To remove a tag from a xml file

You didn't specify that you could only use sed, so if you have access to xml_grep (see Merge multiple XML files from commend line, second answer), I would recommend that because it does a lot of the heavy work for you and for a simple merge job like this can be done in one command:

xml_grep --cond Record --wrap "ns0:collection" --descr 'xmlns:ns0="http://namespace/Service/1.0"' --encoding "UTF-8" *.xml

Test files:

test.xml

<?xml version="1.0" encoding="UTF-8" ?><ns0:collection
xmlns:ns0="http://namespace/Service/1.0""><Record>
Test
</Record></ns0:collection>

test1.xml

<?xml version="1.0" encoding="UTF-8" ?><ns0:collection
xmlns:ns0="http://namespace/Service/1.0"><Record>
Test 1<a>a</a><b c="c">d</b>
</Record></ns0:collection>

Result

<?xml version="1.0" encoding="UTF-8" ?>
<ns0:collection xmlns:ns0="http://namespace/Service/1.0">
<Record>
Test 1<a>a</a><b c="c">d</b></Record><Record>
Test
</Record>
</ns0:collection>

I prefer to use XML-aware tools when dealing with XML files, because the chance of messing up the structure with sed and friends is quite high and you can easily end up with a malformed XML document!

I would suggest that using sed isn't good for processing XML, and instead using a parser.

I would also suggest that you have an XY problem here - it's not about deleting tags, but rather merging XML files.

Personally - I like perl and XML::Twig:

#!/usr/bin/env perl
use strict;
use warnings;

#load the parser
use XML::Twig; 

#get our file list - we use the "first" file as the basis.
#can use sort on this list if desired. 
my ( $first_file, @other_files ) = glob ( 'C://tmp//xmltest/*.xml' ); 

#Our 'parent' document. 
my $doc = XML::Twig -> new -> parsefile ( $first_file ); 


foreach my $file ( @other_files ) { 
   my $mergedoc = XML::Twig -> new -> parsefile ( $file ); 

   #//Record means any <Record> node anywhere in the tree. 
   foreach my $record ( $mergedoc -> get_xpath ( '//Record' ) ) {
      $record -> cut;
      #paste it into our parent doc, as the last node. 
      $record -> paste ( after => $doc -> root -> last_child );
   }
}

#set output formatting (optional)
$doc -> set_pretty_print ('indented_a'); 

#print to STDOUT.
$doc -> print;

#write to output file too
open ( my $output, '>', 'combined.xml' ) or die $!;
print {$output} $doc -> sprint;
close ( $output );

This deliberately extracts the Record elements from the target XML, and just merges those between the documents. However it's a flexible approach - xpath is quite powerful, and is the XML equivalent of regex - but works better because it's context aware, where regex is not.

Solutions:

1. I need to remove only closing tag from the first file, solution:

   sed -i.bak -e 's/<\/ns0:collection>/ /' -e 's/<\/Record>/ /' n0
   

2. remove both <?xml version="1.0" encoding="UTF-8" ?><ns0:collection xmlns:ns0="http://namspace/Service/1.0"> and </ns0:collection> in the next (n-1) files:

   sed -i.bak -e 's/<?xml version=1.0 encoding=UTF-8 ?>.*<ns0:collection/ /' -e 's/xmlns.*/ /' -e 's/<\/R.*>.*>/ /' n1
   

   • Doing so for file name range:

     find . -type f -name "n[1-3]" -exec sed i.bak -e 's/<?xml version=1.0 encoding=UTF-8 ?>.*<ns0:collection/ /' -e 's/xmlns.*/ /' -e 's/<\/R.*>.*>/ /' {} \;
     

3. Have to remove only in the last file and merge all of them together:

   sed i.bak -e 's/<?xml version=1.0 encoding=UTF-8 ?>.*<ns0:collection/ /' -e 's/xmlns.*/ /' ne
   

Then finaly join them:

cat n0 n[1-3] ne > joined

I used the following files: n0,n1, n2, n3, and ne. I added the following text into each:

<?xml version="1.0" encoding="UTF-8" ?><ns0:collection
xmlns:ns0="http://namspace/Service/1.0"><Record>
hello from nigeria
</Record></ns0:collection>

The resulting file joined was as seen below:

<?xml version="1.0" encoding="UTF-8" ?><ns0:collection
xmlns:ns0="http://namspace/Service/1.0"><Record>
hello from nigeria



hello from nigeria



hello from nigeria



hello from nigeria



hello from nigeria
</Record></ns0:collection>

Note:

1. I see from the first problem you need to remove both </Record></ns0:collection> not just </ns0:collection>. So I took the liberty to do, else we would have a duplicate </Record> entry when the files are merged.

2. The filenames here would have to be modified by you so you can run one command over them all, here I used n[1-3]. Pick that which works best for you.

3. PLEASE RUN A TEST FIRST AND SEE THE RESULTS, here I used i.bak so sed creates a backup automatically.