创建正则表达式模式以提取浮点数和整数

Question

I am having an issue in creating a pattern recognition function to extract all the numbers from a data frame column and print them.

I have tried to create a regex pattern after looking at the data camp tutorial and the other questions on stack overflow, but I have not been able to create a pattern that will extract all the numbers and print them. Essentially, the EA patterns that I created and the HR patterns with floats like say 1.12 are not returning results.

import re
import pandas as pd
data = ['1EA @ 3217.45;', 'ST - .63HR@165;', 'ST - .5HR@123;', 'ST - 1.08HR@165;', '1EA @ 3217.45;', 'ST - .85HR@165;', 'ST - .85HR@165;', '1EA @ 3217.45;', 'ST - .12HR@165;', 'OT - 1.12HR @ 165;', 'ST - .55HR@123;OT - 0.82HR @ 123;', 'ST - .5HR@165;', 'OT - 0.45HR @ 123;', 'ST - .6HR@123;', 'ST - 1.42HR@123;', '1EA @ 1500;', 'ST - .3HR@123;', 'ST - 1HR@111;OT - 0.25HR @ 111;']
Travel = pd.DataFrame(data, columns=['Rate Breakup Description'])

for a in Travel['Rate Breakup Description']:
    print(re.search('.(\d+)HR | (\d+)EA | (\d+)HR | (\d+)EA', a, re.I|re.M))

My objective is to be able to have a pattern recognition function that will extract all the numbers regardless of the different string patterns and print them in the order they appear.

Answer 1

You may use

Travel['Result'] = Travel['Rate Breakup Description'].str.findall(r'\d*\.?\d+(?=HR|EA)').apply(', '.join)

The pattern will match

• \\d* - 0+ digits
• \\.? - an optional .
• \\d+ - 1+ digits
• (?=HR|EA) - followed with HR or EA .

The .str.findall will return all matches it finds in an input string, and .apply(', '.join) will join the results with a comma+space.

If there is a single match expected in each input, you might use an alternative solution:

Travel['Result'] = Travel['Rate Breakup Description'].str.extract(r'(\d*\.?\d+)(?:HR|EA)', expand=False)

Here, (\\d*\\.?\\d+) is a capturing group due to the parentheses, this part is returned by .str.extract and (?:HR|EA) is a non-capturing group (so that it is not returned) matching either HR or EA .