I have a log which contains XML lines. Sample format below:
<head> <body> <line> asdasd</line> </body> </head> I want to scan the log file and append the lines that do not start with '<' to the previous line. Output would be like below:
<head> <body> <line>asdasd</line> </body> </head> Thanks
I think I've said this before - but at risk of sounding like a stuck record - DON'T use regular expressions to parse XML. It's brittle and prone to breaking. I would ask first though - why are you trying to do what you're doing? Because it should be irrelevant when working with your XML.
Instead use a parser:
#!/usr/bin/env perl use strict; use warnings; use XML::Twig; my $twig = XML::Twig->parsefile('your_file.xml'); foreach my $elt ( $twig->get_xpath('//#PCDATA') ) { $elt->set_text( $elt->trimmed_text ); } $twig->set_pretty_print('indented_a'); $twig->print; This does what you want... but if you're actually working with the XML normally, that trimmed_text method probably removes the need for this processing anyway.
Perl to the rescue!
perl -pe 'print "\n" if /^\s*+</; chomp;' input > output i.e. newline is removed from each line, and it's printed when the next line starts with whitespace followed by a <.
To keep the final newline, change chomp to chomp unless eof or add END { print "\n" }
Almost standard sed procedure
sed '$!N;s/\n\(\s*[^<[:blank:]]\)/\1/;P;D' log.xml Using the XPath function normalize-space to remove the initial newline of the /head/body/line node:
xmlstarlet edit --update '/head/body/line' --expr 'normalize-space(text())' file.xml Or, using abbreviated names:
xmlstarlet ed -u '/head/body/line' -x 'normalize-space(text())' file.xml The output, given the input in the question, would be
<?xml version="1.0"?> <head> <body> <line>asdasd</line> </body> </head> Use //line in place of the full path from the root node if you want to affect all line nodes in your input document.
Add -O or --omit-decl after edit or ed to discard the <?xml ...> declaration at the start of the resulting document.
