CLI tool to format XML files keeping empty lines and comments

Is there any CLI tool for Linux which formats XML files keeping any empty lines and comments? I have tried xmllint, tidy and xmlstarlet, but all seem to focus completely cleaning XML files rather than just indentation and spacing.

Answer

Try xmlindent. It has several options like -nbe and -nba and others that configure the handling of spaces before and after.

Given an XML input of

<?xml version="1.0"?>
<Response>
  <TroubleResponse>
    <Check>
      <DStatus>
        <GID>123456789</GID>
        <FLAG/>
      </DStatus>
    </Check>
    <RAM>
      <Details>
        <RAMID>5555777788
        </RAMID>
      </Details>
    </RAM>
    <RAM>
      <Details>
        <RAMID>
            5555777788</RAMID>
      </Details>
    </RAM>
  </TroubleResponse>
</Response>

The output can be configured with the following options (an excerpt):

-t     Use tabs instead of spaces
-nas   Suppress newline after start-tag
-nae   Suppress newline after end-tag
-nbs   Suppress newline before start-tag
-nbe   Suppress newline before end-tag
-f     Force newline on elements without children

So xmlindent -f and xmlindent -nba would produce the following output:

<?xml version="1.0"?>
<Response>
    <TroubleResponse>
        <Check>
            <DStatus>
                <GID>123456789       <!-- Change -->
                </GID>
                <FLAG/>
            </DStatus>
        </Check>
        <RAM>
            <Details>
                <RAMID>5555777788
                </RAMID>
            </Details>
        </RAM>
        <RAM>
            <Details>
                <RAMID>             <!-- Change -->
                    5555777788
                </RAMID>
            </Details>
        </RAM>
    </TroubleResponse>
</Response>

And xmlindent -nbe would produce the following output:

<?xml version="1.0"?>
<Response>
    <TroubleResponse>
        <Check>
            <DStatus>
                <GID>123456789</GID>
                <FLAG/>
            </DStatus>
        </Check>
        <RAM>
            <Details>
                <RAMID>5555777788
                </RAMID>
            </Details>
        </RAM>
        <RAM>
            <Details>
                <RAMID>                  <!-- Change -->
                5555777788</RAMID>
            </Details>
        </RAM>
    </TroubleResponse>
</Response>

xmlindent is not perfect as it does not seem to always realize the expected outcome, but it can be somewhat configured.