How to get line strings with special word start and end? [duplicate]

Question

This question already has answers here:

Multiline pattern match using sed, awk or grep [duplicate] (5 answers)

Closed 7 years ago.

Here is the file strings.xml which includeing 3 sentences:

1.string name="schedulelist_nofiles">Não existe agenda de registro! Por favor,

here is missing line tecla "Add" (Adicionar) para adicionar uma. string

2.You should skip this line!

3.string name="Programme_name">GUIA DE PROGRAMA string

I use cat command 'cat strings.xml | grep "string name=" but it only get below lines:

string name="schedulelist_nofiles">Não existe agenda de registro! Por favor,

string name="Programme_name">GUIA DE PROGRAMA string

I want to get the complete line like this , just like sentence 3, start with 'string name' and end with "string". What can I do for this?

• francois P: you can try sed '/first words/,/last words/ !d' to select lines that match this pattern be careful to choose the rigth pattern to not select paragraph :)

Answer 1

Below sed oneliner can be used to achieve the same. Tested it worked fine

First method1

sed -n '/^string name.*string$/p' filename

In method1 It will searches for line beginning with "string name" and ends with "string"

Method 2

sed -n '/^string name/p' filename | sed -n '/string$/p'

In method2 First it will searches for line beginning with "string name" and then it searches for line ending with string

Output

string name="Programme_name">GUIA DE PROGRAMA string

Answer 2

With sed, assuming those string name= ... string take up full lines and that all string name=s are terminated by a string:

sed '
 /^string name=/!d
 :1
 / string$/!{
 N
 b1
 }' < strings.xml

Answer 3

With pcregrep:

pcregrep -Mo '(?s)\bstring name=.*?\bstring\b' < file.xml

• -M: multi-line mode where pcregrep pulls more lines from the input as needed to satisfy the regex
• -o: print the portion that matches the regexp like in GNU grep (you can omit it or replace with -x if the regex matches full lines.
• (?s): activate the s flag that causes . to also match on newline
• .*?: non-greedy version of .*: any number of characters, as few as possible.
• \b: word boundary.

If you don't have pcregrep, you can do the same with perl with something like:

perl -l -0777 -ne 'print for /\bstring name=.*?\bstring\b/gs' < file.xml

Though that means loading the whole file in memory.

• Huang: I use the Linux server to scripting, and I don't have root permission, so the command pcregrep can not be installed, other command can replace it?
• Huang: Thanks very much. If above 3 scentens I just want to get the first one, how can I do? I mean only need to capture the string ID 'schedulelist_nofiles' sentence, don't includeing the last scentens, how can I filter it? Because it is also including keys words 'string name= ... string'.