CODEWITHSUNDEEP
pythonpandas
MaxB
MaxB

Reputation: 458

Perform string replace using column name as string value to replace with

I want to perform a string replace on my dataframe where I find all instances of "X" in a column and replace it with the column name.

ex

Name  FFF1  H0L1
  -    L     -
  -    X     L
  X    -     -
  -    -     X

result df after replace 

Name     FFF1      H0L1
  -      FFF1        -
  -      FFF1      H0L1
Name      -          -
  -       -        H0L1

It seems pretty straightforward, I am just confused on how to "reference" the column name. Thoughts?

Upvotes: 0

Views: 201

Answers (2)

kantal
kantal

Reputation: 2407

The 'apply' method iterates over the columns as series which 'name' attribute corresponds to the column name:

df.apply(lambda col: col.where(~col.str.contains("X"), \
                        col.str.replace("X",col.name)) )

Even better:

df.apply(lambda col: col.str.replace("X",col.name))

Edit: Answering the additional question: Use regular expression: 

#df.apply(lambda col: col.str.replace(r"([^X]|^)(X)([^X]|$)",r"\1"+col.name+r"\3")) # didn't work correctly in all situation, e.g.: "aXbXcXd"
df.apply(lambda col: col.str.replace(r"([^X]|^)(X)(?=[^X]|$)",r"\1"+col.name))


"""  The details:
     We create three pattern groups: (...) 
     [^X] can be any char but X (^ in square br. negates the chars)  
     ^ as a separate char means start of string; 
     $ means end of string; 
     | means 'or'. 
     \1 and \2 mean the corresponding groups;
     (?=...) lookahead check
"""

Edit 2: If there is always one char in the cell to be replaced:

df.apply(lambda col: col.replace(["X","L"],col.name))

Upvotes: 1

Quang Hoang
Quang Hoang

Reputation: 150785

you can use df.where:

df = pd.DataFrame({"A": ['-', 'X'],
                   'B': ['X', '-']})
df.where(df.eq('X'), df.columns)

Output:

   A  B
0  A  X
1  X  B

Upvotes: 0

Related Questions