Python: Replace angular bracket and its string within by curl bracket in each row in pandas

Question

This has a reference to this SO thread .

For the sake of newness, I am reproducing the dataframe with a small change:

ID         Static_Text                                           Params
1      Today, <adj1> is quite Sunny. Tomorrow, <adj2>           1-10-2020  
       may be little <adj3>
1      Today, <adj1> is quite Sunny. Tomorrow, <adj2>           2-10-2020
       may be little <adj3>
1      Today, <adj1> is quite Sunny. Tomorrow, <adj2>           Cloudy
       may be little <adj3>
2      Let's have a coffee break near <adj1>, if I              Balcony
       don't get any SO reply by <adj2>
2      Let's have a coffee break near <adj1>, if I               30
       don't get any SO reply by <adj2> mins

Now I want to replace the whole <adj> by {} where 1st occurrence of i.e. <adj1> shall be replaced by {0}. So the resultant dataframe would look like follows:

ID         Static_Text                                           Params
1      Today, {0} is quite Sunny. Tomorrow, {1}              1-10-2020  
       may be little {2}
1      Today, {0} is quite Sunny. Tomorrow, {1}              2-10-2020
       may be little {2}
1      Today, {0} is quite Sunny. Tomorrow, {1}              Cloudy
       may be little {2}
2      Let's have a coffee break near {0}, if I              Balcony
       don't get any SO reply by {1}
2      Let's have a coffee break near {0}, if I              30
       don't get any SO reply by {1} mins

I am trying the following:

def replace_angular(df):
   if '<' and '>' in df['Static_Text']:
       rep_txt = re.sub(r'\<[^>]*\>',{},df[Static_Text'])
   return rep_txt

df = df.apply(lambda x : replace_angular(x),axis=1)

But I am not too sure about the above code snippet. Especially how to bring 0,1 etc within {}.

Henry Yik · Accepted Answer · 2020-10-14 13:16:43Z

1

IIUC you can pass a lambda function in str.replace as replacement:

df["Static_Text"].str.replace(r"<[A-Za-z]+(\d+)>", lambda m: '{'+f'{int(m.group(1))-1}'+'}')

0                     Today, {0} is quite Sunny. Tomorrow, {1} may be little {2}
1                     Today, {0} is quite Sunny. Tomorrow, {1} may be little {2}
2                     Today, {0} is quite Sunny. Tomorrow, {1} may be little {2}
3         Let's have a coffee break near {0}, if I don't get any SO reply by {1}
4    Let's have a coffee break near {0}, if I don't get any SO reply by {1} mins

edited Oct 14, 2020 at 13:16

answered Oct 14, 2020 at 12:49

Henry Yik

22.6k5 gold badges21 silver badges44 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

pythondumb Over a year ago

Many thanks. Two questions: 1. Can I use my regex while you are using str.replace(r"<adj(\d)>" part? Because the word adj might change. 2. What exactly {int(m.group(0)[-2])-1} is doing? What if instead of 0,1,2 I have 0....15?

Henry Yik Over a year ago

For regex i'd probably use r"<[A-Za-z]+(\d)>" instead. I updated the above and use m.group(1) instead and it should adapt to any number.

pythondumb Over a year ago

Will your regex capture any space or special character within <adj>? Like <some# tag>? Can I use r"<[A-Za-z0-9.#$@]>" this regex instead?

辜乘风 · Accepted Answer · 2020-10-14 13:00:42Z

0

if the number in adj over 9, should change "m.group(0)[-2]" to "m.group(0)[4:-1]".

print (df["Static_Text"].str.replace(r"<adj(\d)>", lambda m: "{"+f"{int(m.group(0)[4:-1])-1}"+"}"))

answered Oct 14, 2020 at 13:00

辜乘风

1526 bronze badges

1 Comment

pythondumb Over a year ago

Many Thanks. I dont want that adj to be part of the regex as this might change. Secondly, what if I have 15 parameters i.e. {0},{1}...{15}? How int(m.group(0)[4:-1] gonna help?

Collectives™ on Stack Overflow

Python: Replace angular bracket and its string within by curl bracket in each row in pandas

2 Answers 2

3 Comments

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related