python format 使用 技巧

https://pyformat.info/

Basic formatting

Simple positional formatting is probably the most common use-case. Use it if the order of your arguments is not likely to change and you only have very few elements you want to concatenate.

Since the elements are not represented by something as descriptive as a name this simple style should only be used to format a relatively small number of elements.

Old

'%s %s' % ('one', 'two')

New

'{} {}'.format('one', 'two')

Output

one two

Old

'%d %d' % (1, 2)

New

'{} {}'.format(1, 2)

Output

1 2

With new style formatting it is possible (and in Python 2.6 even mandatory) to give placeholders an explicit positional index.

This allows for re-arranging the order of display without changing the arguments.

This operation is not available with old-style formatting.

New

'{1} {0}'.format('one', 'two')

Output

two one

Value conversion

The new-style simple formatter calls by default the __format__() method of an object for its representation. If you just want to render the output of str(...) or repr(...) you can use the !s or !r conversion flags.

In %-style you usually use %s for the string representation but there is %r for a repr(...) conversion.

Setup

class Data(object):

    def __str__(self):
        return 'str'

    def __repr__(self):
        return 'repr'

Old

'%s %r' % (Data(), Data())

New

'{0!s} {0!r}'.format(Data())

Output

str repr

In Python 3 there exists an additional conversion flag that uses the output of repr(...) but uses ascii(...)instead.

Setup

class Data(object):

    def __repr__(self):
        return 'räpr'

Old

'%r %a' % (Data(), Data())

New

'{0!r} {0!a}'.format(Data())

Output

räpr r\xe4pr

Padding and aligning strings

By default values are formatted to take up only as many characters as needed to represent the content. It is however also possible to define that a value should be padded to a specific length.

Unfortunately the default alignment differs between old and new style formatting. The old style defaults to right aligned while for new style it's left.

Align right:

Old

'%10s' % ('test',)

New

'{:>10}'.format('test')

Output

      test

Align left:

Old

'%-10s' % ('test',)

New

'{:10}'.format('test')

Output

test      

Again, new style formatting surpasses the old variant by providing more control over how values are padded and aligned.

You are able to choose the padding character:

This operation is not available with old-style formatting.

New

'{:_<10}'.format('test')

Output

test______

And also center align values:

This operation is not available with old-style formatting.

New

'{:^10}'.format('test')

Output

   test   

When using center alignment where the length of the string leads to an uneven split of the padding characters the extra character will be placed on the right side:

This operation is not available with old-style formatting.

New

'{:^6}'.format('zip')

Output

 zip  

Truncating long strings

Inverse to padding it is also possible to truncate overly long values to a specific number of characters.

The number behind a . in the format specifies the precision of the output. For strings that means that the output is truncated to the specified length. In our example this would be 5 characters.

Old

'%.5s' % ('xylophone',)

New

'{:.5}'.format('xylophone')

Output

xylop

Combining truncating and padding

It is also possible to combine truncating and padding:

Old

'%-10.5s' % ('xylophone',)

New

'{:10.5}'.format('xylophone')

Output

xylop     

Numbers

Of course it is also possible to format numbers.

Integers:

Old

'%d' % (42,)

New

'{:d}'.format(42)

Output

42

Floats:

Old

'%f' % (3.141592653589793,)

New

'{:f}'.format(3.141592653589793)

Output

3.141593

Padding numbers

Similar to strings numbers can also be constrained to a specific width.

Old

'%4d' % (42,)

New

'{:4d}'.format(42)

Output

  42

Again similar to truncating strings the precision for floating point numbers limits the number of positions after the decimal point.

For floating points the padding value represents the length of the complete output. In the example below we want our output to have at least 6 characters with 2 after the decimal point.

Old

'%06.2f' % (3.141592653589793,)

New

'{:06.2f}'.format(3.141592653589793)

Output

003.14

For integer values providing a precision doesn't make much sense and is actually forbidden in the new style (it will result in a ValueError).

Old

'%04d' % (42,)

New

'{:04d}'.format(42)

Output

0042

Signed numbers

By default only negative numbers are prefixed with a sign. This can be changed of course.

Old

'%+d' % (42,)

New

'{:+d}'.format(42)

Output

+42

Use a space character to indicate that negative numbers should be prefixed with a minus symbol and a leading space should be used for positive ones.

Old

'% d' % ((- 23),)

New

'{: d}'.format((- 23))

Output

-23

Old

'% d' % (42,)

New

'{: d}'.format(42)

Output

 42

New style formatting is also able to control the position of the sign symbol relative to the padding.

This operation is not available with old-style formatting.

New

'{:=5d}'.format((- 23))

Output

-  23

New

'{:=+5d}'.format(23)

Output

+  23

Named placeholders

Both formatting styles support named placeholders.

Setup

data = {'first': 'Hodor', 'last': 'Hodor!'}

Old

'%(first)s %(last)s' % data

New

'{first} {last}'.format(**data)

Output

Hodor Hodor!

.format() also accepts keyword arguments.

This operation is not available with old-style formatting.

New

'{first} {last}'.format(first='Hodor', last='Hodor!')

Output

Hodor Hodor!

Getitem and Getattr

New style formatting allows even greater flexibility in accessing nested data structures.

It supports accessing containers that support __getitem__ like for example dictionaries and lists:

This operation is not available with old-style formatting.

Setup

person = {'first': 'Jean-Luc', 'last': 'Picard'}

New

'{p[first]} {p[last]}'.format(p=person)

Output

Jean-Luc Picard

Setup

data = [4, 8, 15, 16, 23, 42]

New

'{d[4]} {d[5]}'.format(d=data)

Output

23 42

As well as accessing attributes on objects via getattr():

This operation is not available with old-style formatting.

Setup

class Plant(object):
    type = 'tree'

New

'{p.type}'.format(p=Plant())

Output

tree

Both type of access can be freely mixed and arbitrarily nested:

This operation is not available with old-style formatting.

Setup

class Plant(object):
    type = 'tree'
    kinds = [{'name': 'oak'}, {'name': 'maple'}]

New

'{p.type}: {p.kinds[0][name]}'.format(p=Plant())

Output

tree: oak

Datetime

New style formatting also allows objects to control their own rendering. This for example allows datetime objects to be formatted inline:

This operation is not available with old-style formatting.

Setup

from datetime import datetime

New

'{:%Y-%m-%d %H:%M}'.format(datetime(2001, 2, 3, 4, 5))

Output

2001-02-03 04:05

Parametrized formats

Additionally, new style formatting allows all of the components of the format to be specified dynamically using parametrization. Parametrized formats are nested expressions in braces that can appear anywhere in the parent format after the colon.

Old style formatting also supports some parametrization but is much more limited. Namely it only allows parametrization of the width and precision of the output.

Parametrized alignment and width:

This operation is not available with old-style formatting.

New

'{:{align}{width}}'.format('test', align='^', width='10')

Output

   test   

Parametrized precision:

Old

'%.*s = %.*f' % (3, 'Gibberish', 3, 2.7182)

New

'{:.{prec}} = {:.{prec}f}'.format('Gibberish', 2.7182, prec=3)

Output

Gib = 2.718

Width and precision:

Old

'%*.*f' % (5, 2, 2.7182)

New

'{:{width}.{prec}f}'.format(2.7182, width=5, prec=2)

Output

 2.72

The nested format can be used to replace any part of the format spec, so the precision example above could be rewritten as:

This operation is not available with old-style formatting.

New

'{:{prec}} = {:{prec}}'.format('Gibberish', 2.7182, prec='.3')

Output

Gib = 2.72

The components of a date-time can be set separately:

This operation is not available with old-style formatting.

Setup

from datetime import datetime
dt = datetime(2001, 2, 3, 4, 5)

New

'{:{dfmt} {tfmt}}'.format(dt, dfmt='%Y-%m-%d', tfmt='%H:%M')

Output

2001-02-03 04:05

The nested formats can be positional arguments. Position depends on the order of the opening curly braces:

This operation is not available with old-style formatting.

New

'{:{}{}{}.{}}'.format(2.7182818284, '>', '+', 10, 3)

Output

     +2.72

And of course keyword arguments can be added to the mix as before:

This operation is not available with old-style formatting.

New

'{:{}{sign}{}.{}}'.format(2.7182818284, '>', 10, 3, sign='+')

Output

     +2.72

Custom objects

The datetime example works through the use of the __format__() magic method. You can define custom format handling in your own objects by overriding this method. This gives you complete control over the format syntax used.

This operation is not available with old-style formatting.

Setup

class HAL9000(object):

    def __format__(self, format):
        if (format == 'open-the-pod-bay-doors'):
            return "I'm afraid I can't do that."
        return 'HAL 9000'

New

'{:open-the-pod-bay-doors}'.format(HAL9000())

Output

I'm afraid I can't do that.

本文参与腾讯云自媒体分享计划,欢迎正在阅读的你也加入,一起分享。

发表于

我来说两句

0 条评论
登录 后参与评论

相关文章

来自专栏Netkiller

String boot with Apache kafka 完整的发布订阅例子

本文节选自电子书《Netkiller Java 手札》地址 http://www.netkiller.cn/ 5.21.7. 完整的发布订阅实例 上面的例子仅仅...

37260
来自专栏Netkiller

Spring Cloud Config

摘要: 本文节选自《Netkiller Java 手札》 Spring Cloud Config 本文节选自《Netkiller Java 手札》 https:...

39470
来自专栏一个会写诗的程序员的博客

LoggerFactory is not a Logback LoggerContext but Logback is on the classpath. Either remove Logba...

LoggerFactory is not a Logback LoggerContext but Logback is on the classpath. Ei...

51650
来自专栏Netkiller

Phalcon VS Spring 用法对照手册

Phalcon VS Spring 摘要 Phalcon VS Spring 用法对照表 ---- 目录 1. Install 1.1. Phalcon 1.2...

42860
来自专栏Netkiller

Spring boot with Thymeleaf

本文节选自电子书《Netkiller Java 手札》 5.19. Spring boot with Thymeleaf 5.19.1. Maven <dep...

370130
来自专栏光变

SpringMVC 使用Valid注解校验数据

12130
来自专栏SpringBoot

RESTful

17320
来自专栏云知识学习

kubernetes 基础集群排障

在排错过程中,kubectl 是最重要的工具,通常也是定位错误的起点。这里也列出一些常用的命令,在后续的各种排错过程中都会经常用到。

1.5K120
来自专栏ml

hdu-----(1532)Drainage Ditches(最大流问题)

Drainage Ditches Time Limit: 2000/1000 MS (Java/Others)    Memory Limit: 65536/3...

30490
来自专栏码匠的流水账

redis的sentinel模式故障演练

redis的sentinel模式相对cluster来说比较简单,缺点是需要浪费一些资源来做sentinel节点,对于中小数据量的业务来说,相对比较划算。

15910

扫码关注云+社区

领取腾讯云代金券