Sed，匹配两个模式之间的所有行(包括)，然后将第二个匹配放在顶部。

Question

这就是我需要做的：

输入：(Python)

## This is a function,
## its purpose is... yada yada yada
def function_name(x):
    return x + 1

产出：(减价)

## function_name
This is a function,
its purpose is... yada yada yada

到目前为止我得到了：

sed -n '/## /,/def/ { /## \|def/ p }' TEST.py | cut -d' ' -f2- | sed 's/(.*)\(.*\)://'

它产生：

This is a function,
its purpose is... yada yada yada
function_name

有两个条件：

1.-不匹配这些模式的"##“和"def”之间的行应该被忽略。

示例：

输入：(Python)

## This is a function,
## its purpose is... yada yada yada
# This is a normal comment, nothing to see here! (ignored)
def function_name(x):
    return x + 1

产出：(减价)

## function_name
This is a function,
its purpose is... yada yada yada

2.-第二种模式必须加以推广，例如，不要使用"def"，比方说我要使用"class“。

输入：(Python)

## This is a class,
## its purpose is... yada yada yada
# This is a normal comment, nothing to see here! (ignored)
class class_name:
    __init__(self, x):
        self.x = x

产出：(减价)

## class_name
This is a class,
its purpose is... yada yada yada

Answer 1

使用awk

$ awk '/^## / { comment = comment substr($0, 4) "\n" }
       /^def / { printf "## %s\n%s", substr($0, 5, index($0, "(") - 5), comment
                 comment = "" }' TEST.py  
## function_name
This is a function,
its purpose is... yada yada yada

将以##开头的所有行追加到变量中，当它看到以def开头的行时，打印函数名和前面行的变量，然后空白该变量以重新启动。

使用持有空间存储注释行的GNU sed版本：

$ sed -n -e '/^## / { s/^## //; H }' -e '/^def / { s/^def \([^(]*\).*/## \1/; G; s/\n\n/\n/; p; z; x }' TEST.py
## function_name
This is a function,
its purpose is... yada yada yada

Answer 2

在每个Unix框上使用任何shell中的任何awk (如果awk不支持POSIX字符类，则使用a-zA-Z0-9而不是[:alnum:] )：

$ cat tst.awk
sub(/^## */,"") {
    cmts = cmts $0 ORS
    next
}

sub(/^[[:alnum:]_]+ */,"") {
    sub(/[^[:alnum:]_].*/,"")
    print "##", $0
    print cmts
    cmts = ""
}

$ awk -f tst.awk file
## function_name
This is a function,
its purpose is... yada yada yada

## function_name
This is a function,
its purpose is... yada yada yada

## class_name
This is a class,
its purpose is... yada yada yada

上面的内容是在这个输入文件上运行的：

$ cat file
## This is a function,
## its purpose is... yada yada yada
def function_name(x):
    return x + 1

## This is a function,
## its purpose is... yada yada yada
# This is a normal comment, nothing to see here! (ignored)
def function_name(x):
    return x + 1

## This is a class,
## its purpose is... yada yada yada
# This is a normal comment, nothing to see here! (ignored)
class class_name:
    __init__(self, x):
        self.x = x

Answer 3

awk '
/##/ { 
  gsub( "## *", "" )  
  previous[ ++lines ] = $0 
}
/def/ { 
  gsub( "def ", "")  
  gsub( "[(].*","" ) 
  print "## " $0  
  for (x = 1 ; x <= lines ; x++ ) 
    print previous[x] 
  lines = 0 
}
' TEST.py

看来肖恩比我早找到答案了。答案是相似的，它们的区别主要在于线条是如何累积的。Shawn将前面的行累加在一个变量(注释)中，而我则将其累加到数组中。

我认为Shawn在代码中可能有更多的假设，就像每个##在它之后都有一个空格，或者函数名从一行的第五个字符开始，而##和def都出现在行的开头。

我认为我也应该添加一些锚，因为/def/可能有很多意想不到的匹配，所以/^def/可能要好得多。

有趣的是，我们都忽略了使用sed，因为我个人不知道如何处理sed中的变量。我感兴趣的是，其他解决方案是否可以从纯sed社区获得。