Pandas多级索引列合并代码：为何'{0[0]}|{0[1]}'.format可行而另一种写法报错？

Hey there, let's break down exactly what's happening here and why your approach is throwing that error.

The Root Cause of the Error #

First, let's clarify the critical difference between the working solution and what you tried:

• The correct Pythonic code passes the format function itself to map:

  df.columns = df.columns.map('{0[0]}|{0[1]}'.format)
  

• What you attempted was calling format immediately and passing the resulting string to map (likely in a line like df.columns.map('{}|{}'.format([0],[1]))), which is what triggered the error.

Here's why that breaks:
When you run '{}|{}'.format([0],[1]), you're executing the format call right away, which returns a plain string like '[0]|[1]'. The map function expects a callable function that it can run on each element of df.columns (each element is a tuple from your multi-level index). When you pass a string instead of a function, map tries to "call" that string with each index tuple as an argument—hence the error: a string isn't a callable function.

How the Correct Approach Works #

Let's unpack the working code step by step:

1. df.columns is a multi-level index, so each entry is a tuple (e.g., ('index1', 'subindex1')).
2. '{0[0]}|{0[1]}'.format refers to the format method bound to that string template. This is a callable function that accepts one argument (the tuple from the index).
3. When map runs, it passes each index tuple to this format function:
   • For the tuple ('index1', 'subindex1'), {0[0]} grabs the first element ('index1'), {0[1]} grabs the second ('subindex1'), and they're joined with |.

A More Readable Alternative #

If the {0[0]} syntax feels confusing, you can use a lambda function instead—it does the same thing but is more explicit:

df.columns = df.columns.map(lambda col_tuple: f"{col_tuple[0]}|{col_tuple[1]}")

This lambda takes each index tuple directly, extracts its two elements, and joins them with | using an f-string.

内容的提问来源于stack exchange，提问作者Zac