Find empty tag in `xml` using Unix command

Question

The xml file gets generated with empty tags is like this

<headertag>
</headertag>

It's in consecutive lines and i usually use a perl script to find it based on new line character between tags as pattern (\n) but Unix environment which I'm using now isn't supporting this perl script. Tried the same with sed but i couldn't find the place (line number) where there is a empty tag like this in a huge .xml file. Is there a solution to find this using sed or awk?

Answer 1

To parse XML, use an XML-aware tool. I know you said you have restrictions on the system you're on, but I'm going to give you a solution that will work once you convince your manager/whatever that for robust and reliable operation, you need to install XMLStarlet to do proper XML parsing.

With XMLStarlet, given the XML file

<?xml version="1.0"?>
<root>
  <headertag>
    <subtag/>
  </headertag>
  <headertag>
    <subtag>Don't delete me!</subtag>
  </headertag>
  <headertag>
</headertag>
  <headertag>
Not empty
</headertag>
</root>

the following removes all empty tags, no matter what their names are or where in the document they occur:

$ xmlstarlet ed -d '//*[not(normalize-space())]' file.xml >newfile.xml

$ cat newfile.xml
<?xml version="1.0"?>
<root>
  <headertag>
    <subtag>Don't delete me!</subtag>
  </headertag>
  <headertag>
Not empty
</headertag>
</root>

The xmlstarlet command may sometimes be called just xml depending on how it's packaged on your system.

XMLStarlet is available here: http://xmlstar.sourceforge.net/ ... but see if your default package manager has it first.

Answer 2

Assumptions:

• Empty tags that we are after are on a line by themselves.
• The closing tag is also stands on it's own line + it immediately follows.
• Whitespace is assumed to compose of spaces and not TABs.

---

sed -ne '
   /^ *\(<[^><]*>\) *$/!d                                   # tag opening should be on a line of its own
   s//\1/                                                   # strip away all whitespace
   $d; N                                                    # if the tag opening is on the last line, we dont need it. Otherwise, we grab the next line
   s/^.\(.*\).\n *<\/\1> *$/Empty tag: <\1> on line num\#/p # print only in case the tagnames match across lines => we have found an empty tag
   /\n/!=                                                   # print empty tag line num
   D                                                        # delete the pattern space
' yourfile.xml

Answer 3

You have not told us what you want to do with the empty tags. I am therefore going to assume that you wish to convert such tag pairs into the format <headertag/>

Here is a simple sed script which will do exactly that for the example case you provided us:

sed -E ':a;N;$!ba;s#<([^>]+)>\n</\1>#<\1/>#g' infile > outfile

Answer 4

You can use "pcregrep" instead of grep which supports multiline patterns and has more options.

This pattern finds two consecutive lines containing open and close tags:

pcregrep -M -v "<headertag>\n\s*?</headertag>" file

• which -M stands for Multiline
• -v stands for inverse match
• \n is newline character supported in pcregrep but not grep (I think)
• \s is space character.
• *? any character

this pattern removes all empty tags.