:(\w+)\. (\d{4}) to (\w+)\. (\d{4}))|(?:(\w+)-(\d{2}))|(?:(\d{2})-(\w+))|(?:(\w+)\. (\d+), (\d{4}))|(?:(\w+) (\d+), (\d{4}))|(?:(\w+) (\d{4}))|(?:(\d{4}))'
date_strings = [
'1963 to 1969',
'Aug. 1968 to Sept. 1968',
'1972',
'Mar-73',
'24-Jul',
'Oct. 2, 1980',
'Aug 29, 1980',
'July 1946',
]
# Here you set the group matching functions that will be called for a matching group
regex_groups = {
(1,2): lambda group_matches: {
'start_day': '', 'start_month': '', 'start_year': group_matches[0],
'end_day': '', 'end_month': '', 'end_year': group_matches[1]
},
(3,4,5,6): lambda group_matches: {
'start_day': '', 'start_month': group_matches[0], 'start_year': group_matches[1],
'end_day': '', 'end_month': group_matches[2], 'end_year': group_matches[3]
},
(7,8): lambda group_matches: {
'start_day': '', 'start_month': group_matches[0], 'start_year': group_matches[1],
'end_day': '', 'end_month': '', 'end_year': ''
},
(9,10): lambda group_matches: {
'start_day': group_matches[1], 'start_month': '', 'start_year': group_matches[0],
'end_day': '', 'end_month': '', 'end_year': ''
},
(11,12,13): lambda group_matches: {
'start_day': group_matches[1], 'start_month': group_matches[0], 'start_year': group_matches[2],
'end_day': '', 'end_month': '', 'end_year': ''
},
(14,15,16): lambda group_matches: {
'start_day': group_matches[1], 'start_month': group_matches[0], 'start_year': group_matches[2],
'end_day': '', 'end_month': '', 'end_year': ''
},
(17,18): lambda group_matches: {
'start_day': '', 'start_month': group_matches[0], 'start_year': group_matches[1],
'end_day': '', 'end_month': '', 'end_year': ''
},
(19,): lambda group_matches: {
'start_day': '', 'start_month': '', 'start_year': group_matches[0],
'end_day': '', 'end_month': '', 'end_year': ''
},
}
for ds in date_strings:
matches = re.search(regex_pattern, ds)
start_month = ''
start_year = ''
end_month = ''
end_year = ''
for regex_group, group_func in regex_groups.items():
group_matches = [matches.group(sub_group_num) for sub_group_num in regex_group]
if all(group_matches):
match_data = group_func(group_matches)
print
print 'Matched:', ds
print '%s to %s' % ('-'.join([match_data['start_day'], match_data['start_month'], match_data['start_year']]), '-'.join([match_data['end_day'], match_data['end_month'], match_data['end_year']]))
# match_data is a dictionary with keys:
# * start_day
# * start_month
# * start_year
# * end_day
# * end_month
# * end_year
# If a group doesn't contain one of those items, then it is set to a blank string
产出:
Matched: 1963 to 1969
--1963 to --1969
Matched: Aug. 1968 to Sept. 1968
-Aug-1968 to -Sept-1968
Matched: 1972
--1972 to --
Matched: Mar-73
-Mar-73 to --
Matched: 24-Jul
Jul--24 to --
Matched: Oct. 2, 1980
2-Oct-1980 to --
Matched: Aug 29, 1980
29-Aug-1980 to --
Matched: July 1946
-July-1946 to --
您可以使用正则表达式定义所有可能的日期情况,例如:
import re
s = ['1963 to 1969', 'Aug. 1968 to Sept. 1968',
'1972', 'Mar-73', '03-Jun', '24-Jul', 'Oct. 2, 1980', 'Oct. 26, 1980',
'Aug 29 1980', 'July 1946']
def get_year(date):
mm = re.findall("\d{4}", date)
if mm:
return mm
mm = re.search("\w+-(\d{2})", date)
if mm:
return [mm.group(1)]
def get_month(date):
mm = re.findall("[A-Z][a-z]+", date)
if mm:
return mm
def get_day(date):
d_expr = ["(\d|\d{2})\-[A-Z][a-z]+","[A-Z][a-z]+[\. ]+(\d|\d{2}),"]
for expr in d_expr:
mm = re.search(expr, date)
if mm:
return [mm.group(1)]
d = {}
m = {}
y = {}
for idx, date in enumerate(s):
d[idx] = get_day(date)
m[idx] = get_month(date)
y[idx] = get_year(date)
print "Year Dict: ", y
print "Month Dict: ", m
print "Day Dict: ", d
因此,您可以得到日、月和年的字典。它们可以用来填充行
输出:
Year Dict: {0: ['1963', '1969'], 1: ['1968', '1968'], 2: ['1972'], 3: ['73'], 4: None, 5: None, 6: ['1980'], 7: ['1980'], 8: ['1980'], 9: ['1946']}
Month Dict: {0: None, 1: ['Aug', 'Sept'], 2: None, 3: ['Mar'], 4: ['Jun'], 5: ['Jul'], 6: ['Oct'], 7: ['Oct'], 8: ['Aug'], 9: ['July']}
Day Dict: {0: None, 1: None, 2: None, 3: None, 4: ['03'], 5: ['24'], 6: ['2'], 7: ['26'], 8: None, 9: None}
谢谢你的创新建议。经过考虑,我们决定从数据库中可以搜索的数据中删除日期和月份,因为只有相对较少的数据具有这种详细程度。下面是我用来从一个长而混乱的列表中提取和生成所需数据的代码
import csv, xlwt, re
# create new Excel document and add sheet
from xlwt import Workbook
book = Workbook()
sheet1 = book.add_sheet('Sheet 1')
# populate first row with header
sheet1.write(0,0,"MapYear_(Parsed)")
sheet1.write(0,1,"Map_Number")
sheet1.write(0,2,"As_Entered")
# count variable for populating sheet
rowCount=0
# open csv file and read
yearStr = ''
with open('C:\mapsDateFix.csv', 'rb') as f:
reader=csv.reader(f)
for row in reader:
map = row[0] # first row is map number
dateRaw = row[1] # second row is raw date as entered
# write undated and blank entries
if dateRaw == 'undated':
yearStr = 'undated'
rowCount +=1
sheet1.write(rowCount, 0, yearStr)
sheet1.write(rowCount, 1, map)
sheet1.write(rowCount, 2, dateRaw)
#print rowCount, yearStr, map, dateRaw, '\n'
#yearStr=''
if yearStr != 'undated':
if dateRaw == '':
yearStr = 'NoEntry'
rowCount +=1
sheet1.write(rowCount, 0, yearStr)
sheet1.write(rowCount, 1, map)
sheet1.write(rowCount, 2, dateRaw)
#print rowCount, yearStr, map, dateRaw, '\n'
#yearStr=''
# search and write instances of four consecutive digits
if yearStr != dateRaw:
try:
year = re.search(r'\d\d\d\d', dateRaw)
yearStr= year.group()
#print yearStr, map, dateRaw
rowCount +=1
sheet1.write(rowCount, 0, yearStr)
sheet1.write(rowCount, 1, map)
sheet1.write(rowCount, 2, dateRaw)
#print rowCount, yearStr, map, dateRaw, '\n'
yearStr=''
# if none exist flag for cleaning spreadsheet and print
except:
#print 'Nope', map, dateRaw
rowCount +=1
yearStr='Format'
sheet1.write(rowCount, 0, yearStr)
sheet1.write(rowCount, 1, map)
sheet1.write(rowCount, 2, dateRaw)
#print rowCount, yearStr, map, dateRaw, '\n'
yearStr=''
yearStr=''
dateRaw=''
book.save('D:\dateProperty.xls')
print "Done!"
你有没有看过中的日期格式?具有强大的日期解析功能。我想看看。嘿,谢谢你的建议!我想我会发布我那笨拙的代码。输出.xls作为LU表导入到我们的数据库中,工作正常。缺点是任何空格或额外字符都会破坏它,但缺点是它强制执行原始数据输入符合数据库中单个可搜索年份类别的输出(这是我们在处理范围、天数、月份等问题时必须解决的问题)。谢谢你,Bryce。在时间不多,工作参数改变焦点之前,这正是我想去的地方。不过,我会保留这段代码。我有一堆讨厌的数据要在几个月内清理,而不是更有效的方法,这将非常适合。这绝对是我见过的最好的处理方法,而且比我的代码更好。我希望我能记住你的帖子,当时我们改变了关注点,不将日和月作为可搜索的参数。诚然,我会使用
pandas
,但如果没有访问你的数据,这是我能想到的最好的方法。