1 minute read

Yesterday, I encountered an analysis issue that appeared to be resolvable with a simple pattern replacement technique.

This turned out to be a bit more complex than estimated, as the pattern spanned multiple lines and I struggled to get the regex constructed to match this use case.

I finally figured it out, and although the overall problem remains unresolved (it has many layers), this particular part works well.  My error was in how I viewed use of  /s and /m as mutually exclusive; they aren’t of course, and I have made that mistake one other time over the years.   I also hadn’t set the input record separator ($/) to paragraph read mode.   I’m posting details here to document my specific usage.


  • A literal pattern exists that is present in multiple places in a file
  • The literal pattern spans multiple lines in the file
  • Literal pattern is 
    • Node-path: branches/%%BRANCHES%%
    • Node-kind: dir
    • Node-action: add
  • Perl is being used to read the file and current line of file is $line
  • Objective is to delete all occurrences of this pattern from the file one $line at a time

Here’s how I did that:

[read the file & crawl thru the lines]

$/ = '';

$line =~ s/^Node-path:sbranches/%%BRANCHES%%.*?^Node-kind:sdir.*?^Node-action:sadd.*?//sm;

Tags: , ,