Extract dates in different formats from string using regex in python - python

I need to extract dates from strings using regex in python and the dates can be in one of many formats, and between some random text.
The date formats are:
04/20/2009; 04/20/09; 4/20/09; 4/3/09
Mar-20-2009; Mar 20, 2009; March 20, 2009; Mar. 20, 2009; Mar 20 2009;
20 Mar 2009; 20 March 2009; 20 Mar. 2009; 20 March, 2009
Mar 20th, 2009; Mar 21st, 2009; Mar 22nd, 2009
Feb 2009; Sep 2009; Oct 2010
6/2008; 12/2009
2009; 2010
After extract the dates I need to sort them ascending.
I've tried to use those 6 regex patterns but it seems that it's not doing all the job.
pattern1 = r'((?:\d{1,2}[- ,./]*)(?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)[a-z]*[- ,./]*\d{4})'
pattern2 = r'((?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)[a-z]*[ ,./-]*\d{1,2}[ ,./-]*\d{4})'
pattern3 = r'((?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)[a-z]*[ ,./-]*\d{4})'
pattern4 = r'((?:\d{1,2}[/-]\d{1,2}[/-](?:\d{4}|\d{2})))'
pattern5 = r'(?:(\s\d{2}[/-](?:\d{4})))'
pattern6 = r'(?:\d{4})'

It might be useful to set up some intermediate variables.
import re
short_month_names = (
'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun',
'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'
)
long_month_names = (
'January', 'February', 'March', 'April', 'May', 'June', 'July',
'August', 'September', 'October', 'November', 'December'
)
short_month_cap = '(?:' + '|'.join(short_month_names) + ')'
long_month_cap = '(?:' + '|'.join(long_month_names) + ')'
short_num_month_cap = '(?:[1-9]|1[12])'
long_num_month_cap = '(?:0[1-9]|1[12])'
long_day_cap = '(?:0[1-9]|[12][0-9]|3[01])'
short_day_cap = '(?:[1-9]|[12][0-9]|3[01])'
long_year_cap = '(?:[0-9]{3}[1-9]|[0-9]{2}[1-9][0-9]|[0-9][1-9][0-9]{2}|[1-9][0-9]{3})'
short_year_cap = '(?:[0-9][0-9])'
ordinal_day = '(?:2?1st|2?2nd|2?3rd|[12]?[4-9]th|1[123]th|[123]0th|31st)'
formats = (
r'(?P<month_0>{lnm}|{snm})/(?P<day_0>{ld}|{sd})/(?P<year_0>{sy}|{ly})',
r'(?P<month_1>{sm})\-(?P<day_1>{ld}|{sd})\-(?P<year_1>{ly})',
r'(?P<month_2>{sm}|{lm})(?:\.\s+|\s*)(?P<day_2>{ld}|{sd})(?:,\s+|\s*)(?P<year_2>{ly})',
r'(?P<day_3>{ld}|{sd})(?:[\.,]\s+|\s*)(?P<month_3>{lm}|{sm})(?:[\.,]\s+|\s*)(?P<year_3>{ly})',
r'(?P<month_4>{lm}|{sm})\s+(?P<year_4>{ly})',
r'(?P<month_5>{lnm}|{snm})/(?P<year_5>{ly})',
r'(?P<year_6>{ly})',
r'(?P<month_6>{sm})\s+(?P<day_4>(?={od})[0-9][0-9]?)..,\s*(?P<year_7>{ly})'
)
_pattern = '|'.join(
i.format(
sm=short_month_cap, lm=long_month_cap, snm=short_num_month_cap,
lnm=long_num_month_cap, ld=long_day_cap, sd=short_day_cap,
ly=long_year_cap, sy=short_year_cap, od=ordinal_day
) for i in formats
)
pattern = re.compile(_pattern)
def get_fields(match):
if not match:
return None
return {
k[:-2]: v
for k, v in match.groupdict().items()
if v is not None
}
tests = r'''04/20/2009; 04/20/09; 4/20/09; 4/3/09
Mar-20-2009; Mar 20, 2009; March 20, 2009; Mar. 20, 2009; Mar 20 2009
20 Mar 2009; 20 March 2009; 20 Mar. 2009; 20 March, 2009
Mar 20th, 2009; Mar 21st, 2009; Mar 22nd, 2009
Feb 2009; Sep 2009; Oct 2010
6/2008; 12/2009
2009; 2010'''
for test_line in tests.split('\n'):
for test in test_line.split('; '):
print('{!r}: {!r}'.format(test, get_fields(pattern.fullmatch(test))))
print('')
Which outputs:
'04/20/2009': {'month': '04', 'day': '20', 'year': '2009'}
'04/20/09': {'month': '04', 'day': '20', 'year': '09'}
'4/20/09': {'month': '4', 'day': '20', 'year': '09'}
'4/3/09': {'month': '4', 'day': '3', 'year': '09'}
'Mar-20-2009': {'month': 'Mar', 'day': '20', 'year': '2009'}
'Mar 20, 2009': {'month': 'Mar', 'day': '20', 'year': '2009'}
'March 20, 2009': {'month': 'March', 'day': '20', 'year': '2009'}
'Mar. 20, 2009': {'month': 'Mar', 'day': '20', 'year': '2009'}
'Mar 20 2009': {'month': 'Mar', 'day': '20', 'year': '2009'}
'20 Mar 2009': {'day': '20', 'month': 'Mar', 'year': '2009'}
'20 March 2009': {'day': '20', 'month': 'March', 'year': '2009'}
'20 Mar. 2009': {'day': '20', 'month': 'Mar', 'year': '2009'}
'20 March, 2009': {'day': '20', 'month': 'March', 'year': '2009'}
'Mar 20th, 2009': {'month': 'Mar', 'day': '20', 'year': '2009'}
'Mar 21st, 2009': {'month': 'Mar', 'day': '21', 'year': '2009'}
'Mar 22nd, 2009': {'month': 'Mar', 'day': '22', 'year': '2009'}
'Feb 2009': {'month': 'Feb', 'year': '2009'}
'Sep 2009': {'month': 'Sep', 'year': '2009'}
'Oct 2010': {'month': 'Oct', 'year': '2010'}
'6/2008': {'month': '6', 'year': '2008'}
'12/2009': {'month': '12', 'year': '2009'}
'2009': {'year': '2009'}
'2010': {'year': '2010'}
The main part is the formats variable, where all the different formats are defined. It matches slightly more than what is defined, and can easily be extended.
The overall pattern ends up being:
'(?P<month_0>(?:0[1-9]|1[12])|(?:[1-9]|1[12]))/(?P<day_0>(?:0[1-9]|[12][0-9]|3[01])|(?:[1-9]|[12][0-9]|3[01]))/(?P<year_0>(?:[0-9][0-9])|(?:[0-9]{3}[1-9]|[0-9]{2}[1-9][0-9]|[0-9][1-9][0-9]{2}|[1-9][0-9]{3}))|(?P<month_1>(?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec))\\-(?P<day_1>(?:0[1-9]|[12][0-9]|3[01])|(?:[1-9]|[12][0-9]|3[01]))\\-(?P<year_1>(?:[0-9]{3}[1-9]|[0-9]{2}[1-9][0-9]|[0-9][1-9][0-9]{2}|[1-9][0-9]{3}))|(?P<month_2>(?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)|(?:January|February|March|April|May|June|July|August|September|October|November|December))(?:\\.\\s+|\\s*)(?P<day_2>(?:0[1-9]|[12][0-9]|3[01])|(?:[1-9]|[12][0-9]|3[01]))(?:,\\s+|\\s*)(?P<year_2>(?:[0-9]{3}[1-9]|[0-9]{2}[1-9][0-9]|[0-9][1-9][0-9]{2}|[1-9][0-9]{3}))|(?P<day_3>(?:0[1-9]|[12][0-9]|3[01])|(?:[1-9]|[12][0-9]|3[01]))(?:[\\.,]\\s+|\\s*)(?P<month_3>(?:January|February|March|April|May|June|July|August|September|October|November|December)|(?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec))(?:[\\.,]\\s+|\\s*)(?P<year_3>(?:[0-9]{3}[1-9]|[0-9]{2}[1-9][0-9]|[0-9][1-9][0-9]{2}|[1-9][0-9]{3}))|(?P<month_4>(?:January|February|March|April|May|June|July|August|September|October|November|December)|(?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec))\\s+(?P<year_4>(?:[0-9]{3}[1-9]|[0-9]{2}[1-9][0-9]|[0-9][1-9][0-9]{2}|[1-9][0-9]{3}))|(?P<month_5>(?:0[1-9]|1[12])|(?:[1-9]|1[12]))/(?P<year_5>(?:[0-9]{3}[1-9]|[0-9]{2}[1-9][0-9]|[0-9][1-9][0-9]{2}|[1-9][0-9]{3}))|(?P<year_6>(?:[0-9]{3}[1-9]|[0-9]{2}[1-9][0-9]|[0-9][1-9][0-9]{2}|[1-9][0-9]{3}))|(?P<month_6>(?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec))\\s+(?P<day_4>(?=(?:2?1st|2?2nd|2?3rd|[12]?[4-9]th|1[123]th|[123]0th|31st))[0-9][0-9]?)..,\\s*(?P<year_7>(?:[0-9]{3}[1-9]|[0-9]{2}[1-9][0-9]|[0-9][1-9][0-9]{2}|[1-9][0-9]{3}))'
Which would have been virtually impossible to write by hand.
The bounds for the "between random text" can be added around _pattern.
I would suggest _pattern = r'\b(?:{})\b'.format(_pattern).

Related

start date end date in python validation

1)get date as a string from user
2)convert to date object
3)find the first and last date of the month
4)append the list in the given format
5)find the next month of the date
6)repeat the step 3 and 5 until the date is lesser then the end date
Sample Input:
02/12/2022, 24/02/2023 (Check and validate the start , end dates)
Sample Output:
[
{'year': 2022, 'month': 'Dec', 'start_date': "01/12/2022", 'end_date': "31/12/2022", 'days': 31},
{'year': 2023, 'month': 'Jan', 'start_date': "01/01/2022", 'end_date': "31/01/2023", 'days': 31},
{'year': 2023, 'month': 'Feb', 'start_date': "01/02/2022", 'end_date': "28/02/2023", 'days': 28}
]

Fill in missing values for missing dates in dataframe

I have the following dataframe:
df = pd.DataFrame(
{
'status': ['open', 'closed', 'open', 'closed', 'open', 'closed', 'open', 'closed'],
'month': ['January 2020', 'January 2020', 'February 2020', 'February 2020', 'April 2020', 'April 2020', 'August 2020', 'August 2020'],
'counts': [10, 12, 32, 12, 19, 40, 10, 11]
}
)
status month counts
0 open January 2020 10
1 closed January 2020 12
2 open February 2020 32
3 closed February 2020 12
4 open April 2020 19
5 closed April 2020 40
6 open August 2020 10
7 closed August 2020 11
I'm trying to get a stacked bar plot using seaborn:
sns.histplot(df, x='month', weights='counts', hue='status', multiple='stack')
The purpose is to get a plot with a continuous timeseries without missing months. How can I fill in the missing rows with values so that the dataframe would look like below?
status month counts
open January 2020 10
closed January 2020 12
open February 2020 32
closed February 2020 12
open March 2020 0
closed March 2020 0
open April 2020 19
closed April 2020 40
open May 2020 0
closed May 2020 0
open June 2020 0
closed June 2020 0
open July 2020 0
closed July 2020 0
open August 2020 10
closed August 2020 11
You could pivot the dataframe, and then reindex with the desired months.
import pandas as pd
df = pd.DataFrame({'status': ['open', 'closed', 'open', 'closed', 'open', 'closed', 'open', 'closed'],
'month': ['January 2020', 'January 2020', 'February 2020', 'February 2020', 'April 2020', 'April 2020', 'August 2020', 'August 2020'],
'counts': [10, 12, 32, 12, 19, 40, 10, 11]})
months = [f'{m} 2020' for m in ['January', 'February', 'March', 'April', 'May', 'June', 'July', 'August']]
df_pivoted = df.pivot(values='counts', index='month', columns='status').reindex(months).fillna(0)
ax = df_pivoted.plot.bar(stacked=True, width=1, ec='black', rot=0, figsize=(12, 5))
A seaborn solution, could use order=. That doesn't work with a histplot, only with a barplot, which doesn't stack bars.
import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd
df = pd.DataFrame({'status': ['open', 'closed', 'open', 'closed', 'open', 'closed', 'open', 'closed'],
'month': ['January 2020', 'January 2020', 'February 2020', 'February 2020', 'April 2020', 'April 2020', 'August 2020', 'August 2020'],
'counts': [10, 12, 32, 12, 19, 40, 10, 11]})
plt.figure(figsize=(12, 5))
months = [f'{m} 2020' for m in ['January', 'February', 'March', 'April', 'May', 'June', 'July', 'August']]
ax = sns.barplot(data=df, x='month', y='counts', hue='status', order=months)
plt.tight_layout()
plt.show()

set and sort function in loop

I need to be able to use function (get_years) to iterate a list of reviews such as:
{'rating': 5.0,
'reviewer_name': 'Karen',
'product_id': 'B00004RFRV',
'review_title': 'Bialetti is the Best!',
'review_time': '11 12, 2017',
'images': ['https://images-na.ssl-images-amazon.com/images/I/81+XxFRGyBL._SY88.jpg'],
'styles': {'Size:': ' 12-Cup', 'Color:': ' Silver'}}```
{'rating': 3.0,
'reviewer_name': 'Peter DP',
'product_id': 'B00005OTXM',
'review_title': "Mr. Coffee DWX23 12-cup doesn't have the quality feel as my 13 year old nearly identical 12-cup Mr. Coffee",
'review_time': '04 17, 2015',
'images': ['https://images-na.ssl-images-amazon.com/images/I/71sFKwTW9sL._SY88.jpg'],
'styles': {'Style Name:': ' COFFEE MAKER ONLY'}}
{'rating': 5.0,
'reviewer_name': 'B. Laska',
'product_id': 'B00004RFRV',
'review_title': 'Love my Moka pots!',
'review_time': '07 9, 2015',
'images': ['https://images-na.ssl-images-amazon.com/images/I/719NCqw4GML._SY88.jpg'],
'styles': {'Size:': ' 1-Cup', 'Color:': ' Silver'}}
to be able to return:
print(get_years(reviews)) # [2007, 2008, 2009, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018]
print(type(get_years(reviews))) # <class 'list'>
I have:
def get_years(review):
review_years_set = set()
for review in reviews:
review_years_set.add(review['review_time'][-4:])
review_years_list = list(review_years_set)
review_years_list.sort()
return review_years_list
which gives me what I want but it seems like the longer route. Is there a more Pythonic or efficient way to get a sorted list of set values?
Given an iterable of string-formatted dates, e.g.:
dates = ['07 9, 2007', '04 1, 2008', '01 2, 2007', '08 2, 2014', '01 3, 2004', '01 4, 2004']
A concise way to produce a sorted list of unique years is as follows using set comprehension:
sorted_dates = sorted({int(date[-4:]) for date in dates})
print(sorted_dates)
Output:
[2004, 2007, 2008, 2014]
try this.
def get_years(reviews):
return sorted([review['review_time'][-4:] for review in reviews])
print(get_years(reviews))

max of different classes in pivot table pandas python

I have a dataset as following
df = pd.DataFrame([[2020, 'Jan', 1],
[2020, 'Jan', 2],
[2020, 'Jan', 3],
[2020, 'Feb', 4],
[2020, 'Feb', 5],
[2020, 'Feb', 6],
[2021, 'Jan', 7],
[2021, 'Jan', 8],
[2021, 'Jan', 9],
[2021, 'Feb', 10],
[2021, 'Feb', 11],
[2021, 'Feb', 12],
[2022, 'Jan', 13],
[2022, 'Jan', 14],
[2022, 'Jan', 15],
[2022, 'Feb', 16],
[2022, 'Feb', 17],
[2022, 'Feb', 18]],
columns=['Year', 'Month', 'Sale ($)'])
which shows the sales of different months and years.
Using pandas.pivot_table function, I can create a pivot table to calculate sum of sales for different months and years.
df.pivot_table(index='Month', columns='Year', aggfunc='sum')
And that generates a summary statistics table as following:
2020
2021
2022
Jan
6
24
42
Feb
15
33
51
Here is my question: How can I identify the month at which in each year I had the highest sale?
I know that for the years 2020, 2021, and 2022, my highest sales were 15, 33, and $ 51 respectively.
I can achieve this by adding .max() into the end of the code
df.pivot_table(index='Month', columns='Year',aggfunc='sum').max()
and this returns exactly that:
Year
Sale
2020
15
2021
33
2022
51
And in the case of this example, the maximum sales were all in February so how can I write a function that returns not the maximum value, but the month which had the maximum sale?

How do I convert a list/dictionary into a Dataframe?

I have a JSON response (sample below) that I'm trying to convert into a DataFrame. I've had several issues with the data being listed as columns (1 x 346), etc. I only need the 5 columns listed below:
area_name,
date,
month,
unemployment_rate,
year
Here's my code:
edd_ca_df = pd.DataFrame.from_dict(edd_ca, orient="index",
columns=["area_name", "month", "date", "year", "unemployment_rate"])
and here's a sample of the JSON response:
[[{'area_name': 'California',
'area_type': 'State',
'date': '1990-01-01T00:00:00.000',
'employment': '14099700',
'labor_force': '14953900',
'month': 'January',
'seasonally_adjusted_y_n': 'N',
'status_preliminary_final': 'Final',
'unemployment': '854200',
'unemployment_rate': '5.7',
'year': '1990'},
{'area_name': 'California',
'area_type': 'State',
'date': '1990-02-01T00:00:00.000',
'employment': '14206700',
'labor_force': '15049400',
'month': 'February',
'seasonally_adjusted_y_n': 'N',
'status_preliminary_final': 'Final',
'unemployment': '842800',
'unemployment_rate': '5.6',
'year': '1990'},
Any help would be greatly appreciated.
Since you have a list of dictionaries, this is as simple as passing all the data to a new DataFrame and specifying what columns you want to keep:
import pandas as pd
all_data = [{'area_name': 'California',
'area_type': 'State',
'date': '1990-01-01T00:00:00.000',
'employment': '14099700',
'labor_force': '14953900',
'month': 'January',
'seasonally_adjusted_y_n': 'N',
'status_preliminary_final': 'Final',
'unemployment': '854200',
'unemployment_rate': '5.7',
'year': '1990'},
{'area_name': 'California',
'area_type': 'State',
'date': '1990-02-01T00:00:00.000',
'employment': '14206700',
'labor_force': '15049400',
'month': 'February',
'seasonally_adjusted_y_n': 'N',
'status_preliminary_final': 'Final',
'unemployment': '842800',
'unemployment_rate': '5.6',
'year': '1990'}]
keep_columns = ['area_name','date','month','unemployment_rate','year']
df = pd.DataFrame(columns=keep_columns, data=all_data)
print(df)
Output
area_name date month unemployment_rate year
0 California 1990-01-01T00:00:00.000 January 5.7 1990
1 California 1990-02-01T00:00:00.000 February 5.6 1990

Categories

Resources